State of the Art on Energy Developments

State of the Art on Energy Developments Editors: Professor Abdul Ghani Olabi and Dr Abed Alaswad

SEEP Conference 2015 University of the West of Scotland, Paisley Campus Tuesday 11th – Friday 14th August 2015

State of the Art on Energy Developments Proceedings of the 8th International Conference on Sustainable Energy & Environmental Protection - Part 1

EDITED BY:

Prof Abdul Ghani Olabi & Dr Abed Alaswad

University of the West of Scotland School of Engineering & Computing Institute of Engineering & Energy Technologies

© Abdul Ghani Olabi and Abed Alaswad 2015

“State of the Art on Energy Developments” First Published in 2015 by University of the West of Scotland High Street, Paisley, PA1 2BE UK The authors have asserted their moral rights ISBN: 978-1-903978-52-8 All rights reserved.

Table of Contents Biofuel Technologies Effects of Mechanical Pretreatment on Biogas Production from Waste Paper C. Rodriguez, A. Alaswad, T. Prescott, A.G. Olabi ............................................................1 Farmer’s Willingness to Grow Sweet Sorghum as an Annual Bioenergy Crop under Contract: Assessing Willingness to Pay for Contract Attributes J.S. Bergtold, A. Shanoyan, J.E. Fewell and J.R. Williams ...............................................7 Microalgae Cell Rupture Using High Pressure Homogenization as a Prelude to Lipid Extraction C. Onumaegbu, A. Alaswad, T. Prescott and A. Olabi ......................................................16 Catalytic Cracking of Triglyceride over Sulphated Zirconia Solid Catalyst: Process, Kinetics, and Mechanism Malik Musthofa, Jonathan Lee, and Adam Harvey ...........................................................21 Plasma Assisted Fischer Tropsch Synthesis T. Mukhriza, K. Zhang, A. Phan .........................................................................................26 Ethanol Plant Location Decision in the Brazilian Cerrado G. Granco, A.C.Sant’anna, J.S. Bergtold and M.M.Caldas ...............................................31 Supplying Compressed Natural Gas To The Transport Industry In Ireland: Are the Current Regulatory Licencing Arrangements Facilitating Or Hindering Development? D. Goulding, D. Fitzpatrick, R. O’connor and N.M. Power ..............................................40 Dependence of the Hydrolysis Efficiency on the Lignin Content in Lignocellulosic Material M. Raud, M. Tutt and T. Kikas ..........................................................................................46 Feasibility Analysis for Direct-Fired Absorption Cooling Operation after Decarburization Biogas Produced In the Brewer Li Yingjian, Qiu Qi, Cai Lixing .........................................................................................52 Microbial Community Variations in Thermophilic Anaerobic Digesters Treating Otc Medicated Cow Manure Under Different Operational Conditions G. Turker, O. Ince, Ç. Akyol, E. Ertekin, O. Üstüner and B. Ince ....................................64 Experimental Investigation of Carbon Dioxide Methanation by Sabatier Reaction Duygu Uysal and Bekir Zühtü Uysal .................................................................................70 Biomass Energy: Influence of Acid Treatment on Deashing of Harvested Ash-Rich Algae from Dianchi Lake and Thermal Decomposition Characteristics Chunyan Tian, Zhidan Liu, Yuanhui Zhang, Baoming Li, Haifeng Lu, Na Duan .............74

The Expansion of Sustainable Bioenergy Production (Short Rotation Coppice Willow) Incorporating Waste Water Management and Environmental Protection Christopher R. Johnston, Alistair R. Mccracken, Alexander Gilliland ...............................92 Fermentation Characteristics of Acid Hydrolysates with Different Neutralizing Agents S.Y. Yoon and S.J. Shin .....................................................................................................98 Improving and Optimizing Protein Concentration Yield from Homogenized Yeast at Different Ratios Of Buffer Solution L.E.N. Ekpeni, K.Y. Benyounis, Joseph Stokes, A.G Olabi ..............................................103 The Ignition Characteristics of the Single Coal Slime Particle under Different Oxygen Concentrations and Coflow Temperatures in N2/O2 Atmosphere Kun Zhou, Qizhao Lin, Hongwei Hu ..................................................................................112 Comparision of Hydrochars from Wastewater Sludge and Microalgae for Solid Recovered Fuel by Using Hydrothermal Carbonization Daegi Kim, Jeong Seop Hong, Daeun Bae, Kwanyong Lee, Sun Joo Kim, Ki Young Park .............................................................................................117 A Comparative Technological Review of Hybrid Csp-Biomass CHP Systems in Europe C. M. Iftekhar Hussain, Aidan Duffy, Brian Norton ..........................................................122 The Role of Bioenergy in Germany within Long-Term Energy Scenarios N. Szarka, M. Eichhorn, R. Kittler, A. Bezama and D. Thrän ............................................131 Thermal Susceptibility of Solid Recovered Fuel (SRF) Ljiljana Medic Pejic, Laura Rubio Arrieta, Nieves Fernández Añez, Javier Garcia Torrent ...........................................................................................................137 Revealing Effect of Torrefaction on Biomass Pyrolysis B. Ru, S.R. Wang, G.X. Dai and L. Zhang ..........................................................................142 Potential Assessment Tool of Biomass Electricity Generation for the State Of Paraná Priscila Alves, Susana Viana, Luiz Carlos, Alexandre Aoki ...............................................148 Fast Pyrolysis of Ceylon Tea Waste and Product Biocrude-Oil Characteristics Ramesh Soysa, Yeon Seok Choi, Seock Joon Kim and Sang Kyu Choi .............................154 Investigation and Comparison of Micro-Particulates (Pm) and Gaseous Emissions of Three Biomass Fuels Joanna Relf, Edward Forbes, Rodrigo Olave, Christopher Johnston, Gary Lyons and David Mccall .............................................................................................160 Effect of Steam Exploded Treatment on the Reactivity of Pine Wood Muhammad Azam Saeed, Gordon E. Andrews, Herodotos N. Phylaktou, Bernard M. Gibbs .................................................................................................................166

Ignition Sensitivity of Coal / Waste /Biomass Mixtures Nieves Fernandez-Anez, David J. F. Slatter, Muhammad Azam Saeed, Herodotos N. Phylaktou, Gordon E. Andrews, Javier Garcia-Torrent .................................172

Selective Hydroconversion of Biomass-Derived Levulinic Acid to Gamma-Valerolactone M.R. Mihályi, Gy. Novodárszki and J. Valyon .....................................................................177 Sintering Characteristics of Synthetic Coal Ash Hongwei Hu, Qizhao Lin, Kun Zhou ...................................................................................183 Switchgrass Growth (Alamo Variety) As Affected By Irrigation and N-Fertilization in Central Greece K.D. Giannoulis, D. Bartzialis, E. Skoufogianni, S.N. Sakkou, M. Sakellariou-Makrantonaki and N.G. Danalatos ..............................................................187 The Effect of Fertilizers Containing Urease Inhibitor (Agrotain) on Three Different Industrial Crops’ Productivity D. Bartzialis, E. Skoufogianni, K.D. Giannoulis, S. Kakarantzas, I. Kalkounou, and N.G. Danalatos .......................................................................................191 Energy Storage Aqueous Batteries as Grid Scale Energy Storage Solutions Jorge Omar Gil Posada, Anthony J. R. Rennie, Sofia Perez Villar, Vitor L. Martins, David A. Worsley and Peter J. Hall .........................................................200 Dual Roles of Carbon Coatings in Electrochemical Performances of Hybrid Supercapacitor Using H2TI12O25/Activated Carbon S.H. Lee, E. Baek, H.K. Kim, S.G. Lee, Y.H. Lee and J.R. Yoon .......................................210 Numerical Research on Heat Transfer and Energy Storage in Glass Furnace Regenerator Haitao Zhang, Taohong Ye, Qizhao Lin ..............................................................................218 Nickel-Iron (Ni/Fe) Batteries for Large-Scale Energy Storage A. H. Abdalla, C.I. Oseghale, J.O.G. Posada and P. J. Hall .................................................226 LCA of in-House Produced Small-Sized Vanadium Redox-Flow Battery Michele Dassisti, Piero Mastrorilli, Antonino Rizzuti, Pasqua L’Abbate, Gennaro Cozzolino, Michela Chimienti ...............................................................................232 Hydrogen and Fuel Cell Influence of Carbon Physiological Properties on Its Catalytic Activity Towards Oxygen Reduction: A Comprehensive Review Kiranpal Singh, Fatemeh Razmjooei, Dae-Soo Yang, Min Young Song, and Jong-Sung Yu ..................................................................................238

Application of Fuel Cell Technologies in the Transport Sector. Current Challenges and Developments A. Alaswad, A. Baroutaji, and A.G. Olabi ...........................................................................251 Numerical Simulation of Hydrogen Release and Combustion into an Enclosure Maxim Bragin, Thomas Beard, Weeratunge Malalasekera, Salah Ibrahim ..........................256 Catalytic Conversion of Stearic Acid to Fuel Oil in a Hydrogen Donor Zhentao Huang, Jiangfei Cao, Zhixia Li, Hao Gong, Lingyun Huang, Song Shi, Yue Li ..................................................................................................................261 Catalytic Destruction of Volatile Organic Compound Emissions Using Flow-Through Catalytic Membrane Reactor M. N. Kajama, N. C. Nwogu and E. Gobina ........................................................................267 Hydrogen Generation from Hydrolysis of Nabh4-Nh3bh3 Composite Promoted By Alcl3 Yanmin Xu, Jie Chen, Chaoling Wu, Yungui Chen, Zhenglyu Li .......................................272 A New Synthesis Route for Sustainable Gold Copper Utilization in Direct Formic Acid Fuel Cells C.I. Oseghale, A. H. Abdalla, J. O. G. Posada, and P.J. Hall ...............................................282 Hydrogen Effects on Ignition Delay Of Methyl Butanoate/N-Heptane Mixture Seunghyeon Lee1, Heeseon Kim1, Soonho Song ................................................................289 Modified Silver Catalyst for a Hydrogen Peroxide PEM Fuel Cell J. G. Carton, L. Gonzalez-Macia, A. J. Killard ....................................................................295 Multiple Regression Analysis in the Development of Nife Cells As Energy Storage Solutions for Intermittent Power Sources Such As Wind or Solar Jorge Omar Gil Posada, Abdallah H. Abdalla, Charles I. Oseghale and Peter J. Hall ..........302 Solar Power Plant with Hydrogen Storage Rubal Sambi, A. Alaswad, J. Mooney and A. G. Olabi .......................................................310 Co-Gasification of Coal and Municipal Sewage Sludge M. Agrez, P. Trop, S. Potrc, D. Urbancl and D. Goricanec .................................................317 Optimisation of Pack Chromised Coatings on 304 Stainless Steel for Proton Exchange Membrane Fuel Cell Bipolar Plates Using Box-Behnken Design A.M. Oladoye, J. G. Carton, K. Benyounis, J. Stokes and A.G. Olabi ................................323 Production of Hydrogen Energy from Palm Oil Mill Effluent as Sustainable Biomass Using an Emperical Model N.F. Azman, P. Abdeshahian, N.K.M. Salih, M.G.Dashti, H.Kamyab, N.B.Esfahani, A.A.Hamid and M.S.Kalil ............................................................................329

Analysis of A Ht-Pemfc Range Extender for A Light Duty Full Electric Vehicle (Ld-Fev) F. Millo, S. Caputo, A. Piu ...................................................................................................335 Excess Electrical Energy Storage by Gasification of Sewage Sludge to Syngas B. Arbiter, F. Kokalj and N. Samec .....................................................................................342 Electroactive Biofilm Formation and Electricity Generation in Microbial Fuel Cells S.A.Cheng, D. Sun, W.F. Liu, W.J. Ding, H.B. Huang and F.J. Li ......................................348 Hydrogen Production for Solar Energy Storage. A Proposed Design Investigation Tabbi Wilberforce Awotwe, A. Alaswad, J. Mooney and A. G. Olabi .................................353 Solar Energy and Solar Irradiation Computational Analysis of a Dual Purpose Solar Chimney for Buildings in Nigeria Layeni, A. T, Waheed, M. A, Jeje, O. O, Giwa, S. O. .........................................................363 Peak Temperature Reduction of a Multi-Tubular Solar Receiver by Optimization of Fluid Flow Distribution M. Wei, Y. Fan, L. Luo, G. Flamant ....................................................................................370 Experimental Study on the Different Partial Shading Effects on Different PV Modules Technologies Under Desert Environmental Conditions I. Hadj Mahammed, A. Hadj Arab, S. Berrah, Y Bakelli .....................................................376 Integration of Solar Aided Power Generation Technology into a Steam Power Plant M. A, Sulaiman, M. A Waheed, B A Adewumi and B.O Adetifa ........................................381 Geometric Consideration in Modelling the Effect of Brine Depth on Transient Heat Transfer inside a Basin Type Solar Still A. Madhlopa .........................................................................................................................386 Characterization and Testing Of Mesoporous and Icosahedral Nanomaterials for Enhanced Photovoltaic and Solar Thermal Absorbance Abdul Hai Alami, Meera Almheiri, Afra Alketbi and Jehad Abed ......................................394 Daily Global Solar Radiation Forecasting Over a Desert Area Using Nar Neural Networks: Comparison with Conventional Methods Kacem Gairaa, Abdallah Khellaf, Farouk Chellali, Youcef Messlem, Said Benkaciali .......405 Exemplary Performance of a Pv/T Solar Collector System Contribution in the Energy Balance of a Dwelling H. Jouhara, M. Szulgowska-Zgrzywa, M. A. Sayegh, J. Milko, J. Danielewicz ..................411 Experiments on Polymer Welding Via Concentrated Solar Energy Abdul Hai Alami, Anfal Mahmoud, Jehad Abed .................................................................418 Photovoltaic Generation in the Paraná State, Brazil - An Analysis of the Productive Potential G. M. Tiepolo, O. Canciglieri Junior, J. Urbanetz Junior, Ê. B. Pereira, M. Dassisti ..........423

Wind and Wave Energy Nonlinear Model for On-Seabed Dynamic Stability of Wind Farm Cables H. Zanganeh, N. Srinil and H.D. Nguyen ............................................................................428 On-Bottom Stability of Inter-Array Subsea Cables for Offshore Wind Farms H.D. Nguyen, N. Srinil and H. Zanganeh ............................................................................435

Optimum Methodology for Site Selection of Wave Energy Converters Emmanuel Antai and Iraklis Lazakis ...................................................................................441 Emulation of Expensive Simulation Model for Operation and Maintenance of Offshore Wind Farms Jayanta Majumder, Iraklis Lazakis, Yalcin Dalgic, Iain Dinwoodie, Matthew Revie, David Mcmillan ..............................................................449

Proceedings of SEEP2015, 11-14 August 2015, Paisley

EFFECTS OF MECHANICAL PRETREATMENT ON BIOGAS PRODUCTION FROM WASTE PAPER C. Rodriguez1, A. Alaswad1, T. Prescott1, A.G. Olabi1 1. School of Engineering and Computing, University of the West of Scotland, Paisley; email: [email protected] Abstract In the anaerobic digestion of lignocellulosic materials such as waste paper, the accessibility of microorganisms to the fermentable sugars is restricted by their complex structure. A mechanical pretreatment with a Hollander beater was assessed in order to reduce the biomass particle size and to increase the feedstock‘ specific surface area available to the microorganisms, and therefore improve the biogas yield. The mechanical pretreatment has been applied to a batch of office paper previously shredded and inoculated with sludge from a biogas production plant. A response surface methodology (RSM) was used in order to evaluate the effect of the beating time (BT) and digestion time (DT) on the biogas production; these effects were estimated and discussed using the statistical software Design-Expert v.9. Keywords: Biogas, Waste paper, Pretreatment, Anaerobic digestion cardboard) [14], [15]. Biogas production from waste paper is in most cases studied as codigestion with another substrates to increases the methane yields [16], [17]. Paper materials have a carbon-to-nitrogen (C/N) ratio ranging from 173/1 to greater than 1000/1 [18], these values are very high for anaerobic digestion, while the suggested optimum C/N ratio for anaerobic digestion is in the range of 20/1 to 30/1. Adding as inoculum sewage sludge that has a C/N ratio ranging from 6/1 to 16/1 [19] help to balance the C/N ratios in the reactor. Only two pretreatments have been studied to improve the biodegradability of paper and cardboard: mechanical and biological. The mechanical pretreatment consisted in shred the paper and cardboard fraction of municipal solid waste before anaerobic digestion but no significant effect on biogas yields and on kinetics [20]. Better results was obtained when filter paper (FP), waste office paper (OP), newspaper (NP), and cardboard (CB) were pretreated with a thermophilic cellulose degrading consortium (MC1), after 55 days of anaerobic digestion, the methane yield of pretreated FP, OP, NP, and CB were 277, 287, 192, and 231 ml CH4/gVS respectively, with corresponding increases of 33.2%, 34.1%, 156.0%, and 140.6% [19]. This paper investigates the improvements provided by a Hollander beater pretreatment. This technique is based on the same ‗comminution‘ concept proposed by all other

1 INTRODUCTION Paper and cardboard are a heterogeneous mixture of plant material such as cellulose, hemi-cellulose, and lignin and and filling material such as clay and calcium carbonate. Chemical additives (i.e. rosin, alum, starch) are added to modify quality of the material ans its properties such as brightness, opacity, or glossiness. Some paper such as currency paper is camposed by almost 100% cellulose. Residual contents of chemicals used during processing, such as talc or sodium silicate from may still be found in the paper product and consequently also in waste paper [1]. In Europe the per capita consumption of paper and board was 137 kg in 2012, in United Kingdom the total consumption was 1,0095,000 tonnes, being the sixth country in the world consumption [2]. The biggest source of recovered paper is industry and businesses with the 52% of the total, this covers also the converting losses (cuttings and shavings) and returns of unsold newspapers and magazines. Around 10% comes from offices, and the remaining 38% from households [3]. Although waste paper is mainly derived for recycling in paper mills, some other uses are being investigated such as construction materials [4], [5]; animal bedding [6], composting [7] or as a fuel [8], [9]. Many studies have been carried out about the anaerobic digestion of pulp and paper sludge [10]–[13] and municipal solid waste (MSW) (partially composed by paper and 1

Proceedings of SEEP2015, 11-14 August 2015, Paisley reactors are placed in water-bath to keep the mechanical treatments and increases biogas production. The Hollander beater has never been temperature at 37°C. used as mechanical pretreatment machine on seaweed biomass. Seeing that this proposed pretreatment has already proved its effectiveness when applied to maize silage [21] gaining up to 29% extra biogas volume, in this study it has been applied to seaweed biomass in batch mode. 2

MATERIALS AND METHODS

2.1 Hollander Beater The machine is composed of an oval vessel divided along its major axis by a partition that did not reach the walls, so an elliptic channel is formed (Figure 1). In one of the sides of the channel is placed a bladed drum that spins above a bedplate, churning pulp up over the back fall where it slides down creating momentum to round the curve and continue the loop [22]–[24].

Figure 2. Reactors with collection systems. Reactors are feed with 150 ml of pulp (beated paper) and 200ml of sludge (inoculum), 200ml of water, 1 g of sugar and 2.3 g of Potassium Dihydrogen Phosphate (KDP). Sugar is used to boost the action of microorganisms at the start of the digestion; sludge is collected from a thermophilic plant and the process in the study is under mesophilic conditions so the microorganisms need to acclimatize and sugar is an easy substrate to degrade and start the process. KDP is added as buffer solution to maintain the pH around 8 at the start of anaerobic digestion. The reactors corresponding to the untreated samples are feed with 3g of paper and controls are prepared in the same way except for the paper addition in order to assess their contribution to the biogas production. Flasks are gently shaken during the process in order to favour the degasification of the substrate and the contact between the biomass and the inoculum.

Figure 1. Hollander beater in operation with waste paper and working scheme [25]. 2.2 Feedstock and Inoculum Waste paper was collected from recycle bins at the School of Computing and Engineering at the University of West of Scotland (UWS). This paper was mostly one side printed and was cut by an office shredder in 0.6 x 29.7 cm pieces. Sludge was collected from the Scottish and Southern Energy (SSE) Barkip Biogas Plant, Ayrshire, Scotland. The moisture content (MC) of pulp is calculated to provide a biogas production in term of volume per gram of total solids (TS = 1 - MC (%)) and is obtained by drying 100 ml of paper pulp at 105°C until constant weight.

2.4 Design of experiments The experiment is planned according to a response surface methodology (RSM) for two factors, beating time (BT) and digestion time (DT) with three levels; the response is the biogas production pre g of dry paper (Table 1). The statistical study is performed using the software Design Expert v.9. A second order polynomial is used, Y  b0  bi xi   bii xii2   bij xi x j (1)

2.3 Bioreactors The bioreactor consists of flasks of 500 ml connected through a system of valves and plastic pipes to airtight plastic bags for biogas collection (Figure 2). To clear up any trace of oxygen from the system and preserve the anaerobic conditions, nitrogen is flushed for triplicate for 5 minutes into the reactors. The

where the values of the model coefficients b0, bi, bii and bij are estimated using regression analysis. 2

Proceedings of SEEP2015, 11-14 August 2015, Paisley The adequacy of the models is tested through the compare to the untreated sample, it increase analysis of variance (ANOVA). The statistical compare to 1h pre-treatment. A short beating significance of the models and of each term is time do not disrupt paper structure enough to examined using the sequential F-test and lacklead to an improvement in biogas production. of-fit test. If the Prob. > F of the model and of Because the waste paper has passes through each term in the model does not exceed the level refining during its preparation, its structure is of significance (in this case α = 0.05) then the already disrupted to a large degree. model may be considered adequate within the In the six first days of digestion, there is less confidence interval of (1 - α). than 1.3% difference in biogas production between the three beating times. By day 20, the 3 RESULTS AND DISCUSSION biogas production is 17.20% less for 1h BT and The experiment parameters, beating time (BT) 16.06% for 2h BT compare to non-beated paper. and digestion time (DT) are checked in three At the end of digestion, the biogas production levels. Beating time varies between 0-2 h and for 1h BT is reduced by 21.89% and for 2h BT digestion time between 6-34 d. The response is by 15.97%. The reduction in biogas production the biogas production in ml per g of dry paper may be due to the removal of ink and paper (DP). The values of the biogas volume obtained additives during the beating pre-treatment. These were converted into standard conditions components could be metabolized and produces (101.3kPa, 273.15K). As the biogas produced is biogas. The removal of ink from the paper is saturated with water vapour, the water content visible to the naked eye; a layer of microbubbles was removed from the results as well. accumulates near the rotor with a distinctive Parameters and results are presented in Table 1. black colour. Rosin is an additive present both in The sludge contribution is obtained from the the office paper and in the ink which does not control samples and its value is given in ml of dissolve in water, and may therefore become biogas per g of sludge. All the values are means detached from the fibres in the paper during the of the triplicates. The sugar contribution is Hollander beater treatment forming an quantified theoretically by the following immiscible layer on the surface of the equation: suspension. Rosin is a biodegradable material C12 H 22O11  H 2 O  6CO2  6CH 4 (2) [26], [27], however only aerobic degradation data is available [28]. If it assumed that rosin is Assuming that all sugar is consumed, 1g of anaerobically degradable as well, its removal sucrose will produce 785.6 ml of biogas. during beating will lead to a decrease in biogas production. Beating pre-treatment seems start to Table 1. Experiment parameters and results be effective after 2h being that biogas production Digestion time Beating time Biogas volume for 2h treatment is higher than for 1h. Longer (d) (h) (ml/gDP) beating times disrupt the feedstock structure 6 0 306.54 making it more available for microorganisms. If 6 1 302.67 the effect of structure disruption is higher than 6 2 303.71 the loss of additives and ink, the beating pre20 0 519.22 treatment of waste paper will be feasible. 20 1 429.92 For the optimization through the RSM, a 20 2 435.85 resulted p-value of 7.68805·10-05 indicates model terms are significant. The model terms of 34 0 538.01 R2 = 0.9004, adjusted-R2 = 0.8672, predicted-R2 34 1 455.40 = 0.7297, all these values are very close to 1 and 34 2 489.93 so indicate the adopted model is adequate. The 6 0.56 Control adequate precision, which measures the signal to 20 3.67 (ml/g sludge) noise ratio is 14.9547. A ratio greater than 4 34 4.35 indicates an adequate signal. The analysis of variance indicates that the digestion time (DT), The biogas production decrease when the office the beating time (BT) and the second order paper is treated in the Hollander beater for 1h effect of digestion time (DT) are the most compared with the untreated. Although the 2h significant factors affecting the biogas yield. The pre-treatment reduces the biogas production 3

Proceedings of SEEP2015, 11-14 August 2015, Paisley final mathematical model associated to the and DT. Digestion time affect the biogas yield in response in terms of actual factors determined by an exponential way and beating time affects it in the software is shown below. a linear way. Increasing the beating time has a negative effect on the biogas yield while an increase in the digestion time has a positive Biogas yield  220.46  21.01DT (3) effect. 2  29.88BT  0.34 DT The response surface obtained from the model illustrated in Fig. 3 shows that the optimal condition was located outside the factorial design boundary, at higher digestion times.

Figure 5. Perturbation plot showing the effect of BT and DT on biogas volume. 4 CONCLUSIONS The experimental work shows the biogas yields obtained from the digestion of office paper inoculated with sludge from a biogas production plant. Pre-treated office paper with a Hollander beater for 2h has a negative effect on the biogas production through anaerobic digestion. Loss of ink and paper additives during the pre-treatment can occur. Longer beating times have to be investigated in order to evaluate the positive effects on biogas production.

Figure 3. Response surface plot showing the effect of BT and DT on biogas yield. The predicted vs. actuals plot (Fig. 4) shows that these values were distribute near to a straight line and a satisfactory correlation between them is observed. This demonstrates that the model can be effectively applied for mechanical pretreatment with a Hollander beater for office paper.

REFERENCES

Figure 4. Scatter diagram of biogas yields. The perturbation plot in Fig. 5 shows how the biogas yield is affected by the input variables BT 4

[1]

I. R. Gran, Characterisation of MSW for Combustion Systems. 2001.

[2]

G. Magnaghi, ―RECOVERED PAPER MARKET IN 2012,‖ Brussels, 2014.

[3]

The Bureau of International Recycling, ―Ten questions on paper recovery and recycling.‖ [Online]. Available: http://www.bir.org/industry/paper/tenquestions-on-paper-recovery-andrecycling/. [Accessed: 13-Feb-2015].

[4]

O. P. Folorunso and B. U. Anyata, ―Potential uses of waste paper/sludge as a ceiling material,‖ Res. J. Appl. Sci., vol. 2, no. 5, pp. 584–586, 2007.

[5]

[6]

[7]

[8]

Proceedings of SEEP2015, 11-14 August 2015, Paisley manufacturing operations,‖ Georgia M. Sutcu, J. J. del Coz Díaz, F. P. Álvarez Rabanal, O. Gencel, and S. Akkurt, Institute of Technology, 2009. ―Thermal performance optimization of hollow clay bricks made up of paper [14] M. Kayhanian and D. Rich, ―Pilot-scale waste,‖ Energy Build., vol. 75, pp. 96– high solids thermophilic anaerobic 108, Jun. 2014. digestion of municipal solid waste with an emphasis on nutrient requirements,‖ Biomass and Bioenergy, vol. 8, no. 6, pp. P. L. Ward, J. E. Wohlt, P. K. Zajac, and K. R. Cooper, ―Chemical and physical 433–444, Jan. 1995. properties of processed newspaper compared to wheat straw and wood [15] H. M. Lo, H. Y. Chiu, S. W. Lo, and F. C. shavings as animal bedding.,‖ J. Dairy Lo, ―Effects of different SRT on Sci., vol. 83, no. 2, pp. 359–67, Feb. 2000. anaerobic digestion of MSW dosed with various MSWI ashes.,‖ Bioresour. J. V. L. Alvarez, M. A. Larrucea, P. A. Technol., vol. 125, pp. 233–8, Dec. 2012. Bermúdez, and B. L. Chicote, ―Biodegradation of paper waste under [16] M. O. L. Yusuf and N. L. Ify, ―The effect controlled composting conditions.,‖ of waste paper on the kinetics of biogas Waste Manag., vol. 29, no. 5, pp. 1514–9, yield from the co-digestion of cow dung May 2009. and water hyacinth,‖ Biomass and Bioenergy, vol. 35, no. 3, pp. 1345–1351, V. Brummer, T. Jurena, V. Hlavacek, J. Mar. 2011. Omelkova, L. Bebar, P. Gabriel, and P. Stehlik, ―Enzymatic hydrolysis of [17] H.-W. Yen and D. E. Brune, ―Anaerobic pretreated waste paper--source of raw co-digestion of algal sludge and waste material for production of liquid paper to produce methane,‖ Bioresour. biofuels.,‖ Bioresour. Technol., vol. 152, Technol., vol. 98, no. 1, pp. 130–4, Jan. pp. 543–7, Jan. 2014. 2007.

[9]

Y. Li and H. Liu, ―High-pressure binderless compaction of waste paper to form useful fuel,‖ Fuel Process. Technol., vol. 67, no. 1, pp. 11–21, Jun. 2000.

[10]

C. Priadi, D. Wulandari, I. Rahmatika, and S. S. Moersidik, ―Biogas Production in the Anaerobic Digestion of Paper Sludge,‖ APCBEE Procedia, vol. 9, pp. 65–69, 2014.

[11]

[12]

[13]

[18] P. Stroot, ―Anaerobic codigestion of municipal solid waste and biosolids under various mixing conditions—I. digester performance,‖ Water Res., vol. 35, no. 7, pp. 1804–1816, May 2001. [19] X. Yuan, Y. Cao, J. Li, B. Wen, W. Zhu, X. Wang, and Z. Cui, ―Effect of pretreatment by a microbial consortium on methane production of waste paper and cardboard.,‖ Bioresour. Technol., vol. 118, pp. 281–8, Aug. 2012.

Y. Lin, D. Wang, Q. Li, and L. Huang, ―Kinetic study of mesophilic anaerobic digestion of pulp & paper sludge,‖ Biomass and Bioenergy, vol. 35, no. 12, pp. 4862–4867, Dec. 2011.

[20] S. Pommier, A. M. Llamas, and X. Lefebvre, ―Analysis of the outcome of shredding pretreatment on the anaerobic biodegradability of paper and cardboard materials.,‖ Bioresour. Technol., vol. 101, no. 2, pp. 463–8, Jan. 2010.

T. Meyer and E. A. Edwards, ―Anaerobic digestion of pulp and paper mill wastewater and sludge.,‖ Water Res., vol. 65, pp. 321–49, Nov. 2014.

[21] A. Mohamed, F. Alfarjani, and K. Benyounis, ―Application of Mechanical Pre-treatment to Produce Methane from Maize,‖ ECOS, 2011.

N. Szeinbaum, ―Assessment of anaerobic treatment of select waste streams in paper 5

Proceedings of SEEP2015, 11-14 August 2015, Paisley [22]

M. Cerda, Arqueologia industrial, First. Valencia: Universitat de Valencia, 2008.

[23]

G. A. Osorio, ―Análisis FMECA a sistema de posicionamiento de platinas de refinadora de pulpa de papel diseñada para papelera Gubelin Ltda.,‖ Universidad Austral de Chile, 2010.

[24]

S. Tedesco, ―Mechanical pretreatment assessment of marine biomass,‖ Dublin City University, 2013.

[25]

―Peace Paper Project.‖ [Online]. Available: http://www.peacepaperproject.org/dardhu nter.html. [Accessed: 17-Feb-2015].

[26]

Z. Tiantao, Z. Youcai, Z. Lijie, C. Haoquan, S. Feng, and Z. Haiyan, ―Slowrelease of methanogenic inhibitors derived from encapsulated calcium carbide using paraffin wax and/or rosin: matrix optimization and diffusion characteristics,‖ Waste Manag. Res., vol. 29, no. 11, pp. 1197–1204, 2011.

[27]

H. Pulido González, E. Hernandez, M. Rabelero Velasco, R. Sanjuan Raygoza, and C. Jasso Gastinel, ―Mechanothermal performance evaluation of a biodegradable resin as coupling agent for hydrophobic polymer/cellulosic composites,‖ Maderas. Cienc. y Tecnol., vol. 16, no. ahead, pp. 0–0, 2014.

[28]

Pine Chemical Association, ―Test plan for rosins and rosins salts.‖ US Environmental Protection Agency, Virginia, US, 2001.

6


FARMER’S WILLINGNESS TO GROW SWEET SORGHUM AS AN ANNUAL BIOENERGY CROP UNDER CONTRACT: ASSESSING WILLINGNESS TO PAY FOR CONTRACT ATTRIBUTES 1. 2. 3. 4.

J.S. Bergtold1, A. Shanoyan2, J.E. Fewell3 and J.R. Williams4 Associate Professor, Department of Agricultural Economics, Kansas State University, Manhattan KS, USA; email: [email protected] (Corresponding author) Assistant Professor, Department of Agricultural Economics, Kansas State University, Manhattan KS, USA; email: [email protected] Farm Business Management Director, Lake Region State College, Devils Lake, ND, USA; email: [email protected] Professor, Department of Agricultural Economics, Kansas State University, Manhattan KS; email: [email protected]

Abstract Dedicated annual sorghum crops, such as sweet sorghum or energy sorghum, may provide a way for farmers to help supply cellulosic feedstocks for biofuel production to meet government mandates. Kansas farmers are poised to be major producers of sweet sorghum for biofuels due to favorable agroecological conditions. The purpose of this paper is to assess Kansas farmers’ willingness to grow sweet sorghum under contract as a feedstock for biofuel production. In addition, the paper will examine farmers’ willingness-to-pay for contract attributes and the impact of socio-economic factors on their willingness-to-pay. A stated choice survey was administered to Kansas farmers to assess their willingness to grow sweet sorghum for biofuels under various contracting scenarios. Results show that farmers may be willing to grow biomass for bioenergy under contract, but this will depend on net returns, contract length, insurance availability, potential for biorefinery harvest, and government incentives. Keywords: Biofuels, Contract, Mixed Logit, Sorghum 1. INTRODUCTION The Energy Independence and Security Act of 2007 states, in part, that biofuel production must increase to 36 billion gallons by the year 2022, of which 21 billion gallons must come from “advanced” or second-generation biofuel feedstocks, which includes dedicated annual bioenergy crops [1]. To ensure the viability of biofuel production from cellulosic feedstocks, it is important to know if farm managers are willing to produce cellulosic biofuel feedstocks. That is, how willing are farmers to produce biofuel crops (e.g. sweet sorghum) given that the markets for such crops are underdeveloped or nonexistent. Rajagopal and Zilberman [2] indicate that the factors that lead to farmers’ adoption of biofuel technologies are still not well understood. Among such factors are contract structure and specifications. Farmers are more likely to supply cellulosic biofuel feedstocks if a contract is offered by processors to produce or supply the feedstock and if the

payoff from the enterprise is greater than any other possible land use [3]. Contractual arrangements will be affected by many factors, such as contract pricing, timeframe, acreage commitments, risk, timing of harvest, yield variability, feedstock quality, harvest responsibilities (e.g. custom harvesting), nutrient replacement, location of biorefineries, available cropping choices, technology, and conservation considerations [4-9]. Dedicated annual bioenergy crops provide a potentially viable biofuel feedstock enterprise option for farmers. They can serve as a flexible alternative cash crop for farmers that can be grown in traditional crop rotations. For example, sweet, energy or forage sorghum varieties may serve as annual bioenergy crops. There are several advantages to these types of sorghum crops, including production of high amounts of biomass, drought tolerance and the ability to incorporate them into existing crop rotations [10]. This study focuses on the use of 7


population of farms over 260 acres in size and $50,000 in gross farm sales from the USDANASS farmer list. Farmers already participating in other USDA-NASS enumerated surveys (e.g., ARMS) were removed from the sample and replaced with another randomly drawn name. Prior to the survey entering the field, the stated choice component was field tested with two focus groups at an annual extension conference hosted by the Department of Agricultural Economics at Kansas State University in the summer of 2010. Participants who were identified as farmers were invited to attend the focus group, of which 12 participated. In addition, the entire survey was tested using faceto-face interviews with farmers in the targeted study areas. Potential participants received a fourpage flier via mail asking for their participation in the survey and providing information about cellulosic biofuel feedstock production on-farm one week prior to being contacted by USDANASS enumerators. Enumerators then scheduled one-hour interviews with the farmers to complete the survey and stated choice experiments. Interviews, on average, took 57 minutes to complete. Upon completion of the survey and receipt at the USDA-NASS office, farmers were compensated for their time with a $15 gift card. Of the 485 farmers contacted, 290 completed the survey and 38 were out-ofbusiness, did not farm, or could not be located. Thus, the survey response rate was (290/(48538)) = 0.65 or 65 percent. Of the 290 respondents who completed the stated choice experiment for sweet sorghum, 13 surveys were incomplete due to lack of responses on the experiment or refusal to answer demographic questions, leaving 277 usable surveys for this study. After answering a number of questions about their farming operation, respondents were asked about their willingness to produce cellulosic biofuel feedstocks under contract. Respondents were then asked about biofuel feedstock production preferences and perceptions; conservation on-farm and perceptions; risk management practices and perceptions; crop marketing practices; and demographics. A set of descriptive statistics for a select number of socio-economic and farmer demographic variables used in the study are provided in Table 1.

sweet sorghum specifically, though forage and energy varieties of sorghum would be viable substitutes. Management of sweet sorghum is similar to grain and forage sorghum under dryland conditions. Propheter, et al. [11] found favorable yields for different sorghum varieties in the less than ideal growing conditions in Kansas. They also found that these annual crops produced more usable biomass than perennial options during the study period, which increases appeal of crops such as sweet sorghum for biofuels. The purpose of this study is to examine farmers’ willingness to produce sweet sorghum under alternative contractual, pricing, and harvesting arrangements in Kansas. Assessment of farmers’ willingness to adopt a sweet sorghum enterprise under different contractual arrangements is implemented using an enumerated field survey with stated choice techniques. The survey examines farmers’ willingness-to-pay for different contract attributes and how alternative socio-economic and farm factors may impact their contractual preferences. A stated choice approach following Louviere et al. [12] is used to assess farmers’ willingness to adopt. Survey results are analyzed using a random parameters conditional logistic regression model [13-14]. The estimation of marginal effects of socio-economic and farm characteristics on the willingness-to-pay for specific contract attributes is a unique contribution of this paper. 2

SURVEY METHODS AND DATA A stated choice survey was administered from November 2010 to February 2011 in northeastern, south central and western Kansas by Kansas State University and the USDA, National Agricultural Statistics Service (NASS). The purpose of the survey was to assess farmers’ willingness to produce cellulosic biomass in the form of corn stover, sweet sorghum, and switchgrass for bioenergy production under different contractual arrangements. A total of 485 farmers were contacted to participate in the survey. The areas of Kansas were selected based on the number of farms growing corn and/or sorghum; the mix of irrigated and dryland production; and mix of crops and farming practices adopted. For each area of the state examined, the sample of farms was randomly drawn from the 8


Table 1: Descriptive Statistics for Explanatory Variables (N=277) Variable Name

Definition

East

Equal to 1 if farmer is located in eastern Kansas and 0 otherwise. Equal to 1 if farmer is located in western Kansas and 0 otherwise. Size of farm in acres Percent of farm land rented. Percent of gross farm sales from crop production. Percent of household income earned off the farm. Number of years farming Equal to 1 if farmer uses a crop rotation and 0 otherwise. Equal to 1 if the farmer considers themselves adverse to risk and 0 otherwise. Equal to 1 if the farmer has a college degree and 0 otherwise. Equal to 1 if the farmer has grown sorghum before and 0 otherwise.

West Farm Size Percent Rent Crop Sales Percent Off-Farm Income Experience Crop Rotation Risk Adverse College Grow Sorghum

Mean

Contract Features

Contract B

45% Higher/year

0% Higher/year

5 Years

2 Years

Biorefinery Harvest

Yes

No

Insurance Availability

No

No

Gov. Incentive Payment

None

25%

Your Ranking (1-3)

0.48

2150 57.7 70.8 34.4 32.0 0.38 0.89

1665 34.7 26.5 36.7 13.1 0.49 0.32

0.30

0.46

0.47

0.50

2.1 Stated Choice Experiment A stated choice experiment was designed to assess farmers’ willingness to enter into a contract with a bio-refinery or other biomass processor for producing sweet sorghum following Louviere et al. [12] and Roe et al. [16]. Farmers where presented with information about sweet sorghum production and contract attributes before answering a set of stated choice questions. Survey participants where then asked to consider 5 independent choice scenarios, where they were asked to select between two biomass contracts or an “opt out” option (Figure 1). Each contract option was unlabeled and had five attributes: (1) net returns above corn/sorghum production; (2) contract length;

Contract A

Net Return Above Sorghum/Corn Production (Base: $50/ac) Contract Length

0.34

survey. Farms with the resources and flexibility to produce these types of crops were desired for the survey sample. Larger farms produce most of the crop output in Kansas.

Farmer demographics taken from the 2012 U.S. Census of Agriculture [15] were used to assess the generalizability of the survey sample to the general farmer population with farms above 260 acres in size and $50,000 in gross farm sales. The average age of surveyed farmers was 55.1 years, which was slightly lower than the Census average of 58.6 years. Average farm size and the amount of land rented from the survey sample (2172 acres and 1271 acres, respectively) were larger than the Census averages for Kansas (1553 acres and 1017 acres respectively). While these averages may not be significantly different, the differences may be due to the targeted areas for the survey within the state. The survey asked respondents to choose a category in which their value of agricultural product sales occurred, the Census of Agriculture amount of $448,317 was slightly larger than the most often selected category of $200,000 to $399,999 by respondents on the Sweet Sorghum Scenario:

0.37

Standard Deviation 0.49

2048

2049

Figure 1: Example Choice Scenarios/Questions for Stated Choice Experiment 9

Option C

Do Not Adopt

2050


Table 2: Contract Attributes and Levels for Stated Choice Experiments for Sweet Sorghum Contract Attribute Net Returns (for all features of the contract except cost-share and government payments)

Contract Length Biorefinery Harvest

Insurance Availability Government Incentive Payment

Description Represents the expected percentage gain under the contract above net returns associated with corn/sorghum production on a farmer’s operation. As a reference point, on average, returns from corn/sorghum production are expected to be $50 per acre in Kansas. Represents the time commitment in consecutive years of the contractual agreement. “Yes” indicates the bio-refinery will harvest the biomass at their expense, and “No” means the farmer is responsible for harvest (including cutting, raking, baling and transportation to the bio-refinery). Harvest charges are included in the percentage net return. That is, the charges are considered paid regardless of who harvests the biomass.

Levels 0%, 15%, 30% and 45%

“Yes” indicates crop insurance is available, and “No” otherwise. This incentive payment is provided at two levels for production of cellulosic biofuel feedstocks delivered to a bio-refinery. The incentive levels are either none (0) or 25 percent of the price per dry ton of biomass delivered to the refinery. The incentive received is in addition to the net returns above production.

Yes or No

(3) biorefinery harvest; (4) insurance availability; and (5) government incentive payments. Descriptions and levels for each contract attribute are provided in Table 2. The net returns attribute captures both expected net returns and the opportunity cost of not planting another cash crop. Farmers were asked to assume a base value of $50 for net returns from dryland corn/sorghum production for all scenarios presented. This base level represents the average net returns from Kansas Farm Management Association Farms (KFMA) [17] in the study region(s) for 2008/9. Thus, net returns are not fixed to a specific dollar amount, but can be compared relative to other crops using the attribute level for net returns above expected corn/sorghum production. For the purpose of data analysis, this attribute was recoded into a dollar amount by multiplying the percentage increase by the $50 base. The net returns attribute provides flexibility in allowing the capture of reservation prices and potential opportunity costs. The second attribute was contract length. The increasing length of the contract represents the need of the refinery or biomass processor to mitigate risk by ensuring a long-term supply of biomass. The third attribute was a binary attribute for biorefinery harvest. This attribute provided the option to the farmer of having the biorefinery or intermediate processor harvest the biomass from the field and transport it to the refinery/processor. It was assumed that the cost 10

2,5 and 8 years Yes or No

0% and 25%

of this practice was included in net returns. The fourth attribute was insurance availability, a binary attribute indicating if crop insurance was available to purchase under the contract. All binary attributes were effects coded. The final attribute was a government incentive payment that is similar to that provided under the Biomass Crop Assistance Program (BCAP) [18]. The attribute had two levels, which were set to represent the match the government pays as a percentage of the price paid by the refinery for every ton of biomass produced and delivered. Furthermore, the incentive was not included in the net returns attribute described above. Given the use of substantial subsidies to promote cellulosic biofuel production, this attribute plays an important role for policy analysis. Following Louviere et al. [12], a (24 x 3 2 x 4) fractional factorial experimental design was used to design the choice sets for the sweet sorghum experiment. PROC OPTEX in SAS [19] was used to develop the fractional factorial design from the complete factorial to obtain 90 random choice scenarios, which were then blocked into 18 choice sets of 5 choice scenarios each. The D-optimality criterion was used to obtain an optimal design using a modified Federov search algorithm (see [20]). Optimal blocking was determined following the method outlined in Cook and Nachtsheim [21]. An optimal (treatment) D-Efficiency of 87.1% was obtained. The number of choice sets was


selected in order to identify all main effects and potential interaction effects between contract attributes and levels. Each respondent was faced with 5 choice scenarios for each feedstock alternative, with 18 versions of the survey. Of the 290 surveys completed, 12 to 20 of each version were completed by farm managers. 3 MODEL Following Roe et al. [16], we assume that producers want to maximize expected discounted utility when choosing to enter into a contract to produce sweet sorghum versus produce corn or grain sorghum over time. Let producer’s j’s expected discounted utility for contract option i be given by: ,

= (∆ ( , ),

,

)+

,

,

(1)

where is ∆ is the net returns above dryland corn/grain sorghum production over time, which includes the costs associated with Bi, indicating if a biomass harvest option is part of contract i, and Si, indicating if crop insurance is available. In addition, expected discounted utility is a function of Ci, or the length of the contract in years; and Gi, or the level of government incentive payment. The error term, , represents the nonsystematic part of expected utility that goes unobserved by the modeler and is distributed type I extreme value [12]. It should be emphasized, that the inclusion of ∆ captures the return above the next best alternative for sweet sorghum. This is unique to this experimental set-up and important in that there will be competing uses for the land. Thus, the experimental set-up takes into account the potential next best alternatives for the land. For the purposes of this study we are primarily interested in examining the direct impact of the contract attributes on producers’ willingness to adopt or enter into a contract. Following Roe et al [16], we focus on the reduced-form representation of expected utility. The econometric model is based upon conditional logistic regression model [13]. That is, for producer j and contract i: ,

= + ∆ + + + , , for j=A,B or C.

+

+ (2)

Contract choices A and B represent the randomly assigned unlabeled contract choices presented in each choice scenario, while option C is the “opt out” or “do not adopt” option. Given option C 11

= 0 for k = 0,1,2,3,4,5 has no attribute levels for this option. It may be the case that the parameters of the model may be distributed across individuals due to preference heterogeneity for the different contract attributes. To capture this preference heterogeneity, let the individual specific parameter distribution be captured by , = + ′ + , for k = 0,1,2,3,4,5, where is assumed to be mean zero and distributed , following a one-sided triangle distribution (with = ) for the expected change in net returns (∆ ) parameter and normal distribution for all other attribute parameters for k =0,2,3,4,5. The one-sided triangular distribution restricts the coefficient on expected net returns to be positive and has been found to be useful in practical represents the standard applications [22]; deviation for the kth attribute parameter + ′ is the conditional mean distribution; th of for the k attribute parameter distribution; and is a vector of famer specific characteristics [13]. All attribute parameters were assumed to be independent of each other. Each parameter distribution for the model given by equation (2) is assumed to vary across the population and is conditional on a set of farmer specific characteristics. These characteristics are given by the set of explanatory variables identified in Table 2. This set of farmer specific variables provides a mechanism to incorporate geographical, socioeconomic and farm characteristic variables that may impact farmers’ willingness to produce a cellulosic feedstock as identified in previous studies [23-28]. While each respondent ranked their choices, we only examine the first choice or one with highest likelihood of being chosen. Thus, equation (2) is modeled using a random parameter conditional logistic regression (or mixed logit) model following Greene [13] and Train [29]. NLOGIT 4.0 (Greene, 2007) is used to estimate the model, using simulated maximum likelihood with 1000 Halton draws using the BFGS Quasi-Newton algorithm. Given the large number of parameters in the model (78), full model results are not presented, but are available from the authors upon request. Instead, willingness-to-pay measures for each of the contract attributes is estimated with associated marginal effects.


A common use of the econometric model results is to estimate what a producer would be willing to pay (WTP) for a given contract attribute. For example, what would a producer be willing to pay to reduce their contract length by 1 year? Following Greene [13], a producers willingness to pay for a (one unit change in a) contract attribute would be equal to for k = 2,3,4,5. The WTP estimates provide an estimate of farmers’ preferences for different attributes. Given that the attribute parameters in equation (2) are functions of farmer specific variables, mean WTP across the population will be equal for k = 2,3,4,5 at the mean of the

to

conditional attribute parameter distributions. Thus, mean WTP estimates will vary with changes in farmer specific characteristics. Changes in WTP given changes in farmer specific variables (i.e. marginal effects for WTP) may be of interest, as well. The marginal effect of a binary variable is estimated using a discrete difference. For continuous explanatory variables, the marginal effect can be estimated as: = (

/

)

(

/

)

. All mean WTP and

associate marginal effects are estimated as the average across the individual estimates using the means of conditional attribute parameter distributions. Asymptotic standard error for mean WTP and marginal effects are estimated using the method of Krinsky and Robb [13]. 4

RESULTS Model fit statistics, as well as mean WTP estimates and marginal effect estimates are provided in Table 3. Overall, the model provided good fit with the data with a McFadden Pseudo R2 of 0.50. Not shown in Table 3, the choice data indicates that farmers would have adopted a contract to produce sweet sorghum 41 percent of the time across the 1385 choice situations they were asked to assess. Overall survey respondents were more willing to produce a dedicated annual bioenergy crop than harvest crop residues or produce a dedicated perennial bioenergy crop. Expected returns under the contract above corn/sorghum production had a highly significant and positive effect on a farmer’s willingness to produce sweet sorghum under contract. In addition, the level of returns required decreased as farmer experience

12

increased. These are presented here as they are not readily apparent in the WTP estimates presented next. The WTP estimates provide information about the preferences for the other contract attributes. For contract length, on average, a farmer would be willing to reduce the level of expected net returns earned per acre under the contract by $3.22 for each year they can reduce the length of the contract. This estimate increases in absolute value by $0.13 for each year of farm experience. As expected, farmers prefer a shorter contract, providing more flexibility for their farming operation. Farmers are willing to reduce the amount of expected net returns by $7.50 per acre if the biorefinery or intermediate processor provides a custom harvest option. Having a biorefinery harvest option increases the likelihood of producing sweet sorghum, providing more flexibility for timing of farming operations (e.g. interfering with the harvesting of other crops). Insurance is an important component, especially since markets for cellulosic biomass are very limited in scope or do not exist yet in some locations. If insurance is available in the market, then farmers would be willing to accept a contract with $6.58 less in expected net returns per acre for having the risk protection provided by insurance. The WTP estimate for insurance was the most responsive to the socio-economic and farm characteristics of the farmer respondents. As the percent of land rented increases by 1 percent, farmers would require $0.12 more in expected net returns per acre under the contract when insurance is available. The cost of the land rental would likely require the farmer to seek higher returns from the land. For each additional year of experience the farmer has, they would be willing to give up an extra $0.28 per acre in expected net returns to have insurance. In addition, for each percent increase in household income off the farm, a farmer would be willing to give up an additional $0.13 per acre in expected net returns if insurance was available. It could be the case that additional household income from off the farm provides the farmer with more income security and flexibility, while additional experience provides them with knowledge about dealing with potential difficulties of undertaking a new enterprise [27].


Table 3: Willingness-to-Pay ($/acre), Associated Marginal Effect ($/acre) Estimation Results, and Model Fit Statistics Contract Biorefinery Insurance Government Length Harvest Availability Incentive Payment Mean Willingness-to-Pay Across Respondentsa (Standard Error) Mean -$3.22** $7.50** $6.58* $0.30** ($1.29) ($2.46) ($3.53) ($0.11)

East West Farm Size Percent Rent Crop Sales Percent Off-Farm Income Experience Crop Rotation Risk Adverse College Grow Sorghum

-$1.32 ($2.72) -$2.59 ($3.09) -$2.34E-4 ($3.50E-4) $0.02 ($0.02) -$0.02 ($0.02) -$0.01 ($0.02) -$0.13* ($0.04) -$0.08 ($2.81) -$0.11 ($2.45) $0.65 ($2.36) $1.41 ($2.48)

Marginal Effect on Willingness-to-Paya (Standard Error) $1.03 $1.91 ($5.14) ($7.24) -$4.33 $4.64 ($5.96) ($8.21) $7.26E-4 $0.001 ($7.99E-4) ($9.70E-4) $2.38E-4 -$0.12* ($0.04) ($0.05) $0.07 $0.07 ($0.05) ($0.06) $0.02 $0.13* ($0.04) ($0.06) $0.13 $0.28* ($0.13) ($0.13) -$1.43 -$2.91 ($5.26) ($7.80) $7.85 $3.46 ($5.06) ($6.56) -$1.18 $0.98 ($4.61) ($6.89) -$2.02 $2.37 ($4.93) ($7.23)

$0.12 ($0.24) $0.03 ($0.26) $3.48E-5 ($3.51E-5) -$5.78E-4 ($0.002) $5.73E-4 ($0.002) -$2.93E-4 ($0.002) $0.01* ($0.005) $0.07 ($0.23) -$0.03 ($0.23) -$0.10 ($0.19) -$0.11 ($0.21)

Model Fit Statistics Log Likelihood McFadden Pseudo R2 AIC Number of Observations

-765.7728 0.50 1.217 1385

** and * denote statistical significance at the 5 and 10 percent levels, respectively. a All estimates of willingness-to-pay and associated marginal effects were estimated using the estimated mean of the conditional parameter distribution for each random parameter in the logistic regression model estimated. Asymptotic standard errors were estimated using the method of Krinsky and Robb (Greene, 2012). Both mean and standard error estimates were estimated cutting off the tails at the 5% level of the empirical distributions of WTP to avoid extreme estimates (Greene, 2012).

Government incentive programs have provided a strong incentive to adopt new enterprises and practices by farmers in the past. As expected, for each 1 percent increase in the level of incentive payment per ton of biomass delivered to the refiner, a farmer is willing to give up $0.30 in expected net returns per acre. As with the other attributes, for each additional year of experience a farmer gains, the WTP increases by $0.01 per acre. Thus, incentive payments may provide a mechanism for promoting adoption if the government wants to 13

pursue a policy to increase cellulosic biofuel production. 5. CONCLUSION Bioenergy crops play an important role in crop production on the Great Plains and in Kansas as farmers attempt to help meet the demands for biofuel production from both grain and cellulosic feedstocks. Sweet sorghum is an annual crop that is well-suited to planting in Kansas, but much uncertainty exists as to its viability and the willingness of farmers to grow


such crops for biofuels. A stated choice survey was developed to assess farmers’ willingness to grow crops for biofuels. Results from the estimation showed that farmers are more willing to grow crops if net returns are relatively high and contract length is short. In addition, results show that farmers prefer an insurance option, similar to their existing crop insurance, and that a government incentives can help to promote adoption. Further research includes determining biomass prices that farmers and biorefineries can agree upon under differing contracts. Contract design will be one of the most important, yet most significant aspects of establishing market prices for biomass. Farmers face much risk and uncertainty with growing new crops, especially without well established markets. Therefore, designing insurance contracts for biomass producers that are similar to existing crop insurance is necessary before widespread bioenergy crop adoption will occur on a large scale. USDA’s Risk Management Agency (RMA) must work with farmers and biorefineries to arrive at marketable insurance products. ACKNOWLEDGEMENTS The funding for the primary portion of this project came from the South Central Sun Grant Initiative and Department of Transportation (Award No. DTOS59-07-G-00053), with additional funds from the National Science Foundation, EPSCoR Division, Research Infrastructure Improvement (Award No. 0903806), and a National Science Foundation Grant From Crops to Commuting: Integrating the Social, Technological, and Agricultural Aspects of Renewable and Sustainable Biorefining (I-STAR; Award No. DGE0903701). REFERENCES [1] U.S. Congress, House of Representatives. "Energy Independence and Security Act of 2007. Title II-Energy Security Through Increased Production of Biofuels; Subtitle A—Renewable Fuel Standard." U.S. Government Printing Office, 2007. [2] D. Rajagopal and D. Zilberman. “Review of Environmental, Economic and Policy Aspects of Biofuels.” Policy Research Working Paper No. 4341. Sustainable Rural and Urban Development 14

Team, Development Research Group, The World Bank. 2007. Available at: http://www.wilsoncenter.org/news/docs/worldba nkSept.2007.pdf. [3]D. Rajagopal, S.E. Sexton, D. Roland-Holst and D. Zilberman. “Challenge of Biofuel: Filling the Tank without Emptying the Stomach?” Environmental Research Letters. Vol. 2, 2007. Available at:http://www.iop.org/EJ/article/17489326/2/4/044004/erl7_4_044004.html. [4] I. Altman, C.R. Boessen and D.R. Sanders. “Contracting for Biomass: Supply Chain Strategies for Renewable Energy.” Paper presented at the Southern Agricultural Economics Association Annual Meeting, Mobile, Alabama, February 3 – 6, 2007. Available at: http://ageconsearch.umn.edu/ bitstream/34907/1/sp07al01.pdf. [5] F. M. Epplin, C.D. Clark, R.K. Roberts and S. Hwang. “Challenges to the Development of a Dedicated Energy Crop.” American Journal of Agricultural Economics. Vol. 89, pp. 1296 – 1302, 2007. [6] D.A. Glassner, J.R. Hettenhaus and T.M. Schechinger. “Corn Stover Collection Project.” Paper presented at BioEnergy ’98: Expanding Bioenergy Partnerships, Madison, Wisconsin, October 4 – 8, 1998. Available at: http://ergosphere.files.wordpress.com/2007/04/b io98_corn_stover.pdf. [7]J.A. Larson, B.C. English and L. Lambert. “Economic Analysis of the Conditions for Which Farmers Will Supply Biomass Feedstocks for Energy Production.” Final Report for Agricultural Marketing Center Special Projects Grant 412-30-54, Agricultural Marketing Resource Center, University of Tennessee, 2007. Available at: http://www.agmrc.org/media/cms/2007UTennPr ojDeliverable_9BDDFC4C2F4E5.pdf. [8] J.A. Stricker, S.A. Segrest, D.L. Rockwood and G.M. Prine. “Model Fuel Contract – CoFiring Biomass with Coal.” Paper presented at the Soil and Crop Science Society of Florida and Florida Nematology Forum, 60th Annual Meeting, Tallahassee, Florida, September 20 – 22, 2000. Available at: http://www.techtp.com/ Cofiring/Model%20Contract%20Cofiring%20Bi omass%20with%20Coal.pdf. [9] W.W. Wilhelm, J.M.F. Johnson, J.L. Hatfield, W.B. Voorhees and D.R. Linden. “Crop and Soil Productivity Response to Corn


[22] W.H. Greene, NLOGIT, Version 5 Reference Guide, Econometric Software, Inc. Plainview, NY., 2012. [23] J.E. Fewell, M.K. Lynes, J.R. Williams and J.S. Bergtold. “Kansas Farmers’ Interest and Preferences for Growing Cellulosic Bioenergy Crops.” Journal of American Association of Farm Managers and Rural Appraisers, pp. 132 – 153, 2013. Available at: http://ageconsearch.umn.edu/bitstream/161493/2 /385%20Williams.pdf. [24] P.C. Hipple and M. D. Duffy. "Farmers' Motivations for Adoption of Switchgrass." In: Trends in New Crops and New Uses, edited by J Janich and A Whipkey, Alexandria, VA: AS HA Press, pp. 252-266, 2002. [25] K. Jensen, C. D. Clark, P. Ellis, B. English, J. Menard and M. Walsh. "Farmer Willingness to Grow Switchgrass for Energy Production." Paper presented at the American Agricultural Economics Association Annual Meeting, July 23-26. Long Beach, CA, 2006. [26] D.J. Pannell, G. R. Marshall, N. Barr, A. Curtis, F. Vanclay and R. Wilkinson. "Understanding and Promoting Adoption of Conservation Practices by Rural Landholders." Australian Journal of Experimental Agriculture, Vol. 46, pp. 1407-1424, 2006. [27] S. Paulrud and T. Laitila. "Farmers' Attitudes about Growing Energy Crops: A Choice Experiment Approach." Biomass and Bioenergy, Vol, 34, pp. 1770-1779, 2010. [28] D. J. Qualls, K. L. Jensen, C. D. Clark, B. C. English, J. A. Larson, and S. T. Yen. “Analysis of Factors Affecting Willingness to Produce Switchgrass in the Southeastern United States.” Biomass and Bioenergy, Vol. 39, pp. 159-167, 2012. [29] K.E. Train, Discrete Choice Methods with Simulation. Cambridge, UK: Cambridge University Press, 2003.

Residue Removal: A Literature Review.” Agronomy Journal. Vol. 96, pp.1 – 17, 2004. [10] M. Calvino and J. Messing. "Sweet Sorghum as a Model System for Bioenergy Crops." Current Opinion in Biotechnology Vol. 23, pp. 323-329, 2012. [11] J.L. Propheter, S. A. Staggenborg, X. B. Wu, and D. Wang. "Performance of Annual and Perennial Biofuel Crops: Yields during the First Two Years." Agronomy Journal, Vol. 102, pp. 806-814, 2010. [12] J.J. Louviere, D.A. Hensher and J.D. Swait. Stated Choice Methods: Analysis and Application. Cambridge, UK: Cambridge University Press, 2000. [13] W.H. Greene, Econometric Analysis, 7th Ed. New Jersey: Pearson Prentice Hall, 20012. [14] D.A. Hensher, D.A., J.M. Rose, and W.H. Greene. Applied Choice Analysis: A Primer. Cambridge UK: Cambridge University Press, 2005. [15] National Agricultural Statistics Service, U.S. Department of Agriculture (USDA-NASS). "2012 Census Volume 1, Chapter 1: State Level." 2013. Available at: http://www.agcensus .usda.gov/Publications/2012/Full_Report/Volum e_1,_Chapter_1_State_Level/. [16] B. Roe, T.L. .Sporleder and B. Belleville. “Hog Producer Preferences for Marketing Contract Attributes,” American Journal of Agricultural Economics, Vol 86, pp. 115 – 123, 2004. [17] Kansas Farm Management Association”Summary Reports by Region.”, 2010 Available at: http://www.agmanager.info/kfma/ [18] M. Khanna, X. Chen, H. Huang and H. Onal, “Land Use and Greenhouse Gas Mitigation Effects of Biofuel Policies” University of Illinois Law Review, Vol. 2011, pp. 549-588, 2011. [19] SAS Institute, Inc. SAS for Windows, Version 9.2. Cary, NC., 2008. [20] N.K. Nguyen and A. Miller. “A Review of Exchange Algorithms for Constructing Discrete D-optimal Designs,” Computational Statistics & Data Analysis,Vol. 14, pp. 489-498, 1992. [21] R.D. Cook and C.J. Nachtsheim, “Computer-Aided Blocking of Factorial and Response-Surface Designs,” Technometrics, Vol. 31, pp. 339-346, 1989.

15

Proceedings of SEEP2015, 11-14 August 2015, Paisley MICROALGAE CELL RUPTURE USING HIGH PRESSURE HOMOGENIZATION AS A PRELUDE TO LIPID EXTRACTION C. Onumaegbu, A. Alaswad, T. Prescott and A. Olabi Institute of Engineering and Energy Technologies, School of Engineering and Computing, Univeristy of the West of Scotland, Paisley; email: [email protected] Abstract Microalgae are a potential source for various valuable chemicals for commercial applications ranging from biotechnology purposes to bio-fuels production. The Objective of bio refinery is to utilize biomass ingredients efficiently; this is similar to the process undergone in petroleum refineries in which oil is fractionated into fuels and a variety of products with higher value. Downstream processes in microalgae bio refineries consist of different steps of which cell disruption is the most crucial part. To maintain the functionality of microalgae biomass during cell disruption while obtaining high disruption yields is an important challenge. Cell disruption is an integral part of the downstream pool of unit operations as it enhances the release of intracellular products essential for biodiesel production. Conclusively, the performance of cell disruption method is being evaluated in terms of its efficiency on microalgae biomass cell count reduction which enhances biofuel production. 1

rates, have greater photosynthetic efficiencies, require minimal nutrients and are capable of growth in saline waters which are unsuitable for agriculture [4]. Microalgae utilise a large fraction of solar energy and have the potential to produce 45 to 220 times higher amounts of triglycerides than terrestrial plants. The use of microalgae for biodiesel production requires strain selection, optimisation and viability testing to ascertain the most appropriate organism for large scale cultivation. Despite the various advantages of producing of biofuels using microalgae, an economic feasibility of the microalgae- based biofuels industry comparable to that of either the petroleum or the bioethanol industry has not been achieved. The reason simply the high production cost of algal biofuels and the lack of economic techniques that integrates the multiple steps associated with the harvest, extraction and conversion of biomass to biodiesel. In some review studies, different categories of pre-treatment techniques have been used for algal cell disintegration, depending on the substrate’s morphology, to perform different task while biofuel production were estimated. A mechanical pre-treatment phase is usually the first step not only for biofuel but also bioethanol

INTRODUCTION

Currently, one major source of the global CO2 emissions is the transport sector. The outlook for reduction of emissions from this sector does not look promising as the number of light motor vehicles on the roads globally is estimated to increase to over 2 billion vehicles or more by 2050 [1]. The combustion of fossil fuel produced from the transport sector has large amount of air pollutants which are not environmental friendly. Other detrimental effects of global warming include a high potential increase in sea level and subsequent submerging of lowlands, deltas and islands as well as changing weather pattern [2]. Another issue is the energy depletion of finite fossils fuels resources. The continues use of this fossil fuels as a primary source of energy is widely recognised as unsustainable because of its depleting resources and general effects on environmental degradation [3]. Alternative sources of energy must be developed to replace fossil fuels which will reduce the emissions of greenhouse gases effect. Microalgae have shown great promise as a sustainable alternative to first generation biofuels. They have faster growth 16

Proceedings of SEEP2015, 11-14 August 2015, Paisley fermentation [5]. In particular, bead mills and were harvested/filtered using a round bottled high-pressure homogenizer has been accounted flask that was connect to a pump, an open funnel among the most successful mechanical prewith holes which was fitted with a filter paper treatments on microalgae biomass to produce for algal strain collection. The algal strain was biodiesel. In fact, most studies conducted on oven dried at a temperature of 60°C for 2 h or bead mills and high- pressure homogenizer more and both initial and final weight of filter revealed that both techniques are effective to papers before and after filtration is being 𝑘𝑔−1 considered. enhance biodiesel by 25-145g𝑑𝑤 and 24.90% (w/w) respectively. In general, all these mechanical pre-treatment techniques are high energy demanding [6]. The pre-treatment’s cost has been identified as most outstanding barrier for commercialization of lignocellulosic biofuels [7]. In order developed a large scale energyefficient cell disruption method for bio refineries to be economically feasible, a high pressure homogenizer pre-treatment’s effects on biodiesel yields from microalgae (Scenedesmus sp.) will be investigated in this research. This technique is based on the same ‘comminution’ concept proposed by all other mechanical treatments and has been applied to micro algal feedstock to achieve a positive results. The use of this Fig. 2. Algae Cultivation mechanical technique will be used for algal cell disintegration during the pre-treatment process. Algae Pre-treatment using a Algae biomass Cultivation: Homogenizer: A 5 litre round bottom flask were used for algal The pre-treatment machine consists of a cultivation. The flask were thoroughly wash and modified High pressure homogenizer. The autoclave at a temperature of 40°C, so as to kill machine consists of a sample cylinder cap any form bacteria that will inhibit the growth of through which the algal species passes through algae cells and half filed with a distilled water. A for cell disruption at a pressure of 500 bar for 1.5g of nutrients were added and stirred to only one pass, a pressure gauge (for pressure dissolve the nutrients. The flask was inoculated regulation), mechanically operated by shear with 5 to 10 ml of the algae to be cultured, and stress, cavitation and an impact ring, and an fitted with two pipes in flask which was outlet pipe where the homogenized algal cells is connected to a pump that produces oxygen for being collected after cell disintegration. algal cell growth. During cultivation, the culture medium is being subjected to mixing so as to make sure all algal cells are suspended to identical access to light and also to avoid sticking of the nutrients at the top surface of the flask. The cultivation process was maintained at a temperature of (18-30°C) under a constant illumination throughout the culture period of 21 days. After cultivation period, the algal biomass

17


Fig.2. High-Pressure Homogenizer Fig. 3. Soxhlet Extraction Method. Direct Lipid Extraction USING SOXHLET

Results and discussion:

METHOD: The lipid was extracted using a Soxhlet apparatus. The algal biomass on the oven dried filter papers was cut and fitted into a thimble and measured when empty before extraction. The lipid were extracted from both homogenised and non-homogenized micro-algae using Soxhlet apparatus and petroleum ether as an organic solvent at a temperature of 60°C for at least 4 hours. Initial weight of the flask was measured before and after adding the boiling beads for extraction. When the lipid extraction was completed, the round-bottomed flask that contains the lipid was placed over a steam bath for about 30 minutes to remove the solvent which is mixed up with the lipid during extraction.

Two experimental works have been performed; in the first one, a direct lipid extraction was applied using the sohlet method, while the mechanical pre-treatment method (high pressure homogenizer) was used before the lipid extraction in the second experimental work. The data from the first experimental work suggests that only 0.1573 g of lipid was extracted from samples of micro algae with a combined weight of 2.6483 g, which indicates a lipid content of only 5.9%. While for homogenized algal cells, the microalgae yield was 3.490 g which is being measured as 101.9662 g which means the amount of lipid this time was: 0.3015 g, and a percentage of 8.63896% of the algae was considered as lipid. As a result of using the homogenizer for one pass at 5 bar pressure we were able to get 40.423% lipid increase. Conclusion: Industrial microalgae bio refineries are not yet feasible due to high operational costs. Microalgae cell disruption and its high energy demand, effects on the efficiency of subsequent measures has been the most crucial step in biodiesel production using microalgae biomass. The main aspects of an industrially interesting

18

Proceedings of SEEP2015, 11-14 August 2015, Paisley microalgae cell disruption method include from the lipids after extraction. More research energy efficiency, mildness, selectivity, needs to be done so as to know how to improve controllability and universality. Recent review on microalgae lipid yield. Finally, production of papers indicated that mechanical pre-treatment biodiesel from microalgae biomass using are desired for industrial scale biomass cell mechanical method was reviewed, showing that disruption, but its disadvantage is the high new and improved processes are needed, in specific energy consumption. While nonparticular continues operation, faster reaction mechanical technique may have effect on the times, reduced specific energy demand, zero product quality to a minor extent, but energy disruption agent contamination and lower dissipated during algal cell disruption is much dependence on the water and other impurities. lower as compared mechanical pre-treatment. As To summarize, this project work will aim to a result of non-mechanical inefficiency, energy optimise all the process parameters to get best consumption using high-pressure homogenizers yield of lipid from microalgae biomass. or bead mills is generally preferred for largescale applications. However, these conventional mechanical methods require energy intensive REFERNCES: cooling for the isolation of fragile compounds. Generally speaking, the large energy 1. Balat M, & Bala H. Progress in biodiesel requirements for dewatering, drying, extraction processing. Applied Energy 2010; 87: and further biomass processing pose tremendous pp.1815-35. hurdles for any microalgae-based biofuel 2. Hassan MA, Yacob S, Ghani BA. production technology. Utilization of biomass in MalaysiaMoreover, the choice of the microalgae potential for CDM business. University harvesting and dewatering techniques determine Putra Malaysia, Faculty of the subsequent downstream unit operations, Biotechnology; 2005 including the methods to be used for lipid 3. Sarkar N, Ghosh SK, Bannerjee S, Aikat extraction and possibly the biodiesel production K. Bioethanol production from process itself. In addition, during microalgae agricultural Wastes: an overview. Renew cultivation process, there is need to consider Energy 2012; 37:19e27. measuring the number of cells of each medium 4. Zheng, H., Yin, J., Gao, Z., Huang, H., before cultivation by using spectrophotometer at Ji, X., & Dou, C. (2011). Disruption of least every 7 days during cultivation period. This Chlorella vulgaris cells for the release of will help to know the maximum optimum biodiesel-producing lipids: a comparison growth condition and also the best time/ days for of grinding, ultrasonication, bead maximum cell growth. milling, enzymatic lysis, and Measuring the protein concentration will be used microwaves. Applied Biochemistry and before and after the mechanical pre-treatment Biotechnology, 164(7), 1215–24. will indicate for us the cells breakage percentage doi:10.1007/s12010-011-9207-1 which mainly affects the released lipid 5. Tedesco, S., Marrero Barroso, T., & concentration. In terms lipid extraction using Olabi, A. G. (2014). Optimization of Soxhlet method, there is need to consider using a mechanical pre-treatment of different organic solvent than using petroleum Laminariaceae spp. biomass-derived ether at a different temperature condition during biogas. Renewable Energy, 62, 527–534. extraction. Also, an alternative measure should doi:10.1016/j.renene.2013.08.023. be considered on how to remove the solvent 19

Proceedings of SEEP2015, 11-14 August 2015, Paisley 6. Cheng JJ, Timilsina GR, Status and barriers of advanced biofuels technologies: a review. Renew. Energy 2011; 36:3541-9. 7. Olmstead, I. L. D., Kentish, S. E., Scales, P. J., & Martin, G. J. O. (2013). Low solvent, low temperature method for extracting biodiesel lipids from concentrated microalgal biomass. Bioresource Technology, 148, 615–9. doi:10.1016/j.biortech.2013.09.022.

20


CATALYTIC CRACKING OF TRIGLYCERIDE OVER SULPHATED ZIRCONIA SOLID CATALYST: PROCESS, KINETICS, AND MECHANISM Malik Musthofa1,2, Jonathan Lee1, and Adam Harvey1 1. School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle Upon Tyne, United Kingdom; email: [email protected] 2. Chemical Engineering Department, Faculty of Engineering, Muhammadiyah University of Surakarta, Surakarta, Indonesia. Abstract Heterogeneous catalytic cracking of triglyceride (source: rapeseed oil) was studied over sulphated zirconia and in the absence of catalyst. The maximum yield of cracked oil was 80% (thermal) and 85% (catalytic) at 500oC and 450oC respectively in a stream of helium at a flow rate of 100ml/min. These was supported by the results of kinetics analysis in which the activation energy decreased from 194.6 ± 2.3 to 146.4 ± 0.3 kJ/mol using Kissinger-Akhira-Sunose (KAS) kinetics model. The interesting FTIR spectra suggested a possible mechanism of cracking reaction: triglyceride decomposed to fatty acid, and followed by the formation of another kind of ester. Then, decarboxylation and decarbonylations might be occurred on the fatty acid to produce hydrocarbon, CO, and CO2. The formation of fatty acid and ester compounds was well confirmed by GCMS. In addition, the GC indicated the gases produced during the reactions were CO, CO2, CH4, C2H4, C3H6 and C3H8. Keywords: Guide, Format, Figures, Tables 1 INTRODUCTION Process conversion of triglycerides into biofuels, mainly biodiesel, has been a significant amount of attention [1-3]. The heterogeneous catalytic transesterification is a common technique applied for biodiesel production. The main advantages of this method are high yield of product and the easiness in the separation process. This technology, however, requires plenty of methanol in order to produce biodiesel from the triglycerides [4]. For this reason, catalytic cracking has been proposed as a potential technique for biodiesel production. Moreover, some other valuables chemical could be produced by this technique [4-5]. A lot of research works have been performed on the heterogeneous catalytic cracking of triglycerides [2,5]. However, there are only few publications on the kinetics study of catalytic cracking of triglycerides. In fact there are a lot of controversies on the understanding of this reaction scheme. Therefore, the objective of this research is to study the cracking of triglyceride over sulphated zirconia and to investigate the kinetics model and reaction mechanism of this reaction. This solid catalyst was claimed as an effective and active catalyst for cracking of triglyceride [3].

2

METHODS

2.1 Materials Zirconium oxychloride (ZrOCl2.8H2O) and ammonium sulphate (NH4)2SO4 were purchased from Sigma–Aldrich and Fisher Scientific. Rapeseed oil was from Asda Stores Ltd. 2.2 Catalysts preparation Sulphate zirconia solid catalyst was synthesized according to the solvent free method [6]. Zirconium oxychloride and ammonium sulphate were carefully weighed on the basis of molar ratio 1:6 respectively into a mortar and grinding for 20 min at room temperature. The mixture was left to age for 18 h at room temperature and calcined at 600oC for 5 h. 2.3 Catalysts characterization The structure of sulphated zirconia was determined using a Siemens D 5000 X-ray diffractometer and Cu K a radiation with a wavelength of  = 1.54 Å generated at 40 kV and 40 mA. Fourier Transform Infrared Spectroscopy (FTIR) measurements were performed in a Varian 800 (Scimitar series) spectrometer. The spectra were produced between 4000 cm-1 and 400 cm-1 using a Pike Technologies diamond crystal plate ATR.

21

Proceedings of SEEP2015, 11-14 August 2015, Paisley refers to a liquid or solid that is left and found in the sample boat after completion of the reaction. The surface areas were obtained from N2 BET Figure 1 presents an interesting result of product (Brunauer, Emmett and Teller) isotherms distribution. The highest yield of liquid product determined at 77 K using the Coulter™ (SA for non- catalytic cracking (80% ± 2%) was 3100™ series). The samples were outgassed for obtained at about 500oC. On the other hand, by 2 h at 200oC prior to the analysis. adding sulphated zirconia the maximum yield 2.4 Catalytic cracking of triglyceride can be achieved at lower temperature, 450oC. The cracking of rapeseed oil was performed in a This might be caused by the decrease of fixed bed reactor. Initially, the reactor was activation energy of the reaction due to the role heated from room temperature to the desired of catalyst. This value was better as compared to temperature (300oC-500oC) prior to the feed that reported by Dupain et al. which were 40% placement. When the temperature was attained, and 60% [7-8]. Moreover, the liquid product sample boat containing 3 g of rapeseed oil was obtained in this research was comparable with then carefully inserted into the reactor. The those reported by previous studies [9]. volatile products formed during the reaction were swept by the helium flow of 100 ml/min through a coiled condensing trap. The liquid fraction was analysed using FTIR react iC10 and Perkin Elmer (Clarus) 600/560D Gas Chromatography and Mass spectroscopy (GCMS), while the product gas was analysed by Hewlett Packard 6890 gas Chromatography (GC). 2.5 Kinetics investigation Studies of the kinetics were performed on a Perkin Elmer TGA-STA 6000 Model. Samples (10 mg) were subjected to non-isothermal conditions (50–900oC) at different heating rates (5, 10, 15, 20, and 50oC/min) under flowing nitrogen at constant rate of 30 ml/min. 3

Figure 1: product distribution of triglyceride cracking with and without catalyst.

RESULTS AND DISCUSSION

3.1 Characteristic of sulphated zirconia The XRD diffractograms of sulphated zirconia shows an amorphous phase. It completely agrees with that was reported by Elizabeth et al. [3]. The IR spectra of the catalyst indicates strong, broad bands in the region 3550–3000 cm-1 (due to physisorbed and coordinated water) and weaker absorptions band at 1560–1640 cm-1 (assigned to the bending mode ( δHOH) of coordinated water), this agrees with the previous report [3,6]. The bands between 1250 and 950 cm-1 were observed as well which were typical of sulphate ions groups coordinated to the zirconium cation [6]. In addition, the BET surface area was 158m2/g. 3.2 Product distribution The main products of this reaction were liquid, gas, and coke. Liquid and solid residue, however, were found as well at low temperature which were 300 and 350oC respectively. Residue

3.3 Kinetics model The investigation on the kinetics of cracking of vegetable oils based on Thermogravimetric Analysis (TGA) was carried out and numerous kinetic models have been proposed [10-11]. The kinetic study of thermal decomposition is a very complex process involving a large number of reactions [9-11]. In general, the reaction can be represented by the following reaction scheme: (1) 𝐴 (𝑙) → 𝐵(𝑣𝑜𝑙𝑎𝑡𝑖𝑙𝑒) + 𝐶 (𝑠) The rate of the change of oil weight can be expressed as: 𝑑𝑥 (2) = 𝑘𝑓(𝑥)𝑛 𝑑𝑡 The temperature dependence of the rate of weight loss was expressed by Arrhenius equation: −𝐸𝐴 (3) 𝑘(𝑇) = 𝐴𝑒𝑥𝑝( ) 𝑅𝑇 Combining eq. (2) and (3) generates: 𝑑𝑥 −𝐸𝐴 = 𝐴𝑒𝑥𝑝( )𝑓(𝑥)𝑛 𝑑𝑡 𝑅𝑇

22

(4)

Proceedings of SEEP2015, 11-14 August 2015, Paisley Those plots generated the average activation Under non-isothermal conditions in which energy of 194.6 ± 2.3 kJ/mole for non-catalytic samples are heated at a constant heating rate,  = cracking and 146.4 ± 0.3 kJ/mole for catalytic dT/dt, Eq. (4) can be eliminated to be: cracking. This reduction of activation energy 𝑑𝑥 −𝐸𝐴 possibly is the reason for the increase of liquid 𝛽 = 𝐴𝑒𝑥𝑝( )𝑑𝑇 (5) 𝑓(𝑥)𝑛 𝑅𝑇 product yield as presented in Fig. 1. In addition, the average activation energy value for nonThe integration isoconversional method catalytic cracking is in agreement with that is suggested by Kissinger-Akhira-Sunose (KAS) is reported by Filho et al. for rapeseed oil [12]. based on Eq. (5) leads to: 𝑙𝑛 [

𝛽 𝐴𝑅 𝐸𝐴 ] = 𝑙𝑛 [ ]− 2 𝑇 𝐸𝐴 𝑔(𝑥) 𝑅𝑇

(6)

𝑑𝑥 𝑓(𝑥)𝑛

(7)

𝑔(𝑥) = ∫

Where x is the extent of conversion, T is the temperature (K), A is frequency factor, E is activation energy, R is the gas constant, and f(x) is a function of conversion. When ln (/T2) vs 1/T is plotted for different conversion levels a straight line is obtained and slope of the line corresponds to (-EA/R), thus the activation energy can be calculated. Isoconversional KAS’s plots obtained for this reaction, non-catalytic and catalytic, are shown in Fig. 2 and Fig. 3 respectively.

