International Journal of Green Energy Sunflower Oil Fuel for Diesel ...

3 downloads 1908 Views 2MB Size Report
Dec 12, 2013 - office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK. International Journal of Green Energy. Publication details, including ...
This article was downloaded by: [46.177.89.34] On: 11 January 2014, At: 05:53 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Green Energy Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ljge20

Sunflower Oil Fuel for Diesel Engines: An Experimental Investigation and Optimum Engine Setting Evaluation Using a MultiCriteria Decision Making Approach a

b

b

A. T. Balafoutis , E. Papageorgiou , Z. Dikopoulou , S. Fountas & G. Papadakis

a

a

Department of Natural Resources and Agricultural Engineering , Agricultural University of Athens , Athens , Greece b

Department of Informatics and Computer Technology , Technological Educational Institute of Lamia , Lamia , Greece Accepted author version posted online: 20 Jun 2013.Published online: 12 Dec 2013.

To cite this article: A. T. Balafoutis , E. Papageorgiou , Z. Dikopoulou , S. Fountas & G. Papadakis (2014) Sunflower Oil Fuel for Diesel Engines: An Experimental Investigation and Optimum Engine Setting Evaluation Using a Multi-Criteria Decision Making Approach, International Journal of Green Energy, 11:6, 642-673, DOI: 10.1080/15435075.2013.777912 To link to this article: http://dx.doi.org/10.1080/15435075.2013.777912

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

a

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

International Journal of Green Energy, 11: 642–673, 2014 Copyright © Taylor & Francis Group, LLC ISSN: 1543-5075 print / 1543-5083 online DOI: 10.1080/15435075.2013.777912

SUNFLOWER OIL FUEL FOR DIESEL ENGINES: AN EXPERIMENTAL INVESTIGATION AND OPTIMUM ENGINE SETTING EVALUATION USING A MULTI-CRITERIA DECISION MAKING APPROACH A.T. Balafoutis1 , E. Papageorgiou2 , Z. Dikopoulou2 , S. Fountas1 , and G. Papadakis1 Downloaded by [46.177.89.34] at 05:53 11 January 2014

1

Department of Natural Resources and Agricultural Engineering, Agricultural University of Athens, Athens, Greece 2 Department of Informatics and Computer Technology, Technological Educational Institute of Lamia, Lamia, Greece An experimental investigation on a diesel engine was conducted to evaluate the performance and exhaust emissions of sunflower oil and three blends with diesel fuel (20, 40, and 70% oil content volumetrically) in comparison to diesel fuel. Three injection timing and three injector protrusion settings were tested to study engine performance and exhaust emissions. The work was conducted in a direct injection agricultural tractor engine. Engine operation with sunflower oil-based fuels was unproblematic during the short-term experiments. Torque, brake-specific fuel consumption and NOx were enhanced as oil content was increased in the tested fuel. Early injection timing improved torque output, reduced BSFC, increased thermal efficiency, and NOx emissions. Deep injector protrusion increased torque release in low oil content fuels and shallow injector protrusion and increased torque in high oil content fuels. The experimental results were evaluated using two multi-criteria decision-making techniques (Analytical Hierarchy Process-AHP and Technique for Order Preference by Similarity to the Ideal Solution-TOPSIS) and the optimal fuel type-injection timing-injector protrusion configuration was selected. AHP and TOPSIS were run for three groups of criteria that were focusing on higher engine performance, lower environmental impact, and a balance between the first two, respectively. The results of the two techniques were compared. AHP and TOPSIS gave the same attribute ranking in all three groups of criteria, but did not give the same classification of Fuel/Injection Timing/Injector Protrusion configuration. The 70% oil content blend was selected from both techniques as optimal in the examined groups of criteria. Keywords: Sunflower oil; Injection timing; Injector protrusion; Performance; Emissions; Ranking; AHP; TOPSIS

INTRODUCTION Agriculture has been developed as an economical sector where mechanization plays a significant role in the final output. Agricultural tractors have been the main driver on Address correspondence to A. T. Balafoutis, Agricultural University of Athens, Department of Natural Resources and Agricultural Engineering, 75 Iera Odos Street, GR 11855, Athens, Greece. E-mail: abalafoutis@ aua.gr 642

Downloaded by [46.177.89.34] at 05:53 11 January 2014

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

643

production maximization and they count for 90–95% of the total field agricultural energy consumption (Panagakis et al. 1996), due to high fuel consumption, derived by high power needs during field operations. Unfortunately, their fuel consumption and environmental impact were not considered as significant, because in general agricultural sector stands for only 3.6% in EU energy profile (EEA 2011) and it was very low in comparison to other energy consumers (i.e., industry, transportation). Nevertheless, in accordance to the EU target for 10% biofuels of total transportation fuel consumption until 2020 (2009/28/EC 2009), agricultural tractors should also contribute their share. Biofuels have usually agricultural origin offering environmental benefits (i.e., exhaust emission reduction, non-toxicity), as well as decreasing countries’ dependency on fossil fuels and improving agricultural income. Biofuels to be used for diesel engines, which agricultural tractors use, are vegetable oils (VOs) and their derived methyl esters (biodiesel). Biodiesel from different feedstock have been widely investigated with very positive results in the final product quality (Özcanlı, Keskin, and Aydın 2011). Hence, biodiesel have been a part of the EU and global fuel market for almost a decade. The use of edible VOs for the preparation of biodiesel is debatable due to its conflict with food production, resulting in various attempts to produce biodiesel from non-edible oils (Rao et al. 2007; Amarnath and Prabhakaran 2012) and waste products like pork lard (John Panneer Selvam and Vadivel in Press) and waste cooling oil (Mumtaz et al. 2012). In the recent years, there has been an attempt to combine biodiesel with other fuel types, like bioethanol in blends (Subbaiah and Gopal 2011) and pyrolysis oil (Prakash, Singh, and Murugan 2012) or additives like methanol, ethanol, distilled water and diethyl ether (Vedaraman et al. 2011), and butanol (Liu et al. in Press). As agricultural income could be increased by using fuels produced within the farm, VOs could have an asset in comparison to biodiesel, which has increased production cost, and it is mainly distributed through the traditional fossil fuel market focusing on passenger cars and trucks. From an environmental point of view, the advantage of VOs in comparison to DF is that when combusted in an internal combustion engine there are zero CO2 emissions. However, looking at the whole production line of VOs, there are CO2 and Greenhouse Gas (GHG) emissions, which are calculated using Life Cycle Assessment (LCA) methodology (IPCC 2007) and they should be considered in the final calculation of the produced GHG emissions of each fuel. Nevertheless, even considering this amount of GHG production, there is a positive environmental impact of VOs use as fuels compared to DF (Balafoutis et al. 2010). Many VOs have been considered as diesel fuel (DF) substitutes and various oily seed crops have been tested for this purpose. In Southern Europe, sunflower is the most cultivated oily crop and could be used for in-farm sunflower oil (SO) production and consumption (Kallivroussis Natsis, and Papadakis 2002). VOs utilization as fuel has shown several problems, such as injector cocking, ring sticking, gum formation and lubrication oil contamination and they are mainly related to high viscosity and low volatility of VOs that produce poor atomization and decrease combustion quality (Shehata and Razek 2011). VOs have been proven experimentally that can be used as fuels directly in existing diesel engines, by adapting different methodologies for significant reduction of their viscosity (main VO drawback) to reach DF levels. The most applicable methods to reduce VO viscosity are either the preparation of blends with DF (Agarwal, Kumar, and Agarwal 2008) or the preheating of the VO (Nwafor 2003) or the combination of the two methods (Agarwal and Rajamanoharan 2009). All methods showed

Downloaded by [46.177.89.34] at 05:53 11 January 2014

644

BALAFOUTIS ET AL.

positive results, with the combination of VO preheating and blending with DF being the most appropriate especially when VO percentage was high. Diesel engines are tuned for optimum operation with DF as fuel, which indicates that in order to optimize the engine outcome with VOs as fuel, modified engine tuning should be applied. Engine performance depends on the fuel used, the fuel injection, and the combustion characteristics (Ryan Callaham, and Dodge 1982; Strayer, Craig, and Zoerb 1982). In this study using SO, fuel injection type was modified, which consequently alters the combustion characteristics. There are several methods to change the fuel injection type (fuel pump pressure, injector or nozzle type change, etc.), but spare parts replacement is involved, which on an economical perspective, does not combine with the interest of a farmer. Consequently, other low cost and easy engine setting modifications with significant effect on final combustion characteristics had to be adapted. VOs exhibit longer ignition delay periods and slower burning rate than DF and, as a result, the injection timing (IT) of the fuel should be advanced to provide more time for the VO to be combusted and produce better performance results (Yahya and Marley 1994; Nwafor, Rice, and Ogbonna 2000). However, this advancement was shown to have significant negative effect on the emitted NOx (Shuai et al. 2009). As a result, altering IT was shown to have significant impact on engine operation and is additionally easy to be modified. Piston bowl design in direct injection (DI) diesel engines has significant effect on the combustion characteristics inside the combustion chamber, as it affects the total chamber design. The design process should focus on optimum filling and emptying of the cylinder with fresh unburned charge over the engine operating speed range and create the appropriate cylinder conditions for optimum air/fuel mixing, in order to achieve fuel combustion in the shortest time (Heisler 1999). There are several parameters on piston bowl design (throat diameter, maximum bowl diameter, bowl depth, central pip, main toroidal radius, minor radii, and impingement point) that influence significantly the combustion process (Heywood 1988). However, fundamental piston design interference would be appreciated for new engine production and do not cover the targets of this paper, which is conventional farm tractor modification. The only exception is fuel point of impingement that can be easily altered when changing the injector protrusion (IP) into the chamber by replacing the injector washer thickness. Moreover, the position and geometry of the fuel spray impingement on the bowl vertical surfaces showed large impact on the combustion and emissions performance, due to amendment of the fuel distribution within the bowl and clearance area (Melas 2003). Hence, IP was considered as a parameter that could help on optimizing engine operation with VOs as fuel. Sunflower oil (SO) was selected to be the appropriate VO for in-farm production and consumption under Greek conditions. A deterministic LCA was conducted for SO production in a typical Greek sunflower farm to calculate the real fossil fuel GHG emissions that is produced for each of the tested SO-based fuels. IT and IP were identified as engine operation influential parameters that could be used for engine setting optimization. Therefore, in this work, an agricultural tractor engine was run with pure preheated SO and three preheated SO/DF blends (20/80, 40/60. 70/30 volumetrically) using three IT settings (manufacturer IT, ±2◦ Crank Angle) and three IP settings (manufacturer IP, ±0.5 mm), resulting in 37 Fuel-IT-IP configurations including the reference diesel fuel with manufacturer settings for IT and IP. The selection of blends was based on previous studies (Nwafor and Rice 1996; McDonnell et al. 2000; He and Bao 2005; Agarwal and Rajamanoharan 2009; Venkanna, Swati, and Reddy 2009) and aimed on an overall

