Non-catalytic Hydrothermal Liquefaction of Biomass -

0 downloads 2 Views 407KB Size Report
Hydrothermal liquefaction (HTL) is a promising process to convert high moisture ... time affect biomass conversion and product yields during HTL, while ...

Available online at

ScienceDirect Energy Procedia 105 (2017) 75 – 81

The 8th International Conference on Applied Energy – ICAE2016

Non-catalytic hydrothermal liquefaction of biomass: An experimental design approach Flabianus Hardia,*, Mikko Mäkeläa,b, Kunio Yoshikawaa a

Tokyo Institute of Technology, Department of Environmental Science and Technology, G5-8, 4259 Nagatsuta-cho, Midori-ku, Yokohama 226-8502, Japan b Swedish University of Agricultural Sciences, Department of Forest Biomaterials and Technology, Skogsmarksgränd, Umeå 90183, Sweden

Abstract A face centered central composite design was used for determining the effects of reaction temperature (180–260 °C), reaction time (0–2 h) and sawdust concentration (9.1–25 wt%) on conversion and recovered product yields during hydrothermal liquefaction of pine sawdust. The determined conversion and aqueous product (AP) yield were in the range 23.1–57.2% and 14.6–43.4%, respectively. The results showed that all linear model terms and some interaction terms were statistically significant for sawdust conversion and liquid product yields, whereas all quadratic terms were found statistically insignificant. In general, increasing reaction temperature increased sawdust conversion and AP formation, while increasing sawdust concentration led to a reduction in conversion and liquid product yields. The determined model revealed that the decomposition reactions of lignocellulosic constituents may have occurred competitively, thus resulting in a certain trend in recovered AP and heavy oil (HO) yields. © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. ( responsibility of ICAE Selection peer-reviewofunder Peer-reviewand/or under responsibility the scientific committee of the 8th International Conference on Applied Energy.

Keywords: hydrothermal liquefaction, pine sawdust, central composite design



The development of biomass-based technology to produce chemicals, fuels and energy has become a necessity to counter the dependency on fossil fuels, meet energy demand and reduce carbon dioxide emissions [1]. Biomass is defined as the renewable organic material originating from plants, including their derivatives [2]. It is abundant and is considered as neutral for respective carbon emissions [1,3]. In particular, lignocellulosic biomass has great potential to be utilized since it does not compete with food resources [4,5].

*Corresponding author. Tel.: +81-45-924-5507 ; fax: +81-45-924-5518. E-mail address: [email protected]

1876-6102 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( Peer-review under responsibility of the scientific committee of the 8th International Conference on Applied Energy. doi:10.1016/j.egypro.2017.03.282


Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

Hydrothermal liquefaction (HTL) is a promising process to convert high moisture lignocellulosic biomass into liquid chemicals and fuels. The merits of liquid fuels are widely known due to their high energy content, transportability and applicability. Among other thermochemical conversion processes, HTL allows biomass to be processed without drying at a temperature of a few hundred degrees lower than the one of required for pyrolysis [3,6]. It is currently well-known that reaction temperature and reaction time affect biomass conversion and product yields during HTL, while adjusting the biomass concentration has correlated with energy savings for further upgrading steps (e.g. gasification of the HTL product) [7]. Experimental design offers an efficient and powerful method to evaluate multiple experimental variables, including their interactions [8–11]. In this study, a face centered central composite design at three different levels was utilized to determine the effects of reaction temperature, reaction time and sawdust concentration on the conversion and recovered product yields during HTL of pine sawdust.


Materials and Methods

2.1 Materials Pine sawdust obtained from Northern Sweden (600–710 μm) was dried overnight at 105 °C prior to use. Added water was purified using Milli-Q water purification (0.22 μm Millipak Express 40 filter). The elemental composition (dry ash-free) of pine sawdust was 50.5% carbon, 6.76% hydrogen and 42.8% oxygen. Filter paper Whatman No.3 was dried and pre-weighed before use. Acetone (min 99.5%, Wako Chemical, Japan) was used as received. High purity Argon (>99.9%, Tomoe Shokai) was used during the gas chromatography (GC) analysis and purging the reactor. 2.2 Methods The experimental run order was randomized to prevent systematic errors. In each experiment, 12 g of sawdust and a certain amount of water were loaded into a laboratory scale stirred batch reactor (400 ml effective volume, MMJ-500, OM Labo Tech, Japan) equipped with electric heating (average heating rate is 7 °C/min). The stirrer was set at 120 rpm to ensure adequate mixing. The reactor was sealed and the air inside was replaced by argon. The initial reaction pressure was atmospheric. During the heating, the system may enter the reactive temperature for biomass reaction. As soon as the reactor reached a predetermined target temperature, this temperature was maintained stable (±5 °C from the target temperature) for a certain reaction time. When the reaction time was complete, the reactor was cooled down (about 100 min) using electric fan until the temperature reached 35 °C. The heating and cooling period are excluded in the reaction time. The generated gas product was collected using a sampling bag and its volume was measured. The recovered liquid and solid products were separated by a two step filtration procedure [12]. During the first filtration step, 100 g of water was added for rinsing. The water-soluble filtrate, denoted as aqueous product (AP), was collected. The heavy oil (HO) sticking to the solid residue (SR) was collected during the second filtration step by rinsing the SR with 200 g acetone. The acetone in the HO mixture was removed by using a rotary evaporator. The remaining water and solvent in the SR and HO fractions were furthermore removed by means of oven drying (105 °C) for at least 15 h and 12 h, respectively. An unavoidable loss of light components might have occurred during the drying. 2.3 Feed and product analysis


Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

The CHNS and oxygen compositions of pine sawdust were measured using the Vario EL Cube (Elementar, Germany) and Vario Micro Cube (Elementar, Germany), respectively. The SR and HO yields were calculated from the direct weight measurements, while the AP yield was calculated as a weight difference between used sawdust, added water and recovered SR, gas and HO fractions. The pine sawdust conversion parameter was calculated based on Eq. (1): Conversion = 100% ė yield of SR


The weight of the gas was quantified based on respective volume and composition at 35 °C. The gas composition (H2, CO, CO2, CH4, C2H6, C2H4) was analyzed using a micro gas chromatograph (Micro GC Varian CP-4900) equipped with a TCD detector, MS5A column and PPQ column. 2.4 Experimental design The experimental conditions were determined based on a face centered central composite design including three different levels for each experimental variable (Table 1). A total of 17 experiments were performed including three replicated experiments at the design center. Table 1. The ranges of the experimental variables used in this study Reaction parameters


Reaction temperature (°C)


Sawdust concentration (wt%)


Reaction time (h)


Prior to model calculations, the real values of each experimental variable were transformed into coded values. The vector of measured response values, y, was fitted with the design matrix (X) using a general regression equation (Eq. 2)). The vector of coefficient values, b, was solved by minimizing the sum of squares of model residual, e (Eq.(3)): y = Xb + e


b = (X′X)-1X′Y


All linear model terms and only the significant interaction terms of b were included in the model calculations and final models. The mathematic calculation and the plotting were done using Matlab R2016a (The MathWorks, Inc., USA) and OriginPro 2016 (OriginLab, USA) software.


Results and Discussion

Table 2 shows the ranges of sawdust conversion and product yields obtained during the experiments that aimed to maximize recovered liquid products. The conversion varied in the range of 23.1–57.2%, while the HO and AP yields were found to be 2.34–14.1% and 14.6–43.4%, respectively. The conversion and yield obviously increased with increasing reaction temperature, while prolonging the reaction time resulted in very few improvements. A shorter reaction time had a beneficial effect on HO formation in


Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

most cases. Hence, the later simulation in this study was conducted at the shortest reaction time (0 h). The highest conversion and AP and HO yields were attained at low sawdust concentrations. The weight percentage of generated gas was found to be low and was more likely affected by reaction temperature and reaction time than sawdust concentration. Thus, instead of HO, AP was considered as the main liquid product. Table 2. Measured response values during the HTL experiments Measured responses

Range (wt%)

R2 of the model


23.1 – 57.2


HO yield

2.34 – 14.1


AP yield

14.6 – 43.4


Gas yield

0.22 – 5.54


Model coefficient values were calculated based on used experimental conditions. According to a performed t-test, all linear model terms (except sawdust concentration for the gas yield) and most of the interaction terms were statistically significant on a 95% probability, but all quadratic terms were found to be insignificant. Thus, all of the linear terms and only the significant interaction terms were included in the final models (Fig. 1). Data variation explained by a model was described by the R2 value shown in Table 2. To get clear insight on the profiles of sawdust conversion and recovered product yields, the predicted response values were plotted with a reaction time of 0 h based on the final models (Fig. 2). The conversion and AP yield can be increased not only by increasing the reaction temperature, but also by decreasing the sawdust concentration. The predicted HO yield has a unique trend since it increased by increasing the sawdust concentration only in the low temperature region (180–200 °C). Based on the models, the HO yield was unaffected at around 200–210 °C and eventually decreased by increasing the sawdust concentration at higher reaction temperatures. The gas yield had a positive correlation with reaction temperature but was decreased with increasing sawdust concentration. 12

Conversion HO yield AP yield Gas yield

Coefficient value

10 8 6 4 2 0 -2 -4 -6


Sawdust conc.

Linear terms


Temp. x Sawdust conc.