Figure 2: KAS plot for non-catalytic cracking of rapeseed oil.

Figure 3: KAS plot for catalytic cracking of rapeseed oil.

23

3.4 Cracking mechanism Kinetics parameters for the reaction are clearly can be predicted using TGA study as described earlier. Those results, however, are not able to provide a reaction scheme for this reaction. In order to observe a possible reaction mechanism, FTIR analysis was performed. Spectra of every single liquid product from room temperature to 500oC were collected. The interesting result of FTIR analysis is a presence of significant change for two peaks at wavenumber of about 1745 cm-1 and 1710 cm-1. The peak appearing at 1740-1750 cm-1 indicates ester carbonyl while that is at 1710-1715 cm-1 refers to carboxylic acid carbonyl [13]. In the absence of catalyst, peak of ester carbonyl slightly decreased at 300oC. When temperature increased to 350oC peak of ester considerably decreased, and big peak of carboxylic acid was generated. There was only small decrease and increase for ester and carboxylic acid peaks at 400-500oC, and then tended to be constant. Conversely, in catalytic cracking of rapeseed oil, a new peak of carboxylic acid carbonyl was formatted already event at 300oC when the peak of ester slightly decreased. In this case, at 350oC, the lowest peak of ester and highest peak of carboxylic acid were achieved. Surprisingly, at higher temperature (400oC-500oC), reformation of ester peak happened while the peak of carboxylic acid tended to decrease. All these FTIR analysis results strongly reveal a reduction of triglyceride concentration with the compensation of the carboxylic acid compound formation. In other world, the changes of ester and carboxylic acid peaks indicate a route of the cracking. In order to have better understanding of a possible reaction mechanism of this reaction, a quantitative analysis of FTIR results was done. This quantitative analysis was performed by prepared ester and carboxylic acid solutions at different concentration. Then, IR spectra of these solutions were recorded. By calculating the peak

Proceedings of SEEP2015, 11-14 August 2015, Paisley area of ester and carboxylic acid at various cracking of rapeseed is similar to that was concentrations a calibration curve can be proposed by Gusmau et al. [14] as follow: designed. Based on this calibration curve the change of peak area of ester and carboxylic acid can be better expressed as described in Fig. 4 for non-catalytic cracking and Fig. 5 for catalytic cracking.

Figure 6: A possible scheme of reaction of catalytic cracking rapeseed oil. 4 CONCLUSION Cracking of triglyceride was studied with and without catalyst. In the presence of sulphated zirconia, yield of liquid product was similar to that of without catalyst, but it was obtained at lower temperature. The potential content of fatty acid methyl ester (FAME) and its selectivity is subject of further investigation. The kinetics model and mechanism of reaction were successfully proposed. Detailed kinetics based on the mechanism was not clear yet and would be a further work.

Figure 4: Reduction of ester concentration and increase of carboxylic acid concentration during non-catalytic cracking of rapeseed oil.

ACKNOWLEDGEMENTS The research was supported by Directorate General of Higher Education Ministry of Education and Culture of Indonesia, and Muhammadiyah University of Surakarta, Indonesia. Figure 5: Reduction and reformation of ester concentration and increase of carboxylic acid concentration during catalytic cracking of rapeseed oil. Figure 4 and 5 obviously suggested that the first step of the route of reaction is degradation of triglyceride and carboxylic acid formation. In catalytic reaction, then, the carboxylic acid would be cracked and decreased while ester compound would be produced. Another essential analysis performed for liquid product was GCMS. Ester, carboxylic acid, alcohol, and aldehyde were chemical compound identified by GCMS. Additionally, gas produced during the reaction was analysed by GC. Some gases that are probably produced were including CO, CO2, CH4, C2H4, C3H6, and C3H8. By compiling the results of FTIR, GCMS, and GC analysis so the possible mechanism of this 24

REFERENCES [1] Doronin, V.P. Potapenko, O.V. Lipin, P.V. Sorokina, T.P. and Buluchevskaya, L.A. (2012) ‘Catalytic Cracking of Vegetable Oils for Production of HighOctane Gasoline and Petrochemical Feedstock’, Petroleum Chemistry, Vol. 52, No. 6, pp. 392–400. [2] Taufiqurrahmi, N. and Bhatia, S. (2011) ‘Catalytic cracking of edible and non-edible oils for the roduction of biofuels’, Energy Environ. Sci., 4, pp. 1087-1112. [3] Eterigho, E.J. Lee, J.G.M. and Harvey, A.P. ‘Triglyceride cracking for biofuel production using a directly synthesised sulphated zirconia catalyst’, Bioresource Technology, 102, pp. 6313-6316, 2012. [4] Meher, L. C., Vidya Sagar, D. and Naik, S. N. 'Technical aspects of biodiesel production by transesterification - A review', Renewable and

Proceedings of SEEP2015, 11-14 August 2015, Paisley Sustainable Energy Reviews, 10, (3), pp. 248268, 2006. [5] Ong, Y. K., and Bhatia, S. ‘The current status and perspectives of biofuel production via catalytic cracking of edible and non-edible oils’, Energy, 35, pp. 111–119, 2010. [6] Sun, Y., Ma, S., Du, Y., Yuan, L., Wang, S., Yang, J., Deng, F. and Xiao, F. S. 'Solvent-free preparation of nanosized sulphated zirconia with brønsted acidic sites from a simple calcination', Journal of Physical Chemistry B, 109, (7), pp. 2567-2572, 2005. [7] Dupain, X., Costa, D. J., Schaverien, C. J., Makkee, M. and Moulijn, J. A. 'Cracking of a rapeseed vegetable oil under realistic FCC conditions', Applied Catalysis B: Environmental, 72, (1-2), pp. 44-61, 2007. [8] Lima, D.G. Soares, V.C.D. Ribeiro, E.B. Carvalho, D.A. Cardoso, E.C.V. Rassi, F.C. Mundim, K.C. Rubim, J.C. and Suarez, P.A.Z. ‘Diesel-like fuel obtained by pyrolysis of vegetable oils’, J. Anal. Appl. Pyrolysis, 71, pp. 987–996, 2004. [9] Biswas, S. and Sharma, D.K. ‘Studies on cracking of Jatropha oil’, Journal of Analytical and Applied Pyrolysis, 99, pp. 122-129, 2013. [10] Singh Chouhan, A.P. Singh, N. and Sarma , A.K, ‘A comparative analysis of kinetic parameters from TGDTA of Jatropha curcas oil, biodiesel, petroleum diesel and B50 using different methods’, Fuel, 109, pp. 217–224, 2013. [11] Haykiri-Acma, H. Yaman, S. and Kucukbayrak, S., ‘Effect of heating rate on the pyrolysis yields of rapeseed’, Renewable Energy, 31, pp. 803–810, 2006. [12] Filho, W.L. Mannke, F. Mohee, R. Schulte, V. and Surroop, D., ‘Climate-Smart Technologies: Integrating Renewable Energy and Energy Efficiency in Mitigation and Adaptation Responses’, Springer London, pp. 563-571, 2013. [13] Donald L. Pavia,Introduction to Spectroscopy, Deprt. Of Chemistry Univ. of Belingham, Washington, 2009. [14] Gusmao, J., Brodzki, D., DjegaMariadassou, G. and Frety, R., 'Utilization of vegetable oils as an alternative source for dieseltype fuel: hydrocracking on reduced Ni/SiO2 and sulphided Ni-Mo/γ-Al2O3', Catalysis Today, 5, (4), pp. 533-544, 1989.

25


Plasma Assisted Fischer Tropsch Synthesis T. Mukhriza (1,2), K. Zhang (3), A. Phan (4) 1. School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle upon Tyne UK, e-mail: [email protected] 2. Department of Chemical Engineering, Syiah Kuala University, Banda Aceh, Indonesia 3. School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle upon Tyne UK, e-mail: [email protected] 4. School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle upon Tyne UK, e-mail: [email protected] Abstract Fischer-Tropsch (FT) synthesis has attracted significant interest as alternative process to convert coal and natural gas to liquid fuels. The product mixture is a function of catalyst, process conditions (pressure and temperature, residence time etc.), and synthesis gas composition. This research is to investigate the conversion of syngas to liquid fuels and chemicals using a non-thermal plasma catalytic process at atmospheric pressure and temperature. Plasma technologies have been widely used for many applications such as waste treatment, ozone production and methane reforming. Metal based catalysts including cobalt and nickel were prepared and characterised. Microwave radiation has been used to prepare catalysts and the crystallite size of 3-8 nm was formed, which was reported to be the optimum size for FT synthesis. The catalysts were then applied plasma assisted FT process over a range of power plasma (20-60W). The effects of molar ratios of H2/CO, catalyst composition etc, on the conversion and selectivity were investigated The results showed that the conversion of syngas increase significantly with the increase of power while the selectivity towards C5+ are subject to the catalyst used. Keywords: Syngas, Fischer-Tropsch (FT) synthesis, Plasma Catalytic Reactor 1 INTRODUCTION FT synthesis can be defined as a process to convert synthesis gas containing hydrogen and carbon monoxide to hydrocarbon products using catalyst commonly cobalt and iron. Low Temperature Fischer-Tropsch synthesis (LTFT) has been recognized to have better selectivity toward long chain hydrocarbon, Catalyst play important role on the activity of FT synthesis including increase hydrogen and carbon monoxide conversion and tuning the right selectivity toward desired product. The developments of selective FT catalysts, which can increase the selectivity to desired products that is C5+ for LTFT, contribute to the most challenging research in the field of the FT synthesis. Although catalysts derived from group VIII metals i.e. Ru, Co, Fe and Ni can be used for the FT synthesis, iron and cobalt are commercially used due to their price, of which cobalt offers better resistance to deactivation while nickel catalyst tends to methanation [1]

26

Co particles size on supporting material had a strong effect on the selectivity of C5+ hydrocarbons over CH4 [2]. The CH4 selectivity was higher for the catalysts with smaller Co particles and decreased with increasing Co particle size in the range of 2.6−8 nm. The C5+ selectivity increased with the mean size of Co nanoparticles. A further increase in Co particle size above 6-8nm had little effect of the C5+ selectivity. An increase pore size of the support leads to increase of Co3O4 Crystallite size. This due to large pore size was easy to reduce resulting from weaker interaction between cobalt oxide species with support. The most common method to prepare heterogeneous catalyst is impregnations. Metal is deposited on porous support by contacting dry support with solution containing dissolved cobalt precursor [1]. This step will follow by drying, calcination and reduction. The heating step used in this conventional method, however, may lead to the formation of metal silicate. The potential application of microwave for preparing highly active active catalyst in FT synthesis has

Proceedings of SEEP2015, 11-14 August 2015, Paisley been explored recently . Using Co/SBA catalyst, In this work, dielectric barrier (DBD) plasma has Rodrigues et al [3] found that catalyst prepared been developed for FT synthesis. Synergistic by microwave offer better selectivity of C5+ effect between catalyst and DBD plasma was (78.5%) than those prepared by conventional expected to offer better conversion and high heating (71.9%). The increase of CO conversion selectivity towards long chain hydrocarbon. for microwave prepared catalyst was also Beside cobalt catalyst which has been proven as detected in this research. FT catalyst, the nickel was also used as Difference reactors have been proven to perform alternative catalyst. highly exothermic FT reactors including multitubular and slurry bubble column reactors are used in industrial scale [4]. It is, however, 2 EXPERIMENTAL both of the reactors have several drawbacks. 2.1. Catalyst Preparation Multi tubular reactors fall off from high pressure drop, low catalyst utilization, and insufficient Cobalt silica catalyst were prepared by adding heat removal. While the slurry bubble column cobalt (II) nitrate hexahydrate (Sigma Aldrich reactors suffer from catalyst separation, less with purity ≥98%) into colloidal silica ideal residence time behaviour, and highly (Azkonobel) at molar ratios of cobalt (II) nitrate demanding scale-up. This problem has raised the hexahydrate to silica gel of 1:4 The precursor need for improved reactor. solution was then stirred vigorously using a Non-thermal plasma technology may serve magnetic stirrer and irradiated in a 1000 W alternative to the existing FT reactors. It has commercial microwave (Panasonic) at 1000 W characteristic of low power requirement and the to form catalyst oxide. Cobalt nitrate was ability to induce physical and chemical reaction decomposed and formed black solid cobalt at low temperature. Plasma is produced by oxide. This catalyst was then calcinated in a applying energy to a gas in order to organize the furnace in the presence of air at 550 oC for 2 electronic structure of the species (atoms, hours and followed by a reduction in a hydrogen molecules) and to produce excited species and flow at 550 °C for 24 h. The temperature ramp ions. In non-thermal plasma, e.g. dielectric rates during calcination and reduction were 5 barrier discharge (DBD), all discharged species o C/min [7]. The same procedure as above was are in thermally non-equilibrium state. The used prepare Nickel (II) nitrate hexahydrate electron temperature is in a range of 10,000– (Sigma Aldrich, purity ≥97%) 100,000 K (equivalent to 1–10 eV), which is much higher than the gas temperature (~100 K) 2.2. BET Surface Area [5]. Instead of providing energy to the system, Surfer (Thermofisher) was used in this research the non-thermal plasma generates radicals and to measure surface area and pore size of the exited species, which enable to initiate and to catalyst. The specific surface area of the enhance the chemical reactions. catalysts was measured using the Brunauer– The role of catalyst in plasma reactor was Emmer–Teller (BET) method with nitrogen proposed by Mishusima et al. [6] in the study of adsorption at −196 °C.. plasma ammonia synthesis using rutheniumloaded membrane-like alumina tube as a 2.3. X-ray diffraction catalyst. The amount of ammonia produced by X-ray diffraction pattern were recorded by exposing the N(a) atoms to H2 plasma was Philips PW3040/60 X-ray exposed to Cu-Kα Xnoticeably increased by Ru. Plasma discharge ray radiation at a wavelength () of 1.5418 Å. driven the activation of hydrogen species such as Using A PANalytical X'Pert Pro MPD particle H atoms and H2 molecules. These hydrogen size and phase identified over a range of 0 – 110 o species react with the N(a) atoms directly or via 2θ according to Scherrer equation [8]. adsorption on the alumina to form NH3. When ruthenium catalyst was present on the alumina, ammonia was also produced by reactions of the N(a) atoms with H(a) atoms on Ru. This reaction was faster and higher amount of NH3 was obtained [6]. 27

Proceedings of SEEP2015, 11-14 August 2015, Paisley the catalyst. The decomposition of nitrate groups 2.4. TGA Thermo-gravimetric analysis was performed on (NO3)- on cobalt and nickel started at 150 oC. cobalt and nickel catalysts after microwave Both of the catalysts are stabile to use in the treatment using TGA on a Perkin Elmer STA typical FT Synthesis of 200 – 350 oC o 6000 Model. Samples were heated from 30 C to 500 oC using a heating ramp of 5o C/min with air flow rate of 30 ml/min. The sample loading was typically 4 – 9 mg. 2.5. Plasma and Thermal Catalytic To create plasma in the reactor, a stainless steel mesh was wrapped around the outer surface of the tube as grounded electrode and the high voltage anode was inserted inside the inner tube. Catalyst will be dispersed and placed in the plasma zone. Syngas (CO, H2) was introduced to the reactor filled with glass beads for FT synthesis. The product was analysed online using Variance GC-450 equipped with 2 ovens, 5 columns and 3 detectors (2 TCDs and 1 FID). The GC was connected to the outlet of the reactor. A constant flow of nitrogen (7.0 mL/min) is used as reference gas in order to monitor the change of volume flow during the reaction. Fixed Bed Thermal Reactor was also used as the comparison. All reaction was performed at H2/CO molar ratio of 2 and GSHV= 0.75 Nl grcat-1 h-1. 3. Results 3.1. Composition of the catalyst

Fig. 2. XRD pattern of cobalt and nickel silicasupported catalyst after microwave irradiated Fig 2 illustrated the cobalt and nickel phase after microwave irradiation. XRD patterns exhibited the characteristic of Co3O4 phase and NiO phase. The crystallite sizes were calculated by using scherer equation and listed in table 1. There was no peak attributable to cobalt metallic phase in figure 2. This suggested that either the crystallite very small or the cobalt oxide particles was too amorphous. The crystalline size of Co3O4 was 8.41 nm, which satisfy the characteristic of ideal cobalt FT synthesis catalyst (2 – 8 nm). The crystalline size of NiO was 3.28 nm. BET surface area for cobalt and nickel catalyst were 173 m2/g and 192 m2/g, respectively.

3.2. Catalyst performance Table 2 Performance of cobalt and nickel catalyst on Plasma Catalytic Fischer-Tropsch Synthesis at plasma power of 60 W Fig. 1. TGA Curves of Cobalt and Nickel Silicasupported catalyst. Temperature ramp 2 oC/min.

Catalyst

TGA curves of both catalysts are shown in the Fig. 1. The decomposition profiles for the cobalt and nickel catalyst performed quite similar trend. The endothermic weight loss of 3.25 – 5.19% and 3.4 – 7.18% can be seen at 50 – 82 oC for cobalt and nickel, respectively. This can be explained by the removing of water content from 28

Conversion (%)

Selectivity (%)

H2

CO

CO2

CH4

C2-C4

C5+

Cobalt

31.09

42.06

12.43

16.63

7.76

63.18

Nickel

80.83

96.65

32.57

53.27

14.16

0.00

The catalytyc evaluation of both catalyst were listed in table 1. Nickel catalyst gave much higher conversion than cobalt samples.

Proceedings of SEEP2015, 11-14 August 2015, Paisley However, no long chain hydrocarbon detected in 3.3. Plasma Vs Thermal FT Synthesis The test for both reactors has shown that thermal the product of plasma nickel catalyst. The reactor offer better CO conversion for FT highest selectivity was methane (53.27%) and synthesis. The product selectivity towards C5+ this result corresponded to Khodakov et al [1] hydrocarbon is quite comparable for low that nickel catalyst tend to methanation in the FT temperature FT synthesis. It is interesting to note reaction. that plasma reactor provides better selectivity at The very important factor the influence plasma is higher power compared to thermal reactor. It specific input energy (SIE) which defined as could maintain C5+ selectivity as high as 72% at equation, 90 W while thermal reactor decreased to ~50% at 260 oC. SIE = [Power / Flow rate)] / 1000 (1) Where SIE unit is kJ/L, Power unit is Watt and flow rate unit is L/s.

In order to achieve a higher specific input energy (SIE), the plasma power was varied between 20 – 60 W while flow rate was fixed at 10 ml/min. Fig. 3 showed the effect of SIE on CO conversion and C5+ selectivity for both catalysts. The conversion of CO increase from 22.17 – 42.06% and 91.21 – 96.65% with the increase of SIE in the range of 90 – 360 kJ/L for cobalt and nickel, respectively. Previous study has showed that increasing plasma power at a constant excitation frequency effectively enhances the electric field, electron density and gas temperature in the discharge[9]. It can be noted that rising of SIE decreases the selectivity of C5+ hydrocarbon. While it decrease from ~78 to ~63% for cobalt, the selectivity of C5+ was not detected at SIE of 360 kJ/L for nickel. The methanation reaction taken into effect at high energy input for nickel catalyst.

Fig. 4. FT Synthesis by plasma and thermal reactor on cobalt catalyst 4. CONCLUSION The Fischer-Tropsch synthesis was performed on plasma reactor using cobalt and nickel silicasupported catalyst prepared by microwave irradiation. Nickel exhibited better conversion but methanation took place. Another experiment conducted on thermal reactor for cobalt catalyst. The result showed that plasma reactor is promising reactor as it could maintain high selectivity towards C5+ hydrocarbon at higher CO conversion. Preliminary analysis on the catalyst exposed that microwave irradiation has potential to replace conventional drying method. Further investigation is required to understand the dispersion and reducibility of the metal on the surface. ACKNOWLEDGEMENTS The author thanks Directorate General of Higher Education Ministry of Education and Culture of Indonesia and Syiah Kuala University for providing financial support.

Fig. 3. Effect of Specific Input Energy (SIE) on CO Conversion and C5+ Selectivity

29

Proceedings of SEEP2015, 11-14 August 2015, Paisley Hydrogen Energy, 2014. 39(18): p. 9658REFERENCES 9669. 1. Khodakov, A.Y., W. Chu, and P. Fongarland, Advances in the Development of Novel Cobalt Fischer−Tropsch Catalysts for Synthesis of Long-Chain Hydrocarbons and Clean Fuels. Chemical Reviews, 2007. 107(5): p. 1692-1744. 2. Zhang, Q., W. Deng, and Y. Wang, Recent advances in understanding the key catalyst factors for Fischer-Tropsch synthesis. Journal of Energy Chemistry, 2013. 22(1): p. 27-38. 3. Rodrigues, J.J., F.A.N. Fernandes, and M.G.F. Rodrigues, Study of Co/SBA-15 catalysts prepared by microwave and conventional heating methods and application in Fischer–Tropsch synthesis. Applied Catalysis A: General, 2013. 468(0): p. 32-37. 4. Guettel, R., U. Kunz, and T. Turek, Reactors for Fischer-Tropsch Synthesis. Chemical Engineering & Technology, 2008. 31(5): p. 746-754. 5. Chen, H.L., H.M. Lee, S.H. Chen, Y. Chao, and M.B. Chang, Review of plasma catalysis on hydrocarbon reforming for hydrogen productionInteraction, integration, and prospects. Applied Catalysis B-Environmental, 2008. 85(1-2): p. 1-9. 6. Mizushima, T., K. Matsumoto, H. Ohkita, and N. Kakuta, Catalytic Effects of Metal-loaded Membrane-like Alumina Tubes on Ammonia Synthesis in Atmospheric Pressure Plasma by Dielectric Barrier Discharge. Plasma Chemistry and Plasma Processing, 2007. 27(1): p. 1-11. 7. Akay, G., AMMONIA PRODUCTION BY INTEGRATED INTENSIFIED PROCESSES. 2012, International Patent Publication, PCT/20101/051620 8. Speakman, S.A. Basic of XRD. 2012; Available from: http://prism.mit.edu/xray. 9. Tu, X. and J.C. Whitehead, Plasma dry reforming of methane in an atmospheric pressure AC gliding arc discharge: Cogeneration of syngas and carbon nanomaterials. International Journal of

30


ETHANOL PLANT LOCATION DECISION IN THE BRAZILIAN CERRADO 1. 2. 3. 4.

G. Granco1, A.C.Sant’Anna2, J.S. Bergtold3 and M.M.Caldas4 Department of Geography, Kansas State University, Manhattan, Kansas; email: [email protected] Department of Agricultural Economics, Kansas State University, Manhattan, Kansas; email: [email protected] Department of Agricultural Economics, Kansas State University, Manhattan, Kansas; email: [email protected] Department of Geography, Kansas State University, Manhattan, Kansas; email: [email protected]

Abstract Global demand for biofuels has sharply increased due to higher prices of oil and impacts that fossil fuels have on climate change and the environment. In Brazil, the rise of ethanol production has been more significant across the Brazilian Cerrado, specifically in the states of Goiás and Mato Grosso do Sul. The location decision of ethanol plants has an important role on the success of the operation because ethanol mills are supply-oriented firms that need to be located near agricultural feedstock to ensure access and reduce transaction cost. The purpose of this paper is to model the location decision of ethanol mills in the states of Goiás and Mato Grosso do Sul. The present research draws attention to agricultural production factors, a new set of factors regarding ethanol mills location decisions. We employed two econometric model, probit and SAR probit, to test a set of both industrial and agricultural factors hypothesized to impact the location decision. Our results highlighted the need to incorporate agricultural production factors in location models for ethanol mills to improve the understanding of the sector because there is an increasing participation of mills in the agricultural production, especially in agriculture frontier such as the Brazilian Cerrado. Keywords: Sugarcane, Agricultural Production, Mills 1 INTRODUCTION Global demand for biofuels has sharply increased since the 2000s due to higher prices of oil and impacts that fossil fuels have on climate change and the environment [1,2]. This expanding market has attracted investments to the production of biofuels, resulting in an increase from 0.3 million barrels per day in 2001 to 1.9 million barrels per day of biofuels in 2012 [3]. Among the diverse types of biofuels, fuel ethanol accounts for 77% of worldwide biofuel production in 2012 [3]. This increase in production has come from investments in new ethanol plants and expansion of production capacity of existing plants. The top two producing countries, United States and Brazil, witnessed the construction of 114 and 80 new ethanol plants, respectively, between 2006 and 2012. In this period, ethanol production reached 50 and 23 billion liters in the United States and Brazil, an increase of 170% and 35% respectively [4–6]. Even though the recent development is a significant increase, there is still need to continue to increase ethanol production to meet demand established by 31

blending mandates in several countries and the Brazilian demand for fuel [7]. Brazil is not only the second main producer, but it is also the second largest consumer of fuel ethanol [8]. Brazilian consumers have been using sugarcane ethanol blended with gasoline in theirs vehicles since 1930s. Later on in the 1980s, consumers were able to fuel their cars with ethanol [9]. More recently, with the introduction of flex fuel cars in 2003, consumers can decide the proportion of ethanol and gasoline they will mix in their car [10]. This technology together with competitive ethanol prices1 have stimulated the consumption of ethanol in Brazil. Under these circumstances, the Brazilian sugarcane ethanol industry has increased its production capacity to meet rising domestic and international demand for ethanol. To achieve this growth, the industry has expanded into new frontiers in the Brazilian Cerrado, especially in the states of Goiás and Mato Grosso do Sul [11]. Expanding the 1

Ethanol price in Brazil is competitive against gasoline when it is 30% cheaper than gasoline.

Proceedings of SEEP2015, 11-14 August 2015, Paisley literature by considering the case of an sugarcane ethanol industry into these new integrated decision that takes into account frontier represents a challenge to the Brazilian industrial and agricultural aspects. Additionally, ethanol industry because the agricultural this study is valuable to policy-makers and feedstock, sugarcane, was not initially present. organizations involved with the expansion of Ethanol plants are traditionally placed close to ethanol production to meet the increasing feedstock suppliers in an attempt to reduce demand for ethanol worldwide. acquisition and transportation cost [12]. The The next section of the paper presents a location decision of ethanol mills has an literature review of sugarcane ethanol production important role on the success of the operation, in Brazil and its expansion into the Cerrado. The because ethanol mills are supply-oriented firms third section presents a review of ethanol plant that need to be located near agricultural location literature. The fourth section is feedstock to ensure access and reduced costs of dedicated to the empirical model and data used acquisition. This strategy holds more importance for modelling the location decision in the states to the sugarcane ethanol industry, because of Goiás and Mato Grosso do Sul. The fifth sugarcane needs to be promptly processed after section presents the econometric model results harvest to avoid the loss of the sugar content. and discussion on the effects of location factors Literature has focused on US ethanol plant on the sugarcane ethanol plant location decision. location, showing that the presence of corn was The final section presents some concluding a major factor in the location decision process remarks. [12,13]. Previous studies also related the location decision to infrastructure and 2 BRAZILIAN SUGARCANE ETHANOL governmental policies [14–16]. However, the INDUSTRY literature on ethanol plant location in areas that Brazil has a long and successful history of are developing the needed feedstock production ethanol production and consumption [17]. capacity, such as the Brazilian Cerrado, is Ethanol in Brazil is primarily produced from limited. Therefore, the location decision of sugarcane, which is processed in industrial ethanol plants (sugarcane mills) in the states of plants called mills. Sugarcane was first Goiás and Mato Grosso do Sul differ from introduced in Brazil, by Portuguese settlers in previous studies on ethanol plant location, the fourteenth century, to produce sugar. It was because this decision involves establishing in 1931 that the Brazilian government substantial feedstock production. Furthermore, established the first blend mandate of sugarcane the ethanol industry in Brazil is vertically ethanol and gasoline. But it was during the integrated, as it is responsible for both the 1970s that sugarcane ethanol became a primary management of agricultural production and for alternative to gasoline. With the oil crises in the processing of sugarcane into ethanol. 1973 and 1979, the Brazilian government turned The purpose of this paper is to model the to alternative fuels in an attempt to reduce costly location decision of new ethanol plants in the oil imports [18]. The government created the states of Goiás and Mato Grosso do Sul, two Proalcool Program to support the development very important ethanol producing states that of alternative fuels, which in turn promoted the constitute the new frontier for sugarcane ethanol expansion of sugarcane ethanol [9,17,19,20]. production in Brazil. To achieve this goal, a The Proalcool Program was very important to probit and a spatial autoregressive probit (SAR the development of the sugarcane ethanol probit) regression model are developed, using industry in Brazil by offering subsidized credit mill location in a given municipality as the for production, stimulating demand and dependent variable, and explanatory variables governing ethanol prices [9,17,21]. From 1975, derived from location theory, such as location of the beginning of Proalcool, until 1990, ethanol existing mills, distance to major cities, labor, and production increased from 0.6 billion liters to roads. Moreover, factors related to agricultural 11.9 billion liters. During the Proalcool years, production are also considered as explanatory the ethanol industry had been concentrating its variable, such as amount of area under investments in São Paulo State due to lessen agriculture before the sugarcane expansion and transportation costs, reduce agricultural area suitable for sugarcane expansion. This production costs, favorable agricultural growing paper contributes to the development of the conditions for sugarcane, development of 32

Proceedings of SEEP2015, 11-14 August 2015, Paisley close to their input sources in order to minimize sugarcane varieties specific for the state, and costs associated with obtaining feedstock for its presence of production know-how [10]. operation [22]. Apart from evaluating the Governmental intervention in the sector was potential for a new location based on revenue substantially reduced in the 1990s as the factors, such as market penetration and access to government began the process of deregulation of transportation network, a firm also considers the sugarcane sector [21]. factors that could potentially minimize its costs After deregulation, the sugarcane industry [22]. These factors include the raw products’ concentrated even more in São Paulo. This markets, available infrastructure, state and concentration led to a consolidation of the county incentives, among others [15]. industry, increase in production costs and Rational firms aim at minimizing costs in order stronger competition for sugarcane areas [11]. to maximize profit. Therefore it is assumed that Consequently, sugarcane ethanol firms have a mill will chose a location i based on these started to expand elsewhere. The main factors subject to an indirect cost function destinations have been the states of Goiás and consistent with Leontief production technology Mato Grosso do Sul. This sugarcane expansion and a reduction term given potential L coinitiated in the mid-2000s has transformed these product markets [22,15]: two states into the second largest producing L areas for ethanol and sugar, after São Paulo. i i 1 L (q i , wi )  q i  k wk  k  L p  (1) C (i , j )  From 2005 to 2012, 40 new mills started  i operation in the region promoting the expansion where C is the production cost at location i; qi is of sugarcane area. Total area increased from the mills capacity at location i; wi is the vector of 341,000 ha in 2005 to more than 1.2 million ha input costs; wki is the cost of the kth input at in 2012. As a result, the region produced 6 location i; α and θ are fixed technical billion liters of ethanol in 2012 and represented coefficients for ethanol and for L co-products 16% of the total sugarcane area in Brazil. (such as sugar and electricity) respectively; and With the withdraw of government intervention, p(Li , j ) is the price for the L co-products the relationship between the industry and farmers has become an area of contention [18]. discounted by the transportation costs. A firm The ethanol industry has increasingly become will choose to locate a mill in county i over more vertical integrated, where the mills are another j if the cost associated with locating and i becoming more directly responsible for their operating in county i, C (q i , wi ) , is smaller own agricultural production. This strategy aims j to minimize negotiation costs and to guarantee than other locations j, C (q j , w j ) [22,15]: i.e. feedstock for the mill. In the crop year of i j i j 2011/2012, the industry produced 64% of the (2) C (q , wi)  C (q , w j) sugarcane it processed, the remaining 36% were acquired from nearby farmers [3]. This industrial Using the cost function in (1) and letting model has been implemented in the expansion to x*  qmi  i1 , we have that location decision by a the new frontier [11]. Industry-controlled mill can be represented by S * , a latent choice sugarcane production in Goiás and in Mato Grosso do Sul accounts for 76.8% and 73.4% of variable [22]: * j i the sugarcane crushed by the mills in each state,  (q j , w j )  C (q i , wi )  S i C respectively [3]. Industry participation in (3) x i'* wi  w j    j   i  x *i    i agricultural production is more intense in the new frontier because this region’s farmers do not where   is the expectation of a random term have considerable experience with growing associated with the uncertainty about the sugarcane. location decision. When x * can be exactly 3 LOCATION THEORY MODEL determined, then there is no uncertainty and  Location is a fundamental decision for ethanol will be zero. plants given that ethanol mills are supply1, S i*  0 oriented firms, where input acquisition  (4) S i 0, S *  0 dominates a firm’s cost structure [12]. This, in  i turn, means that mills are likely to be located



33

 



Proceedings of SEEP2015, 11-14 August 2015, Paisley where Si represents the observed location decision by the mill. When S i* is greater than or equal to zero it means the costs associated with other locations are higher than at location i and the firm will choose to locate in county i. Analogously, when S i* is less than zero then the production costs associated with other locations are lower than at county i, thus the firm will prefer not to locate in county i. Therefore the probability of a firm locating a mill in county i is [1]: PrS i  1  Pr S i*  0 Pr  i   xi'  (5)  1  F  xi'  Where F is a cumulative probability distribution function commonly modelled as a normal standard distribution or a logistic distribution. Researchers can only observe if a mill is in fact Figure 1. Counties with ethanol mills located or not in a county and the characteristics of given counties and firms. A set of eight explanatory variables is used to 4 METHODOLOGY model the location decision of a new ethanol In order to estimate how different factors impact plant in the study area. The variables employed the probability of a new mill being installed in a represent both the industrial and the agricultural certain county in the states of Mato Grosso do factors hypothesized to impact the location Sul and Goiás, we conducted a cross-sectional decision. Table 1 presents summary statistics for analysis of the factors impacting mill location. all the variables. This section of the paper describes the relevant The first variable Distance to Mill (dist.Mill) is factors and cross-sectional data collection. The the variable that accounts for the mean distance base year for the analysis was 2012. This year is from each county to the nearest mill. The mean the last year that Compania Nacional de distance was calculated using Euclidean Abastecimento (CONAB) published its official Distance and averaged by the entire county. list of existing mills. The choice of the states of Distance to Mill (dist.Mill) relates the presence Mato Grosso do Sul and Goiás is due the of an ethanol mill with the existence of significant growth of the sugarcane industry sugarcane fields and structure to plant sugarcane since 2005 in this area. These states presented [28]. the highest expansion rate during the period of The next two variables are proxies for distance 2005 to 2012, even without a tradition of to demand markets. Distance to São Paulo growing sugarcane [11]. (dist.SP) is included in the model because São The study area encompass 324 counties, 246 Paulo is the main market for ethanol (and for pertaining in the state of Goiás and 78 in the sugar) in Brazil. Variable Distance to Capitals state of Mato Grosso do Sul. Together both states have 60 mills, 36 in Goiás and 24 in Mato Grosso do Sul, in 33 and 21 counties, respectively. Data on the mills location was collected from [3]. In this dataset, geographic coordinates represent each mill. For our analysis, we do not focus on the exact location, but rather the county where each mill is located (Figure 1). The presence or absence of a mill in a county constitutes the dependent variable for our analysis.



 

 



34

Proceedings of SEEP2015, 11-14 August 2015, Paisley Table 1. Descriptive statistics and source of variable Variables

Unit

Mean

SD

Min

Max

Source

dist.Mills

Kilometers

67.248

47.825

7.344

321.772

CONAB, 2013

dist.SP

Kilometers

876.738

142.085

587.144

1248.094

CONAB, 2013

dist.Capitals

Kilometers

200.548

110.056

0.068

485.013

CONAB, 2013

ruralsyndicate

Dummy

0.602

0.490

0

1

CNA, 2014

labor

Person

13632.63

51824.14

573

748569

IBGE, 2010

roads

Meters/Hectares

0.899

0.514

0.119

3.233

MT, 2010

Zoning

Hectares

72494.78

115246

0

727006.8

Manzatto et al., 2009

agriculture

Hectares

27051.94

54707.26

0

488459.8

MMA, 2008

(dist.Capitals) measures the distance to the nearest state capital. These two variables also make use of the Euclidean Distance to determine the mean distance of the entire county to the destinations. It is important to note that previous studies used county’s centroid to present variable related to distance. However, given the variety of county’s areas and shapes in the study area (see Figure 1), the use of a mean distance for the entire county may be more insightful. Presence of rural syndicate (ruralsyndicate), human capital (labor) and road density (roads) are variables related with a county’s infrastructure. Presence of a rural syndicate (ruralsyndicate) is an indicator variable assigning a value 1 if a syndicate is present in the county and 0 otherwise. This variable provides is an indication of organization of the agricultural sector in the county. For this study is worth knowing that the syndicates can mobilize farmers and landowners, and provide economic and legal information to its members [23]. Human capital (labor) is the population aged between 18 and 69 in each county, providing a proxy for the potential labor force pool as reported in the last Demographic Census collected in 2010. Road density (roads) is the total length of paved roads divided by the county area. Roads are the main transportation mode for the sugarcane industry, both to carry sugarcane to the mill and to ship ethanol to demand markets. The last two variables are related with agricultural production decisions. Variable Sugarcane Agroecological Zoning (Zoning) is the area defined as suitable for sugarcane expansion by the Brazilian Government [26]. Following the fast expansion of the sugarcane industry in the beginning of the 2000s, the Brazilian government issued the Sugarcane Agroecological Zoning [26]. The Zoning 35

mapped the entire country to identify areas suitable for sugarcane expansion. The criteria used were: areas with slope smaller than 12 degrees, facilitating harvest mechanization; areas that had adequate soil; areas that allow for good development of sugarcane without irrigation; areas outside of the Amazon and the Pantanal biomes; and areas without native vegetation [26]. Goiás and in Mato Grosso do Sul are the states with largest areas classified as suitable for sugarcane expansion [26]. The other agricultural production variable is area under agriculture in 2002 (agriculture). This area was classified by the Conservation and Sustainable Use of Diversity Biological Program, a project under the Ministry of Environment of Brazil. This program classified land cover/land use throughout the Cerrado biome for the year 2002 [29]. It is noteworthy that we could not use sugarcane area as a variable in our model due to endogeneity problems. Sugarcane ethanol mills are a driving force for sugarcane production, thus giving rise to endogeneity. However, the use of agriculture area in 2002 does no cause this problem because sugarcane had a small share of all agriculture land in 2002 for the study area. Additionally, during the installation of an ethanol mill, sugarcane fields are planted two years before the plant became on-line. Therefore, the use of agriculture data from 2002 is not affected by the mills that were built during the expansion from 2005 to 2012. 4.1 Empirical model The empirical model takes into account certain location factors that would influence a mills decision in locating in a particular county:

S

i

   1dist.Mill   2 dist.SP

  3 dist.Capitals   4 ruralsyndicate   5labor   6 roads

Proceedings of SEEP2015, 11-14 August 2015, Paisley receive a value of 1, otherwise a value of 0 is assigned in the matrix. The spatial weights matrix W is then row standardized to make the (7) sum of each row in the matrix equal to 1, to help facilitate estimation of the model [30].

  7 Zoning   8 agricultur e   where ε is an error term with mean zero and IID. The model represented by equation (7) is estimated using a probit and a spatial autoregressive probit (SAR probit) given the binary nature of the dependent variable. Even though probit models have been used to model this type of location decision [22], a standard probit model does not account for the presence of spatial dependence between observations, which violates the assumption of independent observations [30]. Two spatial test statistics were calculated to confirm the potential presence of spatial dependence (Table 2). Moran’s I test for global pattern against randomization [31], while Geary’s C test for clustering [32]. Results displayed in Table 2 indicate that ethanol mills are clustered and that its spatial distribution is not due to chance.

Table 2. Moran’s I and Geary’s C statistics for ethanol mills Statistic Moran’s I Geary’s C

z-score 7.317 8.763

p-value 0.000 0.000

To test for spatial dependence in the standard probit model, Moran’s I was used to test for spatial dependence in the fitted regression residuals. Given the potential presence of spatial dependence, a standard probit model and SAR probit model are estimated. The SAR probit model is estimated to capture the spatial dependence present in the data. The latent SAR probit can be defined as:

S

*

5 RESULTS AND DISCUSSION The empirical model was estimated as a probit and as a SAR probit. Table 3 presents the estimation results for each model. After estimating the standard probit model, the fitted residuals were tested for spatial dependence using Moran’s I. The results indicate that the fitted residuals exhibit positive spatial dependence. The Moran’s I z-score was 4.3217 with a p-value 0.00001, indicating the presence of clustering over space in the fitted residuals. Table 2. Results of probit and SAR probit models Probit

SAR Probit

Variable

Coefficient

P>z

Coefficient

P>z

constant dist.Mills dist.SP dist.Capitals ruralsyndicate labor roads Zoning agriculture ρ

0.31547 -0.0896*** 0.00066 0.00343* 0.74387** 0.00000 -0.8404* 0.000005* 0.000009**

0.818 0.000 0.700 0.051 0.023 0.916 0.051 0.051 0.005

1.69 -0.114*** -0.00159 0.0058** 0.586* -0.0000 -0.427 0.000004* 0.00001** -0.462***

0.296 0.000 0.384 0.021 0.071 0.434 0.345 0.047 0.011 0.000

N 324 log likelihood -48.983 Mcfadden's R2 0.598 *** p500°C and steam pressure Combustion plant consists typically of: (up to 130bar) and provide better efficiencies in  Furnace/boiler  Heat recovery/steam generation  Steam engine/turbine with generator (power generation plant) 2.4.2 Biomass Gasification Gasification is a thermochemical process in which a carbonaceous fuel is converted to a combustible gas known as syngas, consisting of H2, CO, CH4, CO2, H2O, N2, higher hydrocarbons and impurities (e.g. tars, NH3, H2S and HCl) [19]. The process occurs when a controlled amount of oxidant (pure O2, air, steam) is reacted at high temperatures with available carbon in a fuel within a gasifier. Gasification converts biomass to a gas, which can then be utilised in advanced power generation systems such as fuel cells thus achieving higher electrical efficiencies compared to combustion based technologies. For this reason, gasification is considered the enabling technology for modern biomass use [23,24]. Furthermore, it offers greater flexibility in terms of applications to electricity, heat, transport fuels and chemicals. Gasification plants typically consists of:  Gasifier  Syngas cleaning units (engine/turbine requirements)  Gas engine/turbine with generator (power generation plant  Heat recovery/steam generation  Steam engine/turbine with generator (combined cycle plant) 3 SYSTEM SELECTION A good number of research have been conducted on working characteristics and performance of different CSP plant in different scenarios [2529]. Peterseim. J. H et.al [3] examined 17 124

electricity and heat production [30]. Solar tower system can operate with Direct Steam Generation (DSG) or Molten Salt for storage system in terms of power generation. DSG is particularly preferable for its higher efficiency, on the other hand molten salt enables power plant to produce electricity during insufficient DNI. Solar tower with molten salt is also and commercially available from different suppliers. Among 17 different combinations which had been studied previously [3,31], solar tower (ST) with direct steam generation (DSG) as primary CSP working fluid combining with biomass gasification gave the highest peak net efficiency of 33.2% followed by the combination of solar tower, molten salt (primary CSP working fluid) and gasification with optimum net efficiency 32.9%. Both systems are able to produce 540°C temperature at 130bar steam pressure. On the other hand at 525°C and 120bar steam pressure ST/DSG/biomass combustion system can provide 33.0% of pick efficiency followed by ST/molten salt/ biomass combustion of 32.8% efficiency. From the above information it appears that biomass gasification gives marginally higher efficiency comparing with combustion system when it merge with CSP. Within the CSP, molten salt as the working fluid is slightly less efficient than DSG. In terms of heat storage, usually molten salt may be best in present time for solar tower technology. On the same research it was found the economically the internal rate of return of DSG with combustion and gasification system is 10.8% and 10.9% respectively in comparison to molten salt with combustion and gasification both 10.5%. The payback period of the first case is 9.7 and 9.6 and the second case gives 10.2. The reason behind the better economic

Proceedings of SEEP2015, 11-14 August 2015, Paisley LCOE of PT-biomass hybrid system could be performance of DSG than molten salt is the more useful in understanding the suitability of capital expenditure of setting up a large storage this system for electricity and heat generation. facilities for molten thermal energy storage (TES) system. 4. COMPARATIVE ANALYSIS CSP/biomass hybridization can lower the capital 3.2 System 2: Linear Fresnel: Biomass cost by sharing the plant equipment such as Linear Fresnel is also an option for hybridization steam turbine, condenser and auxiliary with biomass resource and this systems has also equipment [1-6]. The following presents been investigated in various research [31, 32]. technical comparisons between different CSP Although LF systems is capable of obtaining and biomass technologies. from 400°C to 500 temperature at steam pressure from 90bar to 110bar which is less than 4. 1 Comparison of CSP Technologies ST technology, however no such power plant Table 1 shows that LF have better opportunities had been found which combines linear Fresnel for large scale power plant development in terms with molten salt for heat storage. At 500°C of land use. However, there are very few such temperature and 110bar steam pressure LF with type of reference power plant had been DSG as primary working fluid can provide net developed because of less favourable technical plant efficacy of 32.5% when it combines with features of LF in comparison with its closed biomass combustion system [3,32]. technically similar system PT collector. Among all CSP biomass hybrid system, LF use Table 1: Comparison of different CSP to give the best economic performance. The technology [38] same system can give an IRR of 11.5% with only 8.6 years of payback period. The research System Peak Solar Annual Land also indicates that Fresnel technology offers to Solar to Use much lower investment cost in comparison to Electricity Electricity m²/ other two CSP technology. Conversion Conversion MWh 3.3 System 3: Parabolic Trough: Biomass Efficiency Efficiency Parabolic Trough (PT) technology hybridized Solar 23 -27% 15-17% 8-12 with biomass is most mature system among all Tower of the hybrid technologies as there is one such Linear 18-22% 8-10% 4-6 plant is currently operating in Spain. It had been Fresnel found that PT with DSG in combination with Parabolic 21-25% 15-16% 6-8 biomass combustion system at temperature Trough 450°C and 100bar steam pressure can obtain It was found from other researches that in pick net efficiency of 31.5% [3]. On the other comparison with PT, LF requires 35% smaller hand PT with molten salt at 525°C and 120bar solar field due to smaller row-to-row distance can give the efficiency of 32.7%. If the biomass [35-38].However, it has higher heat loss due to technology adopts gasification, the same its receiver design. Parabolic trough vacuum combination with PT and molten salt can receiver has much lower heat losses than the provide slightly more efficient system of 32.8% atmospheric Fresnel receiver leaving this and able to obtain temperature of 540°C at technology less suitable for large scale heat 130bar steam temperature. It indicates clearly generation. that gasification has higher conversion efficiency Moreover LF observes higher optical losses it is although not very significant [33, 34]. The caused by horizontally placed collectors which economic scenario is not however, as observes higher cosine losses. The cosine losses competitive as other two CSP technologies. PT, generally occurs if the surface is not normal to DSG and biomass combustion will see 8.9% of the sun, the solar irradiance falling on it will be IRR on investment with 14.6 years of payback reduced by the cosine of the angle between the time. Other two combinations will give a little surface normal and a central ray from the sun better IRR which is 9.0% and 9.1% respectively. [36,39]. The shading of a linier Fresnel to The payback period is also marginally better adjacent collector array further reduce optical which is 14.4 years and 14.3 years. No LCOE efficiency. The cosine loss and shading effect had been presented in this particular research. carouses supplying significantly less thermal 125

Proceedings of SEEP2015, 11-14 August 2015, Paisley energy to the power block, especially in the early morning and late afternoon which causes lower dumping rate of thermal energy. However at mid-day with high irradiation, LF is well capable of produce thermal energy which exceeds the power block capacity causing higher upper dumping as shown in Figure 4. To optimize these problems the operating time for linear Fresnel system reduces which increases the costs per kWh.

Figure 5: Hourly solar power production (Psol,el) on a day in July (a) and January (b) [35]

Figure 4: Dumping effect of parabolic trough and linear Fresnel [39] In case of PT and ST, few more research have been carried out to evaluate the performance of each systems [35, 40]. Simulation studies have shown that solar tower performs well in heat generation which allows better cycle efficiency [35]. Figure 5 shows the performance of ST and PT in four different systems in a given day in July and January to understand the performance characteristics in summer and winter time. Systems which have been considered in the model are Solar Rankine Cycle Parabolic Trough Collector (SRC--PTC), Solar Rankine Cycle Solar Tower (SRC_ST), Integrated Solar Combined Cycle Parabolic Trough Collector (ISCC_PTC), Integrated Solar Combined Cycle Solar Tower (ISCC_ST).

126

Figure 6. Hourly solar-to-electric efficiency (ɳsol-el) on a day in July (a) and January (b) [35] The simulation results in Figure 5 show that, in summer time both systems of PT performs better than ST systems. However, parabolic trough energy generation reduces dramatically in winter due to cosine effects and incident angle modifier effects and heat losses.

Proceedings of SEEP2015, 11-14 August 2015, Paisley optimum efficiency. Maximum efficiency An ST performs through-out the year giving spectrum is in between 700K (427°C) to 750K superior yearly solar to energy conversion (477°C). The obtainable maximum efficiency is efficiency. In Figure 6, the efficiency curve of better in solar tower where it offers around 65% both ST and PT are presented. in comparison to 50% efficiency of parabolic Values of ɳsol–el as high as 25% are obtained by trough. The flat plate solar concentrators are the solar tower plants in winter time (Fig. 6b), when least in producing heat and thus less efficient in low ambient temperatures make the condensing CHP generation. pressure fall, thus increasing the The capital costs for the solar field and receiver steam/bottoming cycle efficiency. The solar-tosystem are a larger percentage of the total costs electric efficiency of the PTC plants is strongly in solar tower systems, while the thermal energy affected by the cosine effect: ɳsol-el, whose values storage and power block costs are a smaller are lower than 10% in the central hours of a percentage [3]. As shown in table 1, the area January day, increases up to 23% (SRC) or 25% used to generate per MWh for ST is relatively (ISCC) in July. higher than parabolic trough and significantly Pitz Paal et.al [38] compared different CSP higher than LF and PT, it is apparently clear that technologies from where he presented a ST draws higher capital cost in comparison to correlation between temperature vs efficiency of other two. However, according to International each system. The correlation provides an Renewable Energy Agency report in 2012 there understanding the maximum efficiency on is no CSP power plants using PT and LF are different state of temperatures of each using thermal storage system, which means technology. The efficiency is measured as: those plant only can generate electricity during ɳ max = ɳ th, Carnot × ɳ Absorber (1) day time. Therefore, solar tower can potentially Assuming the obtained absorber temperature is lower the lavalized cost of energy (LCOE) by equal to process temperature. increasing the capacity factor using thermal energy storage system. T Absorber = T Process (2) Figure 7 shows that at higher temperature a Stirling dish gives higher efficiency followed by solar tower. Solar tower performs best between around 1000K (727°C) to 1300K (1027°C) which gives a fare range of options for heat and power generation.

4.2 Biomass Technology Comparison A comparison of gasification, combustion, pyrolysis and pressurised gasification and gas turbine combined cycle, IGCC for power generation was found that the feed expenditure in the combustion systems is the highest of the systems at any capacity which leads to a low system efficiencies shown in Figure 8 [41].

Figure 7: Temperature vs Efficiency curve of CSP system [38]

Figure 8: Comparison of efficiencies for biomass to electricity systems. [41]

In comparison to that the parabolic trough gives a smaller window for CHP generation with

This high feedstock expenditure is countered by low capital expenditure as a result of the low total plant costs shown in Figure 9.

127

Proceedings of SEEP2015, 11-14 August 2015, Paisley 6. CONCLUSION Hybrid CSP and biomass power plants are interesting option for future dispatchable renewable electricity generation. The challenges are the moderate capacity factors or high TES costs, the necessity to build a large biomass collection structure, the volatility of the biomass price and low feed-in tariffs. The hybridization of these technologies increases power plant capacity factors (when compared to a solar only) and reduces biomass consumption (when compared to a biomass only power plant). Figure 9: Comparison of total plant costs for REFERENCES biomass to electricity systems. [41] [1] Peterseim JH, White S, Tadros A, Hellwig U. Low capital payback costs along with low overheads and maintenance costs and relatively lower labour costs are also the advantages of combustion system. Both low capital costs and low labour requirements are the key drivers of various well established power plants using biomass combustion technology. It appears from the study that despite lower system efficiency of biomass combustion, this technology is widely adopted and well proven in the market for power generation due to its economic competitiveness over other biomass systems. 5. DISCUSSION It appears that Solar Tower (ST) is the best possible CSP technology for CHP generation hybrid system. Figure 7 shows that the effective working temperature range is very limited for flat plate solar concentrators. PT efficiency decreases dramatically after 750K (477°C). ST gives relatively better working temperature range over PT and LF. However, as ST is not as proven technology as Parabolic Trough (PT) due to its relatively higher land use and complex technical operations, PT may be the next best option for hybridization. Higher optical and heat losses of linear Fresnel (LF) may not make it due the best option for hybridization. Biomass technology selection is heavily depended on availability of biomass resources, capital and operating cost. Deployment of biomass plant should consider a good availability of biomass resources or the plant may end up with a high operating cost. Regardless the efficiency of biomass systems different research shows that among all biomass technology, combustion system is proved to be most economically proven technology for biomass to electricity conversion.

128

Concentrating solar power hybrid plants Enabling cost effective synergies. Renew Energy; vol.67: pp178–85, 2014. [2] Zhang HL, Baeyens J, Degreve J, Caceres G. Concentrated solar power plants: review and design methodology. Renew Sustain Energy Review, vol.22, 2013. [3] Peterseim. J. H, Hellwig. U, Tadros. A, White.S, Hybridisation optimization of concentrating solar thermal and biomass power generation facilities, Science direct Solar Energy vol.99, pp. 203–214, 2014. [4] Srinivas. T, Reddy.B.V, Hybrid solar -biomass power plant without energy storage, Reference: CSITE22, Appeared in: Case Studies in Thermal Engineering, 23 January 2014. [5] Nixon JD, Dey PK, Davies P a. The feasibility of hybrid solar-biomass power plants in India. Energy. vol.46, pp. 541–54, 2012; [6] Peterseim JH, White S, Tadros A, Hellwig U. Concentrated solar power hybrid plants, which technologies are best suited for hybridisation? Renew Energy; vol.57, pp.520–32, 2013 [7] Cot. A, Amettler. A, Vall-Llovera J, Aguilo J, Arque JM.Termosolar Borges: A Thermosolar Hybrid Plant with Biomass. Third Int Symp Energy from Biomass Waste 2010. [8] Zhang HL, Baeyens J, Degreve J, Caceres G. Concentrated solar power plants: review and design methodology. Renewable and Sustainable Energy Review. vol.22:pp.466-481, 2013; [9] Zhu G, Wendelin T, Wagner MJ, Kutscher C. History, current state, and future of linear Fresnel concentrating solar collectors. Solar Energy vol. 1032014 pp.639–652, 2014 [10] Baharoon DA, Rahman HA, Omar WZW, Fadhl SO. Historical development of concentrating solar power technologies to

Proceedings of SEEP2015, 11-14 August 2015, Paisley [24] Doherty,W.,Reynolds, A.Kennedy, D. generate clean electricity efficiently – A review. Process Simulation of Biomass Gasification Renewable Sustainable Energy Review. 2015. Integrated with a Solid Oxide Fuel Cell Stack. [11] Greenpeace International, SolarPACES and Journal of Power Sources; vol.277, pp. 292-303. ESTELA. Concentrating solar power outlook 2015 2009 – why renewable energy is hot. [25] Isabel Llorente Garcı , Jose´ Luis A ´ [12] Valenzuela L. STE Plants with parabolic lvarez, Daniel Blanco, Performance model for trough collectors. 2012. parabolic trough solar thermal power plants with [13] Morin, G., Mertins, M., Kirchberger, J., thermal storage: Comparison to operating plant Selig, M., 2011. Super NOVA – construction, data, Solar Energy 85 (2011) pp. 2443–2460 control & performance of steam superheating [26] Hussain.C.M.I, Muppala.S, Chowdhury. linear Fresnel collector. In: Presented at the 17th N.H, Adnan. A, Cost Analysis of Concentrated Solar PACES International Symposium, Solar Power Plant with Thermal Energy Storage Granada, Spain. System in Bangladesh, International Journal of [14] Morin, G., Dersch, J., Platzer, W., Eck, M., Enhanced Research in Science Technology & Haberle, A., 2012. Comparison of linear Fresnel Engineering, ISSN: 2319-7463, Vol. 2 Issue 4, and parabolic trough collector power plants. April-2013, pp: 79-87 Solar Energy 86 [27] Andrea Giostri, Marco Binotti, Marco [15] A. Heimsath, F. Cuevas, A. Hofer, P. Nitz, Astolfi, Paolo Silva, Ennio Macchi, Giampaolo W.J. Platzer, Linear Fresnel collector receiver: Manzolini,Comparison of different solar plants heat loss and temperatures, Energy Procedia based on parabolic trough technology, Solar vol.49, pp.386 – 397, 2014. Energy 86 (2012) pp. 1208–1221 [16] Munoz, J, Martinez-Val, J.M, Ramos, A. [28] Coates J, Solar T. SRP and Tessera Solar Thermal regimes in solar-thermal linear Announce 1.5MW Maricopa Solar Plant with collectors. Solar Energy vol. 85, pp.857–870. Stirling Energy Systems SunCatcher TM 2011 Technology in Peoria, Arizona 2009. [17] Pino, F.J., Caro, R., Rosa, F., Guerra, J., [29] Ortega JI, Burgaleta JI, Téllez FM. Central 2012. Experimental validation of an optical and Receiver System Solar Power Plant Using thermal model of a linear Fresnel collector Molten Salt as Heat Transfer Fluid. J Sol Energy system. Applied Thermal Engineering, 1–9 Eng 2008; vol: 130/024501 [18] European Academices Science Advisory [30] J. H. Peterseim, A. Tadros, S. White, U. Council (EASAC). Concentrating solar power: Hellwig, and F. Klostermann, “Concentrated its potential contribution to a sustainable energy solar power / Energy from Waste hybrid plants future. Available at: 〈http://www.easac.eu/file creating synergies,” in SolarPACES Conference, admin/Reports/ Easac_CSP _Web-Final.pdf〉; 2012. 2011 [31] Peterseim JH, Tadros a., White S, Hellwig [19] Wayne Doherty, Modelling of Biomass U, Landler J, Galang K. Solar tower biomass Gasification Integrated with a Solid Oxide Fuel hybrid plants - Maximizing plant performance. Cell System, PhD Thesis Submitted in School of Energy Procedia vol.49, pp.1197–206, 2013 Mechanical and Design Engineering, Dublin [32] Itskhokine D, Lancereau Q, Benmarraze M, Institute of Technology Bichotte M, Dubost L. ScienceDirect LFR500 [20] Basu P. Biomass gasification and pyrolysis: project : getting to high temperature while practical design and theory. Brulington, MA: maintaining the cost-competitive advantage of Academic; 2010. Linear Fresnel Reflectors 2015. [21] Nussbaumer VT. verenum Overview on [33] Peterseim JH, Tadros A, Hellwig U, White Technologies for Biomass Combustion and S. Increasing the efficiency of parabolic trough Emission Levels of Particulate Matter. 2010. plants using thermal oil through external [22] IEA Bioenergy (Task 32), Biomass superheating with biomass. Energy Convers Combustion and Co-firing: An Overview Manag, vol. 77, pp.784–93. 2014 [23] Kirkels AF, Verbong GPJ. Biomass [34] N.S. Suresh, N.C. Thirumalai 1, Badri S. gasification: Still promising? A 30-year global Rao 2, M.A. Ramaswamy, Methodology for overview. Renewable and Sustainable Energy sizing the solar field for parabolic trough Reviews 2011; 15: 471. technology with thermal storage and

129

Proceedings of SEEP2015, 11-14 August 2015, Paisley hybridization, Solar Energy vol. 110, pp.247– 259, 2014 [35] G. Franchini, A. Perdichizzi, S. Ravelli , G. Barigozzi, A comparative study between parabolic trough and solar tower technologies in Solar Rankine Cycle and Integrated Solar Combined Cycle plants, Solar Energy, vol. 98 pp. 302–314, 2013 [36] Morin G, Dersch J, Platzer W, Eck M, Häberle A. Comparison of Linear Fresnel and Parabolic Trough Collector power plants. Sol Energy; vol. 86, pp.1–12, 2012. [37] Nixon JD, Dey PK, Davies P a. Which is the best solar thermal collection technology for electricity generation in north-west India? Evaluation of options using the analytical hierarchy process. Energy; vol.35, pp. 5230– 402, 2010 [38] Pitz-paal R. Parabolic Trough, Linear Fresnel, PowerTower A Technology Comparison. [39] Pitz-paal R, Concentrators L. Line Concentrators for Power Generation : Parabolic Trough and Linear Fresnel. [40] Sargent L. Assessment of Parabolic Trough and Power Tower Solar Technology Cost and Performance Forecasts. Rep No NREL/SR-55034440 2003:47 [41] Bridgwater A. V, Tof A. J, Brammer J.G. A techno-economic comparison of power production by biomass fast pyrolysis with gasification and combustion. vol. 6. 2002.

130


THE ROLE OF BIOENERGY IN GERMANY WITHIN LONG-TERM ENERGY SCENARIOS N. Szarka1, M. Eichhorn2, R. Kittler1, A. Bezama2 and D. Thrän1,2 1. Deutsches Biomasseforschungszentrum, Leipzig, Germany; email: [email protected], [email protected], [email protected] 2. Helmholtz Centre for Environmental Research, Leipzig, Germany; email: [email protected], [email protected], [email protected] Abstract Defining the long-term evolution of the German energy sector towards reaching its development goals by 2050 has been the subject of a series of studies carried out by governmental, industrial and independent interest groups. These studies are actually the basis for evaluating the role of bioenergy in the national energy system in the long term. However, a critical review and comparison of such studies is necessary for a more appropriate political and scientific discussion. Goal of this paper was to review the most relevant studies dealing with long-term energy scenarios, and to analyse and compare their frame conditions and expectations about the energy sector development in Germany and the role of bioenergy by 2050. The results identify the use of heterogeneous sets of key questions, normative goals, and main driving forces as major reasons for the resulting wide range of bioenergy development frameworks. Furthermore, the studies display a higher uncertainty level for characterizing and assessing the bioenergy sector. Considering this we carried out a structured comparison within the results of the different analyzed studies. With this it is expected to support political and scientific discussions about the long-term development of the energy sector and the role of bioenergy in Germany. Keywords: bioenergy, scenario analysis, policy making, decision support 1 INTRODUCTION Germany is aiming to develop and implement an efficient and environmentally sound energy system, characterised by competitive energy prices while also maintaining high living standards and economic prosperity [1-8]. In order to assess possible development options in the energy field to achieve the set goals, a series of energy scenarios have been commissioned independently by the German Government, by environmental groups and by different stakeholders of the energy sector. Energy scenarios can support the development of policy goals by evaluating a broad range of future options. This is the case of explorative scenarios. Another type of scenarios, target scenarios, are implemented in political decisionmaking processes for analysing how a set of goals can be reached or what is the level of achievements at a certain point in time. Any given scenario can be therefore characterized in terms of its main questions and goals, as well as of the methods used and the necessary detail degree of the aspects to be analysed. Because of

131

this it is difficult to understand in depth and interpret directly the results derived from the existing scenarios in order to determine the role and relevance of the bioenergy. For this reason, the present work reviews a total of eight studies carried out in Germany, in which the role of bioenergy in the national energy system has been described, analysed, compared and explained. Goal of this work is to provide with a better understanding of the possible future options of bioenergy within the energy system in Germany, by showing available ranges of the results, and thus providing the appropriate transparency via explanation of resulting differences and understanding the context behind the results and to provide with suggestions for a comparison of energy scenarios. 2

METHODOLOGY

2.1 Scenario selection In the present paper, 18 scenarios from eight studies [9-16] with a political and scientific relevance and acceptance were reviewed. The studies were selected according to their time frame -until the year 2050-, geographical scale

Proceedings of SEEP2015, 11-14 August 2015, Paisley -focus on Germany-, scientific credibility of the 3.1 Domestic biomass potential Figure 1 summarises the results for the indicator applied analytical tools and assumptions, the Domestic biomass potential. To avoid the available comparable information about the transfer of unsustainable effects to other regions, bioenergy sector as well as to account for some scenarios do not allow for an import of different background motivations -studies biomass (e.g., those of Greenpeace [16]). In originating from public bodies, research contrast, BMWi [14] and WWF [15] scenarios institutions, non-governmental institutions and consider biomass imports, for the reason that the energy sector itself were included. within Europe its sustainable production and 2.2 Scenario analysis with indicators trade is assumed to be resolved. As observed, the A comparative analysis of the scenarios has been assumptions on the amount of domestic biomass performed via the selection, definition, potential in 2050 range from 250 PJ (SRU [12]) quantification and explanation of four to 1.760 PJ (BMWi [14]). bioenergy-related indicators, taking into consideration the availability of high quality data. One of the key aspects for the development of the bioenergy sector is the availability of biomass resources which supply the future energy demands. Furthermore, since Germany is aiming at independency in terms of biomass import, the potential from domestic sources should be considered. For this reason the indicator Domestic biomass potential was chosen to differentiate the amount of available national biomass within the studied scenarios. Moreover, since a main objective of this study is to show and analyse the role of biomass within Figure 1. Domestic biomass potential and the different final energy sectors (i.e. power, import in PJ in 2050 heat, transport), the indicator Share of biomass 3.2 Share of biomass in the final energy in the final energy consumption was selected. consumption Among individual sectors only the transport This indicator compares the percentage of could be shown - due to the lack of sufficient bioenergy in the final energy consumption of data available - represented by the indicator one international study and four national studies Biofuels in the transport sector. Finally, the (see Figure 2). All scenarios foresee an increased indicator Allocation of biomass among final share of bioenergy, between 9% and 25% (2010 energy sectors was selected to depict the future level: 8%) in the final energy consumption by use options of biomass among power, heat and 2050, except of UBA (ca. 5%) [11]. The reason transport sector. of this can lie on the different assumptions of the The resulted indicators with their assumptions UBA scenarios: (i) the biomass potential is and the study framework were analysed and a limited to residues and no import is allowed, (ii) discussion is provided for users of scenarios. the electricity generation from biomass is among 3 COMPARISON OF THE BIOENERGY the most expensive and thus its application is SECTOR WITHIN THE SELECTED limited to peak production or balancing power, LONG-TERM SCENARIOS and (iii) it is assumed that biomass will be of The characteristics of the analysed scenarios, more economic value in the field of material use, with their geographical scale, scope, scenario i.e. bio-based products and platform chemicals. types (explorative or target) and names, model All target or explorative scenarios of the BMWi applied, principal goal and main results are [14], WWF [15] and the DLR Lead studies [13] summarized in Table 1. In the following the reach a fraction of bioenergy in the final energy selected indicators are described and their consumption well above 20%. differences among the scenarios are shown in figures and highlighted in description.

132

Proceedings of SEEP2015, 11-14 August 2015, Paisley Table 2 Characterisation of the reviewed scenarios

Germany Electricity Generation

World Entire Energy System

Scale & Scope

Study

Abbreviations

Scenario Type

Goals/Targets  Support of understanding of global developments and the world’s energy supply, use and needs. Most cost-effective technologies:  to stabilise global CO2 emissions (ACT)  reduce CO2 emission by 50% (BLUE) Technically and ecologically feasible electricity system integration options to reach 100% RE in the German electricity generation sector. Technical and ecological feasibility to reach 100% RE in the German electricity generation sector with a demand of: a=500 TWh target/b=700 TWh target

Shell 2011 [9]

(1) Scramble (2) Blueprint

Shell_Scramble Shell_Blueprint

BAU Explorative

IEA 2008 [10]

(1) Baseline Scenario (2) ACT Scenarios (3) BLUE Map Scenarios (1) Regions Network (2) International-Großtechnik (3) Lokal-Autark (1) Self-supply (2.1) Net self-supply and exchange (2.2) Max. 15% import EU (3) Max. 15% import EU & North Africa (1) Scenario 2011 A (2) Scenario 2011 B (3) Scenario 2011 C (4) Scenario 2011 THG95

IEA_Base IEA_ACT IEA_BLUE UBA_Energy_Target

BAU Target* Target** Target

SRU_1a; SRU_1b SRU_2.1a; SRU_2.1b SRU_2.2a; SRU_2.2b SRU_3.a; SRU_3.b

Target Target Target Target

DLR_Leadstudy_A DLR_Leadstudy_B DLR_Leadstudy_C DLR_Leadstudy_THG95

Target

 Options to reach the German Energy Concept Goals explicitly focussing on the transport sector & e-mobility

BMWi 2010 [14]

(1) Reference Scenario (2) IA-IVA (3) IB-IVB (I-IV variation in nuclear phase-out) (A/B variation in retrofit cost)

BWWI_Ref BMWI_IA-IVA BMWI_IB-IVB

BAU Target Target

WWF 2009 [15]

(1) Reference Scenario with CCS (2) Reference Scenario without CCS (3) Innovation Scenario with CCS (4) Innovation Scenario without CCS (5) Modell Deutschland (1) Reference Scenario (2) Greenpeace Scenario

WWF_Ref_CCS WWF_Ref_NoCCS WWF_Inno_CCS WWF_Inno_NoCCS

Explorative* Explorative** Target

 Reduction of GHG emissions of 40% by 2020 and 85% by 2050  Increase the share of RE in the gross final energy consumption: 18% by 2020 and 50% by 2050  Influence of the life time of the German nuclear plants  Technical and economic feasibility to reduce  CO2 emissions by 95%  Influence of CCS technology

Greenpeace_Ref Greenpeace_Tar

BAU Target

UBA 2010 [11] SRU 2011 [12]

DLR 2012 [13]

Germany Entire Energy System

Scenario names

Greenpeace 2009 [16]

Technical and economic feasibility to:  reduce CO2 emissions by 90%  reach 90% RE in final energy consumption  reach 1005 RE in electricity generation

Model examples

 WEM - World Energy Model a  MARKAL and TIMES models for individual countries  Energy System Model SimEE b  Energy System Model REMix c

 VECTOR21-Vehicle c Technologies Scenario c  Energy System Model REMix b  Energy System Model SimEE  European Electricity Market d Model DIME  PANTA RHEI environmentale economic modelling



Based on data from the German Federal Environment Agency (UBA) and Lead Study 2008

Notes: *Low, **High. Sources: aInternational Energy Agency, bFraunhofer IWES, cDLR- German Aerospace Centre, dEWI-Institute of Energy Economics, eGWS- Gesellschaft für Wirtschaftliche Strukturforschung mbH

133

Proceedings of SEEP2015, 11-14 August 2015, Paisley of biofuels in the Greenpeace scenarios [16] can be explained by the expected significant drop in the amount of passenger cars due to an enhanced availability of the public transport, a shift of commercial road transport to rail and a switch of private and commercial transport to e-mobility. Within this context it is discussed that biofuels will be used for transport purposes where the substitution of fossil sources displays the most difficulties (i.e., heavy vehicles). The aviation sector is not included in the Greenpeace [16] analysis. Figure 2. Share of biomass in final energy consumption in 2050 3.3 Biofuels in the transport sector Figure 3 presents the resulting amount of biofuels used in 2050 for transportation purposes which lies between 1 and 15 GJ/capita.

3.4 Allocation of biomass among final energy sectors Figure 4 displays the composition of biomass utilisation among the three major end-use sectors, calculated according to the biomass used in GJ per capita. According to Figure 4 it can be highlighted that despite the current discussions about the future role of biomass, all scenarios foresee that biomass will still be present in all three end-use sectors by 2050. Moreover, in the scenarios which assume a high share of emobility (Greenpeace), the share of biomass in the transport sector remains low. The low use of biomass in the heat sector in the WWF study [15] can be explained by the assumption of very high energy efficiencies in the building sector.

Figure 3. Biofuels in the transport sector in GJ/capita in 2050 All five national and international studies expect an increased utilisation of biofuels, and it is assumed that second-generation biofuels will be produced at a large scale and penetrate the market by 2050. The BMWi [14] target scenarios foresee a share of approximately 85% of biofuels in the transport sector. The WWF [15] innovation scenarios indicate that except for the aviation sector, all mineral oil products in the transport sector will be replaced by biofuels accompanied by natural gas, liquid gas and also hydrogen. The Lead study [13] and reference scenarios of WWF [15] and BMWi [14] do not forecast multiple growth rates in biofuels, rather assuming that a higher share in e-mobility reduces the demand for biofuels. The reference scenarios indicate that fossil fuels remain dominant in the transport sector. The low share 134

Figure 4: Allocation of biomass among final energy sectors in 2050 4

DISCUSSION, CONCLUSION AND SUGGESTION FOR POLICY MAKERS The results indicate that all studies follow key questions and goals, (e.g. reaching a share of renewables or reducing greenhouse gas emissions), considering different main driving forces (e.g. energy supply provided by fossil fuels, level of environmental stress or energy intensity). They are constructed (i) as strategies to reach the set goals or (ii) to show the

Proceedings of SEEP2015, 11-14 August 2015, Paisley consequences of certain (energy-) policy biomass is present in all energy sectors and all decisions. The studies display similarities in the scenarios in 2050. socio-economic framework conditions and in the  The share of biofuels in all studies increases in assumptions and results regarding general relation to the status quo up to 4-14%. aspects of the energy sector in Germany (e.g. Therefore, the biofuel sector can be considered final energy demand and considered technology as important future consumer of biomass. portfolio). However, significant differences in According to the results of this study, selected terms of bioenergy-related parameters can be suggestions for scenario users or commissioners identified among the studies: in terms of the are presented below, for an early introduction of conversion technology development, the measurements to compare scenario results: biomass potential and its use (i.e. energy –  Usually, scenarios are developed by using power, heat or transport- or material pathway), different simulation models. If a comparison of as well as of the utilized methods and data. different scenarios and therewith different Furthermore, the studies differ in terms of their model results is foreseen, it is important to main goals, driving research questions, the type develop similar framing conditions like overall of scenarios (explorative or target), and their goal, socioeconomic conditions and data. specific focus on selected subjects (e.g. electro  Introduce standards to methodology, e.g. the mobility), providing therefore a significant range traceability from input to results should be of results. In addition to the differences, a much described, results should transparently be higher uncertainty level characterises the presented and visualized bioenergy sector, and it is described with less  Standards for definitions in the detail compared with the whole energy system. energy/bioenergy field should be introduced, As an example, Figure 5 shows the statistical and the scenarios should use the same differences within the national studies for the nomenclature and definition contribution of biomass versus renewables in the  Introducing data quality and minimum final energy sector. It can be seen that there is a requirements: a consistent data set of the much higher range in the case of biomass (a related sectors should be used. Since data are factor of 10) than in the case of the renewables constantly changing, a central place of data (factor 2), which explains that the differences collection, storage and preparation should be between the results of the studies and therefore ensured, as well as the development and the uncertainties in the biomass sector are much management of scenario databases higher than in the renewables in general.  Central parameters – minimum parameters – should be developed and included in all models and scenarios for a better comparison  A combination of different models and submodels could be considered to provide with an answer to complex questions  Considering the growing importance of bioeconomy and the combined use of bioenergy with other renewable energy sources, the scenarios and models should be flexible enough to include these sectors, too.  Considering that the type of scenario (explorative or target) has a strong influence on Figure 5: Biomass and renewables in the how the results can be used and interpreted by final energy consumption users, it should clearly be defined, at which Out of the applied analysis some selected stage of the decision making process should conclusions could be drawn: scenarios be applied. At the same time it  The range of the sustainable domestic biomass should be taken into account that goal potential lies between 350-1700 PJ. scenarios do not include a big spectrum of all possible development options, they only  The share of biomass in the final energy include a small range of subjects in focus. consumption lies between 5-28%. Even though this may seem a wide range, it shows that 135

Proceedings of SEEP2015, 11-14 August 2015, Paisley [8] Lechtenböhmer S, Samadi S. Blown by the  Frame conditions and even goals (e.g. purpose wind. Replacing nuclear power in German of bioenergy is GHG reduction, system electricity generation. Environmental Science & stability, rural development) are constantly Policy. 2013;25:234-41. changing. Therefore it is important to [9] BV, S.I., Signals & Signposts Energy constantly update the dataset, frame conditions Scenarios to 2050 – An era of volatile and questions of the study. transitions, 2011: The Hague. p. 40.  The results of a scenario should always be [10] IEA, Energy Technology Perspectives 2008 interpreted considering the goal, type and used – Scenarios and Strategies to 2050, International data. Energy Agency IEA/OECD, 2008.  There is also an option to combine explorative [11] UBA, Energy target 2050: 100% renewable and goal scenarios. For example first to have electricity supply, Federal Environment Agency, an explorative scenario to explore the options Fraunhofer-Institut für Windenergie und for one sector and define goals. These defined Energiesystemtechnik, 2010. goals can then be assessed in a goals scenario [12] German Advisory Council on the (how to reach the goals) and then again the Environment, Pathways towards a 100% resulted (smaller focus) sector explored in an explorative scenario. renewable electricity system, German Advisory Council on the Environment (SRU), 2011. REFERENCES [13] Nitsch, J., Pregger, T., Naegler, T., Heide, [1] VDU Technologiezentrum & Fraunhofer ISI. D., Luca de Tena, D., Trieb, F., Scholz, Y., Zwischenergebnis 3 – Forschungs- und TechnoNienhaus, K., Gerhardt, N., Sterner, M., Trost, logieperspektiven 2030, BMBF-ForesightT., von Oehsen, A., Schwinn, R., Pape, C., Zyklus II. Suchphase 2012-2014. Im Auftrag des Bundesministerium für Bildung und Forschung, Hahn, H., Wickert, M., Wenzel, B., 2014. Langfristszenarien und Strategien für den [2] Ferroukhi R, Nagpal D, Lopez-Peña A, Ausbau der erneuerbaren Energien in Hodges T, Mohtar RH, Daher B, et al. Deutschland bei Berücksichtigung der Renewable energy in the water, energy & food Entwicklung in Europa und global - Leitstudie nexus. IRENA – International Renewable 2011, Deutsches Zentrum für Luft- und Energy Agency; 2015. Raumfahrt (DLR), Institut für Technische [3] Bundesministerium für Wirtschaft und Technologie, Bundesministerium für Umwelt, NaturThermodynamik, Abt. Systemanalyse und schutz und Reaktorsicherheit. Energiekonzept Technikbewertung Fraunhofer Institut für für eine umweltschonende, zuverlässige und Windenergie und Energiesystemtechnik (IWES), bezahlbare Energieversorgung. 2010. Ingenieurbüro für neue Energien (IFNE), 2012. [4] Bundesministerium für Wirtschaft und Tech[14] Schlesinger, M., D. Lindenberger, and C. nologie. Forschung für eine umweltschonende, Lutz, Energieszenarien für ein Energiekonzept zuverlässige und bezahlbare Energieversorgung. der Bundesregierung, Prognos AG, Das 6. Energieforschungsprogramm der Bundes-regierung. In: Technologie BfWu, Energiewirtschaftliches Institut an der editor. 2011. Universität zu Köln (EWI), Gesellschaft für [5] Arbeitsgemeinschaft Energiebilanzen e. V. Wirtschaftliche Strukturforschung mbH (GWS), (AGEB). Energieverbrauch in Deutschland im 2010. Jahr 2011. AGEB, 2012. [15] Institut für angewandte Ökologie Öko[6] Bundesministerium für Umwelt, NaturInstitut e.V., Prognos AG, Blueprint Germany schutz und Reaktorsicherheit. Erneuerbare Energien in Zahlen. Berlin; 2012. A strategy for a climate-safe 2050, A study [7] Winter CJ. SUNRISE – A caesura after commisioned by WWF Deutschland, 2009. nuclear has gone: Energy in Germany (and in [16] Barzantny, K., S. Achner, and S. Vomberg, other potentially de-nuclearizing countries). Klimaschutz: Plan B 2050 - Energiekonzept für International Journal of Hydrogen Energy. Deutschland, 2009. 2012;37(9):7317-42.

