Post on 20-Mar-2020
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473 465
COMPUTATIONAL OPTIMIZATION OF BIODIESEL COMBUSTION USING RESPONSE SURFACE METHODOLOGY
by
Prabhakara Rao GANJI*, V. Rajesh Khana RAJU, S. Srinivasa RAO
Department of Mechanical Engineering, NIT Warangal, Telangana, India Original scientific paper
DOI: 10.2298/TSCI161229031G
The present work focuses on optimization of biodiesel combustion phenomena through parametric approach using response surface methodology. Physical properties of biodiesel play a vital role for accurate simulations of the fuel spray, atomization, combustion, and emission formation processes. Typically methyl based biodiesel consists of five main types of esters: methyl palmitate, methyl oleate, methyl stearate, methyl linoleate, and methyl linolenate in its composition. Based on the amount of methyl esters present the properties of pongamia bio-diesel and its blends were estimated. CONVERGETM computational fluid dynam-ics software was used to simulate the fuel spray, turbulence and combustion phe-nomena. The simulation responses such as indicated specific fuel consumption, NOx, and soot were analyzed using design of experiments. Regression equations were developed for each of these responses. The optimum parameters were found out to be compression ratio – 16.75, start of injection – 21.9° before top dead center, and exhaust gas re-circulation – 10.94%. Results have been compared with baseline case. Key words: CONVERGETM, optimization, response surface methodology,
pongamia biodiesel, variable compression ratio engine
Introduction
Biodiesels are attractive sources of energy to replace fossil fuels used in IC engines. This is because of their desirable attributes such as being sustainable, biodegradable, and car-bon neutral. Biodiesels can also be used in existing CI engines with small or no modifications. Many researchers found that the use of biodiesels can reduce emissions such as unburned hy-drocarbons (UHC), soot, and CO, but a slight increase in NOx was observed. Typically bio-diesels can be assumed to be methyl or ethyl esters but methyl esters were widely used be-cause they show better performance and emission characteristics [1-5].
Simulation of IC engines has gained popularity in recent days due to availability of sophisticated computational facilities and CFD codes for the accurate prediction of spray and combustion phenomena. A suitable 3-D computational model with proper reaction mecha-nism and spray model can accurately predict the combustion characteristics. Physical and chemical properties of biodiesel such as normal boiling point, critical properties, vapor pres-sure, latent heat of vaporization, density, viscosity, thermal conductivity, and surface tension have great impact on the spray and combustion modelling whose accuracy also redefines the –––––––––––––– * Corresponding author, e-mail: ganjiprabhakar@gmail.com, ganjiprabhakar@nitw.ac.in
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … 466 THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473
accuracy of numerical predictions. An et al. [6] presented correlations to predict the properties for 5 methyl esters namely methyl palmitate, methyl oleate, methyl stearate, methyl linoleate, and methyl linolenate.
Brakora et al. [7] described a reduced reaction mechanism for biodiesel combustion which consists of 41 species and 150 reactions. The mechanism predicted well as that of the comprehensive mechanism which consists of 264 species and 1219 reactions. They have also developed reaction mechanism for biodiesel blends which consists of 53 species and 156 reac-tions. The reduced mechanism gave the advantage of lesser computational time with reasona-ble accuracy. Brakora and Reitz [8] performed a reduction in the chemical mechanism which suits to the multi fuel component species like biodiesel and its blends; the reaction mechanism contains 69 species and 192 reactions.
A review of pongamia biodiesel combustion from the open literature by Murugesan et al. [9] reveals the following salient features. Pure pongamia biodiesel can be directly used in diesel engines without any alterations with slightly inferior performance than that of diesel. Addition of little amounts of biodiesel to mineral diesel is a worthy strategy for increasing al-ternative fuel consumption and B20 is the best alternative fuel. Brake thermal efficiency of biodiesel (B20) is slightly increased and brake-specific energy consumption is reduced slight-ly when compared with B100. But in comparison with mineral diesel, the BSFC of B20 could not improve at baseline configuration whereas emission characteristics were reduced and at par with diesel in most of the literature available.
Qi et al. [10] analyzed the effect start of injection timing and exhaust gas re-circu-lation (EGR) rate on the combustion and emissions of a direct injection CI engine using split injection strategy with neat soybean biodiesel. The results show that the increase of EGR rate increases the brake specific fuel combustion (BSFC) and soot emission whereas NOx emis-sions were decreased. Results also reveal that BSFC was slightly increased, NOx emissions were evidently decreased, and soot emissions remained almost same with the retarded main injection timing.