Temp. x Time

Interaction terms

Fig. 1. The coefficients of linear and interaction terms for the conversion and product yields with 95% probability


Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

(b) 25 25


30 35 40


45 50 55

10 180




Sawdust concentration (wt%)

Sawdust concentration (wt%)

(a) 25


20 10


10 6 220



(d) 25 Sawdust concentration (wt%)

Sawdust concentration (wt%)


Reaction temperature (oC)

15 20

20 25 30 15 35

10 180



Reaction temperature (oC)

(c) 25






Reaction temperature (oC)


2.0 1.5

20 1.0 15 0.5 0.0 10 180





Reaction temperature (oC)

Fig. 2. Model predictions for (a) biomass conversion; (b) HO yield; (c) AP yield; (d) gas yield (wt% basis) as functions of reaction temperature and sawdust concentration at a fixed reaction time of 0 h

Two kinds of liquid products (AP and HO) recovered in this study were affected in different ways by the used reaction parameters. Based on the previous work, the hydrolysis of hemicellulose in water starts at around 180 °C and is followed by that of cellulose mainly contributing to the determined AP and its further degradation to HO [13–15]. Meanwhile lignin, a highly aromatic polymer, is hardly decomposed at reaction temperatures lower than 250 °C, thus it contributes to small portion of HO [16]. A plausible reason for the AP and HO yield profiles is the competitive and parallel reactions of biomass decomposition to form AP and HO, mainly. At the reaction temperature below 200 °C, the competition between the hydrolysis reaction of hemicellulose and its further degradation reactions were controlled by sawdust concentration. Therefore, the reverse trend between AP yield and HO yield was observed. On the other hand, as the first step of the hydrolysis was finished, the degradation of the rest components occurred, thus resulting in parallel profiles for the AP and HO yields. The degradation of cellulose contributed to AP yield, while the lignin decomposition likely formed HO, although this needs to be confirmed further by means of compositional analysis. 4.



Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

The effects of three reaction parameters on HTL of pine sawdust were determined and simulated using an experimental design. Determined sawdust conversion was in the range 23.1–57.2%, while the AP yield was 14.6–43.4%. The effects of reaction temperature, sawdust concentration and reaction time were all statistically significant for sawdust conversion and liquid yields. The plausible explanation for the different profile trends of AP and HO was found from the competitive and parallel reactions of lignocellulosic components at certain reaction temperatures. The determined models allow the prediction of HTL product yields for further upgrading steps (e.g. gasification). Acknowledgements This project was supported by JSPS and STNT Japan–Sweden Research Cooperative Program. References [1] McKendry P. Energy production from biomass (part 1): Overview of biomass. Bioresour Technol 2002;83:37–46. [2] Demirbaş A. Biomass resource facilities and biomass conversion processing for fuels and chemicals. Energy Convers Manag 2001;42:1357–78. [3] Zhang L, Xu C (Charles), Champagne P. Overview of recent advances in thermo-chemical conversion of biomass. Energy Convers Manag 2010;51:969–82. [4] Cherubini F. The biorefinery concept: Using biomass instead of oil for producing energy and chemicals. Energy Convers Manag 2010;51:1412–21. [5] Demirbas A. Competitive liquid biofuels from biomass. Appl Energy 2011;88:17–28. [6] Doassans-Carrère N, Ferrasse JH, Boutin O, Mauviel G, Léde J. Comparative study of biomass fast pyrolysis and direct liquefaction for bio-oils production: Products yield and characterizations. Energy and Fuels 2014;28:5103–11. [7] Karagöz S, Bhaskar T, Muto A, Sakata Y, Uddin MA. Low-temperature hydrothermal treatment of biomass: Effect of reaction parameters on products and boiling point distributions. Energy and Fuels 2004;18:234–41. [8] Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA. Response surface methodology (RSM) as a tool for optimization in analytical chemistry. Talanta 2008;76:965–77. [9] Mäkelä M, Benavente V, Fullana A. Hydrothermal carbonization of lignocellulosic biomass: Effect of process conditions on hydrochar properties. Appl Energy 2015;155:576–84. [10] Gan J, Yuan W. Operating condition optimization of corncob hydrothermal conversion for bio-oil production. Appl Energy 2013;103:350–7. [11] Leardi R. Experimental design in chemistry: A tutorial. Anal Chim Acta 2009;652:161–72. [12] Xu, C., Lad N. Production of heavy oils with high calorific values by direct liquefaction of woody biomass in sub/near critical water. Energy Fuels 2007;22:635–42. [13] Gao Y, Wang H, Guo J, Peng P, Zhai M, She D. Hydrothermal degradation of hemicelluloses from triploid poplar in hot compressed water at 180-340 °C. Polym Degrad Stab 2016;126:179–87. [14] Mok WSL, Antal MJ. Uncatalyzed Solvolysis of Whole Biomass Hemicellulose by Hot Compressed Liquid Water. Ind Eng Chem Res 1992;31:1157–61. [15] Minowa T, Zhen F, Ogi T. Cellulose decomposition in hot-compressed water with alkali or nickel catalyst. J Supercrit Fluids 1998;13:253–9. [16] Tymchyshyn M, Xu CC. Liquefaction of bio-mass in hot-compressed water for the production of phenolic compounds. Bioresour Technol 2010;101:2483–90.

Flabianus Hardi et al. / Energy Procedia 105 (2017) 75 – 81

Biography Flabianus Hardi is a PhD candidate in Environmental Science and Technology Department, Tokyo Institute of Technology, Japan. His research interest is thermochemical conversion of biomass for biofuel production.


Suggest Documents