136


THERMAL SUSCEPTIBILITY OF SOLID RECOVERED FUEL (SRF) Ljiljana Medic Pejic1, Laura Rubio Arrieta2, Nieves Fernández Añez3, Javier Garcia Torrent4 1. Department of Energy and Fuels, UPM Technical University of Madrid, C/ Alenza 4, 28003, Madrid, Spain, email: [email protected] 2. Laboratorio de Combustibles y Petroquímica UPM Technical University of Madrid, C/ Eric Kandel, 1, Parque Científico y Tecnológico de la UPM, 28906 Getafe (Madrid), Spain, email: [email protected] 3. Department of Energy and Fuels, UPM Technical University of Madrid, C/ Alenza 4, 28003, Madrid, Spain, email: [email protected] 4. Department of Energy and Fuels, UPM Technical University of Madrid, Alenza 4, 28003, Spain. Laboratorio Oficial Madariaga, LOM (UPM Technical University of Madrid, Spain) C/ Eric Kandel, 1 – (TECNOGETAFE), Parque Científico y Tecnológico de la UPM, 28906 Getafe (Madrid), Spain; email: [email protected] Abstract One of the main environmental problems in all the developed countries is the disposal of the high amount of waste that is generated every day. Energy industries are looking for implementing new Solid Recovery Fuels (SRFs) from municipal, industrial wastes and sewage sludge for sustainable development. The waste management has developed ways to produce secondary fuels as SRFs with reliable qualities, which are used successfully regarding economic and environmental aspects. The aim of this study is the research on the flammability properties of a new solid and homogeneous SRFs obtained from different solid wastes “streams” from the standpoint of fuel recovery. The quality of the fuel is achieved through the combination of different conventional mechanical processes and a thermal treatment that allows obtaining a fuel with better energetic properties for use in a greater number of combustion plants. The results of this study would contribute to the development of adequate strategies improving safety in the storage and handling of SRFs in the coming years. The analysed SRF samples were selected from different locations in Spain and taken during different seasons. Keywords: Solid urban waste, Fuel recovery, Characterization of flammability of SRF 1 INTRODUCTION Sustainable development strategies often consist on the suitable processing of municipal, commercial or industrial wastes. In Europe, it has been necessary to devise new strategies to tackle that promising the reduction and recycling of waste, and also its energy use. Therefore, several Council Directives imposes appropriate specific regulatory obligations in order to reduction in the total amount of biodegradable waste going to landfill in the EU Member States [1, 2], encourages the use of sewage sludge in agriculture, etc. Furthermore, the Council of European Union [4, 5, and 6] has approved recently an energy plan, which includes an objective of 20% of renewable energy in the final energy consumption in Europe by 2020. The use of solid wastes as the energy source can contribute to this challenging target; and especially in Spain. 137

In order to reduce waste generation, to minimize waste transport distance, to promote material or energy recovery and to ban the landfilling of untreated waste or waste that cannot be treated any further it has established the following priority order [3] in solid waste prevention and management: 1) Prevention means measures taken before a substance, material or product has become waste, that reduce the quantity of waste, including through the reuse of products or the extension of the life span of products, the adverse impacts of the generated waste on the environment and human health; or the content of harmful substances in materials and products 2) Re-use means any operation by which products or components that are not waste are used again for the same purpose for which they were conceived

Proceedings of SEEP2015, 11-14 August 2015, Paisley 3) Recycling means any recovery operations by which waste materials are reprocessed into products, materials or substances whether for the original or other purposes. It includes the reprocessing of organic material but does not include energy recovery and the reprocessing into materials that are to be used as fuels or for backfilling operations 4) Recovery means any operation the principal result of which is waste serving a useful purpose by replacing other materials which would otherwise have been used to fulfil a particular function, or waste being prepared to fulfil that Figure 1 Result of Thermogravimetric function, in the plant or in the wider economy. analysis with some characteristic 5) Disposal means any operation which is not parameters [7] recovery even where the operation has as a When analyzing the numerous variables which secondary consequence the reclamation of can be measured in this kind of tests, the substances or energy. temperature to combustion induction (TI) and 2 EXPERIMENTAL WORK the temperature of maximum loss of weight Laboratory tests were carried out to characterize (TDM) have been found as the most significant the flammability of solid recovered fuels in parameters. The onset reaction temperature order to define handling and storage process shows the point where the oxidation reaction requirement. stars to accelerate. The temperature for the Ten samples were taken from different locations maximum loss of weight is a clear indication of in Spain during different seasons from different the reactivity of the product, frequently applied waste streams whose composition varies to solid fuels; it represents the yield of volatile organic-inorganic matter have been subject to matter due to pyrolisis; the higher the thermogravimetric analysis (TG analysis is temperature, the lower the product reactivity. performed in an oxidative atmosphere, air or As a general rule, when higher temperatures are oxygen) used to determine a material’s thermal observed heating reactions will be produced at stability and to differential scanning calorimetry. higher temperatures, that is to say, the product is less prone to self-ignition [8] Differential Scanning Calorimetry (DSC) tests 3 METHODOLOGY are intended to measure heat exchanges 3.1 Characterisation of flammability of produced in the sample by comparison with a SRF reference sample. The test procedure consists of By means of thermogravimetric techniques it is heating the SRF sample in the temperature possible to analyses the thermal susceptibility of ranges from 30 ºC to 550 ºC; at a heating rate 20 solid recovered fuels. K/min and an isothermal period of 10 minutes Thermogravimetric test (TG) consists of was applied when reaching 550 ºC. recording the sample weight changes (weight loss) when the sample is subjected to temperature-programmed heating. Temperature ranges start at 30 ºC and go to 800 ºC at the high end, at a heating rate 5 K/min and an isothermal period of 10 minutes was applied when reaching 800 ºC [7]. Sample weight is graphically represented versus temperature to evaluate results (see Figure 1). Thermograms of different samples are an identity imprint of the materials and can be compared through the recorded values for some variables. 138

Proceedings of SEEP2015, 11-14 August 2015, Paisley Both, activation energy and characteristic Figure 2 Differential scanning calorimetry temperature can be applied to classify the selfresults with the indication of some ignition risk of solid recovered fuels. [10] characteristic parameters [7] The minimum onset temperature of exothermic reaction (initial temperature, TIE), the maximum temperature reached during exothermic reaction (final temperature, TFE) and the onset temperature for rapid exothermic reaction (change of slope temperature, TCP) have been verified as the best characterization parameters for different fuels. The first part of the graph with decreasing values, from the starting point to the TIE, represents an endothermic reaction where the moisture loss occurs. At the TIE, the exothermic reaction starts with a slow reaction that takes place from this point to the TCP. At the TCP the reaction accelerates and becomes a quick exothermic reaction finishing at the TFE. Susceptibility evaluation: activation energy (Ea). Following a conventional thermogravimetric test, the apparent activation energy of the sample is calculated at the point of maximum weight loss by means of J.W. Cummings’ mathematical [9] model applied to a set of points around that of maximum weight loss in a suitable representation of the recorded test points. “Apparent activation energy” can be obtained from the slope of least-squares line fitted to the selected test data. Susceptibility evaluation: oxidation temperature (Tc). Thermogravimetric technique is applied to the study of self-ignition susceptibility by modifying test conditions when air is replaced by oxygen (see Figure 3).

Figure 3 Comparisons between TG and TGO (TG +O2) [7] As a consequence of this oxidant contribution, sample behavior can be very different during testing and a step or sudden loss of weight, associated to a rapid combustion and produced at a characteristic temperature in every substance is observed.

139

4 RESULTS Results from the thermogravimetric analysis and differential scanning calorimetry are shown in Table 1. Table 1. TG and DSC results

L1 L2 L3 L4 L5 L6 L7 L8 L9 L10

TG TDM TIC 319.67 237.91 277.00 231.47 275.83 232.44 278.83 241.92 263.67 233.74 278.00 234.88 269.50 237.25 279.00 229.95 277.25 237.77 279.33 236.14

TIE 82.25 91.97 94.00 88.26 89.39 87.77 100.01 80.46 94.80 83.31

DSC TFE 319.01 348.01 324.35 340.01 288.68 344.68 309.01 328.68 308.68 298.34

TCP 245.60 232.96 235.68 241.84 222.89 251.00 224.80 241.30 234.22 250.46

The numerical results from the TG and DSC analyses show the similarities among the samples. The combustion process starts around 230-240ºC for all studied sample. The temperature of maximum weight loss (TDM) corresponds to the first peak of the differential curve of the TG. Looking at the DSC results, they also show similarities among the samples. The variation of the three parameters obtained from this test is small, and the exothermic reaction starts at less than 100ºC. The change of slope temperature (TCP) represents the temperature at which the quick exothermic reaction starts, and it is also almost similar for all the samples studied. The main difference among samples can be observed by studying the self-ignition risk with the Tcharact vs Ea plot, as shown in Figure 4.

Proceedings of SEEP2015, 11-14 August 2015, Paisley The organic origin of the samples can be observed thanks to the presence of two peaks in the differential curve of the thermogravimetric tests. Only one of the seven parameters obtained from these tests has shown the differences existing among the samples. The activation energy has been used for a long time in coal studies because it gives a good way of classifying these materials looking at them self-ignition risks, but it has not this influence nor in biomass samples, neither in these samples. Looking only at this parameter, all the samples studied are located in the high risk area. Figure 4. Self-ignition risk plot (Tcharact vs.Ea) However, the characteristic temperature shows a This graph shows the similarity among 8 of the differentiation of two samples due to the 10 samples, and how they are located next to the presence of components that hinder the biomasses previously studied. As shown in other oxidation of the samples. studies, the activation energy is not an important parameter because all the samples present less REFERENCES than 79 kJ/mol. However, looking at the [1] EEC Landfill Directive (1991/31/EC) characteristic temperature, two samples are [2] Directive 86/278/EEC on sewage sludge located in the low risk area. [3] Directive 2008/98/EC EU Waste Framework These two samples present a different behaviour Directive. Official Journal L 312, 22/11/2008 P. in the combustion with oxygen. They do not has 0003 – 0030. a sudden loss of weight, but a first stage of [4] EC (1999). Council Directive 1999/31/EC of gradual loss and a final smaller stage of loss, as 26 April 1999 on the landfill of waste. OJ L 182, shown in Figure 5. 16.7.1999, p. 1–19. [5] EC (2011): Commission Decision of 18 November 2011 establishing rules and calculation methods for verifying compliance with the targets set in Article 11(2) of Directive 2008/98/EC of the European Parliament and the Council [6] EC, 2012: Member States’ reporting to the Commission according to Council Directive 1999/31 of 26 April 1999 Landfill Directive and Commission Decision 2000/738/EC concerning a questionnaire for Member States reports on the implementation of Directive 1999/31/EC on the landfill of waste. [7] J. García Torrent, L. Medic Pejic, E. Querol Figure 5. Example of TGO of four samples. Aragón, A self-combustion characterisation The oxidation resistance of some components of index based in thermogravimetric and these two samples causes this process. differential scanning calorimetry techniques, in: V International Symposium on Hazards, Prevention and Mitigation of Industrial 5 CONCLUSIONS Explosions, Krakow, Poland, 2004. Ten samples of organic waste that cannot be [8] U. Krause. Fires in silos: hazards, treated any further have been studied, and their prevention, and firefighting. John Wiley & Sons, similarities have been observed by different 2009. parameters obtained through thermogravimetric and differential scanning calorimetric analyses.

140

Proceedings of SEEP2015, 11-14 August 2015, Paisley [9] J. W. Cumming, "Reactivity assessment of coals via a weighted mean activation energy," Fuel, Vol. 63, pp. 1436-1440, 1984. [10] Medic Pejic, L.; Fernandez Añez, N.; García Torrent, J.; Ramírez Gómez, A. Determination of spontaneous combustion of thermally dried sewage sludge. Journal of Loss Prevention in the Process Industries. Vol. 36 pp. 352-357, 2015

141


REVEALING EFFECT OF TORREFACTION ON BIOMASS PYROLYSIS B. Ru, S.R. Wang*, G.X. Dai and L. Zhang State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, China; email: [email protected] Abstract The influence mechanism of torrefaction on biomass pyrolysis was studied in detail. 13C CP/MAS NMR was employed to characterize the evolution of functional groups during torrefaction, and found that hemicellulose was extensively deacetylated and depolymerized; the crystalline structure of cellulose was destroyed; the aryl ether linkages and propyl side branches in lignin was cleaved. A distributed activation energy model with three linear combined Gaussian functions was introduced to analyze the devolatilization kinetics of biomass. The results showed that the mean activation energies for three parallel reactions remained unchanged after torrefaction, while the weighing factors proved that their contribution to devolatilization largely changed. The final distribution of pyrolytic products was detected by Py-GC/MS. Acetic acid production significantly declined after torrefaction; the yields of HMF and levoglucosan decreased because of the depolymerization of carbohydrates during torrefaction; due to the dissociation of propyl branch and the demethylation of methoxyl in lignin, pyrolysis of torrefied biomass yielded more guaiacol and catechol and less phenols with C4-propyl groups. Keywords: Torrefaction, pyrolysis, 13C CP/MAS NMR, 3G-DAEM 1 INTRODUCTION Torrefaction could improve the quality and grindability of raw biomass, making it easier to be fed into the reactor. Therefore, torrefaction was received great attention in many biomass process, such as combustion, gasification and so forth. However, there was limited research focused on its application in biomass pyrolysis. The crude bio-oil from pyrolysis contained many oxygenated chemicals, such as acids, resulting in its low heating value and strong corrosivity. Torrefaction could remove oxygen in biomass by dehydration and deacetylation at low temperature (200-300°C) [1]. Boateng and Mullen [2] found the main products yielded in torrefaction were water and acetic acid, as well as a few of acetol and levoglucosan. The chemical structure of biomass changed considerably during torrefaction. However, the fundamental research on the evolution of functional groups in biomass torrefaction was inadequate. Melkior et al. [3] proposed 13C solid-state cross-polarization/magic angle spinning (CP/MAS) NMR was a powerful tool to investigate the evolution rules of typical functional groups during torrefaction. Compared with raw biomass, pyrolysis behavior of torrefied biomass changed remarkably. Torrefaction improved the thermal stability and lowered the reactivity of biomass by removing the thermally unstable fragment. However, there 142

were very few kinetic studies on the pyrolysis of torrefied biomass. Broström et al. [4] reported a kinetic model containing three parallel reactions corresponding to the decomposition of three pseudo-components, and found that the activation energy for pyrolysis of each pseudocomponent rarely changed after torrefaction. In this study, the effect of torrefaction on the biomass structure and the subsequent pyrolysis behaviors were revealed in detail. 13C CP/MAS NMR was used to quantitatively characterize the evolution of chemical structure during biomass torrefaction. A distributed activation energy model with three linear combined Gaussian functions (3G-DAEM) was introduced to assess the devolatilization kinetics. The distributions of tar-derived components from pyrolysis of raw and torrefied biomass were discussed in detail by Py-GC/MS. 2 MATERIALS AND METHODS A typical softwood, Pinus bungeana, was select as the raw biomass for torrefaction. The sample was first ground and sieved into 0.18-0.25 mm, then dried at 60°C for several days. Torrefaction pretreatment was performed at a tube furnance with nitrogen purging, about 10 g sample were treated at required temperatures of 200°C, 225°C, 250°C, 275°C and 300°C for 0.5 h. The ultimate analysis was performed on a Micro Elemental Analyzer. The high-heating value was calculated according to the method proposed by

Proceedings of SEEP2015, 11-14 August 2015, Paisley torrefaction temperature increased. The removal Demirbaş [5]. Information about the functional of oxygen significantly improved the HHV of groups in raw and torrefied biomass were 13 biomass, as from original 18.0 kJ/mol to 20.1 quantitatively analyzed by C CP/MAS NMR. kJ/mol for the Torr-300°C biomass. And the The thermal analysis kinetics for pyrolysis of corresponding calculated energy yields were raw and torrefied biomass were studied on a 98.8%, 96.7%, 91.1%, 75.3% and 64.5%, thermogravimetric analyzer by non-isothermal respectively. Thus it was recommended that method. 2 mg sample was loaded and heated 250°C might be a promising torrefaction from 25°C to 800°C with a constant rate of 10 -1 temperature, at which the oxygen content was °C∙min . High purity nitrogen with a flow rate largely removed, and more than 90% energy of of 40 mL∙min-1 was used to provide inert raw biomass was preserved. atmosphere and sweep the released volatiles. Tar-derived pyrolytic products from raw and torrefied biomass were analyzed by Py-GC/MS. In each run, about 0.5 mg sample was pyrolyzed at 600°C for 10 s, the products were swept into GC/MS for analysis. The specific analysis method of GC/MS detection was described in our previous study [6]. 3


3.1 Structural characterization As the torrefaction temperature increased, the molar ratios of O/C and H/C in biomass declined (see Figure 1, raw biomass was assigned to Torr25°C). Their variation in 225-250°C was much less than that in 250-300°C. This phenomenon was also found in the variation of mass yields after torrefaction, they were 94.5%, 91.9%, 85.8%, 70.2% and 57.8%, respectively, for 200300°C. The contents of major components in raw biomass were 44.4%, 20.8%, 25.1% and 9.5% for cellulose, hemicellulose, lignin and extracts, respectively. Obviously, besides hemicellulose (79.2% mass left exclusive of hemicellulose), other components were also largely degraded during torrefaction. It was found that the variation of ultimate compositions could be well fitted by a linear function, indicating that the main reaction types occurred in different torrefaction temperatures were similar, but differed in the reaction extent. The slope of 1.58 for fitted equation was little smaller than 2, the value corresponded to pure dehydration reaction. This suggested that the main reaction during torrefaction was dehydration. And owing to the release of acids, CO2 and CO during torrefaction, the efficiency of deoxygenation was higher than the pure dehydration route (1.58 vs. 2). Nocquet et al. [1] found that the main products yielded from torrefaction of beech wood over 220-300°C included water, acetic acid, formaldehyde and CO2, and their productions were enhanced as the

143

Figure 1. The ultimate compositions and HHV of raw and torrefied biomass.

The evolution of typical functional groups in biomass was recorded by 13C CP/MAS NMR (see Figure 2). None of these detected groups disappeared during torrefaction. Gaussian profile functions were employed to fit the NMR spectra, and the content of carbon in each functional group was quantitatively calculated, followed by normalization by multiplying the corresponding mass yields. The results are listed in Table 1.

Figure 2. 13C CP/MAS NMR spectra.

The content of C=O in ester first declined and then increased during torrefaction (see Table 1). This was attributed to the thermal cleavage of ester linkages during torrefaction, and produced CO2 or CO; on the other hand, the formation of new ester bond induced by dehydration could occur under high temperature, this was probably related to the structural condensation of biomass at high temperature treatment. Guaiacyl-type (Gtype) phenylpropanoids were believed to be the major components in softwood lignin. From the carbon distribution in aromatic region (114.0153.1 ppm), it could be found that as torrefaction

Proceedings of SEEP2015, 11-14 August 2015, Paisley torrefaction of lignin and xylan could yield temperature increased, the amount of etherified methanol via demethoxylation. The G units declined, while that in non-etherified deacetylation of hemicellulose during form increased, indicating that the aryl ether torrefaction resulted in the decline of acetyl linkages were cleaved, and more free phenolic content in biomass (21.1 ppm). This group was hydroxyls were released. The increase of G1-H believed as the main source of acetic acid in bioillustrated that the propyl branch linked at C1 of oil [8]. benzene ring was dissociated during torrefaction. Due to the breakage of aryl ether bonds, Table 1. Carbon content in each functional especially of β-O-4, the amount of G5 and G6 group (g/100g raw biomass). (114.0 ppm) decreased. This reaction could yield Chemical Torrefaction temperature (°C) shift Assignment mono-phenols, such as guaiacol. The signals at 25 200 225 250 275 300 (ppm) C=O in ester 173.7 0.62 0.53 0.57 0.50 0.55 0.60 104.6 ppm and 100.6 ppm were assigned to C1 linkage G3, G4 in etherified 153.1 0.87 0.89 0.91 0.82 0.84 0.76 positions at β-1,4-glycosidic bonds in cellulose form G3,G4 in non147.0 1.19 1.26 1.29 1.30 1.35 1.28 and hemicellulose, respectively. The decline of etherified form 131.5 G1-H 1.15 1.19 1.20 1.19 1.25 1.26 these signals indicated that the two carbohydrate 114.0 G5,G6 2.51 2.45 2.49 2.47 2.40 2.16 104.6 C1 in cellulose 3.41 3.23 3.22 2.70 2.76 1.97 components explored depolymerization during 100.6 C1 in hemicellulose 1.68 1.71 1.66 1.78 0.95 0.56 torrefaction. In which hemicellulose degraded crystalline C4 in 88.4 2.03 1.69 1.46 1.13 1.14 0.98 cellulose C in β-O-4 of more extensively, as the signal at 100.6 ppm lignin and 83.3 3.11 2.94 2.85 2.54 1.95 1.27 amorphous C-4 in decreased much more than that at 104.6 ppm. cellulose C2, C3, C5 in Softwood hemicellulose mainly contained β-1,474.6 carbohydrates and 7.39 7.42 7.29 7.45 4.81 3.22 Cα in lignin glycosidicly linked hexoses, such as mannose C2, 3, 5 in 71.7 9.81 9.05 8.64 7.59 6.23 4.26 carbohydrates and glucose, as backbone; the depolymerization 64.7 C6 in Carbohydrates 2.11 1.91 1.82 1.44 1.50 1.02 61.8 C -OR 3.34 3.33 3.10 2.95 1.76 1.27 of softwood hemicellulose was similar as that of 55.3 methoxyl in lignin 2.31 2.25 2.09 2.03 1.42 1.20 46.3 aliphatic C 0.83 0.80 0.73 0.76 0.56 0.59 cellulose; the C1-cation originated from 36.4 aliphatic C 1.76 1.77 1.64 1.62 1.47 1.38 glycosidic bond breakage would be easily acetyl in 21.1 1.21 1.13 1.08 0.96 0.74 0.35 hemicellulose combined with C6-hydroxyl to form 3.2 Kinetic study levoglucosan [7]. The ordered crystal structure The thermogravimetric analysis indicated that of cellulose was destroyed, as shown in the the thermal stability of biomass was largely decline of signal at 88.4 ppm. The chemical shift enhanced after torrefaction. The initial at 83.3 ppm corresponded to the overlap of aryl decomposition temperatures, corresponding to ether and the amorphous carbon in cellulose. the mass loss of 5%, increased from the original Since the carbon in crystal region might be 247°C to 249°C, 253°C, 256°C, 272°C and transformed into amorphous form, the decline of 292°C, respectively, after torrefied at 200this signal was mainly ascribed to the cleavage 300°C. This was ascribed to the removal of of thermally unstable ether bond in lignin. thermally unstable fragments, such as side Torrefaction lowered the intensities of signals at branches and ether linkages, in biomass. The 74.6 ppm, 71.7 ppm and 64.7 ppm by half, in char yield was also enhanced from 16.8% for accordance with the reduction of C1 in pyrolysis of raw biomass to 28.4% for that of glycosidic bond. This further confirmed the Torr-300°C biomass. However, it was found that significant degradation of carbohydrates. From the normalized char yield only changed a little the variation tendency it could be observed that when the char yield multiplied by mass yield. the degradation reaction mainly occurred above For accurately evaluating the pyrolytic 250°C. The dissociation of propyl side chain in devolatilization behaviors for raw and torrefied lignin was further confirmed by the decline of biomass, 3G-DAEM based on first order three signals, 61.8 ppm (Cγ-OR), 46.3 ppm reaction assumption was introduced to study the (aliphatic Cγ) and 36.4 ppm (aliphatic Cβ). This effect of torrefaction on the kinetics of biomass was in agreement with the decrease of G1-H pyrolysis. According to previous studies [7-11], (131.5 ppm). Methoxyl (55.3 ppm) was also the three Gaussian function distributions dissociated during torrefaction. This group corresponded to three parallel reactions in mainly distributed in lignin, and the uronic acid pyrolysis of raw and torrefied biomass: R1: fragments, such as 4-O-methyl-glucouronic acid, parallel degradation reactions in pyrolysis of in hemicellulose might also contribute a small hemicellulose and lignin. R2: degradation amount. Nocquet et al. [1] also observed that β

γ

γ

β

144

Proceedings of SEEP2015, 11-14 August 2015, Paisley the non-linear fitting. The results showed that reaction of cellulose; R3, parallel condensation the correlation (R2) of 3G-DAEM to pyrolysis of reactions in pyrolysis of hemicellulose and all the six samples were higher than 0.999. lignin. The method was described as Equations n  A T (1) and (2). Weighing factor, wi, was introduced 1 α(T)   w i  exp (  i  e E/RT dT )fi (E)dE (1) 0 β 0 i to quantitatively assess the contribution of three  (E  E0,i ) 1 parallel reactions to devolatilization. The fi (E)  exp[ ] (2) 2 E2 ,i  E ,i 2 algorithm was Pattern research method based on

Figure 3. 3G-DAEM analysis for pyrolysis of raw and torrefied biomass.

It was interesting to find that the mean activation energies (E0) for the three parallel reactions remained unchanged after torrefaction (see Figure 3). This indicated that the chemical structure of biomass was essentially unchanged. The mean activation energy for R1 (E01) located at the lowest region. The corresponding standard deviation (σE1) became narrower as the torrefaction temperature increased. This was due to the dissociation of weak structural fragments in hemicellulose and lignin during torrefaction, and resulted in their reduction of parallel degradation reaction types in pyrolysis. Cellulose was the component that achieved the least invasion during torrefaction, whose decomposition reaction (R2), including E02 and σE2, remained almost unchanged after torrefaction. It was also found that the distributions of R1 and R2 became closer at severer torrefaction conditions. As for the high activation energy stage (R3), similar phenomenon as R1 could be observed, E03 changed little while the distribution width (σE3) became narrower. This was attributed to the elimination of complex side structural fragments, which simplified biomass structure and made it more uniform. Hence the random cross-linking and condensation reactions in pyrolysis of 145

hemicellulose and lignin were reduced, and the corresponding activation energy distribution was therefore narrowed.

Figure 4. Variation of weighing factors.

However, the values of weighing factors significantly changed after torrefaction (see Figure 4). The contribution of R1 to devolatilization (w1) decreased from 0.49 to 0.39 due to the releasing of volatiles in torrefaction. w2 slightly decreased from 0.31 to 0.30. Although cellulose explored depolymerization during torrefaction, its proportion in torrefied biomass might not changed a lot since the hemicellulose degraded more extensively. Thus the contribution of cellulose degradation to devolatilization changed slightly. While w3 corresponding to the condensation reaction leading to char formation largely increased from 0.20 to 0.30, in accordance with the increase of

Proceedings of SEEP2015, 11-14 August 2015, Paisley The main regions of variation were 25-200°C char yield in pyrolysis of torrefied biomass. The and 250-300°C (see Figure 1). variation tendency of weighing factors was similar with the change in structural composition.

Figure 5. Distribution of typical products from pyrolysis of raw and torrefied biomass.

3.3 Pyrolytic product distributions The distribution of pyrolytic products significantly changed after torrefaction (see Figure 5). The abundance of each product was calculated from the GC/MS detected peak area divided by the feedstock weight. For the three small molecular products, hydroxyacetone, glycolaldehyde and acetic acid, their abundances declined as the torrefaction temperature increased. Since hydroxyacetone and glycolaldehyde were mainly formed from the profound decomposition of sugar ring, their decrease was related to the degradation of carbohydrates during torrefaction. It had been proved that the rupture of glycosidic bond to form levoglucosan could occur at higher than 200°C [12]. The obvious decrease of acetic acid was ascribed to the deacetylation of hemicellulose during torrefaction. The evolution of sugar ring during pyrolysis mainly yielded furfural, 5-hydroxymethyl furfural (HMF) and levoglucosan. All of their abundances decreased along the elevated torrefaction temperature. The softwood hemicellulose contained hexoses as the major monosaccaride residues, thus furfural, mainly formed from the dehydration of pentoses, presented a lower abundance in the pyrolysis of softwood. From Figure 5b, it was observed that the yields of HMF and levoglucosan both exhibited slight increase after torrefaction at 225°C. This might be due to the destruction of crystalline region in cellulose, as evidenced in NMR characterization, and the amorphous structure was easier to decompose. This might also explain the slight decrease of E02 for pyrolysis of Torr-225°C biomass in 3G-DAEM study (from 184.4 kJ/mol to 182.4 kJ/mol). Basch and Lewin [13] also found that the activation energy for cellulose pyrolysis decreased linearly as the crystallinity declined. According to the above discussion, the main reactions of lignin in torrefaction were the 146

breakages of aryl ethers and propyl side branches, which significantly affected the resulting distribution of phenols. As shown in Figure 5c, all the four typical phenols with C4propyl side chain showed decrease tendencies due to the dissociation of propyl chain in torrefaction. As for the three mono-phenols without C4-propyl group (see Figure 5d), the abundance of guaiacol presented an initial decrease, followed by an increase. The main reaction to form guaiacol was the cleavage of βO-4 linkage, which could occur at low temperature. Domburg et al. [14] performed pyrolysis of a β-ether type model compound of softwood lignin, and found that a lot of guaiacol was formed by breakage of β-O-4 bond between 180-300°C; and about 30% ether bond broke at 200°C [15]. Therefore, the production of guaiacol from pyrolysis declined after torrefaction. However, as the torrefaction temperature exceeded 250°C, the abundance of guaiacol increased, in accordance with the significant decline of 2-methoxyl-4-propylphenol and 2-methoxyl-4-propenyl-phenol. The original structure in lignin related to the productions of these two chemicals lost propyl side chain during torrefaction, and tended to yield guaiacol. The production of syringol decreased after severe torrefaction, which was attributed to the dissociation of methoxyl during torrefaction, as shown in the above NMR analysis. A previous study also found that syringyl units in lignin structure could be transformed into guaiacyl units through demethoxylation in torrefaction [3]. The abundance of catechol also exhibited an increase for pyrolysis of high temperature torrefied biomass. This might be due to the transformation of methoxyl into hydroxyl through demethylation. Klein et al. [16] found that catechol and methane were yielded from pyrolysis of guaiacol at a low temperature of

Proceedings of SEEP2015, 11-14 August 2015, Paisley [6] S. R. Wang, X. J. Guo, T. Liang, et al., 350°C, and proposed that the reaction pathway Mechanism research on cellulose pyrolysis by was the demethylation of methoxyl in guaiacol. Py-GC/MS and subsequent density functional 4 CONCLUSION theory studies, Bioresource Technology, Vol. Torrefaction significantly changed the chemical 104, pp. 722-728, 2012. structure of biomass and its resulting pyrolysis [7] S. R. Wang, B. Ru, H. Z. Lin, et al., behaviors. Dehydration, deacetylation and Pyrolysis behaviors of four O-acetyl-preserved depolymerization were the main reactions of hemicelluloses isolated from hardwoods and carbohydrates occurred in torrefaction. The softwoods, Fuel, Vol. 150, pp. 243-251, 2015. ordered crystalline structure of cellulose was [8] S. R. Wang, B. Ru, G. X. Dai, et al., destroyed. The aryl ether and propyl chain in Pyrolysis mechanism study of minimally lignin were dissociated. 3G-DAEM analysis damaged hemicellulose polymers isolated from found that the activation energy for each parallel agricultural waste straw samples, Bioresourse reaction remained unchanged, while their Technology, Vol. 190, pp. 211-218, 2015. contributions to volatilization changed [9] S. R. Wang, B. Ru, H. Z. Lin, et al., remarkably. As for the resulting pyrolytic Pyrolysis behaviors of four lignin polymers products, the production of acetic acid was isolated from the same pine wood, Bioresource largely inhibited after torrefaction; the yield of Technology, Vol. 182, pp. 120-127, 2015. HMF and levoglucosan also declined; the [10] M. J. J. Antal, G. Varhegyi, Cellulose abundance of phenols without propyl chain was pyrolysis kinetics: the current state of enhanced and those with side chain decreased as knowledge, Industrial & Engineering Chemistry torrefaction temperature increased. Research, Vol. 34, pp. 703-717, 1995. ACKNOWLEDGEMENTS [11] J. Z. Zhang, T. J. Chen, J. L. Wu, et al., The authors are grateful for the financial support Multi-Gaussian-DAEM-reaction model for from the National Natural Science Foundation of thermal decompositions of cellulose, China (51276166), the National Basic Research hemicellulose and lignin: Comparison of N 2 Program of China (2013CB228101), the and CO 2 atmosphere, Bioresource Technology, National Science and Technology Supporting Vol. 166, pp. 87-95, 2014. Plan Through Contract (2015BAD15B06) and [12] F. Shafizadeh, R. H. Furneaux, T. G. the Program of Introducing Talents of Discipline Cochran, et al., Production of levoglucosan and to University (B08026). glucose from pyrolysis of cellulosic materials, Journal of Applied Polymer Science, Vol. 23, REFERENCES pp. 3525-3539, 1979. [1] T. Nocquet, C. Dupont, J.-M. Commandre, et [13] A. Basch, M. Lewin, The influence of fine al., Volatile species release during torrefaction structure on the pyrolysis of cellulose. I. of wood and its macromolecular constituents: Vacuum pyrolysis, Journal of Polymer Science: Part 1–Experimental study, Energy, Vol. 72, pp. Polymer Chemistry Edition, Vol. 11, pp. 3071180-187, 2014. 3093, 1973. [2] A. Boateng, C. Mullen, Fast pyrolysis of [14] G. Domburg, G. Rossinskaya, V. Sergseva, biomass thermally pretreated by torrefaction, Study of thermal stability of β-ether bonds in Journal of Analytical and Applied Pyrolysis, lignin and its models, in: 4th International Vol. 100, pp. 95-102, 2013. Conference on Thermal Analysis, Budapest, [3] T. Melkior, S. Jacob, G. Gerbaud, et al., 1974, pp. 221. NMR analysis of the transformation of wood [15] C. Amen-Chen, H. Pakdel, C. Roy, constituents by torrefaction, Fuel, Vol. 92, pp. Production of monomeric phenols by 271-280, 2012. thermochemical conversion of biomass: a [4] M. Broström, A. Nordin, L. Pommer, et al., review, Bioresourse Technology, Vol. 79, pp. Influence of torrefaction on the devolatilization 277-299, 2001. and oxidation kinetics of wood, Journal of [16] M. Klein, Lignin thermolysis pathways, in, Analytical and Applied Pyrolysis, Vol. 96, pp. Ph. D. Dissertation, Department of Chemical 100-109, 2012. Engineering, Massachussetts Institute of [5] A. Demirbaş, Calculation of higher heating Technology, 1981. Karagöz, S, 1981. values of biomass fuels, Fuel, Vol. 76, pp. 431434, 1997. 147


POTENTIAL ASSESSMENT TOOL OF BIOMASS ELECTRICITY GENERATION FOR THE STATE OF PARANÁ Priscila Alves 1, Susana Viana1, Luiz Carlos1, Alexandre Aoki 2 1 Priscila Alves , Susana Viana, Luiz Carlos , Electrical Energy Systems Department, University of Campinas 13083-852, Campinas, Sao Paulo, Brazil, [email protected] or [email protected] 2 Alexandre Aoki , Electrical Engineering Department, Federal University of Paraná, Curitiba, Paraná, Brazil

Abstract The growing demand for electricity, coupled with environmental laws, sustainability and economy, establishes the development and improvement of renewable generation technologies. The study of biomass usage for power generation has been growing over the years. Countries such as United States, Germany and Japan study and invest in this area because of the by-products that can be generated, such as fertilizers, gas, and chief among them, carbon credits. Brazil has great potential for generating electric power trough biomass because it has a large diversity of products that are suitable as fuel for electric energy biogeneration. Regarding this potential, some Brazilian states were analysed, and it was found that the southern region of the country has an optimal generation potential for biogas. The state of Paraná was analysed, it is considered the breadbasket of the country due to its grain crops and swine and bovine production. The potential for electric power generation through the waste of swine and cattle was analysed. Some existing tools to biogas calculation were studied and a new tool was proposed. Its main purpose was to be userfriendly to micro and small farmers, so they could easily calculate the potential for biogas and electricity generation, and carbon credits. The proposed tool allows the understanding of distributed generation to be available for small and medium-scale farmers. The results obtained with the proposed tool, which require only 3 information inputs, are compared to the ones from another tool, which requires 14 inputs, showing the developed tool accuracy. This demonstrates the ease of use and applicability for any User to determine his farm, site or enterprise generation potential. Keywords: Biomass, Renewable Energy, Carbon Credits 1

INTRODUCTION

The biomass and all renewable resources that come from organic matter - plant or animal have energy production as main aim [1]. They are one of the most used sources around the world, whether in solid form, as liquid biofuel or as biofuel gas. Some examples of biomass are: wood waste, bagasse, wood, charcoal, animal waste, alcohol and other primary energy sources. This source can be applied in various ways: heating, power generation, and as automotive fuel. A major application is as a source of electric power generation in isolated areas, rural areas, and in urban areas [2][3], using 148

agricultural residues, sanitary landfills or sewage treatment systems. Among the various fonts of renewable energy, biomass is regarded as a Phoenix reborn from its own ashes, due to the carbon cycle. CO2 emissions, released during the burning of biomass, can be considered virtually emissions free because this gas is reabsorbed in the next plant life cycle, through the process of photosynthesis. All dioxide of carbon generated by the biomass process can be reabsorbed by the environment, generating a low impact [4]. The waste generated by agribusiness and large urban centres are residues wasted, leaving their energetic potential unused. Investment in

Proceedings of SEEP2015, 11-14 August 2015, Paisley federal district. Brazil has a total of 211764292 development of renewable energy sources from bovines and 36743593 swines. Four states stand biomass serve as an ally for sustainability, social out for their large concentration of livestock, inclusion and the development of clean energy Paraná (PR), Santa Catarina (SC), Rio Grande and generation of carbon credits. Developed do Sul (RS) and Minas Gerais (MG). countries like the US, Germany, UK and Japan Paraná state will be a case study for it is one of

Figure 1 – Size of the Herd have been giving great importance to biomass for electricity generation [5]. Brazil is a country with diverse biomass sources, like biodiesel, biogas, ethanol and other sources [4]. Some studies have already been carried out to determine the potential for electric power generation in isolated areas in the northern region of the country [6]. With its multiplicity of livestock and crops, there is a great potential for electric power generation, and carbon credits [7]. Biomass study can have multiple branches, the main focus of this work is to find a simple computational tool to determine the potential for raw materials generation and their application (power generation, biogas and carbon credits), as it was detected that there was a lack of such a tool in the market. 2

the states with the largest herd, and the data on the state is easily available. 2.1 State of Paraná The state of Parana, is in the southern region of the country and has an area of approximately 199,554 km2, which corresponds to 2.3% of the total Brazilian area. The state is divided into 39 micro-regions, has can be seen in Figure 2, showing information on each micro-region. This data has been provided by INPARDES - Paraná Institute for Economic and Social Development [9].

BIOMASS AND THE BRAZIL

Brazil is a country consisting of 26 states and a federal district (DF). This country’s dominant features are farming and ranching. The focus in the production of biogas is due to the large number of livestock production. Brazil is considered the second largest holder of bovine and buffalo herds, the 3rd largest poultry producer, and the 4th in swine products [8]. Figure 1 shows the percentage of the size of the swine and bovine herds, at the 26 states and the 149

Figure 2 – State of Paraná


Figure 3 – Size of the Herd – Paraná State Data from herds is presented in Figure 3 and the micro-regions with the highest concentrations of bovine and swine are shown. In micro region 22 there is a swine predominance due to a slaughterhouse that is installed in that region. 3

4

BIOGAS

The biogas production occurs through the digestion process, where anaerobic bacteria (bacteria that live in the absence of air) are used in the process of breaking large molecules. This is a simple process that occurs naturally. Animal residues are the largest methane gas producers, and the swine and bovine waste have the greatest potential. Table 1 presents daily waste and methane production for these two species. The biogas calorific value varies between 5500 kcal/m3 and 7000 kcal/m3, this interval is related to the gas CH4 concentration variations.

Table 1- Daily Waste and Biogas Production Cattle Bovine Swine (50 kg)

Daily Waste Production [kg] 10 2,25

EXISTING TOOL

A tool developed by CENBIO [10] was studied, for it solves the problem that has been taken into consideration. The tool is developed on an Excel platform, with several fields to be filled. For a simple simulation this tool needs 14 information inputs. As the final simulation result it submits a report with information about the digester, the gas production, the biogas surplus burning, the carbon credits and the electricity generation. 5

CALCULATION

Using basic equations it is possible to determine the biogas volume, the electric power generation and the amount of carbon credits. Equation (1) allows for the biogas volume calculus using the information presented in Table 1. In (1), Nc is the number of animals, Pt, the daily production of waste, Qg, the amount of gas produced per residue mass unit. Vp 

Gas Volume [l/kg] 36 78

Nc.Pdr .Qg m³ [ ] 1000 day

(1)

Knowing the biogas volume, the electric power generation potential, Wg, can be determined by Equation 2. For this calculation the produced gas volume, Vp, is needed, alongside with c, that is the conversion factor from kcal/m³ to kWh, the generator efficiency n and the biogas calorific value, K. For this last parameter, usually the value of 6500 kcal/m³ is used.

150

Proceedings of SEEP2015, 11-14 August 2015, Paisley 6.2 CASE II - SMALL FARMERS (2)

Wg  Vp.K .c.n

For the calculation of carbon credits it’s necessary to determine the CH4 concentration, usually it varies from 55% to 75%. For the conversion of CH4 in carbon credits, Equation 3 is used, where the methane equivalent value is used to obtain the conversion to CO2. In this equation, 𝑃 is the methane concentration and 𝐶𝑀 is the conversion value for tons of carbon credits [11].

tCO2  Vp.P.CM

PARANÁ STATE CASE STUDIES

Two case studies applying both tools will be presented, considering a whole micro region and small farmers from Paraná. The application for these two cases intends to prove that the proposed tool is simple, easy to understand, and can be applied to any study case. 6.1 CASE I – MICRO REGION The results from the application of the two tools for the micro region 22 – Toledo, which has 1.617.549 swine heads and 379.813 bovine heads, are shown in Table 2. Note that the tool developed by CENBIO does not calculate the generated power without previous electricity consumption values assignment, which is difficult to do for a whole micro region. Hence, this value is missing in the table below. Table 2 – Comparison between CENBIO tool and the Proposed Tool for Case I Case I Power generation [kWh] Carbon Credits [tCO2/day]

Table 3 – Comparison between CENBIO tool and the Proposed Tool for Case II

(3)

This set of three equations determines, in a simple way, the potential of a small farm, city, state or country. 6

Here, both tools are applied to a small farm that has 1180 swine, and an average monthly energy consumption of 540 kWh. Table 3 shows the results obtained using the two tools. When compared with CENBIO tool, the proposed tool shows a satisfactory result, considering that the only information needed is the number of animals on the property.

CENBIO tool

Proposed tool

-

917.202,90

689,88

Case II Electric Power Generation [kWh] Carbon Credits [tCO2/day] 7

CENBIO tool

Proposed tool

470,350

451,580

0,282

0,256

RESULTS

The bioenergy atlas was used to validate the results. The municipalities that have higher potential for electric power generation are consistent with those presented in the atlas. The State of Paraná could generate 26.617.151,90 kWh of electricity through bovine waste use and 19,294,640.76 kWh from swine waste exploitation. The deviation for the amount of electric power generation calculated by the proposed tool, relative to CENBIO tool is of 4%, whereas for the carbon credits the difference is higher, approximately 8%. The potential for the state of Paraná has been assessed and Figure 4 and Figure 5 show the results for electric power generation and carbon credits for its 39 micro regions. 8

CONCLUSIONS

The developed tool proved to be satisfactory. The need for two to three information inputs, namely the number of animals and their average weight makes it simple to use. This methodology can be laid down in a brochure or booklet, explaining the processes

521,56

151

Proceedings of SEEP2015, 11-14 August 2015, Paisley Other studies may be related to this work, such within a digestion system, not requiring a as the impact of the electric power generated in computer, nor software to be used. It meets the rural distribution networks. The social impact, needs and the resources available for small and environmental development and new policies micro farmers in rural areas. involving renewable energy markets and isolated Compared to other tools, the errors are areas. considered permissible for it isn’t a refined tool.

Figure 4- Electricity generation – Paraná State

Figure 5 – Carbon Credit – Paraná State Figure 5 – Carbon Credit – State Parana The aim to be simple and understandable was acknowledge when this tool was taken into the field. People with basic education understood the principle and the biodigestion system operation, and were able to make the calculations without effort or the need for a machine (computer, calculator, etc.). 9

FUTURE WORK

The next steps of the project intend to engage the cost and a technical viability analysis. 152

ACKNOWLEDGEMENTS Researchers from the Electrical Engineering Department of UNICAMP, especially Susana Viana and Professor Luiz da Silva for their support and incentive, and also Professor Alexandre Aoki's from the Electrical Engineering Department of UFPR, for guiding the project.

REFERENCES

Proceedings of SEEP2015, 11-14 August 2015, Paisley http://www.usp.br/portalbiossistemas/?p=1579. [Accessed: 04-Dec-2014].

[1] MINISTÉRIO DO MEIO AMBIENTE, “Biomassa.” [Online]. Available: http://www.mma.gov.br/clima/energia/energiasrenovaveis/biomassa.

[11] ESO, “Greenhouse Gases.” [Online]. Available: http://www.eso.org.om/UserFiles/files/ESO_GH Gs_En.pdf.

[2] V. G. García and M. M. Bartolomé, “Rural electrification systems based on renewable energy: The social dimensions of an innovative technology,” Technol. Soc., vol. 32, no. 4, pp. 303–311, Nov. 2010. [3] L. A. H. Nogueira, Biodigestao: a alternativa energetica., 1 ed. São Paulo, 1986. [4] L. A. B. Cortez, E. E. S. Lara, and G. E. Olivares, Biomassa para energia, 2 Edicção. Campinas: 2011, 2008. [5] A. Evans, V. Strezov, and T. J. Evans, “Sustainability considerations for electricity generation from biomass,” Renew. Sustain. Energy Rev., vol. 14, no. 5, pp. 1419–1427, Jun. 2010. [6] A. A. Bacellar and B. R. P. Rocha, “Wood-fuel biomass from the Madeira River: A sustainable option for electricity production in the Amazon region,” Energy Policy, vol. 38, no. 9, pp. 5004–5012, Sep. 2010. [7] S. T. Coelho, M. B. Monteiro, and M. da R. Karniol, “Atlas de Bionergia do Brasil,” 2012. [Online]. Available: http://cenbio.iee.usp.br/download/atlasbiomassa 2012.pdf. [8] “MINISTÉRIO DA AGRICULTURA,” 2014. [Online]. Available: http://www.agricultura.gov.br/. [Accessed: 03Apr-2014]. [9] IPARDES, “INPARDES - Instituto Paranaense de Desenvolvimento Econômico e Social,” 2011, 2011. [Online]. Available: http://www.ipardes.gov.br/. [Accessed: 10-Oct2014]. [10] Sebrae, “Cartilha ensina como calcular a produção de biogás a partir de resíduos animais.” [Online]. Available:

153


FAST PYROLYSIS OF CEYLON TEA WASTE AND PRODUCT BIOCRUDEOIL CHARACTERISTICS Ramesh Soysa1, Yeon Seok Choi1,2*, Seock Joon Kim1,2 and Sang Kyu Choi2 1. Department of Environment and Energy Mechanical Engineering, Korea University of Science and Technology, Daejeon 305-350, Republic of Korea; email: [email protected] 2. Environmental and Energy Systems Research Division, Korea Institute of Machinery and Materials, Daejeon 305-343, Republic of Korea; email: [email protected] Abstract Biomass is the main practical source of renewable liquid fuel, that could replace fossil fuels in the immediate future and is carbon neutral. Biocrude-oil production through fast pyrolysis is a direct and sustainable way of producing liquid fuel. Biocrude-oil has high water content along with high O/C atomic ratios that generally attributes to a smaller heating value in comparison to that of conventional petroleum fuels. Abundantly found Ceylon Tea Waste can be utilized as a resourceful feedstock in the production of biocrude-oil through fast pyrolysis, as well as an effective waste disposal mechanism while producing a sustainable and useful biochar. In the present study, Ceylon Refuse Tea Waste was fast pyrolyzed in order to establish optimum mass and energy yields for biocrude-oil production. The product biocrude-oil results can be a platform for a novel sustainable biomass based, renewable energy model and an integrated waste to energy system for tea industry in Sri Lanka. The reaction temperature of 773 K converted the highest biocrude-oil energy yields. Keywords: Biocrude-oil, Fast pyrolysis, Tea Waste, Biochar 1 INTRODUCTION Biomass is one of the largest sources of energy in the world historically, and currently estimates to be 14% of world's energy consumption [1] as well as being carbon neutral, low in sulfur and biomass can be changed into secondary liquid fuels by thermochemical and biochemical techniques. Thermal conversion of biomass through fast pyrolysis in which biomass is rapidly heated to moderate temperatures followed by rapid quenching of the vapours is developed today as a replacement for fossil fuels in the near future with mass yields of 34-67 wt% [2]. woodcrops, agricultural and forestry residues, biodegradable components of municipal solid wastes (MSW) and commercial and industrial wastes are some of the main renewable and bioenergy resources available [3]. Agricultural waste to secondary liquid fuels have been extensively looked into as a solution to waste disposal with studies done on rice husk [4], palm oil waste [5], coffee ground [6]. Tea known as Camellia sinensis was first grown in Ceylon first 1867 and has been grown ever since, with annual Ceylon Tea production in 2014 reaching 338 mil. kg of black tea, out of which roughly 10% of the yield ending up as waste tea technically known as refused tea in tea factories [7]. At this rate the country produces

154

27,397 kg/day of Ceylon Refuse Tea (henceforth known as CRT) which is commonly fed back to the tea plantations as a soil enhancement [8], insecticide and insect repellent by converting to biochar through slow pyrolysis. Currently Dilmah Conservation and Sustainable Agriculture Centre pyrolyzes 10,000 kg of tea waste per annum into porous biochar with 75% reduced weight [8]. Biochar can also be produced as a byproduct of biocrude-oil productions. Pyrolysis of Turkish tea waste yielded 29.6% at reaction temperature of 773 K of biocrude-oil output, while there was no HHV information available [9]. A study conducted to find out the composition of pyrolyzate from Japanese green tea residue found it to be alkaline, 70% nitrogen containing compounds of which caffeine being major constituent [10]. Low mass yields and low HHV of tea biocrudeoil are the main reasons for fewer studies. This is a comprehensive study of the characteristics of waste tea and it's product biocrude-oil that could potentially lead to many metric tons of CO2 reductions per day to the atmosphere. This study lays a platform for a sustainable renewable energy model through an integrated waste to energy CRT biocrude-oil plant. This can lead towards a sustainable low carbon economy.

2

EXPERIMENTAL

Proceedings of SEEP2015, 11-14 August 2015, Paisley inert nitrogen atmosphere using LECO TGA701 thermogravimetric analyzer.

2.1 Sample Preparation CRT waste was collected from Gartmore tea estate factory, Central Highlands, Sri Lanka. Initial moisture of CRT was about 5 ~ 7 wt% since the process of withering extracts most of the moisture in the production of black tea. The CRT was dried further in a dry oven at 105°C for 24 hours that reduces the moisture further to 3wt%. The moisture content was investigated each time based on ASTM D5142 moisture analysis using LECO TGA701 thermogravimetric analyzer. Sizing of CRT were carried out using testing sieves to size the particles within 0.85 ~ 2 mm which is 34% of the natural particle size distribution. 2.2 CRT Biomass Characteristics Physico-chemical properties of CRT is illustrated in Table 1. Proximate analysis was carried out using LECO TGA701 (Precision ±0.02% RSD). CRT contains a high level of ash of 11 wt% in comparison to other biomass. The elemental analysis was carried out by Flash EA 1112 series. CRT consists of nitrogen-containing compounds predominantly found in the tea leaf, such as caffeine, amines, pyrroles, pyridines, indole [10]. A bomb calorimeter (LECO AC500) was used to measure the HHV of the feed and biocrude oil. Bulk density of CRT was 207 kg/m3. Table 1: Proximate and Elemental analysis of Ceylon Refuse Tea Proximate Analysis (wt. %) Water content Volatile Fixed Carbon Ash Elemental Analysis (wt. %)a C H Ob N S HHV (MJ/kg) a dry, ash-free basis b by difference

3.55 ± 0.07 66.49 ± 0.16 19.15 ± 0.50 10.80 ± 0.50 58.48 ± 1.15 6.54 ± 0.17 31.36 ± 1.56 3.62 ± 0.24 0.0 19.27 ± 0.08

The effect of pyrolysis heating rates (TG and DTG curves) on CRT feed was studied under 155

2.3 Pyrolysis CRT was pyrolyzed in a bubbling fluidized bed reactor (BFBR) illustrated in Figure 1 [6,11]. The reactor chamber is cylindrical and SUS 316 with an inner diameter and height of 0.1 and 0.4 m respectively. Sand with average particle size of diameter 0.78 mm, bulk density of 1382 kg/m3 and voidage fraction of 0.43 was used as the fluidizing material. The total sand weight used in the experiment was 1.35 kg which resulted in a sand bed height of 0.12 m. Nitrogen gas, preheated by an electric heater was used as the fluidization gas and superficial velocity of 0.15 m/s was adjusted to operate within the regime of bubbling fluidization. The vapour residence time was 1~2 seconds. CRT were fed into the reactor by a two stage screw feeder at a feeding rate of 220 g/hr. An electrostatic precipitator was used to trap fine oil mist from receding volatile gas, after the condenser phase.

Figure 1: Schematic diagram of Bubbling Fluidized Bed Reactor. (1) Silo. (2) Screw feeder. (3) N2 cylinder. (4) Electric heater. (5) Fluidized bed reactor. (6) Cyclone. (7) Condenser. (8) Electrostatic precipitator. (9) Filter. (10) Accumulative Gas flow meter [6]. 2.4 Biocrude-oil Analytical Techniques CRT biocrude-oil functional groups were analyzed using Bio-Rad Excalibur FTS 3000 MX spectrophotometer. Samples were prepared as a diluted liquid sample on a 1:10 volume basis. The standard IR-spectra of hydrocarbons were used to identify functional groups of CRT biocrude-oils at different reaction temperatures. Water content of biocrude-oil was measured using Karl-Fischer 870 KF Titrino plus coulometer from Metrohm.

Activation Energy (kJ/mol)

Proceedings of SEEP2015, 11-14 August 2015, Paisley thus establishing a new generation, secondary 3 RESULTS AND DISCUSSION liquid fuel denoted in Figure 3. The sudden increase of Ea in conversion values of over 0.8 is 3.1 Thermogravimetric Analysis due to devolatilization of char. Figure 2 through DTG and TG curves, shows three distinct stages of CRT in compliance with 700 Douglas Fir other lignocellulosic biomass which can be 600 classified as vaporization and surface volatile Coffee Waste removal stage (Stage A occurs from room 500 Ceylon Tea Waste temperature to 500 K), active pyrolysis stage 400 (Stage B occurs from 500 K to 723 K) and passive pyrolysis stage (Stage C occurs beyond 300 723 K). First DTG peak is attributed to mainly 200 hemicellulose decomposition, peaks at lower temperatures (473 ~ 623 K) and second DTG 100 peak is attributed to mainly cellulose 0 decomposition (573 ~ 723 K). The 0.0 0.3 0.5 0.8 1.0 decomposition of lignin which generally has a Conversion (𝛼) broad decomposition temperature range, occurs from 473 ~ 773 K and spans into the passive Figure 3: CRT activation energy comparison pyrolysis stage, where it decomposes slowly with Douglas fir and coffee ground. alongside char. 1.0

18

A

C

B

DTG, d𝛼/dt (%/ min)

16 14

DTG 20 K/min

12

DTG 15 K/min

10 8

0.9

3.2 Mass and Energy Conversions

0.8

A digital balance with 0.01 g accuracy was used to measure the weights of the product biocrudeoil along with the feed. To determine the efficiency of the process in converting the feedstock into the products, the % mass yield and % energy yield was calculated according to Equation (1) and (2) summarized in Table 2.

0.7 0.6 0.5 0.4

DTG 10 K/min

6

DTG 5 K/min

4

0.3

0.1

0

0.0 1100

500

700

900

% 𝑀𝑎𝑠𝑠 𝑌𝑖𝑒𝑙𝑑 =

0.2

2 300

Conversion (𝛼)

20

𝑊𝑡. 𝑜𝑓 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑖𝑙 × 100% 𝑊𝑡. 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑓𝑒𝑒𝑑 (1)

% 𝐸𝑛𝑒𝑟𝑔𝑦 𝑌𝑖𝑒𝑙𝑑

𝑊𝑡. 𝑜𝑓 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑖𝑙 × 𝐻𝐻𝑉𝑜𝑖𝑙 𝑊𝑡. 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝑓𝑒𝑒𝑑 × 𝐻𝐻𝑉𝑓𝑒𝑒𝑑 × 100% (2) =

Temperature (K)

Figure 2: effect of pyrolysis rate (TG and DTG curves) of CRT. A, B and C represents three stages of pyrolysis. Based on this result, the lower limit of the pyrolysis reaction temperatures was set at 723 K. The rest reaction temperatures were set to be 773, 823 and 873 K to study the reaction temperature dependency of CRT biocrude-oil production. Kinetic study on the thermal degradation of CRT shows average activation energy (Ea) for conversion values of 0.1 - 0.7 to be 108 kJ/mol, which results in less activation energy needed for pyrolysis, to that of Douglas fir 149 kJ/mol 156

Table 2: Characteristics of reaction temperature in making CRT Biocrude-oil Reaction Temp

723K 773K 823K 873K

% Mass Yield % Energy Yield % Biochar Mass Yield

27.6 23.8

32.1 25.9

33.3 17.4

32.6 15.1

38.5

35.7

31.8

29.0

As with the Turkish tea study [9], CRT biocrude-oil has low HHV along with low mass yields. This is the main reason for limited studies been done and is not ranked as a popular

Mass Yield (%)

Proceedings of SEEP2015, 11-14 August 2015, Paisley content increases. At Optimum reaction biocrude-oil. CRT Biocrude-oil highest energy temperature of 773 K, the HHV is 15.6 MJ/kg yield denotes optimum conditions and which still results in an average to low liquid establishes reaction temperature 773 K as secondary fuel with a high moisture content of optimum reaction temperature. As a byproduct, 36.8 wt%. In comparison to other biocrude-oil 35.7 wt% of biochar is left behind which can be produced from agricultural wastes such as used for soil enhancement, composting, Douglas fir [12] (17.3 MJ/kg, 25.4wt%) and rice insecticide, insect repellent and water husk [13] (16.5 MJ/kg, 28.0wt%), CRT biocrude purification. Figure 4 illustrates the temperature oil is almost similiar. Table 2 illustrates the dependency on CRT biocrude-oil production and proximate and ultimate analysis of CRT CRT biochar productions. biocrude-oil for reaction temperature range of Biocrude-oil 723 K - 873 K. For the optimum reaction Char temperature of 773 K, H/C ratio is 2.61 and O/C 41 ratio is 1.01. This is also similar with other Non-condensable Gas 39 biocrude-oil such as rice husk [13] (2.45, 0.96), palm EFB [5] (3.14, 0.86), Mallee [6] (2.33, 37 0.88), Douglas fir [12] (2.01, 0.81) and Palm 35 shell [5] (2.87, 1.02). 10% ash in the feed is removed in the product oil. 33 Table 3: Biocrude-oil properties of pyrolysis of CRT at varying reaction temperatures.

31 29

Parameters

27 25 700

750 800 850 Reaction Temperature (K)

900

Figure 4: Temperature dependency on yield of biocrude-oil, biochar and biogas from fast pyrolysis of CRT. 3.3 CRT Biocrude-oil Characteristics 18.0

70

16.0

HHV (MJ/kg)

14.0 50

12.0

40

10.0 8.0

30

6.0

HHV

20

Water Content

10

4.0 2.0 0.0

Water Content (wt %)

60

0 700

750 800 850 Reaction Temeperature (K)

900

From Figure 5, as the reaction temperature increases, the HHV reduces while the moisture 157

773 K

823 K

873 K

Proximate Analysis (wt.%) Water Content Volatile Fixed Carbon Ash

24.21 36.84 52.88 57.23 58.89 51.86 40.10 37.78 13.60 10.80 6.86 4.99 3.30 0.50 0.16 0.00

Ultimate Analysis (%w/w) C

42.46 36.06 30.95 25.80

H

7.16

7.87

8.19

8.45

N

6.31

7.08

6.02

5.14

S

0.0

0.00

0.0

0.0

a

O

H/C molar ratio O/C molar ratio HHV (MJ/kg) a

Figure 5: HHV and water content of CRT biocrude-oil.

723 K

44.07 48.99 54.84 60.61 2.02

2.61

3.17

3.93

0.78

1.01

1.32

1.76

16.6

15.6

10.1

8.9

by difference

FTIR results of CRT biocrude-oil in Figure 6 confirms that it is similar to that of other biocrude-oils such as Douglas fir and coffee ground in terms of functional groups elaborated

Table 4: FTIR functional groups of CRT bioc rude-oil produced at reaction temperature 7 73 K. Frequency Type of Class of range Functional Compound (cm-1) Group polymeric O-H, water O-H stretching 3600 - 3200 impurities, N-H stretching Phenols -NH2 3200 - 2800 C-H stretching Alkanes 2350 - 2000 C≡C stretching Alkynes Carbonyl Aldehyde, 1740 - 1640 Stretchings Ketone, Ester 1470 - 1350 C-H bending Alkanes Alcohols, 1300 - 950 C-O stretching phenols, ethers C-H in plane Aromatic 900 - 650 benching compounds Presence of water, carboxylic acids, hydroxyaldehydes, hydroxyketones and phenols which are highly oxygenated compounds in biocrude-oil, marks the main differences between physical and chemical properties of hydrocarbon fuels and biomass pyrolysis oils These compounds result in low energy density with heating values of 16 - 18 MJ/kg nearly 50% that of conventional fuel oil [14]. Study on Douglas fir confirmed that the yields of the products from secondary thermochemical reactions such as phenol compounds increase with increasing reaction temperature [15]. The FTIR results in this study confirms this which explains the lower HHV in CRT biocrude-oil. It is evident that aldehydes and ketone absorption peaks at 1740~1640 cm-1 for the CRT biocrude-oil at 773 K is greater than that of other biocrude-oil of Douglas fir and coffee ground. This can contribute to chemical instability and reactive carbonyl compounds can readily undergo oligomerization [16]. However, the 158

Absorption

Proceedings of SEEP2015, 11-14 August 2015, Paisley in Table 3. The O-H stretching vibrations presence of alcohols were found to improve the between 3600 ~ 3200 cm-1 means confirmation homogeneity, decrease the viscosity, density and of the presence of large quantity of water flash point and also increase the heating value of through phenolic and alcoholic compounds, pyrolysis liquids [17]. In the experiment higher to that of Douglas fir. Strong Carbonyl concluded on CRT at reaction temperature 773 stretching absorption, peak between 1740~1640 K, moderately high FTIR absorption levels were -1 cm indicates presence of aldehydes, ketones observed of alcohols and ether compounds at and ester compounds. The C-H in plane 1300 ~ 950 cm-1 to that of experiments biocrude-1 benching from 900~650 cm is single ring and oil productions of Douglas fir and coffee ground. polycyclic aromatics in biocrude-oils. 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

Coffee Ground

CG

Douglas fir CRT - 773 K

CRT DF 0

1000 2000 3000 Wave Number (cm-1)

4000

Figure 6: CRT Biocrude-oil from reaction temperature 773 K FTIR spectra comparison with biocrude-oil made of Douglas fir, coffee ground 3.4 Integrated CRT biocrude-oil plant Currently the process of making biochar is through slow pyrolysis with the use of an updraft pyrolyzer and is in need of an industrial scale pyrolyzer in making biochar [18]. A study done on steam activated tea waste biochar used for removal of sulfamethazine from water is another use of biochar [19]. In making CRT biocrude-oil at 773 K, CRT biochar becomes a byproduct with a mass yield of 35.7 wt% thus making biochar a value added chemical byproduct, integrated into the waste to energy CRT biocrude-oil plant. It would not be long before plants like these would be economically viable as the general trends of liquid fossil fuels are increasing. 4

CONCLUSIONS

Although CRT biocrude-oil at reaction temperature 773 K has a lower mass and energy yields of 32.1% and 25.9% respectively and an average to low HHV of 15.6 MJ/kg with moisture content of 37 wt%, with abundant potential feed CRT results in a secondary liquid

Proceedings of SEEP2015, 11-14 August 2015, Paisley [10] M. Sakasegawa, M. Yatagai, Composition fuel that could directly be used as burner fuel for of pyrolyzate from Japanese green tea, J. Wood heat applications in tea industry. The byproduct Sci., Vol. 51, 73-76, 2005. biochar could be utilized as value added [11] H. S. Choi, Y. S. Choi, H. C. Park, chemicals for soil enhancement, an insecticide or Influence of fast pyrolysis condition on a water purifier. This in turn can be a platform biocrude-oil yield and homogeneity, Korean J. for the future when its economically favourable, Chem. Eng., Vol. 27 (4), 1164-1169, 2010. to develop a renewable and sustainable, biomass [12] J. P. Bok, Y. S. Choi, S. K. Choi. Y. W. based, integrated waste to energy model and Jeong, Fast pyrolysis of Douglas fir by using system that could lead to a reduction of CO2 tilted-slide reactor and characteristics of emissions to the atmosphere. biocrude-oil fractions, Renewable Energy, Vol. 65, 7-13, 2014. ACKNOWLEDGEMENTS [13] Q. Lu, X. L. Yang, X. F. Zhu, Analysis on The authors would like to thank The Korea chemical and physical properties of bio-oil Ministry of Science, ICT and Future Planning, pyrolyzed from rice husk, J. Anal. Appl. Pyrol., and University of Science and Technology under Vol. 82, 191-198, 2008. the Ministry of Education and Human Resources [14] L. Moens, S. K. Black, M. D. Myers, S. for supporting this study. Czernik, Study of the neutralization and stabilization of a mixed hardwood bio-oil, REFERENCES Energy & Fuels, Vol. 23, 2695-2699, 2009. [1] E. Salehi, J. Abedi. T. Harding, Bio-oil [15] S. S. Liaw, Z. Wang, P. Ndegwa, C. Frear, from sawdust: pyrolysis of sawdust in a fixedS. Ha, C. Z. Li, M. Garcia-Perez, Effect of bed system, Energy & Fuels, Vol. 23, pp.3767pyrolysis temperature on the yield and properties 3772, 2009. of bio-oils obtained from the auger pyrolysis of [2] D. Carpenter, T.L. Westover, S. Czernik, Douglas Fir wood, J. Anal. Appl. Pyrol., Vol. 93, W. Jablonski, Biomass feedstocks for renewable 52-62, 2012. fuel production: a review of the impacts of [16] J.P. Diebold, S. Czernik, Additives to lower feedstock and pretreatment on the yield and and stabilize the viscosity of pyrolysis oils product distribution of fast pyrolysis bio-oils and during storage, Energy & Fuels, Vol. 11, 1081vapors, Green Chemistry, Vol. 16, 384-399, 1091, 1997. 2014. [17] A. Oasmaa, E. Kuoppala, J. F. Selin, S. [3] T. Bridgewater, Biomass for energy, J. Sci. Gust, Y. Solantausta, Fast pyrolysis of forestry Food Agric., Vol. 86, 1755-1768, 2006. residue and pine. 4. improvement of the product [4] Q. Lu, X. L. Yang, X. F. Zhu, Analysis on quality by solvent addition, Energy & Fuels, chemical and physical properties of bio-oil Vol. 18, 1578-1583, 2004. pyrolyzed from rice husk, J. Anal. Appl. Pyr., [18] A. Abayakoon, Using biochar to improve Vol. 82, 191-198, 2008. soil health and leaf production at tea plantations [5] F. Abnisa, A. Arami-Niya, W. M. A. W. in Sri Lanka, International Biochar Initiative, Daud, J. N. Sahu, Characterization of bio-oil and June 2012. Retrieved from the website: bio-char from pyrolysis of palm oil wastes, http://www.biocharBioEnergy Research, Vol. 6, 830-840, 2013. international.org/profiles/Sri_Lanka [6] J. P. Bok, H. K. Choi, Y. S. Choi, H. C. [19] A.U. Rajapaksha, M. Vithanage, M. Zhang, Park, S. J. Kim, Fast pyrolysis of coffee M. Ahmad, D. Mohan, S.X. Chang, Y.S. Ok, grounds: characteristics of product yields and pyrolysis condition affected sulfamethazine biocrude-oil quality, Energy, Vol. 47, 17-24, sorption by tea waste biochars, Bioresource 2012. Technol., Vol. 166, 303-308, 2014. [7] Ceylon Tea Brokers PLC. Annual Report 2013/2014. Colombo, Sri Lanka, 2014. [8] Ceylon Tea Services PLC. Sustainability Report 2013/2014. Colombo, Sri Lanka, 2014. [9] B. B. Uzun, E. Apaydin-Varol, F. Ates, N. Özbay, A. E. Pütün, Synthetic fuel production from tea waste: Characterisation of bio-oil and bio-char, Fuel, Vol. 89, 176-184, 2010.

159


INVESTIGATION AND COMPARISON OF MICRO-PARTICULATES (PM) AND GASEOUS EMISSIONS OF THREE BIOMASS FUELS Joanna Relf1, Edward Forbes1, Rodrigo Olave1, Christopher Johnston1, Gary Lyons1 and David McCall1 1. Agri-Food and Biosciences Institute, Hillsborough, Northern Ireland, United Kingdom; [email protected]

email:

Abstract The Renewable Heat Incentive (RHI) was recently introduced in the UK to support heat generation from renewable energy. The RHI has set criteria for emissions from biomass combustion of 150 g/GJ of nitrous oxides (NOx) and 30 g/GJ of particulate matter 1000 >1000 >1000 660 170 74 190 74 74 240

The determination of MEC was achieved by testing decreasing mass of powder, repeating each experiment three times, until a concentration was found at which no ignition occurred in any of the three tests. Consequently a curve representing probability of explosion against equivalence ratio could be drawn. From such curve the MEC could be found as the ratio at which the probability of explosion is 0%, which provides the safest values. MEC for 50%

175

NFA-103

8.20

0.30

0.48

0.60

1.37

NFA-104

12.74

0.39

0.45

0.77

2.31

NFA-105

14.27

0.41

0.46

0.50

4.00

NFA-106

17.22

0.42

0.48

0.67

4.84

NFA-107

11.00

0.38

0.40

0.43

3.19

NFA-108

12.52

0.25

0.39

0.44

3.72

NFA-109

19.45

0.22

0.33

0.45

5.37

NFA-110

12.61

0.32

0.37

0.38

3.45

Lower MEC suggests a higher reactivity and a greater hazard presented by the material, since less concentration of powder is necessary for the mixture to be ignitable. For typical hydrocarbon fuels (gases and vapours) the lower flammability limit is of the order of =0.5. If the value determined is near this value, the evolved gases may be hydrocarbons (such as methane), but if the value is lower, the evolved gases are flammable gases with much wider flammability range, such as hydrogen.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [2] Handbook of biomass combustion and coThe most reactive mixture is this one where the firing. Sjaak van Loo and Jaap Kippejan. Twente peak rates of pressure rise and flame speed were Universty Press, 2002 found. [3] K.L. Cashdollar, Coal dust explosibility, Looking at the first three mixtures, formed by Journal of Loss Prevention in the Process coal and sewage sludge (NFA-101 to 103), the Industries, higher the concentration of coal, the higher the [4] N. Fernandez-Anez, J. Garcia-Torrent, L. value of MEC, so the lower the reactivity of the Medic-Pejic. Flammability properties of samples is. When the mixtures are formed by thermally dried sewage sludge, Fuel, Vol. 134, coal and torrefied wood pellets (NFA-104 to pp. 636-643, 2014. 106), a double effect is shown. In one hand, the [5] J. Garcia Torrent, N. Fernandez Anez, L. higher the concentration of biomass, the higher Medic Pejic, L. Montenegro Mateos. the (dP/dt) max and the Sf values so the samples Assessment of self-ignition risks of solid present more reactivity. But on the other hand, biofuels by termal analysis, Fuel, Vol. 143, pp. the MEC is higher too, so the reactivity seems to 484-491, 2015. be lower. In relation to the sewage sludge and [6] B. Binkau, C. Wanke, U. Krause. Influence torrefied wood pellets mixtures, the addition of of inert materials on the self-ignition of pellets made a more reactive sample, increasing flammable dusts, Journal of Loss Prevention in the values of (dP/dt) max and the Sf and the Process Industries, 2014, decreasing the values of MEC. Finally, looking at http://dx.doi.org/10.1016/j.jlp.2014.11.017 the sample formed by the three components, the [7] UNE 32004:1984. Solid mineral fuels. values of both parameters are located in between Determination of ash. the two-sample mixtures, but closer to the [8] UNE 32019:1984. Hard coal and coke. highest reactive values. Determination of volatile matter content. [9] UNE 32002:1995. Solid mineral fuels. 4 CONCLUSIONS Determination of moisture in the analysis The influence of the percentage of the different sample. materials varies depending on the way of [10] EN 50281-2-1:1999. Electrical apparatus distribution of the samples. In one hand, the for use in the presence of combustible dust. Part parameters related with the formation of clouds 2-1: Test methods – Methods for determining show that the flammability of biomass clouds is the minimum ignition temperatures of dust. more dangerous than the others, so the formation [11] A. Janes, J. Chaineaux, D. Carson, P.A. Le of cloud dusts present more risks as the Lore, MIKE 3 versus HARTMANN apparatus: percentage of biomass in the mixtures increases. Comparison of measured minimum ignition On the other hand, looking at the values of the energy (MIE), Journal of Hazardous Materials, minimum ignition temperature on a layer, the Vol. 152, pp. 32-39, 2008 values of this parameter are higher by adding [12] C. Huéscar Medina, H.N. Phylaktou, H. torrefied wood pellets. Sattar, G.E. Andrews, B.M. Gibbs, The This variation is caused by the shape of the development of an experimental method for the particles, and how the heating process may determination of the minimum explosible influence to the long and thin biomass particles concentration of biomass powders, Biomass and or to the round sewage sludge and coal particles. Bioenergy, Vol. 53, pp. 95-104, 2013. [13] M.A. Saeed, C. Huéscar Medina, G.E. ACKNOWLEDGEMENTS Andrews, H.N. Phylaktou, D. Slatter, B.M. N. Fernandez-Anez was supported by the Gibbs, Agricultural waste pulverised biomass: mobility program “Estancias breves en España y MEC and flame speed, Journal of Loss en el extranjero 2014” of the Technical Prevention in the Process Industries, 2014, University of Madrid (UPM). http://dx.doi.org/10.1016/j.jlp.2014.12.007 REFERENCES [1] ec.europa.eu/clima/policies/package/index_en.ht m 176


SELECTIVE HYDROCONVERSION OF BIOMASS-DERIVED LEVULINIC ACID TO GAMMA-VALEROLACTONE M.R. Mihályi, Gy. Novodárszki and J. Valyon Institute of Materials and Environmental Chemistry, Research Centre for Natural Sciences, Hungarian Academy of Sciences, Budapest, Hungary; email:[email protected]

Abstract Co/silica catalyst was used for the hydroconversion of levulinic acid (LA) using a flow-through microreactor at 200-300 °C, 30 bar total pressure and 1.1 h-1 WHSV for LA. Silica, impregnated with Co(II)nitrate solution, was dried, calcined at 500°C and then reduced in H2 flow in-situ in the catalytic reactor at 450 oC. Samples were characterized by XRD, TEM, and H2-TPR measurements. Chemisorption of CO was examined by FT-IR to sound the cobalt surface. The catalytic activity and selectivity was function of the Co dispersion and time-on-stream (TOS). The product distribution could be controlled by adjusting the reaction conditions. Product -valerolactone (GVL) was obtained with high selectivity (98 %) at high LA conversion (99%) at 200°C. Reaction network and the possible pathways of GVL formation were proposed. Keywords: Levulinic acid (LA), -valerolactone (GVL), Co/SiO2 catalyst, LA hydroconversion 1 INTRODUCTION Lignocellulosic biomass is an environmentally benign substitute for fossil carbon sources in the production of fuel, chemicals and carbon-based products. From the polysaccharid components (cellulose and hemicellulose) valuable platform molecules can be prepared for the chemical industry. Technology, known as Biofine process, has been developed for large-scale production of levulinic acid (LA) [1]. Hydrolysis of polysaccharides and sulphuric acid catalysed reactions of the product 6- and 5-carbon sugars provide LA with high yield. In consecutive hydrogenation and dehydration steps LA can be converted to cyclic compound -valerolactone (GVL). GVL can be used as gasoline surrogate, solvent, flavoring agent, and building block of polymers. In consecutive reaction steps GVL can be converted to 1,4pentanediol (1,4-PD), 2-methyl tetrahydrofurane (2-MTHF) and, on another reaction route, to pentanoic acid (PA) and its pentylester. The 1,4-PD may find application in the production of biobased polymers, whereas the 2-MTHF, having relatively high energy density, is potential fuel component. The selective conversion of LA to useful product is a real challenge for research. Both homogeneous and heterogeneous metal catalysts, particularly noble metal catalysts, have been applied for LA transformation to GVL [2, 3]. Among them Ru, Pt and Pd showed high 177

activity. Mainly conventional batch reactors were used. However, heterogeneous catalytic flow reactors were especially suited to commercial-scale production of GVL because, in this case, separation of the product and the catalyst would not require additional operation. Although noble metals show good performance, their high cost and low abundance limit their large-scale application. Few reports have been published on the continuous catalytic process of LA conversion to GVL, utilizing oxidesupported non-precious metals, such as, Ni [4] and Cu [5,6]. Recently, a cobalt catalyst prepared from commercially available Co3O4 was shown to be active in the hydrogenation ethyl levulinate to GVL in batch reactor [7]. Cobalt nanoparticles embedded in HZSM-5 zeolite crystals have found an efficient application in the one-pot conversion of LA to ethyl valerate and valeric acid in the presence of ethanol [8]. The use of Co on oxide supports, such as silica or alumina, for solvent-free LA transformation in H2 is reported here for the first time. The catalyst was characterized by XRD, TEM, H2TPR methods. The chemisorption of CO was followed by transmission FT-IR spectroscopy. A high-pressure, flow-through fixed-bed microreactor was used. The effect of time-onstream (TOS) and temperature on the activity and the selectivity of the catalyst were studied.

Proceedings of SEEP2015, 11-14 August 2015, Paisley samples to calibrate the system for H2 2 EXPERIMENTAL consumption. 2.1 Catalyst Transmission Fourier-transform infrared (FT-IR) Silica support (CAB-O-SIL EH5 type silica spectra were recorded by a Nicolet Impact (Cabot Corp.), having a BET surface area of 385 Type 400 spectrometer applying self-supported m2/g) was impregnated with an aqueous solution wafer technique. Spectra were recorded at room of Co(NO3)2*6H2O. 10 g of silica was temperature averaging 32 scans at a resolution of suspended in 40 cm3 0.4 M solution. The 2 cm-1. Sample was reduced in an H2 flow of 50 suspension was dried at 110°C for 12 h. The cm3/min for 1 h at the temperature given in surface-bound nitrate was decomposed by Figure 4, then evacuated in high vacuum (HV), heating the sample in air at 500°C for 3 h. The cooled to room temperature and a spectrum was air-calcined sample was designated as collected. Five mbar CO was introduced into the Co3O4/SiO2. Before catalytic test the supported cell. After contacting the CO and the catalyst for oxide was reduced in situ in the applied catalytic 10 min, the cell was degassed in HV for 10 min microreactor (vide infra) by flowing H2 at 450 and a spectrum was recorded. The spectrum o C for 1 h. The Co content of the catalyst was 9 recorded before CO admission in the infrared wt %. The reduced catalyst is referred to as cell was subtracted from that, recorded after CO Co/SiO2. adsorption to get the spectrum of the adsorbed 2.2 Methods CO (Figure 4). X-ray diffraction (XRD) patterns (Figure 1) The catalytic experiments were carried out at 30 were recorded by a Philips PW 1810/3710 bar total pressure using a down-stream fixed-bed powder diffractometer, equipped with a graphite stainless steel tube reactor (12 mm ID). The monochromator, applying CuKα1+α2 radiation reactor was loaded with 1 g of 0.315-0.630 mm (λ(average)=1,541862 Å) using a step of 0.02 size sieve fraction of Co3O4/SiO2 grains. Prior to 2 and a count time of 4.0 s. Crystallite size was catalytic runs the supported oxide was reduced determined by the Scherrer equation evaluating in-situ in an H2 flow of 100 cm3/min at 30 bar the FWHM values of the diffraction lines and 450 °C for 2 h. Levulinic acid was then fed applying full profile fitting method. into the reactor by a micro pump (Gilson, Model TEM images (Figure 2) were taken by a 302) at a WHSV of 1.1 h-1. The H2/LA molar Morgagni 268D electron microscope. ratio was 12.8. At room temperature the reactor Adsorption isotherms were determined by N2 effluent was separated to liquid and gas sorption at – 196°C using automated gas products. Products were analyzed by GC-MS adsorption instrument (SURFER, Thermo (Shimadzu QP-2010) using a 25 m FFAP-CB Fischer Scientific). Before measurements the capillary column. samples were degassed by evacuation at 350°C for 12 h. The specific surface area was 3 RESULTS AND DISCUSSION calculated by the B-point method. 3.1 Catalyst properties The reducibility of the surface bound cobalt was The XRD pattern of the catalyst was recorded in studied by temperature-programmed reduction its oxide form, after pre-reduction and also after experiment (TPR) using H2 as reducing agent use for LA hydroconversion (Figure 1). (Figure 3). About 100 mg of Co3O4/SiO2 sample The presence of XRD lines in the oxide form was treated in 30 cm3/min O2 flow in a quartz confirmed that cobalt(II) nitrate precursor was reactor tube (6 mm ID) at 500°C for 1 h. In order decomposed to Co3O4 in air at 500°C (Figure to obtain TPR curve the sample was cooled to 1a). The low intensity reflection at 44.5 o comes 40°C flushed with N2 for 10 min, exposed to a from Co phase (Figure 1b). Furthermore, the flow of 9.0 vol% H2/N2 mixture, and then heated lines of CoO at 36.5o, 42.4o and 61.5o also up to 800 oC at a rate of 10 oC/min. The water appeared. Only reflections of Co were detected was removed from the reactor effluent by in the pattern of the used catalysts (Figure 1c). passing it through a liquid nitrogen (-196 °C) The crystallite size of Co3O4, determined from trap. The rate of hydrogen uptake was recorded the broadening of the diffraction lines was 26 using a thermal conductivity detector (TCD). nm. Finely dispersed metallic Co particles (12 Measured amounts of Cu(II)O (Merck) were nm) were obtained upon in-situ reduction in reduced under the same conditions as the

178

Proceedings of SEEP2015, 11-14 August 2015, Paisley TEM images confirm the XRD results. Particle size of about 20-30 nm was observed for the Co3O4 supported on silica (Figure 2). The TPR profile of the Co3O4/SiO2 sample is presented in Figure 3. Two strongly overlapping reduction peaks give a single maximum at 354°C. In accordance with Li et al. [9] result suggests that the reduction of Co3O4 to Co0 is a two-step process. (Co3O4 + H2→ 3CoO + H2O and CoO + H2→ Co + H2O). The shoulder in temperature range of 500-700°C has to be ascribed to Co species being in strong interaction with the silica support. It was found that the Co3O4/SiO2 sample consume 2.54 H/Co up to 800°C. This corresponds to reduction degree of 95 % assuming that 2.67 H/Co is needed for the total reduction of Co3O4 to Co0. Figure 1. XRD patterns of the (a) Co3O4/SiO2 , (b) Co/SiO2, and (c) Co/SiO2 used for 50 h in catalytic test. Reduction was carried out in H2 flow 30 bar H2 pressure for 1 h. hydrogen before catalytic run. The higher intensity of the Co reflections in the used catalyst indicates, that during the long ( 50 h) catalytic experiment (vide infra) sintering proceeded. The size of Co crystals was 40 nm. The presence of CoO phase in the reduced catalyst suggest that Co3O4 reduction to Co was not complete at 450°C, one-third of the Co atoms remained in Co2+ state. It cannot be excluded that some of the surface Co atoms became oxidized when the sample was exposed to air. (In-situ XRD experiments are in progress.)