Downloaded by [46.177.89.34] at 05:53 11 January 2014

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

645

blend range. The IT was changed by only ±2◦ CA to avoid erratic behavior (Nwafor, Rice, and Ogbonna 2000) and IP was changed by ±0.5 mm, due to injector washer specifications. In view of the number of parameters monitored, multivariate data analysis techniques was applied, in order to rank the engine settings under investigation. Analytical Hierarchy Process (AHP) and Technique for Order Preference by Similarity to Ideal Situation (TOPSIS) methods were applied, as they have found a large number of applications in many domains. In the energy sector, a number of applications have been found (Nigim, Munier, and Green 2004; Pohekar and Ramachandran 2004; Zhou, Ang, and Poh 2006). A new methodological framework of multi-participatory and multi-criteria decision-making (MCDM) was proposed to evaluate renewable energy options (Polatidis and Haralambopoulos. 2004) . The problem of selecting the most suitable electricity generation alternative for Turkey was dealt using an integrated decision aid framework using MCDM (Topcu and Ulengin 2004). Recently, different MCDM techniques were implemented to determine the best alternative technique in Turkey energy planning (Kahraman and Kaya 2010; Kaya and Kahraman 2011). A thorough description on MCDM methods for strategic planning in sustainability energy assessment was also presented highlighting the multi-dimensional nature of sustainability issues in strategic public energy plans (Laes and Verbruggen 2010). The aim of the present work was, therefore, to study how an agricultural tractor diesel engine could be optimized in terms of performance and emissions to work on SO-based fuels using simple and low-cost engine tuning (IT and IP) and to rank the Fuel-IT-IP configurations using MCDM techniques. MATERIALS AND METHODS Experimental Apparatus The experimental apparatus used for the purpose of this work is described in detail in reference (Balafoutis et al. 2011) and shown in Figure 1. The engine selection was based on the average tractor engine power in Greek agriculture (Mourtzinis, Fountas, and Gemtos 2007) and its specifications are listed in Table 1. The engine was coupled on a test bench, using a hydraulic dynamometer to measure load, a proximity speed sensor to measure the engine speed and a burette and a stopwatch for fuel consumption measurement. Temperatures in all cylinders, inlet and outlet coolant, air filter, exhaust gas, air inlet manifold (after turbo charger), fuel line, and engine oil were recorded using a series of k-type thermocouples. Pressures in air inlet manifold, exhaust manifold, and fuel line were also recorded, using pressure transducers. Finally, airflow before the air filter was being recorded using an insertion calometric sensor. All sensors were connected to data loggers and data were stored. A VO preheating kit based on heat exchange between the engine coolant and the alternative fuel was installed for the preheating of the SO-based fuels during the experimental work. Details on the kit are shown in the study by Balafoutis et al. 2011. An exhaust gas analysis was carried out measuring NOx and CO2 with an MRU Delta 1600L gas analyzer (Table 2). Experimental Procedure After warming up the engine using DF, until it reached operating temperature, the alternative fuel [pure preheated SO and three preheated SO/DF blends (20/80, 40/60.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

646

BALAFOUTIS ET AL.

Figure 1 Experimental rig (1–hydrokinetic dynamometer, 2–engine, 3–gas analyzer, 4–air filter, 5–fuel pump, 6–burette, 7–diesel fuel tank, 8–alternative fuel tank, 9–air flow meter, 10–engine oil temperature, 11–exhaust gas temperature, 12–engine speed meter). Table 1 Engine Specifications Make

Case New Holland

Type No of cylinders Aspiration Cooling system Stroke (mm) Bore (mm) Engine capacity (cm3 ) Compression ratio Type of fuel pump

Four-stroke direct injection diesel engine 4 Turbocharged Water cooled 132 104 4485 17.5 RDP

Table 2 Accuracy of Measurements and Uncertainty of Computed Results Measurement CO CO2 NO NO2 HC O2

Unit

Measurements range

Accuracy

% vol. % vol. ppm ppm ppm n-hexane % vol.

0–15 0–20 0–5000 0–1000 0–20000 0–25

0.01 0.01 1 1 1 0.1

70/30 volumetrically)] was tested and after the experimental procedure the fuel was again switched to DF, to flush out the fuel lines, the injection pump, and the injectors before shutting down. The engine load was set in the point of maximum torque release, which was detected by small changes around the given manufacturer maximum torque engine speed. In every test, power, torque, fuel consumption, engine temperatures and pressures, airflow, and gas emissions were measured. Atmospheric pressure and air temperature were also recorded

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

647

and measured torque was corrected according to them. At each test, the engine was stabilized for 3 min and then the measurements were taken. Each test was conducted three times and the average value of each measurement was taken for eliminating statistical errors. The procedure that was followed has been described minutely in the study by Balafoutis et al. 2011.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Greenhouse Gases Life Cycle Assessment In order to calculate the amount of CO2 produced by combusting each of the five fuels of this study (three SO/DF blends, pure SO, and conventional DF), it was important to consider a full life cycle assessment (LCA) accounting for direct and indirect (also called hidden) Greenhouse Gases (GHG) emissions (Malça and Freire 2006; Malca and Freire 2009) through the whole SO production (Figure 2). This would enable the calculation of the total avoided GHGs. There are several GHGs that contribute in global warming. Though, in this study we took into account only Carbon Dioxide (CO2 ), Methane (CH4 ), and Nitrous Oxide (N2 O), as they are more relevant to biofuel’s production and other gases are in reduced quantities. According to Life Cycle Impact Assessment (LCIA) methodology, all GHGs are aggregated in kg of CO2 equivalent, using the corresponding global warming potential (GWP) for a 100-year horizon, which is CO2 = 1, CH4 = 25, and N2 O = 298 (IPCC 2007). The most commonly used time horizon (100 years) for GWP estimation was chosen considering the short-term to mid-term implications of the present study in terms of global warming effect. Fuel-IT-IP Configuration Ranking Using Multiple-Criteria Decision Making Methods An application of Multiple-Criteria Decision-Making (MCDM) approach to the problem of selecting the best Fuel-IT-IP configuration for the engine of this study was

Fuel Seed Fertilizers Pesticides

Fuel

Electricity

Transportation

Cold Pressing

Sunflower Oil

Electricity

Cultivation Procedure

Sunflower Meal

Figure 2 Flow chart of the life cycle chain of sunflower oil (color figure available online).

Downloaded by [46.177.89.34] at 05:53 11 January 2014

648

BALAFOUTIS ET AL.

elucidated. MCDM is a well-known method and used to find the best selection from all of the possible alternatives in the presence of multiple decision criteria (Isiklar and Büyüközkan 2007). Two popular MCDM methods, Analytic Hierarchy Process (AHP) and Technique for Order Preference by Similarity to the Ideal Solution (TOPSIS) were applied for ranking the engine settings. AHP technique, which was developed by Saaty (Saaty 1980), is a MCDM technique, exporting the expert knowledge. The AHP develops a structure to collect all the important factors and choose the best option. The main advantage of this method is the ability to manage a large number of different factors (qualitative and quantitative data) in order to make a decision (Saaty 2004). AHP decomposes a complex problem into smallest individual problems (criteria), which can be solved separately. The criteria are analyzed until they can be used effectively for the comparison of alternative solutions or decision alternatives. The AHP uses as input the answers from the decision maker to a series of questions of the general form: “How important is criterion A relatively to criterion B?” or “How important is attribute A relatively to attribute B with respect to criterion A?” These are called pairwise comparisons and are used to create (a) weights for attributes and (b) performance scores for alternatives on the different criteria (Janic and Reggiani 2002). In Appendix A, the steps of AHP process for determining the best attribute are presented. TOPSIS technique, which was proposed by Hwang and Yoon (Hwang and Yoon 1981), selects the best alternative, which is nearest to the positive ideal solution and farthest from the negative ideal solution. TOPSIS flows from the concept of selecting best alternative or ranking the alternatives that have the shortest distance from the positive ideal solution (PIS) and the farthest from the negative ideal solution (NIS) in a geometrical (Euclidean) sense (Karimi, Yusop, and Law 2010). In TOPSIS method, all attributes are considered simultaneously regarding the distances to both PIS and NIS (Shih 2008). The ideal and non-ideal solutions are worked out by using normalized matrix. Next, the Euclidean attributes distances from the ideal and non-ideal points are calculated and relative closeness to the ideal solution is obtained, which is in range of 0 to 1. If the examinee solution approaches to 1, then it is close to the ideal point; otherwise if it approaches to 0 then it is close to the non-ideal point (Li Hao 2006). In TOPSIS method, the importance of every index is not equal, so they must be set different weight factors. The weight factors are very important because they directly influence the results of output. There are many weights to set the weight factors, like in AHP (Li Hao 2006). TOPSIS uses m attributes (options) and n criteria and gets as input the score of each option with respect to each criterion. The steps of this method are depicted in Appendix B. In this study, the evaluation of the Fuel-IT-IP configurations was executed using the opinion of fuel-engine experts. There were five attributes derived from the engine tests (Torque, BSFC, Thermal Efficiency, NOx , and CO2 ). The motivation criteria for fuel-engine tuning had to be set, for the MCDM models to prioritize parameters regarding their importance. Therefore, three main criteria were determined using these five attributes. The first criterion aimed in high engine performance with limited attention in gas emissions (Criterion 1), the second aimed in low gas emissions with limited attention in engine performance (Criterion 2), and the third was an intermediate situation between the two first extreme cases (Criterion 3). Additionally, two auxiliary criteria referring in the new EU policy on engine emissions (Criterion 4) and in the new engine design trends (Criterion 5) were selected. Furthermore, a questionnaire of five questions corresponding to the five criteria was developed and evaluated using the five attributes defined above (Figure 3). Each of the attributes was assessed by the experts giving a grade according

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

649

Fuel-IT configuration ranking

Group 1

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Criterion 1: Increased Engine Performance

Group 2

Criterion 2: Reduced environmental effects

Group 3

Criterion 3: Balanced between engine and environment

Criterion 4: Expected EU legislations

Criterion 5: New trends in engine design

Torque

Torque

Torque

Torque

Torque

T.E.