Gopal and Kanupparaj [11] conducted experiments on CI engine with blends of pongamia biodiesel and diesel without any modification of existing engine design. The PME 20 (pongamia 20% and diesel 80%) allows for better performance and lower emissions over the various blends and it is also on a par with petro diesel. Wilson [12] optimized diesel en-gine control parameters such as clearance volume, fuel injection pressure (FIP), nozzle-hole diameter, start of injection (SOI), and load using Taguchi design of experiments (DOE) in or-der to get the best performance in terms of NOx and BSFC. They also derived relations be-tween operating parameters and responses and identified that the results of experimental data had a good agreement with the predicted results. It was also analyzed that among all the pa-rameters FIP has a great influence on NOx. They concluded that the DOE is an effective and efficient way to develop a robust design in order to get optimum performance and emissions.
The complexity of the in-cylinder combustion phenomena can be modeled using CONVERGETM [13] CFD software, in which turbulence modeling, combustion modeling, and spray modeling are used simultaneously to predict the performance and emission charac-teristics precisely. Response surface methodology (RSM) is also a proven technique which reduces the number of experiments to be conducted and saves time and human effort. It can also predict the model with reasonable accuracy [14].
There is plenty of literature available on each of these individual effects on perfor-mance and emissions. But the interaction effects of these parameters were not discussed much on the performance and emissions. The present work focuses on CFD simulations of ponga-
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473 467
mia biodiesel combustion in a variable compression ratio (VCR) engine and to an-alyze the effect of parameters such as CR, SOI, and EGR on performance and emis-sions. The present study also uses output responses from the CFD study and ana-lyzed using a systematic approach RSM [15]. These effects are studied in the pre-sent simulation study and also aimed to op-timize the operating parameters in order to obtain minimum emissions and improved performance.
The CFD simulation
The engine sector model shown in fig. 1, with the dimensions given in tab. 1, is considered for the present study. The fuel properties are given in tab. 2. The simula-tions are carried out using CONVERGETM CFD software. The models used in CFD package for different physical phenomena are shown in tab. 3. The numerical simula-tion is validated with experimental data and also compared with literature [11] with PB 20 (pongamia methyl ester 20% and 80% diesel) as a fuel. The properties of the selected biodiesel (PB 20) are estimated using MATLAB code based on the correla-tions developed by An et al. [6]. These properties are given as input to numerical simulations.
Validation of model
Comparison of pressure with respect to crank angle for baseline configuration of experimental and simulation model are shown in fig. 2. It can be observed that simulation results are in good agreement with the experimental. The difference in peak pressure for the experimental and simulation is observed to be around 7%. The emissions are also compared for the experi-mental and simulation models in tab. 4. The difference between experimental and simulation is found to be 2% and 8%, respectively, for NOx and soot emissions.
Results and discussion
Three parameters, CR, SOI, and EGR were chosen to optimize the engine configura-tion with a view to attaining better performance and emissions. A total of 17 experiments
Figure 1. Computational domain of VCR engine sector model; (a) near top dead center, (b) near bottom dead center
Table 1. Specifications of the VCR engine
Make Kirloskar
Bore 87.5 mm
Stroke 110 mm
Connecting rod length 234 mm
Compression ratio 17.5:1
RPM 1500
Piston bowl shape Hemispherical
Piston bowl diameter 52 mm
Cylinder head curvature Flat
Injection timing 23° bTDC
Load 3.5 kW (100%)
Fuel injection opening pressure 200 bar
Table 2. Properties of mineral diesel and PB20
Properties Diesel PB20
Density [kgm–3] 830.4 836.4
Calorific value [MJkg–1] 45.2 43.3
Viscosity [cSt] 3.61 4.1
Flash point [C] 77.6 95
Cetane number 48 56
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … 468 THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473
were conducted and their responses tabulated in tab. 5. The CR is varied from 14 to 18, SOI is varied from 17° bTDC to 26° bTDC and EGR percent-age is varied from 0 to 20 in three lev-els. The RSM was used in the current work for analyzing the output respons-es to obtain the regression equations.