Figure 3. H2-TPR curve of the Co3O4/SiO2 sample pre-treated in an O2 flow of 30 cm3/min at 500°C for 1 h. Figure 4 shows the FT-IR spectra of the adsorbed CO over Co3O4/SiO2 samples reduced at different temperatures. No vibration band of adsorbed CO appeared in the infrared spectrum of the sample reduced at 300 °C. Reduction to metallic cobalt particles (Co0) begins at 350 °C that is confirmed by the appearance of the CO band at 2006 cm-1, assigned to CO linearly bound to cobalt metal [10]. Hydrogen treatment at 400°C resulted in a significant increase in the intensity of this band, whereas the band became slightly weaker upon treatment at 450°C.

Figure 2. TEM image of the Co3O4/SiO2 sample. 179

Proceedings of SEEP2015, 11-14 August 2015, Paisley 3.2 Levulinic acid hydroconversion The metal catalyzed transformation of LA to GVL proceeds either through 4-hydroxypentanoic acid (4-HPA) or through unsaturated lactone intermediates (Scheme 1) [12]. On Path I LA is first hydrogenated to 4-HPA that is readily dehydrocyclized to GVL. On Path II cyclization starts with the formation of pseudo-LA transition state that relaxes by dehydration giving unsaturated lactone. Hydrogenation of the ring CC double bond gives GVL. Rh and Ru are the most investigated metals as catalysts of LA hydroconversion. Their complexes are used in batch reactor in liquid phase as homogeneous catalysts or they are deposited on carbon and used as solid catalysts in batch or continuous flow-through reactor Figure 4. FT-IR spectra of CO adsorbed on systems. Ruthenium was reported to be selective the Co3O4/SiO2 sample (room temperature, in the hydrogenation of a carbonyl group even in 5 mbar) after in-situ reduction in H2 at the indicated temperature. Temperature of a the vicinity of CC double bond or aromatic single pellet was increased in successive ring. It is also selective for partial hydrogenation steps. of aromatic ring to cycloalkene [13]. Consequently, Ru is able to catalyse both A weak band was also observed at 2181 cm-1, n+ pathways of LA to GVL reaction. which is due to the CO adsorbed on Co (n=2,3) species [11]. These results suggest that after H2 reduction at 450°C majority of Co-oxide was reduced to Co0, but small amount of finely dispersed CoOx remained on the silica surface.

Scheme 1. Pathways of levulinic acid hydroconversion.

180

Proceedings of SEEP2015, 11-14 August 2015, Paisley pressure, contact time) 4-HPA has never been detected among the products, Path I for GVL formation cannot be excluded (experiments are in progress). In contrast to 4-HPA, angelica lactone intermediates were identified in most product mixtures. Angelica lactones in product mixtures, obtained by using noble metal catalysts, were not always detected. It can be explained by their rapid hydrogenation to GVL and to further products on reaction Path III. Metal dispersion is known to have decisive role in the activity and the selectivity of supported metal catalysts. The conversion decreasing with the TOS and the varying selectivity can be interpreted by the decreasing dispersion while the catalyst is used. Another possible reason of activity loss and/or selectivity change during reaction is catalyst coking. Figure 5. LA hydroconversion and product XRD results suggested that in-situ reduction of yields over Co/SiO2 catalyst at 200°C, 30 -1 the Co3O4/SiO2 sample resulted in finely bar total pressure, and 1.1h WHSV of LA, dispersed particles of Co metal on the support. as a function of time-on-stream (TOS). In the early stage of the reaction the dispersion Figure 5 shows the conversion of LA and the and thus, the hydrogenation activity of the product yields as a function of TOS. Full LA catalyst was higher than at the end of the conversion was achieved under the indicated catalytic experiment. Catalysts having lower Co reaction conditions. The high activity of the dispersion were found to be of lower ring catalyst was maintained during 20 h time on opening activity and, therefore, higher GVL stream, while the product distribution drastically selectivity. In our opinion ring opening changed. The ring opening activity decreased. polymerization of angelica lactones, leading to After 16 h TOS only GVL was formed virtually catalyst coking, strongly contributes to the found at full conversion. In the early stage of the selectivity change. catalytic run the primary product GVL was further converted to 2-MTHF via 1,4-PD intermediate (Scheme 1, Path III). Small amounts of pentanols were also formed, indicating that 2-MTHF ring opening reactions proceeded both at bonds IV and V (Scheme 1). Under reaction conditions, specified in the legends of Figure 5, the pure silica support did not induce LA conversion at all. The calcined Co3O4/SiO2 sample exhibited very low activity. The LA conversion was about 8 %. No GVL only angelica lactones were formed, confirming the dehydration activity of the CoOx species. Surface metal cobalt atoms are supposed to be the catalytically active sites of the reaction. However, XRD, H2-TPR and FT-IR results suggest that the reduction of Co3O4 to metal Co was not complete at 450°C, thus CoOx species Figure 6. LA hydroconversion and product remained on the catalyst surface. These yields as a function of temperature over unreduced Co-oxide sites may also contribute to Co/SiO 2 catalyst at 30 bar total pressure, the LA transformation. and 1.1 h-1 WHSV of LA. Although LA hydroconversion was studied at varied reaction conditions (temperature, 181

Proceedings of SEEP2015, 11-14 August 2015, Paisley [3] W. R. H. Wright and R. Palkovics, Gamma-alumina support, known about its high Development of Heterogenouos Catalyst for the dehydration activity, favors angelica lactone formation and polymerization explaining the Conversion of Levulinic Acid to found fast and complete deactivation of the Valerolactone, Chem Sus Chem, Vol. 5, pp. 112, 2012. Co/-Al2O3 catalyst in the LA hydroconversion [4] V. Mohan, V. Venkateshwarlu, C. V. reaction (not shown). Pramod, B. D. Raju and K. S. R. Rao, Vapour The silica-supported cobalt catalyst is less active in generation and polymerization of unsaturated phase hydrocyclisation of levulinic acid to lactones. However, results show that the valerolactone over Ni supported catalysts. hydrogenation activity and, thus, the activity of Catalysis Science  Technology, Vol. 4, pp. the catalyst in the ring opening reactions of GVL 1253-1259, 2014. and 2-MTHF slowly decrease in time (Figure 5). [5] B. Putrakumar, P. Bijayanand, K. V. P. As a result the GVL yield increases with TOS on Kumar and K. V. R. Chary, Hydrogenation of the expense of the 2-MTHF yield. biomass-derived levulinic acid to -valerolactone After 20 h, when steady state was achieved, as over copper catalysts supported on ZrO2, shown in Figure 5, the temperature was Journal of Chemical Technology and increased from 200°C up to 300°C in 25 °C Biotechnology, DOI 1.1002/jctb.4643. intervals. The product distribution at each [6] B. Putrakumar, N. Nagaraju, V. P. Kumar temperature is presented in Figure 6. The GVL and K. V. R. Chary, Catalysis Today, Vol. 250, yield decreased drastically, because increased pp. 209-217, 2015. temperature accelerated ring opening, [7] H. Zhou, J. Song, H. Fan, B. Zhang, Y. hydrogenolysis, dehydration, and hydrogenation Yang, J. Hu, Q. Zhu and B. Han, Co catalysts: reactions. very efficient for hydrogenation of biomassderived ethyl levulinate to gamma-valerolactone under mild conditions, Green Chemistry, Vol. 4 CONCLUSIONS 16, pp. 3870-3875, 2014. Efficient hydroconversion of LA to GVL was [8] P. sun, G. Gao, Z. Zhao, C. Xia and F. Li, achieved over Co/SiO2 catalyst applying a Stabilization of Cobalt Catalysts by Embedment continuous-flow fixed-bed tube microreactor. for Efficient Production of Valeric Biofuel, ACS The activity, especially the selectivity could be Catalysis, Vol. 4, pp. 4136-4142, 2014. controlled by adjusting the reaction conditions [9] H. Li, Sh. Wang, F. Liang and J. Li, Studies and tuning the surface properties of the catalyst. on MCM-48 supported cobalt catalyst for The LA conversion could be stopped at GVL Fischer–Tropsch synthesis, Journal of formation. High selectivity (98%) to GVL was Molecular Catalysis A, Vol. 244, pp. 33-40, obtained at high LA conversion (99%) at 200°C 2006. and 30 bar. [10] K. R. Mohana, G. Spoto and A. Zecchina, The low-cost Co/SiO2 catalysts are promising IR investigation of CO adsorbed on Co particles alternatives to noble metal catalysts in the obtained via Co2(CO)8 adsorbed on MgO and selective hydroconversion of LA to GVL. SiO2 Journal of Catalysis, Vol. 113, pp. 466474, 1988. ACKNOWLEDGEMENTS [11] S. Ischi, Y. Ohno and B. Viswanathan, An The authors thank the financial support of the overview on the electronic and vibrational National Development Agency, Grant No. properties of adsorbed CO, Surface Science, KTIA_AIK_12-1-2012-0014. Vol. 161, pp. 349-372, 1985. [12] D. M. Alonso, S. G. Wettstein and J. A. REFERENCES Dumesic, Gamma-valerolactone, a sustainable [1] S. W. Fitzpatrick, (Biofine Technologies platform molecule derived from lignocellulosic LLC). U.S. Patent No. US2010234638A1, 2010. biomass, Green Chemistry, Vol. 15, pp. 584[2] F. Liguori, C. Moreno-Marrodan and P. 595, 2013. Barbaro, Environmentally Friendly Synthesis of [13] P. Kluson and L. Cerveny, Selective -Valerolactone by Direct Catalytic Conversion hydrogenation of ruthenium catalysts, Applied of Renewable Sources, ACS Catalysis, Vol. 5, Catalyst A: General, Vol. 128, pp.13-31, 1995. pp. 1882-1894, 2015.

182


SINTERING CHARACTERISTICS OF SYNTHETIC COAL ASH Hongwei Hu1, Qizhao Lin*2, Kun Zhou3 1. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected] 2. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected] 3. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected] Abstract Sintering problems in boiler and heat exchanger cause damage to heat transfer and stability of facility, and seriously lead to security risks. Three synthetic coal ashes (SGA): Helinghe (HLH), Xiaolongtan (XLT) and Huaibei coal blending with coal slime (CBS), were kept in a drop tube furnace (DTF) to make sintering samples. Ash fusion test，X-Ray Diffraction (XRD) and scanning electron microscope (SEM) were carried out, respectively, to measure ash fusion temperature (AFT), chemical composition and microstructure of these sintering samples. Results show that, both deformation temperature (DT) and soften temperature(ST) of coal ash increase with SiO2/Al2O3（Si/Al）； the forming of eutectic, such as anorthite, augite, diopside etc., is the essential condition of sintering. Still, morphology distinctions of three samples are mainly due to different relative content of each eutectic. Keywords: Sintering, Synthetic, Fusion temperature, Eutectic 1. INTRODOCTION Sintering problems in boiler and heat exchanger cause damage to heat transfer and stability of facility, and seriously lead to security risks. Factors that affect sintering are complicated, such as flow field structure, geometry, chemical components and operation parameters etc. Pisarev G I etc. [1] and Wu X etc.[2] reported that the ‘end of vortex’, causing temperature and pressure fluctuation, can affect the flow structure of ash-gas flow. Jing N, Wang Q etc. [3] found out that operation temperature and atmosphere play a decisive role during sintering, while the influence of pressure is complicated. Matjie R H etc.[4] tried to discuss how the cooling rate affects the process of mineral crystallization. Yang T etc. ’s [5] result shows that basic oxide that react with SiO2 to form silicates, such as CaSiO3, MgSiO3 etc., can lower ash fusion temperature. Eutectics that act as a role of adhesive, such as anorthite, augite, diopside etc., were found in almost every literature [6-8]. Traditional research methods of sintering include ash fusion test [9], XRD [9] and SEM [9-11]. Recently, for better controlling the operation parameter, thermal mechanical analyser (TMA)[12] and DTF [13] were utilized to study sintering characteristics. Xuan W etc.[14, 15] used synthetic coal ash in their research to ana-

183

lyse the interaction between compounds, neglecting the influence of microelement. In this paper, SGA were used to form sintering samples. Different methods were tried to elaborate the sintering characteristics. 2. EXPERIMENT 2.1 Experimental appraratus DTF: As shown in Fig.1, air flows through the quartz tube（8） and exits from outlet pipe (9), the volume flow-rate of which is controlled by flow meter (2). Crucible (5) is embedded in the pallet (6) which is located in the bottom of the tube. Furnace temperature is controlled by the combine effect of temperature controller (4) and electric strip heater (9). Two adiabatic plugs (7, 11) were installed in the top and bottom of quartz tube (8) to prevent heat loss. DTF can deal with 3 samples in the same atmosphere and at the same temperature each time. Ash Fusion Test: TMA, NETZSCH DIL, 402C. XRD : High power X-ray powder diffractometer, TTRⅢ，Scanning Rage: 3º~70º. FESEM:

Proceedings of SEEP2015, 11-14 August 2015, Paisley Field emission scanning electron microscope, with rise of Si/Al. This is because of the comJSM-6700F; Scale: 200X, 500X, 1000X. bine action of Al2O3 and SiO2. SiO2 has dual influence on ash fusion temperature because it exists in two forms: quartz and cristobalite[8]. Quartz is very stable and has high fusion temperature, up to 1800K,which would elevate AFT; however, cistobalite is relatively active which can act with basic oxide to form eutectics, such as anorthite, augite, diopside etc., which will lower AFT[9]. The existence of quartz and cristobalite is certified by XRD graph in Fig.3. On the other hand, Al2O3 will always elevate AFT through two approaches, one of which is that Al2O3 itself is very stable and has high fusion temperature, up to 2270K; the other is that Al2O3 acts as “skeleton”, which will obstruct the formation and conversion of eutectics[6]. Fig.1. DTF system: 1-Valve 2-Flow meter 3Mixer 4-Temperature controller 5-Pallet 6Crucible 7-Adiabatic plug 8-Quartz tube 9Electric strip heater 10-Window 11-Adiabatic plug 12-Outlet pipe 2.2 Progress of experiment The chemical composition of SGA samples is shown in Table 1: Table1. Chemical composition of SCA Ash compositions of lignites m/%

SCA

SiO2

Al2O3

Fe 2O3 CaO

M gO

Si / Al

XLT HLH

39.90 21.30 26.00 10.80

2.00

1.87

57.30 20.30

8.70

11.70

2.00

2.82

CBS

51.00 23.90 15.10

8.00

2.00

2.13

Fig.2. AFT of XLT, CBS and HLH

SCA samples were moulded into 4mm*4mm*12 mm shape sample with 10MPa pressure and then sent to take AFT measurement. These shaped samples were kept in DTF for 6 hours to make sintering samples, the temperature of which was set at 1223K,with air flow (30%O2/70%N2) through. Then these sintering samples were naturally cooled to room temperature. Half of the cooled samples were polished into powder and then sent to take XRD measurement while the other half were directly sent to take SEM scanning. 3. RESULTS AND DISCUSSION 3.1 Ash fusion test The result of ash fusion test is displayed in Fig.2. From left to right, respectively, the points in Fig.2 represent the DT and FT of XLT, CBS and HLH. As is shown, both DT and ST increase 184

3.2 XRD ANALYSIS Fig.3 shows the XRD graphs of XLT, CBS and HLH sintering ash. As is shown, number 1-10 represents cristobalite, corundum, calcium oxide, anorthite, diopside, wollastonite, augite, iron, quartz and orrite, respectively. The chemical components of sintering ash have been decided by SCA, so each species exists on different graphs. However, there are distinctions among peak heights of different species, especially corundum , cristobalite and quartz, which means different relative content of each species exists in different samples.

Proceedings of SEEP2015, 11-14 August 2015, Paisley 3.3 SEM ANALYSIS SEM graphs of XLT, CBS and HLH were displayed on Fig 5. From top to down, graphs show different scales of sintering samples: 200X, 500X and 1000X, respectively. There are crystal material which is detected as crisobilite and “skeleton cell” which is detected as corundum on every image. In the graph of XLT sample, grey fibber material which is detected as anorthite exists while in CBS and HLH coal it disappears； this is mainly because anorthite has turned into other eutectics, such as anorthite, augite etc.[8]. Distinctions of morphology of three samples are mainly due to relative content of each eutectics Fig. 3. XRD graphs of XTL, HLH and CBS: which is consistent with the results of AFT and 1-cristobalite 2-corundum 3-calcuium oxide XRD analysis. 4-anorthite 5-diopside 6-wollastonite 7augite 8-iron 9-quartz 10-dorrite

The fusion temperature of different compounds and eutectics which are detected by XRD is shown in Fig.4. The scatter points reflect the fusion temperature of different compounds and eutectics, while the blue line denotes the furnace operation temperature (FOT) which is also equal to furnace exit gas temperature (FEGT). In Fig 4, none of the scatter point lies below the blue line, which means that FOT or FEGT is lower than AFT (FEGTAFT)[11]. So the formation and conversion of eutectics should be the essential condition of sintering, which is also consistent with the literature conclusion [8].

Fig. 4. Fusion temperature of different compound: 1-cristoblite 2-corundum 3-calcuium oxide 4-anorthite 5-diopside 6-wollastonite 7-augite 8-iron 9-quartz 10-dorrite

Fig. 5. SEM graphs of XLT, CBS and HLH 4. CONCLUSIONS Three SGA: HLH, XLT, CBS were kept in a DTF for 6 hours to get sintering samples. Ash fusion test， XRD, and SEM were carried out, respectively, to get ash fusion temperature, chemical composition and microstructure of sintering samples. Results show that: 1. Both DT and ST of coal ash increase with Si/Al. 2. The forming of eutectics, such as anorthite, augite, diopside etc., is the essential condition of sintering, rather than FEGT>AFT. 3. Distinctions of morphology of three samples are mainly due to different relative content of each eutectic which is consistent with the results of AFT and XRD analysis. ACKNOWLEDGEMENTS This work is supported by National Natural Science Foundation of China (No. 51376171), and National Key Basic Research Program (No. 2010CB227300) founded by MOST. REFERENCES

185

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

Proceedings of SEEP2015, 11-14 August 2015, Paisley scale laboratory furnace [J]. Fuel, 2014, 117 Pisarev G I, Hoffmann A C. Effect of the ‘ :269-277. end of the vortex’ phenomenon on the part [13] Liu Y, Gupta R, Elliott L, et al. Thermomec icle motion and separation in a swirl tube se hanical analysis of laboratory ash, combusti parator [J]. Powder Technology, 2012, 222: on ash and deposits from coal combustion[J 101-107. ]. Fuel Processing Technology, 2007, 88(11 Wu X, Liu J, Xu X, et al. Modelling and ex -12):1099-1107. perimental validation on pressure drop in a r [14]Xuan W, Whitty K J, Guan Q, et al. influen everse-flow cyclone separator at high inlet s ce of isothermal temperature and cooling rat olid loading [J]. Journal of Thermal Science es on crystallization characteristics of a synt , 2011, 20(4):343-348. hetic coal slag [J]. Fuel, 2014, 137:193-199. Jing N, Wang Q, Cheng L, et al. Effect of te mperature and pressure on the mineralogica l and fusion characteristics of Jincheng coal ash in simulated combustion and gasificatio n environments[J]. Fuel, 2013, 104:647-655 . Matjie R H, Li Z, Ward C R, et al. Chemica l composition of glass and crystalline phase s in coarse coal gasification ash[J]. Fuel, 20 08, 87(6):857-869. Yang T, Ma J, Li R, et al. Ash Melting Beh aviour during Co-gasification of Biomass a nd Polyethylene [J]. Energy & Fuels, 2014, 28(5):3096-3101. Rao Su, Cao Xinyu, Lan Zequan, et al. Char acteristics of mineral distribution in furnace and its influence on the slagging during co mbustion[J]. Journal of Fuel Chemistry and Technology, 2004,32(04):395-399. Lan Zequan, Cao Xinyu, Zhao Xanqiao, et al. Experimental study on fusion of coal wat er slurry[J].Thermal Power Generation, 200 3,32(4):13-16. Wang Quanhai, Qiu Jianrong, Li Fan, et al. Conformation change and phase transforma tion of minerals during coal blends combust ion [J]. Journal of Chemical Engineering of (china), 2000, 51(6):840-843. Ji S, Li F, Wang T, et al. Investigation on th e sintering behaviours of low-temperature li gnite ashes[J]. Journal of Thermal Analysis and Calorimetry, 2014, 117(3):1311-1320. Luan C, You C, Zhang D. An experimental investigation into the characteristics and de position mechanism of high-viscosity coal a sh [J]. Fuel, 2014, 119:14-20. Bilirgen H. Slagging in PC boilers and deve loping mitigation strategies [J]. Fuel, 2014, 115:618-624. Wang G, Pinto T, Costa M. Investigation on ash deposit formation during the co-firing of coal with agricultural residues in a large-

186


SWITCHGRASS GROWTH (ALAMO VARIETY) AS AFFECTED BY IRRIGATION AND N-FERTILIZATION IN CENTRAL GREECE

K.D. Giannoulis1, D. Bartzialis1, E. Skoufogianni1, S.N. Sakkou1, M. SakellariouMakrantonaki1 and N.G. Danalatos1 1. University of Thessaly, Dept. of Agriculture, Crop Production & Rural Environment, Volos, Greece, corresponding author email: [email protected]

Abstract In the present paper is presenting the growth and the biomass yield of the perennial energy crop Switchgrass (variety Alamo) as affected by two irrigation and four N-fertilization levels in its third year of establishment. For the purpose of the study, a factorial split plot field experiment in four blocks was established in Karditsa (west Thessaly) plain, central Greece, in which plant growth and crop productivity data were collected throughout the growing season of 2011. Switchgrass cultivation reached an average yield of 64 and 24.6 t ha1 fresh and dry biomass, respectively. The averaged growth rates of the fresh and dry biomass from the emergence to the harvest period were 267 and 132 kg ha-1 d-1, respectively. These rates were higher from emergence to the stage of anthesis (437.5 and 168.4 t ha-1 d-1, respectively). Considering the biomass yield of switchgrass under the low inputs required and the prices of biomass for energy purposes or the prices for feedstock, it can be concluded that switchgrass may be comprise an alternative energy crop in Greece in the immediate future. Keywords: Switchgrass, yield, biomass, irrigation, fertilization, Greece 1 INTRODUCTION Switchgrass is a perennial grass which has an average height of approximately 1.5 m, but can reach up to three meters under favorable conditions. The root system is extensive and reaches a depth of three meters [1], while produces many new culms each year, which enrich the soil with organic matter. The underground biomass production, the full development of culture, is equal to or greater than the aboveground [2]. Crop management depends mainly on soil and climatic conditions of the region, which will determine the choice of the genotype and the variety. The relationship genotype - environment has proven to be very significant [3]. Switchgrass prefers deep soils with good water holding capacity. However, it can grow in a variety of soils. Contrast thrives in very heavy clay soils [4]. The biomass yield is significantly reduced in soils of high pH. The pH range that switchgrass prefers, is between 4.5 and 7.6 [5]. Despite the wide range of soil where switchgrass can grow, it has been shown that biomass production can significant vary, even within the same field [6]. The main sources that affect switchgrass productivity, except sunlight, are water and nitrogen. The requirements of nutrients mainly depend on soil fertility, climatic conditions, anticipated production and the nature of the 187

cultivation [7]. The main nutrient that affect biomass yield is nitrogen. Nitrogen management is a very important parameter because affect the final yield and production costs [8]. Water is another important factor that affects the performance of the crop. A special feature of switchgrass, is the rich root system which reaches a depth of 3 meters and helps it to exploit moisture from greater depths. Stroup et al. found that the level and availability of other factors affecting production determines the minimum water requirement [9]. The aim of this study was to test the adaptability of the perennial plant switchgrass, in central Greece and more specifically in the plain of Thessaly. Main research topic was the growth and development of the crop under different irrigation and N-fertilization treatments. 2 MATERIALS AND METHODS The experiment was conducted at Palamas area, in west Thessaly plain, central Greece. With coordinates 39o04' N, 22o04'E and altitude of 107.5m asl. Palamas soil is characterized as a deep, calcareous sandy loam to loam, fertile soil representing a large part of the West Thessaly Plain. The tested variety was the Alamo and the applied seed quantity was 7 kg ha-1 using a cereal seeding machine on 2009. This means that the cultivation was on its 3rd year of establishment. The experiment was designed as

27/6/2011(2 sampling)

178

804

11/7/2011(3rd sampling)

192

1036

25/7/2011(4th sampling)

206

1301

th

9/8/2011(5 sampling)

221

1556

29/9/2011(6th sampling)

272

2326

3

Ten days period Precipitation

Temperature oC

September

557

1

162

nd

11/6/2011(1 sampling)

2

0

July

83

st

1

24/3/2011(Emergence)

2

GDD

May

JD

1

DATES

30 25 20 15 10 5 0

100 80 60 40 20 0

March

Table 1. Sampling dates and growing degree days (GDD)

2

Precipitation (mm)

Proceedings of SEEP2015, 11-14 August 2015, Paisley randomized split plot with four replications and Average air temperature and precipitation eight plots per replication. 120 35

Average Temperature

Figure 1. Ten days average temperature and precipitation recorded at Palamas during the growing season in ten days

RESULTS AND DISCUSION

3.1 Climatic conditions Figure 1 illustrates 10-day average air temperature and precipitation data prevailing during the growing season of 2011 at Palamas site. As it is shown the mean air temperature during the whole growing period was 20.6oC. It has to be mentioned that the maximum air temperature during the summer was higher than 35C, while the minimum air temperature ranged to 16-21C. The precipitation during the last ten days of March and April was around 60mm. The higher precipitation was noticed on the last ten days of May with an amount of almost 100mm. Thereafter July was a drought month, while in the first ten days of August a precipitation of 45mm was noticed. The total precipitation during switchgrass growth was 295mm. Therefore, from the above it can be concluded that the irrigation levels of the experiment may not be able to show clearly if switchgrass cultivation can be used as rainfed cultivation or if it is necessary to be irrigated.

3.2 Plant height The emergence (>50%) of switchgrass was on 24/3/2011. As it is shown in Table 2, there was not noticed any significant difference. The highest height was noticed at the final sampling on 29/9/2011 (2,06m) for the treatment with fertilization of 240 kg N ha-1 (Ι2Ν1). The average growing rate of the height from the emergence till the 5th sampling (9/8/2011) was 1.45 cm per day. The reason that there was not found any significant difference was the high precipitation and air temperature that were prevailed. Table 2. Plant height for the irrigation and Nfertilization levels Samplings Irrigation levels Rainfed Irrigated (250 mm) LSD0.05 Fertilization levels (kg N ha-1) 0 80 160 240 LSD0.05

2nd

3rd

4th

5th

6th

1.52 1.62

1.66 1.72

1.89 1.92

1.92 2.03

2.01 2.01

ns

ns

ns

ns

ns

1.36 1.63 1.66 1.62 ns

1.61 1.64 1.73 1.78 0.12 2

1.84 1.85 1.93 2.00 ns

1.92 1.95 2.01 2.02 ns

2.05 1.92 2.01 2.06 ns

3.3 Fresh and Dry Weight Swithgrass was growing with high rates and the average fresh weight at beginning of June was up to 37.4 ton ha-1. During the period end of June – end of July, where the precipitation was equal to 0mm, the rainfed treatment had decreased growth rate in comparison with the irrigated. The average growth rate for the period from emergence (24/3/2011) till flowering 188

Proceedings of SEEP2015, 11-14 August 2015, Paisley (22/7/2011) was 420 kg day ha . Growth rate is 4 CONCLUSIONS decreasing for the period from flowering till the Switchgrass cultivation has a particular interest mature stage. As it is illustrated in Figure 2 the as an alternative crop that will have a primary higher fresh weight was noticed on 9/8/2011 (5th role in biomass production. The plant has low sampling) for the irrigated unfertilized treatment. water and nitrogen requirements. Switchgrass is growing with high rates and it can reduce them under adverse conditions. 70 As a perennial crop, and due to the rich and deep 60 root system, switchgrass has favourable effects 50 on soil erosion and protection against 40 desertification, a future problem that will face many regions. Finally it has to be mentioned that 30 further in depth research must be carried out for 20 this crop; to study more factors that may affect 10 the final yield. -1

-1

0 50

100

150

200

250

300

Figure 2. Plant fresh weight as affected by 2 irrigation and 4 N-fertilization levels (■ I1N1, ♦ I1N2, ▲ I1N3, ■I1N4, × I2N1, Ж I2N2, I2N3, + I2N4) in 2011. The growth rate of the dry weight from the emergence till the beginning of July was 300 kg day-1 ha-1. Due to the drought conditions that were observed during July the rainfed treatments decreased their growth rates and this had as result to find significant differences between the irrigation levels in the 5th sampling, as it is shown in Table 3. Thereafter, due to the precipitation that was observed at the first ten days of August, the rainfed treatments increased again their growth rates. This had as result to found non-significant differences to the final sampling on 29/9/2011. Table 2. Switchgrass dry weight as affected by 2 irrigation and 4 N-fertilization levels Samplings Irrigation levels Rainfed Irrigated (250 mm) LSD0.05 Fertilization levels (kg N ha-1) 0 80 160 240 LSD0.05

2nd

3rd

4th

5th

6th

11.6 11.8

18.3 17.5

21.2 20.5

19.3 27.4

23.2 26.5

ns

ns

ns

1.84

ns

9.3 10.2 13.1 14.1 ns

15.7 18.6 19.1 18.4 ns

20.5 18.0 22.0 22.9 ns

21.8 22.4 23.6 25.8 ns

22.5 26.2 27.7 22.9 ns

189

REFERENCES [1] M. A. Liebig, H. A. Johnson, J. D. Hanson and A. B. Frank, “Soil carbon un-der switchgrass stands and cultivated cropland”, Biomass and Bioenergy, Vol. 28, pp. 347-354, 2005. [2] A. B. Frank, J. D. Berdahl, J. D. Hanson, M. A. Liebig and H. A. Johnson, «Biomass and carbon partitioning in switchgrass», Crop Science, Vol.44, pp. 1391-1396, 2004. [3] J. H. Fike, P. J. David, D. D. Wolf, J. A. Balasko, J. T. Green, M. Rasnake M. and J. H. Reynolds, «Long-term yield potential of switchgrass for biofuel systems», Biomass and Bioenergy, Vol.30, pp. 198-206, 2006. [4] H. Eldersen, D. Cristian, N. Bassam, G. Sauerbeck, E. Alexopoulou, N. Sharma, I. Piscioneri, «A management guide for planting and production switchgrass as a biomass crop in Europe», 2nd Conference on Biomass for Energy Industry and Climate Protection, 10-14 May 2004, Rome Italy, 2004. [5] N. Virgilio, A. Monti and G. Venturi, “Spatial variability of switchgrass”, (Panicum virgatum L.) yield as related to soil parameters in a small field. Field Crops Research, Vol.101, pp. 232-239, 2007. [6] J. R. Kiniry, K. A. Cassida, M. A. Hussey, J. P. Muir, W. R. Ocumpaugh, J. C. Read, R. L. Reed, M. A. Sanderson, B. C.Venuto, J. R. Williams, «Switchgrass simulations by the ALMANAC model at diverse sites in the southern US», Biomass and Bioenergy, Vol. 29, pp. 419-425, 2005. [7] R. Lemus, E. C. Brummer, C. Lee Burras, K. J. Moore, M. F. Barker, M. E. Neil, “Effects of nitrogen fertilization on biomass yield and

Proceedings of SEEP2015, 11-14 August 2015, Paisley quality in large fields of established switchgrass in southern Iowa, USA”, Biomass and Bioenergy, Vol. 32 (12), pp. 1187-1194, 2008. [8] R. G.Nelson, J. C. Ascough and M. R. Langemeier, “Environmental and economic analysis of switchgrass production for water quality improvement in north-east Kansas”, Journal of Environmental Management, Vol. 79, pp. 336-347, 2006. [9] J. A. Stroup, M. A. Sanderson, J. P. Muir, M. J. McFarland and R. L. Reed, “Comparison of growth and performance in upland and lowland switchgrass types to water and nitrogen stress”, Bioresource Technology, Vol.86, pp. 65-72, 2003.

190


THE EFFECT OF FERTILIZERS CONTAINING UREASE INHIBITOR (AGROTAIN) ON THREE DIFFERENT INDUSTRIAL CROPS’ PRODUCTIVITY D. Bartzialis, E. Skoufogianni, K.D. Giannoulis, S. Kakarantzas, I. Kalkounou, and N.G. Danalatos 1. University of Thessaly, Dept. of Agriculture, Crop Production & Rural Environment, Volos, Greece, responding author email: [email protected] 2. DIM. S. GAVRIEL & CO LTD Abstract In this research was studied the effect of fertilizer types containing urease inhibitor (AGROTAIN) and conventional fertilizers at 3 different N-levels on durum wheat, maize and cotton productivity at 2 different sites (Palamas and Velestino) in central Greece. Urease inhibitor ensures crops N-nutrition for longer period compared to conventional fertilizers. This results in better utilization of supplied nitrogen, achieving ultimately higher yields. In all cases, was found supremacy of fertilizers containing urease inhibitor against conventional fertilizers for the majority of the studied characteristics. Durum wheat and maize that were fertilized with fertilizers containing urease inhibitor reached higher biomass-grain yield and chlorophyll content. Even cotton, which is a hermit plant and covers a large part of its needs from soil available nitrogen, gave higher seed cotton performance and chlorophyll content on fertilizers containing urease inhibitor. The differences between the examined fertilizers are possibly due to smoother and stable N-nutrition and the higher photosynthesis rates. Finally, it must been mentioned that the presented results belongs to the 1st experimentation year where only the trend can be noted. Safer conclusions should be drawn from the results of the 2nd experimentation year, where the residual effect of the previous crop will be absent. Keywords: Urease inhibitor, wheat, maize, cotton, fertilization, yield 1 INTRODUCTION The global demand for food is expected to increase while it has been reported that this increase may rich the double by 2050 [1, 2]. However, it is reported that yields of important crops are stagnating [3, 4 5]. Therefore, it is critical to understand the yield differences between potential and actual yield (e.g. harvested yield) [6, 7]. The knowledge about yield potential and actual yield will help to guide to a sustainable intensification of agriculture [8]. Nitrogen fertilizer is essential for the high rate of food production delivered by modern agriculture. It contributes 20–80 billion of profit per year for EU farmers [9]. The use of synthetic N fertilizers has increased in response to intensification of agricultural systems [10]. Farmers apply lots of different N fertilizers such (e.g. urea, ammonium nitrate, ammonium sulphate, di-ammonium phosphate etc) to increase yields but this brings with it high associated N losses from agricultural soils to the environment [11]. Nitrogen is a mobile nutrient and susceptible to gaseous losses and leaching. However, the increase in N use, with N-response 191

efficiency reported to be between 33 and 50%, is contributing to higher worldwide N losses via NH3 volatilization and NO3− leaching [12, 13, 14]. Such a low N response efficiency shows that a large percentage of the applied fertilizer N is not being used for productive purposes and is lost to air, water, having negative impact to the quality recipient ecosystems [15, 16, 17], while increases production costs [18]. In the crowd of the different types of N fertilizer that exist, urea has become the predominant source of inorganic N used throughout the world [15], meeting almost half of the world's N requirement. Continued growth is expected in the use of urea fertilizer owing to its high-N content and ease of application in a dry granular form or as an aqueous solution. Nevertheless, there is a need to improve the efficiency of ureabased fertilizers through new technologies and management approaches due to its emissions, the environmental pollution and the increasing production cost. One of the most promising approaches is to apply urea in combination with urease inhibitor at low concentrations [19, 20]. During the last 2 decades a variety of compounds have been reported as urease

Proceedings of SEEP2015, 11-14 August 2015, Paisley Sowing of durum wheat took place using a inhibitors. The use of urease inhibitors retards modern seeding machine applying 20 kg ha-1 at urea hydrolysis by the soil enzyme urease and the end of November 2013 and the variety was allow the urea to diffuse further into the soil. the “Simeto”. Corn and cotton was sown using a Urease inhibitor (NBPT) is commercially pneumatic precision seeder machine “Gaspardo available under the trade name of Agrotain. 520”, in both sites. The sowing of corn held at Agrotain applications have been reported to be distances of 75cm between rows and 15cm on effective in delaying urea hydrolysis as well as each line at the end of March, while the hybrid increasing productivity [21, 22, 23]. Urease “PR32P26” of Pioneer Hi-Bred was used. inhibitors inhibit the enzyme urease, decrease Sowing of cotton held at distances of 95cm the urease activity and block the hydrolysis of between rows and 4,5cm on each line in both urea to NH3 [24]. Urea can damage the seedlings regions during the first week of May, where the after it hydrolyses by the enzyme urease, where “Flora” variety of Bayer Crop Science was used. the produced ammonia (NH3) and ammonium In both regions was performed pre- and post(NH4+) can cause ammonia toxicity and osmotic emergence herbicide application, as well as damage [25]. Urea toxicity can be reduced by manual control of weeds. applying urease inhibitor to the fertilizer granule Basic fertilization took place one-two days [26, 27, 28]. before sowing using a dispenser and then the Finally, this study was designed to evaluate the fertilizer was incorporates using a rotary effect of different fertilizers containing urease cultivator. Finally, the irrigation dose for the inhibitor on the yield of the main arable crops emergence applied using a sprinkler system and (durum wheat, cotton and maize) in the main then a drip irrigation system was established. agricultural plain (Thessaly) in Greece. 2 MATERIALS AND METHODS For the purposes of the project, field experiments were established at two sites in East Thessaly (Velestino, Volos) and West Thessaly (Palamas, Karditsa). The selected crops, to assess the impact of a new fertilizer type in their performance were durum wheat, corn and cotton, which are the most prevalent arable crops in Greece. 2.1 Soil characteristics Velestino soil is a clay loam (sand 19-21%; clay 39-41%, silt 38-42%) calcareous (pH = 8.1-8.3) rich in organic matter (2.3-2.7% in soil profile of 40cm). On the other hand, Palamas soil is a deep, sandy loam to loamy (37-45% sand, clay 51-43%, silt 12%), calcareous (pH = 8,3), poor (organic matter content 0.9% in soil profile of 40cm). Furthermore, Palamas area is characterized by a shallow underground aquifer and is classified as Aquic Xerofluvent, while the soil in Velestino as Calcixerollic Xerochrept according to USDA (1975). 2.2 Cultivation practices The experimental plots were demarcated by fixed points both on the outer perimeter and the sub-plots of each replication (block), as to be able to remain stable the treatments for the following year of the conducting experiments. The cultivation practices that took place were plowing, harrowing, seeding. 192

2.3 Experimental design 2.3.1 Durum Wheat The experimental design of durum wheat was a completely randomized design with 25 fertilization treatments and three replications (blocks), for both regions. In basic fertilization during seeding period, there were applied three different N-levels (60, 120 and 180 kg ha-1), using four different types of fertilizers (2 conventional simple: 20-10-0 and 16-20-0, and 2 with urease inhibitor: 30-15-0 and 20-24-0). The rest amount of the N-fertilization (top dressing) was applied using two simple fertilizers (calcium ammonium nitrate 26-0-0 and ammonium nitrate 34,5-0-0), and two with urease inhibitor (40-0-0 and 46 -0-0). Of course in each block there was a plot of zero fertilization. 2.3.2 Maize The same experimental design was used for maize using different amounts and types of fertilizers. Therefore, three levels of Nfertilization was applied for basic fertilization (120, 240 and 360 kg ha-1), using two simple (20-10-0 and 27-7-5M+0.5 Zn) and two with inhibitor urease (30-15-0 and 24-8-8M+0.5 Zn). The top dressing applied using two simple fertilizers: the ammonium nitrate (34,5-0-0) and urea (46-0-0), while in the case of urease inhibitor were used the 40-0-0 and 46-0- 0. Of

2.5 Meteorological data Meteorological data were recorded in Velestino from the established meteorological station of University of Thessaly, while the meteorological data in Palamas from the meteorological station of NAGREF in Karditsa. 3

RESULTS AND DISCUSION

3.1 Meteorological data In Figures 1, 2 are illustrated the average temperature and precipitation during crop growth (durum wheat, maize, cotton) at Velestino and Palamas, respectively.

193

20

30

15

20

10

10

5

0

0

Temperature (oC)

40

1 OCT 2 3 1 NOV 2 3 1 DEC 2 3 1 JAN 2 3 1 FEB 2 3 1 MAR 2 3 1 APR 2 3 1 MAY 2 3.0 1.0 JUN 2 3.0 1 JUL 2 3 1 AUG 2 3 1 SEP 2 3 1 OCT 2 3

25

Precipitation

Average annual Precipitation

Temperature

Average annual Temperature

Figure 1. Average air temperature and precipitation at Velestino Palamas 2013-2014 30

70

25

60 20

50 40

15

30

10

Temperature (oC)

80

20 5

10 0

0 1 NOV 3 1 DEC 2 3 1 JAN 2 3 1 FEB 2 3 1 MAR 3 1 APR 2 3 1 MAY 3.0 1.0 JUN 2 3.0 1 JUL 2 3 1 AUG 2 3 1 SEP 2 3 1 OCT 2 3 1

2.4 Yield To calculate Durum wheat yield in Velestino 1m2 was harvested by hand in each plot and after that a threshing of whole plot took place using an experimental harvester machine of the University of Thessaly on 20 June 2014. In Palamas site only 1m2 was harvested by hand. In case of maize and cotton, 3,75m2 and 3,8 m2 were harvested, respectively.

50

10 days

Precipitation (mm)

2.3.3 Cotton In cotton case was used a completely randomized experimental design with 37 treatments for both regions, in three replications blocks. There were applied three levels of Nfertilization for basic fertilization (70, 140 and 210 kg ha-1), using six types of fertilizers. Three of them were simple (20-10-0, 16-20-0 and 1515-15) and the other three with urease inhibitor (30-15-0 and 20-24-0 and 15-15-15). The top dressing applied on the plots with the simple basic fertilization using the simple fertilizers ammonium nitrate (34,5-0-0) and urea (46-0-0), while for the other plots the fertilizers 40-0-0 and 46-0 -0 with urease inhibitor were applied. Also in cotton case, in each block there was a plot of zero fertilization.

Precipitation (mm)

Proceedings of SEEP2015, 11-14 August 2015, Paisley course in each block there was a plot of zero Velestino 2013-2014 fertilization. 60 30

10 μέρες Precipitation

Average annual precipitation

Temperature

Average annual Temperature

Figure 2. Average air temperature and precipitation at Palamas 3.1.1 Durum wheat Low air temperatures were prevailing during the sowing of durum wheat, resulting in delayed germination without particular problems with respect to the final plant population in both sites. Notable rainfall occurred mainly in the first ten days of March (53 and 90 mm for Velestino and Palamas, respectively), as shown in Figures 1, 2, which favored the first growing stages and development of wheat in conjunction with the rise in temperature and application of top dressing. Later on, at Velestino a dry season was followed until the end of April which suspended the further growth and development. It was clearly viewed in field that the crop needed water especially at the second ten days of April.

3.1.3 Cotton The progress rainfall in the last days of April and the low temperatures during that period, delayed the sowing of cotton in about a week. After sowing, irrigation was applied to assist germination, and to lead to successful establishment of the crop with the desired plant population. There were not presented extreme weather events during crop development, except the heavy rain the first ten days of September at Velestino, which hampered a uniform and satisfactory mature of cotton.

Fertilizers

194

3130 3860 4070 3730 204

20-10-0 & 26-0-0

65

23,8

10700

4020

30-15-0 & 40-0-0 Inhibitor

62

27,5

10840

4210

16-20-0 & 26-0-0

60

26,4

9840

3840

20-24-0 & 40-0-0 Inhibitor

62

27,2

10320

4040

20-10-0 & 34,5-0-0

64

24,3

9500

3820

30-15-0 & 46-0-0 Inhibitor

59

26,8

9220

4010

16-20-0 & 34,5-0-0

58

25,4

9060

3360

20-24-0 & 46-0-0 Inhibitor

59

28,3

9130

3780

LSD0.05

ns

2,52

1049

333

CV (%)

8,2

10,1

11,2

8,9

3.2 Yield and growth characteristics 3.2.1 Durum wheat The observation of plant height (Table. 1), it appears that the supply of 30 kg ha-1 in spring applied at the low N-level (60 kg ha-1) gained an increase in height about 8 cm with respect to the control (no fertilization). The supply of higher N-doses in combination with the drought period created a toxic environment in the soil, inversely proportional increase in height relative to the applied nitrogen. As for biomass production and grain yield, the amount of 120 kg ha-1 found to have statistically significant superiority -1 compared to the level of 180 kg ha . It seems that the drought period was particularly critical for the crop, while rainfall towards the end of April failed to reverse the already formed state of cultivation. However, this amount of

Grain (kg ha-1)

Biomass (kg ha-1)

Chlorophyll

Nitrogen (kg ha-1)

Height (cm)

Proceedings of SEEP2015, 11-14 August 2015, Paisley water that applied during the delayed Finally, the rainfall that occurred in late April precipitation helped the crop to reach the yield prevented the crop destruction. At Palamas of 4000 kg ha-1. unlike Velestino, trop benefited from the precipitation in early April (about 50 mm), Table 1. Height, chlorophyll content, resulting in achieving satisfactory yields at biomass and grain yield of durum wheat harvest. under different N-fertilization levels and fertilizer combinations at Velestino. 3.1.2 Maize The germination period at Velestino and \ Characteristic Palamas was helped by the low precipitation Factor occurred in early April (10mm at Velestino and 40mm at Palamas) gaining satisfactory germination in both regions. Although there was 0 56 18,6 8310 occurred some rainfall during the summer, crop 60 64 24,5 9940 required significant addition of water through 120 61 25,9 10230 irrigation for satisfactory crop development. 180 59 28,3 9320 Finally, the rainfall in September especially at Velestino delayed the harvest. 3,3 1,04 673 LSD0.05

Different combinations of simple and with urease inhibitor fertilizers did not show any clear trend (data not shown). Instead, only numerical superiority of fertilizers with inhibitor was noticed, leading to thoughts of better power plant with nitrogen. Moreover, at Palamas site, the observation of plant height (Tab. 2), it also appears that the supply of 30 kg ha-1 in spring applied at the low N-level (60 kg ha-1) gained an increase in height about 6 cm with respect to the control (no fertilization). The supply of higher N-doses gave a slight increase in plant height of about 2-3 cm from the lowest to the highest level but this was not significant. At Palamas site significant rainfall benefited crop, giving greater plant height relative to Velestino. The three fertilization

Grain (kg ha-1)

Biomass (kg ha-1)

Chlorophyll

Fertilizers

Nitrogen (kg ha-1)

Height (cm)

Proceedings of SEEP2015, 11-14 August 2015, Paisley levels gave significant differences in chlorophyll efficient combination is the basic fertilization but not in biomass and grain yield, even though with 20-24-0 and top dressing with 46-0-0. the second nitrogen level (120 kg ha-1) gave higher amounts (200 kg ha-1 biomass and 60 kg 3.2.2 Maize ha-1 grain) than the first N-level (60 kg ha-1). The -1 Plant height was not shown statistically third N-level (180 kg ha ) showed no difference -1 significant differences between different than the second one, indicating that 120 kg ha are sufficient especially at Palamas site where a nitrogen levels, nor among the different fertilizer legume crop was cultivated before wheat combinations (Tab. 3). Combinations of seeding. This is also evident from the control fertilizer with urease inhibitor seem to produce where 3710 kg ha-1 seeds were produced. higher plants in some cases, but without Table 2. Height, chlorophyll content, significant difference. In case also of chlorophyll biomass and grain yield of durum wheat measures, although fertilizers with urease under different N-fertilization levels and inhibitor showed higher chlorophyll levels over fertilizer combinations at Palamas. simple fertilizers, the differences were not Characteristic statistically significant. Finally, the grain yield, Factor which is the economic product of the maize, produced higher yields but not statistically significant in the case of using fertilizers with 0 75 21,3 9100 3710 urease inhibitor. 60 81 24,2 11940 4840 In the case of the of the three N-fertilization 120 83 26,2 13700 5430 levels it was found that the supply of 120 kg ha-1 180 86 29,1 13300 5390 to maize produced 4000 kg ha-1 seed, the level of ns 2,37 ns ns 240 kg ha-1 increased the yield for 2500 kg ha-1 LSD0.05 and final the level of 360 kg ha-1 reached the 20-10-0 & 26-0-0 87 24,6 11400 4770 yield of 9000 kg ha-1. Even if the seed yield is 30-15-0 & 40-0-0 84 27,4 13640 5460 inhibitor increasing by increasing the fertilization level it 16-20-0 & 26-0-0 83 26,8 12180 4970 is cleary shown that the more we apply fertilizer 20-24-0 & 40-0-0 the smaller is the degree of performance. 84 28,1 13300 5530 inhibitor At Palamas site maize height had the same trend 20-10-0 & 34,5-0-0 85 25,6 13150 4850 as at Velestino (Tab 4). Among the different 30-15-0 & 46-0-0 81 26,4 14270 5650 fertilization levels it was found significantly inhibitor 16-20-0 & 34,5-0-0 83 25,6 12260 4940 difference of the lower level (120 kg ha-1) 20-24-0 & 46-0-0 compared to the others. The combination of 82 27,4 13680 5580 inhibitor fertilizers with urease inhibitor against the ns 1,82 1778 638 LSD0.05 simple shown plants of bigger height in the three 6, 7,2 14,4 12,9 CV (%) 1 of the four cases, but there was not found significant differences. In case of chlorophyll the Therefore, it could be concluded that the Nthree nitrogen levels statistically differ with the fertilization level of 120 kg ha-1 was more highest level having higher measures which profitable, while fertilizers with urease inhibitor means greenest plants. produced higher biomass and ultimately higher Total biomass production was increased and seed yield in durum wheat. Moreover, demonstrates a statistically significant upward chlorophyll measurements showed better plant trend since the lowest to the highest Nnutrition which fertilized with fertilizers with fertilization level. Growth biomass rate urease inhibitor. Finally, the most efficient decreased with nitrogen dosing. Comparison as combination of fertilizers is the basic with 30pairs between the simple and the fertilizers 15-0 and the top dressing with 40-0-0 at Velestino site, while at Palamas the most 195

Chlorophyll

Biomass (kg ha-1)

Seed yield (kg ha-1)

9610

120

236

51.1

32720

13760

240

244

57.8

36020

16260

360

245

63.5

37540

17040

LSD0.05

3.3

6.30

1413

694

238

55.8

34060

14820

0

196

14.0

18530

7320

30-15-0 & 40-0-0 inhibitor

243

59.7

38010

16370

120

231

46.7

29860

11210

26-7-5 & 34,5-0-0

244

54.9

33330

14640

24-8-8 & 40-0-0 inhibitor

248

60.3

35720

16340

20-10-0 & 46-0-0

236

55.1

35220

15160

30-15-0 & 46-0-0 inhibitor

241

59.0

37410

16690

26-7-5 & 46-0-0

243

56.3

33850

15260

24-8-8 & 46-0-0 inhibitor

241

58.7

35800

16240

LSD0.05

ns

ns

ns

ns

CV (%)

5.4

10.7

10.2

12.2

.

240

238

55.2

32280

13720

360

236

61.8

33350

15190

ns

4.18

ns

2999

20-10-0 & 34,5-0-0

234

52.4

30070

12350

30-15-0 & 40-0-0 inhibitor

237

57.5

32670

14220

26-7-5 & 34,5-0-0

237

50.0

31120

12790

24-8-8 & 40-0-0 inhibitor

237

55.8

32400

13850

20-10-0 & 46-0-0

228

52.4

30810

13010

30-15-0 & 46-0-0 inhibitor

234

57.1

26-7-5 & 46-0-0

234

53.3

30310

12960

24-8-8 & 46-0-0 inhibitor

239

58.2

35230

13890

LSD0.05

ns

ns

ns

ns

CV (%)

4.4

11.0

12.9

18.0

LSD0.05

32020

Fertilizer

Seed yield (kg ha-1)

22110

Biomass (kg ha-1)

18.3

Chlorophyll

213

Height (cm)

Nitrogen (kg ha-1)

0

20-10-0 & 34,5-0-0

Characteristic Factor

Fertilizer

Nitrogen (kg ha-1)

Table 3. Plant height, chlorophyll content, biomass and grain yield of maize under different N-fertilization levels and fertilizer combinations at Velestino

Height (cm)

Proceedings of SEEP2015, 11-14 August 2015, Paisley containing urease inhibitor showed superiority Table 4. Plant height, chlorophyll content, -1 biomass and grain yield of maize under of the second in the range of 200 to 400 kg ha . different N-fertilization levels and fertilizer Similarly to biomass were the seed yield results. combinations at Palamas There was found a statistically significant Characteristic difference between nitrogen levels and . superiority of fertilizers with urease inhibitor Factor -1 against the simple (100 to 170 kg ha ).

13910

Therefore, it could be concluded that Palamas and Velestino site shown the same results depending on the effect of the fertilizers which contain urease inhibitor. The N-fertilization level of 240 kg ha-1 using fertilizers with urease inhibitor lead to the same yield with the fertilization of 360 kg ha-1 using simple fertilizers, which means probably less leaching and production costs.

196

3.3 Cotton It is well known that cotton is a less demanding crop in nitrogen, therefore it is not expected any major differences in fertilization. Table 5 shows the chlorophyll content and the seed yield at Velestino. In chlorophyll cases there were found statistical significant differences between Nfertilization levels while among the different combinations of the different types of fertilizers there were not found statistically differences. The cotton seed yield was lower in the first Nlevel (70 kg ha-1) than the top two N-levels (140, 210 kg ha-1). Supplying a dose of 70 kg ha-1 leads to the seed yield of 3200 kg ha-1, while adding extra 70 kg ha-1 the final seed yield increases more or less 400 kg ha-1. Finally at the top N-level the seed yield remains at the same level as with the previous nitrogen dressing (Tab. 5). Among the different fertilizer combinations, compared them in pairs of similar simple and including urease inhibitor fertizers,

Proceedings of SEEP2015, 11-14 August 2015, Paisley recorded an increase in yield to the urease Table 6. Chlorophyll content and seed cotton yield of cotton under different Ninhibitor fertilizers by 4-11%. fertilization levels and fertilizer combinations Table 5. Chlorophyll content and seed at Palamas cotton yield of cotton under different NCharacteristic Cotton seed fertilization levels and fertilizer combinations Chlorophyll yield Factor at Velestino (kg ha-1) Chlorophyll

Fertilizer

0

33.1

2790

70

47.5

3500

140

50.3

3990

210

51.2

4050

1.37

265

0

31.4

2680

70

47.3

3200

140

49.9

3620

210

51.4

3570

20-10-0 & 34,5-0-0

47.7

3580

LSD0.05

0.74

327

30-15-0 & 40-0-0 inhibitor

49.4

3800

20-10-0 & 34,5-0-0

48.7

3310

16-20-0 & 34,5-0-0

49.8

3730

50.4

4030

LSD0.05

30-15-0 & 40-0-0 inhibitor

49.6

3500

20-24-0 & 40-0-0 inhibitor

16-20-0 & 34,5-0-0

49.3

3330

15-15-15 & 34,5-0-0

50.2

3820

20-24-0 & 40-0-0 inhibitor

50.4

3710

15-15-15 & 40-0-0 inhibitor

50.1

4140

15-15-15 & 34,5-0-0

49.6

3350

20-10-0 & 46-0-0

48.8

3530

49.4

3870

Fertilizer

Nitrogen (kg ha-1)

Factor

Cotton seed yield (kg ha-1)

Nitrogen (kg ha-1)

Characteristic

15-15-15 & 40-0-0 inhibitor

50.1

3520

30-15-0 & 46-0-0 inhibitor

20-10-0 & 46-0-0

48.9

3290

16-20-0 & 46-0-0

48.7

3700

51.9

4000

30-15-0 & 46-0-0 inhibitor

49.0

3640

20-24-0 & 46-0-0 inhibitor

16-20-0 & 46-0-0

48.1

3440

15-15-15 & 46-0-0

48.9

3800

20-24-0 & 46-0-0 inhibitor

50.8

3580

15-15-15 & 46-0-0 inhibitor

50.6

4150

15-15-15 & 46-0-0

48.8

3320

LSD0.05

2.09

ns

15-15-15 & 46-0-0 inhibitor

50.4

3550

CV (%)

4.5

11.8

LSD0.05

ns

ns

CV (%)

4.1

16.7

At Palamas site the differences in chlorophyll measures between N-levels showed statistically significant differences as at Velestino. Among the different combinations those of the fertilizers with urease inhibitor prevail in almost all cases. The results in cotton seed yield are the same as at Velestino site, while among the different fertilizer combinations, compared them in pairs of similar simple and including urease inhibitor fertizers, recorded an increase in yield to the urease inhibitor fertilizers by 6-10%.

197

Therefore, it could be concluded that Palamas and Velestino site shown the same results depending on the effect of the fertilizers which contain urease inhibitor. The N-fertilization level of 140 kg ha-1 using fertilizers with urease inhibitor lead to higher cotton seed yield comparing to the fertilization of 210 kg ha-1 using simple fertilizers, which means probably less leaching and production costs. 4 CONCLUSIONS It was recorded for all crops in both sites constant voltage supremacy of fertilizers with urease inhibitor against simple fertilizers in almost all the studied characteristics. The second N-fertilization level was the most effective, while the second N-fertilization level using fertilizers with urease inhibitor gained greater than or equal odds with the high N_fertilization

Proceedings of SEEP2015, 11-14 August 2015, Paisley fields: summary of available measurement data. level with simple fertilizers. Generally fertilizers Global Biogeochem. Cycles, Vol. 16, pp. 1080, 2002. with urease inhibitor increased efficiency about [12] W. R. Raun, G. V. Johnson, Improving nitrogen 10%. Due to the fact that the above results were use efficiency for cereal production. Agron. J. Vol. found through the first experimentation year, 91, pp. 357–363, 1999. safer conclusions expected to arise after the [13] V. C. Baligar, N. K. Fageria, Z. L. He, Nutrient processing of the second experimentation year use efficiency in plants. Soil Sci. Plant Anal. Vol. 32, results. In case that the results of the coming pp. 921–950, 2001. year will show statistically significant [14] Intergovernmental Panel on Climate Change superiority, then fertilizers with urease inhibitor (IPCC), Changes in Atmospheric Constituents and in should be proposed in future fertilization Radiative Forcing. Cambridge University Press, UK, schemes. 2007. REFERENCES [1] H. C. J. Godfray, J. R. Beddington, I. R. Crute, L. Haddad, D. Lawrence, J. F. Muir, J. Pretty, S. Robinson, S. M. Thomas, C. Toulmin, Food security: the challenge of feeding 9 billion people. Science, Vol. 327, pp. 812–818, 2010. [2] D. Tilman, C. Balzer, J. Hill, B. L. Befort, Global food demand and the sustainable intensification of agriculture. Proc. Natl. Acad. Sci. U. S. A. Vol. 108, pp. 20260–20264, 2011. [3] K. G. Cassman, A.D. Dobermann, D. Walters, H. Yang,. Meeting cereal demand while protecting natural resources and improving environmental quality. Annu. Rev. Environ. Resour. Vol. 28, pp. 315–358, 2003. [4] N. Brisson, P. Gate, D. Gouache, G. Charmet, F. X. Oury, F. Huard,. Why are wheat yields stagnating in Europe? A comprehensive data analysis for France. Field Crops Res. Vol. 119, pp. 201–212, 2010. [5] D. K. Ray, N. Ramankutty, N. D. Mueller, P. C. West, A. Foley, Recent patterns of crop yield growth and stagnation. Nat. Commun. Vol. 3, pp. 1293, 2012. [6] K. G. Cassman, Ecological intensification of cereal production systems: yield potential soil quality, and precision agriculture. Proc. Natl. Acad. Sci. U. S. A. Vol. 96, pp. 5952–5959, 1999. [7] D. B. Lobell, K. G. Cassman, C. B. Field, Crop yield gaps: their importance, magnitudes, and causes. Annu. Rev. Environ. Resour. Vol. 34, pp. 179–204, 2009. [8] M. K. Van Ittersum, K. G. Cassman, P. Grassini, J. Wolf, P. Tittonell, Z. Hochman, Yield gap analysis with local to global relevance–a review. Field Crops Res. Vol. 143, pp. 4–17, 2013. [9] M. A. Sutton, O. Oenema, J. W. Erisman, A. Leip, H. van Grinsven, W. Winiwarter, Too much of a good thing. Nature, Vol. 472, pp. 159–161, 2011. [10] J. W. Erisman, A. Bleeker, J. Galloway, M. S. Sutton, Reduced nitrogen in ecology and the environment. Environ. Pollut., Vol. 150, pp. 140– 149, 2007. [11] A. F. Bouwman, L. J. M. Boumans, N. H. Batjes, Emissions of N2O and NO from fertilized 198

[15] R. Harrison, J. Webb, A review of the effect of N fertilizer type on gaseous emissions. Adv. Agric. Vol 73, pp. 65–108, 2001. [16] R. Howarth, R. Marino, Nitrogen as the limiting nutrient for eutrophication in coastal marine ecosystems: evolving views over three decades. Limnol. Oceanogr. Vol. 51, pp. 364–376, 2006. [17] D. A. Turner, R. B. Edis, D. Chen, J. R. Freney, O. T. Denmead, R. Christie, Determination and mitigation of ammonia loss from urea applied to winter wheat with N-(n-butyl) thiophosphorictriamide. Agric. Ecosyst. Environ. Vol. 137, pp. 261–266, 2010. [18] B. Van der Stelt, E. J. M. Temminghoff, W. H. Riemsdijk, Measurement on ion speciation in animal slurries using theDonnanmembranetechnique. Anal. Chim. Vol. 552, pp. 135–140, 2005. [19] C. D. L. Rawluk, C. A. Grant, G. J. Racz, Ammonia volatilization from soils fertilized with urea and varying rates of urease inhibitor NBPT. Can. J. Soil Sci. Vol. 81, pp. 239–246, 2001. [20] A. Sanz-Cobena, T. H. Misselbrook, A. Arce, J. I. Mingot, J. A. Diez, A. Vallejo, An inhibitor of urease activity effectively reduces ammonia emissions from soil treated with urea under Mediterranean conditions. Agric. Ecosyst. Environ. Vol. 126, pp. 243–249, 2008. [21] C. J. Watson, N. A. Akhonzada, J. T. G. Hamilton, D. I. Matthews, Rate and mode of application of the urease inhibitor N-(n-butyl) thiophosphoric triamide on ammonia volatilization from surface-applied urea. Soil Use Manage. Vol. 24, pp. 246–253, 2008. [22] D. Chen, H. Suter, A. Islam, R. Edis, J. R. Freney, C. N. Walker, Prospects of improving efficiency of fertiliser nitrogen in Australian agriculture: a review of enhanced efficiency fertilisers. Aust. J. Soil Res. Vol. 46, pp. 289–301, 2008. [23] R. J. Martin, V. D. Weerden, M. U. Riddle, R. C. Butler, Comparison of Agrotain treated and standard urea on an irrigated dairy pasture. Proc. N.Z. Grassl. Assoc. Vol. 70, pp. 91–94, 2008. [24] V. H. Varel, Use of urease inhibitors to control nitrogen loss from livestock waste. Bioresour. Tech. Vol. 62, pp. 11–17, 1997.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [25] Bremner, J.M., 1995. Recent research on problems in the use of urea as a nitrogen fertilizer. Fert. Res. 42, 321–329. [26] C. A. Grant, L. D. Bailey, Effect of seed-placed urea fertilizer and N-(n- butyl) thiophosphoric triamide (NBPT) on emergence and grain yield of barley. Can. J. Plant Sci. Vol. 79, pp. 491–496, 1999. [27] S. S. Malhi, E. Oliver, G. Mayerle, G. Kruger, K. S. Gill, Improving effectiveness of seedrowplaced urea with urease inhibitor and polymer coating for durum wheat and canola. Commun. Soil Sci. Plant Anal. Vol. 34, pp. 1709–1727, 2003. [28] R. E. Karamanos, J. T. Harapiak, N. A. Flore, T. B. Stonehouse, Use of N-(n-butyl) thiophosphoric triamide (NBPT) to increase safety of seed-placed urea. Can. J. Plant Sci. Vol. 84, pp. 105–116, 2004.

199


AQUEOUS BATTERIES AS GRID SCALE ENERGY STORAGE SOLUTIONS Jorge Omar Gil Posadaa, Anthony J. R. Renniea, Sofia Perez Villara, Vitor L. Martinsa,b, David A. Worsleyc and Peter J. Halla a Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK b Instituto de Química, Universidade de São Paulo - C.P. 26077, CEP 05513-970, São Paulo, SP, Brazil c SPECIFIC, College of Engineering Swansea University, Baglan Bay Innovation and Knowledge Centre, Port Talbot SA12 7AZ, UK Abstract This paper explores selected rechargeable aqueous batteries as a means to store large amounts of energy coming from intermittent sources such wind power or solar energy. Herein, we identify challenges and opportunities for selected aqueous-based batteries (emphasising lead acid and Nickel-Iron cells) and their potential to help balancing the future electric grid. Lead acid batteries are by far the most widely used aqueous-based batteries on the market and represent a very mature battery technology; however, there remain some challenges that cast doubt on their future use at the large scale. Nickel-Iron batteries are well known for their long cycle life and ability to tolerate abuse, but require significant improvements in efficiency. Other technologies that use aqueous electrolytes and have the potential to be useful in large-scale applications are also introduced. Keywords: NiFe, electrolyte decomposition, cell performance, hydrogen evolution 1

providing services to support balancing and manage network utilization [15, 16].

INTRODUCTION

Due to climate change and the depletion of fossil fuel reserves, governments have started to reevaluate global energy policy. Therefore, we are experiencing an increasing demand of energy from renewable sources such as solar and wind power [1-3] and the majority of countries face challenges in the integration of an increasing share of energy coming from these intermittent sources [4-13]. Renewable sources (such as solar, wind power, etc.) are changing the energy market and they may displace significant amounts of energy that are currently produced by conventional means; this is, for example, an staggering 57% of the total demand of electricity in Denmark by 2025 [1], around 15% of the total UK energy demand by 2015 and almost 16% of China by 2020 [2]. Energy storage technologies are required to facilitate the move towards the supply of low carbon electricity [14], and are particularly useful when exploiting intermittent energy sources. Incorporating energy storage has been shown to be beneficial to various sectors of the electricity industry, including generation, transmission and distribution, while also 200

It is well known that organic electrolyte based batteries exhibit much larger energy and power density than their conventional aqueous based counterparts. Therefore, non-aqueous batteries has become the industry standard for most mobile applications (portable computers, smart telephones, etc.). Due to the flammable nature of organic solvents, safety measurements are of overriding importance when dealing with nonaqueous batteries. Implementing safety measurements to prevent short circuit conditions in your laptop’s battery (leading to a thermal runaway) is one thing, but securing energy for Heathrow airport or for an entire city is a completely different story. In large scale energy storage systems operational safety is of prime importance and characteristics such as energy (Wh/L) and power density (W/L), which are major drivers in the development of devices for mobile applications, are of lesser concern. Other desirable characteristics for large scale energy storage systems are a low installed cost, long operating life, high energy efficiency and that they can be easily scaled to meet increasing demand.

Proceedings of SEEP2015, 11-14 August 2015, Paisley Different battery chemistries demonstrated for 2 LEAD ACID BATTERIES Invented in 1859 by French physicist Gaston use at this scale include lead-acid, lithium-ion Planté, lead acid is the oldest type of a practical and sodium-based batteries. Lithium-ion and commercial rechargeable battery [20]. batteries exhibit very high round trip efficiencies Nowadays, this technology dominates the global (as high as 99%), energy densities in the range market for small-medium scale reversible energy of 100-200 Wh kg-1 and can typically withstand storage applications; in particular, they are 1000 cycles before fading [17]. Sodium-based widely used in automotive applications for batteries (sodium-sulphur, ZEBRA) operate at engine starting, lightning and ignition [21, 22]. temperatures in the region of 300-350ºC and are characterised by a round trip efficiency of 80%, In their most basic form, lead acid batteries energy densities up to 150 Wh kg-1 and lifetimes consist of a positive electrode composed of leadin excess of 3000 cycles [17]. dioxide (PbO2), a negative electrode composed of metallic lead (Pb), and a solution of sulphuric Not only are there safety concerns with these acid 10% acting as the electrolyte. During the lithium and sodium based cell chemistries, but discharge the cathode is reduced to Pb2+ that these technologies are also associated with high precipitate as PbSO4. costs due to the materials used, manufacturing processes and auxiliary systems required for their operation. As a result of these PbO2  4H   2e   Pb 2  2H 2 O E 0  1.46V considerations, the inherent safety and potential (1) low cost of the aqueous based electrochemical energy storage devices discussed in the Pb 2  SO42  PbSO4 (2) following sections, renders these systems extremely promising for large-scale applications. and the anode is oxidize to Pb2+ and also At present, lead acid cells are the most recognisable aqueous- based battery system and account for the majority of aqueous battery sales worldwide. For example, it was reported that during 2010 the use of lead acid batteries in China reached a staggering 75% usage of all new photovoltaic systems [18]; likewise, during 2008, lead acid technology held 79% of the US rechargeable battery market share [19]. This manuscript focuses on aqueous based electrochemical energy storage technologies suitable for large-scale applications and discusses some of the challenges faced in the development of viable systems. These technologies have the potential to be integral components in future electricity supply systems providing that substantial reductions in cost can be achieved, and that safe and reliable operation can be assured. Finally, the authors do not intend to rule out any competing technology, or to imply any of the batteries considered here are going to provide the ultimate solution for large scale energy storage. It is more likely that the final solution would consist of a combination of different technologies.

201

precipitate as PbSO4.

Pb  Pb 2  2e 

E 0   0.13V (3)

Pb 2  SO42  PbSO4 (4) Figure 1 provides a schematic representation of a lead acid cell [20, 23].

Figure 1. Schematic representation of a lead-acid cell

Proceedings of SEEP2015, 11-14 August 2015, Paisley Lead acid batteries are well known for their Negative Electrodes Most of the problems that are usually encounter small self-discharge properties, sophisticated with lead acid batteries, are strongly dependent production process, low cost of raw materials, upon the the negative electrode. Sulfation is, by recyclability, and high performance at high and far, the most common ageing phenomenon low temperatures. inherent to lead acid batteries. This problem consists in the formation of nearly insoluble After important improvements, as the use of a crystals of lead sulphate, which during re-charge valve-regulated system, the most significant would regenerate the electro-active material to a change to increase performance was the very small extent at best. introduction of carbon material in the anode together with Pb. The use of carbon additives in Basically, the negative electro-active material the Pb anode improves cyclability and life cycle have been mixed with a highly conductive due to minimization of the PbSO4 accumulation additive (usually of carbonaceous nature) to [24]. prevent lead sulfation while maintaining high electronic conductivity [21-23]; in addition, Collector Grids for Lead Acid Batteries some authors have nano-structured the negative A great deal of effort has been made in order to electrode with very promising results [40]. The improve the energy density and to reduce the exchange of lead anode by a carbon electrode weight of lead-acid batteries. In particular, was investigated, this assembly is similar to an ample work has been done in order to optimize asymmetric supercapacitor and resulted in a the composition of the electro-active material battery with higher cycle life [41]. and the structure of the collector grid. Current developments in grids for lead acid batteries hinge on the development of lighter grids, by electro-depositing layers of lead on highly conductive and low specific gravity substrates such as copper, aluminium, carbon, barium, indium, among other materials. [23, 2534]. Broadly speaking there are two main types of grid for the positive terminal for a lead-acid battery: lead-antimony and lead-calcium based grids. Unfortunately, lead-calcium grids render lead acid cells with a short deep-discharge cycle life. Likewise, lead-antimony grids are well known for reduced hydrogen overpotential, for which considerable amounts of hydrogen are produced especially during the charging of the battery. It has been reported that elements such as strontium, cadmium, silver and in general most rare earth elements can be used to produce lead-antimony or lead-calcium based alloys with enhanced performance [25, 35-38]. Vitreous carbons coated with lead have been used as a current collector for lead-acid batteries; however, due to oxygen evolution problems, these systems are not well suited as positive plate collectors [34, 39].

202

Another promising design is the use of a double anode, containing a foil of Pb and a second foil of carbonaceous material; such design allow the battery works at high power due to supercapacitor-like behaviour as at extend period in partial state of charge operation [41-43]. Although, the corrosion of the positive electrode has always been regarded as a major concern in lead-acid battery technology, the corrosion of the negative electrode has drawn attention recently, and scientists have recognized the importance of its control through the optimization of the formulation of the electroactive material [32, 33, 44]. Positive Electrodes In order to enhance the performance of the leadacid battery, low antimony grids are commonly used in most countries. Unfortunately, low antimony grids are prone to develop a passivation film between the grid and the electroactive material of the electrode; in addition, the corrosion of the positive electrode is well known to play a detrimental effect on the performance of the lead-acid battery. Therefore, the production of new formulations based on lead oxides [45], different sets of conductive [46], and non-conductive additives [46], play a pivotal role in controlling corrosion and

Proceedings of SEEP2015, 11-14 August 2015, Paisley NiFe cells. Figure 2 provides a schematic preventing passivation of positive electrodes for representation of a NiFe cell. lead acid batteries [47]. Electrolyte As is well known, sulphuric acid, the electrolyte in lead-acid batteries, actively participates in the cell reactions. As such, the concentration of sulphuric acid changes on battery charge and discharge. Research challenges Lead acid batteries are known for their low energy density (close to 30 Wh/kg which is only about 25% of lithium-ion batteries), reduced cycle life, weight of the battery (not crucial for stationary applications), toxicity, and low charge/discharge efficiency remain as major drawbacks inherent to this technology [22, 25, 48, 49].