T.E.

T.E.

T.E.

T.E.

BSFC

BSFC

BSFC

BSFC

BSFC

NOx

NOx

NOx

NOx

NOx

CO2

CO2

CO2

CO2

CO2

Figure 3 The hierarchical structure of Fuel-IT configuration ranking (3 Groups, 5 Criteria, 5 attributes for each criterion).

to its importance. The grading was 1–5, with 1 counting for low importance and 5 for high importance. Nine fuel-engine experts (USA: 1, Spain: 1, Italy: 1, Germany: 2, Denmark: 1, Greece: 3) answered the questionnaire and the results were used for both AHP and TOPSIS models. In order to achieve the overall goal of ranking the Fuel-IT-IP configurations, it was decided to study the five criteria into three groups. Specifically, each group included the combination of a main criterion (Criteria 1,2,3) with the two auxiliary criteria (Criteria 4,5). In particular, the three groups were Group 1 (Criteria 1–4–5), Group 2 (Criteria 2–4–5), and Group 3 (Criteria 3–4–5). It is normal to cluster criteria in a value of three (www. communities.gov.uk 2009). These combinations were selected as it was considered important to combine the main criteria of fuel-engine tuning together with the modern perception on engine development (technical and legal). These three groups were used in the analysis both for AHP and TOPSIS techniques (Figure 3).

RESULTS AND DISCUSSION Greenhouse Gases Life Cycle Assessment A typical sunflower farm in northern Greece, where sunflower is mainly grown, was chosen and an inventory of all inputs and their respective GHGs (materials, fuels, electricity) is presented in Table 3. The oil extraction GHG contribution due to electricity consumption was calculated for the oil expeller used in this study that consumed 0.229 MJ/kg of sunflower seed. The total GHG emissions produced by SO production

650

BALAFOUTIS ET AL.

Table 3 Agricultural and Industrial Inputs of the Sunflower Farm

CO2

CH4

N2 O

Unit GHGs

3.5 45 20

316 2620 1200

0 13.8 2.38

1 0.118 0.235

g/kg g/kg N g/kg P2 O5

2.15 135 265.9

1.18 74.4 14.6

Kg/ha Kg/ha

2 80

179 3423

0.004 0.18

0.016 0.047

g/kg g/kg

107.5 275.3

5.92 151.7

MJ/ha

1867.62

283

0.173

g/MJ

536.6

295.7

986.4

543.5

Input

per ha

Seed N fertilizer P2 O5 Fertilizer Pesticide Diesel Fuel Electricity

Kg/ha Kg/ha Kg/ha

Sunflower farm

Total GHG Energy Allocation (kgCO2 eq/ha) (gCO2 eq/ha)

Downloaded by [46.177.89.34] at 05:53 11 January 2014

TOTAL

Table 4 Agricultural and Industrial Outputs of the Sunflower Farm Output Seed Oil content∗ Oil Meal ∗ from

kg % kg kg

1800 41 738 1062

Energy content (MJ/kg)

Total energy content (MJ/ha)

39.4 19.6

29.08 20.8

mechanical extraction (60–70% of total oil content in seed)

was 986.4 kgCO2 eq/ha. However, according to EU RED Directive of 2009, energy allocation has to be applied when LCA is used for biofuel production systems (2009/28/EC 2009). Hence, energy allocation has been used and only 55.1% of the total burdens were allocated to SO. Therefore, the final GHG emission produced by SO production was 543.5 kgCO2 eq/ha. On the other hand, sunflower farm outputs (seed, SO, meal) are illustrated in Table 4 together with their energy content. The GHG emissions that reflected to the production of every MJ of SO (18.23 gCO2 eq/MJ) was calculated by dividing the total GHG emissions/ha (543.5 gCO2 eq/ha) with the energy content of SO/ha (29.08 GJ/ha). When comparing this value with the fossil diesel respective GHG emissions (84 gCO2 eq/MJ) (Malca and Freire 2009; Malça and Freire 2006), we came to the conclusion that SO produced 21.68% of the CO2 that DF would emit for the same energy content. This significant difference of GHG emission production between SO and DF gave an extra bonus to the fuels that contained SO. Therefore, CO2 production of each blend were calculated and further used as percentage of the CO2 production of DF (the 20/80 blend emitted 85.1% of the respective value of DF, the 40/60 blend 69.02%, the 70/30 44.22%, and pure SO 21.68%).

Engine Performance In order to evaluate the alternative fuels in terms of engine performance, a comparison of the five fuel types (three SO/DF blends, pure SO, and conventional DF) was conducted, in terms of the three first attributes related to engine performance, namely Torque output, Brake Specific Fuel Consumption (BSFC), and total Thermal Efficiency (TE). In Figures 4–6 each engine performance attribute is presented in a set of nine bars

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

651

14 10

IT-2IP0

8

IT-2IP-0.5

6

IT-2IP+0.5

4

IT+2IP0

2

IT+2IP-0.5

0

IT+2IP+0.5

–2

IT0IP0

–4

IT0IP-0.5

–6

IT0IP+0.5

Figure 4 Comparison of Torque between the 37 Fuel-IT-IP configurations.

20

15 B.S.F.C. difference from DF (%)

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Torque difference from DF (%)

12

IT-2IP0 IT-2IP-0.5

10

IT-2IP+0.5 IT+2IP0

5

IT+2IP-0.5 IT+2IP+0.5 IT0IP0

0

IT0IP-0.5 IT0IP+0.5

–5

–10 Figure 5 Comparison of BSFC between diesel fuel and the three oil types.

presenting the 9 IT-IP combinations for the three SO/DF blends and pure SO, where every parameter is shown as a difference percentage from the respective DF reference value. Torque. The engine torque variations using SO fuels are presented in Figure 4. A general observation from all nine IT-IP combinations is that as the oil content was increased, torque outcome was also increased. There are several reasons for such a result. This could be explained by the higher density and viscosity and oxygen content of SO in comparison to DF (Kalam Husnawan, and Masjuki 2003). In particular, SO density is approximately 13% higher than DF and since the fuel is sent to the chamber on volumetric basis, the total fuel mass pumped into the cylinders with the SO is larger. Moreover, the fact that SO contain oxygen, improves combustion, which is an important factor for the torque increase observed in this study. Another reason for higher engine performance could be

Th. Efficiency difference from DF (%)

652

BALAFOUTIS ET AL.

6 4 IT-2IP0 2

IT-2IP-0.5

0

IT-2IP+0.5 IT+2IP0

–2

IT+2IP-0.5 –4

IT+2IP+0.5

–6

IT0IP0

–8

IT0IP-0.5

Downloaded by [46.177.89.34] at 05:53 11 January 2014

IT0IP+0.5

Figure 6 Comparison of Thermal Efficiency between diesel fuel and the three oil types.

that higher viscosity of the SO fuels reduced internal leakages (Al-Widyan, Tashtoush, and Abu-Qudais 2002; Wagner, Clark and Schrock 1984). Finally, combustion is always enhanced by high mixture of air and fuel in the chamber before combustion is initiated, as it provides good atomization and spray characteristics that boosts combustion process (Venkanna, Swati and Reddy 2009). Therefore, SO based fuels that contain high content of SO (unsaturated oil with single bonds between molecules) improve mixing and performance (Puhan et al. 2010). According to Figure 4, there are significant differences between the 9 IT-IP configurations. In terms of IT alteration, it was shown that early IT (IT+2◦ CA) increased Torque output for all SO fuels in comparison to the manufacturer’s IT (IT0), when late IT (IT-2◦ CA) indicated a much lower torque output. This phenomenon could be explained by the fact that the early IT setting of the engine gave SO more time to complete its combustion, as it always has a higher ignition delay and combustion rate (Nwafor, Rice, and Ogbonna 2000). In addition, VOs have beneficial effect on the combustion process when it is mixing controlled (Desantes et al. 1999), something which seem to be the case in this study as the engine is turbocharged with high compression ratio (high temperatures and pressures in the air/fuel mixture in the cylinder) (D. C. Rakopoulos et al. 2011). Nonetheless, the oil content effect was more influential in the final torque release than the late IT setting, illustrated in the increased torque of all tested fuels, except SO20. In this case, the effect of late IT on the engine torque was much more significant than the oil content, which was demonstrated in lower torque release than DF. This phenomenon indicated that late IT does not allow SO fuels to combust properly due to lack of time, which is aggravated in fuels with low SO content. Regarding IP impact on torque output, there is a trend of deep protrusion (IP– 0.5 mm) to be dominant in low oil content fuels (SO20 and SO40) and of shallow protrusion (IP+0.5 mm) to lead in high oil content fuels (SO70 and SO). This result could be explained by the fact that VOs have larger droplet size due to their high kinematic viscosity and if the injector was in the same protrusion as with DF using the same injection pressure, then the impingement point on the wall of the piston bowl would be in a lower point due to gravity. A lower impingement point would increase the delivered fuel on the piston bowl floor, causing poor mixing with air due to lack of air near the center of the