Evaluation of the regression model
The regression models are ana-lyzed based on analysis of variance (ANOVA) which gives the p value for different response parameters such as ISFC, NOx, and soot emissions. The p values for output responses are given in tab. 6. These p values tell us whether the particular parameter is significantly affecting the output responses or not. The reference limit for p value was chosen as 0.05. The models are signifi-cant as their p values are less than 0.05 [25, 26].
It can be observed from tab. 6 that the individual parameters (CR, SOI, and EGR) have significant effect on all the responses since their p value is less than 0.05. Interaction effects of (CR·SOI) and (SOI·EGR) also have significant effect on all the responses. The CR·EGR has no significant effect on the output responses since their p value is greater than 0.05. So the inter-action effects that are significant are only discussed in the present section. The regression statistics goodness of fit, R2, adjusted R2, and predicted R2 for
all the responses is shown in tab. 7. For a model to fit the data well, the difference between predicted and adjusted R2 should be less than 0.2.
The quadratic models have been developed for ISFC, NOx, and soot in terms of in-put parameters in eqs. (1), (2), and (3), respectively.
2 2
2
ISFC 312.5 5.72 CR 4.33 SOI 1.720 EGR 0.114 0.0236 SOI
0.0318 EGR 0.2650 CR SOI 0.0366 CR EGR 0.0031 SOI EGR
CR= + ⋅ − ⋅ + ⋅ + ⋅ − ⋅ −
− ⋅ − ⋅ ⋅ − ⋅ ⋅ − ⋅ ⋅ (1)
Table 3. Key sub models used in the CFD analysis
Turbulence model RNG k-ε [16]
Injection drop distribution χ2 distribution [13]
Drop drag Dynamic drag model [17]
Droplet collision model NTC model [18]
Collision outcomes model Post collision outcomes [19]
Drop turbulent dispersion O’Rourke model [20]
Drop/wall interaction Rebound/slide model [21]
Evaporation model Chiang drop correlations [22]
Spray breakup KH-RT [23]
Combustion SAGE [24]
Figure 2. Validation of pressure vs. crank angle for VCR engine with PB20 as fuel
Table 4. Comparison of emissions for VCR engine with PB20 as fuel
Experimental Simulation
NOx [gkg–1 of fuel] 30.15 29.45
Soot [gkg–1 of fuel] 0.678 0.625
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473 469
x2 2 2
NO 410.68 51.27 CR 2.94 SOI 0.417 EGR 0.198 CR SOI
0.033 CR SOI 0.002 SOI EGR 1.72 CR 0.0047 SOI 0.0213 EGR
= − ⋅ − ⋅ − ⋅ + ⋅ ⋅ −
− ⋅ ⋅ − ⋅ ⋅ + ⋅ + ⋅ + ⋅ (2)
2
2 2
Soot 35.34 3.61 CR 0.174 SOI 0.044 EGR 0.0163 CR SOI
0.0011 CR EGR 0.0038 SOI EGR 0.11 CR
0.0047 SOI 0.00154 EGR
= − ⋅ − ⋅ − ⋅ − ⋅ ⋅ −
− ⋅ ⋅ − ⋅ ⋅ + ⋅ +
+ ⋅ + ⋅ (3)
Table 5. Experimental design and their responses
Table 6. The p values for model terms of ANOVA analysis
S. No. Parameter Responses
CR SOI EGR ISFC [gkW–1h–1]
NOx [gkg–1 of fuel]
Soot [gkg–1 of fuel]
1 16 26 0 235.00 40.01 0.81
2 18 17 10 239.48 48.72 0.98
3 18 21.5 20 231.23 51.12 0.64
4 18 26 10 225.12 56.52 0.42
5 16 21.5 10 248.12 32.52 0.95
6 14 17 10 265.27 22.84 2.45
7 14 26 10 260.45 23.52 2.48
8 14 21.5 0 257.15 27.45 2.12
9 16 21.5 10 247.15 32.52 0.95
10 16 21.5 10 248.12 32.52 0.95
11 14 21.5 20 267.16 20.02 2.53
12 16 21.5 10 247.18 32.52 0.95
13 16 21.5 10 247.68 32.52 0.95
14 16 17 20 253.15 26.12 1.54
15 18 21.5 0 224.15 61.15 0.32
16 16 26 20 243.45 30.0 1.25
17 16 17 0 244.15 36.45 1.03
ISFC NOx Soot
Regression model <0.001 <0.001 <0.001
SOI <0.001 <0.001 <0.001
CR <0.001 <0.001 <0.001
EGR <0.001 <0.0001 <0.0001
CR·SOI 0.019 0.005 <0.0001
CR·EGR 0.381 0.187 0.255
SOI·EGR 0.001 0.002 0.02
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … 470 THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473
Table 7. The RSM model evaluation
Interaction effect of SOI and CR
It can be observed from fig. 3 that at lower CR, the interaction of SOI and CR is sig-nificant, but not fruitful in reducing ISFC. But at higher CR, the interaction is beneficial in re-ducing ISFC. Advancing the injection timing to 26° bTDC increased NOx emissions and re-duced the soot emissions. This is due to the fact that in-cylinder temperature and pressures were low at the time of fuel injection, which increases the ignition delay. At the same time increase in compression ratio resulted in increased in-cylinder temperature and thus NOx emissions were increased, whereas soot emissions are decreased due to efficient combustion. From figs. 4 and 5 it can be interpreted that the combination of high compression ratios and early fuel injection timings resulted high NOx emissions but less soot emissions. Soot emissions were high and NOx emissions were low at retarded injection timings and low compression ratios.