3 NIFE BATTERIES Successfully developed and commercialised in the early 20th century, NiFe cells are secondary batteries that fell out of favour with the advent of cheaper lead acid cells, there is a renewed interest on this batteries due to their environmentally friendliness, longevity, and tolerance to electrical abuse. It is also believed this technology could provide cost effective solution for large scale energy where relatively low specific energy (in the order of 30-50 Wh/kg) is required. The relative abundance of the raw materials required to produce NiFe cells is another aspect favouring its use. Relatively easy to shape into different forms, iron is the fourth most abundant element in the Earth’s crust and probably the major component of its liquid core [50, 51]. Nickel is believed to be the second most abundant element in the Earth's core and large deposits of nickel ore can be found in countries such as Brazil, Russia, Philipines, Canada, Australia, Indonesia, etc. [52, 53]. Considered to be the 17th most abundant element on the Earth’s crust, sulphur is probably one of the least studied elements in upper mantle [54-57]. Finally, bismuth is considered the 64th most abundant element in the Earth’s crust, so it is not truly abundant [58, 59], but only small amounts of this element are required in the production of

203

Figure 2. Schematic representation of a NiFe cell Negative Electrodes The main process taking place during the charging process of an iron electrode under basic conditions is the reduction of ferrous ion (Fe2+) to elemental iron (Fe0). Similarly, the oxidation of elemental iron to ferrous ions, take place during the discharging process of the same electrode. Equation (1) illustrates the charging and discharging (forward and backward reactions respectively) processes of an iron electrode under strong alkaline conditions [60, 61]. Fe(OH ) 2  2e   Fe  2OH 

E 0   0.87V (5)

Unfortunately, during the charging of an iron electrode, water can be decomposed yielding hydrogen. So part of the energy that was originally intended to be stored in the cell, is wasted in decomposing the electrolyte. Through this process, electrolyte decomposition (and therefore hydrogen evolution) accounts for a drastic reduction in the overall performance of the battery, as indicated by Equation (2). 2H 2O  2e   H 2  2OH 

E 0   0.83V (6)

The mitigation of electrolyte decomposition on NiFe cells has been traditionally achieved by either modification of the iron electrode

Proceedings of SEEP2015, 11-14 August 2015, Paisley rechargeable Li batteries, also called 'rockingformulation or by the addition of elements (such chair batteries [71], has been the replacement of as sulphur or bismuth) that are capable to organic-based by aqueous-based electrolytes. increase the activation energy for electrolyte The advantage of aqueous batteries is the low decomposition [62, 63] cost materials and safety because aqueous electrolytes are mainly not flammables. The Positive Electrodes first group to report these aqueous systems in LiBasically, the positive electrode on NiFe cells is ion batteries was Dahn’s group in 1994 [72]. based upon nickel hydroxide. Two polymorphs They developed the system with a VO2 as anode of Ni(OH)2 exist, they are α-Ni(OH)2 and βand the spinel LiMn2O4 as cathode in 5M LiNO3 Ni(OH)2; they can be transformed into γ-NiOOH solution reaching an energy density around 75 and β-NiOOH, respectively. However, due to Wh/kg with an average voltage of 1.5, values the low stability of α-Ni(OH)2 in alkaline media, higher than in Pb-acid and Ni-Cd (30-45 Wh/kg) the β-Ni(OH)2 is usually used as a precursor batteries, but with a poor cycling life [73]. A material in alkaline batteries [64-66]. clear limitation in these batteries is the restricted stability of the stable voltage window of aqueous Electrolyte electrolytes. The decomposition of electrolyte or NiFe cells use strongly alkaline solutions of electrolysis of H2O occurs at 1.23 V and potassium and lithium hydroxide and selected involves H2 or O2 gas evolution. This potential additives (such as potassium sulphide) to prevent defines the stable operating voltage window in electrolyte decomposition. aqueous systems, although it is very low compared with that of the organic-based systems In general terms, the mitigation/prevention of (3.0 V for Li-ion batteries) indicating a lower Eq. (2) has been achieved by either modification energy density storage (75 Wh/kg) [73]. The of the anode or by the addition of electrolyte enhancement of the energy density storage have additives that are capable to increase the been addressed by expanding the operating activation energy for electrolyte decomposition, voltage window up to 1.23 V. Hou et al. reported such as wetting agents [67], long chain thiols values around 342 Wh/kg at average discharge [68], organic acids [69], etc., have been voltage of 3.32 V, which compete with the investigated [62]. potential of 3.0 V in organic batteries, when the lithium anode is covering by a polymer and Research challenges LISICON film. They acted as a protective Nowadays, the major concern about NiFe coating to avoid the formation of lithium batteries is the evolution of hydrogen, which dendrites and as a separator of the lithium metal renders low charge/discharge efficiency (in the from the aqueous electrolyte releasing lithium order of 50-60%) [60, 70], low specific energy ions [74]. Another strategy to increase the (30-50 Wh/kg), weight of the battery (not crucial energy density has been by developing for stationary applications), toxicity of nickel. alternative batteries which deliver a large energy density storage such as aqueous Li-air batteries (1910 Wh/kg) [75]. 4. AQUEOUS ‘ROCKING-CHAIR’ BATTERIES The complexity of the electrochemical and The replacement of organic-based by aqueouschemical reactions during the insertion/debased electrolytes offers the alternative to fulfil insertion processes in aqueous systems makes the need of low cost materials with higher ionic them a very challenging systems compared to conductivity. Several alkaline aqueous batteries organic-based batteries. There are many side (lead-acid, nickel-metal hydride, and nickel reactions that limit their performance and their metal) systems have been developed and long cycling life. The main drawbacks of this commercialized, but they fail meeting the technology are (1) the reaction of molecular standards for large-scale energy storage (in hydrogen or oxygen with electrolyte terms of toxicity, cycle life, etc.) [65]. + decomposition, (2) proton (H ) co-intercalation into the host electrode, (3) reactions between An additional approach solution based on the electrode material and water or residual concept of insertion electrode-based

204

Proceedings of SEEP2015, 11-14 August 2015, Paisley molecular oxygen, and (4) the dissolutions of ion insertion in detriment of the capacity electrode materials [76]. resulting in an obstruction of the Li ion pathways [78]. The structure of the host material To gain deeper insights into the intercalation is very important because each structure behave mechanisms, similar electrode materials different in the insertion processes. As an employed in non-aqueous batteries have been example, structures such as spinel Li1-xMn2O4 selected in the aqueous technology. and olivine Li1-xFePO4 cannot host H+ meanwhile layered structures (Li1-xCoO2, Li1Thus, to choose the convenient material as an xNi1/3Mn1/3Co1/3O2) presented a large amount of electrode in aqueous systems should be consider H+ concentration in the framework at acidic not only its stability at operating potential also a media [79, 80]. determined pH, which has strongly influence in the H2 and O2 evolution potentials. For instance, Moreover, dissolution of the electrode in the a theoretical method to evaluate the stability of electrolyte is other limiting factor for long-term the materials in these systems by using cyclability. As a solution the addition of a thermodynamic considerations, in function of protective surface coating onto the electrode has the pH and Li concentration has been reported demonstrated to be an effective approach to [77]. They demonstrate that in presence of O2, maintain the cycle life [81, 82]. negative electrode materials would theoretically be chemically oxidized by the O2 and H2O In the past decades, the increase of lithium instead to go through the electrochemical redox demand and its low earth availability has leaded process, resulting in capacity fading upon to find other elements most abundant and cycling. In the absence of O2, during the economical, and the sodium is the most insertion reaction the lithium ion intercalated promising candidate to replace it because its compounds may react with H2O as in the similar chemistries and ion sizes (90 pm for Na+ following reaction: and 116 for Li+) [83]. Whitaker’s group has developed important research inside of the Naion in aqueous media [84]. They found that the Li  H 2 O  2e   Li   OH   0.5H 2 (7) sodium could be reversible inserted into the tunnel structure Na0.44MnO2 delivering a The potential of the lithium ion intercalated capacity of 45 mAh/g at 0.125C. Later, they compound with H2O in function of the pH is as reported a large format hybrid/asymmetric follows: aqueous intercalation batteries by using λ-MnO2 as cathode materials and an active carbon in V ( x)  3.089  0.059 pH (V) (8) Na2SO4-based electrolyte (saltwater) but with low energy density and a wider operational In agreement to these theoretical results, it is range. possible to estimate that materials with a redox potential higher than equation (8) (2.626 V vs. Li+/Li of equilibrium voltage) are chemically stables and intercalation processes are occurring spontaneously in aqueous solution at a defined pH. For instance, the anode LiTi2(PO4)3 showed the lithium ion intercalation potential at 2.45 V vs. Li+/Li and theoretically it is not chemically stable in pH 7. By applying the previous consideration, the anode can became a stable material in absence of O2 if the pH is tuning to an alkaline media. For the case of cathode electrode materials, which are quite stables in water solutions, protons (H+) or water molecules insertion into the host structure are competing with the lithium 205

Recently, Chen et al. have shown that the use of Li+/Na+ mixed-ion electrolytes would render good stability [85]. In these systems, one side Li+ are released between the electrolyte and electrode whereas other side exchange Na+ between electrode and electrolyte. The concentration of Na+/Li+ is constant upon cycling. They used two system based on Li2SO4/Na2SO4 mixed electrolytes (LiMn2O4/Na0.22MnO2 and Na0.44MnO2/TiP2O7). Commercially systems are already in the market as an example is the supplier Aquion Energy. They employed a cubic spinel LiMn2O4 as cathode and activated carbon/NaTi2(PO4)3 composite as anode and they are able to have

Proceedings of SEEP2015, 11-14 August 2015, Paisley thousands of cycles without significant loss. The Zinc Air Primary zinc-air cells are a fairly mature smallest product (Aquion S-Line Battery Stack) technology that has found commercial with eight batteries in series have a nominal applications in medical and telecommunications. energy of 2.4 kWh at 20 hour of discharge and at As with other metal-air cells, a major driver for 30 C with a round trip of efficiency around development is their high theoretical energy >85%. density (1086 Wh kg-1 including oxygen). The compatibility of zinc with an aqueous alkaline electrolyte allows for substantially reduced 5. OTHER TECHNOLOGIES manufacturing costs in comparison with nonaqueous based cells. The development of Nickel-Cadmium electrically rechargeable zinc-air cells has been Exploiting the same reaction as the positive hindered by the propensity of zinc to form electrode in NiFe cells, are devices with alkaline dendrites upon repeated charge-discharge aqueous electrolytes that use metal hydride or cycling and their low output power [87]. A cadmium based negative electrodes. These further drawback of aqueous alkaline chemistries may be familiar, as they have been electrolytes is that carbon dioxide is absorbed by employed for many years in the consumer the solution and produces insoluble, electrode electronics sector, and were integral to the blocking compounds that reduce electrolyte development of electric vehicles in the 1990s. conductivity and impede cell performance. As a This represents a mature battery technology that consequence, the process of air purification has been identified as suitable for power quality needs to be considered alongside cell design. applications and grid support. An example of this technology is the system developed by Improvements in performance require the “GVEA” that uses almost 14,000 NiCd cells identification of suitable robust catalysts and providing backup power of 27MW for up to 15 electrolyte additives. Zinc-air cells have been minutes. This system has been in operation since proposed as a suitable alternative to lithium-ion 2003. Research efforts for these batteries are for use in electric vehicles and were successfully centred in preventing their gradual loss of energy demonstrated by “Electric Fuel” in 2004. capacity across cycles (memory effects), Currently, “Eos Energy Storage” are developing improving their energy density, and preventing a grid scale zinc-air system using a hybrid zinc self-discharge. In addition, the major drawback electrode and a near neutral pH aqueous of this technology is the toxicity of cadmium. electrolyte. Nickel Metal Hydride Recent improvements to cell architecture have Iron Air An alternative cell chemistry that has received focused on increasing the power density of attention of late is the iron-air cell that also NiMH cells and it remains a viable choice for operates in an aqueous alkaline electrolyte. Ironuse in light rail vehicles. NiMH cells are air cells do not exhibit the same characterized by energy densities in the region of -1 stripping/redeposition problem as seen in zinc250-330Whl , a specific energy of up to air cells but have a lower theoretical energy 100Whkg-1 and are limited to around 1000 density of 764 Wh kg-1 and electrically charge-discharge cycles. By comparison NiCd rechargeable cells exhibit relatively low energy cells can perform roughly twice the number of efficiencies (ca.35%) [88]. As with zinc-air cells cycles but are associated with a lower energy the development of more efficient oxygen density. A major drawback of this technology electrodes is required. has been identified by a recent study that showed NiCd cells being associated with substantially greater CO2 and SO2 emissions during Copper-Zinc Another noteworthy technology utilizing production, in comparison with lithium based aqueous electrolytes is the development of a cells [86]. Finally, these batteries are also rechargeable copper-zinc battery by “Cumulus susceptible to memory effects but to a lesser Energy Storage”. Based around processes used extent than their Ni-Cd counterparts. in metal refining, this project aims create safe,

206

Proceedings of SEEP2015, 11-14 August 2015, Paisley low cost battery systems with capacities in the ACKNOWLEDGEMENTS The authors would like to acknowledge the U.K. region of 1MWh to 100MWh. Engineering and Physical Sciences Research Council for supporting this work This battery is powered by the transfer of (EP/K000292/1; SPECIFIC Tranche 1: electrons from zinc atoms to copper ions, and it Buildings as Power Stations). consists of two metal plates, each in its own container of salt soluble ion (joined through a VLM thanks FAPESP (2014/14690-1) for salt bridge or semi-permeable membrane). One fellowship support. of the metal plates is zinc made, and it sits in a water solution of zinc sulphate. Likewise, the copper plate sits in a water solution of copper sulphate. REFERENCES [1] K. Hedegaard, P. Meibom, Renewable Energy, 37 (2012) 318-324. The major drawbacks of these batteries are the [2] H. Chen, T.N. Cong, W. Yang, C. Tan, Y. relatively large internal resistance (from which Li, Y. Ding, Progress in Natural Science, 19 only a moderate current can be taken from these (2009) 291-312. batteries). Deposition of copper in places other [3] A. Jacob, Reinforced Plastics, 45, than the copper electrode is a problem that Supplement 1 (2001) 10-13. renders the battery inactive. In addition, when [4] P. Seljom, A. Tomasgard, Energy the battery is not in action, copper (coming from Economics, 49 (2015) 157-167. the salt) diffuses and ends in contact with zinc, [5] M. Carrasco-Díaz, D. Rivas, M. Orozcothis process is followed by oxidation and Contreras, O. Sánchez-Montante, Renewable deposition of copper on the zinc plate as black Energy, 78 (2015) 295-305. cupric oxide (CuO), thus impairing the [6] J. Waewsak, M. Landry, Y. Gagnon, performance of the battery; this process is Renewable Energy, 81 (2015) 609-626. usually minimised by taking down the battery [7] J. Haas, M.A. Olivares, R. Palma-Behnke, when not in use. Journal of Environmental Management, 154 (2015) 183-189. [8] M. De Prada Gil, J.L. Domínguez-García, F. 6. CONCLUSIONS Díaz-González, M. Aragüés-Peñalba, O. GomisEnergy storage technologies are required to Bellmunt, Renewable Energy, 78 (2015) 467facilitate the adoption of renewable energy 477. sources. For large-scale applications, the safety [9] T.R. Ayodele, A.S.O. Ogunjuyigbe, and low installed costs of aqueous-based Renewable and Sustainable Energy Reviews, 44 batteries make them desirable propositions if (2015) 447-456. some of the limitations discussed above can be [10] S. Sun, F. Liu, S. Xue, M. Zeng, F. Zeng, overcome. Due to the existing manufacturing Renewable and Sustainable Energy Reviews, 45 capacity, lead acid cells are likely to remain a (2015) 589-599. viable option for many applications, however as [11] A. Ahadi, N. Ghadimi, D. Mirabbasi, this is a mature technology only incremental Journal of Power Sources, 264 (2014) 211-219. advances in performance are likely. More [12] M. Fisac, F.X. Villasevil, A.M. López, substantial improvements in performance with Journal of Power Sources, 252 (2014) 264-269. respect to efficiency, longevity and cost are [13] J.J. Vidal-Amaro, P.A. Østergaard, C. likely to be seen in other aqueous-based Sheinbaum-Pardo, Applied Energy, 150 (2015) chemistries, and these technologies have the 80-96. potential to be integral components in future [14] P. Gao, Y. Liu, W. Lv, R. Zhang, W. Liu, electricity supply systems. X. Bu, G. Li, L. Lei, Journal of Power Sources, 265 (2014) 192-200. In summary, aqueous systems are a clear route [15] D. Rastler, Electricity energy storage to energy storage that is emerging as an technology options: a white paper primer on alternative to other chemistry technologies. applications, costs and benefits, Electric Power Research Institute, 2010.

207

Proceedings of SEEP2015, 11-14 August 2015, Paisley Wark, E. Rühl, Journal of Power Sources, 222 [16] G. Strbac, M. Aunedi, D. Pudjianto, P. (2013) 554-560. Djapic, F. Teng, A. Sturt, D. Jackravut, R. [34] K. Ji, C. Xu, H. Zhao, Z. Dai, Journal of Sansom, V. Yufit, N. Brandon, Report for Power Sources, 248 (2014) 307-316. Carbon Trust, (2012). [35] D.G. Li, J.D. Wang, D.R. Chen, Journal of [17] D. Linden, New York, McGraw-Hill Book Power Sources, 235 (2013) 202-213. Co., 1984, 1075 p. No individual items are [36] W.X. Guo, D. Shu, H.Y. Chen, A.J. Li, H. abstracted in this volume., 1 (1984). Wang, G.M. Xiao, C.L. Dou, S.G. Peng, W.W. [18] M. Beaudin, H. Zareipour, A. Wei, W. Zhang, H.W. Zhou, S. Chen, Journal of Schellenberglabe, W. Rosehart, Energy for Alloys and Compounds, 475 (2009) 102-109. Sustainable Development, 14 (2010) 302-314. [37] A. Li, Y. Cheni, H. Chen, D. Shu, W. Li, H. [19] N.-K.C. Nair, N. Garimella, Energy and Wang, C. Dou, W. Zhang, S. Chen, Journal of Buildings, 42 (2010) 2124-2130. Power Sources, 189 (2009) 1204-1211. [20] P. Kurzweil, Journal of Power Sources, 195 [38] D.G. Li, J.D. Wang, D.R. Chen, Journal of (2010) 4424-4434. Power Sources, 210 (2012) 163-171. [21] M. Fernández, J. Valenciano, F. Trinidad, [39] P. Wang, X. Luo, X. Wu, X. Wei, L. Zhou, N. Muñoz, Journal of Power Sources, 195 X. Zheng, J Porous Mater, 20 (2013) 1-6. (2010) 4458-4469. [40] A. Moncada, M.C. Mistretta, S. Randazzo, [22] M. Saravanan, M. Ganesan, S. S. Piazza, C. Sunseri, R. Inguanta, Journal of Ambalavanan, Journal of Power Sources, 251 Power Sources, 256 (2014) 72-79. (2014) 20-29. [41] B.B. McKeon, J. Furukawa, S. [23] M. Saravanan, P. Sennu, M. Ganesan, S. Fenstermacher, Proceedings of the IEEE, 102 Ambalavanan, Journal of The Electrochemical (2014) 951-963. Society, 160 (2013) A70-A76. [42] L.T. Lam, R. Louey, Journal of Power [24] B. Zakeri, S. Syri, Renewable and Sources, 158 (2006) 1140-1148. Sustainable Energy Reviews, 42 (2015) 569[43] L.T. Lam, R. Louey, N.P. Haigh, O.V. Lim, 596. D.G. Vella, C.G. Phyland, L.H. Vu, J. [25] B. Hong, L. Jiang, K. Hao, F. Liu, X. Yu, Furukawa, T. Takada, D. Monma, T. Kano, H. Xue, J. Li, Y. Liu, Journal of Power Sources, Journal of Power Sources, 174 (2007) 16-29. 256 (2014) 294-300. [44] Y. Kamenev, G. Shtompel, E. Ostapenko, [26] I.A. Carlos, T.T. Matsuo, J.L.P. Siqueira, V. Leonov, Journal of Power Sources, 257 M.R.H. de Almeida, Journal of Power Sources, (2014) 181-185. 132 (2004) 261-265. [45] D. Yang, J. Liu, Q. Wang, X. Yuan, X. Zhu, [27] E. Gyenge, J. Jung, B. Mahato, Journal of L. Li, W. Zhang, Y. Hu, X. Sun, R.V. Kumar, J. Power Sources, 113 (2003) 388-395. Yang, Journal of Power Sources, 257 (2014) 27[28] G. Barkleit, A. Grahl, M. Maccagni, M. 36. Olper, P. Scharf, R. Wagner, H. Warlimont, [46] T. Woodland, M. Sorge, S. Zhang, T. Bean, Journal of Power Sources, 78 (1999) 73-78. S. McAllister, J. Canning, Y. Xie, D. Edwards, [29] S.M. Tabaatabaai, M.S. Rahmanifar, S.A. Journal of Power Sources, 230 (2013) 15-24. Mousavi, S. Shekofteh, J. Khonsari, A. Oweisi, [47] M. Sorge, T. Bean, T. Woodland, J. M. Hejabi, H. Tabrizi, S. Shirzadi, B. Cheraghi, Canning, I.F. Cheng, D.B. Edwards, Journal of Journal of Power Sources, 158 (2006) 879-884. Power Sources, 266 (2014) 496-511. [30] S.W. Swogger, P. Everill, D.P. Dubey, N. [48] D.R. Battlebury, Journal of Power Sources, Sugumaran, Journal of Power Sources, 261 80 (1999) 7-11. (2014) 55-63. [49] L. Chen, Z. Xu, M. Liu, Y. Huang, R. Fan, [31] I. Paleska, R. Pruszkowska-Drachal, J. Y. Su, G. Hu, X. Peng, X. Peng, Science of The Kotowski, Z. Rogulski, J.D. Milewski, A. Total Environment, 429 (2012) 191-198. Czerwiński, Journal of Power Sources, 129 [50] F. Birch, Journal of Geophysical Research, (2004) 326-329. 69 (1964) 4377-4388. [32] L. Zhao, B. Chen, D. Wang, Journal of [51] G. Morard, J. Siebert, D. Andrault, N. Power Sources, 231 (2013) 34-38. Guignot, G. Garbarino, F. Guyot, D. [33] E. Ebner, D. Burow, J. Panke, A. Börger, A. Antonangeli, Earth and Planetary Science Feldhoff, P. Atanassova, J. Valenciano, M. Letters, 373 (2013) 169-178.

208

Proceedings of SEEP2015, 11-14 August 2015, Paisley [72] W. Li, J. Dahn, D. Wainwright, Science, [52] A. Oxley, N. Barcza, Minerals Engineering, 264 (1994) 1115-1118. 54 (2013) 2-13. [73] N. Alias, A.A. Mohamad, Journal of Power [53] C.-L. Huang, J. Vause, H.-W. Ma, Y. Li, Sources, 274 (2015) 237-251. C.-P. Yu, Journal of Cleaner Production, 84 [74] Y. Hou, X. Wang, Y. Zhu, C. Hu, Z. Chang, (2014) 450-458. Y. Wu, R. Holze, Journal of Materials [54] J.P. Lorand, Geochimica et Cosmochimica Chemistry A, 1 (2013) 14713-14718. Acta, 54 (1990) 1487-1492. [75] N. Imanishi, O. Yamamoto, Materials [55] J.P. Lorand, Earth and Planetary Science Today, 17 (2014) 24-30. Letters, 93 (1989) 50-64. [76] H. Kim, J. Hong, K.-Y. Park, H. Kim, S.[56] J.P. Lorand, Earth and Planetary Science W. Kim, K. Kang, Chemical reviews, 114 Letters, 119 (1993) 627-634. (2014) 11788-11827. [57] D.H. Eggler, J.P. Lorand, Geochimica et [77] J.-Y. Luo, W.-J. Cui, P. He, Y.-Y. Xia, Cosmochimica Acta, 57 (1993) 2213-2222. Nature chemistry, 2 (2010) 760-765. [58] H. Braunschweig, P. Cogswell, K. Schwab, [78] J. Shao, X. Li, Q. Qu, Y. Wu, Journal of Coordination Chemistry Reviews, 255 (2011) Power Sources, 223 (2013) 56-61. 101-117. [79] J.Y. Luo, Y.Y. Xia, Advanced Functional [59] N.M. Leonard, L.C. Wieland, R.S. Mohan, Materials, 17 (2007) 3877-3884. Tetrahedron, 58 (2002) 8373-8397. [80] J. Choi, E. Alvarez, T. Arunkumar, A. [60] A.K. Shukla, S. Venugopalan, B. Manthiram, Electrochemical and solid-state Hariprakash, Journal of Power Sources, 100 letters, 9 (2006) A241-A244. (2001) 125-148. [81] L. Tian, A. Yuan, Journal of Power [61] G. Halpert, Journal of Power Sources, 12 Sources, 192 (2009) 693-697. (1984) 177-192. [82] P. He, J.-L. Liu, W.-J. Cui, J.-Y. Luo, Y.-Y. [62] A.K. Manohar, C. Yang, S. Malkhandi, B. Xia, Electrochimica Acta, 56 (2011) 2351-2357. Yang, G.K.S. Prakash, S.R. Narayanan, Journal [83] A.F. Cotton, G. Wilkinson, M. Bochmann, of the Electrochemical Society, 159 (2012) C.A. Murillo, Advanced inorganic chemistry, A2148-A2155. Wiley, 1999. [63] J.O. Gil Posada, P.J. Hall, Journal of Power [84] J. Whitacre, A. Tevar, S. Sharma, Sources, 262 (2014) 263-269. Electrochemistry Communications, 12 (2010) [64] B. Shruthi, V. Bheema Raju, B.J. Madhu, 463-466. Spectrochimica Acta Part A: Molecular and [85] L. Chen, Q. Gu, X. Zhou, S. Lee, Y. Xia, Z. Biomolecular Spectroscopy, 135 (2015) 683Liu, Scientific reports, 3 (2013). 689. [86] G.P. Hammond, T. Hazeldine, Applied [65] U. Köhler, C. Antonius, P. Bäuerlein, Energy, 138 (2015) 559-571. Journal of Power Sources, 127 (2004) 45-52. [87] Y. Li, H. Dai, Chemical Society Reviews, [66] P. Xu, X.J. Han, B. Zhang, Z.S. Lv, X.R. 43 (2014) 5257-5275. Liu, Journal of Alloys and Compounds, 436 [88] R.D. McKerracher, C. Ponce de Leon, (2007) 369-374. R.G.A. Wills, A.A. Shah, F.C. Walsh, [67] A.K. Manohar, C. Yang, S. Malkhandi, ChemPlusChem, 80 (2015) 323-335. G.K.S. Prakash, S.R. Narayanan, Journal of the Electrochemical Society, 160 (2013) A2078A2084. [68] S. Malkhandi, B. Yang, A.K. Manohar, G.K.S. Prakash, S.R. Narayanan, Journal of the American Chemical Society, 135 (2012) 347353. [69] M.J. Mackenzie Jr., Salkind, Alvin J., in, 1969. [70] A. Chaurey, S. Deambi, Renewable Energy, 2 (1992) 227-235. [71] M. Armand, Intercalation electrodes, in: Materials for advanced batteries, Springer, 1980, pp. 145-161.

209


DUAL ROLES OF CARBON COATINGS IN ELECTROCHEMICAL PERFORMANCES OF HYBRID SUPERCAPACITOR USING H 2 TI 12 O 25 /ACTIVATED CARBON S.H. Lee1, E. Baek1, H.K. Kim1, S.G. Lee2, Y.H. Lee1 and J.R. Yoon3,* 1. Dept. of Electronics Materials Engineering, Kwangwoon University, Seoul, South Korea; email: [email protected], [email protected], [email protected], [email protected] 2. Dept. of Ceramic Engineering, Eng. Res. Insti., Gyeongsang National University, Gyeongsang-do, South Korea; email: [email protected] 3. R&D center, Samwha Capacitor, Gyeonggi-do, South Korea; email: [email protected] Abstract Cylindrical hybrid supercapacitor is fabricated using pristine and different coating amounts of carbon (1.5, 3, 4.5 and 6 wt%) coated H2Ti12O25 anode and activated carbon cathode to investigate the effect of carbon coating on electrochemical performance of hybrid supercapacitor. The electrochemical performances of hybrid supercapacitor indicate that the 4.5 wt% carbon coating (approximately 3.09 nm) exhibit the superior electrochemical performance. Moreover, the carbon coating also plays an important role in suppressing swollen phenomenon. These can be attributed to the dual roles of carbon coating, including physical and chemical obstacle and high conductivity pathways. Therefore, we can suggest that carbon coated H2Ti12O25 anode can be applied to various energy storage devices. Keywords: Cylindrical hybrid supercapacitor, Carbon coating, Dual roles, Swollen Phenomenon. 1 INTRODUCTION Around the world, the studies regarding energy storage devices for alternative to fossil fuels and high efficiency in energy usage are being actively conducted. Among them, the lithium ion secondary battery and supercapacitor are currently being used in electric vehicles (EVs), hybrid electric vehicles (HEVs) and energy storage devices (ESS), uninterruptible power supplies (UPS) [1-4]. The lithium ion secondary batteries and supercapacitors have merits of high energy density and high power density, respectively due to their different principles of capacity implementation. The hybrid supercapacitor is designed as an energy storage device that combines the advantages of lithium ion secondary battery and supercapacitor. The spinel Li4Ti5O12 is one of the promising materials being investigated for anode material of energy storage device anode because of the excellent Li-ion insertion/extraction reversibility and high structural stability. However, Li4Ti5O12 has a drawback to apply to actual product due to its low theoretical specific capacity of 175 mAh g-1 [5-8]. Recently, Akimoto et al. [9] reported that H2Ti12O25 as anode which has a specific capacity of 229 mAh g-1 with comparable cycle performance to Li4Ti5O12 [10]. These can be explained by tunnel structure [11]. In order to 210

improve the conductivity of the metal oxidebased electrode, several methods such as reducing particle size [12,13], doping with metal ions [14,15], and coating with conductive materials [16,17] have been studied. Among various methods, in order to consider the not only improving conductivity related rate capability but also swollen phenomenon related cycle performance, we selected the carbon coating method to H2Ti12O25 anode. In this paper, we fabricated hybrid supercapacitor using carbon coated H2Ti12O25 as anode and activated carbon (AC) as cathode and investigated the effects of the carbon coating on the electrochemical performance. 2 EXPERIMENTAL The carbon coated H2Ti12O25 anode material was first prepared by mixing Na2CO3 (99.5%) and TiO2 in a molar ratio of 1:3. This mixture, Na2Ti3O7, was synthesized at 800 oC for 20 h in air. The H2Ti3O7 sample was fabricated from Na2Ti3O7 via Na+/H+ ion exchange reaction using a 1M HCl solution for 3 days at 60 oC. The prepared H2Ti3O7 was washed with deionized water until a pH equal to 7, and then dried at 100 o C for 24 h. The H2Ti12O25 was obtained by heating the H2Ti3O7 sample at 300 oC for 5 h. The H2Ti12O25 powders were mixed with

Proceedings of SEEP2015, 11-14 August 2015, Paisley different amounts (1.5, 3, 4.5, and 6 wt%) of beta cyclodextrin (β-CD) as the carbon source and then calcined at 800 oC for 2 h. The activated carbon (AC) and the carbon-coated H2Ti12O25 were used as the cathode electrode and anode electrode of the hybrid supercapacitor, respectively. The anode electrodes of the hybrid supercapacitor were prepared using the following process. To fabricate a slurry, a conductive carbon black binder (Super P) and polyvinylidene fluoride (PVDF) were mixed at the weight ratio of 83:7:10. N-Methyl pyrrolidinine (NMP) solvent was then added. The mixed slurry was cast on aluminum foil to a thickness of 125 μm and dried at 100 oC to remove the NMP solvent. The aluminum foil Figure 1. XRD patterns of H2Ti12O25 anode was pressed to a thickness of 70 - 80 μm. materials with different content of carbon Subsequently, the cell was dried in a vacuum coating oven for 48 h to eliminate moisture and was impregnated with a 1.5 M LiBF4 solution in 1:1 ethylene carbonate (EC): dimethyl carbonate (DMC) as the electrolyte. Lastly, the cylindrical hybrid supercapacitors were assembled with a cathode electrode/separator/anode electrode structure in an argon-gas-filled glove box. The structural properties of the carbon-coated H2Ti12O25 particles were measured using x-ray diffraction (XRD), a scanning electron microscope (SEM), and a transmission electron microscope (TEM) and the electrochemical performances were measured using Arbin BT 2042 battery test system at various current densities with a cut off voltage of 0-2.8 V. The electrochemical impedance spectroscopy (EIS) was done in the frequency range of 10-1 to 10-6 Hz. 3 RESULTS AND DISCUSSION Figure 1 shows the x-ray diffraction patterns of the synthesized pristine and carbon-coated H2Ti12O25. All peaks were corresponded with the H2Ti12O25 phase and no impurity phase was observed, as previously reported [18]. It can be inferred that carbon does not affect the crystallinity of H2Ti12O25. The microstructures of pristine and carbon-coated H2Ti12O25 particles are shown in Figure 2. All samples show almost similar morphologies and sizes regardless of different amounts of carbon coating. The average particle size was around 3 μm in all samples.

211

Figure 2. SEM images of the (a) carbon-free H2Ti12O25, (b) 1.5 wt% carbon-coated H2Ti12O25, (c) 3 wt% carbon-coated H2Ti12O25, (d) 4.5 wt% carbon-coated H2Ti12O25, (e) 6 wt% carbon-coated H2Ti12O25 anode materials. Figure 3 shows the transmission electron microscopy (TEM) images of the carbon-coated H2Ti12O25 materials to identify the thickness and distribution of carbon coating. The TEM results confirm that the H2Ti12O25 particle was randomly coated with carbon layer ranged from approximately 1.28 to 3.9 nm. The thickness of the carbon layers is in proportional to the carbon amounts, as shown in Figure 3. This carbon

Proceedings of SEEP2015, 11-14 August 2015, Paisley coating layer can be expected to increase were observed after the dwelling time. The IR conductivity of H2Ti12O25 than pristine one. drop can occur by differences operating principles between cathode and anode. The IR drop can be calculated using the following equation: [22]

Figure 3. TEM images of the (a) 1.5 wt% carbon-coated H2Ti12O25, (b) 3 wt% carboncoated H2Ti12O25, (c) 4.5 wt% carboncoated H2Ti12O25, (d) 6 wt% carbon-coated H2Ti12O25 anode materials. Figure 4 shows the initial charge/discharge curves for the hybrid supercapacitor using pristine and the carbon-coated H2Ti12O25 between 0 and 2.8 V. The discharge specific capacitances of the hybrid supercapacitor using the pristine and carbon-coated H2Ti12O25 anode were obtained using the following relationship: [19] (1) where C is the capacitance (Fg-1), ΔV is the voltage change, m is the mass of the active materials in both electrodes, q is the total charge, i is the current, and t is time. The discharge capacitances of the 0, 1.5, 3, 4.5, and 6 wt% carbon-coated H2Ti12O25 anodes were 57, 58, 60, 63, and 61 Fg-1, respectively at current density of 0.5 Ag-1. Compare to pristine, the capacitance of carbon-coated H2Ti12O25 increased with increasing carbon amounts, except for the 6 wt% carbon-coated H2Ti12O25. However, the capacitance of the 6 wt% carbon-coated H2Ti12O25 reversely decreased than 4.5 wt% carbon amount due to the excessive thickness of carbon layer. Jin and Zhu et al. [20, 21] reported that the conductivity of lithium ion is reduced by the excessive carbon layer due to the holdbacks for the lithium ion transport. Therefore, the optimal thickness of carbon layer can be determined to a 4.5 wt% carbon amounts. In addition, the sudden voltage drops (IR drops)

212

(2) Vcharge is the voltage of the cell at the end charge, Vdischarge is the voltage of the cell at the starting discharge, and I is the absolute value of the charge and discharge currents. The calculated values of the 0, 1.5, 3, 4.5, and 6 wt% carboncoated H2Ti12O25 anodes are 0.26, 0.255, 0.252, 0.241, and 0.247 Ω, respectively. The IR drop can reduce the performance of hybrid supercapacitor through a voltage drop.

Figure 4. Initial charge-discharge curves of the hybrid supercapacitors with the carbonfree and carbon-coated H2Ti12O25 anode materials. The cyclic voltammetry (CV) curves of the hybrid supercapacitors with pristine and carboncoated H2Ti12O25 anodes between 0 and 2.8 V are shown in Figure 5. The CV curve is a technique used to study the electrochemical reaction. According to previous researches, the activated carbon is a major factor contributing to charging the cells up to about 1.55 V. However, the very small amount of charging occurred at this time [3]. On the other hand, the redox reaction occurred H2Ti12O25 anode due to Li ion become major factor above 1.55 V and the large amounts of charging generated. The oxidation and reduction peaks indicated the Li ion intercalation and deintercalation into the H2Ti12O25 anode was observed around 2.8 V and 2.4 V versus Li/Li+, respectively. These CV results show the similar tendency to Fig. 4. The electrochemical reaction of carbon coated

Proceedings of SEEP2015, 11-14 August 2015, Paisley Faraday’s constant, C is the concentration of Li H2Ti12O25 up to 4.5 wt% was more activated ion, and σ is the Warburg impedance coefficient. than pristine one. The calculated diffusion coefficients of lithiumion are 2.937*10-7, 3.998*10-7, 5.757*10-7, 8.995*10-7, and 6.874*10-7 cm2 s-1 at the 0, 1.5, 3, 4.5, and 6 wt% carbon-coated H2Ti12O25 anode, respectively. Based on these Rct and DLi values, the carbon-coating can be expected to affect rate capabilities of hybrid supercapacitor [16, 17, 25].

Figure 5. Cyclic voltammetry curves of the hybrid supercapacitors with the carbon-free and carbon-coated H2Ti12O25 anode materials. To compare the effect of the pristine and carboncoated H2Ti12O25 anodes on the interface resistances at electrolyte/H2Ti12O25, the electrochemical impedance spectra (EIS) curves were performed after charging at 2.8 V, as seen in Figure 6. The EIS curves consist of a semicircle and sloping line at the low frequency indicate the lithium diffusion resistance inside H2Ti12O25 so called Warburg impedance [23, 24]. The inset of Fig. 6 shows the equivalent circuit of the hybrid supercapacitor where W is the Waburg impedance, CPEL and CPEC are the double layer capacitance of H2Ti12O25 and the activated carbon components under the applied potential, respectively. The semicircle indicates the charge transfer resistance (Rct) and electrolyte resistance (Rs) at high frequency. The Rct values of the 0, 1.5, 3, 4.5, and 6 wt% carbon-coated H2Ti12O25 anodes were approximately 0.06, 0.048, 0.037, 0.023, and 0.029 Ω, respectively. The Rct values of H2Ti12O25 anodes decreased with the increasing amounts of carbon contents. Therefore, we can expect that the carbon coating has a positive influence on anode by increasing conductivity. To support argument, we calculated the diffusion coefficient (DLi) of lithium-ion using the following equation [11, 18]: (3) where R is the gas constant, T is the absolute temperature, A is the surface area of the anode electrode, n is the number of electrons, F is the 213

Figure 6. EIS curves of the hybrid supercapacitors with the carbon-free and carbon-coated H2Ti12O25 anode materials.

Figure 7. Different charge-discharge rates of the hybrid supercapacitor with the carbonfree and carbon-coated H2Ti12O25 anode materials. Figure 7 shows the rate capabilities of the pristine and carbon-coated H2Ti12O25 at various charge-discharge rates of 0.1, 0.5, 1, 2, and 3 Ag-1. At slow charge-discharge rates, all retentions are almost same. However, with charge-discharge rates increasing, the difference of retention gradually increased. The capacitance retention increases in proportion to the carbon amounts. The capacitance retention of the pristine H2Ti12O25 considerably dropped with

Proceedings of SEEP2015, 11-14 August 2015, Paisley electrolyte by HF attack [30]. Yan-Bing et al. increasing charge-discharge rate and showed a capacitance retention of 52 % at the 3 Ag-1 [31] reported that the carbon coating on -1 compared to the 0.1 Ag . In the case of carbon electrode surface is an effective method to coated the H2Ti12O25, the capacitance retentions suppress the swollen phenomenon because of the were 53.5, 55.5, 61.5, and 58 % for the 1.5, 3, separation of H2T12O25 surface and electrolyte. 4.5, and 6 wt% coated carbon at the 3 Ag-1, We can get identical result and conclude that respectively. As a result, the carbon-coated carbon layer played a role barrier well, as shown H2Ti12O25 anodes have a much superior rate in Figure 8 (b). capability than the pristine one, as mentioned in Figure 6. Figure 8 (a) shows the cycle performance of the hybrid supercapacitors with pristine and carboncoated H2Ti12O25 anode after 1000 cycles at 3.0 Ag-1. The cycling behavior of carbon-coated H2Ti12O25 anodes appeared to show more flat compared with the pristine during 1000 cycles. In the case of 1.5 wt% carbon coated H2Ti12O25, it shows the similar behavior with pristine due to not only too little amounts carbon cover all H2Ti12O25 surface but also weak adhesion [26]. In the case of 6 wt%, the thickness of carbon Figure 8. (a) Cycling behaviors of the hybrid layer was too thick to move Li ion ordinarily. supercapacitor with the carbon-free and Among carbon coated anodes, the 4.5 wt% carbon-coated H2Ti12O25 anode materials. carbon-coated H2Ti12O25 which can allow maximize kinetics of Li ion shows the best cycling performance of 94.18%, superior to that of the others. Figure 8 (b) shows the photographs of cylindrical hybrid supercapacitors using the pristine and 4.5 wt% carbon coated H2T12O25 anode after 1000 cycles at 30 oC. The significant difference is shown in two cylindrical hybrid supercapacitors. The swollen phenomenon was produced during the reduction decomposition process from the interfacial reactions between the H2T12O25 anode and surrounding electrolyte. Many researchers reported [27-29] that the generation mechanisms of the swollen phenomenon are caused by the (b) Photographs of hybrid supercapacitors ( reduction decomposition process of the mixed ⅰ) 4.5 wt% carbon-coated H2Ti12O25 and ( salt solutions (EC/DMC 1.5M LiBF4) according to the following equations: ⅱ) pristine H2Ti12O25. (4) (5) (6) (7) The generated gases of C2H4, CO and CH4 are detected in the EC/DMC electrolyte, as shown in equations (4)-(7). Therefore, the carbon coating on H2T12O25 surface effectively play an important role in obstacle protecting the interfacial reactions between the electrode and 214

Proceedings of SEEP2015, 11-14 August 2015, Paisley energy but also suppress of swollen phenomenon. The 6 wt% carbon-coated H2Ti12O25 has lower electrochemical properties than the 4.5 wt% one, because the thicker carbon coating layer makes the decrease the lithium ion kinetics. The 4.5 wt% carbon-coated H2Ti12O25 shows the energy and power density of 5.71 Wh kg-1, 5439.8 W kg-1, respectively and capacitance retention of 94.18% after 1000 cycles at 3 Ag-1. Therefore, the hybrid supercapacitor using 4.5 wt% carbon-coated H2Ti12O25 anode can be expected to excellent Figure 9. Ragone plots of various hybrid energy storage device. supercapacitors with 4.5 wt% carbon-coated REFERENCES H2Ti12O25. [1] K. Karthikeyan, V. Aravindan, S.B. Lee, I.C. Figure 9 shows the Ragone plots of hybrid Jang, H.H. Lim, G.J. Park, M. Yoshio, Y.S. Lee, supercapacitor using 4.5 wt% carbon-coated A novel asymmetric hybrid supercapacitor based H2Ti12O25/activated carbon to compare various on Li2FeSiO4 and activated carbon electrodes, J. hybrid supercapacitors previously reported. The Alloy. Comp., Vol. 504, pp. 224-227, 2010. energy and power density of 4.5 wt% carbon[2] K. Karthikeyan, V. Aravindan, S.B. Lee, I.C. coated H2Ti12O25 were calculated at various Jang, H.H. Lim, G.J. Park, M. Yoshio, Y.S. Lee, charge and discharge currents using the Electrochemical performance of carbon-coated following relationship: [19] lithium manganese silicate for asymmetric hybrid supercapacitors, J. Power Sources, Vol. (8) 195, pp. 3761-3764, 2010. [3] S.H. Lee, H.K. Kim, Y.S. Yun, J.R. Yoon, (9) S.G. Lee, Y.H. Lee, A novel high-performance cylindrical hybrid supercapacitor with Li4(10) xNaxTi5O12/activated carbon electrodes, Int. J. Hydrogen Energy, Vol. 39, pp. 16569-16575, is the potential at the starting where 2014. discharge, is the potential at the end [4] S.H. Lee, S.G. Lee, J.R. Yoon, H.K. Kim, discharge, I is the charge and discharge currents, Novel performance of ultrathin AlPO4 coated m is the mass of active materials including the H2Ti12O25 exceeding Li4Ti5O12 in cylindrical anode and cathode electrodes, and t is the hybrid supercapacitor, J. Power Sources, Vol. discharge time in the hybrid supercapacitor. The 273, pp. 839-843, 2015. energy and power density of the hybrid [5] L. Cheng, J. Yan, G.N. Zhu, J.Y. Luo, C.X. supercapacitor using 4.5 wt% carbon-coated Wang, Y.Y. Xia, General synthesis of carbonH2Ti12O25 anode was 38.8-5.71 Wh kg-1 and coated nanostructure Li4Ti5O12 as a high rate 182-5439.8 W kg-1 at 0.1-3 Ag-1. The hybrid electrode material for Li-ion intercalation, J. supercapacitor using 4.5 wt% carbon-coated Mater. Chem., Vo. 20, pp. 595-602, 2010. H2Ti12O25 anode is superior to the other [6] T.F. Yi, H. Liu, Y.R. Zhu, L.J. Jiang, Y.g composition hybrid supercapacitors. Xie, R.S. Zhu, Improving the high rate performance of Li4Ti5O12 through divalent zinc substitution, J. Power Sources, Vol. 215, pp. 4 CONCLUSION 258-265, 2012. As a new energy storage device, the cylindrical [7] J. Lim, E. Choi, V. Mathew, D. Kim, D. hybrid supercapacitors using pristine and Ahn, J. Gim, S.H. Kang, J. Kim, Enhanced highcarbon-coated H2Ti12O25 anode and activated rate performance of Li4Ti5O12 nanoparticles for carbon cathode were fabricated. Among them, rechargeable Li-ion batteries, J. Electrochem. the 4.5 wt% carbon-coated H2Ti12O25 with 3 nm Soc., Vol. 158, pp. A275-A280, 2011. thickness is confirmed as the optimum [8] Z. Chen, I. Belharouak, Y.K. Sun, K. Amine, condition. The 4.5 wt% carbon-coated H2Ti12O25 Titanium-based anode materials for safe lithiumshows not only the electrochemical performance 215

Proceedings of SEEP2015, 11-14 August 2015, Paisley [19] V. Aravindan, Y. L. Cheah, W. F. Mak, G. ion batteries, Adv. Funct. Mater., Vol. 23, pp. Wee, B.V.R. Chowdari, S. Madhavi, Fabrication 959-969, 2013. of high energy-density hybrid supercapacitors [9] J. Akimoto, K. Kataoka, N. Kojima, S. using electrospun V2O5 nanofibers with a selfHayashi, Y. Gotoh, T. Sotokawa, Y. Kumashiro, supported carbon nanotube network, A novel soft-chemical synthetic route using ChemPlusChem, Vol. 77, pp.570-575, 2012. Na2Ti6O13 as a starting compound and [20] J. Wang, X.M. Liu, H. Yang, X.D. Shen, electrochemical properties of H2Ti12O25, J. Characterization and electrochemical properties Power Sources, Vol. 244, pp.679-683, 2013. of carbon-coated Li4Ti5O12 prepared by a citric [10] S.H. Lee, H.K. Kim, J.H. Lee, S.G. Lee, acid sol-gel method, J. Alloy. Comp., Vol. 509, Y.H. Lee, Fabrication and electrochemical pp.712-718, 2011. properties of cylindrical hybrid supercapacitor [21] G.N. Zhu, C.X. Wang, Y.Y. Xia, A using H2Ti12O25 as anode material, Mater. Lett., comprehensive study of effects of carbon Vol. 143, pp.101-104, 2015. coating on Li4Ti5O12 anode material for lithium[11] J. Akimoto et al. U.S. Patent No. 8,309,253. ion batteries, J. Electrochem. Soc., Vol. 158, (2012) pp.A102-109, 2011. [12] A. Guerfi, S. Sévigny, M. Lagacé, P. [22] J.Y. Luo, Y.Y. Xia, Electrochemical profile Hovington, K. Kinoshita, K. Zaghib, Nanoof an asymmetric supercapacitor using carbonparticle Li4Ti5O12 spinel as electrode for coated LiTi2(PO4)3 and active carbon electrodes, electrochemical generators, J. Power Sources, J. Power Sources, Vol. 186, pp.224-227, 2009. Vol. 119-121, pp.88-94, 2003. [23] S.S. Zhang, K. Xu, T.R. Jow, EIS study on [13] Y. Lan, X. P. Gao, H. Y. Zhu, Z. F. Zheng, the formation of solid electrolyte interface in LiT. Y. Yan, F. Wu, S. P. Ringer, D. Y. Song, ion battery, Electrochim. Acta, Vol. 51, pp.1636Titanate nanotubes and nanorods prepared from 1640, 2006. rutile powder, Adv. Funct. Mater., Vol. 15, [24] S. Zhang, M.S. Ding, K. Xu, J. Allen, T.R. pp.1310-1318, 2005. Jow, Understanding solid electrolyte interface [14] B. Zhang, H. Du, B. Li, F. Kang, Structure film formation on graphite electrodes, and electrochemical properties of Zn-doped Electrochem. Solid-State Lett., Vol. 4, pp. A206Li4Ti5O12 as anode materials in Li-ion battery, 208, 2001. Electrochem. Solid-State Lett., Vol. 13, pp.A36[25] H.G. Jung, S.T. Myung, C.S. Yoon, S.B. A38, 2010. Son, K.H. Oh, K. Amine, B. Scrosati, Y.K. Sun, [15] Q. Zhang, C. Zhang, B. Li, S. Kang, X. Li, Microscale spherical carbon-coated Li4Ti5O12 as Y. Wang, Preparation and electrochemical ultra high power anode material for lithium properties of Ca-doped Li4Ti5O12 as anode batteries, Energy Environ. Sci., Vol. 4, pp.1345materials in lithium-ion battery, Electrochim. 1351, 2011. Acta, Vol. 98, pp.146-152, 2013. [26] Z. Chen, B. Du, M. Xu, H. Zhu, L. Li, W. [16] H.G. Jung, J. Kim, B. Scrosati, Y.K. Sun, Wang, Polyacene coated carbon/LiFePO4 Micron-sized, carbon-coated Li4Ti5O12 as high cathode for Li ion batteries: Understanding the power anode material for advanced lithium stabilized double coating structure and enhanced batteries, J. Power Sources, Vol. 196, pp.7763lithium ion diffusion kinetics, Electrochim. Acta, 7766, 2011. Vol. 109, pp.262-268, 2013. [17] L. Cheng, X.L. Li, H.J. Liu, H.M. Xiong, [27] J.S. Shin, C.H. Han, U.H. Jung, S.I. Lee, P.W. Zhang, Y.Y. Xia, Carbon-coated Li4Ti5O12 H.J. Kim, K. Kim, Effect of Li2CO3 additive on as a high rate electrode material for Li-ion gas generation in lithium-ion batteries, J. Power intercalation, J. Electrochem. Soc., Vol. 154, Sources, Vol. 109, pp.47-52, 2002. pp.A692-697, 2007. [28] H. Yoshida, T. Fukunaga, T. Hazama, M. [18] J. Akimoto, K. Chiba, N. Kijima, H. Terasaki, M. Mizutani, M. Yamachi, Hayakawa, S. Hayashi, Y. Gotoh, Y. Idemoto, Degradation mechanism of alkyl carbonate Soft-chemical synthesis and electrochemical solvents used in lithium-ion cells during initial property of H2Ti12O25 as a negative electrode charging, J. Power Sources, Vol. 68, pp.311material for recharge lithium-ion batteries, J. 315, 1997. Electrochem. Soc., Vol. 158, pp.A546-549, [29] D. Aurbach, Y. Ein-Eli, O. Chusid, Y. 2011. Carmeli, M. Babai, H. Yamin, The correlation

216

Proceedings of SEEP2015, 11-14 August 2015, Paisley between the surface chemistry and the performance of Li-carbon intercalation anodes for rechargeable ‘rocking-chair’ type batteries, J. Electrochem. Soc., Vol. 141, pp.603-611, 1994. [30] Y.B. He, B. Li, M. Liu, C. Zhang, W. Lv, C. Yang, J. Li, H. Du, B. Zhang, Q.H. Yang, J.K. Kim, F. Kang, Gassing in Li4Ti5O12-based batteries and its remedy, Sci. Rep., Vol. 2, pp.913-921, 2012. [31] Y.B. He, F. Ning, B. Li, Q.S. Song, W. Lv, H. Du, D. Zhai, F. Su, Q.H. Yang, F. Kang, Carbon coating to suppress the reduction decomposition of electrolyte on the Li4Ti5O12 electrode, J. Power Sources, Vol. 202, pp.253261, 2012.

217


NUMERICAL RESEARCH ON HEAT TRANSFER AND ENERGY STORAGE IN GLASS FURNACE REGENERATOR Haitao Zhang1, Taohong Ye2, Qizhao Lin3 1. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected] 2. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected] 3. Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei, China; email: [email protected]

Abstract Glass furnace is highly energy-consuming equipment for melting glass raw materials, which usually uses regenerator to recover energy from hot flue air to preheat combustion air. The aim of our study is to improve the heat transfer efficiency, and provide a comprehensive energy-saving program and optimization scheme for glass furnace regenerator. CFD method was used to simulate the actual performance and heat transfer efficiency of glass furnace regenerator in different conditions. The results show that the method which concerns about regenerator wall loss agree with the experimental data well, the inlet velocity of flue gas and combustion air has an apparently effect on the outlet temperature of flow stream, while the cycle time and the material of the checker bricks have little effect on the efficiency of heat transfer. These results have important significance for energy conservation of industrial glass production. Keywords: CFD modeling, Energy storage, Regenerator, Industrial glass furnace 1

INTRODUCTION

As the core device melting glass raw material in glass production process, glass furnace is highenergy-consuming and high emission equipment. A 500 TPD (Ton per Day) glass furnace consumes nearly 30,000 t coal each year, and the specific energy consumption is about 6530-6780kJ/kg for melting glass in the furnace [1]. A few approaches for energy saving have been studied, for example, oxy-fuel combustion can decrease 22-23% of total consumption [2], waste heat power generation can reduce the output temperature of the flue to 430 K, and save 12-13% of fuel[3]. However, as the renovation during operation or cold repair period is quite complex and costs huge, the methods mentioned above are not very popular in glass production industry. Thus, varied kinds of regenerators are widely used. Regenerative system can accumulate the waste heat of flue gas and release it to combustion air for preheating, improve the temperature of combustion air, decrease the specific energy consumption, and help to obtain higher flame temperature. About 60–65% of input heat and 1500 kJ/kg of specific energy consumption can be recovered by a welldesigned regenerator system [4]. Furthermore, 218

Flue gas with lower temperature is beneficial to avoid the greenhouse effect and environmental pollution. Conventional glass furnace regenerator is usually composed of several elementary channels which are fixed by the solid packing of checker bricks. A couple of regenerator matrices operate at the same time to guarantee the continuous alternation of hot and cold periods. Energy is delivered from the hot flue air (combustion exhaust gas) to the solid packing during hot periods, while during cold periods the energy previously stored in the bricks is restored to the cold combustion air. A hot period succeeded by a cold period forms one cycle of operation. Several cycles are necessary to reach the thermal equilibrium of the regenerator system, industrial glass furnace regenerators usually operate for years under the stabilized thermal equilibrium. A lot of research work has been studied in the past decades, not only include the improvement of theoretical model, but also experimental measurement and analysis in different conditions. Hausen [5-7] started the study of regenerators by theoretical approach with the development of closed methods. However, this

Proceedings of SEEP2015, 11-14 August 2015, Paisley During the two different periods of one cycle in method can hardly calculate the successive the regenerator, flue gas and combustion air flow alternation of hot and cold periods until the as two separate streams without being mixed, cyclic equilibrium is obtained, which is which are assumed to be viscous turbulence. The particularly difficult in glass furnace regenerator operation schematic diagram of glass furnace because of the complex flow and heat transfer regenerator system in half cycle is presented in phenomena. For instance, it is troublesome to Figure 1. Mass rate and temperature of input calculate the mean heat transfer coefficients flow are maintained at constant value, and the air during both periods by using correlations with leakage is assumed to be negligible. Mass closed methods. Based on the remarkable balance equation and momentum balance progress on numerical simulation by using equation are shown as follows, computers over the past twenty years, an open method using computational fluid dynamics   (CFD) code was developed to change this (1)  (  ui )  0  t  x i predicament. Following this method, modelling of heat transfer in regenerators using numerical u    p ( ui )  ( ui u j )  ( i )  Su  Qk (2) techniques was studied by Foumeny and t x j x j x j xi Pahlevanzadeh [8]. The k-ε RNG turbulence Where Sui represents the generalized source of model combined with the enhanced wall functions was used by Reboussin [9] to describe the component velocity ui . the turbulent aiding mixed convection k-ε RNG turbulence model is used to describe phenomenon in glass furnace regenerator. Wall the turbulent flow, the value of constant in the heat losses from the regenerator surface were model refers to the literature [13]. considered by Sardeshpande [10], the results of his study suggested that a drop in regenerator efficiency due to leakages can be significant and the degree of drop depends on the kind of leakages. Experimental measurements of the thermal performance of regenerators with various kinds of checker bricks in an industrial scale glass furnace system were presented by Zanoli et al. [11]. In addition, the influence of using different checker brick materials is studied by Wu Zi-ying [12]. As far as we know, there is no research concerns the comprehensive Figure 1. Operation schematic diagram of discussion on the effect of different conditions glass furnace regenerator system in half on the output temperature of flue gas and cycle. combustion air, while the different conditions contains varies boundary conditions, different 2.2 Energy modeling operation cycle time and packing materials. This During the hot period, thermal exchange in the is one of the innovations brought by this paper. regenerator is dominated by convection and radiation combined with heat conduction of the In this paper, CFD method was used to simulate checker bricks, whereas during the cold period, operation performance and heat transfer the thermal exchange is consistent with efficiency of glass furnace regenerator in convection between combustion air and checker different conditions, the results are compared bricks. Wall loss from regenerator cannot be and analyzed to sum up a comprehensive and negligible during each period. Heat transfer is effective energy-saving program for glass assumed to be in an unsteady state under the furnace system, which can provide a degree of thermal equilibrium condition. The total heat reference in energy-saving design on industry transfer from flue gas to combustion air is the glass production. function of superficial area and mean 2 MODELING temperature difference, Qreg  U reg Areg ,surf t 2.1 Flow field modelling (3) i

219

Proceedings of SEEP2015, 11-14 August 2015, Paisley Where the overall ideal heat transfer coefficient Brick dimensions (length × 450×160×40 mm width × thickness) U reg is calculated using gas side and air side Density of brick 2900 kg/m3 heat transfer coefficients, refers to the literature [14]. The mean temperature difference between the outlet and inlet air is (T flue,in  Tair ,out )  (T flue ,out  Tair ,in ) t  (4) (T T ) ln flue,in air ,out T flue,out  Tair ,in During heat storage and air heating period, energy loss from regenerator is detected in checker bricks, such as wall heat loss and the heat absorbed by leaked air. Therefore the energy transmission from flue gas to combustion air in the regenerator for a whole cyclic can be represented by the following equation, Q flue,reg  Qwall ,loss,hot  Qair ,leak ,hot  Qair ,reg  Qwall ,loss,cold  Qair ,leak ,cold

(5) While Q flue,reg and Qair ,reg represent the energy stored in regenerator from flue gas and recovered to combustion air ， respectively. Generally air leakage in the operating process will be negligible. The equations mentioned above form the general regenerator model . Given the input conditions, the equations are solved by using SIMPLE algorithm and semi implicit difference scheme with second order accuracy in every time and space, while pressure is corrected. Staggered grid and triangle matrix algorithm are used to linear the source term which contains the thermal resistance. Wall function is considered to deal with the near-wall grid. Energy equations are solved by Hybrid scheme and Runge-Kutter method respectively in time and space. Regenerator is heated by one-way preheating so that every point on the solid packing wall is uniformly linear distribution of temperature with height and time. The transient calculation goes on until the difference between the energy stored during the hot period and recovered during the cold period is less than 1%, the convergence criterion is satisfied and the thermal equilibrium is reached. The lists of model input parameters and operating parameters are presented in Table 1 and Table 2. Table 1. Designed parameters of regenerator system.

Table 2. Inlet parameters in hot and cold period. Design variable Mass flow rate of flue gas Mass flow rate of combustion air Average inlet temperature of flue gas Average inlet temperature of combustion air

Design value 8687 kg/h 7345 kg/h 1300 K 300 K

2 RESULTS AND DISCUSSION The thermal equilibrium is reached at 97200s, and the temperature changes from inlet to outlet at different radial positions of the regenerator are shown in Figure 2 and Figure 3. These two figures indicate that the closer to the wall, the more linearly the temperature increases or decreases along the length. However, it does not completely become a linear change. These results are consistent with the research results of Delrieux [15]. The nonlinear change of the temperature of flue gas and combustion air in the regenerator is the results of the overall heat transfer coefficient and specific heat value change with temperature, thus we infer that the linear change of the flue gas and combustion air in the traditional exchanger theory, which is held by Rummel and Hausen [5-7] is not reasonable enough here.

Figure 2. The temperature change from inlet to outlet at different radial positions of the regenerator (hot period).

Design variable Design value Regenerator dimensions 4.9×3.1×7.5 m (length × width × height)

220

Proceedings of SEEP2015, 11-14 August 2015, Paisley On the contrary, during the cold period, thermal exchange between the fluid and the solid packing is dominated by convection, and there is no thermal exchange by radiation between the air and the walls during this period. Thus the heat transfer coefficients in this period should be calculated by the following equation: c  4.5(1.115  0.244 a de0.6 )(tg  273)0.25 (10) Figure 4 and Figure 5 show the temperature changes from inlet to outlet with different inlet flow rates. The results both show that the inlet flow rate has an effect on the output temperature of the regenerator. From the equation (8) and (10) we can see that the heat transfer coefficients in cold and hot periods are influenced by the flow rate of flue gas and combustion Figure 3. The temperature change from inlet respectively, and consequently lead to the to outlet at different radial positions of the different effect on heat transfer. Although the regenerator (cold period). heat transfer coefficient are proportional to the As the overall heat transfer coefficient is defined flow rate, the figures show that the input flow by the relationship rate has the optimal value, which is between 1 0.15-0.225m/s. (6) U reg  R hot  Rcold  Rsolid Where the thermal resistance between flue gas 1 and brick wall Rhot  , the thermal

 hot hot

resistance between combustion air and brick 1 , and the thermal wall Rcold 

 cold cold

conductivity

resistance

of the solid     C packing Rsolid  (1  ). 3    108    During the hot period the thermal exchange between the fluid and the solid packing is characterized by convection and radiation. This thermal radiation is due to the presence of waste gas which becomes a semi-transparent medium. This thermal radiation reaches 80% to 90% of the global heat exchange. Thus the heat transfer coefficients should be composed of convection heat transfer coefficient and radiation heat transfer coefficient: (7)  h   hk   hr Where the convection heat transfer coefficient is calculated by the following equation: (8)  hk  4.5(1.115  0.244 g de0.6 )(tg  273)0.25 And the radiation heat transfer coefficient is calculated by the following equation: t (9)  hr  3.6(21PCO2  35PH2O )( g  2) L0.5 100 2

221

Figure 4. Temperature changes from inlet to outlet of the regenerator with different inlet flow rates (hot period).


Figure 5. Temperature changes from inlet to outlet of the regenerator with different inlet flow rates (cold period). For different cycle time, the output temperature of the flue gas and combustion air does not change significantly in Figure 6 and Figure 7. The results indicated that the outlet temperature of flue gas is maintained between 825 and 875k, and the outlet temperature of combustion air between 750 and 800 K for every cycle time tested in the calculation. It confirms that cycle time has little influence on heat transfer of regenerator, which was proposed in the previous literature [14]. Currently 20 minutes for the cycle time is widely used in industrial glass production.

Figure 6. Temperature changes with time for different cycle time in the inlet of flue gas.

222

Figure 7. Temperature changes with time for different cycle time in the inlet of combustion air The kind of checker bricks directly affects the thermal conductivity resistance of the solid packing, and its matching pattern has a significant effect on its service life, storage capacity and the energy-saving efficiency of regenerator. Table 3 is the comparison between the heat transfer coefficients of different checker bricks. The thermal conductivity coefficient and the specific heat C are determined corresponding to 1123 K, which is the average temperature of the regenerator. The diathermic depth S is half of the thickness of the checker bricks. Figure 8 shows the temperature change on the surface of checker bricks along the length of regenerator. This figure shows that the material of checker bricks has little effect on the efficiency of heat transfer. In this case, silicon carbide seems to be the optimal choice of the four material types, but it is very susceptible to be corroded in the regenerator channel, since the flue gas with high temperature has a large proportion of thenardite water and alkali vapor. Taking account of the erosion resistance, thermal shock resistance, and creep resistance of these materials, magnesia brick should be the optimal choice.

Proceedings of SEEP2015, 11-14 August 2015, Paisley Table 3. Comparison between heat transfer coefficients of different checker bricks Materials Clay Magnesia brick Fused zirconia Silicon carbide

 / KJ 

C / KJ 

(m  h  K )1

(kg  K )1

/ kg  m3

S/m

 h / hr

 c / hr

4.47

1.059

2300

0.03

1/3

14.04

1.153

2700

0.02

14.94

1.126

3450

4.47

1.059

2300

 h / KJ 

 c / KJ 

(m2  h  K )1

(m2  h  K )1

1/4

90

55

1/5

1/7

86

42

0.01

1/7

1/9

45

39

0.06

1/6

1/8

73

51

According to the previous equation (3) we find that increasing the heat exchange area appropriately with reducing brick thickness can also improve heat transfer. We made a comparison about heat exchange efficiency between different brick thickness, and the simulation results verified it. From Figure 9and Figure 10 we can see that the outlet temperature of flue gas with the thickness 20mm is significantly lower than that of 40mm, and the outlet temperature of combustion air also has a clearly change. However, given the problem that SiO2, V2O5 and CaO in hot flue gas are easy to corrode checker bricks, which results in collapse and blocking in the regenerator, the thickness of bricks cannot be reduced substantially. To avoid these problems, it should be based on the actual situation of glass furnace regenerator system to choose suitable thickness.

Figure 9. Outlet temperature of flue gas changes with different brick thickness.

Figure 10. Outlet temperature of combustion air changes with different brick thickness.

Figure 8. Temperature change of the surface of checker brick along the length of regenerator using different materials in half cycle.

An experiment study on the heat exchange in glass furnace regenerator is carried out and the measurement data was compared with the numerical simulation results. The furnace in this 223

Proceedings of SEEP2015, 11-14 August 2015, Paisley results are of great significance in the field of energy saving of industrial glass production.

experiment is a 200 TPD (Ton per Day) regenerative glass furnace located in Zhang jiagang, China. The regenerator consists of 24 elementary single pass channels, which is made of cylinder bricks. The height of the regenerator is 7.54 m. Other input parameters are approximately the same as the physical model. The outlet temperature of hot flue gas is measured and the comparison with simulation results is shown in Figure 11. We can see that calculation results match the actual situation well.

ACKNOWLEDGEMENTS My deepest gratitude goes first to my advisor, Prof. Lin Qizhao, a respectable, responsible scholar who was one of the most famous scholars researching on energy engineering in China, provided me with valuable guidance in every stage of the writing of this thesis. Without his enlightening instruction and patience, I could not have completed my thesis. His keen and vigorous academic observation enlightens me not only in this thesis but also in my future study. My sincere appreciation also goes to the teachers and students of Department of Energy Engineering, University of Science and Technology of China, who participated this study with great cooperation. Last but not least, I would like to thank all my friends, especially my lovely roommate, for their encouragement and support.

Figure 11. Outlet temperature profile of hot flue gas in hot period. 4 CONCLUSIONS In conclusion, comprehensive study on the effect of boundary condition, operation cycle time, and packing material on the heat transfer efficiency of glass furnace regenerator has been discussed. The CFD method which is used to simulate actual performance and calculate heat transfer efficiency in different conditions has proved to be effective and has a high accuracy. The model is based on the mass balance and energy balance of streams along with heat transfer characteristic equations. Regenerator wall loss is also considered in the model for the accuracy of results. We identified that the temperature increases and decreases nonlinearly in the regenerator channels, the inlet velocity of flue gas and combustion air has an apparently effect on the outlet temperature of the flow stream. The cycle time and material of the checker bricks has little effect on the efficiency of heat transfer. These

224

REFERENCES [1] Abbassi A, Khoshmanesh Kh. Numerical simulation and experimental analysis of an industrial glass melting furnace. Applied Thermal Engineering, Vol. 28, pp. 450–459, 2008. [2] Hu F, Shi QX, Ren D, et al. Discussion of key technical issues on waste heat recovery from glass-melting furnaces. Journal of Chinese Society of Power Engineering, Vol. 31, pp. 381386, 2011. [3] Xie JL, Jin MF, Mei SX. Numerical Simulation of the Effect Rule of the Burners' Installation Height on Combustion Space in Oxy-fuel Glass Furnace. Journal of Wuhan University of Technology, Vol. 33, pp. 26-31, 2011. [4] Tata Energy Research Institute. Practical energy audit manual, glass industry, prepared for Indo-German energy efficiency project, August 1999. [5] Hausen H, Berechnung der Steintemperatur in winderhitzern, Archiv. Fur das Eisenhu¨ ttenwesen, Vol.10, pp. 473–480, 1939. [6] Shah RK, Thermal design theory for regenerators, Heat Exchangers, Thermal Hydraulic Fundamentals and Design, Hemisphere, New York, pp. 721–763,1981,.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [7] Heggs PJ, Experimental techniques and correlations for heat exchangers surfaces: Regenerators, Fundamentals of Low Reynolds Number Forced Convection, Hemisphere, New York, pp. 369–394, 1982. [8] Foumeny EA, Pahlevanzadeh. Performance evaluation of thermal regenerators. Heat Recov Syst CHP, Vol. 14(1), pp. 79–84, 1994. [9] Reboussin Y, Fourmigue JF, Marty Ph, Citti O. A numerical approach for the study of glass furnace regenerators. Apply Thermal Engigeering, Vol. 25, pp. 2299–320, 2005. [10] Vishal Sardeshpande, Renil Anthony, Gaitonde UN, Rangan Banerjee. Performance analysis for glass furnace regenerator. Applied Energy, Vol. 88, pp. 4451-4458, 2011. [11] Zanoli A, Leahy WD, Vidil R, Lagarenne D. Experimental studies of thermal performance of various cruciform regenerator packing. Glass Technology, Vol. 32(5), pp. 157–162, 1991. [12] Wu ZY, Zhao HS, Gao XP. Research on glass furnace regenerator refractories. Journal of Wuhan University of Technology, Vol. 32(22), pp. 96-101, 2010. [13] Balabel A, El-Askary WA. On the performance of linear and nonlinear k-epsilon turbulence models in various jet flow applications. Vol. 30(3), pp. 325-340, 2011. [14] Schmidt FW, Willmott AJ. Thermal energy storage and regeneration. New York, McGrawHill Book Company, 1981. [15] Delrieux, J. The influence of the thermalproperties of refractories and their mode of utilization on the heat-balance in regenerators. Glass Technology, Vol. 21(4), pp. 162-172, 1980.

225


NICKEL-IRON (NI/FE) BATTERIES FOR LARGE-SCALE ENERGY STORAGE A. H. Abdalla1, C.I. Oseghale, J.O.G. Posada and P. J. Hall Department of Chemical and Biological Engineering. Faculty of Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK; email: [email protected]

Abstract Due to their low cost, robustness and eco-friendliness, Nickel/Iron batteries can be used for large-scale energy storage. Aside these advantages, the commercial use of these batteries have been limited by their relatively low charging efficiency. In fact, the performance of these batteries is drastically reduced by the parasitic evolution of hydrogen. By using a hot-pressed route, different additives, such as Bi2S3, CuSO4 were used as a means to tackle the problem mentioned above. Iron electrodes under strong alkaline conditions were examined by means of galvanostatic charge/discharge, cyclic voltammetry and X-ray diffraction (XRD). It is found that the additions of these additives enhance the performance of the NiFe battery. Keywords: Renewable Energy Sources; Energy Storage, Additives; iron-based electrode batteries 1 INTRODUCTION There is an increasing demand of energy coming from renewable sources; unfortunately, the intermittent nature of these resources restricts their use [1-6]. Basically, energy generation from renewable sources and demand are not easily matched. The electrochemical energy storage emerges as a possible solution to overcome the problem mentioned above. In fact, a great deal of efforts has been made to develop highly efficient, robust and safe energy storage (ES) [7, 8]. Of course, the intended solution would also require abundant, non-toxic and ecofriendly materials [9]. NiFe cells are secondary batteries that left out of favour with the advent of cheaper lead acid batteries; however, there is renewed interest in these cells due to their durability, environmentally friendliness, compatibility with renewable sources (such as wind power), low cost of raw materials, long life-cycle, low cost and good resistance to electrical abuse (overcharge and deep discharge) [10, 11]. Unfortunately, these batteries have been limited by the poor charge efficiency related to hydrogen evolution. Essentially during the charging of the battery Equation (1), the 226

electrolyte is decomposed on the surface of the iron electrode as indicated by Equation (2) [1214].

Fe(OH)2 + 2e- « Fe + 2OH 2H 2O  2e   H 2  2OH 

E 0 = - 0.88V (1) E 0   0.828V

(2)

The mitigation of hydrogen evolution and electrolyte decomposition as illustrated by Equation (2), has been traditionally achieved by utilizing electrode additives such as bismuth, bismuth sulfide, cobalt, copper, iron sulphide, carbon-black, etc. [12, 13, 15-17]. Undoubtedly, bismuth sulphide is one of the most effective additives to reduce the evolution of hydrogen under strong alkaline conditions [13, 15, 17, 18], as batteries with staggering capacities (reaching 0.3 Ah g-1) and very high coulombic efficiencies (in the order of 96%) have been reported. The Iron sulphide has also been successfully used [17, 19] but to a lesser degree than bismuth sulphide. Equation (3) illustrates the reduction process that bismuth sulphide undergoes under alkaline conditions [15, 17].

Proceedings of SEEP2015, 11-14 August 2015, Paisley Bi2 S3  6e  2Bi  3S 2 E  0.81V (3) polyethylene powder and potassium carbonate as a binder. The coating process was repeated until a constant amount of iron was reached (about The addition of additives such as Na2S, Li ions 0.3 grams). More experimental design details to the electrolyte has been effective not only in can be found here[15]. controlling the reduction of Fe(II) to Fe(III), but 

2

0

in increasing the capacity of the iron electrode as well [17, 20]. Due to its relatively low cost, high electrical conductivity, chemical activity, and capacity to prevent electrolyte from decomposing, copper has attracted the attention of NiFe battery researchers [21-29]. This manuscript reports the effect of bismuth sulphide and copper composites on the electrochemical performance of NiFe batteries. The formation of a good conductive network of copper is crucial to improving the reversibility of the active material. Iron electrodes were investigated by galvanostatic charge/discharge, cyclic voltammetry, X-ray - diffraction methods (XRD). 2

EXPERIMENTAL

2.1 Materials The chemicals and materials used were of the following specifications: iron powder (Fe, 99% ≤ 10µm, Sigma-Aldrich), Bismuth sulphide (purity 99.5%) from Sigma-Aldrich, Potassium carbonate (purity 99.0%) from Sigma-Aldrich, Potassium carbonate (purity 99.0%) from Sigma-Aldrich, copper sulphate (CuSO4 5H2O, 98% ≤ 10µm, Alfa Aesar), nickel foam (Ni, purity 99.0%, density 350 g/m2, Sigma Aldrich), potassium hydroxide (KOH, purity ≥ 85.0%, pellets, Sigma-Aldrich). The deionized water was produced by using an Elix 10-Milli-Q Plus water purification system (Millipore, Eschborn, Germany). 2.2 Preparation of the working electrodes Iron electrodes from this active material are obtained as follows; strips of nickel foam (4cm x 1 cm) were coated and hot pressed (140℃ at, a pressure of 10Kg cm-2 ) with differing amounts of electrode materials as shown in Table 1. An electroactive paste consisting of iron powder, bismuth sulphide, copper and with a mixture of 227

Table 1. Experimental determinations of factors and levels Factors

Components weight fraction Low High

Fe Bi2S3 Cu PP K2CO3

60 4.5 0 3 3

90 5 10 5 5

2.3 Electrochemical testing methods Electrodes assembled and tested into a three electrode configuration cell. Our in house made iron electrodes were used as working electrodes (WE). Commercially available nickel oxide electrodes were used as counter electrodes (CE). All readings were made against a mercury/mercury oxide reference electrode (RE) (EHg/HgO =+0.098V vs. the normal hydrogen electrode). KOH (28.5 w/v%) solution used as at the electrolyte. Cells were charged to their rated capacity (0.35 Ah/g-1) at C/5, discharged to cut off voltage of 0.8 vs. Hg/HgO reference electrode to a potential of 1.4V, at room temperature using a 64Channel Arbin (Model-SCTS) electrochemical cycling system operating in galvanostatic mode. The performance of the battery was calculated by using the following expression: Q -Q hQ = ch H (5) Qch Where, ηQ is the coulombic efficiency, Qch is the total charge, QH is the charge wasted in hydrogen evolution [17]. The Tafel relationship was used to calculate the charge used for hydrogen evolution with the current of hydrogen evolution. Cyclic voltammetry measurements were performed under potentiostatic conditions on a Solartron Cell Test System. The electrochemical measurements were made using

Voltage(V)

Proceedings of SEEP2015, 11-14 August 2015, Paisley a conventional three-electrode glass cell. All 1.2 readings were made against a mercury/mercury 1.1 oxide reference electrode (RE) (EHg/HgO=+0.098V vs. the normal hydrogen 1.0 electrode). A platinum wire was used as a counter electrode (CE). Finally, a concentrated 0.9 with additives solution of potassium hydroxide (28.5% KOH) was used as the electrolyte. 0.8 No additives

The X-ray diffraction measurements were used to characterize the electrode and determining the crystal phase of the materials. X-ray data were obtained on a Bruker D2-Phaser - Cu-Kα, λ = 1.5406 nm, radiation in, and step time of 0.1s in a 2θ at range between 10 and 85°(2θ); detector set to 0.27 V of the lower detection limit. The XRD data analysis was performed by ICDD PDF-4+ and Sieve+ software 3

RESULT AND DISCUSSION

3.1 Galvanostatic charging/discharging As shown in Fig. 1, under galvanostatic conditions of charge/discharge curves, NiFe cells exhibit good cycling properties in the sense that performance tends to increase with the cycle number during the first 30 cycles, after that they stabilise and reach the steady state. 1.2

0.7 1135000

1140000

1145000

1150000

Time (s)

Figure 2. Discharge curves of hot-pressed type iron electrodes with and without additives at C/5 rate. Fig. 2 reveals that electrodes produced without additives exhibit only one plateau; however, electrode formulations using additives display two well-defined plateaus. The appearance of the second plateau suggested that the presence of bismuth sulfide and copper sulfide would promote an electron transfer network across the electrode (for the oxidation reaction of Fe/Fe(II) species). Basically, the first plateau would correspond to the conversion of Fe to Fe(OH)2. The second plateau would suggest the presence of bismuth sulfide acting on the construction of the electron transfer network for the oxidation iron species. Fig. 3 illustrates specific discharge capacities for selected iron electrodes. 1.00

1.0

0.95

0.9 0.8 800000

1000000

1200000

Time (s)

Figure 1. Galvanostatic charge and discharge profile for a NiFe cell vs. Hg/HgO reference electrode at a C/5 rate.

228

Voltage(V)

Voltage(V)

1.1

40

0.90 0.85

5

0.80 244000

245000

246000

50 30 20 10 247000

Time (s)

Figure 3. Constant current discharge curves of an iron electrode (loaded 3%wt CuSO4, at C/5).

Proceedings of SEEP2015, 11-14 August 2015, Paisley Our experimental results are in line with previous publications, not only in the sense that bismuth sulphide increases the performance of the battery, but also in our observation of a conditioning period (25-30 cycles) that is required for the batteries to achieve the steady state [15, 30, 31].

25

ηQ %

20 15

3%CuSO4 Fe-PTFE Fe-Bi2S3 5.5%CuSO4

10 5 0

0

20

40

Cyclic voltammetry of in-house made iron electrodes was used to investigate the electrochemical properties of the electroactive material.

60

Number of cycles

Figure 4. Discharge capacity for the iron electrode versus cycle number.

0.00010

Aiming to develop electrode formulations that minimise the evolution of hydrogen, sulphurcontaining additives were tested at different concentrations. Our final results indicate that the addition of 3.5%wt CuSO4 on electrode formulations render a reduced production hydrogen. Likewise, the use of bismuth sulphide at a concentration of 4.5wt% would have similar effects. This experimental observation can be rationalised in terms of a heterogeneous reaction between different species as indicated in Equation (6).

S 2- + Fe(OH)2 « FeS + 2OH -

(6)

Table 2. Experimental results Content QT ηQ -1 Bi2S3 CuSO4 (mAh g ) 5 % 3% 5% 5%

210 190

22 19

229

Ox1

5th

30th-Fe-Bi2S3 30th-Fe-Bi2S3-CuSO4

Ox2

Ox0

0.00000 0.00002

a)

-0.00005

Red1

-0.00010 -0.00015

without additives

0.00001 Current/mA

As can be seen from Fig. 4, copper sulphate would render electrode formulations with comparable efficiency to its bismuth sulphide counterparts. However, in the long run, electrodes fall apart, and electrode capacity decreases rapidly after the conditioning period (first 30 cycles) is finished; this issue should be tackled by increasing the structural integrity of the electrode without compromising its electrical conductivity, nor increasing electrolyte decomposition. These aspects merit further investigation and are proposed for future work.

Current/mA

0.00005

b)

0.00000

-1.2

-1.0

-0.8

-0.6

-0.00001

-0.00002

-1

Potential / V vs.Hg/HgO

0

1

Potential / V vs.Hg/HgO

Figure 5. The voltammograms of the hotpressed type iron powder electrode with and without additives. Fig. 5 reveals two oxidation peaks -0.81 V (Ox1), -0.6 V (Ox2) (vs./HgO), which are related to the oxidation of Fe0 to FeII and FeII to FeIII; and two reduction peaks -0.91 V (Red1), -1.07 V (Red2) corresponding to the reduction of FeIII to FeII and FeII to Fe0. The presences of additives (copper sulphate and bismuth sulphide) render oxidation peaks slightly displaced, this is, -0.78 V (Ox1) and 0.64 V (Ox2), and reduction peaks -0.9 V (Red1). No reduction peak corresponding to a reduction of FeII to Fe0 (Red2) was visible; one possible explanation is that the hydrogen evolution overshadowed the peak, as has been suggested [19, 32-34]. Figure 6. The XRD measurement on the prepared iron electrodes modified with bismuth sulfide and copper sulfide after had been exposed to charging and discharged after 60 cycles.

500

Proceedings of SEEP2015, 11-14 August 2015, Paisley Council for supporting this work (EP/K000292/1; SPECIFIC Tranche 1: Buildings as Power Stations.

Fe

intensity

400 300

Return the manuscript by 5th July 2015 to:

200 100 0

Fe2S4

20

Bi Cu2O

40

Fe 2S4 Fe Cu2O

60

Cu 2O

Fe

Dr Abed Alaswad Email: [email protected] Conference Secretary, SEEP2015

80

2θ (degrees)

Figure 6. X-ray powder diffraction pattern for the iron electrode after extended cycling. The X-ray powder diffraction pattern for the iron electrode modified with bismuth sulfide and copper sulfide after cycling showed the present of Fe, Fe3S4 and Cu2O, and the elemental bismuth. The XRD analysis reveals that no definite evidence of reactions between iron and copper were found.

4

CONCLUSIONS

Aiming to develop highly efficient NiFe cells to store large amounts of energy coming from intermittent sources such as the wind or the sun, iron electrodes based on Fe/Bi2S3CuSO4 were developed. Our experimental results reveal that the addition of bismuth sulfide and copper sulfide to the iron electrodes increases the overall performance of the battery. The relatively large capacity of nearly 210 mA/g was achieved by using either 3% to 5%wt copper sulfide or 5% wt. bismuth sulfide additives. Coulombic efficiencies were relatively modest (25%); however, due to the fact that we have used commercial grade reactants and materials, this technology definitely have the potential to be further developed so NiFe cells, in the long run, could provide a cost-effective solution to large scale energy storage.