Downloaded by [46.177.89.34] at 05:53 11 January 2014

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

653

combustion chamber (Zhu, et al. 2004). Therefore, the engine would rather behave better with high oil content fuels when IP+0.5 mm was used. Brake-Specific Fuel Consumption (BSFC). Brake-specific fuel consumption (BSFC) was found to be in all cases, except one (SO20-IT+2-IP0 configuration), higher than with DF (Figure 5). It is also illustrated that as oil content was increasing, BSFC was also increased. This reaction could be explained by the higher density of SO, which leads to increased gravimetric fuel consumption due to increased fuel mass for the same volume of fuel injected by the RDP fuel pump (Puhan et al. 2007). In addition, lower calorific value might also increase the volumetric fuel consumption to keep similar energy input to the engine (Agarwal and Agarwal 2007; Fontaras et al. 2006). In the case of SO20-IT+2-IP0, BSFC was lower than DF that was attributed to higher thermal efficiency, thus less fuel consumption for the same load constraint (D. C. Rakopoulos et al. 2011). The impact of IT on BSFC had a clear trend in favor to early IT (IT+2◦ CA), which decreased BSFC significantly in comparison to the manufacturer setting of IT0 and late IT (IT-2◦ CA). This improvement in fuel consumption with the early IT was thought to be due to increased time availability for the slow burning rate of SO that the extra 2◦ CA gave to the fuel (Nwafor, Rice, and Ogbonna 2000). In terms of IP, the impact on BSFC was blear. The least BSFC with early IT (IT+2◦ CA) was found with IP0, which indicated that IT impact was very significant in comparison to IP. In the case of late IT (IT-2◦ CA), which gave the worst BSFC results with all blend types, the best IP was IP+0.5 mm. Injection was executed 2◦ CA ATDC and the impingement point using this IP was on the highest point of the bowl wall. Therefore, fuel droplets were distributed more to the cylinder head side, took advantage of the air in this part of the chamber, and counteracted partially the time lose for the combustion process due to late IT. Finally, when the engine run with IT0, it was shown that in low oil content fuels (SO20 and SO40) IP0 prevailed, when in high oil content fuels (SO70 and SO) IP–0.5 mm gave the least BSFC. Thermal Efficiency (TE). In Figure 6, there is an illustration of TE difference of the 36 Fuel-IT-IP configurations from conventional DF. It is noticeable that in most cases TE was lower than DF. On the other hand, this reduction was up to 6.5 % that kept TE in acceptable levels for diesel engines. The reason for such a reduction was that even if brake power output was increased with most of the configurations (numerator), there was simultaneous fuel consumption increase (denominator), which had more significant influence on the final TE outcome. However, it is worth mentioning that all early IT configurations (IT+2◦ CA) have shown the best TE results, which was due to higher power production in combination to lower fuel consumption. Especially, the IT+2-IP0 configurations, with all fuel types, have given TE increase between 2.5–4.5%.

Engine Emissions Investigation of performance characteristics has to be combined with engine emissions examination, in order to have an overall view of SO as fuel in comparison to DF. Two engine emissions attributes were measured to be used in the selection analysis, the nitrogen oxides (NOx ) and the carbon dioxide (CO2 ). Nitrogen Oxides (NOx ). Figure 7 presents the difference of NOx emissions between the 36 Fuel-IT-IP configurations from conventional DF.

654

BALAFOUTIS ET AL.

NOx difference from DF (%)

80 60 IT-2IP0 40

IT-2IP-0.5 IT-2IP+0.5

20

IT+2IP0 IT+2IP-0.5

0

IT+2IP+0.5 IT0IP0

–20

IT0IP-0.5 IT0IP+0.5

Downloaded by [46.177.89.34] at 05:53 11 January 2014

–40

Figure 7 Comparison of NOx between diesel fuel and the three oil types.

It can be observed that the introduction of SO in the content of the fuel used in this engine had a significant influence in the NOx emission level. It has to be pointed out that the main factor to receive increased NOx emissions is the elevated cylinder temperature (Heywood 1988). In addition, fuels with low cetane number (SO in this case) were expected to produce higher NOx levels due to higher ignition delay that creates longer premixed combustion, which is the main phase of NOx production (C. D. Rakopoulos et al. 2006). Furthermore, the fact that the engine under testing exhibits a very high air to fuel ratio added to NOx increment, due to the O2 excess available for chemical reaction with the atmospheric N. As the oil percentage was increased, NOx emissions were augmented. However, there was a noticeable phenomenon in the case of SO20, where NOx values were lower or very close to DF. It seems that the combination of DF and SO reduced the combustion temperature and resulted in reduced NOx levels. This might happen due to bad fuel mixture with the air that produced incomplete combustion of lower temperature. An indication of such incomplete combustion was the very low torque release (see Figure 4) combined with very high fuel consumption (see Figure 5). The influence of IT on NOx production had a clear trend. Early IT (IT+2o CA) gave the highest values and late IT (IT-2◦ CA) the lowest values. As mentioned above, early IT had all indications of better combustion, which is always followed by higher combustion temperatures resulting in NOx augmentation. The opposite phenomenon was demonstrated in the case of late IT. In terms of IP, it was seen that in most cases IP0 had the highest NOx emissions, which illustrated that IT had more important role on NOx emission production. However, IP+0.5 mm produced the lowest NOx values with low oil content fuels (SO20 and SO40), while IP–0.5 mm gave the least NOx emissions with high oil content fuels (SO70 and SO). These settings seemed to make more incomplete combustion (lower combustion temperature), which was accompanied by lower NOx release. Carbon Dioxide (CO2 ). In Figure 8, CO2 emissions produced by the SO fuels in comparison to DF are illustrated. In general, with a few exceptions, CO2 emissions

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

655

15

CO2difference from DF (%)

10 IT-2IP0 5

IT-2IP-0.5 IT-2IP+0.5

0

IT+2IP0 IT+2IP-0.5

–5

IT+2IP+0.5 IT0IP0

–10

IT0IP-0.5 IT0IP+0.5

Downloaded by [46.177.89.34] at 05:53 11 January 2014

–15

Figure 8 Comparison of CO2 emission between diesel fuel and the three oil types.

were increased compared to DF (Venkanna and Reddy in Press). This is an indication of better combustion, due to SO oxygen content. It can be observed that high oil content fuels had the tendency to increase the CO2 levels. There was a unique case (SO20-IT-2-IP+0.5), where CO2 emissions were a lot lower than DF, that could be attributed to very low quality combustion, as indicated by the low torque release (Figure 4), low thermal efficiency (Figure 6), and NOx levels (Figure 7). In terms of IT and IP, there were no certain trends followed in CO2 emissions.

AHP Application in Fuel-IT-IP Configuration Ranking Below is analyzed the AHP method using Group 1 (Criteria 1–4–5) and five attributes (Torque, Thermal Efficiency, BSFC, NOx , and CO2 ). The other two groups were calculated by the same way. First, it was required to set the goal of this work, which was the 37 Fuel-IT-IP configurations ranking. Then, the criteria in terms of which outcomes are evaluated were defined. Subsequently, the attributes according to which the criteria could be evaluated were selected. The criteria and attributes of this study are stated in Figure 3, where the hierarchy tree followed by the AHP is illustrated. Next, a pairwise comparison of the evaluation criteria and attributes with respect to each criterion was conducted (Uzoka, Osuji, and Obot 2011). The answers of the fuel-engine experts were summed and the pairwise comparison matrix (RWC) was prepared for each of the three (3) cases under investigation (Table 5). Table 6 shows the normalized PWC, by dividing each value by the sum of the column in which it belongs (Steps 1–2 in Appendix A). The sum of each row was divided by the number of criteria (Step 3 in Appendix A) to find the weights of criteria (see Table 7), in order to check for judgment consistency. Following the AHP steps (as described in Appendix A), the necessary indexes (CI and CR) were calculated. Table 8 shows the PWC matrix for Criterion 1 and Table 9 shows the normalization of Criterion 1. The weight of each attribute was calculated as the mean

656

BALAFOUTIS ET AL.

value of each line of the normalized PWC and it is shown in Table 10. The Consistency Ratios for all cases are 0. After measurements of preferences, the overall priority vector of each attribute was obtained by multiplying criteria weights matrix with the priority vector of the attributes to find the most influential attribute (Table 11). The calculations were executed as follows: Torque = 0.909 × (0.546) + 0.759 × (0.182) + 0.816 × (0.273) = 0.171 TE = 1.091 × (0.546) + 0.949 × (0.949) + 1.105 × (0.273) = 0.214

Downloaded by [46.177.89.34] at 05:53 11 January 2014

BSFC = 1.091 × (0.546) + 0.981 × (0.182) + 0.868 × (0.273) = 0.202 NOx = 0.939 × (0.546) + 1.139 × (0.182) + 1.105 × (0.273) = 0.204 CO2 = 0.970 × (0.546) + 1.171 × (0.182) + 1.105 × (0.273) = 0.209 Clearly, in Group 1 of criteria, TE was the most important attribute for the Fuel-IT-IP configuration ranking, followed by CO2 , NOx , BSFC, and last Torque. This result was in line with the objective of the main Criterion 1 (increase engine performance), as TE is the best indicator of high engine performance (combination of engine power and fuel consumption). Using the same methodology, Groups 2 and 3 gave the following results in attribute ranking (Tables 12 and 13 respectively). From Table 12, it was observed that in Group 2 of criteria, CO2 was ranked first in importance. It was followed closely by NOx . Then, TE and BSFC were of almost the same significance and last was Torque. The attribute ranking was also according to the purposes of the main Criterion 2 (reduced environmental impact), like in Group 1, as the most important attributes were the emissions. Concerning Group 3, NOx was in the first place, followed by CO2 , BSFC, TE, and Torque. Once more, the attribute ranking was according to the aim of main Criterion 3, because although the pollutants were found more important, the performance indicators (BSFC, TE, and Torque) were also significant. The produced result was in between Group 1 and Group 3 and was the expected one. Finally, the overall priority vector of each attribute was multiplied with the attribute value of the experimental data for each Fuel-IT-IP configuration and then these values were summed to a Ranking Factor Index (RFI). The highest the RFI was, the highest was the ranking position of each configuration. This way, a total ranking of 37 Fuel-IT-IP configurations was obtained. However, for the purposes of this study only the first 10 best configurations are illustrated below for the three Groups of scenarios. Table 5 PWC of Selected Criteria A/A Criterion 1 Criterion 4 Criterion 5

Criterion 1

Criterion 4

Criterion 5

1 0,33 0,5

3 1 1,5

2 0,67 1

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

657

Table 6 Normalized PWC A/A

Criterion 1

Criterion 4

Criterion 5

0,55 0,18 0,27

0,55 0,18 0,27

0,54 0,18 0,27

Criterion 1 Criterion 4 Criterion 5

Table 7 Eigenvector X Weights of Criteria

Downloaded by [46.177.89.34] at 05:53 11 January 2014

0,546 0,182 0,273

Table 8 PWC for Criterion 1 A/A

Torque

TE

BSFC

NOx

CO2

Torque TE BSFC NOx CO2

1 1.2 1.2 1.033 1.067

0.833 1 1 0.861 0.889

0.833 1 1 0.861 0.889

0.968 1.161 1.161 1 1.032

0.958 1.125 1.125 0.969 1

Table 9 Normalized PWC for Criterion 1 A/A

Torque

TE

BSFC

NOx

CO2

Torque TE BSFC NOx CO2

0.219 0.208 0.208 0.179 0.185

0.219 0.208 0.208 0.179 0.185

0.219 0.208 0.208 0.179 0.185

0.219 0.208 0.208 0.179 0.185

0.219 0.208 0.208 0.179 0.185

Table 10 Eigenvectors for the Three Cases A/A Torque TE BSFC NOx CO2

Priority Vi1

Priority Vi2

Priority Vi3

0.909 1.091 1.091 0.939 0.970

0.759 0.949 0.981 1.139 1.171

0.816 1.105 0.868 1.105 1.105

In Figures 9–11, it can be observed that the best configuration in all three Groups of criteria was SO70, combined with early IT (IT+2◦ CA) and manufacturer IP (IP0). High oil content fuels were ranked in the first places, which showed that regardless of the criterion used, these fuels were preferred by AHP technique. Another remark would be that 9 of 10 best configurations had early IT (IT+2◦ CA). This outcome showed that the very good engine performance, which is early IT provoked, overcame the increased NOx effect on the final decision of AHP. In terms of IP, the AHP results were scattered. However, IP0 was usually appeared in the top 10 configurations.

658

BALAFOUTIS ET AL.

Table 11 Attribute Ranking A/A Torque TE BSFC NOx CO2

Priority Vi1

Priority Vi2

Priority Vi3

0.909 1.091 1.091 0.939 0.970

0.759 0.949 0.981 1.139 1.171

0.816 1.105 0.868 1.105 1.105

×

Weights of criteria

−→

0,546 0,182 0,273

Score

Rank

0,171 0.214 0.202 0.204 0.209

5 1 4 3 2

Downloaded by [46.177.89.34] at 05:53 11 January 2014

TOPSIS Application in Fuel-IT-IP Configuration Ranking The criteria, the attributes and the groups of criteria were the same as AHP method. Below, only the analysis using TOPSIS for Group 1 is presented (Criteria 1–4–5) and the same methodology was followed for Groups 2 and 3. Before the creation of the decision matrix D (Step 1 in Appendix B), it was necessary to set the weights of each criterion as used in AHP method (as shown in Table 7). Then, the criteria were classified into benefit and cost categories. The benefit criteria were: J+ = Criterion 1, Criterion 4, Criterion 5 (the larger value of the criterion is, the greater performance has), while none of the evaluating criteria belong to cost criteria J− . Finally, a decision matrix D (Table 14) was created. Then (Step 2 in Appendix B), the square root of the addition element value squares (R) was calculated, according to each criterion (Table 15) and the decision matrix rij , was normalized (Table 16). Afterwards (Step 3 in Appendix B), the values in the normalized decision matrix and the criteria weights were multiplied using the type vij = wj rij (Table 17). In step 4 of the Appendix B, the ideal and the negative ideal point were determined. The symbol “+” at Tables 18 and 19 means that Table 12 Attribute Ranking Using AHP (Group 2) A/A

Score

Rank

Torque TE BSFC NOx CO2

0.143 0.195 0.191 0.234 0.236

5 3 4 2 1

0.75 0.7 0.65 0.6 0.55 0.5

Figure 9 Fuel-IT-IP configuration ranking using AHP (Group 1).

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

659

Table 13 Attribute Ranking Using AHP (Group 3) A/A

Score

Rank

Torque TE BSFC NOx CO2

0.149 0.200 0.203 0.228 0.220

5 4 3 1 2

RFI = 0.171 × Torque + 0.214 × TE + 0.202 × BSFC + 0.204 × NOx + 0.209 × CO2 RFI = 0.143 × Torque + 0.195 × TE + 0.191 × BSFC + 0.234 × NOx + 0.236 × CO2 RFI = 0.149 × Torque + 0.200 × TE + 0.203 × BSFC + 0.228 × NOx + 0,220 × CO2

Group 1 Group 2 Group 3

Downloaded by [46.177.89.34] at 05:53 11 January 2014

0.75 0.7 0.65 0.6 0.55 0.5

Figure 10 Fuel-IT-IP configuration ranking using AHP (Group 2).

0.75 0.7 0.65 0.6 0.55 0.5

Figure 11 Fuel-IT-IP configuration ranking using AHP (Group 3) (color figure available online).

the evaluated criterion belong to J+ , so in each row the greater value was considered. So, the ideal point was: A+ =[0.261, 0.098, 0.133]. Respectively, the negative ideal point considered the smaller value in each row. Hence, the negative ideal point was: A− =[0.232, 0.076, 0.103]. The separation measures Si+ and Si− (Table 20) of each attribute ai from the ideal point A+ and the negative ideal point A− were computed respectively (Step 5 in Appendix B). First, the values of each column were subtracted by the max value of each

660

BALAFOUTIS ET AL. Table 14 Matrix D with size 5∗ 3

D = [xij ] =

Weights

0.546

0.182

0.273

a↓/c→ Torque TE BSFC NOx CO2

Criterion 1 8 9 9 8 8

Criterion 4 6 7 7 8 9

Criterion 5 7 9 7 9 9

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Table 15 R Calculation Weights

0.546

0.182

0.273

a↓/c→ Torque TE BSFC NOx CO2  m  xij2

Criterion 1 64 81 81 64 64

Criterion 4 36 49 49 64 81

Criterion 5 49 81 49 81 81

18.81

16.70

18.47

i=1

Table 16 Normalized Decision Matrix Weights

0.546

0.182

0.273

a↓/c→ Torque TE BSFC NOx CO2

Criterion 1 0.43 0.48 0.48 0.43 0.43

Criterion 4 0.36 0.42 0.42 0.48 0.54

Criterion 5 0.3 0.49 0.38 0.49 0.49

Table 17 Weighted Normalized Decision Matrix a↓/c→

Criterion 1

Criterion 4

Criterion 5

Torque TE BSFC NOx CO2

0.232 0.261 0.261 0.232 0.232

0.065 0.076 0.076 0.087 0.098

0.103 0.133 0.103 0.133 0.133

column and raised in the power of two (Table 21). Max values were selected because all criteria belong to J+ . For Si− , the value of each column was subtracted by the min value of each column (min value selected because all criteria belong to J + ) and raised in the power of two. Then, the square root of the sum of each row was calculated. The relative closeness of the attributes from ideal point was calculated (Table 21) in Step 6 of the Appendix B. The attributes were ranked based on closeness to ideal point Ci . The attribute with maximum Ci was selected (Table 21).

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

661

Table 18 Ideal Points J+ a↓/c→

+ Criterion 1

+ Criterion 4

+ Criterion 5

Torque TE BSFC NOx CO2 A+ A+

0.232 0.261 0.261 0.232 0.232 max 0.261

0.065 0.076 0.076 0.087 0.098 max 0.098

0.103 0.133 0.103 0.133 0.133 max 0.133

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Table 19 Negative Ideal Points J+ a↓/c→

+ Criterion 1

+ Criterion 4

+ Criterion 5

Torque TE BSFC NOx CO2 A− A−

0.232 0.261 0.261 0.232 0.232 min 0.232

0.065 0.076 0.076 0.087 0.098 min 0.076

0.103 0.133 0.103 0.133 0.133 min 0.103

Table 20 Separation MeasuresSi+ a↓/c→

Criterion 1

Criterion 4

Criterion 5

Sum of rows

Si+

Torque TE BSFC NOx CO2

0.00084 0.00000 0.00000 0.00084 0.00084

0.00106 0.00047 0.00047 0.00012 0.00000

0.00087 0.00000 0.00087 0.00000 0.00000

0.00278 0.00047 0.00135 0.00096 0.00084

0.053 0.022 0.037 0.031 0.029

Table 21 Relative Closeness to the Ideal Solution Attributes Torque TE BSFC NOx CO2

Si+

Si−

Ci

Rank

0.053 0.022 0.037 0.031 0.029

0.000 0.043 0.031 0.037 0.044

0.00 0.66 0.46 0.54 0.60

5 1 4 3 2

Clearly, using TOPSIS method, in Group 1 of criteria the best attribute was TE, followed by CO2 , NOx , BSFC, and last Torque. This result was in line with the objective of the main criterion 1 (increase engine performance), as TE is the best indicator of high engine performance (combination of engine power and fuel consumption). Using the same methodology, Groups 2 and 3 gave the following results in attribute ranking, shown in Tables 22 and 23.

662

BALAFOUTIS ET AL. Table 22 Attribute Ranking Using TOPSIS (Group 2) A/A

Score

Rank

Torque TE BSFC NOx CO2

0.00 0.48 0.45 0.86 1.00

5 3 4 2 1

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Table 23 Attribute Ranking Using TOPSIS (Group 3) A/A

Score

Rank

Torque TE BSFC NOx CO2

0.00 0.59 0.67 0.84 0.74

5 4 3 1 2

From Table 22, it is observed that in Group 2 of criteria, CO2 was ranked first in importance. It was followed closely by NOx . Then, TE and BSFC were of almost the same significance, and last came Torque. The attribute ranking was also according to the purposes of the main criterion 2 (reduced environmental impact), like in Group 1, as the most important attributes were the emissions. However, the score of CO2 was very high in comparison to the rest of attributes. Concerning Group 3, NOx was in the first place, followed by CO2 , BSFC, TE, and Torque. Once more, the attribute ranking was according to the aim of main criterion 3 (balanced situation between increased engine performance and reduced environmental impact), because although the pollutants were found more important, the performance indicators (BSFC, TE) were also significant. The produced result was in between Group 1 and Group 2, as expected. Finally, the overall priority vector of each attribute was multiplied with the attribute value of the experimental data for each Fuel-IT-IP configuration and then these values were summed to a Ranking Factor Index (RFI). It has to be pointed out that TOPSIS technique gives no score (0) to the least important attribute, which in this work was Torque in all Groups of criteria. This result could produce problems in the final Fuel-IT-IP configuration, as the experimental data of Torque are not taken into account in the final ranking.