ACKNOWLEDGEMENTS The authors would like to acknowledge the U.K. Engineering and Physical Sciences Research

230

REFERENCES

[1] M.E. Kintner-Meyer, P. Balducci, V. Viswanathan, C. Jin, X. Guo, T. Nguyen, Energy Storage for Power Systems Applications: A Regional Assesement for the Northwest Power Pools, 2010. [2] A. Castillo, D.F. Gayme, Energy Conversion and Management, 87 (2014) 885-894. [3] D. Fernandes, F. Pitié, G. Cáceres, J. Baeyens, Energy, 39 (2012) 246-257. [4] M.-M. Titirici, R.J. White, C. Falco, M. Sevilla, Energy & Environmental Science, 5 (2012) 6796-6822. [5] P.P. Varaiya, F.F. Wu, J.W. Bialek, Proceedings of the Ieee, 99 (2011) 40-57. [6] Z.G. Yang, J.L. Zhang, M.C.W. KintnerMeyer, X.C. Lu, D.W. Choi, J.P. Lemmon, J. Liu, Chemical Reviews, 111 (2011) 3577-3613. [7] A.S. Rajan, S. Sampath, A.K. Shukla, Energy & Environmental Science, 7 (2014) 1110-1116. [8] N. Yabuuchi, S. Komaba, Science and Technology of Advanced Materials, 15 (2014) 043501. [9] H. Pan, Y.-S. Hu, L. Chen, Energy & Environmental Science, 6 (2013) 2338-2360. [10] D. Linden, T.B. Reddy, Handbook Of Batteries, McGraw,2002, 2002. [11] C.-Y. Kao, Y.-R. Tsai, K.-S. Chou, Journal of Power Sources, 196 (2011) 5746-5750. [12] J.O. Gil Posada, P.J. Hall, Journal of Power Sources, 262 (2014) 263-269. [13] A.K. Manohar, C. Yang, S. Malkhandi, G.K.S. Prakash, S.R. Narayanan, Journal of the Electrochemical Society, 160 (2013) A2078A2084.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [14] S. Malkhandi, B. Yang, A.K. Manohar, G.K.S. Prakash, S.R. Narayanan, Journal of the American Chemical Society, 135 (2012) 347353. [15] J.O. Gil Posada, P.J. Hall, Journal of Power Sources, 268 (2014) 810-815. [16] G. Huo, X. Lu, Y. Huang, W. Li, G. Liang, Journal of The Electrochemical Society, 161 (2014) A1144-A1148. [17] A.K. Manohar, S. Malkhandi, B. Yang, C. Yang, G.K. Surya Prakash, S.R. Narayanan, Journal of The Electrochemical Society, 159 (2012) A1209-A1214. [18] D.M. Rice, Journal of Power Sources, 28 (1989) 69-83. [19] C.A. Caldas, M.C. Lopes, I.A. Carlos, Journal of Power Sources, 74 (1998) 108-112. [20] U. Casellato, N. Comisso, G. Mengoli, Electrochimica Acta, 51 (2006) 5669-5681. [21] J.O.M. Bockris, N. Pentland, Transactions of the Faraday Society, 48 (1952) 833-839. [22] K.-S. Chou, C.-Y. Kao, Meeting Abstracts, MA2009-02 (2009) 269. [23] F. Gros, S. Baup, M. Aurousseau, Hydrometallurgy, 106 (2011) 127-133. [24] Q. Meng, G.S. Frankel, Journal of The Electrochemical Society, 151 (2004) B271B283. [25] C.G. Morales-Guio, S.D. Tilley, H. Vrubel, M. Gratzel, X. Hu, Nat Commun, 5 (2014). [26] I.N. Popescu, C. Ghiţă, V. Bratu, G. Palacios Navarro, Applied Surface Science, 285, Part A (2013) 72-85. [27] U. Sarac, M.C. Baykul, Advances in Materials Science and Engineering, 2013 (2013) 7. [28] S. Virtanen, H. Wojtas, P. Schmuki, H. Böhni, Journal of The Electrochemical Society, 140 (1993) 2786-2790. [29] Q. Zhao, Y. Liu, E.W. Abel, Materials Chemistry and Physics, 87 (2004) 332-335. [30] J.O.G. Posada, P.J. Hall, Sustainable Energy Technologies and Assessments. [31] J.O. Gil Posada, P.J. Hall, Journal of Power Sources, 262 (2014) 263-269. [32] J. Černý, K. Micka, Journal of Power Sources, 25 (1989) 111-122. [33] E. Bednarkiewicz, Z. Kublik, Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, 106 (1980) 61-69. [34] J. Černý, J. Jindra, K. Micka, Journal of Power Sources, 45 (1993) 267-279.

231


LCA OF IN-HOUSE PRODUCED SMALL-SIZED VANADIUM REDOX-FLOW BATTERY Michele Dassisti1, Piero Mastrorilli2, Antonino Rizzuti2, Pasqua L’Abbate2, Gennaro Cozzolino3, *, Michela Chimienti3 1

DMMM, Politecnico di Bari (michele.dassisti)@poliba.i DICATeCh, Politecnico di Bari (piero.mastrorilli; Antonio.rizzuti; pasqualabbate)@poliba.it 3 InResLab, (gennaro.cozzolino; michela.chimienti)@inreslab.org

2

(*) corresponding author

Abstract Storage systems are becoming one of the most critical components in the energy management, mainly due to the discontinuity characteristics of renewable sources. Redox flow batteries (RFB) are one of the most promising technologies as electrochemical energy storage system, because of the independence of energy and power rating, fast response, room temperature operation, and extremely long life. It is susceptible of a strong increase of its use in the next future, in particular in combination to a wide range of renewable energy sources (e.g. solar, eolic, tidal). Key factors such as the energy density and the operating temperature range depend on the properties of the battery electrolyte. This promising feature brings new questions to solve: the sustainability is one of the most critical to be faced. In this work, the vanadium electrolyte environmental sustainability is analyzed. Electrolyte is one of the most important component of the battery, having a strong impact particularly on its use. A comparative LCA analysis is performed to appreciate its potentialities for domestic use under safety conditions, providing some useful indication on its drawbacks. Keywords: Life Cycle Assessment; Energy and environmental planning tools. 1 INTRODUCTION The energy production from renewable sources have grown dramatically in these last years. They present new challenges regarding the possibility to match the power requirements because of their fluctuation over time independently from demand [1]. This characteristic requires suitable storage methods to become completely reliable as primary sources of energy. Essentially, it is necessary to decouple the production from the use: energy from these sources must be stored when is produced in excess and then released when production levels are less than the required demand [2]. Redox flow batteries seems to be a promising device to win the challenge put to endeavour the renewable energy production systems. This descends from their almost infinite storage capability based on liquid electrolyte properties, the high reliability due to their

technological simplicity as well as from their economical sustainability, being the overall investment competitive along the whole life cycle. Vanadium electrolytes have in fact an almost unlimited life cycle. All these facts make vanadium redox-flow batteries very interesting as a diffuse storage mean for most civil uses. Therefore, the question of their sustainability becomes of utmost importance, being these devices doomed to become our life companion in a near future. A preliminary life cycle assessment (LCA) study is here presented for an homemade short stack vanadium redox flow battery (VRFB). The battery was developed within an industrial funded research project aimed at optimising smallsize VRFB for future widespread civil application. Before new technologies enter the market, in fact, their environmental superiority over competing technologies 232


interpretation with proposals for improvement. Most of the data are direct, as measured in the laboratory where the battery was built and operates. Upon necessity, additional data, have been integrated from Eco-invent database or scientific literature. In particular, this latter approach was adopted for vanadium composites, and the reagents: VCl3 andVOSO4. The inventory of emissions and energies has been performed using the CMLCA inventory tool [6]. The impact assessment was made by classifying the inventory results with respect to toxicity categories, damages on human health, damages on eco-system, resources and global warming potential. The following calculation methods were used to this aim: ECOtox, IMPACT2000+, ReCiPe.

must be asserted based on a life cycle approach to appreciate the environmental impacts and resources. Life Cycle Assessment (LCA) investigates environmental impacts of e.g., systems or products from cradle to grave throughout the full life cycle, from the exploration and supply of materials and fuels, to the production and operation of the investigated objects, to their disposal/recycling. In the present paper the whole product’s life cycle has been considered, from raw material acquisition to the final battery disposal. Few LCA studies has been performed for redoxflow batteries so far. In [3] a comparison is provided for big size Vanadium Redox Flow Battery (VRFB) against a lead-acid (PbA) battery systems for a Swedish scenario. Still strong uncertainties affect the analysis, since two different stages of technology maturity are compared: in particular, VRFB are even today far from being a largely industrialised systems. As such, stable process parameters as well as maintenance experiences are rarely comparable. A wider perspective is discussed in [4] where all components of a batteryphotovoltaic system are addressed. It is clearly stated that, when dealing with relatively new technologies, strong uncertainties on production efficiency are present. This in turn may bring to inconsistent system boundaries. This question is clearly discussed in [5] that proposes a dynamic LCA approach against the diffused status-quo LCA All the scarce literature sources related to LCA of VRFB refer to conventional batteries: they never address a self-produced battery.

Goal and scope definition The goals of this study is to quantify environmental as well as the effect on the human-health of a small size in-house made vanadium redox-flow battery (VRFB) as in Fig. 1.

2 LCA METHODOLOGY ADOPTED This section outlines the main LCA steps performed. The LCA study was conducted according to the ISO 14040 series, considering four phases: goal and scope definition, life-cycle inventory (LCI), lifecycle impact assessment (LCIA), and

Figure 1. VRFB made by the authors. The scope was to characterise a small-unit VRFB against the existing large batteries and to evaluate its impact against these latter. The life-cycle phases of considered to this aim

233


were production and assembly of materials and components, use and disposal of VRFB. The cumulative environmental performance of the VRFB were considered.

components come in fact from South Italy; the same was for electrolyte reagents. LCI - Assumptions and hypotheses A cradle-to-grave cycle was assumed for the LCA analysis of the VRFB: Cell Casing; Anode and cathode production; Carbon felt production; Hydraulic system assembly; Electrolyte synthesis; overall assembly; Use and Disposal. Cell casing: includes bipolar plates in polypropylene charged with graphite, cells in polypropylene, terminal plates in stainless steel, steel screws, brass current collectors made, End Cell in polypropylene, and a fluoropolymer elastomer. Gasket for the stack hydraulic sealing. Anode and cathode electrodes are made of carbon felt; The Separator is a SPEEK membrane in sulfonated peek; The Hydraulic System consists of PVC pipes, valves, and flux meters, and polypropylene tanks. As concern, only pump energy consumption was considered. Electrolyte: the synthesis of the Vanadium electrolyte (V2S2Cl5) was included. The electrolyte was prepared starting from, VOSO4•2H2O, VCl3, HCl, H2SO4 and deionized water. In order to obtain the electrolyte solution, the required amount of VO2+ and V3+ precursors was dissolved in 1M H2SO4 + 2M HCl solution. The mixture was magnetically stirred for 3 h. Reaction and mixing energy has been considered too, expressed in terms of power and storage energy capacity to enable evaluation independently from sizing. In addition, all the raw material extraction and their production were taken into account. Energy: input energy to the hydraulic system and the charge energy per cycle. Transportation and packaging: all transportation for raw materials were considered. Packaging were not considered,

Functional Unit and system boundaries The analysis was referred to a unit power in order to assure comparability of others similar applications. The functional unit is in fact an electricity storage system (see tab. 1 for the other characteristics) with a rated power of 0,15kW. These specifications refers to an electricity requirements of a small domestic facility. Tab.1 FU characteristics Characteristics Value Number of stack 1 Nominal Power 150 W # cells/stack 5 Average voltage at end 6V discharge (SoC=0.2) Energy density of 36,18 Wh/l electrolyte electrolyte Volume 6l Overall efficiency 0,85 Average Current 25, A Charge energy 176.47 Wh Discharge energy 127.5 Wh Cycle time(Charge and ~3 h Discharge) This allowed us to provide a reference for relating process inputs and outputs to the inventory, and impact assessment for the LCIA (EPA, 2013). The analysis considers only the storage system, including the hydraulic system. No control unit were included in the analysis. The impact of electricity production for charging the battery was included. The VRFB is assumed – as in reality - to be produced, assembled and used in Apulia (South Italy). All the VRFB

234


and reporting categories impact: Human Health Photochemical Oxidation (Ph Ox), Human Health Ionizing Radiation (Ion Rad), Human Respiratory Health Effects, Human Health Human Toxicity (see figure 2).

provided its impact was not significant on the overall life cycle. Use and Disposal The hypothesis of a continuous running for 24 hrs/day over the period of 20 years was made, with an average energy delivery of 1,2kWh/day for 20 years. Vanadium electrolyte is assumed to have a very long lasting life and its only treatments are filtering before re-use, provided it is selfrecovery. The electrolyte with active material is thus assumed to last indefinitely. A deionized water refill was assumed of 100 ml in 20 years. The only consumables are the SPEEK ionic membranes, that were assumed to be replaced every 5 years. As concerns the hydraulic system, a 5-years maintenance is assumed with replacement of seals. At the end-of-life membranes are brought to landfill as well as the pumps. All the other materials has been considered as fully reusable. LCIA impact categories The LCIA phase involves translating the environmental burdens identified in the LCI into environmental impacts with the help of an impact assessment method. There are many well-established methods and models in LCIA; in our analysis 3 method were used: USEtox ™ model (Rosenbaum et al., 2008), Impact 2000+ method and ReCiPe Method. The first two were selected to analyze toxicity in distinct impact categories, with particular concern to carcinogenic fraction of human toxicity. From ReCiPe there are 21 impact categories, that have been aggregated into 3 environmental damage indicators: Damage to human health, Damage to ecosystems, Damage to the availability of resources. From these a single indicator was derived: the climate change.

Porduction

Use

Disposal

TLC

1.0E-02 7.5E-03 5.0E-03 2.5E-03 0.0E+00 Ph. Ox. Ion. Rad. Resp. Eff. U. Tox.

tot

Figure 2. Toxicity category (Impact 2002+) From the graph it is observed that the total damage on human health comes more from category Human Respiratory Health Effects. The main contribution is given by the use phase. The high values of the use phase originates by the use of fossil fuels for electricity employed at the start of the battery at each cycle. The production process contribution concern 75% from the electrolyte component production. With the USEtox™ method distinguish between carcinogenic and non–carcinogenic Human Toxicity (see figure 3).

Carc

Non Carc

HU Tox Tot

6.0E-05 4.5E-05 3.0E-05 1.5E-05 0.0E+00 Porduction

3 RESULTS AND DISCUSSION The results obtained are grouped in three graphs. The first show categories of toxicity obtained using the method Impact 2002+ ,

Use

Disposal

TLC

Figure 3. Toxicity category (USEtox™)

235


renewable sources can significantly reduce the resulting impact, as it is the case for the use of this kind of batteries.

The graph in figure 3 confirms the most impactful from the point of view of toxicity due to the use phase. ReCiPe method results are shown in the figure 4 and 5. The figure 4 shows the various parts that make up the battery and its environmental weight calculated aggregate indicators: Human Health total, Resources Depletion Total, Total Quality Ecosystem.

AKNOWLEDGEMENTS The innovative contents described in this paper are disclosed after the permission of Duferco Energia SpA company, which committed to InResLab the research project called "VaESS - Vanadium-Flow batteries Energy Storage System"

Note that the component that has a greater weight in the environmental impact of the production phase is the electrolyte. In addition, the highest damage is over exhaustion of natural resource, followed by the damage on human health. In the next figure 5 the three phases are reported: battery production, use and disposal and the complete life cycle ( TLC ). Also in this case , the data from the Recipe method show that the phase highest impact is the use of the VRFB.

REFERENCES [1] H. Ibrahim, A. Ilinca, e J. Perron, «Energy storage systems—characteristics and comparisons», Renew. Sustain. Energy Rev., vol. 12, n. 5, pagg. 1221– 1250, 2008. [2] I. Hadjipaschalis, A. Poullikkas, e V. Efthimiou, «Overview of current and future energy storage technologies for electric power applications», Renew. Sustain. Energy Rev., vol. 13, n. 6, pagg. 1513–1522, 2009. [3] C. J. Rydh, «Environmental assessment of vanadium redox and lead-acid batteries for stationary energy storage», J. Power Sources, vol. 80, n. 1, pagg. 21– 29, 1999. [4] C. J. Rydh e B. A. Sandén, «Energy analysis of batteries in photovoltaic systems. Part I: Performance and energy requirements», Energy Convers. Manag., vol. 46, n. 11, pagg. 1957–1979, 2005. [5] M. Pehnt, «Dynamic life cycle assessment (LCA) of renewable energy technologies», Renew. Energy, vol. 31, n. 1, pagg. 55–71, 2006. [6] «http://www.cmlca.eu/.» .

4 . CONCLUSIONS The LCA results show how the production of the battery has a moderate impact, including the effect toxicity, which usually is an important aspect for other types of existing types of batteries. The environmental footprint of the materials used for VRFB production is distributed over time: therefore the longer its use the lower the impact results. At the end of life being the material and the electrolyte fully reusable, only a small fraction goes to landfill disposal. The improvement should then be for the use phase. In our case, the highest impact is due to the use of Italian electricity mix to operate the battery at each cycle over the 20 years considered. Surely using energy from

236


Human Health, Total Ecosistem Quality, Total Total

Resources Depletation, Total E.Q. Climate Change

1.00E+01

7.50E+00

5.00E+00

2.50E+00

0.00E+00 Production

Hyd. Syst.

An-Cath.

Separator

Elect. V

Figure 4. Environmental weight calculated aggregate indicators

Human Health, Total Ecosistem Quality, Total Total

Resources Depletation, Total E.Q. Climate Change

1.0E+02

7.5E+01

5.0E+01

2.5E+01

0.0E+00 Production

USE

Disposal

TLC

Figure 5. Battery weight calculated aggregate indicators

237


Influence of carbon physiological properties on its catalytic activity towards oxygen reduction: A comprehensive review Kiranpal Singh, Fatemeh Razmjooei, Dae-Soo Yang, Min Young Song, and Jong-Sung Yu* Department of Energy Systems Engineering, DGIST, Daegu 711-873, Republic of Korea E-mail address: [email protected] (J.-S. Yu)

Abstract Realizing energy devices without carbon is next to impossible, due to its unique properties including a variety of platforms on which various applications can be carried out. Among all conductivity and surface area are the most highly desired properties of a carbon for electrocatalysis and energy storage. Herein we report various steps taken towards enhancing these physiological properties of carbon and their ultimate effect on the electrocatalytic reduction of oxygen in cathode part of fuel cell. Although surface area and conductivity are two properties of carbon which are usually inversely proportional to each other, our adopted methodologies have proven to optimize both the properties of carbon. Furthermore it was also found that the surface area and conductivity are the major contributor in the enhancement of ORR activity, and the optimized value of these two parameters have huge implications on the electrocatalytic activity of the carbon.

1. Introduction: Fuel cells are regarded as the most

The biggest hurdle in these fuel cells has

promising renewable energy conversion

been the effective achievement of oxygen

devices due to their high energy efficiency

reduction reaction (ORR) in the cathode,

along with negligible pollutant emission.[1]

which otherwise needs a high amount of

238


catalyst to decrease high overpotential.[2]

conductivity and surface area of a carbon

To date, Pt loaded on carbon has been

support.[7–11]

considered to be the best catalyst for this

provide a direct 4e- pathway for the ORR,

purpose. However, the high cost of Pt and

while a high surface area enables the

the ~ 300 mV of over potential required to

maximum

formation of a triple-phase

gain the minimum current hamper the

boundary

and

commercialization of these fuel cells. On the

obtaining both surface area and conductivity

other hand, after the first report R. Jasinski,

at a same time is found to be very

worldwide research on heteroatom-doped

challenging.[12]

carbon was sparked.[3] Hence, in recent

conductive carbon nanotubes are found to

years, research in this field has focused not

have limited surface area (100 m2 g-1 for N-

only on nonprecious metals but also on

doped carbon nanotubes) with a relatively

metal-free doped carbon catalysts for the

low density of active catalytic sites.[13] On

ORR.[4-6] Interestingly, from the first use to

the other hand, harsh chemical oxidation of

replace Pt in cathode part to today's

the graphitic structure results in an enhanced

applications

heteroatom-doped

surface area carbon, but will significantly

carbons are used, the one common substrate

reduce the graphitic integrity of the carbon

“carbon” is irreplaceable, and till now no

structure, which results in reduced electrical

other catalyst support as efficient and cheap

conductivity.[14] Highly porous carbon

as carbon has been found.

fibers with high surface area demonstrated

where

High

conductivity

active

sites.

For

can

However,

instance,

highly

poor electrochemical conductivity, which

Carbon has been always considered

eventually

to be a fundamental unit, due to its

resulted

in

a

poor

ORR

activity.[15]

omnipresence in every aspects of moving and static entities. The physiochemical

Herein, we will present several

properties of the carbon substrates play an

studies that we have undertook to increase

important role in enhancing the activity and

the surface area without significantly losing

durability of both precious and non-precious

the

catalysts towards the ORR. Several studies

Furthermore, the effect of these properties

have proven/shown the improvement in the

on the capability of carbon to carry out the

activity and selectivity of a catalyst for the

reduction of oxygen is also evaluated. There

ORR, as a result of the high electrical

are some literature already available, which

239

conductivity

and

vice-versa.


have shown the several methodologies to

carbon by carbonizing a conductive polymer

improve the conductivity of the carbon

polyaniline (PANI) in presence and absence

substrate without significantly losing the

of iodine. To understand the effect of iodine

surface area.[16,17] It was reported that

treatment we have also used varying amount

doping the conducting polymer with iodine

of iodine for the present study. The

can significantly improve the conductivity,

immediate effect of iodine treatment was

which was attributed to the formation of

found on the amount of carbon obtained,

iodine polarons between the band gap of

furthermore with various analysis techniques

polymers. Similar results were obtained for

it was confirmed that the iodine treatment

iodine-doped CNTs, where the increase in

increases

conductivity was attributed to the formation

conductivity of the resultant CPANI.

of a charge complex between iodine and the

obtained carbon. It was found that the ID/IG value, which measures the degree of

was found that the conductivity was

disorder in graphitic structure, is decreased

proportional to the amount of iodine

from 1.11 for CPANI to 1.06 for CPANI-02I,

used.[18] Another approach that we have the

treatment

of

the

huge effect on the physical characteristics of

area of the

polyaniline-derived carbon (I-CPANI). It

is

and

of iodine during carbonization of PANI have

the effect of iodine treatment on the

applied

graphiticity

It can be seen from table 1, presence

carbon matrix. Therefore, we have evaluated

conductivity and surface

the

and to 1.01 for CPANI-05I, which clearly

ordered

signifies that not only presence of iodine but

mesoporous carbon with the Fe metal

the concentration of iodine also have huge

species (Fe-OMC). During this approach we

implications on the graphiticity of the

found that this approach results in an OMC

obtained carbon. Similarly, sharp increase in

with much improved conductivity with the

the C/O ratio up to the CPANI-05I sample is

minimum loss of surface area.[19]

observed with the iodine treatment. It was found that the presence of iodine at high temperature helps in the reduction of oxygen

2. Effect of iodine treatment on

groups present on the carbon surface.[20]

conductivity and ORR

However, a further increase in the amount of

In present approach we have tried to

iodine has an adverse effect and the amount

synthesize highly porous and conductive

of oxygen slightly increases. We propose

240


that this increase in the C/O ratio helps in

iodine treatment apart from increasing

increasing the conductivity of the prepared

conductivity of CPANI also increases its

catalysts.

iodine

surface area significantly. This increase in

treatment, area of the sp -hybridized carbon

surface area can be attributed to the etching

peak (Ac–c) decreased, and the peak area of

of heteroatoms including oxygen due to the

the sp2-hybridized carbon peak (Ac=c)

iodine treatment at high temperature.

Furthermore,

after

3

increased. Consequently, the Ac=c/Ac–c

The comparative variations in the

ratio of the iodine treated samples increased

physical,

from 1.63 for CPANI to 2.71 for CPANI-05I

chemical

and

electrochemical

properties of the studied CPANI samples,

(Table 1), which indicates that the iodine

each as a function of iodine concentration,

treatment helps to increase the amount of

are plotted in Fig. 1. It can be seen with an

sp2-hybridized carbon in the CPANI-XI

increase in the iodine concentration there is

samples.[17]

a linear decrease in the nitrogen-heteroatom

The conductivity of the prepared

concentration, but a continuous increase in

carbon catalysts were measure by measuring

the surface area and conductivity and a

their current–voltage characteristics under

slight increase in S content have also been

the application of pressure. Table1 shows

observed. Oxygen reduction by carbon can

the comparative conductivity profiles of the

be explained in terms of the interplay

untreated and I-treated samples determined

between the heteroatom content, surface

using a home-built 4-point probe apparatus.

area, and electrical conductivity of the

The conductivity of CPANI increases with

carbon

increasing iodine treatment and reaches a

CPANI-05I shows the best ORR catalytic

maximum value for CPANI-05I. In fact, the

activity among the studied samples although

CPANI-05I shows an exceptionally high

it has a considerably lower N content

conductivity of 19.8 S cm-1, which is about 3

compared to CPANI and CPANI-02I. The

times that of the untreated CPANI (6.5 S

better

-1

cm )

at

18

MPa.

This

electrocatalyst.

ORR

activity

Interestingly,

of

CPANI-05I

compared with CPANI and CPANI-02I can

improved

conductivity of the CPANI-05I sample was

be

attributed to the presence of higher amount

conductivity, surface area and improved S

of graphitic carbon and reduced amount of

content.

oxygen. Furthermore, it was also found that

241

attributed

to

its

higher

electrical


Table1. Influence of iodine treatment on the physical properties of obtained CPANI Samples

C1s

O1s

Ac=c/Ac-c

Prepared

ID/IG

Conductivity Surface area

ratio

(S/cm)

(m2/g)

87.8

6.4

1.6

1.1

6.5

855

CPANI-02I

90.2

5.1

1.9

1.06

14.05

1104

CPANI-05I

93.4

2.8

2.7

1.01

19.8

1130

CPANI-10I

91.9

5.07

2.03

1.07

17.8

1060

Untreated CPANI

Therefore, it can be concluded that a high N content, which creates active sites, is not the only determining

factor

for

improved

activity. An electrocatalyst should also have a high conductivity, as well as a high surface area for an efficient ORR,[21-23] which can be efficiently and easily achieved by treating PANI with a controlled iodine concentration. On the other hand, although CPANI-10I has a slightly lower or similar surface area and conductivity and a slightly higher S content compared with CPANI-05I, the former possesses a lower N content and a higher

Figure 1: The effect of the iodine

amount of ORR-inactive SOx among the

concentration on the nitrogen content,

different S species thus, has a lower

surface area, conductivity, sulfur content

electrocatalytic activity than the latter

and onset potential of I-treated and untreated CPANI samples.

242


3. Iron

treated

OMCs

for

conductivity was ascribed to the catalytic

conductivity and ORR

graphitization of OMC surface in presence of Fe during carbonization. This claim was

In present approach we have shown the

further supported by the increase in C/O

exact role of Fe in the preparation of Fe-N

ratio in N-OMC (FePc) in comparison to

catalyst. For this thin platelet Fe–N-OMC

that of untreated N-OMCs. It was also found

catalysts are synthesized for the first time

that the surface area of Fe treated samples

using Fe-phthalocyanine as a nitrogen, iron

are negligibly lesser than that of untreated

and carbon precursor and mesoporous

N-OMCs, however, the micro pore volume

platelet SBA-15 silica as a hard template,

of Fe treated sample is much higher than

wherein the silica template and Fe were

that of untreated counterparts. This increase

etched out using HF washing (N-OMC

in micropore volume was due to the removal

(FePc)). After carrying out the physical and

of Fe particles from the N-OMCs (FePc)

electrochemical analysis on the prepared Fe-

surface which leaves behind micro cavities.

N-OMC catalyst it was found that the physical presence of Fe is not necessary for the N-doped carbon catalyst to be active for ORR in both alkaline and acidic media, however it is necessary for the preparation of a highly active N-doped carbon catalyst. The presence of Fe was found to have negligible effect on the ORR performance of the prepared catalyst. Interestingly, the samples prepared in presence of Fe showed better performance than the sample prepared in absence of Fe. After measuring physical properties

Figure 2: The effect of the Iron treatment

of the OMCs it was found that the

on the nitrogen content, surface area,

carbonization of N-OMCs in presence of Fe

conductivity, and onset potential of OMC,

have huge effect on the conductivity of the

OMC (Fe), N-OMC (Pc) and N-OMC

final product Table 2. This increase in

(FePc).

243


Table 2. Influence of iron treatment on the physical properties of obtained OMCs. Samples

C1s

O1s

Prepared

Conductivity Surface area Vmicro (S/cm)

(m2/g)

(cm3/g)

OMC

94.68

5.32

0.14

1567

0.70

N-OMC (Pc)

92.89

3.17

0.18

1550

0.94

N-OMC (FePc)

93.24

3.14

0.56

1530

1.15

Figure 2 portrays the cumulative

(FePc), in terms of better conductivity and

effect of surface area, Nitrogen content and

surface area, however, it does not take part

conductivity on the ORR performance of the

in the ORR actively.

catalyst in basic medium. Similar to iodine

4. Effect

treated carbon here also it is clear that the

of

carbonization

temperature on surface area

ORR performance of the catalyst depends

and conductivity

widely on the surface area and the A little

Apart from chemical factors there are

variation in OMC-Fe is seen where,

physical factors as well which can affect the

conductivity is higher than N-OMC (Pc) but

physiological properties of a carbon among

the ORR activity is lesser than the later one.

which one is high temperature carbonization.

This is due the absence of N-heteroatom in

Generally with increase in carbonization

OMC-Fe, this result give a direct indication

temperature the crystallite growth of carbon

that N-doping is the most important

also

parameter for the ORR. Whereas the

graphitized and conductive carbon. The

conductivity and surface area are the

other interesting phenomenon that occurs

supporting parameters

in

during carbonization of biomass precursors

enhancing the activity but cannot give rise to

is the, evaporation of salts and organic

ORR activity. Therefore it can be easily

moieties which leaves behind huge amount

decipher from the above results that

of fissures in the carbon framework and

involvement of Fe in the present system is

eventually leads to enhanced surface area in

only in giving superior edge to the N-OMC

the same.[24-26] In this regard we have tried

conductivity of the samples.

which helps

244

increases

which

leads

to

much


to make carbon from few biomass sources and tried to optimize the carbonization temperature

to

get

the

Table 3 clearly portrays the effect of

maximum

temperature on the conductivity and the

conductivity, surface area, nitrogen content

surface area of the obtained leaves carbon.

and ORR activity.

Yellow leaves (LY-900) carbonized at 900 0

Razmjooei et.al. had shown that the

C

shows

the

minimum

conductivity

natural drying of biomass also have huge

whereas the LY-1100 shows the maximum

implication on the surface retention property

conductivity hence the direct correlation of

of the precursor during carbonization.[27] In

conductivity

an experiment performed by our group we

temperature can be established here. The

have collected leaves from Ginkgo tree in

increase in conductivity in LY0-1100 can be

two different seasons, summer and winter.

understood from the enhanced graphitization

The leaves turn colors from green during

of carbon at high temperature which is clear

spring and summer to yellow during fall

from increasing ID/IG and C/O value from

with short day-time sunlight and cold

900 to 1100 0C carbonization temperature.

weather. These Yellow leaves structure

Another important conclusion that can be

undergoes gradual changes by natural drying

derived from the above table is the effect of

during the short and cold days of autumn,

natural drying on the surface area of the

which induces gradual changes in leaf's

obtained carbon. From the total surface area

veins and structure. This can make it

as well as micropore volume of green leaves

possible for the leaves to maintain their

(LG-100) and LY-1000 it can be concluded

original structure gently and naturally

that the natural drying process can help

compared with that of green leaves prepared

maintaining the pore structure. The direct

by simple oven drying. This phenomenon

implication of these improved physiological

can create more well-preserved pores,

properties can be seen on the ORR activity

consequently giving larger surface area to

of these carbons.

the resulting carbon obtained from yellow leaves.

245

and

the

carbonization


Table 3. Influence of carbonization temperature and natural drying on the physical properties of the biomass carbon. Samples

C1s

O1s

ID/IG

Prepared

Conductivity Surface (S/cm)

area

Vmicro

ORR onset

3

(cm /g)

potential

(m2/g)

(V

vs.

Ag/AgCl) LY-900

91.01

5.61

0.98

5.04

367.12

261.42

0.83

LY-1000

91.4

4.98

1.02

6.56

486.44

377.72

0.91

LY-1100

92.75

4.6

1.11

7.08

501.06

368.53

0.86

LG-1000

90.45

6.55

0.97

6.12

138.17

100.52

0.91

which gives a better control over the

In other set of work we have pyrolysis

chemical properties of the obtained carbon.

temperature on the seaweed carbon.[21] The

In this regards N-doped porous carbon

pyrolysis temperature was a key parameter

nanofibers were prepared by electrospinning

that determined the trade-off between

the

heteroatom doping, surface properties, and

oxide)/poly(acrylonitrile)

electrical conductivity. The trade-off was

followed by carbonization, and the porosity

analyzed

ORR

was tuned by varying the ratio of PEO to

performance. The best performance of the

PAN.[14] As the PEO ratio increases, the

carbon electrodes obtained for SCup-1000

surface area also increases but electrical

(seaweed carbon carbonized at 1000 C) can

conductivity decreases. Thus, there is a

be attributed to its highly porous nature with

trade-off between porosity and electrical

a high mesopores surface area and good

conductivity. The carbonization temperature

electrical

with

was varied, and the effect of temperature,

catalytically active nitrogen and sulfur sites,

which in turn controls the nitrogen content

which significantly enhance the

and the amount of the graphitic phase, on

demonstrated

the

to

effect

obtain

conductivity

the

best

along

ORR

the

activity.

solution

oxygen

mixture

reduction

of

poly(ethylene (PEO/PAN)

reaction

(ORR)

performance was studied. Hence, as the

Apart from using the biomass it is

temperature increases, more graphitic phase

also possible to use organic compounds,

246


is formed, which is beneficial to the ORR,

carbon. The carbonization temperature is

but the nitrogen content decreases, which

found to play a very important role in

may not be favorable for the ORR. The

determining the physiological properties of

0

carbonization temperature of 1000 C was

carbon, and therefore it was found that

found to be the best for this system.

choosing a carbonization temperature which can provide optimized surface area and

5. Conclusion

conductivity is a very essential stage in

Achieving highly porous carbon with not

preparing the optimum ORR catalyst.

much affecting electrical conductivity has been one of the main subjects in carbon science.

Here,

comprehensive enhancement

we study

of

properties of carbon.

presented directed

the

a

6. Acknowledgments

towards

We thank all our group members for

physiochemical

generous contribution in one way or another.

It was shown that

This work was supported by NRF grant

surface area and conductivity are usually

(NRF 2010-0029245) and Global Frontier

inversely proportional to each other, but by

R&D Program on Center for Multiscale

implementing some techniques these carbon

Energy

properties can be optimized. Treatment of

System

(NRF-2011-0031571)

funded by the Korea Ministry of Education,

carbon surface with iodine and iron is shown

Science and Technology. Authors also

to have immense effect its surface area and

would like to thank KBSIs at Jeonju,

conductivity, which in turn shows excellent

Daejeon, and Chuncheon for SEM, TEM

ORR performance. It is also shown that

and XPS measurements.

these physiological properties are highly needed to further enhance the ORR activity. Carbon surface with only high surface area

7. References:

and conductivity cannot be an efficient [1] M. Lefe`vre, E. Proietti, F. Jaouen and J-

catalyst, and hence active sites are needed

P.

for carrying out the ORR efficiently.

Dodelet,

Iron-based

catalysts

with

improved oxygen reduction activity in

Furthermore, apart from chemical treatment,

polymer electrolyte fuel cells, Science, 324,

physical treatment is also an alternative to

71-74, 2009.

enhance the physiological properties of

247


[2] B.C.H. Steele and A. Heinzel, Materials

[8] D-S.

Yang,

D.

Bhattacharjya,

S.

for fuel-cell technologies, Nature, 414, 345-

Inamdar, J. Park and J-S. Yu, Phosphorus-

352, 2001.

doped ordered mesoporous carbons with different lengths as efficient metal-free

[3] R. Jasinski, A new fuel cell cathode

electrocatalysts

catalyst, Nature, 201, 1212-1213, 1964.

for

oxygen

reduction

reaction in alkaline media, J. Am. Chem. [4] Z. Lin, G.H. Waller, Y. Liu, M.L. Liu,

Soc., 134, 16127-16130, 2012.

and C.P. Wong, Simple preparation of nanoporous graphene

few-layer for

use

[9] F. Razmjooei, K.P. Singh, M.Y. Song

nitrogen-doped as

an

and J-S. Yu, Enhanced electrocatalytic

efficient

activity due to additional phosphorous

electrocatalyst for oxygen reduction and

doping

oxygen evolution reactions, Carbon, 53,

in

nitrogen

and

sulfur-doped

graphene: A comprehensive study, Carbon,

130-136, 2013.

78, 257-267, 2014. [5] S.A. Wohlgemuth, T.P. Fellinger, P. [10]

Jaker, and M. Antonietti, Tunable nitrogendoped

carbon

aerogels

as

Daems,

X.

Sheng,

I.F.J.

Vankelecom, and P.P. Pescarmona, Metal-

sustainable

free

electrocatalysts in the oxygen reduction

doped

carbon

materials

as

electrocatalysts for the oxygen reduction

reaction, J. Mater. Chem. A, 1, 4002-4009,

reaction, J. Mater. Chem. A, 2, 4085, 2014.

2013.

[11] Y. Zheng, Y. Jiao, M. Jaroniec, Y. Jin,

[6] R. Balgis, G.M. Anilkumar, S. Sago, T.

and S.Z. Qiao, Nanostructured metal‐free

Ogi, and K. Okuyama, Ultrahigh oxygen reduction

N.

activity

of

electrochemical catalysts for highly efficient

Pt/nitrogen-doped

oxygen reduction, Small, 8, 3550-3566,

porous carbon microspheres prepared via

2012.

spray-drying, J. Power Sources, 229, 58-64, 2013.

[12] S. Maldonado and K.J. Stevenson. Influence of nitrogen doping on oxygen

[7] Y. Cheng, H. Zhang, C.V. Varanasi, and

reduction

J. Liu, Highly efficient oxygen reduction

electrocatalysis

at

carbon

nanofiber electrodes, J. Phys. Chem. B, 109,

electrocatalysts based on winged carbon

4707, 2005.

nanotubes, Sci. Rep., 3, 3195, 2013.

248


[13] Y. Li, W. Zhou, H .Wang, L. Xie, Y.

[18] K.P. Singh, M.Y. Song and Yu J-S.

Liang, F. Wei, J-C. Idrobo,

Iodine-treated

Pennycook

S.J.

J. Mater. Chem. A, 2, 18115-18124, 2014.

reduction electrocatalyst based on carbon complexes.

carbon:

conductivity driven electrocatalytic activity,

and H. Dai, An oxygen

nanotube-graphene

heteroatom-doped

Nat.

[19] D-S. Yang, M.Y. Song, K.P. Singh and

Nanotechnol, 7, 394-400, 2012.

Yu J-S. The role of iron in the preparation

[14] D-S. Yang, S. Chaudhari, K.P. Rajesh

and oxygen reduction reaction activity of

and J-S. Yu, Preparation of nitrogen‐doped

nitrogen-doped carbon, Chem. Commun.,

porous carbon nanofibers and the effect of

51, 2450-2453, 2015.

porosity,

electrical

conductivity,

and

[20] S. Pei, J. Zhao, J. Du, W. Ren, and

nitrogen content on their oxygen reduction

H.M. Cheng, Direct reduction of graphene

performance, ChemCatChem, 6, 1236, 2014.

oxide films into highly conductive and

[15] O.K. Park, M.G. Hahm, S. Lee, H.I.

flexible graphene films by hydrohalic acids,

Joh, S.I. Na, Vajtai R, H. L. Joong, B-C. Ku,

Carbon, 48, 4466-4474, 2010.

and P.M. Ajayan, In situ synthesis of

[21] M.Y. Song, H.Y. Park, D-S. Yang and

thermochemically reduced graphene oxide

Yu J-S, Seaweed‐derived heteroatom‐doped

conducting nanocomposites, Nano Lett. 12,

highly porous carbon as an electrocatalyst

1789-1793, 2012.

for

the

oxygen

reduction

reaction,

ChemSusChem, 7, 1755-1763, 2014.

[16] Y.M. Choi, H. Choo, H. Yeo, N.H. You, D.S. Lee, B.C. Ku, H.C. Kim, P.H.

[22] C. You, S. Liao, X. Qiao, X. Zeng, F.

Bong, Y. Jeong and M. Goh, Chemical

Liu, Zheng R, H. Song , J. Zeng and Y.Li,

method for improving both the electrical

Conversion of polystyrene foam to a high

conductivity and mechanical properties of

performance doped carbon catalyst with

carbon nanotube yarn via intramolecular

ultrahigh surface area and hierarchical

cross-dehydrogenative coupling, ACS Appl.

porous structures for oxygen reduction, J.

Mater. Interfaces, 5, 7726-7730, 2013.

Mater. Chem. A, 2, 12240-12246, 2014.

[17] S. Adhikari, and P. Banerji, Enhanced

[23] D-S. Yang, D. Bhattacharjya, M.Y.

conductivity in iodine doped polyaniline

Song and J-S. Yu, Highly efficient metal-

thin film formed by thermal evaporation,

free

Thin Solid Films, 518, 5421-5425, 2010.

249

phosphorus-doped

platelet

ordered


mesoporous

carbon

for

electrocatalytic

oxygen reduction, Carbon, 67, 736, 2014. [24]

D.

Bhattacharjya

and

J-S.

Yu,

Activated carbon made from cow dung as electrode

material

for

electrochemical

double layer capacitor, J. Power Sources, 262, 224-231,2014. [25] C. Ruan, K. Ai and L. Lu, Biomassderived

carbon

materials

for

high-

performance supercapacitor electrodes, RSC Adv. 4, 30887-30895, 2014. [26] N.K. Chaudhari, M.Y. Song and J-S. Yu, Heteroatom-doped highly porous carbon from human urine, Sci. Rept., 4, 5221, 2014. [27] F. Razmjooei, K.P. Singh and J-S. Yu, Superior pore network retention of carbon derived from naturally dried ginkgo leaves and

its

enhanced

performance,

oxygen Catal.

reduction Today

http://dx.doi.org/10.1016/j.cattod.2015.06.0 12, 2015.

250


Application of Fuel Cell Technologies in the Transport Sector. Current Challenges and Developments.

A. Alaswad1 , A. Baroutaji2, and A.G. Olabi1 1. Institute of Energy and Engineering Technologies, University of the West of Scotland, Paisley; email: [email protected] 2. Cork Institute of Technology, Cork, Ireland; email: [email protected] Abstract The demand for clean power source which can be used to run the various types of vehicles on the road is increasing on a daily basis due to the fact that high emissions released from internal combustion engine play a significant role in air pollution and climate change. Fuel cell devices, particularly Proton Exchange Membrane (PEM) type, are strong candidates to replace the internal combustion engines in the transport industry. The PEMFC technology still has many challenges including high cost, low durability and hydrogen storage problems which limit the wide-world commercialization of this technology. In this paper, the fuel cell cost, durability and performances challenges which are associated with using of fuel cell technology for transport applications are detailed and reviewed. Recent developments that deal with the proposed challenges are reported. In the extended version of this paper, problems of hydrogen infrastructure and hydrogen storage in the vehicle will be covered. Keywords: fuel cell, PEM, cost, durability 1 BACKGROUND Due to the growing global concerns on the depletion of petroleum based energy resources, and the environmental pollution and climate change caused by the burning of fossil fuels, fuel cell technologies have received much attention in recent years owing to their high efficiencies and low emissions. A fuel cell is an electrochemical power source which converts chemical energy in the form of fuel directly into electrical energy. However, unlike other electro-chemical power sources such as batteries which store their reactants within a cell, the reactants are fed continuously to it from external stores. Also, the electrodes in a fuel cell are not consumed as in a battery, irreversibly in a primary cell and reversibly in a secondary cell, and do not take part in the reaction. Fuel cells are already used to generate electricity for other applications, including in spacecraft and in stationary uses, such as emergency power generators. Although the concept of a fuel cell was developed in England in the 1800s by Sir William Grove, the first workable fuels cells were not produced until much later, in the 1950s. During this time, interest in fuel cells increased, as NASA began searching for ways to generate power for space flights [1]. Several types of fuel cells are classified according to the electrolyte employed. The most popular type of fuel cells is 251

the Proton exchange membrane fuel cells, also known as polymer electrolyte membrane (PEM) fuel cells (PEMFC). PEMFC use a solid polymer as an electrolyte and porous carbon electrodes usually containing a platinum or platinum alloy catalyst. They are typically fuelled with pure hydrogen supplied from storage tanks or reformers. Hydrogen fuel is processed at the anode where electrons are separated from protons on the surface of a platinum-based catalyst. The protons pass through the membrane to the cathode side of the cell while the electrons travel in an external circuit, generating the electrical output of the cell. On the cathode side, another precious metal electrode combines the protons and electrons with oxygen to produce water, which is expelled as the only waste product; oxygen can be provided in a purified form, or extracted at the electrode directly from the air. PEM fuel cells are used primarily for transportation applications and some stationary applications. Due to their fast start up time and favourable power-to-weight ratio, PEM fuel cells are particularly suitable for use in passenger vehicles, such as cars and buses. Transport consumes about one quarter of the world total energy. In the case of internal combustion engines, a large part of the fuel energy is emitted as heat due to friction loss and exhaust gas.

Proceedings of SEEP2015, 11-14 August 2015, Paisley efficiency as long as the decrease of the vehicle 2 FUEL CELL IN TRANSPORTATION Road transport today is responsible for a weight in EVs over internal combustion vehicles significant and growing share of global warming are believed to have a significant effect on the and climate change. Moreover, it is almost GHG pollution reduction. entirely dependent on oil-derived fuels and Electric vehicles can be powered either by therefore highly vulnerable to possible oil price batteries or by fuel cells. It has been stated that shocks and supply disruptions. In cities and for any vehicle range greater than (100 miles), urban area, transport vehicles generate more fuel cells are superior to batteries in terms of pollution than any other single human activity mass, volume, cost, initial greenhouse gas and greenhouse gases emitted by or attributable reduction, refuelling time, well-to-wheels energy to this source are CFCs, carbon dioxide, nitrous efficiency using natural gas or biomass as the oxide, methane, hydrocarbons, nitrogen oxides source and life cycle costs. Several types of fuel (NO,) and carbon monoxide. However, as traffic cell exist, but PEMFCs are particularly welllevels are predicted to increase, road transport suited to automotive applications due to their will continue to be a significant contributor to low weight, small size, high durability, quiet greenhouse gas emissions (GHG). One approach operation, and responsive power output to fight against that is to increase the energy compared with other fuel cells types. However, efficiency of vehicles; another one is to decrease hybridization of fuel cell systems with high the GHG emissions resulting from operating specific energy-storage devices was found to them. For both approaches, electric vehicles have important advantages overcoming the (EVs), which are driven by batteries or fuel cell relatively slow FCS transient response, units, are a potential option for reducing improving the hydrogen economy and reducing emissions from the transportation sector. the warm-up time of the FCS to reach full EVs are proven to be significantly more efficient power. than internal combustion engines. Automobile Fuel Cell usage in the transport sector includes manufacturers have been compelled to shift part the application of this technology in forklift of their production from internal combustion trucks and other goods handling vehicles such as engines to (EVs) [1- 2]. The electrical motor airport baggage trucks, two- and three-wheeler converts more than 90% of the energy in its vehicles such as scooters, light duty vehicles storage cells to motive force, whereas internal (LDVs), such as cars and vans, buses and trucks, combustion drives use less than 25% of the trains and trams, ferries and smaller boats, energy in a gallon of gasoline. Additionally, in manned light aircraft, unmanned aerial vehicles contrast with internal combustion engine, the (UAVs) and unmanned undersea vehicles electric motor can be directly connected to the (UUVs), for example, for reconnaissance. wheels, so that no energy consumption is taken According to Carbon Trust, fuel cells could be place when the car is at rest or freewheeling. powering up to 491 million cars by 2050 [5] – a Moreover, the regenerative braking system can third of all the cars on the road. Governments return as much as half an electric vehicle’s around the world increasingly recognise the role kinetic energy to the storage cells. On the other FCEVs could play in the future low carbon hand, EVs are considered to be more transport mix, and several countries have made environmentally friendly than internal serious commitments to accelerate their uptake. combustions vehicles [3- 4]. Combusting fossil 3 CURRENT CHALLENGES AND fuels to power conventional vehicles releases DEVELOPMENTS GHG emissions and other pollutants from the Fuel cell technology is showing year-on-year vehicle exhaust system. In addition, there are growth, with more prototypes being unveiled. also emissions associated with producing Successful application of these technologies in petroleum-based fuels, notably emissions from the transport sector has taken place in Europe oil refineries. Electrical vehicles produce no and USA. However, the fuel cell industry is still GHG pollution when operating. However, it facing a number of challenges to depends on how electrical energy is produced, commercialization. Fuel cell cost is one major there can be substantially lower upstream GHG challenge, the durability of the unit and its emissions associated with producing hydrogen performance is another important one. Both fuel. The important increase of the energy challenges will be covered in this paper. 252

Proceedings of SEEP2015, 11-14 August 2015, Paisley densities by replacing the perfluorosulfonic acid 3.1 Fuel Cell Cost membranes that are the current industry standard Fuel cell costs can be broken into three with a membrane fabricated by ionic polymers. elements: the material and component costs, Higher power densities translate into more labor (design and fabrication), and capital cost of power per cell; hence a much smaller, lighter, the manufacturing equipment [6, 7]. It is clearly and cheaper stack can meet the same power seen that labor and capital costs can be reduced output. The proposed fuel cell cost given for by through mass-manufacturing. Material and ITM Power’s fuel cell was 35 $/kW. component costs, such as catalysts, membrane In a further patent, fuel cell stack design was and bipolar plates, are dependent on improved by Imperial College & University technological innovations and the market [8, 9]. College London [5]. The so called ‘FlexiAccording to Carbon Trust, in order to be Planar’ design uses a layered arrangement of competitive with internal combustion engine laminated, printed circuit board materials, vehicles, automotive fuel cells must reach bonded on top of each other to create a fuel cell approximately $36/kW. Platinum (Pt) is a stack with internal fuel, water and air channels. precious metal, with around 250 tonnes annually These boards lead to cost benefits over production. It is currently mined in South Africa, conventional fuel cell systems by eliminating the Russia and North America. Estimated world need for several components that are normally reserves of Pt are >30,000 tonnes. Given its high used in a conventional fuel cell. The biggest value, the majority of Pt used in FCEVs is likely areas for potential cost reduction are air-, fuelto be recycled at the end of life of the vehicle. and water-management, sealing (no gaskets or Cost savings can be achieved by reducing frame required) and stack assembly. The planned material costs (in particular: platinum use), cost of the proposed fuel cell according to this increasing power density, reducing system project is 26 $/kW. complexity and improving durability. Platinum content of fuel cells was reduced by more than 3.2 Durability and performances doubling catalyst specific power from the 2008 Both low durability and reliability are caused by baseline of 2.8 kW/g of platinum group metal accumulated degradation of materials and (PGM) to 5.8 kW/g in 2012. Current catalyst catalyst due to water and heat issues [6, 12]. The specific power is approaching the 2017 target of degradation of materials and catalyst are mainly 8.0 kW/g, and it reflects more than 80 percent because of poor water management, fuel and reduction in PGM content since 2005. Several oxidant starvation, corrosion and chemical UK organisations are focused on achieving a reactions of cell components that cause step-change in PEM fuel cell system costs, by dehydration or flooding. The dehydration can developing technologies that reduce platinum damage the membrane and flooding can use, increase power densities and radically facilitate corrosion of the electrodes, the catalyst simplify system designs. layers, the gas diffusion media and the ACAL Energy’s patented FlowCath® [10, 5] membrane. Lifetime of fuel cells can be fuel cell design uses a liquid polymer cathode extended either by controlling the flow solution, which replaces the platinum-based conditions (i.e. humidity, flow rates and solid cathode used in standard PEM fuel cells. temperature) or by changing the materials and This represents a fundamental design the flow design. breakthrough that has the potential to reduce The carbon corrosion in the catalyst layer is a expensive platinum use by at least two thirds, major degradation source in operating a PEM reduce the number of components within the fuel cell. It is well known that Platinum overall system (by avoiding fuel humidification nanoparticles supported on carbon black (Pt/C) and water recovery), and increase durability (as were found the most promising electrocalyast it replaces the solid cathode of typical systems, applied on PEMFC. However, platinum which usually suffers performance degradation nanoparticles in catalyst layers must have that limits product lifetime). The projected fuel simultaneous access to the gases, electrons, and cell cost given for the FlowCath® fuel cell was protons to be effectively utilized. When 36 $/kW. operating under extremely high current Another patent [11, 5] was made by ITM conditions, platinum nanoparticles in the thin Power’s to demonstrate exceptional high power catalyst layers may detach from the support 253

Proceedings of SEEP2015, 11-14 August 2015, Paisley Department of Energy (DoE) industry target for carbon and accelerate the degradation of the fuel cell powered vehicles to last 5,000 hours, electrochemical performance. It was found that equivalent to 150,000 road miles, with an mixing of graphene with conventional Pt/C was expected degradation threshold of approximately able able to transport electrons effectively and to provide better pathways under high current 10%. Unlike a conventional PEM hydrogen fuel density conditions [13]. Graphene displays low cell design, ACAL Energy’s technology does electrical resistance and provides channels with not rely on platinum as the catalyst for the better conductivity for large amount of electrons. reaction between oxygen and hydrogen. The Graphene can potentially provide much higher liquid acts as both a coolant and catalyst for the durability than carbon black with its unique cell’s, ensuring that they last longer by removing graphitized basal plane [14]. most of the known decay mechanisms. Some minimum level of hydration is required to 4 SUMMARY facilitate efficient ionic conductivity in the This paper highlighted the main challenges proton exchange membrane. However, excess associated with using the PEM fuel cell for hydration will be related to reliability issues transport applications. PEMFC provides several such as voltage loss at high current density, advantages over the traditional internal voltage instability at low current density, combustion engine, which are the formal power unreliable start-up under freezing conditions, source in transport industry, including higher and will promote the corrosion of the carbon in efficiency and lower emissions. However, to the catalyst support due to hydrogen starvation meet the full requirements as power sources for [15]. Therefore, the design of membrane and its transport applications, the fuel cell researchers material selection must comprehend the critical have to overcome serious challenges related to balance between too little and too much high cost, low durability and fuelling of fuel cell hydration, especially for automotive applications vehicles (FCVs). In this paper, problems of the where the fuel cell can be subjected to wide fuel cell cost, and the durability and performance variations in load demand and ambient were detailed. Recent developments that deal conditions during its lifetime. Another approach with the proposed challenges were reported. In is to develop the proton exchange membranes to the extended version of this paper, problems of improve the performance and durability of the hydrogen infrastructure and hydrogen storage in standard membranes made of Nafion. However, the vehicle will be covered. in spite of their high proton conductivities and fuel cell performances, sulfonated statistical copolymers are generally characterized by a high REFERENCES degree of dimensional change and poor [1] US DOD, Fuel Cell Test and Evaluation durability derived from a poorly connected nonCenter. hydrated phase and inordinate dimensional History.” http://www.fctec.com/fctec_history.as variation in the hydrated phase. Alternatively, p. Accessed 31 December 2010. The sPP-b-PAES(6.5K)-3.0 membrane had been [2] LI Bing, LI Hui, MA Jianxin, WANG developed by [16]. The developed membrane Haijiang. PEM Fuel Cells: Current Status and showed high proton conductivity, stable Challenges for Electrical Vehicle Applications. J dimensional variation, along with high Automotive Safety and Energy, 2010, Vol. 1 No. performance and durability superior to those of 4. Nafion. [3] Hofman P, Elzen B, Geels F. Sociotechnical In June 2013, ACAL Energy Ltd announced that scenarios as a new tool to explore system it enabled a PEM hydrogen fuel cell to reach innovations: Co-evolution of technology and 10,000 hours runtime on a third party society in the Netherlands energy system [J]. automotive industry durability test without any Innov Manag Policy Pract, 2004, 6: 344-360. significant signs of degradation. 10,000 hours, [4] Elzen B, Geels F W, Hofman P S, et al. the equivalent of 300,000 driven miles, is the Sociotechnical scenarios as a tool for transition point at which hydrogen fuel cell endurance is policy: An example from the traffic and comparable to the best light-weight diesel transport domain [C]// Evidence and Policy, engines under such test conditions. This Edward Elgar, Cheltenham, 2004: 19-47. endurance far exceeds the current 2017 US

254

Proceedings of SEEP2015, 11-14 August 2015, Paisley [5] Polymer Fuel Cells – Cost reduction and [16] Jang Yong Lee, Duk Man Yu, Tae-Ho Kim, market potential. A report by the Carbon Trust Sang Jun Yoon, Young Taik Hong. Multi-block based on independent analysis September 2012. copolymers based on poly(p-phenylene)s with Carbon Trust. excellent durability and fuel cell performance. [6] Barriers of scaling-up fuel cells: Cost, Journal of Membrane Science 492 (2015) 209– durability and reliability. Junye Wang. Energy 219. 80 (2015) 509-521. [7] Marcinkoski J, James BD, Kalinoski JA, Podolski W, Benjamin T, Kopasz J. Manufacturing process assumptions used in fuel cell system cost analyses. J Power Sources 2011;196(12):5282e92. [8] Odeh AO, Osifo P, Noemagus H. Chitosan: a low cost material for the production of membrane for use in PEMFCea review. Energy Sources, Part A: Recovery, Util Environ Eff 2013;35(2):152e63. [9] Sun Y, Delucchi M, Ogden J. The impact of widespread deployment of fuel cell vehicles on platinum demand and price. Int J Hydrogen Energy 2011;36(17):11116e27. [10] ACAL Energy News. Hydrogen fuel cell that’s as durable as a conventional engine. http://www.acalenergy.co.uk/news/release/acalenergy-system-breaks-the-10000-hourendurance-barrier/en [11] Fuel cell membrane performance update. ITM Power. http://www.itm-power.com/newsitem/fuel-cell-membrane-performance-update [12] Bae SJ, Kim S-J, Lee J-H, Song I, Kim N-I, Seo Y, et al. Degradation pattern prediction of a polymer electrolyte membrane fuel cell stack with series reliability structure via durability data of single cells. Appl Energy 2014;131: 48e55. [13] U.S. Department of Energy. Pathways to Commercial Success: Technologies and Products Supported by the Fuel Cell Technologies Program. September 2011. [14] Graphene for energy conversion and storage in fuel cells and supercapacitors. Hyun-Jung Choi, Sun-Min Jung, Jeong-Min Seo, Dong Wook Chang, Liming Dai, Jong-Beom Baek. Nano Energy (2012) 1, 534–551. [15] Jon P. Owejan, Jeffrey J. Gagliardo, Jacqueline M. Sergi and Thomas A. Trabold. Two phase flow consideration in PEMFC design and operation. Proceedings of the Sixth International ASME Conference on Nanochannels, Microchannels and Minichannels ICNMM2008 June 23-25, 2008, Darmstadt, Germany.

255


NUMERICAL SIMULATION OF HYDROGEN RELEASE AND COMBUSTION INTO AN ENCLOSURE 1. 2. 3. 4.

Maxim Bragin1, Thomas Beard2, Weeratunge Malalasekera3, Salah Ibrahim4 Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Loughborough; email: [email protected] Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Loughborough; email: [email protected] Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Loughborough; email: [email protected] Department of Aeronautical and Automotive Engineering, Loughborough University, Loughborough; email: [email protected]

Abstract This paper presents results of numerical simulations of hydrogen release, dispersion and combustion in a partially enclosed space that resembles a garage. The present work is being performed as part of the Engineering Safe and Compact Hydrogen Energy Reserves (ESCHER) project, which is working towards a safe residential hydrogen refuelling station for off-peak topping up of hydrogen vehicles. Intrinsically safer concepts for hydrogen refuelling indoors are based on the use of metal hydride compressor, which allows minimising the amount of hydrogen inventory that is momentarily stored in the refueler cabinet. The numerical simulation is based on the k-ω SST turbulence and the standard EBU combustion models. Results from the numerical simulations are validated against experimental data provided by the UK Health and Safety Laboratory (HSL). It has been found that dispersion is in good agreement with experimental findings whilst modelling of heat transfer through garage walls is essential for correct predictions of the temperature field inside of the enclosure. Keywords: Hydrogen, Safety, Dispersion, Combustion 1 INTRODUCTION The use of hydrogen is viewed as a viable, green option to replace fossil fuels as an energy carrier. However the main implication of using hydrogen is safety due to wide flammability and detonability ranges coupled with relatively low ignition energy [1]. The scope of the ESCHER project is to design a residential installation for hydrogen production, storage and compression for delivery to a vehicle and to determine the safety implications related to its use indoors. As the hydrogen is very hard to contain due to high storage pressures and diffusivity of gas, it is envisaged that releases might occur. Therefore the main focus is on the consequences post-leak, as such dispersion and combustion simulations need to be performed. The dispersion simulations are assessed against the lower flammability limit of 4% of hydrogen content by volume, whilst combustion simulations are compared against the minimum harmful temperature to a human, 400 K [2]. Numerous studies have been performed on the dispersion of hydrogen within a partially

256

enclosed space [3, 4, 5] with only a few investigations taking into account atmospheric conditions [6, 7]. These studies have never progressed past two vents and are mainly validations for large geometries. To the authors knowledge there has only ever been one paper on hydrogen combustion within a partially enclosed geometry [8]. This paper was a blind simulation as there was no relevant experimental data available at the time. This is still a problem although it is envisaged that better data will be available in the future. The current paper presents results of numerical simulations of release, dispersion and combustion in partially enclosed space and validation against available experimental data [9, 10]. 2 PROBLEM DESCRIPTION Experiments were performed by the UK Health and Safety Laboratory as part of the EU JTI Hyindoor project [9, 10]. The experimental set up consisted of a steel container (5x2.5x2.5m) with numerous vents (0.82x0.27m); the vents could be open or closed. Hydrogen was released

Proceedings of SEEP2015, 11-14 August 2015, Paisley vertically upward from a pipe with internal diameter 10mm located in the centre of the floor at a height of 0.5 m as shown in (Figure 1).

Figure 3. Positions of thermocouples within enclosure Table 1. Summary of experimental data

Figure 1. Sketch of the HSL test facility Atmospheric conditions, wind speed and direction, were measured. Two types of experiments were carried out: non-ignited and ignited. For non-ignited experiments the main interest was the dispersion of hydrogen inside of the enclosure. Hydrogen concentrations were recorded using an array of 27 O2 sensors; the locations are given in (Figure 2).

Figure 2. Positions of oxygen sensors within enclosure (for concentration measurements) In ignited experiments the main interest was in recording the temperature levels inside of the enclosure and for that an array of 24 type K thermocouples was used; the locations of which are shown in (Figure 3). Four experiments were chosen – two nonreacting and two with combustion and simulations were carried out for those. Details of the experimental runs, which are studied here, are given in Table 1.

257

Test Reacting or Non- Vents Open Reacting 1 Non-Reacting V1 2 Non-Reacting V4 & V5 3 Reacting V1 4 Reacting V3 & V5

Flow Rate (g/s) 0.2 1.6 0.2 0.88

3 NUMERICAL MODEL The non-reacting and reacting computations were performed using the commercial CFD package, STAR-CCM+ Ver. 9.06.011. Both sets of computations solved transient equations for continuity, momentum and energy. The turbulence model used was the k-ω Shear Stress Transport (SST) model developed by Menter [11]. Diffusion was taken into account using individual diffusivities for the species. Air was modelled as 21% O2 and 79% N2 by volume. The inlet boundary condition for hydrogen was a constant mass flow rate, whilst all outlets were pressure outlets. The outlets were on a bigger geometry, 8 m by 8 m by 8m, which encased the container; this has proven to account for the piezometric pressure at the vents. The total number of control volumes used was 674102, these were arranged such that externally they were very coarse whilst internally they were refined with the smallest control volumes in the major areas of interest. The mesh inside of the enclosure and in vents contained hexahedral elements clustered towards the point of hydrogen release and above it. Externally a tetrahedral mesh was used with rapidly increasing cell volume. When wind was modelled the inlet for this was an external boundary, which was set to a velocity inlet, the simulations was then

Proceedings of SEEP2015, 11-14 August 2015, Paisley performed with no release for 100 seconds to The comparison was carried out for the first 800 make sure that the wind profile was enforced. seconds of the release. Numerical results for all All vents that were open were treated as internal of the sensors along with experimental curves interfaces, such that there is no restriction on the are presented in Figure 4. fluid flow through the vent. Simulation results showed good agreement for For combustion simulations the numerical model the two upper layers and slight under-prediction also incorporated the standard Eddy Break Up in hydrogen concentrations in the bottom one. (EBU) model to calculate the combustion. This 4.2 Non-reacting 1200 NL/min (1.6 g/s) model is solely dependent on the turbulent release. Two vents and opposing wind mixing. A one step reaction mechanism was The second case, which was simulated, had a used. The EBU model coefficients used were 1 stronger release rate of 1200 NL/min (1.6 g/s) and 0.5 for A and B respectively. The ignition and two vents – one lower V4 and one upper V5 source was placed 0.5 m above the release point, (see Figure 1). ignition started 1 second after release for the The experimental data for this case, however, duration of 0.1 seconds. In order to account for was only available for the bottom layer located heat transfer, the temperature of internal garage 1m above the floor of the enclosure. In the walls was set to atmospheric temperature. current case wind was experimentally recorded 4


4.1 Non-reacting 150 NL/min (0.2 g/s) release. One vent and no wind The first case had 150 NL/min (0.2 g/s) release rate and one vent V1 open (see Figure 1). This case was specifically chosen to minimise effects of external wind as experimentally recorded wind direction was blowing on the wall of the garage opposite to the vent. Experimental data from [9] was digitised and three experimental curves for averaged hydrogen concentration at 2.25m, 1.75m and 1m heights were obtained.

to be fluctuating between 2 and 5 m/s as shown in Figure 5. The wind was directed into the upper vent V5. For the purpose of numerical simulation a constant average value of 3.2 m/s was imposed.

Figure 5. Recorded wind speed during the experiment with 1200 NL/min case. The comparison was carried out for the quasisteady release stage between 400 and 800 s. Simulation results and experimental data are shown in Figure 6. It can be seen that the experimental measurements fluctuate due to the varying wind profile and results of the numerical simulation show nearly constant value. Authors believe that this is due to the numerical approximation for constant wind speed. It can be seen that the value predicted by simulation lies within the experimentally predicted range.

Figure 4. Averaged hydrogen concentrations at different heights for 150 NL/min release.

258


Figure 6. Hydrogen concentrations at 1m height inside of the enclosure for 1200 NL/min release.

Figure 7. Averaged temperature profiles at different heights for 150 NL/min release.

4.3 Combustion of hydrogen for the release rate of 150 NL/min (0.2 g/s) and one vent The experimental data for hydrogen fires indoors is currently very limited. For the purpose of validation of numerical code, some preliminary unpublished data for hydrogen combustion in an enclosure was obtained [10]. The third case modelled has the same flow rate as the case described in section 4.1 and the same vent configuration – V1 open. In the initial runs walls of the enclosure were modelled as adiabatic boundary. This assumption was proven to be incorrect as the timescale of the release is of the order of half an hour and the heat “trapped inside” resulted in a much higher temperature in the enclosure exceeding the experimental measurements by 250-300 degrees in the top layer. As modelling of heat transfer is computationally expensive, the simplified way to account for this phenomenon was taken in fixing the temperature of enclosure walls at ambient value of 282 K. The results for this case are presented in Figure 7. Thermocouple readings in numerical simulations were averaged across the level of a given height. The results generally showed good agreement with experimental data with slight over-prediction of 20-40 degrees in all three layers. This is most likely associated with a simplified way of modelling the heat transfer to walls.

259

4.4 Combustion of hydrogen for the release rate of 648 NL/min (0.88 g/s) and two vents The fourth case involved higher release rate of 648 NL/min and two vents – one lower V3 and one upper V5. The experimental data is available only as an average temperature in the layer for the steady state regime (200 - 1200 s), excluding unsteady stages of ignition and layer formation during the first 200 seconds and temperature dissipation once the hydrogen release (and jet fire) ceases. The comparison of numerical simulations against the experimental measurements of temperature are presented in Figure 8.

Figure 7. Averaged temperature profiles at different heights for 648 NL/min release.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [6] K. Matsuura, H. Kanayama, H. Tsukikawa, It can be seen that experimental temperature M. Inoue, Numerical simulation of leaking profiles are fluctuating, which is most likely due hydrogen dispersion behavior in a partially open to the effect of wind. However no information is space, International Journal of Hydrogen available for this case regarding the wind profile. Energy, Vol 33, pp 240-247, 2008. Nevertheless, the agreement on the average [7] S. G. Giannissi, J. R. Hoyes, B. temperature in the top layer is good with Chernyavskiy, P. Hooker, J. Hall, A. G. overprediction of only 20-50 degrees in Venetsanos, V. Molkov, CFD benchmark on numerical simulations. The middle layer results hydrogen release and dispersion in a ventilated lie in the median of the experimental readings, enclosure: Passive ventilation and the role of an where results for the bottom layer are slightly external wind, International Journal of under-predicted; however, the difference is only Hydrogen Energy, Vol 40, pp 6465-6477, 2015. 5 to 9 degree. [8] V. Molkov, V. Shentsov, S. Brennan, D. Makarov, Hydrogen non-premixed combustion 5 CONCLUSION in enclosure with one vent and sustained release: Overall, the numerical simulations showed good Numerical experiments, International Journal of agreements against available experimental data Hydrogen Energy, Vol 39, pp 10788-10801, and future work will involve simulation of the 2014. project specific scenarios. [9] P. Hooker, D. Willoughby, J. Hall, J. Hoyes, It has been found that simulations of dispersion Accumulation of hydrogen release into a vented are in good agreement with experimental enclosure – Experimental results, Proc. Int. Conf findings whilst modelling of heat transfer on Hydrogen Safety, Brussels, 2013. through garage walls is essential for correct [10] P. Hooker, FCH JU Hyindoor Project – predictions of the temperature field inside of the HSL input, Presentation at the ESCHER project enclosure in combustion simulations. meeting, October 2014. [11] F. R. Menter, Two-equation eddy-viscosity ACKNOWLEDGEMENTS turbulence modeling for engineering This work has been supported by the UK applications, AIAA Journal, Vol 32, pp 1598Engineering and Physical Sciences Research 1605, 1994. Council through the ESCHER project under grant No. EP/K021117/1. REFERENCES [1] V. Molkov, Fundamentals of Hydrogen Safety Engineering I, www.bookboon.com, 2012. [2] J. L. Bryan, Damageability of buildings, contents and personnel from exposure to fire, Fire Safety Journal, Vol 11, pp. 15-31, 1986. [3] C. R. Bauwens, S. B. Dorofeev, CFD modeling and consequence analysis of an accidental hydrogen release in a large scale facility, International Journal of Hydrogen Energy, Vol 39, pp 20447-20454, 2014. [4] M. R. Swain, E. S. Grilliot, M. N. Swain, Risks incurred by hydrogen escaping from containers and conduits, Proc. 1998 U.S. DOE Hydrogen Program Review, Alexandria, Virginia, 1998. [5] P. Middha, O. R. Hansen, I. E. Storvik, Validation of CFD-model for hydrogen dispersion, Journal of Loss Prevention in the Process Industries, Vol 22, pp 1034-1038, 2009.

260


CATALYTIC CONVERSION OF STEARIC ACID TO FUEL OIL IN A HYDROGEN DONOR

Zhentao Huang1, Jiangfei Cao2, Zhixia Li1,*, Hao Gong1, Lingyun Huang1, Song Shi1, Yue Li1 1. School of Chemistry and Chemical Engineering, Guangxi University, Nanning, Guangxi 530004, China 2. State Key Laboratory of Non-food Biorefinery Enzymolysis, National Engineering Research Center for Non-food Biorefinery, Guangxi Academy of Sciences, Nanning 530007, China * E-mail: [email protected] Abstract Production of diesel or gasoline from renewable raw materials e.g. vegetable oil or waste food oil has attracted considerable attention. Stearic acid as a typical component in vegetable oil was dispersed in tetralin (a hydrogen donor) and hydrotreated in a batch reactor with 1.6 wt% sulfided NiMo/-Al2O3-zeolite catalyst for obtaining the diesel-range component. A series of sulfided NiMo/-Al2O3--zeolite (Si/Al=25) catalysts were prepared with maceration method. The total NiMo mass ratio was changed from 15 to 25 wt%, and -Al2O3/-zeolite mass ratio (A/Z ratio) changed from 8:2 to 6:4. The effects of properties of catalysts (acid amount, acid strength), reaction conditions and solvent on the deoxygenation of stearic acid were investigated. A nearly complete conversion of stearic acid was achieved at 350 C, 2.0 MPa H2 for 2 h. The total yield of C15-C18 alkanes and alkenes was more than 95%, and the product also contained 30-40 wt% isomerized compounds. Increase of -Al2O3/-zeolite mass ratio in supports shows little effect on conversion, but decreases the isomerized ratio in product oil, which is likely due to the change of acid property of catalysts. An increase of NiMo mass ratio trends to change the production distribution, especially increasing the formation of alkenes compounds. Keywords: Stearic acid, catalytic hydrogenation, fuel oil, hydrogen donor 1 INTRODUCTION With the depletion of traditional energy and increase of environment pollution, a lot of attention has been attracted to catalytic conversion of renewable raw materials e.g. vegetable oil or waste food oil to green fuel oil [1-3]. The main techniques used are catalytic hydrogenation and catalytic cracking. The most common used catalysts for hydrodeoxygenation are sulfided supported metal (e.g.Co, Ni and Mo) catalysts with γ-Al2O3 or zeolite as support [1,2], and noble metal (e.g. Pt and Pd) catalysts with active carbon as support [3]. Especially, the sulfided Ni, Mo and NiMo-based catalysts have caused wide concern for its low cost and high activity on catalytic conversion [1]. However, so far, a few of work has been done on investigating the effect of acidic properties and metal content in catalysts on the catalytic activity. Tetralin as a hydrogen donor, widely used as solvent in liquefaction process of biomass, can disperse and stabilize intermediate product from liquefaction process by supplying active

261

hydrogen, and thus, improving liquefaction efficiency [6]. However, little attention is paid to tetralin as solvent on catalytic hydrogenation of fatty acid or plant oil to fuel. In this study, a serious of catalysts with different acidic properties and metal active sites are prepared by adjusting γ-Al2O3/-zeolite mass ratio (A/Z ratio) and NiO and MoO3 contents in catalysts. The effect of acidic properties and active sites of catalysts on conversion of stearic acid (SA) and product distribution are investigated. The effect of tetralin as solvent is discussed. 2

EXPERIMENTAL

2.1 Synthesis of NiMo/-Al2O3--zeolite catalysts γ-Al2O3 and H-type -zeolite (Si/Al=25) were purchased from Nankai University Catalyst Co., Ltd. (Tianjin, China). Other reagents were analytical grade and used as received. NiMo/-Al2O3--zeolite was prepared by the conventional impregnation method. The Al2O3--zeolite composite supports were firstly

Proceedings of SEEP2015, 11-14 August 2015, Paisley 2.4 Product analysis prepared by mechanically mixing γ-Al2O3 and After methyl esterifying the liquid products zeolite with different A/Z ratio in deionized (GB/T 17376-2008), the components in products water. The mixed suspension was evaporated at were analyzed by GC-MS (GC 7820A, MS 80 C to remove water, and dried in an oven at 5977E, Agilent) and GC-FID (9792II, Fuli, 105 C for 1 h. The obtained solid sample was China). The products were separated on a then calcined at 450C for 3 h. Different A/Z column of HP-5MS (50m×0.2mm×0.3μm). The ratio of 8:2, 7:3 and 6:4 gave the composite column oven temperature was set at 60 C for 5 support samples denoted as 8A-2Z, 7A-3Z and min and then programmed with a heating rate of 6A-4Z, respectively. Afterward, a certain 10 C/min until it reached 300 C, which was amount of Ni(NO3)2, (NH4)2MoO4 and citric held for 15 min. The injector and interface acid (CA) were dissolved in deionized water to temperature were maintained at 290 C and 300 form metallic salt solution. The obtained C, respectively. composite supports were dispersed into the Conversion and isomerization are considered as metallic salt solution. The mixed suspensions important index of catalytic experiments. were rotary-evaporated to remove water, and the Conversion is defined as the percent of loss of obtained solid samples were dried, tableted, and SA during reaction, which transforms to other calcined at 380 C for 5 h. The total NiO and compounds e.g. alkanes, alkenes, acohol and MoO3 content of 15%, 20% and 25% with aldehyde et. Isomerization is defined as the ratio Ni/NiMo molar ratio of 0.2 were loaded in of the branched alkanes in all alkanes composite supports, and thus produced samples 20%NiMo-CA/8A-2Z, 20%NiMo-CA/7A-3Z, 3 RESULTS AND DISCUSSION 20%NiMo-CA/6A-4Z and 15%NiMo-CA/7A3.1 Characterization of catalysts 3Z, and 25% NiMo-CA/7A-3Z, respectively. The XRD patterns of catalysts with different 2.2 Characterization of catalysts A/Z ratio and different NiNo mass ratio are The obtained catalysts were characterized by Xshown in Figure 1a and Figure 1b, respectively. Ray Diffraction (XRD, UItima IV, Rigaku, When A/Z ratio decreases from 8/2 to 6/4 Japan) for metal sites, and temperature(Figure 1a), the peaks due to -Al2O3 (2θ at programmed desorption of ammonia (NH3-TPD) 39.5° and 45.9°) become weak, and the peaks connected with a Residual Gas Analyzer due to -zeolite (2θ at 7.8° and 22.5°) become (RGA200, Agilent) as detector for acid sites. intense. As total NiMo mass fraction increases The desorbed NH3 was absorbed in 100 ml HCl from 15% to 25% (Figure 1b), the peaks of (0.01M), and then NaOH (0.01M) was used to MoO3 (2θ at 23.4°, 25.7° and 27.3°) become titrate the absorption liquid. The acid amount intense and shift to high angles, but the peaks of was calculated by consumption amounts of HCl NiO (2θ at 37.3°) become weak and shift to high and NaOH. The distribution of weak acid, angles. This indicates that some Ni atoms likely medium acid and strong acid was calculated entered the lattice of MoO3 through a from the peak area from Gaussian Fitting of the homogeneous substitution of Mo and some Mo NH3-TPD profiles acquired from RGA200. atoms likely entered the lattice of NiO through a 2.3 Catalytic hydrotreating stearic acid homogeneous substitution of Ni with Mo Catalytic experiments were carried out in a 20because the radius of Mo atom (0.145 nm) is ml batch-type reactor, which was fixed on a larger than that of Ni (0.135nm). Incorporation shaker and shaken at a constant speed. The of Mo in the lattice of NiO was reported to catalysts were sulfided in situ prior to the promote the dispersion of NiO and reduce the experiments by means of dimethyl disulfide crystallinity of NiO [4]. (DMDS) at 5 MPa H2 pressure and 320 C for 3 The NH3-TPD profiles of catalysts with different h. After sulfuration reaction, the reactor was A/Z ratio and different NiMo mass fraction are shown in Figures 2 and 3. Weak acid (desorbed cooled to about 100 C, and the remanent DMDS was released. After dispersed in tetralin temperature at about 200C), medium acid in a tetralin/SA mass ratio of 4:1, SA was added (about 260C) and strong acid (about 400C) are into reactor and reacted under various observed in each figure. The main desorbed peak temperature (300–350 C), H2 pressure (2-5 occurs at about 260C. And the distribution of MPa) and time (1-3 h). weak, medium and strong acid is depicted in 262

Torr signal(a.u.)

(22.5)

Proceedings of SEEP2015, 11-14 August 2015, Paisley 60 Table 1 and Table 2. It is found that acid amount (a) 8:2 and intensity of medium acid increases with 50 increasing the ratio of -zeolite. Medium acid is 40 main acid site for all catalysts. —γ-Al2O3 —β-zeolite







0

6:4

200

7:3

2θ(°)

40

50

60

—α-MoO3 —NiO

(12.8)



 



30 20 10

2θ(°)

50

T(℃ )

400

500

60

600

(c) 6:4

50

15%

40

300

60

40

Figure 1. XRD patterns of NiMo/γ-Al2O3-zeolite catalysts, (a) A/Z ratio changed ranged from 6:4 to 8:2; (b) total amount of NiO and MoO3 increased from 15% to 25%.

Torr signal(a.u.)

30

600

0

20%

(37.3) 20

500

(b) 7:3

200

10

400

40

25%



T(℃ )

50

Torr signal(a.u.)

30

(23.4) (25.7) (27.3)

20

300

60

8:2

10

20 10

(45.9)

(39.48)

(7.8)



30

30 20 10 0 200

Table 1. Effect of A/Z ratio on the acid properties of catalysts. Catalyst Total Percentage of acid Acid sites(% of total acid) A/Z mmol/g Peak I Peak II Peak III ratio 200C 260C 400C 8A-2Z 0.3515 5 60 35 7A-3Z 0.6351 15 61 24 6A-4Z 0.7086 6 67 27 Table 2. Effects of NiMo mass ratio on acid properties of catalysts. Catalyst Total Percentage of acid Acid sites(% of total acid) (NiO+ mmol/g Peak I Peak II Peak III MoO3) 200C 260C 400C 15% 0.4782 26 48 26 20% 0.6351 15 61 24 25% 0.5214 15 60 25

263

300

T(℃ )

400

500

600

Figure 2. NH3-TPD profiles of NiMo/A12O3zeolite catalysts with different A/Z ratio, (a) 20%NiMo-CA/8A-2Z; (b) 20%NiMo-CA/7A3Z; (c) 20%NiMo-CA/6A-4Z. 3.2 Catalytic Hydrotreating of stearic acid Catalytic reaction of SA was carried out at 350 C, 5MPa H2 pressure for 2 h. Catalyst used in this experiment was sulfided 20%NiMo-CA/7A3Z. Table 3 shows the conversion and product distribution for catalytic hydrotreating of SA. It can be seen that conversion decreases from 88.6% to 81.7%, alkanes yield decrease and alkenes increase with increasing -zeolite content from 20% to 40%. These changes may be related to the changes on acid density and distribution of acid intensity of composite zeolite catalysts (Figure 2 and Table 1). It has been suggested that increase of medium acid site and strong acid sites in catalysts accelerated isomerization [7] and cracking reaction [8].

Torr signal(a.u.)

Torr signal(a.u.)

Torr signal(a.u.)