Group 1 Group 2 Group 3

RFI = 0.00 × Torque + 0.66 × TE + 0.46 × BSFC + 0.54 × NOx + 0.60 × CO2 RFI = 0.00 × Torque + 0.48 × TE + 0.45 × BSFC + 0.86 × NOx + 1.00 × CO2 RFI = 0.00 × Torque + 0.59 × TE + 0.67 × BSFC + 0.84 × NOx + 0.74 × CO2

By this way, a total ranking of 37 Fuel-IT-IP configurations was obtained. However, for the purposes of this study, only the first 10 best configurations are illustrated below for the three Groups of scenarios. In Figures 12–14, it can be observed that the best configuration in Groups of criteria 1 and 3 was SO70, combined with early IT (IT+2◦ CA) and manufacturer IP (IP0).

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

663

1.85 1.75 1.65 1.55 1.45 1.35 1.25

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Figure 12 Fuel-IT-IP configuration ranking using TOPSIS (Group 1).

1.85 1.75 1.65 1.55 1.45 1.35 1.25

Figure 13 Fuel-IT-IP configuration ranking using TOPSIS (Group 2).

Nonetheless, Group 2 indicated the same fuel type as optimum (SO70), but with late IT (IT-2◦ CA) and shallow protrusion (IP+0.5 mm). In Groups 1 and 3, the second place was taken by an SO20 configuration, which is a result of the torque influence absence (torque score was 0) on the final RFI given by TOPSIS. In Group 2, high oil content fuels were ranked in the first places, mainly due to low CO2 emissions produced by them. Another remark would be that in Groups 1 and 3 the majority of selected configurations were using early IT (IT+2◦ CA). On the other hand, Group 2 showed all types of IT selection with IT-2◦ CA prevailing. This was the case because the two emissions (NOx and CO2 ) were assigned with very high scores in the attribute ranking, which in the final Fuel-ITIP configuration ranking obviously favored IT-2 that produced very low quantities of both emissions. Comparison of AHP and TOPSIS In this paper, AHP and TOPSIS techniques were applied to classify the 37 tested Fuel-IT-IP configurations. It was important to compare the results of the two techniques primarily in terms of attribute ranking and, second, for the final Fuel-IT-IP configuration ranking for each of the three Groups of Criteria that were utilized. Table 24 gathers the rankings of AHP and TOPSIS techniques for the attributes concerning each Group. It is

664

BALAFOUTIS ET AL.

1.85 1.75 1.65 1.55 1.45 1.35 1.25

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Figure 14 Fuel-IT-IP configuration ranking using TOPSIS (Group 3). Table 24 Ranks of Attributes with AHP and TOPSIS Techniques (Best Attribute Assigned with 1, Worst Attribute Assigned with 5) Group 1 AHP A/A Torque TE BSFC NOx CO2

Ranking 5 1 4 3 2

Group 2

Group 3

TOPSIS

AHP

TOPSIS

AHP

TOPSIS

5 1 4 3 2

5 3 4 2 1

5 3 4 2 1

5 4 3 1 2

5 4 3 1 2

observed that the classification of both techniques is the same. This is an indication that both techniques could be applied in such decision-making cases with similar results. On the other hand, the final Fuel-IT-IP configuration classification did not have the same exact results (Table 25), because the AHP and TOPSIS attributes scores were multiplied with their tested value in each Fuel-IT-IP configurations and the final outcome was differentiated. However, what is important to be stressed is that both methods indicated the same fuel blend as the optimal option (SO70), combined with early IT (IT+2◦ CA) and manufacturer’s protrusion (IP0). This result could be attributed to the use of CO2 values indicated after a complete GHG LCA that gave an advantage to high oil content fuels. In Group 2, AHP and TOPSIS showed significantly different ranking. In particular, TOPSIS indicated late IT (IT-2◦ CA) with shallow protrusion (IP+0.5 mm) as the best configuration. Considering the TOPSIS attribute ranking, it was observed that the two emissions (NOx and CO2 ) were assigned with very high scores, which in the final FuelIT-IP configuration ranking obviously favored IT-2 that produced very low quantities of both emissions. In addition, in TOPSIS technique there were no influence of torque data (torque score in attribute ranking was 0) and there were no rival to overcome the emission influence. Referring to fuel type selection, it could be observed that the majority of the 10 predominant configurations were high oil content fuels (SO and SO70), which indicated that oil increase was favored from both MCDM techniques. As mentioned above, this outcome was mainly due to low CO2 emissions produced by high oil content fuels that

665

A/A 1 2 3 4 5 6 7 8 9 10

Ranking 70/30 IT+2IP0 70/30 IT+2IP+0.5 40/60 IT+2IP+0.5 SO IT+2IP0 SO IT+2IP–0.5 40/60 IT+2IP0 70/30 IT+2IP–0.5 20/80 IT+2IP0 70/30 IT-2IP+0.5 SO IT+2IP+0.5

AHP

TOPSIS

70/30 IT+2IP0 20/80 IT+2IP0 40/60 IT+2IP+0.5 70/30 IT+2IP+0.5 20/80 IT+2IP–0.5 40/60 IT+2IP0 70/30 IT+2IP–0.5 SO IT+2IP–0.5 40/60 IT+2IP–0.5 70/30 IT-2IP+0.5

Group 1 TOPSIS

70/30 IT-2IP+0.5 SO IT+2IP–0.5 SO IT-2IP+0.5 SO IT-2IP–0.5 70/30 IT+2IP0 SO IT0IP–0.5 70/30 IT+2IP+0.5 40/60 IT-2IP+0.5 20/80 IT-2IP+0.5 70/30 IT+2IP–0.5

Group 2

70/30 IT+2IP0 70/30 IT+2IP+0.5 SO IT+2IP–0.5 40/60 IT+2IP+0.5 SO IT+2IP0 70/30 IT+2IP–0.5 70/30 IT-2IP+0.5 40/60 IT+2IP0 20/80 IT+2IP0 SO IT+2IP+0.5

AHP

Table 25 Ranks of Fuel-IT-IP Configurations with AHP and TOPSIS Techniques

Downloaded by [46.177.89.34] at 05:53 11 January 2014

70/30 IT+2IP0 70/30 IT+2IP+0.5 SO IT+2IP–0.5 40/60 IT+2IP+0.5 SO IT+2IP0 70/30 IT+2IP–0.5 40/60 IT+2IP0 20/80 IT+2IP0 70/30 IT-2IP+0.5 20/80 IT+2IP–0.5

AHP

TOPSIS

70/30 IT+2IP0 20/80 IT+2IP0 70/30 IT+2IP+0.5 40/60 IT+2IP+0.5 20/80 IT+2IP–0.5 70/30 IT-2IP+0.5 70/30 IT+2IP–0.5 SO IT+2IP–0.5 20/80 IT-2IP+0.5 40/60 IT+2IP0

Group 3

666

BALAFOUTIS ET AL.

were indicated by LCA intervention. Nevertheless, comparing the three Groups of criteria, it was observed that Group 2, which was focusing on environmental protection, had more high oil content fuels in the top places and Group 3 promoted significantly SO20 in comparison to the other two Groups.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

CONCLUSIONS An experimental investigation was conducted on a diesel engine to evaluate the performance and exhaust emissions of sunflower oil and three blends with diesel fuel (20, 40, and 70% volumetrically) and compare the results with the reference fuel, which was diesel fuel. Three injection timing and three injector protrusion settings were tested, in order to optimize the engine results in both performance and emissions. The work was conducted in a fully instrumented direct injection agricultural tractor engine. The experimental results were evaluated using two multi-criteria decision-making techniques (AHP and TOPSIS) and the optimal fuel type, injection timing, and injector protrusion were selected. AHP and TOPSIS were run for three groups of criteria that were focusing, respectively, in higher engine performance, lower environmental impact, and a balance between the first two. The results of the two techniques were compared. The conclusions extracted were:

r Sunflower oil-based fuels have resulted in unproblematic normal operation during the short-term experiments.

r The 20% blend showed unstable results, in comparison to higher oil content fuels. r Torque and BSFC were enhanced as oil content was increased in the tested fuel. r NOx emissions were augmented as oil percentage in the fuel was increased. r CO2 emissions showed an increase tendency as the oil content was evolved. r Early injection timing improved torque ouput, reduced BSFC, and, as a result, increased thermal efficiency.

r Early injection timing increased rapidly NOx emissions. r Deep injector protrusion increased torque release in low oil content fuels and shallow injector protrusion increased torque in high oil content fuels. protrusion showed bleared results in many cases, due to its interaction with injection timing alteration, which had more severe influence on the engine behavior. AHP and TOPSIS gave the same attribute ranking in all three groups of criteria. AHP and TOPSIS did not give the same results in the final Fuel-Injection TimingInjector Protrusion configuration. The 70/30 blend was selected from both techniques as optimal in all groups of criteria. Main reason was the LCA intervention in the CO2 values used in the multi-criteria decision-making methods. Early injection timing was indicated in most cases as the optimum by both AHP and TOPSIS.

r Injector r r r r

NOMENCLATURE ATDC AHP BSFC BTDC CA

After Top Dead Center Analytical Hierarchy Process Brake-Specific Fuel Consumption Before Top Dead Center Crank Angle

Downloaded by [46.177.89.34] at 05:53 11 January 2014

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

CH4 CI CO CO2 CR DF DI DM GHG GWP HC IP IT LCA LCIA NOx N2 O NIS PIS RDP RFI RI RWC SO TE TOPSIS VO

667

Methane Consistency Index Carbon monoxide Carbon dioxide Consistency Ratio Diesel Fuel Direct Injection Decision Maker Greenhouse Gas Global Warming Potential Hydrocarbons Injector Protrusion Injection Timing Life Cycle Assessment Life Cycle Impact Assessment Nitrogen oxides Nitrous Oxide Negative Ideal Solution Positive Ideal Solution Rotary Distribution Pump Ranking Factor Index Random Consistency Index Pairwise Comparison Matrix Sunflower Oil Thermal Efficiency Technique for Order Preference by Similarity to Ideal Situation Vegetable Oil