Proceedings of SEEP2015, 11-14 August 2015, Paisley It can be observed from Table 3 that conversion 3.3 Effect of reaction condition on of SA increases from 76.7% to 87.3% and then catalytic conversion of stearic acid The catalyst used in these experiments was decreases with increasing the total amount of 20%NiMo-CA/6A-4Z with Ni/Mo molar ratio of NiMo mass ratio from 15% to 25%, and the conversion goes through a maximum (87.3%) 60 over the catalyst 20%NiMo-CA/7A-3Z. This (a) wt=15% trend is similar to that acid amount change of 50 catalysts with increasing the NiMo mass ratio as 40 shown in Table 2. This suggests that addition of 30 metallic oxides changed the acidic property of catalysts significantly, and thus affected the 20 activity of catalysts. However, excess addition of 10 Ni and Mo may accelerate the growth of NiO and/or MoO3 crystals, and contrarily restrain 0 200 300 400 500 600 their good dispersity on surface, consequently, T(℃ ) decrease the catalytic activity of catalysts. 60 (b) wt=20% Above analysis shows that at present test range, 50 both A/Z ratio and total amount of NiMo in 40 catalysts affected the acidic property of catalysts significantly. Compared with the effect caused 30 by A/Z ratio, metal site (NiMo amount) in 20 catalysts shows more obvious influence on the 10 catalytic activity. It is considered that high dispersity of metal oxides and modest acidic 0 200 300 400 500 600 properties of composite supports in catalysts are T(℃ ) responsible for the high catalytic activity of 60 (c) wt=25% catalysts. 50 In addition, isomerized fraction in products keeps a positive correlation with the acid amount 40 of catalysts as seen in Tables 1-3, indicating that 30 increasing acid density, especially medium acid 20 site and strong acid sites, is helpful to enhance isomerization reaction. 10 According to Table 3, when the NiMo mass 0 fraction increases from 15% to 25%, the yield of 200 300 400 500 600 T(℃ ) C15-C18 alkanes decreases from 88.6% to 55.2%, Figure 3. NH3-TPD profiles of NiMo/A12O3meanwhile, alkenes yield increases from 9.8 to zeolite catalysts with different NiMo mass 43.5%, indicating a possible reaction pathway: fraction, (a) 15%NiMo-CA/7A-3Z; (b) the isomerized alkanes in this reaction are 20%NiMo-CA/7A-3Z; (c) 25%NiMo-CA/7Amainly achieved via hydroisomerization, in 3Z. which alkanes are obtained through hydrodeoxygenation of SA, and then alkanes are 3:7. Effects of reaction conditions (temperature, transformed to alkenes by metal sites [5]. H2 pressure, time, solvent) on conversion and However, the formed alkenes cannot transform production distribution were investigated. to isomerized alkanes without enough acid sites. Naphthalene as a dehydrogenation product of Furthermore, the presence of an intermediate of tetralin was also analyzed by GC-MS. Table 4 hydrodeoxygenation of SA (1-Octadecanol) in shows the conversion and product selectivity reaction product, indicates that hydrogenationobtained under various reaction conditions. It is deoxygenation reactions are the main route for found that conversion and production producing C18 alkanes without carbon loss. distribution are significantly affected by reaction temperature, and a temperature higher than 325C is needed for effective catalytic

264

Proceedings of SEEP2015, 11-14 August 2015, Paisley Table 3. Effect of ratio of composite support and metal mass fraction on products γ-Al2O3/ β-zeolite The total NiO and MoO3 mass fraction Conversion and Yields (%) 8:2 7:3 6:4 15% 20% 25% Conversion 88.6 87.3 81.7 76.7 87.3 77.5 C15-C18 alkanes 85.9 75.4 73.5 88.6 75.4 55.2 C16-C18 alkenes 12.9 20.8 24.9 9.8 20.8 43.5 Alcohols 1.24 3.8 1.7 1.2 3.8 1.1 Isomerized fraction 13.72 31.0 10.7 13.7 11.8  Table 4. Effect of reaction conditions on conversion and yields of products Conversion and Time1 /h H2 pressure 3/MPa Temperature 2/C Yields (%) 1 2 3 300 325 350 2 3 5 Conversion 87.8 95.2 100 22.5 80.8 95.2 96.3 100 95.2 C15-C18 alkanes 83.7 83.6 84.0 61.8 77.3 83.6 78.2 84.4 83.6 C16-C18 alkenes 16.3 16.5 16.0 38.2 22.7 16.5 21.8 15.6 16.5 Naphthalene 1.4 3.3 5.8 0.3 1.0 3.32 5.9 4.8 3.3 Isomerized fraction 24.7 33 41.7 18.2 33 41.8 32.6 33  1

, 350C and 5 MPa; 2, 5 MPa for 2 h; 3, 350 C for 2 h.

hydrogenation reaction by using present selfmade 20%NiMo-CA/6A-4Z catalysts. Temperature influenced product distribution likely via affecting metal site density and its dispersity on catalysts, as well as the hydrogenreleasing activity of tetralin. H2 pressure 2MPa resulted in a high conversion of 96.3%, and the obtained product contained less alkanes, more alkenes and naphthalene as compared with those obtained at higher H2 pressure. These results suggest that tetralin as a hydrogen-donor solvent, effectively exert its hydrogen-supplying action, and accelerate catalytic conversion reaction of SA at a lower H2 pressure condition. In addition, when reaction time increases from 1 h to 3 h, conversion increases from 87.8% to 100%, the yields of alkanes and alkenes remain at about 84% and 16%, respectively, meanwhile naphthalene and isomerized fraction in product increase. These results means that self-made catalysts maintained high catalytic activity during 3 h of reaction time, and the generated alkanes possibly undergo further reaction to transform to isomerizaed products. Since liquid products contain a trace of polycyclic aromatic hydrocarbon, naphthalene and their derivatives, demonstrating that tetralin participated in catalytic hydrotreating reaction of SA. Some work is being carried out to clarify the role of tetralin.

β-zeolite catalysts. The acidic properties and metallic activity of catalysts were adjusted by changing the total NiMo mass ratio and γAl2O3/β-zeolite mass ratio in catalysts. The effects of properties of catalysts (acid site density, acid strength, metallic site), reaction conditions and solvent on the deoxygenation of SA were investigated. A nearly complete conversion of SA was achieved at 350C, 2.0 MPa H2 for 2 h by using self-made 20%NiMoCA/6A-4Z catalysts. The total yield of C15-C18 alkanes and alkenes is more than 95%, and the products also contain 3044 wt% isomerized compounds. Increase of acid density by decreasing γ-Al2O3/β-zeolite mass ratio in supports shows little effect on conversion, but increases the isomerized fraction in product oil. NiMo mass fraction has important effect on acid density of catalysts and product distribution, and 20% NiMo mass fraction in catalysts resulted in a higher conversion and more isomerized products, which is likely due to the enough acid sites and high dispersion of metal site. Reaction time and temperature have positive effect on improving conversion, alkanes yield and isomerized products. Tetralin as solvent shows high efficiency on catalytic conversion of SA at a low H2 pressure. Additional research is required to study the reaction pathway.

4 CONCLUSION Catalytic conversion of SA into fuel oil was studied using self-made sulfided NiMo/γ-Al2O3-

ACKNOWLEDGEMENTS The authors are grateful for supports from the National Natural Science Foundation of China

265

Proceedings of SEEP2015, 11-14 August 2015, Paisley (No. 21266002) and the Scientific Research Foundation of Guangxi University (Nos. [6] S. N. Ali; M. F. Yusop; K. Ismail; Z. A. XGZ120081 and XTZ140787). Ghani; M. F. Abdullah; M. A. M. Ishak; A. R. Mohamed, Tetralin-glycerol as Solvent in Direct Liquefaction of Mukah Balingian Coal, Energy REFERENCES [1] D. Kubicka and L. Kaluza, Deoxygenation of Procedia, 52, 618-625, 2014. vegetable oils over sulfided Ni, Mo and NiMo [7] S. C. C. Wiedemann.; A. Munoz-Murillo; R. catalysts, Applied Catalysis, A: General, 372(2), Oord; T. van Bergen-Brenkman; B. Wels; P. C. 199-208, 2010. A. Bruijnincx; B. M. Weckhuysen, Skeletal [2] E. Santillan-Jimenez; T. Morgan; J. Lacny; S. isomerisation of oleic acid over ferrierite: Mohapatra; M. Crocker, Catalytic Influence of acid site number, accessibility and deoxygenation of triglycerides and fatty acids to strength on activity and selectivity, Journal of hydrocarbons over carbon-supported nickel, Catalysis, 329, 195-205, 2015. Fuel, 103, 1010-1017, 2013. [8] A. Janda; B. Vlaisavljevich; L. C. Lin; S. [3] G. W. Huber and A. Corma, Synergies Mallikarjun Sharada; B. Smit; M. Head-Gordon; between bio-and oil refineries for the production A. T. Bell, Adsorption Thermodynamics and of fuels from biomass, Angewandte Chemie, Intrinsic Activation Parameters for International Edition, 46(38), 7184-7201, 2007. Monomolecular Cracking of n-Alkanes on [4] J. Chen; Y. Yang; H. Shi; M. Li; Y. Chu; Z. Bronsted Acid Sites in Zeolites, Journal of Pan; X. Yu, Regulating product distribution in Physical Chemistry C, 119(19), 10427-10438, deoxygenation of methyl laurate on silica2015. supported Ni-Mo phosphides: Effect of Ni/Mo ratio, Fuel, 129, 1-10, 2014. [5] C. Z. Ai; R. A. Sun; C. S. Wang, DFT studies on hydrogen overfall mechanism for catalyzed hydroisomerization of pentane, Chinese Journal of Structural Chemistry, 26(2), 239-247, 2007.

266


CATALYTIC DESTRUCTION OF VOLATILE ORGANIC COMPOUND EMISSIONS USING FLOW-THROUGH CATALYTIC MEMBRANE REACTOR M. N. Kajama1, N. C. Nwogu2 and E. Gobina3 1. M. N. Kajama, Centre for Process Integration and Membrane Technology, School of Engineering, Robert Gordon University, Aberdeen, U K. email: [email protected] 2. N. C. Nwogu, Centre for Process Integration and Membrane Technology, School of Engineering, Robert Gordon University, Aberdeen, U K. email: [email protected] 3. E. Gobina, Professor of Chemical Engineering and the Chair/Director of Centre for Process Integration and Membrane Technology, School of Engineering, Robert Gordon University, Aberdeen, U K. email: [email protected] Abstract A catalytic membrane has been prepared by impregnating Pt on the pore walls of γ-Al2O3 through the evaporative-crystallization deposition method for volatile organic compounds (VOCs) destruction. Scanning electron microscopy and energy dispersive X-ray analysis (SEM-EDAX) observation, BET measurement, permeability assessment and the catalytic oxidation of selected VOCs (propane and propylene) have been obtained. The VOCs conversion obtained by varying the reaction temperature showed that the flow-through membrane reactor outperforms conventional technologies in catalytic abatement of VOCs. Keywords: Volatile organic compounds (VOCs), propane conversion, propylene conversion, flow-through membrane reactor, platinum supported alumina catalysts. 1

INTRODUCTION

The emissions of VOCs into the atmosphere are caused from a variety of sources but mainly mobile and industrial processes. VOC constitute serious hazards to the environment [1]. VOCs lead to more hazardous compounds when they react with other compounds thereby contributing to poor air quality [1]. Eliminating these organic compounds is of paramount importance in order to safeguard the environment. Over the past decade, VOC emissions have witnessed ever stricter regulations globally. For example, the air quality standards developed by the United States Environmental Protection Agency (USEPA) stipulates that a maximum 3hour concentration of hydrocarbon content of 0.24 parts per million (ppm) should not to be exceeded for a period of more than a year [2, 3]. Also, the Gothenburg protocol states that by 2020 the European Union (EU) should reduce VOCs emission levels by 50% compared to the year 2000. In comparison with thermal oxidation, catalytic destruction is the favoured alternative for VOC emission destruction. Catalytic destruction requires lower energy, thus resulting in lower operating and/or capital cost [2, 4-6]. Also, catalytic destruction does not emit pollutants 267

such as carbon monoxide and nitrogen oxides [7]. Noble metal and transition metal oxides are well-known in the oxidation reaction of VOCs because of their high activity [8]. The most efficient metal for VOC combustion is the platinum supported on gamma-alumina (Pt/γAl2O3) which can operate at a lower temperature and achieve total VOC conversion [1, 9-12]. For example, in the case of flow-through contactor configuration, the reactants gas mixtures are forced to go through the catalytic pores after been heated to the desired temperature. The catalyst (Pt) lowers the reaction activation energy and the porous membrane produce a wide dispersion of the catalytically active metal. Oxidation of the reactants will occur on the catalyst surface whereby heat will be released as the VOCs are converted to yield the product from the exit side of the reactor as shown in Figure 1. Nonetheless, numerous authors have revealed that platinum metal supported on alumina are more superior for the catalytic combustion of VOCs [1, 7, 12-16]. The performance of the catalysts robustly relies on the method of preparation. The degree of metal dispersion on the surface of the support and the metallic nanoparticles size are important factors for determining the efficiency of the process. The

Proceedings of SEEP2015, 11-14 August 2015, Paisley content of the noble metal should be low due to 2.2 Platinum Activation its high cost. Consequently, particle size and Metallic platinum is obtained after thermal dispersion are among the key parameters treatment of the sample under flowing hydrogen ensuing in preparing such catalysts [1]. at 400 0C for at least 10 min followed by nitrogen flow for 10 min at 400 0C.

Figure 1. Flow-through catalytic membrane reactor [11]. The present work has been conducted by depositing a low Pt on Al2O3 support using flowthrough membrane reactor for the oxidation of propane and propylene as a representative of VOC in the presence of oxygen to yield carbon dioxide and water. 2

EXPERIMENTAL

2.1 Materials and Membrane Preparation Commercially available porous supports of tubular configuration supplied by Ceramiques Techniques et Industrielles (CTI SA) France, consisted of 77% alumina + 23% TiO2 have been used in this study. The support had an internal and outer diameter of 7 and 10 mm respectively with a permeable length of 348 mm and a porosity of 45%. A solution of hexachloroplatinic acid (H2PtCl6) has been used as platinum precursor. The tubular support is first dried at 65 0C. After weighing, it is dipped for 2 hours in deionised water before Pt introduction. The deposition method used was based on evaporation-crystallisation method. This method was based on the so-called “reservoir” method proposed by Uzio et al. [17] and Iojoiu et al. [18]. The tube was first dipped for 2 hours in pure water (in our case we used deionised water) afterwards the tube was dipped for 10 hours in a 10 g Pt/l of H2PtCl6 precursor solution. The sample was then dried at room temperature to favour evaporation from the inner side and deposition in the top layer.

268

2.3 Membrane Characterization Scanning electron microscopy (Zeiss EVO LS10) has been used to determine the position of platinum particles inside the porous structure of the multi-layered ceramic material. Samples for crosswise Pt EDX analyses were prepared by breaking the tube after depositing a film on the section. SEM and EDXA results indicated the presence of Pt. The surface area was measured using Brunauer-Emmett-Teller (BET) method from nitrogen adsorption–desorption at 77 K using automated gas sorption analyzer (Quantachrome instrument version 3.0) (not shown). All samples were first degassed at 400 0 C for 2 hours prior to the nitrogen adsorption analysis. Gas permeation measurements of propylene were performed before and after Pt deposition using a conventional setup [17, 18]. The gas permeate flow was measured [11, 17] by a digital flowmeter (Cole-Parmer). The catalytic tests were carried out on a shell and tube configured reactor in a flow-through contactor configuration. The VOC reactants mixture was composed of propylene and oxygen as well as propane and oxygen. The products were analysed by CO2 analyser (CT2100Emissions Laser Sensor). 3


3.1 Pt Membrane Characterization SEM micrograph of the inner and outside surface of the membrane after Pt impregnation is shown in Figures 2 and 3. Pt particles are clearly visible. 3.2 Gas Permeation Permeation experiments were carried out at 25 0 C using propylene as the permeating gas in order to quantify the viscous and Knudsen flow contributions. Figure 4 depicts the permeate flux of the untreated and Pt/Al2O3 membranes. Eqn. (1) was used to relate the permeation flux and average pressure [16]. (1) where β and k equals;

Proceedings of SEEP2015, 11-14 August 2015, Paisley From Figure 4, it can be seen that the slope of , and (2) the line corresponding to the untreated membrane is high indicating a large viscous Where, F is the permeation flux per unit of time flow contribution. On the other hand, after 3.52 and area, ε is the porosity of the membrane, r is wt% Pt impregnation, a lower slope is obtained the mean pore radius (m), Pav = (p1+p2)/2 is the which indicates a higher Knudsen flow average pressure (Pa), μ is the viscosity (Pa-s) contribution. The obtained results almost and L is the thickness of the membrane (m), τ is corroborate with the literature [16]. the tortuosity, M is the molecular weight of the diffusing gas (g/mol), R gas constant (8.314 J.K1 .mol-1) and T the permeation temperature (K). β and K can be regarded as viscous and Knudsen contributions to the permeation flux.

Figure 2. SEM image of the inner surface of Pt membrane.

Figure 2. SEM image of the outer surface of Pt membrane.

269

Figure 4. Propylene permeation flux Vs average pressure across the membrane for an untreated membrane (γ-Al2O3) and for platinum membrane (Pt/γ-Al2O3) membrane at 25 0C. 3.3 Reaction Results Figure 5 depicts the relationship between VOC conversion and the reaction temperature. It can be seen that 72% and 71% propane and propylene conversion have been achieved at a temperature of 232 0C and 255 0C respectively on 3.52-wt% Pt/Al2O3 catalysts. Gluhoi et al. [15] obtained 72% propane conversion at nearly 275 0C on 1-wt% Pt/Al2O3. Furthermore, Saracco and Specchia [19] obtained 72% propane conversion at nearly 365 0C on 5-wt% Pt/γ-Al2O3. They also achieved 71% propylene conversion at nearly 265 0C on the same 5-wt% Pt/γ-Al2O3. Therefore, the temperature at which the catalytic combustion takes place for these VOCs in this study are lower than those obtained in literature [15, 19] for the same organic compounds on Pt/γ-Al2O3 catalysts.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [3] M. Tamaddoni, R. Sotudeh-Gharebagh, S. Nario, M. Hajihosseinzadeh, and N. Mostoufi, Experimental study of the VOC emitted from crude oil tankers, Process Safety and Environmental Protection, vol. 92, pp. 929-937, 2014. [4] A. O. Rusu, and E. Dumitriu, Destruction of volatile organic compounds by catalytic oxidation, Environmental Engineering and Management Journal, vol. 2(4), pp. 273-302, 2003. [5] E. N. Ruddy, and L. A. Carroll, Select the best VOC control strategy, Chemical Engineering Progress, Vol. 89(7), pp. 28-35, 1993. [6] http://www.meca.org/galleries/files/hapwp. pdf Catalytic oxidation for the control of hazardous organic air pollutants, 1995, Figure 5. VOC conversion against reaction [Accessed on 29th October 2014]. temperature on Pt/γ-Al2O3. [7] S. F. Tahir, and C. A. Koh, “Catalytic 4 CONCLUSION destruction of volatile organic compound Catalytic membrane was prepared by emissions by platinum based catalyst,” impregnating a Pt on the pore walls of γ-Al2O3 Chemosphere, vol. 38(9), pp. 2109-2116, 1999. support through reservoir technique. The Pt/γ[8] V. P. Santos, S. A. C. Carabineiro, P. B. Al2O3 membrane was tested for its performance Tavares, M. F.R. Pereira, J. J. M. Órfão, and J. towards the catalytic oxidation of selected VOC L. Figueiredo, Oxidation of CO, ethanol and compounds. The result of the selected VOC toluene over TiO2 supported noble metal conversion confirms that the flow-through catalysts, Applied Catalysis B: Environmental, membrane reactor operation is a promising vol. 99, pp. 198-205, 2010. alternative using this simple but effective [9] L. F. Liotta, M. Ousmane, G. Di. Carlo, G. “reservoir technique”. The reservoir method Pantaleo, G. Deganello, A. Boreave, and A. combined with proper selection of the support Giroir-Fendler, Catalytic removal of toluene has resulted in high conversion and reduced over C03O4-CeO2 mixed oxide catalysts: precious metal content. This will reduce cost and comparison with Pt/Al2O3, Catal Lett., vol. 127, enhance commercialization. pp. 270-276, 2009. [10] P. Marécot, A. Fakche, B. Kellali, G. ACKNOWLEDGEMENTS Mabilon, M. Prigent, and J. Barbier, Propane The authors gratefully acknowledge Petroleum and propene oxidation over platinum and Technology Development Fund (PTDF) Nigeria palladium on alumina: Effects of chloride and for funding this research, and School of water, Applied Catalysis B Environmental, vol. Pharmacy & Life Sciences RGU Aberdeen for 3, pp. 283-294, 1994. the SEM and EDXA results. [11] S. Benard, A. Giroir-Fendler, P. Vernoux, N. Guilhaume, and K. Fiaty, Comparing REFERENCES monolithic and membrane reactors in catalytic [1] S. Benard, M. Ousmane, L. Retailleau, A. oxidation of propene and toluene in excess of Boreave, P. Vernoux, and A. Giroir-Fendler, oxygen, Catalysis today, vol. 156, pp. 301-305, Catalytic removal of propene and toluene in air 2010. over noble metal catalyst1, Can. J. Civ. Eng., [12] M. Paulis, L. M. Gandia, A. Gil, J. vol. 36, pp. 1935-1945, 2009. Sambeth, J. A. Odriozola, and M. Montes, [2] F. I. Khan, and A. Kr. Ghoshal, Removal of Influence of the surface adsorption-desorption volatile organic compounds from polluted air, processes on the ignition curves of volatile Journal of Loss Prevention in the Process organic compounds (VOCs) complete oxidation Industries, vol. 13(6), pp. 527-545, 2000.

270

Proceedings of SEEP2015, 11-14 August 2015, Paisley over supported catalysts, Applied Catalysis B: Environmental, vol. 26, pp. 37-46, 2000. [13] N. Radic, B. Grbic, and A. TerleckiBaricevic, Kinetics of deep oxidation of nhexane and toluene over Pt/Al2O3 catalysts platinum crystallite size effect, Applied Catalysis B: Environmental, vol. 50, pp. 153-159, 2004. [14] D. H. Kim, M. C. Kung, A. Kozlova, S. D. Yuan, and H. H. Kung, Synergism between Pt/Al2O3 and Au/TiO2 in the low temperature oxidation of propene, Catalysis Letters, vol. 98(1), pp. 11-15, 2004. [15] A. C. Gluhoi, N. Bogdanchikova, and B. E. Nieuwenhuys, Total oxidation of propene and propane over gold-copper oxide on alumina catalysts: Comparison with Pt/Al2O3, Catalysis Today, vol. 113, pp. 178-181, 2006. [16] M. P. Pina, M. Menendez, and J. Santamaria, The Knudsen-diffusion catalytic membrane reactor: An efficient contactor for the combustion of volatile organic compounds, Applied Catalysis B: Environmental, vol. 11, pp. L19-L27, 1996. [17] D. Uzio, S. Miachon, and J.-A. Dalmon, Controlled Pt deposition in membrane mesoporous top layers, Catalysis Today, vol. 82, pp. 67-74, 2003. [18] E. E. Iojoiu, J. Walmsley, H. Raeder, R. Bredesen, S. Miachon, and J.-A. Dalmon, Comparison of different support types for the preparation of nanostructured catalytic membranes, Rev. Adv. Mater. Sci., vol. 5, pp. 160-165, 2003. [19] G. Saracco, and V. Specchia, Catalytic filters for the abatement of volatile organic compounds, Chemical engineering science, vol. 55, pp. 897-908, 2000.

271


HYDROGEN GENERATION FROM HYDROLYSIS OF NABH4-NH3BH3 COMPOSITE PROMOTED BY ALCL3 1. 2.

Yanmin Xu1, Jie Chen2, Chaoling Wu1,*, Yungui Chen1, Zhenglyu Li1 Department of Advanced Energy Materials, College of Materials Science and Engineering, Sichuan University, Chengdu, 610064, China; email: [email protected] Skills Training Center of Sichuan Electric Power Company of State Grid, Chengdu, 611130, China

Abstract In this work, AlCl3 is used as a catalyst for hydrolysis of NaBH4-based composite xNaBH4-NH3BH3 (xSB-AB, x=4,6,8). Hydrogen generation performance are affected by at least four factors, including different addition method of AlCl3, molar ratio of SB/AB by varying the value of x, AlCl 3 amount and temperature. The experimental results demonstrate that direct addition of AlCl3 by hand mixing leads to better hydrogen generation properties than that by ball-milling owing to concentrated heat caused by uneven distribution of AlCl3. The optimized composite achieves complete dehydrogenation within 1h at 50℃, whose maximum hydrogen release reaches 2314ml/g. The improved hydrogen generation properties are attributed to thermal stimulation induced by dissolving AlCl3 in water, decline of pH value caused by the formation of Al(OH)3 precipitation, and the synergy between SB and AB. The kinetic simulation results of the xSB-AB/yAlCl3 system show that the hydrolysis kinetics is influenced by composition design. The activation energies of 15.12~29.5 kJ/mol are obtained from 4SB-AB/yAlCl3 (y=5, 10, or 20 in wt%), and 12.59~15.12 kJ/mol from xSB-AB/20AlCl3 (x=4, 6, or 8). The main solid byproducts of this hydrolysis system are Na2B4O7·10H2O, NaCl, Na2ClB(OH)4 and Al(OH)3. Keywords： catalyst, hydrolysis, sodium borohydride, Ammonia borane 1. INTRODUCTION Sodium borohydride (NaBH4, denoted as SB) and ammonia borane (NH3BH3, denoted as AB) are high-profile hydrolysis materials owing to their high theoretical hydrogen yields (THY) of 21.3 wt% and 19.5 wt% respectively (without reacted water in calculation). Even taking the reacted water into account, the THY of the system SB–H2O and AB–H2O can reach 10.8 wt% and 9.0 wt%, respectively. The features of stability under ambient conditions, safe handling and controllable hydrogen desorption process make them promising for practical application, such as portable power supply, fuel cell and on-board hydrogen storage system. But the low hydrogen productivity of either SB or AB at room temperature hinders their commercialization. Therefore, extensive

research has been conducted in recent decades to improve their respective hydrolysis properties. In our previous work, a catalyst-free hydrolysis system of solid-state xSB-yAB composite (molar ratios of x/y ranged from 1:4 to 8:1) was first proposed and studied [1,2]. Our researches demonstrate that the xSB-yAB composites without any catalyst show much better hydrolysis kinetics than their monomers due to a synergetic effect between SB and AB. And among them, 4SB-AB (i.e. the molar ratio of x/y is 4:1) releases about 99% of hydrogen in an hour at 70℃ and the corresponding hydrogen yield reaches 10wt% (with reacted water in calculation), who shows more excellent hydrogen generation performance than others [2]. Based on the facts mentioned above,

272

Proceedings of SEEP2015, 11-14 August 2015, Paisley amount of AlCl3 is added into it directly, which 4SB-AB is selected as the reference hydrolysis is described as direct addition. A SPEX8000 material in this paper. high-energy ball miller is employed for making However, with respect to the excellent samples and with the ball-to-powder weight dehydrogenation property at 70℃, the ratio of 30:1 under Ar atmosphere for 15min. performance improvement of the catalyst-free xSB-yAB composite is not significant at 0~50℃, and the hydrogen production rate needs to be 2.2 Hydrolysis Test An experimental apparatus used for hydrolysis impoved as well. Many metal salts were used of xSB-AB/yAlCl3 (x is the molar ratio of and evaluated as accelerators for SB or AB SB/AB; y is weight percentage of AlCl3) hydrolysis and normally utilized as catalyst composite is described in the previous work [1]. precursors, such as Ru(III)salts [3], Co(II) salts The composites with different ingredients (as [4-6], Cu(II) salts [7], Ni(II) salts [8,9] and shown in Table1) are weighted, and transferred Fe(III) salts [10]. In this work, AlCl3 is selected into a 100ml round-bottom flask sealed by a from various metal salts to accelerate the dual-port tube with one water inlet plug and one hydrolysis of xSB-AB due to the following hydrogen outlet plug. The reaction starts by reasons: 1). Dissolving AlCl3 in water releases stirring the mixture after 10ml water is injected large amounts of heat (the dissolution heat is into the flask through a syringe pump. The 76.85 kcal/mol at 18℃ [11]), which is hydrolysis temperatures are set at 0,15,25,35 conducive to a quick start of the hydrolysis and 50℃, respectively. The evoluted gas goes reaction of the composite; 2). AlCl3 reacts through a CuSO4 solution, a condenser and a vigorously with water: dry tube with CaCl2 in sequence. These (1) AlCl3 + 3H 2O → Al(OH )3 + 3HCl procedures aim at absorbing ammonia, The protons generated from AlCl3 separating steam and drying the gas, self-hydrolysis decrease the pH value of the respectively. The generated hydrogen is solution , and the catalytic mechanism is similar collected and measured in an inverted 1.5L to acid catalysis. graduated cylinder immersed in a water-filled tray finally. Each test is repeated at least twice in order to ensure the accuracy of the results. 2. EXPERIMENTAL Table1. Ingredients of xSB-AB/yAlCl3 Composites.

2.1 Sample Preparation In our experiments, NaBH4 (SB,≥97.0%) and AlCl3(≥99.0%) are used as received. NH3BH3 (AB) is synthesized by the procedure mentioned in literature[12], whose purity is about 90%. All raw materials are stored and all manipulations are carried out in an argon glove box with 95%, Aldrich), sodium hydroxide (General purpose grade, Fisher Scientific), polyvinylpyrrolidone (PVP, Mw = 40,000), Gold nanoparticles-5nm diameter, OD 1, stabilized suspension in citrate buffer (SigmaAldrich). All materials were of analytical grade and used without any further purification. 2.2 Synthesis of AuCu-C catalysts In the synthesis procedure, the AuCu-C catalysts were prepared by measuring, 140 mg of polyvinylpyrrolidone (PVP, 1.30 mmol of monomeric units, Mw = 40,000), 10 mL of fresh deionized water (15.0 M.cm-1, Purelab option ELGA) and 1.1 mL of 0.062 mol/L copper sulphate (CuSO4.5H2O in H2O). The obtained solution was purged with argon gas for approximately 1h in order to remove dissolved O2. Then a freshly prepared 10 mL of 1M solution of sodium borohydride (NaBH4) was added dropwise to the CuSO4-PVP solution with continuous stirring and protected with argon gas. The temperature of the solution was then raised to about 50 oC and the reaction was allowed to proceed for 1h to yield Cu nanoparticles. Deaerated freshly prepared 0.35 mL of (0.05 mol/L) HAuCl4.3H2O solution was quickly added to the CuSO4-PVP nanoparticle solution and stirred for 1h under argon protection (see figure 1). The redox potential of Au3+/Au is higher than that of Cu2+/Cu, Au (III) is reduced to Au and therefore Cu-Au particles were formed. Within a few minutes of sol generation, the sol is immobilized by adding KetJen carbon black under vigorous stirring for 30 min and then allowed to settle for 30 min. The amount of support is calculated as having a final Au-Cu

Proceedings of SEEP2015, 11-14 August 2015, Paisley loading of 20 wt.%. The resulting Carbon black2.4 Cyclic Voltammetry (CV) Experiment supported AuCu catalysts were obtained by CV experiments were performed in 0.5 filtrating the resultant solution using a Whitman M H2SO4 in the potential range of -0.2 to 1.0 V cellulose nitrate filter paper. The AuCu-C (Ag/AgCl, KClstd) at a scan rate of 20 mVs-1, catalysts were washed several times with and in 0.5M HCOOH + 0.5M H2SO4 in the deionized water to remove any free PVP potential range of -0.4 – 1.6V (Ag/AgCl, KClstd) macromolecules not bounded to AuCu-C at a scan rate of 2 mVs-1. The kinetics catalysts and until no chloride ion (Cl-) was parameters and active surface area are evaluated detected in the washing solution. The Au-Cu-C from cyclic voltammetry based on the hydrogen catalysts were dried under vacuum at 80oC desorption area corresponding to a monolayer of overnight to obtain the final catalysts. adsorbed hydrogen. Cu-PVP Au nanoparticle core shell 2+

Cu

Cu

Cu

(PVP Stabilizer)

NaBH4 + NaOH reducing agent

Au

3+

AuCu-C nanoparticles

Figure 1. Schematic illustration of the formation of AuCu-C core-shell nanoparticles 2.3 Working Electrode Preparation 2.3.1

Electrode Polishing The working electrode was polished to a mirror finish using 0.5 and 0.05m alumina suspensions sequentially before use. Then the electrodes are washed ultrasonically with ethanol, acetone and deionized water, sequentially. After that, dry in air and the electrode is ready use underlying substrate of the of the catalyst paste. 2.3.2

AuCu-C Catalyst Ink The appropriate dilution ratio of AuCu-C catalyst and analytical grade ethanol were mixed for 1h. On the surface of the GCE (OD: 6 mm ID: 3.0 mm), 9µL slurry was spread. After drying the electrode at 80 oC overnight, 4.5µL of 5wt% Nafion solution was covered on the surface of the catalyst electrode and dried overnight at 80 oC, to the obtain the working electrode. The apparent surface area of glassy carbon electrode was 0.07cm2. The 20wt% AuCu loading of the electrode was 0.1 mg/cm2.

284

Fig. 2. Schematic of the Electrochemical setup 2.5 Physical Characterization of Catalysts The structure and morphology of the prepared electrocatalysts were examined by 2.5.1

Atomic Force Microscopy (AFM) Samples were prepared for AMF imaging by deposition of AuCu-C catalyst slurry dispersed in ethanol and ethylamine solution on a freshly cleaved mica surface. The freshly cleaved circular mica sheets glued to a metal pad were used for sample analysis on the AFM imaging. After deposition, samples were allowed to dry in air and then transferred for AFM imaging. AFM imaging was performed on the Dimension icon with ScanAsyst system (Bruker Ltd, Germany) operating in soft tapping mode in air at room temperature. All cantilevers used throughout the experiments were silicon reflective aluminum coating with 3.7 m thickness. According to producer's specifications, spring constant was 26 N/m, a resonance frequency of 300 kHz and radius tip

Proceedings of SEEP2015, 11-14 August 2015, Paisley was 7 nm. All images were flattened, and then This shows the complete coverage of the copper used section analysis and particle size (core) nanoparticle by gold (shell). AuCu determination. nanoparticles will show reflections according to the selection rules for crystal diffraction, where 2.5.2 X-ray Diffraction (XRD) the Miller indices (hkl) are all odd or all even Au X-ray diffractometer (model D2 Phaser (111), Au (200), Au (220), Au (311), and Au Bruker Ltd) was employed for the phase (222), respectively. The absence of the identification with CuKα1 radiation (λ = 1.5406 superlattice reflections of 001 and 110, and the Armstrong) and a graphite monochromator were splitting reflections of 200/002 and 220/202 maintained at a tube voltage and current of 30 evidently proved the formation of core-shell kV, 10 mA, respectively to obtain X-ray structure instead of ordered intermetallic diffraction (XRD) patterns of the sample. The 2θ structure of an alloy [26, 31, 33]. AuCu angular region between 10o – 90o were explored exhibited a shift from pure Au nanoparticles (2 at a scan rate (1o min-1), with a step size of 0.1 = 38.3, 44.4, 64.6 and 77.7) indicative of the and increment (i.e. step size between data point) absence of alloy structure formation. Table 1 of 0.02. The primary divergence slit of 0.6 mm presents the particle size analysis of AuCu and was used, and Ni K-beta filter were not fitted Au catalysts on the carbon support. The mean because of the carbon black support. particle size (d), were calculated from the X diffraction plane using Debye-Scherrer equation 3 RESULTS AND DISCUSSIONS (equation 3)[23, 34]. 3.1 Crystallography of AuCu-C and AuC catalysts 0.94 l (3) d= b cos q The crystal phases of the AuCu-C catalysts were characterised using XRD. The diffraction peaks where, d is the average particle size, nm,  is the matched very well with the tetragonal facex-ray wavelength (1.54056 Å for Cu K centred cubic Au crystalline structure (JCPDS 4radiation),  is the full width at half-maximum in 0784) and did not match the AuCu alloy (JCPDS 1-072-5241). radians (FWHM) and  is the angle of Au (200) peak. The peak width  and peak position  were A u -C obtained from curve fitting using Fityk software. There were no impurity phases detected, indicating the formation of pure and highly crystalline AuCu nanoparticles. In te n s ity /a .u

A u (1 1 1 )

C u (0 0 2 )

20

A u (2 0 0 )

A u (2 2 0 )

40

A u (3 1 1 )

60

The relative crystallinities of Au nanoparticles are the ratio of the intensities of Au (111), and carbon peaks are 3.20 for AuCu-C and 2.93 for Au-C catalysts, respectively.

80

2 th e ta /D e g r e e

A u C u -C

In te n s ity /a .u

A u (1 1 1 )

Table 1 Particle size analysis of catalysts

C u (1 1 1 ) C u (0 0 2 )

A u (2 0 0 ) C u (2 0 0 )

20

40

A u (2 2 0 ) A u (3 1 1 ) A u (2 2 2 )

60

80

2 t h e t a /D e g r e e

Fig. 3. The XRD patterns of Au-C and AuCu-C

285

Catalysts Peak position (degrees): 2θ

Peal FWHM, (degree s): β2θ

d, SXRD Particle (m2/g) size (nm)

Au-C AuCu-C

1.50 1.40

1.00 1.06

44.22 44.19

312.52 294.39

Proceedings of SEEP2015, 11-14 August 2015, Paisley Amp/cm2, respectively.

3.2 Surface Morphology

Current density, Amps/cm2

Figure 4a shows the surface morphology of the AuCu-C catalyst deposited on mica sheet. The cross section analysis of the image (figure 4b) indicates showing the height of carbon black and the peaks of AuCu nanoparticles arising from the surface. A

0.014 0.012 0.010

AuCu-C AuNPs-C

0.008 0.006 0.004 0.002

-0.4 -0.2 -0.002

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Potiential, V, (vs. Ag/AgCl)

Fig.5. Cyclic Voltammogram of AuCu-C catalyst in 0.5 M H2SO4 solution and 0.5 M HCOOH + 0.5 M H2SO4 solution at scan rate of 2mV/s, 25 oC The results clearly demonstrate that the catalytic activity of AuCu-C is 3.14 time higher than commercial AuNPs dispersed on carbon black. We think this enhanced catalytic activity is due to the synergistic effect of Cu component in the AuCu-C catalysts. It is observed that the lower onset potential for formic acid electrooxidation above confirmed AuCu-C has better electrocatalytic activity of AuNPs catalyst. 4

Fig. 4. AFM image of AuCu-C catalyst The peaks observed from the section analysis show AuCu nanoparticles anchored on carbon black of height between 1-14 nm. 3.3 Electrocatalytic performance of AuCu-C for formic acid oxidation

CONCLUSION

The results clearly demonstrated that catalysts gold nanoparticles can play an important role in advancing green oxidations. The XRD evidently demonstrated the absence of superlattice and splitting reflections formation of AuCu coreshell structure. Cyclic voltammetry confirm that AuCu-C have superior catalytic activity for formic acid oxidation to commercial AuNPs dispersed on carbon black. ACKNOWLEDGEMENTS

Figure 5 shows the electrooxidation activity of formic acid on AuCu-C and AuNPs-C catalyst in the positive oxidation scan direction. The first oxidation peaks are located at 1.36 and 1.41 V, and current density of 0.0132 and 0.0042

The authors gratefully acknowledge funding support from Petroleum Technology Development Fund (PTDF) Scholarship of Nigeria and the University of Port Harcourt, Nigeria. Finally, the authors would like to thank Dr. Anthony J. R. Rennie of the Department of Chemical & Biological Engineering, University of Sheffield, UK.

REFERENCES [1] Rees NV, Compton RG. Sustainable energy: a review of formic acid electrochemical fuel

286

Proceedings of SEEP2015, 11-14 August 2015, Paisley [14] Zhu Y, Ha SY, Masel RI. High power cells. Journal of Solid State Electrochemistry. density direct formic acid fuel cells. Journal of 2011;15(10):2095-100. Power Sources. 2004;130(1‚Äì2):8-14. [2] Evans A, Strezov V, Evans TJ. Assessment [15] Choi HJ, Kim J, Kwon Y, Han J. of utility energy storage options for increased Comparative study of three different catalyst renewable energy penetration. Renewable and coating methods for direct methanol fuel cells. Sustainable Energy Reviews. 2012;16(6):4141Journal of Power Sources. 2010;195(1):160-4. 7. [16] Baik SM, Han J, Kim J, Kwon Y. Effect of [3] Hadjipaschalis I, Poullikkas A, Efthimiou V. deactivation and reactivation of palladium anode Overview of current and future energy storage catalyst on performance of direct formic acid technologies for electric power applications. fuel cell (DFAFC). International Journal of Renewable and Sustainable Energy Reviews. Hydrogen Energy. 2011;36(22):14719-24. 2009;13(6–7):1513-22. [17] Bauskar AS, Rice CA. Impact of anode [4] Rahman F, Rehman S, Abdul-Majeed MA. catalyst layer porosity on the performance of a Overview of energy storage systems for storing direct formic acid fuel cell. Electrochimica Acta. electricity from renewable energy sources in 2012;62(0):36-41. Saudi Arabia. Renewable and Sustainable [18] Oko DN, Zhang J, Garbarino S, Chaker M, Energy Reviews. 2012;16(1):274-83. Ma D, Tavares AC, et al. Formic acid electro[5] A. SMaR. Comparison between Batteries and oxidation at PtAu alloyed nanoparticles Fuel Cells for Photovoltaic System Backup. synthesized by pulsed laser ablation in liquids. World Academy of Science, Engineering and Journal of Power Sources. 2014;248(0):273-82. Technology. 2007;1:277-81. [19] Shi M, Liu W, Zhao D, Chu Y, Ma Ca. [6] Zhang S, Shao Y, Yin G, Lin Y. Recent Synthesis of palladium nanoparticles supported progress in nanostructured electrocatalysts for on reduced graphene oxide-tungsten carbide PEM fuel cells. Journal of Materials Chemistry composite and the investigation of its A. 2013;1(15):4631. performance for electrooxidation of formic acid. [7] Andújar JM, Segura F. Fuel cells: History Journal of Solid State Electrochemistry. and updating. A walk along two centuries. 2014;18(7):1923-32. Renewable and Sustainable Energy Reviews. [20] Huang J ZZ, Yang Song, Xiongwu Kang, 2009;13(9):2309-22. Ke Liu, Wancheng Zhou, Shaowei Chen. [8] Aslam NM, Masdar MS, Kamarudin SK, Electrocatalytic Activity of Palladium Daud WRW. Overview on Direct Formic Acid Nanocatalysts Supported on Carbon Fuel Cells (DFAFCs) as an Energy Sources. Nanoparticles in Formic Acid Oxidation. Journal APCBEE Procedia. 2012;3:33-9. of electochemistry. 2012;18(6):508-14. [9] Rice C, Ha S, Masel RI, Waszczuk P, [21] Feng L, Yao S, Zhao X, Yan L, Liu C, Xing Wieckowski A, Barnard T. Direct formic acid W. Electrocatalytic properties of Pd/C catalyst fuel cells. Journal of Power Sources. for formic acid electrooxidation promoted by 2002;111(1):83-9. europium oxide. Journal of Power Sources. [10] An H, Cui H, Zhou D, Tao D, Li B, Zhai J, 2012;197(0):38-43. et al. Synthesis and performance of Pd/SnO2– [22] Kwon Y, Lai SCS, Rodriguez P, Koper TiO2/MWCNT catalysts for direct formic acid MTM. Electrocatalytic Oxidation of Alcohols on fuel cell application. Electrochimica Acta. Gold in Alkaline Media: Base or Gold 2013;92(0):176-82. Catalysis? Journal of the American Chemical [11] Ha S, Larsen R, Masel RI. Performance Society. 2011;133(18):6914-7. characterization of Pd/C nanocatalyst for direct [23] Lee JK, Lee J, Han J, Lim TH, Sung YE, formic acid fuel cells. Journal of Power Sources. Tak Y. Influence of Au contents of AuPt anode 2005;144(1):28-34. catalyst on the performance of direct formic acid [12] Rice C. Catalysts for direct formic acid fuel fuel cell. Electrochimica Acta. 2008;53(9):3474cells. Journal of Power Sources. 8. 2003;115(2):229-35. [24] Sardar R, Funston AM, Mulvaney P, [13] Yu X, Pickup PG. Recent advances in direct Murray RW. Gold Nanoparticles: Past, Present, formic acid fuel cells (DFAFC). Journal of and Future. Langmuir. 2009;25(24):13840-51. Power Sources. 2008;182(1):124-32.

287

Proceedings of SEEP2015, 11-14 August 2015, Paisley [25] Chandra P, Singh J, Singh A, Srivastava A, Goyal RN, Shim YB. Gold Nanoparticles and Nanocomposites in Clinical Diagnostics Using Electrochemical Methods. Journal of Nanoparticles. 2013;2013:12. [26] Bauer JC, Mullins D, Li M, Wu Z, Payzant EA, Overbury SH, et al. Synthesis of silica supported AuCu nanoparticle catalysts and the effects of pretreatment conditions for the CO oxidation reaction. Phys Chem Chem Phys. 2011;13(7):2571-81. [27] Yu X, Pickup PG. Codeposited PtSb/C catalysts for direct formic acid fuel cells. Journal of Power Sources. 2011;196(19):7951-6. [28] Haruta M. Size- and support-dependency in the catalysis of gold. Catalysis Today. 1997;36(1):153-66. [29] Zhu W, Michalsky R, Metin Ö, Lv H, Guo S, Wright CJ, et al. Monodisperse Au Nanoparticles for Selective Electrocatalytic Reduction of CO2 to CO. Journal of the American Chemical Society. 2013;135(45):16833-6. [30] Zhu Y, Jin R, Sun Y. Atomically Monodisperse Gold Nanoclusters Catalysts with Precise Core-Shell Structure. Catalysts. 2011;1(1):3-17. [31] Wang G, Xiao L, Huang B, Ren Z, Tang X, Zhuang L, et al. AuCu intermetallic nanoparticles: surfactant-free synthesis and novel electrochemistry. Journal of Materials Chemistry. 2012;22(31):15769-74. [32] Long NV, Hien TD, Asaka T, Ohtaki M, Nogami M. Synthesis and characterization of Pt–Pd nanoparticles with core-shell morphology: Nucleation and overgrowth of the Pd shells on the as-prepared and defined Pt seeds. Journal of Alloys and Compounds. 2011;509(29):7702-9. [33] Chen CH, Liou WJ, Lin HM, Wu SH, Mikolajczuk A, Stobinski L, et al. Carbon nanotube-supported bimetallic palladium-gold electrocatalysts for electro-oxidation of formic acid. Physica Status Solidi (A) Applications and Materials Science. 2010;207(5):1160-5. [34] Liu Z, Hong L, Tham MP, Lim TH, Jiang H. Nanostructured Pt/C and Pd/C catalysts for direct formic acid fuel cells. Journal of Power Sources. 2006;161(2):831-5.

288


HYDROGEN EFFECTS ON IGNITION DELAY OF METHYL BUTANOATE/NHEPTANE MIXTURE Seunghyeon Lee1, Heeseon Kim1, Soonho Song2 1. Graduate School, Department of Mechanical Engineering, Yonsei University, Seoul, Republic of Korea 2. Department of Mechanical Engineering, Yonsei University, Seoul, Republic of Korea; email: [email protected] Abstract The interest in alternative fuels is increasing due to the concerns about air pollution and the lack of fossil fuels. Hydrogen, which has a high octane number and does not emit carbon dioxide, is one of the promising clean alternative fuels. Many researches have been conducted about the effect of hydrogen addition on the combustion of conventional fuels. Researches on dual fuel combustion are gaining more and more attention as a new combustion technology called Reactivity Controlled Compression Ignition (RCCI) showed the possibility of high efficiency and low emission. For this type of combustion, the combustion process is mostly controlled by the chemical kinetics of the fuel itself. Therefore, conducting a research on the chemical kinetics of the fuel is essential for the development of the RCCI combustion. In this study, the effects of hydrogen on the ignition delay of methyl butanoate/n-heptane mixture will be investigated. An experiment has been conducted using the Rapid Compression Machine (RCM) by varying the temperature and fuel composition, and a numerical analysis was also executed through the CHEMKIN-PRO software to understand how the combustion process changes. Both the experimental and numerical results show that with hydrogen addition, the ignition delay increases at a low temperature (1030K). The difference in the reaction paths at different temperatures is the reason why the results show opposite tendencies Keywords: Ignition delay, Hydrogen, Methyl butanoate, N-heptane, Rapid Compression Machine 1 INTRODUCTION New combustion technologies with high fuel efficiency and low emissions are getting more and more attention due to climate change, air pollution, and instability of oil supply. The HCCI (Homogeneous Charge Compression Ignition) combustion is one of promising technologies which has advantages of high efficiency and low NOx and soot emission due to high compression ratio and low temperature combustion. However, the control of ignition timing in the HCCI combustion is so hard because it doesn’t have any controlling device such as a fuel injector or a spark plug. Ignition timing of the HCCI combustion is solely dependent on chemical kinetic of fuel itself. The RCCI (Reactivity Controlled Combustion Ignition) combustion is able to control ignition timing by blending a couple of fuels having different reactivity, and it has been widely investigated to make up for the weakness of the HCCI combustion. The research on the ignition delay time of fuel should be needed because it is

289

one of the most important parameters of the RCCI combustion. Researches on biofuels have been extensively conducted in order to reduce the dependence on fossil fuels. Biodiesel, one of the most widely used biofuels, has advantages of less carbon monoxide (CO), unburned hydrocarbon (UHC), and soot emission. It is also said to be a carbon neutral fuel. One of the biggest structural differences between biodiesel and diesel is that biodiesel contains oxygen atoms, so the combustion reaction path of biodiesel is considerably different to diesel. Therefore, research on reaction pathway of oxygenated fuels, especially methyl esters, is essential for understanding on biodiesel combustion. Dooley et al. studied the autoignition pathway of methyl butanoate (C5H10O2) using diverse type of experimental equipment and kinetic analysis [1]. In addition, kinetic analysis on large methyl esters such as methyl decanoate (C11H22O2), methyl palmitate (C17H34O2), and methyl linolenate (C19H32O2) have been conducted. These studies, however, were challenging

Proceedings of SEEP2015, 11-14 August 2015, Paisley vessel. Figure 1 describes the experimental setup because these components require large for this study. mechanisms. Brakora et al. suggested a fuel mixture of 1 mole of methyl butanoate and 2 moles of n-heptane as a biodiesel surrogate since it has similar physical properties to biodiesel [2]. Hydrogen has been drawing attention as a fuel additive due to its superior ignition and emission characteristics. It can reduce CO2 and soot emission, and has high flame propagation speed. Effects of hydrogen blending to the combustion of hydrocarbon fuels have been investigated by many researchers. Aggarwal et al. did a research Figure 1. The schematic diagram of on ignition characteristics of heptane-hydrogen experimental setup fuel blends [3]. Pan et al. studied hydrogen addition effects in ignition characteristics of Experiments using inert gas (nitrogen) are dimethyl ether [4]. conducted five times to identify the In this study, hydrogen effects on ignition delay reproducibility of the RCM. Figure 2 shows that of methyl butanoate/n-heptane mixture were the in-cylinder pressure increases during the investigated. Through RCM experiment and compression process and then decreases due to CHEMKIN analysis, it was analysed how added heat loss through the cylinder wall [6]. Pressure hydrogen changes autoignition reaction. profiles of five experiments are almost same, so the reproducibility of this RCM is confirmed. 2 EXPERIMENTAL SPECIFICATIONS 2.1 Experimental equipment The RCM is used to figure out hydrogen effects on methyl butanoate/n-heptane mixture combustion. This machine is useful for studying combustion phasing because it can directly measure the in-cylinder pressure from which the ignition delay can be directly calculated. The incylinder pressure is measured by a pressure transducer (6125C; Kistler). Then the pressure signal is amplified and converted to digital signals by a data acquisition (DAQ) hardware. The RCM is driven by pneumatic pressure and stopped by hydraulic pressure. The creviced piston is used to prevent roll-up vortex [5]. The compression time is 25ms and the compression ratio is 16.5 in this study. Fuels are mixed with air in the premixing vessel. The inside of the vessel is made high vacuum state by a vacuum pump. Then, methyl butanoate and n-heptane is injected into the vessel using a micro syringe and is vaporized immediately due to the extremely low pressure. 99.999% purity hydrogen and air are directly entered into the vessel through the gas line. The amount of entered gas is regulated by a metering valve (SS-4BMW-VCR; Swagelok). The inside temperature of vessel can be set by wrapping the vessel with three pieces of heating jackets controlled by a thermostat. The magnetic stirrer is used to make homogeneous charge in the

290

Figure 2. Pressure profiles of five same experiments. Gas : pure air(99.9%), PC=26bar, Compression ratio=16.4 2.2 Definition of the Ignition Delay The ignition delay is one of the most important parameters representing autoignition characteristics. In this study, the ignition delay is defined as the time interval from when the peak pressure is occurred by compression to when the pressure rise rate(PRR) has a maximum value. Figure 3 shows the definition of the ignition delay.

Proceedings of SEEP2015, 11-14 August 2015, Paisley physical properties to soy-based biodiesel as represented in Table 1. Hydrogen was added to this mixture to understand effects of hydrogen to methyl butanoate/n-heptane combustion. Table 1. Comparison of suggested fuel mixture and soy-based biodiesel (reproduced from [2])

Figure 3. The example of the pressure profile in the RCM experiment. Fuel: Isooctane, Equivalence ratio = 0.5, P0=1bar, T0=296K, Compression ratio=15.2 2.3 Temperature at the End of Compression In chemical kinetics, the Arrhenius equation is as follows,

where k is the rate constant, Ea is the activation energy, R is the universal gas constant, and T is the temperature. According to this equation, the rate constant is highly sensitive to the temperature which is in the exponential term. Therefore, inaccurate estimation of temperature will create non-negligible error, thus accurate estimation of the temperature is important. The adiabatic compression assumption is used in the compression process [6]. The compressed temperature is calculated using the following equation,

where TC is the calculated compressed temperature, PC is the pressure at end of compression, T0 is the initial temperature, P0 is the initial pressure, and γ is the ratio of specific heat. The value of γ is slightly different according to the equivalence ratio and temperature. 2.4 Experiment conditions In this study, a fuel mixture of 1 mole of methyl butanoate and 2 moles of n-heptane is used as a biodiesel surrogate. This fuel mixture has similar 291

1 mole of methyl butanoate / 2 moles of n-heptane mixture

Soy-based biodiesel

Chemical Formula

C19H42O2

~C19H34O2

Lower H.V. (MJ/kg)

39

36-38

Mol. Mass (kg/kmol)

302

292

% Oxygen (by mass)

~11%

~11%

In the RCM experiment, compression pressure was 15bar and equivalence ratio was 0.5. Temperature range was changed from 681K to 773K by initial temperature and mixture composition. Mole fractions of hydrogen were 0%, 25%, 50%, and 75%. 3 KINETIC MODELING METHOD In the compression stroke, the adiabatic compression assumption was used for the analysis as explained in the section 2.3. After the compression was finished, closed homogeneous reactor model was used to simulate in-cylinder condition of the RCM. Heat loss effect was included by applying the constant heat transfer coefficient. The computing method was necessary for analysing the chemical kinetics including thousands of reactions and hundreds of chemical species. A closed homogeneous reactor model in CHEMKIN-PRO was used for chemical kinetic analysis in this study [7]. For the exact numerical calculation, validated data of thermodynamic properties of chemical species and reaction rate of various chemical reactions related to methyl butanoate/n-heptane combustion were needed. The detailed n-heptane mechanism was developed by the Lawrence Livermore National Laboratory (LLNL) consisting of 504 species and 2827 reactions [8]. Dooley et al. subtracted C5-C7 species and related reactions from LLNL n-heptane

Proceedings of SEEP2015, 11-14 August 2015, Paisley mechanism, and added methyl ester-related radical ( • C7H14OOH). The hydroperoxy alkyl species and reactions to develop methyl radical reacts with the molecular oxygen butanoate mechanism consisting of 275 species forming the peroxy alkyl hydroperoxy radical (• and 1549 reactions [1]. In this study, methyl butanoate/n-heptane mechanism, including 711 OOC7H14OOH). Then, peroxy alkyl species and 3064 reactions, was developed by hydroperoxy radical is broken up successively to adding methyl ester-related species and reactions keto-hydroperoxide, hydroxyl, and oxy-radical developed by Dooley et al. to LLNL n-heptane producing an OH radical. This is the main mechanism. This mechanism included 711 reaction path in low temperature cycle. Under species and 3064 reactions. It was confirmed high temperature, decomposition of H2O2 into that most of important reaction mechanisms two OH radicals is important. The number of related to hydrogen were included in this radicals can be increased by chain-branching mechanism, so other mechanisms of hydrogen through these reactions. Initial hydrogen reaction were not added. abstraction reaction is important to activate these successive reactions. 4 RESULT AND DISCUSSION 4.2 Experimental Results 4.1 Oxidation Path of Methyl Butanoate and N-heptane In the oxidation of methyl butanoate under RCM condition, alkyl radicals are mainly formed by reaction with oxygen molecule or oxygen atom,

and β-scission. Then, these alkyl radicals break into smaller molecules and radicals. Most significant reactions in methyl butanoate oxidation are MB+CH3O2•→MBXJ+CH3O2H and MB+HO2•→MBXJ+H2O2, where MBXJ means four types of fuel alkyl radicals. CH2O2H breaks into CH3O and OH, and this exothermic reaction increases the system temperature, makes hydrogen peroxide decompose to OH radicals.

Figure 4. Main reaction path of n-heptane (reproduced from [6]) Figure 4 represents main oxidation path of nheptane. Under low temperature, the first step is H abstraction from n-heptane producing alkyl radical. This alkyl radical reacts with the molecular oxygen then forms alkyl-peroxy radical (C7H15OO • ), and the isomerization of alkyl-peroxy radical forms hydroperoxy alkyl

292

Figure 5. Pressure profiles of methyl butanoate/n-heptane mixture combustion with hydrogen addition. Equivalence ratio = 0.5, TC=681-683K Some of pressure profiles obtained from experiment are represented in Figure 5. All of pressure profiles show second-stage ignition characteristics, so it is confirmed that low temperature reactions responsible for 1st stage ignition were occurred. Reaction is fastest at 0% of hydrogen, and it is slowing down as hydrogen is added. Maximum pressure is decreased with hydrogen addition, and there are two reasons for change. First is decreased reaction speed by hydrogen, and second is reduced heating value. Molar heating value of hydrogen is lower than methyl butanoate and n-heptane, so heating value of fuel mixture is decreased by hydrogen.

4.3 Results of Numerically Analysis

Proceedings of SEEP2015, 11-14 August 2015, Paisley On the other hand, n-heptane clearly shows NTC tendency [8]. Figure 6(a) represents that fuel mixture used in this study has NTC characteristic at 770-870K. Numerical result qualitatively well correspond to experimental result. Ignition delay is lengthened by hydrogen when the temperature is lower than 1030K. Interestingly, a tendency of ignition delay change with hydrogen addition is wholly reversed above 1030K. In this temperature range, ignition delay is shortened by hydrogen. 4.4 Sensitivity Analysis

Figure 6. Ignition delay of methyl butanoate/n-heptane/hydrogen mixture at (a) low temperature range (b) high temperature range. Equivalence ratio = 0.5, PC=15bar. Symbol: Experiment, Line: Numerical analysis Experiment was conducted varying temperature (681-773K) and hydrogen mole fraction (0-75%) at constant pressure and equivalence ratio. Ignition delay times obtained from experiment and numerical analysis in relatively low temperature range are plotted in Figure 6(a). Figure 6(b) shows the result of numerical analysis in high temperature range. Generally, chemical reactions are activated by increasing temperature, and therefore, ignition delay is shortened. However, ignition delay times of some types of fuel are lengthened with temperature rise at particular temperature range. The region showing this tendency is called NTC (Negative Temperature Coefficient) region. Biodiesel has NTC characteristic, but methyl butanoate does not show NTC characteristic [1]. 293

Figure 8. Normalized sensitivity of ignition delay of methyl butanoate/nheptane/hydrogen mixture at (a) 800K (b) 1200K. Equivalence ratio = 0.5, PC = 15bar In section 4.2, a tendency of ignition delay change as hydrogen addition at high temperature

Proceedings of SEEP2015, 11-14 August 2015, Paisley H+O2=O+OH reaction. Therefore, the overall is different from those at low temperature. To reactivity is increased by hydrogen addition. find a reason for different tendency, sensitivity analysis of ignition delay is performed at 800K and 1200K. Equation for normalized sensitivity ACKNOWLEDGEMENTS for ignition delay is, This work NRF-2014R1A2A1A11051130 was supported by Mid-career Researcher Program where k is an original reaction constant and τ is through NRF grant funded by the MEST ignition delay. As a result of sensitivity analysis, top 20 REFERENCES reactions are represented in Figure 8. At 800K, [1] S. Dooley et. al., Autoignition measurements H abstraction of methyl butanoate by reaction and a validated kinetic model for the biodiesel with OH radical is main decomposition pathway surrogate, methyl butanoate, Combustion and of methyl butanoate. Low temperature reactions Flame, Vol. 153, pp. 2-32, 2008 of n-heptane explained in section 4.1 are also [2] J. L. Brakora et. al., Development and important. When hydrogen is added, sensitivity Validation of a Reduced Reaction Mechanism of the OH+H2=H+H2O reaction is rapidly for Biodiesel-Fueled Engine Simulations, SAE increased. Most of added hydrogen molecules International, Detroit, 2008 react with OH radicals to produce H and H2O, [3] S. K. Aggarwal et. al., Ignition and therefore, production of OH radical is characteristics of heptane-hydrogen and heptaneretarded by H2 addition. At 1200K, H methane fuel blends at elevated pressures, abstraction reactions are less important than International Journal of Hydrogen Energy, Vol. those at 800K. H+O2=O+OH and 36, pp. 15392-15402, 2011 CH3+HO2=CH3O+OH reactions become [4] L. Pan et. al., Kinetic modeling study of important. The sign of sensitivity of hydrogen addition effects on ignition OH+H2=H+H2O reaction is reversed, having a characteristics of dimethyl ether at engineminus sign. When hydrogen is added, hydrogen relevant conditions, International Journal of molecules react with OH radicals and H and Hydrogen Energy, Vol. 40, pp. 5221-5235, 2015 H2O are produced. However, produced H [5] G. Mittal et. al., A rapid compression radicals immediately react with oxygen machine for chemical kinetic studies at elevated molecules to produce O and OH radicals. These pressures and temperatures, Combustion Science successive reactions are mostly same as a radical and Technology, Vol. 179, pp. 497-530, 2007 branching reaction, so overall reaction is [6] S. Tanaka et. al., A reduced chemical kinetic stimulated by added hydrogen. As a result, model for HCCI combustion of primary ignition delay is shortened by hydrogen addition. reference fuels in a rapid compression machine, Combustion and Flame, Vol. 133, pp 467-481, 5 CONCLUSION 2003 In this study, hydrogen effects on methyl [7] CHEMKIN-PRO 15112, Reaction Design: butanoate/n-heptane oxidation were investigated. San Diego, 2011. The main conclusions of this study are as [8] M. Mehl et. al., Kinetic modeling of gasoline follows: surrogate components and mixtures under 1. Methyl butanoate/n-heptane mixture shows a engine conditions, Proceedings of the two stage ignition characteristic. Methyl Combustion Institute, Vol. 33, pp. 193-200, butanoate does not have NTC region, but methyl 2011 butanoate/n-heptane mixture has NTC region like biodiesel. 2. Below 1000K, ignition delay is lengthened by hydrogen addition. OH radicals are consumed by OH+H2=H+H2O reaction, so the overall reactivity is reduced by hydrogen addition. 3. Above 1050K, ignition delay is shortened by hydrogen addition. H radicals, made by OH+H2=H+H2O reaction, react with oxygen molecules to produce O and OH radicals by 294


MODIFIED SILVER CATALYST FOR A HYDROGEN PEROXIDE PEM FUEL CELL J. G. Carton 1, L. Gonzalez-Macia2, A. J. Killard2 1. Department of Manufacturing and Mechanical Engineering, Dublin City University, Dublin, Ireland; email: [email protected] 2. Department of Biological, Biomedical and Analytical Sciences, University of the West of England, Coldharbour Lane, Bristol BS16 1QY, UK Abstract A cathode carbon gas diffusion layer (GDL) electrode, modified with a silver paste and a combination of surfactant (dodecylbenzenesulphonic acid) and KCl, for the enhanced electrocatalytic reduction of hydrogen peroxide, is described. Physical analysis of the cathode GDL shows modifications to both the surface morphology and chemical composition of the electrode surface. The modified GDL has been demonstrated as a suitable cathode electrocatalyst for a direct hydrogen peroxide fuel cell. Preliminary results from the fuel cell indicate an open circuit voltage approaching 1 volt and current densities in the range of 0.005A/cm2 can be achieved at 0.3 volts. This material may have potential to replace or reduce costly precious metal catalysts and improve cathode oxygen reduction efficiency. Keywords: PEM, fuel cell, silver catalyst, hydrogen peroxide, GDL 1 INTRODUCTION Proton exchange membrane (PEM) fuel cells have proven to be very advantageous compared to other fuel cell types because of their low temperature operation, high power density, fast start up, system robustness, flexibility of fuel type (with reformer) and reduced sealing, corrosion, shielding or leaking concerns [1, 2]. A conventional PEM fuel cell consists of a membrane electrode assembly (MEA), which contains a proton exchange membrane, an electrically conductive porous gas diffusion layer (GDL) and an electrocatalyst layer, sandwiched between two flow plates [3, 4]. Traditionally, hydrogen has been used as the fuel and oxygen as the oxidant in the development of fuel cells[5]. The slow kinetics of the oxygen reduction has long been known as the main limiting factor of an oxygen fuel cell system. The direct oxygen reduction at the cathode involves four electron transfer, equation (1), which implies a high activation barrier to overcome. O2 + 4H+ + 4e- → 2H2O

(1)

An alternative pathway is a series of twoelectron transfers in which Oxygen (O2) is reduced to water (H2O) via hydrogen peroxide (H2O2), has been suggested, equation (2) and (3):

295

O2 + 2H+ + 2e- → H2O2

(2)

H2O2 +2H+ + 2e- → 2H2O

(3)

A parallel pathway in which both the direct and series pathways take place simultaneously has also been reported [6, 7]. In order to achieve an acceptable cell performance, to improve the electrode life and to decrease the polarisation created in the cathode due to the H2O2 formation, many catalysts for both oxygen and hydrogen peroxide reduction have been investigated as electrode materials [8, 9]. Platinum (Pt) has been reported as one of the best electrocatalysts for PEM fuel cells, but its high cost has hindered more widespread commercialization [6, 7]. Porous electrodes fabricated with high-area carbon as a conductive support and a catalyst for the O2 reduction and H2O2 decomposition have been also examined. Venkatachalapathy et al. [9] investigated the catalytic activity of several materials (carbon, Ptsupported on high-area carbon, Pt, lead ruthenate and ruthenium oxide) towards H2O2 decomposition. Pt showed the best catalytic effect although ruthenate activity was also comparable. Therefore, lead ruthenate could be used in a dual catalyst on the cathode side of fuel cells to decompose the produced H2O2. Nevertheless a fundamental problem with the use of O2 (or any gaseous reactant) is the

Proceedings of SEEP2015, 11-14 August 2015, Paisley For the preparation of the cathode catalyst, 1.3 g availability of the reactant at the active catalyst silver paste (Electrodag PF-410) was diluted in 1 site on the electrode surface. Liquid reactants are mL diethylene glycol monoethyl ether acetate notably more abundant at the active site than (99%, Sigma–Aldrich) and applied on the gaseous reactants, so the use of H2O2 as the fuelcathode GDL electrode (Ag-GDL). Further cell oxidant is an alternative [10, 11]. H2O2 as an catalyst modifications were carried out by oxidising agent in fuel cells is not a new concept, immersing the Ag-GDL electrodes in a 3.3 x 10-2 H2O2 fuel cells are air-independent energy M DBSA (DBSA-D0989, TCI Europe) and 0.1 generation systems which could be potentially M KCl mixed solution for 3 hours. Following employed in anaerobic environments such as this treatment the Ag-GDLs were thoroughly submersible and space applications [12-17]. rinsed with distilled water to remove excess Kjeang et al. [18] developed a microfluidic fuel modification solution and left to dry at room cell incorporating H2O2 as an oxidant. Hightemperature. All solutions were prepared using surface area electrodeposited Pt and Pd on Au 18 MΩ Milli-Q water. electrodes were evaluated as catalysts showing Surface morphology of the samples was similar catalytic activity towards H2O2 reduction. examined using SEM, carried out with a Hitachi The H2O2-based fuel cell showed higher power S-3400N using secondary electron (SE) and current densities than cells based on detection. An acceleration voltage of 20 kV and dissolved oxygen. 100, 1.0k and 5.0k x magnification were used to Reciently, Gonzalez-Macia et al. [19, 20] have obtain the surface images. reported a significant enhancement in the The H2/H2O2 fuel cell was assembled as a catalytic activity of a silver-based electrode typical PEM fuel cell [2, 3]; the cathode GDL towards H2O2 reduction after exposure to a with varying Ag loads, was placed into the mixed surfactant/salt solution. The silver paste cathode chamber and the anode GDL, with 60% electrodes which were modified with a wt Pt (0.5 mg/cm2) was placed onto the opposite dodecylbenzenesulphonic acid (DBSA) and side of the proton conducting membrane (Nafion potassium chloride (KCl) lyotropic solution 212). Gasketing was then applied to a double exhibited up to 80-fold higher amperometric serpentine flow plates with each flow plate responses to H2O2 at – 0.1 V vs. Ag/AgCl, in attached to the anode or cathode sides, PBS pH 6.8. Consequently, the DBSA/KCl respectively. Backing plates were finally secured modified silver paste could form the basis of a with bolts. The assembly was leak tested under novel catalyst for H2O2 reduction in the low pressure in a water bath. Flow rates of 50 manufacture of fuel cells that use O2 reduction as ml/min (99.99%) and 0.1 ml/min (0.1 M) were well as H2O2-based fuel cells. set for H2 (BOC) and H2O2 (30% v/v, Sigma– In the present work, carbon GDLs were modified Aldrich), respectively. with silver (Ag) paste and a DBSA/KCl solution. The experimental setup used (Figure 1) was The modified GDL (Ag-GDL) was then similar to that of Carton & Olabi [21]. The H2 employed as the cathodic electrode of a reactant gas was stored in a compressed cylinder. hydrogen fuel cell with hydrogen peroxide as the H2 pressure regulators and a volumetric flow oxidant. GDL modifications with different silver controller managed the H2 gas flow. H2 gas was concentrations were evaluated as the platform humidified as stated by the manufacturer of the for the catalyst. Scanning electron microscopy PEM. A syringe pump (CMA 100) was used to (SEM) measurements were performed to deliver exact flow rates of H2O2 to the cathode. characterise the GDLs surfaces before and after The open circuit voltage and the fuel cell the surfactant-based modification. operating voltage were detected by the DAQ hardware and analysed through the software. 2 EXPERIMENTAL The open circuit voltage reading was also A PEM fuel cell with an active area of 14.45 cm2 checked at the anode and cathode using a and serpentine flow plates was used in the multimeter (Fluke 8808A digital multimeter). experiments (Model: ECOFC-1 - H2Economy, The fuel cell current was measured using the Armenia). The MEA (same supplier) consisted same multimeter in series with the external load. of a Nafion 212 membrane and carbon paper 2 GDLs. The anode catalyst contained 0.5 mg/cm 60% wt Pt on paper GDL. 296

Proceedings of SEEP2015, 11-14 August 2015, Paisley A

Figure 1. Experimental set up of the H2/H2O2 PEM fuel cell. In order to enhance the catalytic activity of GDL membranes towards H2O2 reduction, the GDLs were first covered with a 1.3 g/mL silver paste (Electrodag PF-410) solution and left to dry at room temperature overnight. Several dilutions (1:20, 1:50 and 1:400) of the silver preparation equivalent to 6.5 x 10-2, 2.6 x 10-2 and 3.25 x 103 g/mL, respectively, were also made in diethylene glycol monoethyl ether acetate and applied to the GDLs. Figure 2 shows the SEM images of GDL substrates modified with silver paste solutions containing various loadings of silver. As can be observed, the GDL modified with a 1:400 dilution of silver paste (Figure 2E) had a surface similar to the unmodified GDL, where only the carbon fibres were observed (Figure 2A). When the content of silver in the modification solution increased, the GDL surfaces exhibited a higher coverage of the carbon platform with high contrast silver. The GDL electrodes were subsequently modified with a 3.3 x 10-2 M DBSA and 0.1 M KCl mixed solution in order to enhance the catalytic activity of the silver structures towards H2O2 reduction, as was previously reported in the literature [19, 20]. The Ag-GDLs were immersed in the surfactant-based solution and left to react for 3 hours. The surfaces were then rinsed and left to dry before further characterisation. Figure 3 shows the SEM images of the Ag-GDL substrates initially covered with a 1.3 g/mL silver paste solution before and after further modification with DBSA/KCl solution. As can be seen, spheroidal structures appeared on the Ag-GDL surfaces following the surfactant-based modification.

297

B

C

D

E

Figure 2. SEM images using secondary electron (SE) detection of GDL electrodes unmodified (A) and modified with preparations of silver paste 1.3 g/mL (B); 1:20 (C); 1:50 (D) and 1:400 (E). Accelerating voltage of 20 kV (1.0k×magnification). This phenomenon has been already observed when silver screen printed electrodes were modified with similar surfactant/salt solutions [19, 20]. From these images it can be noticed that DBSA/KCl modified electrodes exhibited more surface nano-structuring compared to the unmodified ones. This effect may be attributed to the possible interaction of micellar, or more likely, hexagonal or lamellar structures formed in the surfactant/salt modification solution and the silver paste electrodes, creating an enhanced surface for the catalytic reduction of H2O2, as reported by Gonzalez-Macia et al. [19] SEM analysis of the DBSA/KCl modified Ag-GDL was used as the basis for the selection of a loading to be used for fuel cell testing. SEM analysis of the DBSA/KCl modified Ag-GDL was used as the basis for the selection of a loading to be used for fuel cell testing. The 1:400 and 1:20 dilution pastes did not show enough GDL surface coverage, while the 1.3 g/mL initial preparation created crusts on the GDL that could block the GDL pores.

Proceedings of SEEP2015, 11-14 August 2015, Paisley A

B

C

D

Figure 3. SEM images using secondary electron (SE) detection of Ag-GDL electrodes modified with a preparation of 1.3 g/mL silver paste before (A), (C) and after (B), (D) further modification with a DBSA/KCl solution. Accelerating voltage of 20 kV (1.0k and 3.0k×magnification). Reducing the GDL pores may affect the H2O2 mass transport through the catalyst and therefore, the fuel cell operation [22]. Therefore, the 1:50 silver preparation (equivalent to 2.6 x 10-2 g/mL) was use for the H2O2 fuel cell testing. 3 RESULTS AND DISCUSSION Figure 4 shows the I-V results of the DBSA/KCl modified Ag-GDL in the H2/H2O2 PEM fuel cell. The results indicate that an open circuit voltage approaching 1 V could be achieved. A number of repeats using the same modified Ag-GDL were performed, where the current density increased from one test to the next, until a maximum level of 5 mA/cm2 at 0.3 V was reached. This may have been due to the increasing temperature of the cell from test to test or to an autocatalytic process during the H2O2 reduction. The following mechanism has been proposed by several authors for the electrochemical reduction of H2O2 on Ag [23, 24]: H2O2 + e- → OHads + OH-

(4)

OHads + e- ↔ OH-

(5)

2OH- + 2H+ ↔ 2H2O

(6) 298

with the first step as the rate-determining one [24]. Therefore, any structure that stabilizes OHads would favour the reaction.