REFERENCES 2009/28/Ec, D. 2009. Promotion of the use of energy from renewable sources Parliament, E. Ed. Agarwal, A.K. and K. Rajamanoharan. 2009. Experimental investigations of performance and emissions of karanja oil and its blends in a single cylinder agricultural diesel engine. Applied Energy 86 (1):106–12. Agarwal, D. and A.K. Agarwal. 2007. Performance and emissions characteristics of jatropha oil (preheated and blends) in a direct injection compression ignition engine. Applied Thermal Engineering 27(13):2314–23. Agarwal, D., L. Kumar, and A.K. Agarwal. 2008. Performance evaluation of a vegetable oil fuelled compression ignition engine. Renewable Energy 33(6):1147–56. Al-Widyan, M.I., G. Tashtoush, and M.D. Abu-Qudais. 2002. Utilization of ethyl ester of waste vegetable oils as fuel in diesel engines. Fuel Processing Technology 76(2):91–103. Amarnath, H.K. and P. Prabhakaran. 2012. A study on the thermal performance and emissions of a variable compression ratio diesel engine fuelled with karanja biodiesel and the optimization of parameters based on experimental data. International Journal of Green Energy 9(8): 841–63. Balafoutis, A., S. Fountas, A. Natsis, and G. Papadakis. 2011. Performance and emissions of sunflower, rapeseed, and cottonseed oils as fuels in an agricultural tractor engine. ISRN Renewable Energy 531510. Balafoutis, A., F. Freire, and G. Papadakis. 2010. Life cycle assessment of sunflower oil produced in Greece and used as fuel in tractor diesel engines. Paper presented at the AgEng 2010, September 2010, Clermont-Ferrand.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

668

BALAFOUTIS ET AL.

Desantes, J.M., J. Arregle, S. Ruiz, and A. Delage. 1999. Characterization of the injectioncombustion process in a d.I. Diesel engine running with rape oil methyl ester. SAE no. 1999–01–1497. Eea. Website: http:www.Eea.Europa.Eu/data-and-maps/figures/final-energy-consumption-bysector-4. Fontaras, G., A. Mamakos, L. Ntziachristos, G. Miltsios, and Z. Samaras. 2006. Evaluation of cottonseed oil - diesel fuel blends as fuel for automotive diesel engines. Paper presented at the FISITA 2006 World Automotive Congress, 22–27 October, Yokohama, Japan. He, Y. and Y.D. Bao. 2005. Study on cottonseed oil as a partial substitute for diesel oil in fuel for single-cylinder diesel engine. Renewable Energy 30(5):805–13. Heisler, H. 1999. Vehicle and engine technology. London, UK: Arnold. Heywood. 1988. Internal combustion engines fundamentals. New York: McGraw Hill. Hwang, L.C. and K. Yoon. 1981. Multi attribute decision-making: A methods and applications. Lecture series in economics and mathematical systems. Berlin, Germany: Springer-Verlag. Ipcc. 2007. Climate change 2007: The physical science basis. Fourth assessment report. . New York: Press, CU. Isiklar, G. and G. Büyüközkan. 2007. Using a multi-criteria decision making approach to evaluate mobile phone alternatives. Computer Standards & Interfaces 29(2):265–74. Janic, M. and A. Reggiani. 2002. An application of the multiple criteria decision making (mcdm) analysis to the selection of a new hub airport. European Journal of Transport and Infrastructure Research 2(2):113–42. John Panneer Selvam, D. and K. Vadivel. 2013. Experimental investigation on performance, emission and combustion characteristics of diesel engine fueled with methyl esters of waste pork lard and diesel blends. International Journal of Green Energy10(9):908–23. Kahraman, C. and I. Kaya. 2010. A fuzzy multicriteria methodology for selection among energy alternatives. Expert Systems with Applications 37(9):6270–81. Kalam, M.A., M. Husnawan, and H.H. Masjuki. 2003. Exhaust emission and combustion evaluation of coconut oil-powered indirect injection diesel engine. Renewable Energy 28(15):2405–15. Kallivroussis, L., A. Natsis, and G. Papadakis. 2002. Rd–rural development: The energy balance of sunflower production for biodiesel in Greece. Biosystems Engineering 81(3):347–54. Karimi, M.S., Z. Yusop, and S.H. Law. 2010. Location decision for foreign direct investment in Asean countries: A topsis approach. International Research Journal of Finance and Economics 36:196–207. Kaya, T. and C. Kahraman. 2011. Multicriteria decision making in energy planning using a modified fuzzy topsis methodology. Expert Systems with Applications 38(6):6577–85. Laes, E. and A. Verbruggen. 2010. Decision support for national sustainable energy strategies in an integrated sustainability assessment framework. In: Paths to Sustainable Energy, Nathawani, J. et al. (eds.). Rijeka, Croatia: Intech, pp. 93–116. Li Hao, Q.-S.X. 2006. Application of topsis in the bidding of manufacturing enterprises. In 5th International Conference on e-Engineering & Digital Enterprise Technology. Guiyang, China: e-ENGDET2006. Liu, Y., H. Liu, M. Huo, C.F. Lee, and J. Li. in Press. Spray and combustion characteristics of butanolbiodiesel-diesel in a constant volume combustion chamber. International Journal of Green Energy. Malca, J. and F. Freire. 2009. Energy and environmental benefits of rapeseed oil replacing diesel. International Journal of Green Energy 6(3):287–301. Malça, J. and F. Freire. 2006. Renewability and life-cycle energy efficiency of bioethanol and bio-ethyl tertiary butyl ether (bioetbe): Assessing the implications of allocation. Energy 31(15):3362–80. Markovi´c, Z. 2010. Modification of topsis for solving of multicriteria tasks. Yugoslav Journal of Operations Research 20(1):117–43.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

669

Mcdonnell, K.P., S.M. Ward, P.B. Mcnulty, and R. Howard-Hildige. 2000. Results of engine and vehicle testing of semirefined rapeseed oil. Transactions of the American Society of Agricultural 43:1309–16. Melas, D. 2003. The impact of some geometrical features of the piston bowl on combustion and emissions of an automotive diesel engine. Ph.D. thesis, Brunel University, UK. Mourtzinis, S., S. Fountas and, T. Gemtos. 2007. Greek farmers’ perceptions for precision farming. Paper presented at the 5th National Agricultural Engineering Conference, 18–20 October, Larissa, Greece. Mumtaz, M.W., A. Adnan, Z. Mahmood, H. Mukhtar, M.F. Malik, F.A. Qureshi, and A. Raza. 2012. Biodiesel from waste cooking oil: Optimization of production and monitoring of exhaust emission levels from its combustion in a diesel engine. International Journal of Green Energy 9(7):685–701. Nigim, K., N. Munier, and J. Green. 2004. Pre-feasibility mcdm tools to aid communities in prioritizing local viable renewable energy sources. Renewable Energy 29(11):1775–91. Nwafor, O.M.I. 2003. The effect of elevated fuel inlet temperature on performance of diesel engine running on neat vegetable oil at constant speed conditions. Renewable Energy 28(2):171–81. Nwafor, O.M.I. and G. Rice. 1996. Performance of rapeseed oil blends in a diesel engine. Applied Energy 54(4):345–54. Nwafor, O.M.I., G. Rice, and A.I. Ogbonna. 2000. Effect of advanced injection timing on the performance of rapeseed oil in diesel engines. Renewable Energy 213–4:433–44. Özcanlı, M., A. Keskin, and K. Aydın. 2011. Biodiesel production from terebinth (Pistacia terebinthus) oil and its usage in diesel engine. International Journal of Green Energy 8(5):518–28. Panagakis, P., G. Papadakis, T. Natsis, and S. Kyritsis. 1996. Strategic r&d for better energy efficiency in the agricultural sector. Csmc), EPCNJ-C—GD, .editor Pohekar, S.D. and M. Ramachandran. 2004. Application of multi-criteria decision making to sustainable energy planning—A review. Renewable and Sustainable Energy Reviews 8(4):365–81. Polatidis, H. and D.A. Haralambopoulos. 2004. Local renewable energy planning: A participatory multi-criteria approach. Energy Sources 26:1253–64. Prakash, R., R.K. Singh, and S. Murugan. 2012. An experimental investigation on a diesel engine fueled by biodiesel and its emulsions with wood pyrolysis oil. International Journal of Green Energy 9(8):749–65. Puhan, S., G. Nagarajan, N. Vedaraman, and B.V. Ramabramhmam. 2007. Mahua oil (Madhuca indica oil) derivatives as a renewable fuel for diesel engine systems in India: A performance and emissions comparative study. International Journal of Green Energy 4(1):89–104. Puhan, S., N. Saravanan, G. Nagarajan, and N. Vedaraman. 2010. Effect of biodiesel unsaturated fatty acid on combustion characteristics of a di-compression ignition engine. Biomass and Bioenergy 34(8):1079–88. Rakopoulos, C.D., K.A. Antonopoulos, D.C. Rakopoulos, D.T. Hountalas, and E.G. Giakoumis. 2006. Comparative performance and emissions study of a direct injection diesel engine using blends of diesel fuel with vegetable oils or bio-diesels of various origins. Energy Conversion and Management 47(18–19):3272–87. Rakopoulos, D.C., C.D. Rakopoulos, E.G. Giakoumis, A.M. Dimaratos, and M.A. Founti. 2011. Comparative environmental behavior of bus engine operating on blends of diesel fuel with four straight vegetable oils of Greek origin: Sunflower, cottonseed, corn and olive. Fuel, 90:3439–3446. Rao, G.L.N., B.D. Prasad, S. Sampath, and K. Rajagopal. 2007. Combustion analysis of diesel engine fueled with jatropha oil methyl ester-diesel blends. International Journal of Green Energy 4(6):645–58. Ryan, T.W., T.J. Callaham, and L.G. Dodge. 1982. Characterization of vegetable oil for use as fuels in diesel engines. ASAE 49085:71. Saaty, T.L. 1977. A scaling method for priorities in hierarchical structures. Journal of Mathematical Psychology 15:19–43.

Downloaded by [46.177.89.34] at 05:53 11 January 2014

670

BALAFOUTIS ET AL.

Saaty, T.L. 1980. The analytic hierarchy process: Planning setting priorities. New York, McGrawHill. Saaty, T.L. 2004. Decision making—The analytic hierarchy and network processes (ahp/anp). Journal of Systems and Systems Engineering 13(1):1–34. Shehata, M.S. and S.M.A. Razek. 2011. Experimental investigation of diesel engine performance and emission characteristics using jojoba/diesel blend and sunflower oil. Fuel 90(2):886–97. Shih, H.-S. 2008. Incremental analysis for mcdm with an application to group topsis. European Journal of Operational Research 186(2):720–34. Shuai, S., N. Abani, T. Yoshikawa, R.D. Reitz, and S.W. Park. 2009. Evaluation of the effects of injection timing and rate-shape on diesel low temperature combustion using advanced cfd modeling. Fuel 88(7):1235–44. Strayer, R.C., W.K. Craig, and G.C. Zoerb. 1982. Engine deposit and pour point studies using canola oil as a diesel fuel. ASAE 49085:349. Subbaiah, G.V. and K.R. Gopal. 2011. An experimental investigation on the performance and emission characteristics of a diesel engine fuelled with rice bran biodiesel and ethanol blends. International Journal of Green Energy 8(2):197–208. Topcu, Y.I. and F. Ulengin. 2004. Energy for the future: An integrated decision aid for the case of turkey. Energy 29(1):137–54. Uzoka, F.-M.E., J. Osuji, and O. Obot. 2011. Clinical decision support system (dss) in the diagnosis of malaria: A case comparison of two soft computing methodologies. Expert Systems with Applications 38(3):1537–53. Vedaraman, N., S. Puhan, G. Nagarajan, and K.C. Velappan. 2011. Preparation of palm oil biodiesel and effect of various additives on nox emission reduction in b20: An experimental study. International Journal of Green Energy 8(3):383–97. Venkanna, B.K. and C.V. Reddy. in Press. Performance, emission and combustion characteristics of diesel engine running on blends of honne oil and diesel fuel. International Journal of Green Energy. Venkanna, B.K., B.W. Swati, and C.V. Reddy. 2009. Effect of injection pressure on performance, emissions and combustion characteristics of direct injection diesel engine running on blends of Pogomia pinnata linn oil (honge oil) and diesel fuel. Agricultural Engineering International: The CIGR Ejournal XI:1316. Wagner, L., S. Clark and M. Schrock. 1984. Effect of soybean oil esters on the performance, lubricating oil, and wear of diesel engines.SAE, Warrendale, PA 841385. Website: http:www.Communities.Gov.Uk Multi-criteria analysis: A manual. (accessed August 2, 2013) Yahya, A. and S.J. Marley. 1994. Performance and exhaust emissions of a compression ignition engine operating on ester fuels at increased injection pressure and advanced timing. Biomass and Bioenergy 6(4):297–319. Zhou, P., B.W. Ang, and K.L. Poh. 2006. Decision analysis in energy and environmental modeling: An update. Energy 31(14):2604–22. Zhu, Y., H. Zhao, D.A. Melas, and N. Ladommatos. 2004. Computational study of the effects of the geometry of the piston bowl pip for a high-speed direct injection diesel engine. Journal of Automobile Engineering 218:875–90.

Appendix A AHP process steps for eigenvector calculation and best attribute determination. Step 1: All values in each column are summed. Step 2: The matrix A is normalized. The values in each column of the matrix A are divided by the corresponding column sums (Eq. A.1).

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

rij = aij /

n 

671

aij ,

(A.1)

i=1

rij is the normalized rating of the ith attribute for the jth criterion. The sum of each column should counterbalance with 1. A normalized matrix Anorm is created. Step 3: The Eigenvector X (or the related weight) is found. Each row is divided by n (n represent the number of criteria) to find the average. For example, the weight for the ith row of the matrix w, wi is determined as the average of elements in row i of the matrix Anorm as follows in Equation A.2: 1 rij for i = 1, . . . , n. n j=1 n

Downloaded by [46.177.89.34] at 05:53 11 January 2014

wi =

(A.2)

The new matrix contains the weights of criteria. Criteria weighting must be determined using (n∗ (n–1))/2 pairwise comparisons. Step 4: The Eigenvalue λmax is found. Eigenvalue is obtained from the summation of products between each element of Eigenvector and the sum of columns of the reciprocal matrix. Step 5: (a) The Consistency Index (CI) is found. Consider that Criterion 1 is A, Criterion 2 is B, and Criterion 3 is C. If A > B and B > C, according to the transitive property, the logic preference A > C is come up. Saaty (1977) (T.L. Saaty 1977) proved that for consistent reciprocal matrix, the largest Eigenvalue is equal to the size of comparison matrix, or λmax = n. Then, he gave a measure of consistency, called CI as deviation or −n . degree of consistency using the following formula: CI = λmax n−1 (b) The Consistency Ratio (CR) is found, which is obtained from the deviation of  CI and Random Consistency Index (RI), CR = CI RI . Saaty (1977) (T.L. Saaty 1977) randomly generated reciprocal matrix using scale, 1, . . . , 9 and got the RI to see if it is about 10% or less. The average RI of sample size 500 matrices is shown in the table below:

Table A.1: Random Consistency Index n

1

2

3

4

5

6

7

8

9

10

RI

0

0

0.58

0.9

1.12

1.24

1.32

1.41

1.45

1.49

If the CR is less than 0.1, then the results are trustworthy and further evaluation is not needed. However, if greater than 0.1, then the results are unreliable and the pairwise comparison should be re-evaluated or rejected (Saaty 1977). Step 6: The relative ranking (priority) of attributes is found. AHP uses pairwise comparison to create relative performance scores for each of the attributes on each criterion. If there are m attributes and n criteria, then n separate m∗ m matrices must be created and worked out (Steps 1–3 are repeated with the new matrices). Alternative scoring is taken using n∗ ((m∗ (m–1))/2) pairwise comparisons between attributes for each criteria.

672

BALAFOUTIS ET AL.

Step 7: The synthesis of priorities is carried out by multiplying the criteria’s weights matrix with the priorities of attributes matrix (with respect to each criteria) as follows in Eq. A.3: Si =

m 

wj vij for i = 1, . . . , n,

(A.3)

j=1

where, Si is the overall score of the ith attribute and vij is the element of a priority vector of the ith attribute with the jth criterion. The higher score enters first in the list of classification. Appendix B

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Steps of TOPSIS method for ranking alternatives: Step 1: The decision matrix D is created by each decision maker.

D=

a1 a2 . .. am

C1

C2

...

Cn

x11 x21 . .. xm1

x12 x22 . .. xm3

... ... . .. ...

x1n x2n . .. xmn

where ai indicates the ith attribute, i =1,2, . . ., m; Cj typifies the jth criterion, j =1, 2, . . . , n; xij represents the performance of the ith attribute as regards jth criterion. Criteria Cj may be of benefit or cost type. Benefit type criteria means that the larger value of attribute is, the greater performance it has (represented by J+ ) while cost type criteria means that the smaller value of attribute is, the greater performance it has (represented by J− ) (Hao and Qing-Sheng 2006; Markovi´c. 2010). Step 2: The decision matrix is normalized to transform various attribute dimensions into non-dimensional attributes, which allows comparisons across criteria (Karimi et al. 2010). The R represents the square root of the addition element value squares, according to each criterion. The R is calculated for each j criterion of decision making matrix using Equation B.1 (Markovi´c. 2010).   m  xij2 for i = 1, . . . , m; j = 1, . . . , n. Rj = 

(B.1)

i=1

 Then, each column is divided by Rj =

m 

xij2 to get rij ,which represents the elements

i=1

of new normalized decision making matrix and calculated as in Equation B.2:

SUNFLOWER OIL FUEL FOR DIESEL ENGINES

  m  rij = xij  xij2 for i = 1, . . . , m; j = 1, . . . , n.

673

(B.2)

i=1

Step 3: weighted normalized decision matrix is calculated by multiplying each column of the normalized decision matrix with normalized weight coefficients wj for j = 1, n  . . ., n, such as that: wj = 1. The weight normalized value vij is calculated as: vij = wj rij j=1

(Markovi´c. 2010).

Downloaded by [46.177.89.34] at 05:53 11 January 2014

Step 4: The ideal and non-ideal points are specified as shown below (Markovi´c. 2010). Ideal point:

+ + + A+ = {(max vij | j ∈ J1 ), (min vij |j ∈ J2 )|i = 1, 2, ..., m} = {v+ 1 , v2 , ..., vj , ..., vn }

Negative ideal point:

− − − A− = {(min vij | j ∈ J1 ), (max vij |j ∈ J2 )|i = 1, 2, ..., m} = {v− 1 , v2 , ..., vj , ..., vn }

i

i

i

i

whereas J1 ⊂ {1, 2, ..., n| j − max}for the max type criteria, J2 ⊂ {1, 2, ..., n| j − min} for the min type criteria

Step 5: The Euclidean distances Si+ and Si− of each criteria ai from the ideal point A+ and from the negative ideal point A− ,respectively, are calculated. Each attribute distance from the ideal point is indicated in Eq. B.3:    n 2 Si+ =  (vij − v+ j ) , i = 1, . . . , m.

(B.3)

j=1

Whereas, each attribute distance from the negative ideal point is indicated in Eq. B.4:    n − 2 (vij − v− Si =  j ) , i = 1, . . . , m,

(B.4)

j=1

where rj+ and rj− means the distances from jth criterion to the ideal and non-ideal solution. Step 6: The relative closeness of the attributes from ideal and non-ideal points is calculated (Eq. B.5): Ci =

Si− − ; 0 ≤ Ci ≤ 1; i = 1, . . . , n. S+ i + Si

(B.5)

If Ci is equal to 1 then ai is the ideal point (A+ ), and if Ci is equal to 0 then ai is the negative ideal point (A− ). Step 7: Attributes are ranked according to Ci and select the attribute with maximum Ci (Hao and Qing-Sheng 2006; Markovi´c. 2010).