Figure 4. H2O2 Fuel Cell experimental I-V (▼) and power density (●) results: (a) repeat 1; (b) repeat 2; (c) repeat 3 The autocatalytic reduction of H2O2 on silver electrodes has been previously reported by Flatgen et al. [25] They observed two mechanisms of H2O2 reduction operating at different over-voltages. At potentials more negative than – 0.45 V (vs Ag/AgCl) the “normal” reduction takes place whereas a second mechanism was operative at E ≥ – 0.35 V. They proposed that the rate of H2O2 reduction in this

Proceedings of SEEP2015, 11-14 August 2015, Paisley (PVA) film with Ag nanoparticles in it. They second case was increased by the presence of the related that process to the oxygen reduction adsorbate (OH)ad formed during the reduction formed from H2O2 decomposition catalyzed by process on the silver surface, leading to an silver and it was confirmed with a Clark autocatalytic reaction on the electrode. electrode. No evidences of oxygen influence It is thought that DBSA/KCl modified Ag during H2O2 electrochemical reduction on surfaces lead to the formation or stabilisation of DBSA/KCl modified Ag paste electrodes were OHads or OH radical, generally implicated in the previously reported. Amperometric electrochemical reduction of H2O2. Although the measurements were performed in the presence exact mechanism is not yet fully elucidated, it is and absence of oxygen in the solution and no believed that catalytic structures composed of noticeable differences were observed, which dodecyl benzene sulphonic acid (or other may be due to the low H2O2 concentrations (1 to appropriate amphiphilic molecules) and salts 5 x 10-3 M) used for the electrochemical (potassium and sodium chlorides or others) are measurements. However, surfactant/salt formed at the surface of the silver paste modified silver paste electrodes also showed an electrode. These structures would catalyse the increase in their catalytic activity towards H2O2 initial reduction of H2O2 which in turn would decomposition at concentrations of 1 M (data not oxidise the surface of the silver electrode. The shown). The modification of the spheroidal oxidised silver electrode is then returned to its structures after the two processes might suggest neutral state via electrochemical reduction that those structures were involved in both resulting from an applied cathodic potential at phenomena, but not necessarily following a the electrode. common pathway. Once the H2O2 fuel cell stabilised, current In this work, H2O2 concentrations of 0.1 M were densities in the range of 5 mA/cm2 at 0.3 V used. This might imply that not only H2O2 could be achieved, as shown in Figure 4. electrochemical reduction, but also Although these measured current densities are decomposition was taking place, which would not as high as those obtained with Pt cathode lead to a decrease in the current density and a PEM fuel cells (~ 800 mA/cm2 at 0.5 V or possible polarization of the cathode by oxygen higher) [26], Ag is approx. one hundredth the accumulation. Reducing H2O2 concentration in cost of Pt and therefore, the Ag-GDL stands as a the oxidant flow might improve the fuel cell promising material for future developments of a operation and subsequently, enhance the voltage Pt-less fuel cell. and current density outcomes. The losses in the H2O2 fuel cell (Figure 4) The electrocatalytic reduction of hydrogen corresponded to “activation loss”, and this could peroxide has proven to play a key role in a wide be attributed to a parallel H2O2 decomposition range of applications including chemical sensors, reaction [15]. As it is well known, H2O2 is biosensors, chemical synthesis, batteries and fuel unstable with respect to disproportionation and it cells. Further optimizations of the DBSA/KCl decomposes to oxygen and water rapidly in the silver-based catalyst including silver layer presence of catalysts, as follows [27]: homogenization on GDL surfaces and catalyst formation and deposition will be performed in 2H2O2 → O2 + 2H2O (7) order to improve its applicability in the area of energy generation and fuel cells. Some authors in the literature have reported the effect of H2O2 decomposition during the electrochemical reduction of H2O2. Welch et al. 4 CONCLUSION [23] ascribed the shoulder exhibited at – 0.7 V The application of DBSA/KCl modified silver vs. SCE (approx. – 0.66 V vs. Ag/AgCl) in the paste has been demonstrated as a suitable reduction wave to the electron-reduction of electrocatalyst for a H2O2 fuel cell. Physical oxygen, produced via the silver-catalysed analysis of the DBSA/KCl modified Ag-GDL decomposition of H2O2. Guascito et al. [28] substrates showed modifications to both the observed a new reduction process between 0 and surface morphology and chemical composition – 0.1 V vs. SCE for H2O2 concentrations larger of the silver paste electrode surface. than 2.5·10-3 M when H2O2 was reduced on a Pt Preliminary fuel cell results have shown that an electrode modified with a polyvinyl alcohol initial open circuit voltage approaching 1 V and 299

Proceedings of SEEP2015, 11-14 August 2015, Paisley peroxide in alkaline solutions," Electrochemistry current densities in the range of 5 mA/cm (at Communications, vol. 1, pp. 614-617, 1999. 0.3 V) could be achieved. Although these current [10] D. N. Prater and J. J. Rusek, "Energy densities are relatively low when compared to density of a methanol/hydrogen-peroxide fuel optimised Pt cathode PEM fuel cells, it does cell," Applied Energy, vol. 74, pp. 135-140, hold promise for future developments of an 2003/2// 2003. inexpensive fuel cell with low platinum metal [11] D. N. Prater and J. J. Rusek, " Systematic loading. examination of a direct methanol - Hydrogen peroxide fuel cell. ," in Proceedings of the First ACKNOWLEDGMENT International Conference on Green Propellants The authors would like to acknowledge the for Space Propulsion, European Space Agency, financial assistance of EU, FP7/2007-2013, Noordwijk, The Netherlands, 2001. under grant number 257372. [12] S. Fukuzumi, Y. Yamada, and K. D. Karlin, "Hydrogen peroxide as a sustainable energy REFERENCES carrier: Electrocatalytic production of hydrogen [1] R. O'Hayre, S. Cha, W. Colella, and F. B. peroxide and the fuel cell," Electrochimica Acta, Prinz, Fuel Cells Fundamentals. USA: John vol. 82, pp. 493-511, 2012. Wiley & sons, 2006. [13] S. Hasegawa, K. Shimotani, K. Kishi, and [2] M. S. Basualdo, D. Feroldi, and R. Outbib, H. Watanabe, "Electricity Generation from PEM Fuel Cells with Bio-Ethanol Processor Decomposition of Hydrogen Peroxide," Systems. A Multidisciplinary Study of Modelling, Electrochemical and Solid-State Letters, vol. 8, Simulation, Fault Diagnosis and Advanced pp. A119-A121, February 1, 2005 2005. Control. London: Springer, 2012. [14] S. P. Kumaraguru, N. Subramanian, H. [3] S. Litster and G. McLean, "PEM fuel cell Colon-Mercado, K. Hansung, and B. N. Popov, electrodes," Journal of Power Sources, vol. 130, "Novel non precious metal catalyst for PEMFC pp. 61-76, 2004. applications," in Joint International Meeting [4] M. Grujicic and K. M. Chittajallu, "Design 206th Meeting of the Electrochemical and optimization of polymer electrolyte society/2004 Fall Meeting of the membrane (PEM) fuel cells," Applied Surface Electrochemical Society of Japan Honolulu, HI, Science, vol. 227, pp. 56-72, 2004. United States, 2004. [5] S. Srinivasan, "Fuel Cells for Extraterrestrial [15] T. I. Valdez, S. R. Narayanan, C. Lewis, and Terrestrial Applications," Journal of The and W. Chun, "Hydrogen peroxide oxidant fuel Electrochemical Society, vol. 136, pp. 41C-48C, cell systems for ultraportable applications," in February 1, 1989 1989. Proceedings of the Electrochemical Society, [6] E. J. Lamas and P. B. Balbuena, "Oxygen Pennington, NJ, USA, 2001, pp. 265-273. Reduction on Pd0.75Co0.25 (111) and [16] A. Woerner and N. N. Lloyd, "Fuel cells for Pt0.75Co0.25 (111) Surfaces:â€‰ An ab Initio underwater propulsion applications," in Comparative Study," Journal of Chemical Proceedings of the 30th Power Sources Theory and Computation, vol. 2, pp. 1388-1394, symposium, Atlantic City, NJ, USA, 1982. 2014/07/21 2006. [17] N. Luo, G. H. Miley, R. J. Gimlin, R. L. [7] D.-H. Lim and J. Wilcox, "Mechanisms of Burton, J. Rusek, and F. Holcomb, "Hydrogenthe Oxygen Reduction Reaction on Defective Peroxide-Based Fuel Cells for Space Power Graphene-Supported Pt Nanoparticles from Systems," Journal of Propulsion and Power, vol. First-Principles," The Journal of Physical 24, pp. 583-589, 2014/10/06 2008. Chemistry C, vol. 116, pp. 3653-3660, [18] E. Kjeang, A. G. Brolo, D. A. Harrington, 2014/07/21 2012. N. Djilali, and D. Sinton, "Hydrogen Peroxide as [8] H. Falcon and R. E. Carbonio, "Study of the an Oxidant for Microfluidic Fuel Cells," Journal heterogeneous decomposition of hydrogen of The Electrochemical Society, vol. 154, pp. peroxide: its application to the development of B1220-B1226, December 1, 2007 2007. catalysts for carbon-based oxygen cathodes," [19] L. Gonzalez-Macia, M. R. Smyth, A. Journal of Electroanalytical Chemistry, vol. 339, Morrin, and A. J. Killard, "Enhanced pp. 69-83, 1992. electrochemical reduction of hydrogen peroxide [9] R. Venkatachalapathy, G. P. Davila, and J. on silver paste electrodes modified with Prakash, "Catalytic decomposition of hydrogen 2

300

Proceedings of SEEP2015, 11-14 August 2015, Paisley surfactant and salt," Electrochimica Acta, vol. 56, pp. 4146-4153, 2011. [20] L. Gonzalez-Macia, M. R. Smyth, and A. J. Killard, "A Printed Electrocatalyst for Hydrogen Peroxide Reduction," Electroanalysis, vol. 24, pp. 609-614, 2012. [21] J. G. Carton and A. G. Olabi, "Design of experiment study of the parameters that affect performance of three flow plate configurations of a proton exchange membrane fuel cell," Energy, vol. 35, pp. 2796-2806, 2010. [22] L. Cindrella, A. M. Kannan, J. F. Lin, K. Saminathan, Y. Ho, C. W. Lin, and J. Wertz, "Gas diffusion layer for proton exchange membrane fuel cellsâ€”A review," Journal of Power Sources, vol. 194, pp. 146-160, 2009. [23] C. M. Welch, C. E. Banks, A. O. Simm, and R. G. Compton, "Silver nanoparticle assemblies supported on glassy-carbon electrodes for the electro-analytical detection of hydrogen peroxide," Analytical and Bioanalytical Chemistry, vol. 382, pp. 12-21, 2005. [24] M. Honda, T. Kodera, and H. Kita, "Electrochemical behavior of H2O2 at Ag in HClO4 aqueous solution," Electrochimica Acta, vol. 31, pp. 377-383, 1986. [25] G. Flatgen, S. Wasle, M. Lubke, C. Eickes, G. Radhakrishnan, K. Doblhofer, and G. Ertl, "Autocatalytic mechanism of H2O2 reduction on Ag electrodes in acidic electrolyte: experiments and simulations," Electrochimica Acta, vol. 44, pp. 4499-4506, 1999. [26] F. Barbir, PEM Fuel Cells: Theory and Practice, 2nd Edition ed.: Academic Press Elsevier, 2013. [27] P. W. Atkins, T. Overton, J. Rourke, M. Weller, and F. Armstrong, Shriver & Atkins Inorganic Chemistry, 4th ed. New York: Oxford University Press, 2006. [28] M. R. Guascito, E. Filippo, C. Malitesta, D. Manno, A. Serra, and A. Turco, "A new amperometric nanostructured sensor for the analytical determination of hydrogen peroxide," Biosensors and Bioelectronics, vol. 24, pp. 1057-1063, 2008.

301


MULTIPLE REGRESSION ANALYSIS IN THE DEVELOPMENT OF NIFE CELLS AS ENERGY STORAGE SOLUTIONS FOR INTERMITTENT POWER SOURCES SUCH AS WIND OR SOLAR Jorge Omar Gil Posada, Abdallah H. Abdalla, Charles I. Oseghale and Peter J. Hall Chemical and Biological Engineering, University of Sheffield, Sir Robert Hadfield Building, Mappin Street, Sheffield S1 3JD, England, UK Abstract Multiple regression analysis was used to investigate the effect of bismuth sulphide and iron sulphide as anode additives for NiFe cells. With this in mind, in-house made Fe/FeS/Bi2S3 based electrodes were cycled against commercially available nickel electrodes. A simplex centroid design was used to investigate the combined effects of any of the aforementioned additives on cell performance. The manuscript ends with an initial look at electrolyte systems as a means to further improve the performance of our cells. Finally, our findings support the idea that HS- ions improve the overall performance of NiFe cells. Keywords: NiFe, electrolyte decomposition, cell performance, hydrogen evolution 1 INTRODUCTION Although, there is a continuously increasing demand of energy coming from renewable sources; the intermittent nature of these resources restrict their use (temporary wind profiles, availability of sun light, sufficient/constant supply of water, etc); so energy generation and demand are not easily matched [1-3]. Nonetheless, we are facing a transition to integrate an increasing share of energy coming from renewable sources to balance the electric grid [4-10]. Three different strategies have been proposed to tackle the aforementioned problem: energy storage, transmission and full back up capacity (for example by using fossil fuels); this manuscript looks at energy storage as a means to overcome this issue.

energy which must be reverted back into electricity for further use. This process is not 100% efficient and prices are still very high. Compared with our current technologies, a more practical way to store large amounts of energy is very much needed. This is because, unfortunately, modern batteries would utilize:

Among all renewable sources, wind power is, undoubtedly, one the world’s fastest growing technologies [3]. It is well known that offshore wind is stronger and steadier than its onshore counterpart, so offshore wind farms could convert larger amounts of wind energy into more useful forms of energy (such as electricity) [8, 11].

NiFe cells are rechargeable aqueous batteries that were successfully commercialized by Edison back in the early 20th century. Although, this technology was superseded by cheaper (and more toxic) lead-acid cells, there is a renewed interest on them, arising from their environmental friendliness, low cost of raw materials, long life and tolerance to electrical abuse (such as overcharge, over-discharge, being idle for extended periods and short-circuit conditions) and compatibility with intermittent power sources, such as wind power and photovoltaic’s (PV’s). In addition, this technology

Unfortunately, large scale energy storage is still very expensive and inefficient. Broadly speaking, energy storage demands electricity to be converted into some non-electrical form of 302

   

organic/flammable electrolyte systems ultrapure and/or non-abundant (costly) reactants environmentally unfriendly raw materials and components costly nano-structuring procedures

These aspects will increase the price of the battery and would polarize general opinion against the intended solution.

Proceedings of SEEP2015, 11-14 August 2015, Paisley The prevention of electrolyte decomposition has would be suitable for relatively low specific been achieved by modifying the iron electrode energy applications (30-50 Wh kg-1) [12]. by either nano-structuring the anode or by the addition of elements (such as sulphur or The low cost and abundance of the raw materials bismuth) that would increase the overpotential required to produce NiFe cells are two important for hydrogen evolution [26, 27]. Electrolyte aspects that further encourage their use. Iron is modification also permits preventing Eq. (2) the fourth most abundant element in the Earth’s from happening. In fact, different electrolyte crust [13, 14]. Nickel less abundant than iron, additives such as wetting agents [28], long chain but it is still relatively abundant [15, 16]. Other thiols [29], organic acids [30], etc., have been materials/compounds such as potassium investigated for such end [26]. With this in hydroxide, sulphur and iron sulphide are also mind, NiFe batteries with exceptional capacities very abundant [17-20]. Bismuth is relatively of nearly 800 mAh g-1 have been reported [31, scarce [21, 22] but only small amounts of it are 32]. Although, highly efficient, these batteries required to produce NiFe cells. Basically, there require costly reactants and nano-structuring are not good reasons to foresee a shortage of any techniques. These aspects would certainly of the aforementioned elements any time soon. influence the final price of the battery [31, 32]. There are, of course, many challenges In our previous publications, the role of selected preventing a large scale utilization of NiFe cells electrode additives (such as bismuth and iron for offshore wind applications, such as low cell sulphide) on battery performance was explored efficiency, electrolyte decomposition (hydrogen [27, 33]. Our experimental observations evolution), low energy and power densities [23, suggested that bismuth sulphide and iron 24]. Broadly speaking, the major challenge sulphide effectively improve the performance of preventing this technology from claiming its the NiFe cell. However, the remaining question righteous place as a large scale energy storage is whether a synergistic effect could be found solution is the low performance of the iron between them. Basically, most manuscripts on electrode, which is strongly related to the NiFe cell technology would consider one single evolution of hydrogen. electrode additive at a time, thus neglecting the possibility that the combined effect of the Under strong alkaline conditions, the main additives would be greater than the sum of the process taking place during the charging of an parts. This manuscript ends with an initial look iron electrode is the reduction of ferrous ions at electrolyte selection and improvement. (Fe2+) to metallic iron (Fe0); conversely, during discharge, metallic iron is oxidized to ferrous ions. Eq. (1) illustrates the charging and 2 EXPERIMENTAL discharging (forward and backward reactions By using a similar procedure as the one respectively) processes of an iron electrode described in our previous publications [11, 27], under alkaline conditions [24, 25]. iron based anodes were produced by using Fe(OH ) 2  2e   Fe  2OH 

E 0   0.87V (1)

Unfortunately, during the charging of the iron electrode, water is decomposed to yield hydrogen. Because of that, part of the energy that is intended to be stored in the battery, ended up wasted in the parasitic evolution of hydrogen. Hydrogen evolution, therefore, accounts for a dramatic reduction in the overall performance of the battery, as indicated by Eq. (2). 2H 2 O  2e   H 2  2OH 

E 0   0.83V (2)

varying amounts of Fe, PTFE, FeS and Bi2S3. Essentially, strips of nickel foam (10 mm × 40 mm × 1.8 mm) were coated and then vacuum dried until approximately 0.2 – 0.25 g of iron powder were loaded on an area of approximately 1 cm2. The chemicals and materials used were of the following specifications.   

303

Iron powder (purity 99.5%, < 10μm) from Alfa Aesar Iron sulphide (purity 99.5%) from Sigma Aldrich Bismuth sulphide (purity 99.5%, < 5μm) from Sigma Aldrich

Proceedings of SEEP2015, 11-14 August 2015, Paisley System Composition  PTFE (Teflon 30-N, 59.95% solids) from Alfa Aesar S1 5.1 M KOH S2 5.1 M KOH + 0.1M LiOH  Nickel foam (purity 99.0%, density 350 2 g/m ) from Sigma Aldrich S3 5.1 M KOH + 0.1M K2S S4 5.1 M KOH + 0.1M LiOH+ 0.1M K2S  Potassium hydroxide (purity ≥ 85.0%, pellets) from Sigma Aldrich The electrode and electrolyte formulations were  Potassium sulphide (purity ≥ 99.5%) tested on a three electrode cell. In-house made from Sigma Aldrich anodes were tested in a three-electrode cell.  Lithium hydroxide (purity ≥ 98.0%) from Nickel electrodes, obtained from a commercial Sigma Aldrich nickel iron battery, where used as cathodes. All potentials were measured against a In house deionized water was produced by using mercury/mercury oxide (Hg/HgO) reference an Elix 10-Milli-Q Plus water purification electrode (E0MMO = + 0.098 V vs. NHE). Cells system (Millipore, Eschborn, Germany). were cycled on a 64 channel Arbin SCTS. Basically, varying amounts of each electrolyte components were dissolved in our in-house Experiments of charge and discharge were produced deionized water. The electrolyte conducted under galvanostatic conditions at system used for electrode development consisted room temperature until the steady state was of an aqueous solution of 5.1 M KOH. reached. Cells were cycled from 0.6 to 1.4 V vs. Hg/HgO at a C/5 rate. Formation and Essentially, we have kept the composition of stabilization of the electrodes were found to be PTFE at a constant value of approximately 6%w. th complete by the 50 cycle of charge and Based on the constancy of the binder and our discharge [27, 33]. Fig. 1 provides a sketch of desire to explore the composition space defined the cell test configuration. by formulations not exceeding 15%w on each additive, Table 1 was constructed. Table 1. Experimental determinations of factors and levels (free-PTFE basis). Factor Concentrations (%w) Low High Fe 70 100 FeS 0 15 Bi2S3 0 15 By using six replicates per electrode formulation and the two standard deviation criteria for rejection, a surface response model that allows the finding of electrode formulations that reduce electrolyte decomposition was obtained. Broadly speaking, electrolyte systems for NiFe cells are concentrated solutions of potassium hydroxide and other minor constituents, we decided to keep the concentration of KOH constant (5.1 M), and to explore the effect of both lithium hydroxide and potassium sulphide (six replicates per electrolyte formulation were used). Table 2 reports electrolyte systems under consideration. Table 2. consideration

Electrolyte

systems

under

304

Figure 1. Test cell configuration.

3


3.1 Electrode formulation It has been recognised that any NiFe cell requires a relatively long conditioning period (in the order of 30 cycles of charge and discharge)

Proceedings of SEEP2015, 11-14 August 2015, Paisley before it reaches the steady state [11, 27, 33]. In order to determine whether a relationship Fig. 2 confirms this observation. existed between the factors and responses (coulombic efficiency and electrode additives), polynomial functions, as the one represented by Eq. (3) were used

 Q    i , j Yi j  (3) n

m

i 1 j 0

Where ηQ represents coulombic efficiency, the α terms are the expansion coefficients, the Y terms represent the weigh per-cent of each component, n the number of component in the electrode formulation and m is the order of the polynomial. Figure 2. Galvanostatic charge and discharge profile for a NiFe cell (sample G from Table 3, electrolyte system S1 from Table 2) versus mercury/mercury oxide (Hg/HgO) reference electrode. Based upon Table 1, and by using the mixing rules in a three dimensional concentration space, a simplex centroid design based on a conventional central composite design was proposed. Table 3 reports experimental values of coulombic efficiency calculated for our electrodes developed by considering Table 1 and utilizing electrolyte system S1. It is important to note that collected data exhibit large variability so a relatively large number of replicates (six in this case) were used to increase the statistical force of the analysis. With this in mind, any sample whose coulombic efficiency lays more than two standard deviations from the mean was rejected. Table 3. Experimental design matrix (PTFE-free basis, electrolyte system S4, 50th cycle). Cell %wFeS %wBi2S3 ɳQExp ɳQModel A 7.5 15.0 43.5 ± 2.3 43.2 B 7.5 7.5 47.0 ± 1.2 46.6 C 7.5 0.0 42.9 ± 2.4 42.2 D 0.0 7.5 28.5 ± 2.3 26.9 E 15.0 7.5 45.7 ± 1.8 45.8 F 12.8 12.8 48.3 ± 3.0 46.5 G 12.8 2.2 46.1 ± 2.3 45.9 H 2.2 12.8 32.6 ± 1.9 33.2 I 2.2 2.2 32.4 ± 1.9 32.5

305

We began by investigating the simplest forms of Eq. (3) until the predicted by the model coulombic efficiencies closely mirrored the experimental values (results not shown). The final model is represented by Eq. (4): ηQ = 22.514 + 3.996YFeS + 1.117 YBi2s3 – 0.183 Y2FeS – 0.0702 Y2Bi2s3 (4) The regression analysis reveals not only a relatively high multiple correlation coefficient (r2 = 0.9533), but also a highly significant model (F statistic 250.2), and as expected, all terms from the model are significant. Finally, no evidence against normality was found. Note that at first glance, Eq. (4) seems to ignore the composition of iron in the electrode formulation. But this observation is not correct, for all components within each formulation must add up to 100%. The model establishes that iron, as an electrode component (factor) is not significant. This can be rationalised in terms of its dominance within the formulations, for iron accounts for at least 70% of the electroactive material. It is important to recognise that although, large variability on cell performance was always noted, the second order model given by Eq. (4) renders a relatively good fit, and this is because by increasing the number of replicates, the likelihood of having used true values is enhanced. Fig. 3 provides a three dimensional representation of Eq. 4.

Proceedings of SEEP2015, 11-14 August 2015, Paisley Fig. 4 compares electrolyte formulations (from Table 2) based on groups of 6 replicates per formulation. Each group corresponds to a different box; each horizontal line within the box represents the mean coulombic efficiency for that formulation. Likewise, whiskers indicate extreme values and empty circles outliers (data that lays more than two standard deviations from the mean).

Figure 3. Second order efficiency representation

coulombic

We are not going to give a presentation on how to find the maximum value of a differentiable function subject to a constraint (such as %Bi2S3 + %FeS + %Fe = 100). The details of such procedure can be found in most books of calculus, and this subject is out of the scope of this manuscript. The final electrode formulation that reduces electrolyte decomposition was found to be 10.3% FeS + 7.5% Bi2S3 + 76.5% Fe + 5.7% PTFE, and was denoted by formulation M. 3.2 Electrolyte system Now that we have developed an electrode formulation that renders low electrolyte decomposition, the next step is to investigate which electrolyte system would further reduce hydrogen evolution, thus increasing cell performance. We shall use electrode formulation M. Basically, electrolyte systems for NiFe cells, consist of aqueous solutions of concentrated KOH (in the order of 5.0M) and lithium hydroxide. It has been reported that the soluble bisulfide anion (HS-) would increase the performance of the NiFe cells by mitigating electrolyte decomposition; however, our previous experimental results have shown that this is not entirely correct, for at low concentrations, the performance of the cell seems to be independent of the presence of this additive [11, 33], but this aspect require further investigation. In order to do that, we are going to compare electrolyte formulations and then we will find out whether or not meaningful differences in battery performance exist.

306

Figure 4. Battery performance as a function of electrolyte formulation. For information on electrolyte system composition, please refer to Table 2. A visual inspection of Figure 4 reveals that electrolyte formulations based on potassium hydroxide (electrolyte system S1) are outperformed by any of the remaining electrolyte systems. It seems clear that lithium hydroxide increases the performance of the cells. Likewise, potassium sulphide clearly increases the performance of the NiFe cells. Although, these conclusions seem to be logical, we require a more formal way of drawing conclusions. In this case, this can be achieved by using post-hoc comparisons. In particular, we shall use the Tukey’s HSD to find out whether meaningful differences across electrolyte formulations exist. To begin with, we must check that our data is normally distributed; a residuals analysis indicates no evidence against normality. As

Proceedings of SEEP2015, 11-14 August 2015, Paisley The practical implications of the fact that there shown in Fig. 5, data are randomly dispersed and are meaningful differences neither between a linear regression model is appropriate for the formulations S1 and S2 nor between S3 and S4 data. are tremendous, for they imply that lithium hydroxide does not enhance the performance of the battery (well, not at least under our experimental conditions). It is important to mention here that it has been suggested that lithium hydroxide would enhance the cycling life of the battery; this claim, of course, would require long testing and it is propose as a future work.

Figure 5. Normal probability plot for coulombic efficiency residuals. The same conclusion can be drawn by using more rigorous normality tests, such as the Shapiro-Wilk (SK) test, which is a nonparametric test for normality. In this case the SK test consolidates the lack of evidence against normality, as all p-values are significant (data not shown).

Unlike lithium hydroxide, the Tukey-HSD reveals the use of potassium hydroxide does indeed enhance battery performance. From this statement, it necessarily follows that soluble species coming from the aforementioned reactant would be responsible for the prevention of electrolyte decomposition and thus the evolution of hydrogen. It has been proposed that the soluble bisulphite anion (HS-) is indeed responsible for such behaviour. And our experimental results are in-line with such claims.

The Tukey HSD test reveals no meaningful differences among electrolyte systems S1 and S2 or between S3 and S4. The results are shown in Fig. 6.

Figure 7. Battery performance versus cycling number for NiFe cell formulation (electrolyte system S4 and anode formulation M). Fig. 7 confirms the tendency described in Fig. 2, in the sense that battery performance tends to increase with the cycle number during the conditioning period (first 20-30 cycles). Moreover, battery performance tends to increase in a rather unpredictable manner; however, in the long run, the performance of the battery would increase until the steady state was reached at the end of the conditioning period. The

Figure 6. 95% family-wise confidence level

307

Proceedings of SEEP2015, 11-14 August 2015, Paisley incidence on the performance of the battery; in authors believe this is because electrodes tend to other words, the presence of potassium sulphide fall apart until they reached a configuration in the electrolyte increases coulombic efficiency. where they can handle the current requirements This conclusion also supports the idea that the of the charge/discharge test without major soluble bisulfide anion (HS-) do improve the structural changes. overall performance of NiFe cells. Finally, a large specific charge storage capacity From our experimental findings, we can close to 290 mAh g-1 was observed. Although, -1 conclude there is a link between electrode larger capacities (close to 800 mAh g ) have performance (coulombic efficiency) and: been reported [31, 34-36], our manufacturing process is a more cost-effective solution, for it only utilises commercial grade reactants, which  conditioning of the iron electrode makes it ideal for grid-scale energy storage (development of new surface area) applications.  electrode composition  electrolyte composition (presence of The renewal of metal surface area on iron potassium sulphide) materials has been identified as a key parameter in the evolution of hydrogen under strong The data gathered during this project is subject alkaline conditions [37, 38]. We believe, to large variability; therefore, a relatively large therefore, that due to the breaking up of the number of replicates (6 in total) and the two electrode, new electrode surface area is available standard deviation criteria for rejection were for the electrochemical reactions to take place. used to increase the statistical force of the analysis. 4 CONCLUSIONS AND FUTURE WORK By pursuing the development of cost effective offshore wind energy storage solutions, we have achieved NiFe cells that render coulombic efficiencies in the order of 49% and capacities in the order of 290 mAh g-1. These results are very promising as we have used neither ultra-pure reactants, nor we have nano-structured the electrode, so our batteries hold a promise for a real cost effective solution to store large amounts of energy coming from intermittent sources such as wind power. The experimental approach used and explained in this manuscript has been successful in facilitating the development and improvement of secondary NiFe cells, by reducing electrolyte decomposition and thus hydrogen evolution. The formulation developed corresponds to an iron electrode formulation consisting of 10.3% FeS + 7.5% Bi2S3 + 76.5% Fe + 5.7% PTFE. It was also found that the aqueous electrolyte 5.1 M KOH + 0.1M LiOH+ 0.1M K2S further reduced the evolution of hydrogen, thus improving the overall performance of the aqueous NiFe cell. It was found that under our experimental conditions, potassium sulphide has a real 308

Our experimental findings would suggest that at the level of confidence α = 0.05, potassium sulphide, as electrolyte additive, does enhance the performance of the battery. Finally, the evidence for lithium hydroxide supports the idea that this additive only marginally improves cell performance. However, it has been reported this additive, in the long run, would enhance cell life and performance, so extended testing is recommended as a future work.

ACKNOWLEDGEMENTS The authors would like to acknowledge the U.K. Engineering and Physical Sciences Research Council for supporting this work (EP/K000292/1; SPECIFIC Tranche 1: Buildings as Power Stations).

REFERENCES [1] K. Hedegaard, P. Meibom, Renewable Energy, 37 (2012) 318-324. [2] H. Chen, T.N. Cong, W. Yang, C. Tan, Y. Li, Y. Ding, Progress in Natural Science, 19 (2009) 291-312.

Proceedings of SEEP2015, 11-14 August 2015, Paisley [24] G. Halpert, Journal of Power Sources, 12 [3] A. Jacob, Reinforced Plastics, 45, (1984) 177-192. Supplement 1 (2001) 10-13. [25] A.K. Shukla, S. Venugopalan, B. [4] P. Seljom, A. Tomasgard, Energy Hariprakash, Journal of Power Sources, 100 Economics, 49 (2015) 157-167. (2001) 125-148. [5] M. Carrasco-Díaz, D. Rivas, M. Orozco[26] A.K. Manohar, C. Yang, S. Malkhandi, B. Contreras, O. Sánchez-Montante, Renewable Yang, G.K.S. Prakash, S.R. Narayanan, Journal Energy, 78 (2015) 295-305. of the Electrochemical Society, 159 (2012) [6] J. Waewsak, M. Landry, Y. Gagnon, A2148-A2155. Renewable Energy, 81 (2015) 609-626. [27] J.O. Gil Posada, P.J. Hall, Journal of Power [7] J. Haas, M.A. Olivares, R. Palma-Behnke, Sources, 262 (2014) 263-269. Journal of Environmental Management, 154 [28] A.K. Manohar, C. Yang, S. Malkhandi, (2015) 183-189. G.K.S. Prakash, S.R. Narayanan, Journal of the [8] M. De Prada Gil, J.L. Domínguez-García, F. Electrochemical Society, 160 (2013) A2078Díaz-González, M. Aragüés-Peñalba, O. GomisA2084. Bellmunt, Renewable Energy, 78 (2015) 467[29] S. Malkhandi, B. Yang, A.K. Manohar, 477. G.K.S. Prakash, S.R. Narayanan, Journal of the [9] T.R. Ayodele, A.S.O. Ogunjuyigbe, American Chemical Society, 135 (2012) 347Renewable and Sustainable Energy Reviews, 44 353. (2015) 447-456. [30] M.J. Mackenzie Jr., Salkind, Alvin J., in, [10] S. Sun, F. Liu, S. Xue, M. Zeng, F. Zeng, 1969. Renewable and Sustainable Energy Reviews, 45 [31] H. Wang, Y. Liang, M. Gong, Y. Li, W. (2015) 589-599. Chang, T. Mefford, J. Zhou, J. Wang, T. Regier, [11] J.O.G. Posada, P.J. Hall, Sustainable F. Wei, H. Dai, Nature Communications, 3 Energy Technologies and Assessments. (2012). [12] P. Gao, Y. Liu, W. Lv, R. Zhang, W. Liu, [32] A.K. Manohar, S. Malkhandi, B. Yang, C. X. Bu, G. Li, L. Lei, Journal of Power Sources, Yang, G.K.S. Prakash, S.R. Narayanan, Journal 265 (2014) 192-200. of the Electrochemical Society, 159 (2012) [13] F. Birch, Journal of Geophysical Research, A1209-A1214. 69 (1964) 4377-4388. [33] J.O. Gil Posada, P.J. Hall, Journal of Power [14] G. Morard, J. Siebert, D. Andrault, N. Sources, 268 (2014) 810-815. Guignot, G. Garbarino, F. Guyot, D. [34] P. Periasamy, B. Ramesh Babu, S. Antonangeli, Earth and Planetary Science Venkatakrishna Iyer, Journal of Power Sources, Letters, 373 (2013) 169-178. 62 (1996) 9-14. [15] A. Oxley, N. Barcza, Minerals Engineering, [35] Z. Liu, S.W. Tay, X. Li, Chemical 54 (2013) 2-13. Communications, 47 (2011) 12473-12475. [16] C.-L. Huang, J. Vause, H.-W. Ma, Y. Li, [36] M.K. Ravikumar, T.S. Balasubramanian, C.-P. Yu, Journal of Cleaner Production, 84 A.K. Shukla, Journal of Power Sources, 56 (2014) 450-458. (1995) 209-212. [17] J.P. Lorand, Geochimica et Cosmochimica [37] I. Flis-Kabulska, Electrochemistry Acta, 54 (1990) 1487-1492. Communications, 11 (2009) 54-56. [18] J.P. Lorand, Earth and Planetary Science [38] R. Solmaz, G. Kardaş, Electrochimica Acta, Letters, 93 (1989) 50-64. 54 (2009) 3726-3734. [19] J.P. Lorand, Earth and Planetary Science Letters, 119 (1993) 627-634. [20] D.H. Eggler, J.P. Lorand, Geochimica et Cosmochimica Acta, 57 (1993) 2213-2222. [21] H. Braunschweig, P. Cogswell, K. Schwab, Coordination Chemistry Reviews, 255 (2011) 101-117. [22] N.M. Leonard, L.C. Wieland, R.S. Mohan, Tetrahedron, 58 (2002) 8373-8397. [23] A. Chaurey, S. Deambi, Renewable Energy, 2 (1992) 227-235.

309


SOLAR POWER PLANT WITH HYDROGEN STORAGE Rubal Sambi, A. Alaswad, J. Mooney and A. G. Olabi Institute of Engineering and Energy Technologies, Scholl of Engineering and Computing, University of the West of Scotland, Paisley, United Kingdom; email: [email protected] ABSTRACT Due to the continues use of finite energy resources and their environmentally disastrous nature, the need for equally stable environmentally friendly power generation techniques are required, Currently PV solar panels and wind turbines are leading the way for renewable power generation, however the major problem which hinders the PV solar panel technology is that it suffers from fluctuating power generation as it only generates power when the sun shines. The wind turbine also shares a fundamental problem as its power generation fluctuates as it is dependent on the sporadic wind current. By combining a PV solar panel and a wind turbine, it would allow power generation to operate day and night, additionally by introducing a storage method, a final system would be created which would tackle the power fluctuation and offer a continuous power supply. By using engineering software and experimental studies, the proposed project will model and optimise a solar power plant with hydrogen storage which has the aim of supplying power demands throughout the year. The optimisation will be carried out and the focus on the power output and cost output allowing the plants maximum power generation at its lowest cost to will be formulated. Keywords: renewable energy, wind turbines, solar, standalone 1

due to technological advancements this need for

INTRODUCTION

power has been increasing at an exponential rate. The world has been facing a dangerous economic and environmental problem for the last years. Fossil fuels are running out and the environment has been devastated. The UK has been using finite fossil fuel energy production such as coal, gas, oil for many years. As a result of generating power from plants and using these hydrocarbon fuels for necessities such as industry and transportation, toxic emissions known as greenhouse gases have caused detrimental environmental changes. Alarmingly 310

Due to the dangerous environmental problem and the finite nature of these fuels the world has been striving to find alternative means of generating pollution free, renewable energy which will be readily useable for commercial use. Since the last few decades much research and development has gone into optimising technologies such tidal energy production, wind turbine energy production, solar panels and bio fuel. However this project will focus on the

Proceedings of SEEP2015, 11-14 August 2015, Paisley an optimal design for the solar energy plant were optimization, design of PV solar panels and later different variables will be optimized and the aim

on will look into the wind turbines.

is to reduce the cost and increase the overall The PV solar panel is a device which uses the

output.

sun’s rays and converts them into a usable current this process is known as photovoltaic,

2

SOLUTION

this method generates electrical power by using semiconducting materials. In essence the solar panel is a set of solar photovoltaic (PV) modules electrically connected and mounted on a

Renewable

energy

replacement

of

is

fast

fossil

fuels

becoming and

the

carbon

emissions. Currently wind generation and biomass are the most used renewable enrgy

supporting structure.

resources. Unfortunately there is a problem with A wind turbine is a device that converts kinetic energy from the wind into electrical power, it uses wind currents to turn a blades which in turn creates electricity by use of electrical generators

renewable energy on the large scale. Wind is not predictable, it can have direct effect of the generation capabilities of a wind farm. Nuclear power could also be an option however disposing of hazardous waste is a major problem

and gearboxes.

too (1). The problem with these two renewable energy production

methods

is

that

there

energy

production tends to fluctuate due to weather and orbital changes, hence electrolysis will be used to convert the electrical current into a storable hydrogen gas, and this gas will be securely compressed and stored into a gas storage tank. Lastly this hydrogen gas will finally be converted back onto an electrical current when the load requires it, by use of a technology known as a hydrogen turbine, which converts the

It has been seen that fossil fuels are finite and cause much pollution, however it is a necessity at

the

moment

manufacturing

to

allow

process,

for

the

new

however

as

new

technology emerge; wind farms, biomass and solar

panels

unsustainable

will fuels.

eventually New

replace

technological

advancement are emerging such as the hybrid technology, to tackle the power needs of the population, hence this project will focus on the generating electricity from solar system and looking onto different storage systems.

chemical energy from a gas into electricity. This entire system will be a standalone system which will be used to supply power.

The next stage of this project is the experiment and optimisation phase this will make use of engineering software and experiments to develop 311

the

Proceedings of SEEP2015, 11-14 August 2015, Paisley Solar cells consist of Silicon wafers which are doped with an additive to form a positive and negative electric field on the opposite sides. When lights shines on the cell it creates an electric field across the layers, the more electricity is produced depending on the stronger sunlight, atoms let go of their electrons, these electrons then flow through a circuit. Fig1: Operation diagram of a Hybrid system PV/WD (9) 3

PV SOLAR PANEL

By use of experiment the photovoltaic effect was first demonstrated by Edmond Becquerel, later in 1905 Albert Einstein went into detail and explained how the mechanism of light instigated carrier excitation works, in 1921 for his work he won the Nobel Prize in physics. In 1958, solar cells became more prominent when they were proposed for satellite named vanguard I satellite. Solar cells began to sell well thereafter, improvements to the design also continued, hence costs began to drop allowing affordability. The semiconductor industry continued to boom

Fig 2: PV cell. (10)

allowing further declines in price eventually

The level of current and voltage produced by the

reaching $0.62/watt in 2012. (2)

conversion is too small for any useful purpose however when the solar cell are connected in a

The PV solar panel is a device used to collect

serial and parallel which is known as a solar

energy so that it can convert it into electrical

module, 12 to 40 V can be produced. As the

current. This panel consists of a series of an

electricity produced is directly proportional to

array of PV module. They are made up of a

the amount of light energy that falls on the

semiconductor which exhibits the photovoltaic

panel, solar modules are also connected to form

effect, light directly is converted to current as

solar array to collect more photons and hence

electrons are allowed to flow in a closed circuit

produce more electricity.

when materials such as semiconductors absorb light photons and electrons are released. 312

Proceedings of SEEP2015, 11-14 August 2015, Paisley There is much research which is being done to 4 ELECTROLYSER optimise the efficiency of PV solar panels technology,

currently

technology is

being

multifunction experimented

cell on

as

currently in the single cell junction technology the lower energy photons are not utilised as their energy is not greater than the band gap to create an electric field.

The electrolyser which will be used for this set up is a piece of equipment which will convert electrical energy into chemical energy. The electrolyser will make use of water also known as

, where it will extract hydrogen and

oxygen, so that it can be stored and then used at

The way to solve this problem is to use multi-

a alater date..

junction where materials which have different band gaps are used so that a reaction to wider range of energy levels of photons can occur. Multi-junction cell can achieve an efficiency of 35%. Examples of common solar panels which are available in the market are polycrystalline, monocrystalline, amorphous or thin film and hybrid silicon. These panels have their own application, with advantages and disadvantages (4). As the area of the solar panel is directly proportional to the generation of power this is a crucial factor when deciding to use them when designing a power plant, as polycrysatal are less efficient as they require more area to produce electricity then they are of the monocrystalline which produce more electricity with much less area however monocrystalline cost more, but they might worth buying due to their overall benefit (3).

Fig3: Electrolysis of water (5)

There are different technologies available for electrolysis of water electrolysers such as 

Alkaline Electrolyser



High Pressure Electrolyser



High Temperature Electrolyser



PEM Electrolyser



Solid Oxide Electrolyser

This device has many studies and developments which are into the research stage however there are many factors which are in relationship to the feasibility of this device factors such as limits to 313

Proceedings of SEEP2015, 11-14 August 2015, Paisley large scale hydrogen production investment cost, efficiency, life time and operation load range. 5

CHOSEN ELECTROLYSER

Many large scale chemical and power plants use electrolysis for hydrogen production to cool down generators, further more for renewable

For our chosen set up the alkaline electrolyser from NEL hydrogen, the electrolyser has benefits such as Hourly hydrogen production

energy use hydrogen is also stored as use a fuel.

capacity of 500 Nm3, low power consumption of There are many different types of electrolysers however the atmospheric electrolyser has been well documented as a reliable and efficient piece of equipment which has been used for a multitude of different loads. Water mixed alkaline Electrolyser is made from an electrolyte mixed with water, two electrodes; the cathode (+) and the anode (-) and wiring which creates a

4.35 KW/h per Nm3, High reliability and lifespan with Flexible operation, maximum output can be achieved, ideally fitted for integration with H2-energy applications, low maintenance costs and relatively low space requirement. (7) Some of the benefits of using Alkaline over others are as follows:

loop. The electrodes are placed in the solution and when direct current at a certain value is passed through them and the solution water is split into hydrogen and oxygen gas, hydrogen is collected at the negative terminal and oxygen at the positive terminal. The gas which is produced is directly proportional to the current passed through the electrodes (6) The equations for this chemical reaction are displayed in the following equatin:_ When current passes in to the electrolyte, H2O (l) ↔ H+ (aq) + OH- (aq) (2) At anode the reaction is, 4OH- ↔ 2H2O + O2 + 4e- (3) At cathode the reaction is, 2H+ (aq) + 2e- ↔ H2 (g) (4) Overall equation is, 2H2O → 2H2 + O2

314

Table1: Characteristics of Alkaline vs. PEM electrolysis.

Proceedings of SEEP2015, 11-14 August 2015, Paisley adds the advantage to the system reliability and Also the main aim of this section of the report is to highlight the optimal number of solar arrays

future use.

that meets the demand of small community. It will focuses on finding the best characteristics of solar power system so that it can power its users.

7

REFERENCES 1. BP. (2012) BP statistical review of world energy. 2012 edition. London: BP.

A relationship between solar and wind power has already been depicted in many experiments

2. Gil Kliner., 2002. How do Photovoltaic

conducted in the past. The main aim of the first

work. [online] NASA science news.

step in our research is to produce the same

Available at:

amount of power with solar panels alone as

http://science1.nasa.gov/science-

much it would generate if they would generate when combined with wind turbine. This

news/science-at-nasa/2002/solarcells/. [Accessed at: 12-July- 2014].

arrangement will also result in reduced overall maintenance & operational costs. However, the

3. Alan Goodrich n, Peter Hacke, Qi Wang,

complementary benefits of the hybrid system

Bhushan Sopori, Robert Margolis, TedL.

outweighed single model therefore, it was

James, Michael Woodhouse nn The

concluded hybrid model will be used.

National Renewable Energy Laboratory,

6

Golden, COUSA Solar Energy Materials

CONCLUSION

& Solar Cells 114 (2013) 110–135. A

The project aim is to model a renewable system

wafer-based monocrystalline silicon

using the solar power plant with hydrogen

photovoltaics road map: Utilizing known

storage later on will add wind power plant will

technology improvement opportunities

be added to the system to get the maximum

for further reductions in manufacturing

output from the system, which would be used to

costs.

supply power day and night without any fluctuation in the output. Experiments will be

4. S.Middleman, A.K.Hochberg, 1993.

done on the different technologies available for

Process Engineering Analysis in

electrolysis of water to show ability of chosen

Semiconductor Device Fabrication ,

electrolyser to produce hydrogen at large scale.

McGrawHill , NewYork.

Economically

this

system

should

deliver

competitively against other more harmful fuels

5. Water, K. Electrolysis – Non

such as fossil fuel power plant. The lifetime of

Spontaneous Reaction.

wind turbine, solar panels, inverter, electrolyser

http://www.examinationtoday.com/electr

and fuel cell should work continuously which

315

Proceedings of SEEP2015, 11-14 August 2015, Paisley olysis-non-spontaneous-reaction/ (accessed 08 July 2015).

6. G.Lozza, P. Chiesa.,2005. Using Hydrogen as Gas Turbine Fuel. Journal of Engineering for Gas Turbines and Power, vol 173.

7. NEL Hydrogen., 2012. Technical Data sheet of Electrolyser., NEL Hydrogen. Available at : http://www.nelhydrogen.com/home/?pid=75

8. Yunus A. Cengel and Michael A. Boles., 2007. Thermodynamics 6th edition Table A-2 Appendix (Cp values can be in error up to 1%)

9. Barbie, F. PEM electrolysis for production of hydrogen from renewable energy sources, 19 October 2004.

10. G E living. Photovoltaic (PV) Solar Panel Installation in Scotland. http://www.geoliving.co.uk/solar/photovo ltaic-cells (accessed 20 July 2015).

11. Tsiplakides, D. PEM water electrolysis fundamentals. Available at: http://research.ncl.ac.uk/sushgen/docs/su mmerschool_2012/PEM_water_electroly sis-Fundamentals_Prof._Tsiplakides.pdf (Accessed at: 08-July- 2015)

316


CO-GASIFICATION OF COAL AND MUNICIPAL SEWAGE SLUDGE 1. 2. 3. 4. 5.

M. Agrez1, P. Trop2, S. Potrc3, D. Urbancl4 and D. Goricanec5 ELES, Ltd., Electricity Transmission System Operator, Ljubljana, Slovenia [email protected] Faculty of Chemistry and Chemical Engineering, University of Maribor, Slovenia; [email protected] Faculty of Chemistry and Chemical Engineering, University of Maribor, Slovenia; [email protected] Faculty of Chemistry and Chemical Engineering, University of Maribor, Slovenia; [email protected] Faculty of Chemistry and Chemical Engineering, University of Maribor, Slovenia; [email protected]

email: email: email: email: email:

Abstract Municipal sewage sludge is a renewable resource that can be used for the production of synthetic fuels. Simulations regarding the co-gasification of coal and municipal sewage sludge within entrained-flow slagging gasifiers (two 500 MWth) were performed using AspenPlus software. This study presents a comparison between the gasification of coal and the co-gasification of coal and municipal sewage sludge. A ratio between sewage sludge and coal of 1:9 was chosen for the co-gasification case. Methanol was produced from the synthesis gas in both cases. The comparison was based on net present values. The results showed that co-gasification case outperformed coal-only case. Keywords: gasification, municipal sewage sludge, coal, simulation 1 INTRODUCTION The European legislation regarding wastewater treatment has become increasingly restrictive. Consequently the number of wastewater treatment plants has increased substantially, thereby increasing the amount of municipal sewage sludge throughout the EU. It is because of this fact that a lot of research is being done on developing alternative ways of handling sewage sludge. It is considered as a renewable energy source and consequentially its usage has become very popular. Usually it is incinerated, combusted in a thermal power plant, landfilled, or even used in agriculture. However, recent studies have aimed at using it as a source of carbon for producing synthetic fuels such as methanol, DME, etc. In Slovenia alone, around 28,600 tons of sewage sludge was generated from wastewater treatment plants in 2009. The majority of this amount, around 16,800 tons, was combusted or incinerated whilst the remainder was deposited in landfills, composted or exported. The Slovenian main targets regarding this area are:  Biological pre-treatment of less polluted sewage sludge at appropriate wastewater treatment plants at a regional level;

317



To ensure sufficient capacities for thermal treatment of 70,000 tons of sewage sludge dried to 30% of dry matter. The gasification of coal and municipal sewage sludge has already been investigated by some researchers. Co-gasification of mixtures of sewage sludge with two types of coal (bituminous and lignite) were performed in a laboratory-scale fluidised bed reactor, where the influence of the feedstock composition was determined on key parameters of the gasification [1]. The gasification of sewage sludge was investigated within near and super-critical water in a batch reactor and the results showed that the formation of gaseous products could be affected by temperature [2]. Gasification of sewage sludge in supercritical water was also studied in a fluidised bed reactor. The increase in temperature and the decrease in feedstock concentration were both favourable for gasification [3]. Experiments were also conducted within an atmospheric fluidised bed reactor using air and air + steam as gasification agents [4]. The sewage sludge gasification process was investigated in a bubbling fluidised bed gasifier and they studied the influences of

Metals content (mg/kgdry)

Ultimate analysis (%)

Proximate analysis (%)

Proceedings of SEEP2015, 11-14 August 2015, Paisley different operating variables on the synthesis  Gasifier was modelled using an RGibbs gas' properties. The variables studied were the reactor model, which minimises Gibbs free equivalence ratio, steam to biomass ratio, and energy in order to calculate the equilibrium the temperature [5]. composition at the reactor outlet. Formations of tar and char can be very  The temperature at the gasifier outlet would problematic in low-temperature gasification be 1450°C in order to allow for the ash slag systems such as fluidised bed gasifiers, causing to flow out of the gasifier. pipe-blockages, and consequently frequent  Carbon conversion was set at 100%. shutdowns [6]. On the other hand, there are no  Heat losses of the gasifier were estimated to problems with tar and char within entrained flow be 10MW for a 500MWth gasifier. slagging gasifiers, thus presenting well established technology for converting solid Table 1: Properties of coal and sewage feedstock into clean syngas [7]. It is a leading sludge gasification technology in terms of conversion efficiency, availability, and operational Sewage Coal flexibility [8]. sludge This paper presents a comparison between coal Moisture 14.2 60.0 gasification and the co-gasification of coal and Fixed C 34.7 9.0 municipal biomass within an entrained-flow Volatiles 40.5 20.0 gasifier. The ratio between biomass and coal, Ash 10.6 11.0 calculated on the basis of LHW (lower heating Carbon 56.4 52.5 value), was set at 1:9 for the co-gasification case. Hydrogen 4.4 6.4 The comparison is based on the simulations of Nitrogen 0.8 9.2 gasification and methanol productions. Methanol Sulphur 1.2 0.8 was chosen because it is a key raw material for Oxygen 12.5 31.1 the syntheses of many different chemicals and is 7.16 Heating value (MJ/kg) 23.02 a global commodity with a large and liquid 70.9 Ca 11.9 market. Methanol is used for the production of 9.2 Mg 1.1 formaldehyde, acetic acid, MTBE (Methyl tert43.3 Cr 0.1 butyl ether), propylene, DME (Dimethyl ether) 1.0 Hg ND etc. Furthermore, methanol can be used as fuel 1.1 Cd ND within internal combustion engines and gasoline 53.8 can also be produced from methanol. Pb ND 2 METHODS Simulation of co-gasification within an entrained flow gasifier was performed using Aspen Plus software. Table 1 states the proximate and ultimate analyses of coal and sewage sludge that were used in the model. 2.1 Simulation Figure 1 presents a simplified flowsheet of the model, which includes the drying of biomass and coal, gasification, a two-stage Rectisol process with WGS reaction and the production of methanol. It was assumed that coal and biomass would be dried in a mill drier to moisture contents of 1.5%, and 3% respectively. CO 2 was assumed as the conveyance-gas for transporting the dried pulverised coal and sewage sludge to the gasifier. The following assumptions were considered during the creation of the gasifier model:

318

The obtained syngas was quenched to 600°C using a certain part of the syngas cooled to 300°C. Syngas should then be sent to a ventury scrubber for the removal of dust and heavy metals. However the removals of dust and heavy metals (Cr, Hg, Cd, Pb) are outside the scope of this work. A two stage Rectisol process was chosen during this study. In the first stage sulphur compounds were removed from the synthesis gas. A certain percentage of the synthesis gas without sulphur was sent directly to the methanol synthesis section and the remainder to the WGS section for the production of H2 from CO by the following equation: CO + H 2 O

CO 2 + H 2

(1)

The WGS reaction was necessary for obtaining synthesis gas with a suitable ratio between H 2,

Proceedings of SEEP2015, 11-14 August 2015, Paisley was purged from the synthesis loop in order to CO, and CO2 for the production of methanol [9]. eliminate the inert material. The purged gas was A two stage WGS process was used. It was then conveyed to a gas turbine for producing assumed that the concentration of CO would be electricity. Flue gases coming from the gas 4% after the high temperature WGS reactor, and turbine were used for drying the coal. In the case 0.03% after the low temperature WGS reactor. of co-gasification the gas turbine was replaced The main components of the gas stream were H 2 with a furnace because larger amounts of heat and CO2 after the elimination of water. The were needed to dry the sewage sludge. The heat majority of the CO2 was removed during the that was released during the cooling of the second stage of the Rectisol process. The synthesis gas, and the heat released during the produced hydrogen stream was then added to the WGS-reaction and synthesis of methanol was stream coming directly from the first stage of the used to produce steam that was partially used for Rectisol process and the recycle stream coming heating the re-boilers of the distillation columns. from the methanol reactor. This mixture was The remaining heat was used to raise steam that then heated and conveyed to the reactor for the was applied in a triple-stage steam turbine to synthesis of methanol which is fully described in produce electricity. Isentropic and mechanical [9]. Crude methanol was obtained after efficiencies of 0.8 and 0.98, respectively were condensation in HX2. The majority of unreacted assumed for all the turbines and compressors. gases coming from the flash unit F were recycled to the synthesis loop, whilst a smaller percentage

Figure 1: A schematic illustration of methanol production from coal and sewage sludge 2.2 Economic analysis Economic analysis was based on NPV (Net present value). NPV values were calculated covering a 30 year period with an annual discount rate of 5%, and 8400 working hours per year.

319

Estimation of revenues and operating costs The prices for raw materials, methanol and electricity that were assumed for the estimation of revenues and operating costs are listed in Table 2 The price of sewage sludge has been reported as negative in previous studies [10], [11], therefore

Proceedings of SEEP2015, 11-14 August 2015, Paisley it was assumed that the price of sewage sludge equally-sized units, C0 is the cost of a reference would be (-20) €/t. Electricity was consumed for unit, S is the capacity of a process unit, S0 is the the compressors, pumps, and refrigeration units capacity of a reference process unit, e is the cost for cooling the methanol within the Rectisol scaling exponent for different numbers of process. equally-sized units, and f is the cost- scaling factor [15]. All the prices of equipment were Table 2: Estimated prices converted to values in 2013 using CEPCI Coal price [12] 15 €/MWh (Chemical Engineering Plant Cost Index). It was SS price (-20) €/t assumed that the same type of dryer would be Methanol price [13] 365 €/t used in the coal-only case and co-gasification case. Table 3 presents estimated cost of Electricity price [12] 54 €/MWh equipment for the coal-only case and coCarbon tax [14] 14 €/t gasification case. The capital costs of both It was assumed that the captured CO 2 would be processes are nearly the same. The main emitted into the atmosphere. The amount of the difference was in the usage of purge gas from CO2 that would be emitted to the atmosphere the synthesis loop. Drying of the sewage sludge during the combustion of the used sewage demands high amounts of high-temperature heat. sludge was subtracted from the overall CO 2 It is for this reason that only the triple-stage emissions in the case of co-gasification, due to steam turbine was used in the co-gasification carbon neutrality of the sewage sludge. This process because the purge gas needed to be burnt resulted in specific emissions of 1.7kg and for the production of drying heat. On the other 1.95kg of CO2 per kg of methanol produced hand, a gas turbine was used in the coal-only during the co-gasification and coal-only case, process because there was enough heat left in the respectively. flue gases in order to dry the coal. However, Estimation of investment costs low-temperature dryer could also be used, which Investment cost were calculated by the following would increase the investment cost and decrease f e the need for high temperature heat that was equation: C  n C0  S / (n  S0 ) , where C is the supplied from the purge stream. cost of a process unit, n is the number of Table 3: Capital cost of synthesising equipment Process/Unit

Scaling parameter

Coal and sorbent handling ASU and O2 compression Gasification island Rectisol process Claus process WGS Methanol production Methanol distillation Combined cycle power plant Total capital cost

Coal (t/day) O2 (t/day) MW based on LHV CO2 (t/h) S (t/day) MW based on LHV MeOH (t/day) MeOH (t/day) MWe

Coal, sewage sludge, and sorbent handling ASU and O2 compression Gasification island

Coal and sewage sludge (t/day) O2 (t/day) MW based on LHV

S

So

Co(M€)

Methanol synthesis from coal 3976 5447 30,9 2749 2035 81,5 1000 500 112,8 205 2771 806,2 50 96 6,1 582 815 7,1 2671 5000 76,9 2671 5292 16,1 84 69 30,0

Reference /year

Co’ (M€)

n

[15] / 2008 [15] / 2008 [*] / 2012 [16] / 2011 [17] / 2007 [15] / 2008 [18] / 2005 [19] / 2008 [20] / 2003

31,4 82,8 112,8 804,3 6,8 7,2 96,0 16,4 44,7

1 1 2 1 1 2 1 1 2

C (M€) 25,4 101,3 210,5 140,5 4,4 6,7 63,1 10,3 59,8 622,1

Methanol synthesis from coal and sewage sludge

Rectisol process Claus process WGS Methanol production Methanol distillation Steam turbine Total capital cost

CO2 (t/h) S (t/day) MW based on LHV MeOH (t/day) MeOH (t/day) MWe

4783

5447

30,9

[15] / 2008

31,4

1

28,8

2745 1000

2035 500

81,5 112,8

[15] / 2008 [*] / 2012

82,8 112,8

1 2

101,2 210,5

205 49 576 2672 2672 70

2771 96 815 5000 5292 350

806,2 6,1 7,1 76,9 16,1 105,0

[16] [17] [15] [18] [19] [21]

804,3 6,8 7,2 96,0 16,4 161,1

1 1 2 1 1 2

140,4 4,3 6,7 63,1 10,3 54,8 620,1

/ 2011 / 2007 / 2008 / 2005 / 2008 / 1995

*Personal communication (Anton Haberzettl, Head of Business Development, Siemens Fuel Gasification Technology GmbH)

320

Proceedings of SEEP2015, 11-14 August 2015, Paisley 2400 2000 1600

NPV [M€]

1200

800 Coal + Sewage Sludge

400

Coal

0 -400

-800

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

Time [years]

Figure 2: Net present values In all the calculations of net present value, Figure 2, it was assumed that all prices for raw materials and products, and also annual discount rate would be constant during the 30 year-long period. This assumption is slightly unrealistic but has been used for simplicity. Our future work would include sensitivity analysis on those parameters that would have the greatest impact on the NPV. 3 CONCLUSION Co-gasification of sewage sludge in entrainedflow slagging gasifiers (two 500MWth) was presented in this work and compared to a coal to methanol process. It can be seen from Figure 2 that the co-gasification case performed better with a net present value of 2,151M€ after a 30 year period, whilst the coal-only case’s net present value amounted to 1.827M€. This result was expected because the price of sewage sludge is negative. It is also considered as a carbon neutral material, therefore carbon tax was lower in the case of co-gasification. However, the heavy metals that are present in sewage sludge in much bigger quantities than in coal were unconsidered during this work. The drying of the sewage sludge is one of the critical steps, because it is very energy-intensive. For this reason the development of an efficient low-temperature drier for sewage sludge would be of great importance in the future, which could increase the economic efficiency of the process even further. The ratio between sewage sludge and coal is also a very important parameter, which could be optimised in the future work. 321

REFERENCES [1] G. García, J. Arauzo, A. Gonzalo, J. L. Sánchez, and J. Ábrego, Influence of feedstock composition in fluidised bed cogasification of mixtures of lignite, bituminous coal and sewage sludge, Chemical Engineering Journal, vol. 222, no. 0, pp. 345–352, Apr. 2013. [2] Y. Chen, L. Guo, H. Jin, J. Yin, Y. Lu, and X. Zhang, An experimental investigation of sewage sludge gasification in near and super-critical water using a batch reactor, International Journal of Hydrogen Energy, vol. 38, no. 29, pp. 12912–12920, Sep. 2013. [3] Y. Chen, L. Guo, W. Cao, H. Jin, S. Guo, and X. Zhang, Hydrogen production by sewage sludge gasification in supercritical water with a fluidized bed reactor, International Journal of Hydrogen Energy, vol. 38, no. 29, pp. 12991–12999, Sep. 2013. [4] E. Roche, J. M. de Andrés, A. Narros, and M. E. Rodríguez, Air and air-steam gasification of sewage sludge. The influence of dolomite and throughput in tar production and composition, Fuel, vol. 115, no. 0, pp. 54–61, Jan. 2014. [5] J. M. de Andrés, A. Narros, and M. E. Rodríguez, Air-steam gasification of sewage sludge in a bubbling bed reactor: Effect of alumina as a primary catalyst, Fuel Processing Technology, vol. 92, no. 3, pp. 433–440, Mar. 2011.

[6]

[7]

[8]

[9]

[10]

[11]

[12]

Proceedings of SEEP2015, 11-14 August 2015, Paisley [13] Pricing | Methanex Corporation. [Online]. L. Shen, Y. Gao, and J. Xiao, Simulation of Available: https://www.methanex.com/ourhydrogen production from biomass business/pricing. [Accessed: 01-Jul-2015]. gasification in interconnected fluidized [14] Ministry of Finance, Environmental Taxes, beds, Biomass and Bioenergy, vol. 32, no. May-2015. . 2, pp. 120–127, Feb. 2008. [15] E. Martelli, T. Kreutz, and S. Consonni, D. Bi, Q. Guan, W. Xuan, and J. Zhang, Comparison of coal IGCC with and without Combined slag flow model for entrained CO2 capture and storage: Shell gasification flow gasification, Fuel, vol. 150, pp. 565– with standard vs. partial water quench, 572, Jun. 2015. Energy Procedia, vol. 1, no. 1, pp. 607– M. Troiano, R. Solimene, P. Salatino, and 614, Feb. 2009. F. Montagnaro, Multiphase flow patterns in [16] W. Zhou, B. Zhu, D. Chen, F. Zhao, and entrained-flow slagging gasifiers: Physical W. Fei, Technoeconomic assessment of modelling of particle–wall impact at nearChina’s indirect coal liquefaction projects ambient conditions, Fuel Processing with different CO2 capture alternatives, Technology. Energy, vol. 36, no. 11, pp. 6559–6566, P. Trop, B. Anicic, and D. Goricanec, Nov. 2011. Production of methanol from a mixture of [17] Betty, N., Matusik, K., and Nagpurwala, torrefied biomass and coal, Energy, vol. 77, N., Coal-to-methanol process: final report. pp. 125–132, Dec. 2014. 2008. Andreas Lüschen and Reinhard Madlener, [18] Elton Amirkhas, Raj Bedi, Steve Harley, Economics of Biomass Co-Firing in New and Trevor Lango, Methanol production in Hard Coal Power Plants in Germany, trinidad & tobago, Final Report: Phase II. Institute for Future Energy Consumer University of California, Davis, 07-JunNeeds and Behavior (FCN), AACHEN, 2006. 2010. [19] X. Zhang, L. Zhong, Q. Guo, H. Fan, H. D. R. McIlveen-Wright, Y. Huang, S. Zheng, and K. Xie, Influence of the Rezvani, J. D. Mondol, D. Redpath, M. calcination on the activity and stability of Anderson, N. J. Hewitt, and B. C. the Cu/ZnO/Al2O3 catalyst in liquid phase Williams, A Techno-economic assessment methanol synthesis, Fuel, vol. 89, no. 7, pp. of the reduction of carbon dioxide 1348–1352, Jul. 2010. emissions through the use of biomass co[20] Zachary Hoffman, Simulation and combustion, Fuel, vol. 90, no. 1, pp. 11–18, Economic Evaluation of Coal Gasification Jan. 2011. with Sets Reforming Process for Power Proizvodni proces - Termoelektrarna Production, Master Thesis. Louisiana State Trbovlje. [Online]. Available: University, 2005. http://www.tet.si/si/elektricna[21] L. Alexander, Stream Turbines for modern energija/proizvodni-proces/. [Accessed: 14fosil fuel power plants. The Fairmont press, Feb-2015]. 2008.

322


OPTIMISATION OF PACK CHROMISED COATINGS ON 304 STAINLESS STEEL FOR PROTON EXCHANGE MEMBRANE FUEL CELL BIPOLAR PLATES USING BOX-BEHNKEN DESIGN. A.M. Oladoye1, J. G. Carton1, K. Benyounis1, J. Stokes1 and A.G. Olabi2 1. School of Mechanical & Manufacturing Engineering , Dublin City University, Ireland [email protected],[email protected],[email protected], [email protected] 2. School of Engineering, University of the West of Scotland, Paisley; [email protected] Abstract Proton exchange membrane fuel cells (PEMFC) are clean and efficient power sources that have the potential to reduce greenhouse emissions if successfully commercialised for automotive, portable and distributed/stationary power generation. However, the high cost and low durability of the bipolar plates are key challenges to be addressed for PEMFCs to compete with conventional power generation technologies. Bipolar plates are currently fabricated from graphite but the material is brittle, permeable to gases and expensive to mass produce. Alternative materials such as stainless steels offer lighter weight, higher mechanical strength and lower productions cost advantages over graphite but have low surface conductivities and are prone to corrosion attack in PEMFC environments. Hence in this study, AISI304 stainless steel alloy is chromised via pack cementation in order to enhance its performance as PEMFC bipolar plate material. Box–Behnken experimental design was employed to optimise the deposition parameters to produce chromium diffusion coatings on AISI304 stainless steel with the maximum corrosion resistance in simulated PEMFC environment of 0.5M H2SO4 + 2ppmHF at 70oC. The optimal coating was characterised and electrochemically polarised in simulated PEMFC environments and working potentials. Surface conductivity was evaluated in terms of the interfacial contact resistance between the optimised coated surface and carbon paper. The results are compared with that of the substrate and the implication for bipolar plate application is discussed. Keywords: pack chromising, stainless steel, corrosion, Box-Behnken design, interfacial contact resistance, bipolar plates, PEM fuel cells 1

INTRODUCTION

Stainless steel alloys are considered as potential candidate material to replace graphite bipolar plates in PEMFCs as they offer lighter weight, higher mechanical strength and lower production cost advantages over graphite. Conversely, stainless steel alloys corrode in the humid and acidic PEMFC environments producing metallic ions that could degrade the efficiency of the membrane electrode assembly (MEA). These ferrous based alloys also possess low surface conductivities due to the existence a semiconductive surface oxide layer, hence, they exhibit high interfacial contact resistance with the gas diffusion layer (GDL) resulting in cell power degradation [1 -5]. A widely adopted method for enhancing the performance of stainless steels alloys in PEMFC 323

environments has been surface modification via coatings and treatments. However, coatings and treatments for bipolar plates must be cheap, corrosion resistant, and electrically conductive amongst other requirements [1-3].Chromised coatings produced via pack cementation are a class of diffusion coatings that could satisfy these requirements. Indeed, previous works on chromised stainless steel alloys for PEMFC bipolar plates evidently showed that chromised stainless steels produced either at low or high temperatures performed better than the bare substrate in both in-situ and ex-situ PEMFC testing [6-10]. Nevertheless, it is well established that the surface properties and thickness of chromised coatings depends on the alloying content of the substrate amongst other factors. For instance, the thickness of chromised coatings in austenitic stainless steels is restricted compared to ferritic stainless steels because

Proceedings of SEEP2015, 11-14 August 2015, Paisley with flowing Argon gas. The crucibles were heated according to the design matrix based on chromium is a ferrite stabiliser, hence, its solubility in austenite is limited [11-13]. Also, three-level, three-factor Box–Behnken design austenitic stainless steels with higher as shown in Table 2. After the heat treatment concentration of ferrite stabilisers (e.g. 316 cycle, the chromised samples were cleansed in stainless steel) exhibit lesser thickness than their acetone and dried in open air. other counterparts with lesser amounts of ferrite stabilisers (e.g. 304 stainless steel).This fact was Corrosion resistance of the chromised samples in demonstrated in the work of Ralston et al. [11] aerated 0.5MH2SO4 +2ppm HF was evaluated wherein the thickness of chromised coating by Tafel polarization at a scanning rate of 1mv/s produced on 304SS was twice that of using CHI 630C potentiostat (CH Instruments, 316stainless steel. Although, Mo enhances USA). A three conventional electrode system pitting corrosion resistance of stainless steel, a consisting of Ag/AgCl (saturated KCl) as thicker chromised coating on AISI304 may be reference electrode, platinum mesh as the able to compensate for such effect. Hence, in counter electrode and the chromised samples as this study we examine the pack chromising of the working electrodes. Open circuit potential 304stainless steel. It is also noted that pack measurements were conducted for ~ 30minutes chromised 304stainless steel has not been after which Tafel measurements was conducted. reported in the literature before now. In order to produce chromised coatings with the maximum The thickness of the chromised samples was corrosion resistance in simulated PEMFC measured using line measurements techniques environment of 0.5M H2SO4 + 2ppmHF at 70oC on ZEISS EVOLS 15 Scanning electron , Box-Behnken experimental design (BBD) Microscope (SEM). The results are presented in coupled with Derringer’s desired function Table 1. Development and verification of the methodology was employed to optimise the model as well as optimisation were conducted deposition parameters. with Design Expert statistical software package version 9.0.2.0 (State-Ease Inc., USA). This Box–Behnken designs are a family of rotatable paper deals with only the optimisation process. or nearly rotatable three-level second-order designs based on incomplete factorial. This The phases present in the optimal coating was method reduces experimental cost and time as characterised using a D8 Advance Buker XRD it requires fewer experimental runs and avoids with a Cu Kα radiation tube with a wavelength combinations involving extreme conditions of of 0.154nm using 5o glancing angle of incidence higher or lower levels of factors investigated method. Potentiostatic tests for the optimised [14-16]. Hence, the aim of this study is to sample was conducted at -0.1V with hydrogen optimise the corrosion resistance of pack gas bubbling and 0.6V with air bubbling for chromised coatings deposited on AISI304 60minutes to simulate anodic and cathodic stainless steel in PEMFC environments. The PEMFC working potentials and environments. optimised coated surface is thereafter tested in simulated PEMFC working environments. Surface conductivity of the chromised steel and bare substrate was evaluated in terms of 2 EXPERIMENTAL PROCEDURES interfacial contact resistance between carbon paper and the sample. Experimental set up for Commercially available AISI 304 stainless steels the measurement consisted of two of pieces of coupons were ground with SiC abrasive papers, Toray Teflon treated carbon paper (TGP-H-90, ultrasonically cleansed in acetone and dried in Fuel cell Earth, USA) sandwiched between the air. The cleaned coupons were thereafter chromised sample and two copper sheets. A embedded in a previously ball-milled powder direct current of 1A was supplied to the copper pack of 50wt% Cr powder Wt % of NH4Cl as sheets via a XHR 300-3.5DC power source. The shown in Table 1 with the balance being voltage drop across the setup was measured with alumina. The crucible was sealed with fire clay, Tektronix DMM912 digital multi-meter while cured and afterwards loaded into a tube furnace the compaction force was gradually applied by a

324

Proceedings of SEEP2015, 11-14 August 2015, Paisley Table 1: Design matrix based on 3- level, 3 –factor Box–Behnken design

Run

Time (Hours)

Activator content (wt%)

Temperature (oC)

Thickness (µm)

Corrosion current (µA/cm2)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

6 6 6 9 3 6 9 3 6 6 6 3 3 6 6 9 9

3 5 7 5 5 3 5 5 7 5 5 3 7 5 5 7 3

1100 1050 1000 1000 1000 1000 1100 1100 1100 1050 1050 1050 1050 1050 1050 1050 1050

78.19 47.30 31.23 48.01 19.04 22.96 92.09 50.90 80.73 43.25

2.05 1.85 1.99 9.13 3.31 9.99 4.54 2.74 4.40 1.01

46.31 34.89 34.17 45.21 47.75 55.02 55.13

1.02 0.41 0.76 1.06 1.26 1.88 2.35

ZwickRoell (Z5KN) universal testing machine. ICR between carbon paper and the bipolar plate material and correction for the carbon paper were calculated as described in [17]. 3 3.1

RESULTS & DISCUSSION Optimisation

Optimum conditions for producing chromium diffusion with maximum corrosion resistance in simulated PEMFC environment was obtained by numerical simulation using the desirability function. This method determines the optimum condition by first locating the level of each factor that can achieve the predicted response after which the overall desirability is maximised based on the set goals [16]. In this study our objective was to minimise corrosion current density while keeping the thickness in range as shown in Table 2.The optimal process parameter with the highest desirability value of 1 was given 325

as Temperature: 1040oC, Time 3.04hrs and activator content: 6.78wt% with a predicted corrosion current density value of 0.398 µA/cm2 and thickness of 30.5µm. Model validity experiments showed that the average corrosion current density of the coating produced at the optimal process parameter was 0.414 µA/cm2 (Fig 1) with a thickness of ~30µm (Fig 2).These experimental values are in good agreement with the predicted values hence validating the model. 3.2

Simulated PEMFC test

Interfacial contact resistance between the bipolar plates and the GDL is vital to the operation of PEMFCs as it accounts for ohmic losses in the cell. Increase in ICR would lead to a corresponding increase in ohmic losses and ultimately cell performance degradation. From the ICR of the coated surface produced with the optimal process parameters obtained in section

Log current density (log A/cm2)

Proceedings of SEEP2015, 11-14 August 2015, Paisley 3.1(hereafter called Chromised 304) and SS304 current decay which is related to the formation (Fig.3), it can be observed that chromising of passive oxide as soon as the potential is treatments significantly reduced the ICR of applied. After the current decay, SS304 SS304.The high ICR of SS304 is attributed to stabilised at 19.45µA/ cm2 after one hour of the semi-conductive nature of the passive oxide polarisation while chromised 304 stabilised at layer found on stainless steels. At typical 1.28µA/cm2 an order of magnitude lower. 2 PEMFC compaction force of 150N/cm , the Similarly, in the anodic conditions of -0.1V and chromised surface exhibited ICR value of hydrogen gas bubbling, chromised 304exhibited 2 24N/cm which was about five times lower than low cathodic currents and stabilised at -6µA/cm2 that of the uncoated substrate. Notably, the ICR .On the other hand, SS304 exhibited high value for chromised 304 stainless steel was corrosion current stabilising at 200.30µA/cm2. moderately higher than those earlier reported for The poor performance of SS304 is due to the other chromised stainless steels [6 -10]. This inability to passivate under reducing conditions. could be ascribed to the difference in contact resistance measurements, for instance, it is well The reduction in ICR and the enhanced known that Teflon treated paper such as used in corrosion behaviour of SS304 after chromising this study increases ICR [18]. treatments can be attributed to the formation of chromium based nitrides and carbides at the chromising temperature as seen in the 5o (a) -5 glancing XRD pattern (Fig 4). (Cr, Fe)2N was formed from the reaction of Cr and decomposed -6 nitrogen from the activator while (Cr, Fe)3C7 , (Cr, Fe)23C6 were formed from the reaction between inwardly diffused Cr and outwardly -7 diffused carbon at chromising temperatures. -8

Table 2: constraints used for optimisation -9 -0.2

0.0

0.2

0.4

0.6

variables

criteria In range

Lower limit 3

Upper limit 9

Importance 3

Time Activator content

In range

3

7

3 3 3

0.8

Potentials (V)

Temp. In range 1000 1100 Thickness In range 19.035 92.085 Corrosion 0.41 9.99 minimise 2 current µA/cm µA/cm2

Fig. 1: (a) Tafel plot for coating produced with optimal process parameters (b) cross- section of the coating

Fig 4a and b displays the static polarisation curves for chromised 304 and SS304 under simulated PEMFC cathodic & anodic conditions. In simulated cathodic conditions of 0.6V and air purging (Fig 6a), It can be seen that both chromised 304 and SS304 experienced transient 326

5

These chromium based compounds are known to possess higher electrical conductance than the native chromium oxide layer on stainless steel. Comparing the static polarisation result obtained in this study with that of 316/316L in the literatures, wherein values within 10 µA/ cm2 were reported for chromised coatings produced at 1100oC and 1050oC [6,7] .The enhanced corrosion resistance of chromised 304 can also be attributed to a thicker coated layer.

Proceedings of SEEP2015, 11-14 August 2015, Paisley 600

-5

4.0x10

-5

3.5x10

2

3.0x10

Current densiy, A/cm

ICR , mohms.cm

2

-5

400 304 stainless steel 200

Chromised 304 stainless steel

-5

2.5x10

304 stainless steel

-5

2.0x10

-5

1.5x10

-5

1.0x10

0

-6

0

50

100

150

200

Compaction Force , N/cm

250

chromised 304 stainless steel

5.0x10

300

2

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Time,hrs

Fig. 2: ICR of chromised 304 stainless steel and the substrate vs compaction pressure -3

5.0x10

BIPOLAR

Optimisation of pack chromising process parameters using Box–Behnken designs combined with desirability function yielded a continuous chromised layer produced at 1040oC for ~3hrs with an activator content of 6.78wt%. The chromised surface exhibited corrosion current densities of ~2µA/cm2 and -6µA/cm2 when tested in the PEMFC cathode and anodic working conditions respectively. Considering the < 1µA/cm2 target for bipolar plate material, chromised 304 could be considered for bipolar plate as the testing environments (pH 1) are very aggressive compared to the actual environment in PEMFC (pH 3). However, the ICR of chromised 304 needs to be further reduced to meet the target of < 10mΩ /cm2 target for materials for bipolar plates within the compaction force range of interest for PEMFCs (100-200N/cm2). 4

(b) -3

2

IMPLICATION FOR PLATE APPLICATION

CONCLUSION

Box–Behnken experimental design was successfully employed to optimise pack chromising process parameters to produce chromised coatings on 304stainless steel which exhibited maximum resistance in simulated PEMFC environment. The coating produced at 1040oC for ~3hours with pack powder composition of 50%wtCr, 6.78wt% NH4Cl, balance Al2O3 yielded a continuous chromised layer composed of a mixture of (Cr,Fe)2N, 327

Current density , A/cm

3.2

4.0x10

-3

3.0x10

-3

2.0x10

-3

1.0x10

chromised 304 stainless steel

304 stainless steel

0.0 0

12

24

36

48

60

Time, minutes

Fig.3: polarisation curve of chromised 304 and 304SS in PEMFC (a) cathode and (b) anode conditions

(Cr,Fe)23C6 and (Cr,Fe)3C7. Consequently, the chromised layer exhibited ICR of 24mΩ/cm2 at compaction force of 150N/cm2 and exhibited excellent corrosion resistance in simulated PEMFC environments. The results were significantly better than that of the substrate. However, the ICR of chromised 304 needs to be further reduced to meet the target of < 10mΩ /cm2 target for materials for bipolar plate application.

Proceedings of SEEP2015, 11-14 August 2015, Paisley cell by lower temperature chromizing treatment, J. Power sources, Vol. 195, Issue 9, pp. 2810– 2814, 2010. [9] T-M Wen , K-H Hou , C-Yuan Bai , M-Der  - (Cr,Fe) C Ger , P-H Chien , S-J Lee, Corrosion behaviour  - (Cr,Fe) N and characteristics of reforming chromized  - (Cr,Fe) C  coatings on SS 420 steel in the simulated environment of proton exchange membrane fuel  cells, Corrosion Sci., Vol. 52, Issue 11, pp   3599–3608, 2010.      [10] C-Y. Bai, T-M. Wen, M.-S. Huang, K-H.       Hou, M-D. Ger, S.-J. Lee, Surface modification & performance of inexpensive Fe-based bipolar plates for proton exchange membrane fuel cells, 40 50 60 70 80 90 J. Power sources, Vol. 195, Issue 17, pp 56862degrees o 5691, 2010. Fig. 4: 5 GIXRD of chromised 304 stainless steel [11] K.D. Ralston, D. Fabijanic, R.T. Jones , N. REFERENCES Birbilis, Achieving a chromium rich surface upon steels via FBR-CVD chromising [1] H. Tawfik Y. Hung, D. Mahajan, Metal treatments, Corrosion Sci., Vol. 5,pp. 2835– bipolar plates for PEM fuel cell—A review, J. 2842, 2011. Power sources, Vol. 163, Issue 2, 2007 755 [12] G.H. Meier, C. Cheng, R.A. Perkins, W. 767. Bakker, Diffusion chromizing of ferrous alloys [2] D. J. L. Brett and N. P. Brandon, Bipolar Surf. Coat. Tech, Vol. 39-40 part 1,pp 53 -64 , Plates: The Lungs of the PEM Fuel Cell, The 1989. Fuel Cell Review, Vol. 2, pp 15-23, 2005. [13] R.C. Jongbloed, Chromizing, Mater. Sci. [3] S. Karimi, N. Fraser, B. Roberts, F.R. A Forum 163-165 pp. 611- (1994) 611-618. Foulkes, Review of Metallic Bipolar Plates for [14] J.G. Carton, A.G. Olabi, Design of Proton Exchange Membrane Fuel Cells: experiment study of the parameters that affect Materials and Fabrication Methods, Advances in performance of three flow plate configurations Materials Sci. & Engr ,2012, Article ID 828070, of a proton exchange membrane fuel cell, http://dx.doi.org/10.1155/2012/828070. Energy, Vol. 35, pp.2796-2806, 2010. [4] H. Wang and J. A. Turner Reviewing [15] K.Y. Benyounis, A.G. Olabi, M.S.J. Metallic PEMFC Bipolar Plates, Fuel cells Vol. Hashmi, Effect of laser-welding parameters on 10, Issue 4, pp. 510–519, 2010. the heat input and weld-bead profile, J. Mater. [5] R. A. Antunes, M. Cristina, L. Oliveira G. Process. Tech., Vol. 164–165, pp.978–985, Ett, V. Ett , Corrosion of metal bipolar plates for 2005. PEM fuel cells: A review, Int. J. of hydrogen [16] R.H. Myers and D.C. Montgomery, energy Vol. 35, pp. 3632–3647, 2010 Response surface methodology, Wiley series [6] K.H. Cho, W.G. Lee, S.B. Lee, H. Jang 2002. Corrosion resistance of chromized 316L [17] H. Wang , M.A. Sweikart , J.A. Turner, stainless steel for PEMFC bipolar plates J. of Stainless steel as bipolar plates for PEM fuel Power sources, Vol. 178, pp. 671–676, 2008. cells, J. Power sources, Vol. 115,Issue2, pp 243[7] S.B. Lee, K.H. Cho, W.G. Lee, H. Jang, 251, 2003. Improved corrosion resistance and interfacial [18] D. Ye, E. Gauthier, J. B. Benziger, M. Pan, contact resistance of 316L stainless-steel for Bulk and contact resistances of gas diffusion proton exchange membrane fuel cell bipolar layers in proton exchange membrane fuel cells, plates by chromizing surface treatment, J. Power J. Power sources, Vol. 256, pp 449-456, 2014. sources, vol. 187 pp.318–323, 2009 [8] L. Yang, H. Yu, L. Jiang, L. Zhu, X. Jian, Z. Wang, Improved anticorrosion properties and electrical conductivity of 316L stainless steel as bipolar plate for proton exchange membrane fuel 23

2

1-x

3

Intensity, a.u

7

6

328


PRODUCTION OF HYDROGEN ENERGY FROM PALM OIL MILL EFFLUENT AS SUSTAINABLE BIOMASS USING AN EMPERICAL MODEL N.F. Azman1, P. Abdeshahian2, N.K.M. Salih3, M.G.Dashti4, H.Kamyab5, N.B.Esfahani6, A.A.Hamid1 and M.S.Kalil1 1. Department of Chemical and Process Engineering; National University of Malaysia (UKM), Malaysia; email: [email protected]; [email protected]; [email protected] 2. Faculty of Chemical Engineering, Universiti Teknologi Malaysia, Malaysia; email: [email protected] 3. Faculty of Applied Sciences and Foundation Studies, Infrastructure University Kuala Lumpur, Malaysia; email: [email protected] 4. Institute for Research in Molecular Medicine (INFORMM), Universiti Sains Malaysia, Malaysia; email: [email protected] 5. Centre for Environmental Sustainability and Water Security, Universiti Teknologi Malaysia, Malaysia; email: [email protected] 6. School of Nutrition and Food Science, Isfahan University of Medical Sciences, Iran; email: [email protected] Abstract Hydrogen generation was studied using palm oil mill effluent (POME) as agro-industrial waste biomass obtained from palm oil industry. POME was subjected to an acid treatment (HCl 37% v/v). POME hydrolysate obtained was used as a substrate for hydrogen generation. Hydrogen production was performed in a fermentative process by cultivation of new microbial strain Clostridium acetobutylicum YM1 on POME hydrolysate according to central composite design (CCD). CCD was constructed by considering three pivotal process variables, namely incubation temperature, initial pH of culture and microbial inoculum size. A second-order polynomial regression model was generated to adjust to CCD data. The empirical model analysis showed that linear and quadratic terms of temperature had a highly significant on hydrogen generation (P