Figure 3. The ISFC variations against SOI and CR
Figure 4. The NOx variations against SOI and CR
Effects of SOI and EGR
Figure 6 shows the interactive effects of injection timing and exhaust gas recircu-lation on ISFC. As the start of injection is advanced from 17° to 26° bTDC, ISFC is decreased. Also, as the EGR is increased from 0% to 25%, ISFC increased. But this phenomenon is dominant at lower EGR. This is due to fact that the EGR reduces the oxygen concentration while also increasing the specific heat of the charge and thus drops the in cylinder temperature which re-sults in inefficient combustion.
Model ISFC NOx Soot
R2 0.9923 0.9976 0.9989
Adjusted R2 0.9823 0.9945 0.9975
Predicted R2 0.9267 0.9617 0.9823
Figure 5. Soot variations against SOI and CR
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473 471
From the results shown in figs. 7 and 8, it can be interpreted that the combination of low EGR and early fuel injection tim-ings resulted in high NOx emissions but very little soot emissions. Soot emissions were high and NOx emissions were low at retarded SOI and higher EGR.
Optimization
The previous discussions on the ef-fects of SOI, CR, and EGR on the perfor-mance and emission characteristics re-vealed that the smallest CR of 14 and re-tarded SOI of 17° bTDC resulted in lower
Figure 7. The NOx variations against SOI and EGR
Figure 8. Soot variations against SOI and EGR
NOx, higher ISFC, and higher soot emission. On the other hand, higher CR and advanced SOI resulted in higher NOx, reduced ISFC and decrease in soot values. There is an abso-lute trade-off observed be-tween NOx and soot. So parameters must be optimized in order to get lower emissions without any cost of its performance. In desirability based approach, different solutions were obtained. The solution with highest desirability is preferred. The lowest from the experimental data (in case of emissions/minimization objective) is taken as desirability of 1. The value of desirabil-ity tells us how close the optimum is to the lowest value. Desirability of 0.97 was obtained at the following parameters CR – 16.75, SOI – 21.9° bTDC, and EGR – 10.94%. Table 8 shows the comparison of optimized and baseline cases and it is observed that simultaneous reduction in soot and NOx is attained with some improvement in ISFC. Optimum configuration reduced ISFC, NOx and soot by 2%, 31%, and 34%, respectively.
Figure 6. The ISFC variations against SOI and EGR
Table 8. Comparison of optimized case and base case
ISFC [gkW–1h–1]
NOx [gkg–1 of fuel]
Soot [gkg–1 of fuel]
Base line 235.87 29.45 0.5625
Optimized case 231.17 20.15 0.370
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … 472 THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473
Conclusions
The following conclusions were drawn based on the numerical simulations using CONVERGETM software. The DOE RSM methodology was employed for designing the min-imum number of simulations and their levels. • The DOE is very useful in analyzing the simulation results in a statistical approach and
eventually help us to determine the most influential parameters that affect the perfor-mance and emissions.
• The NOx emissions were evidently decreased and soot emissions increased with increas-ing rate of EGR.
• Increase in SOI increases NOx and reduces soot and ISFC. • Decrease in CR increases soot emissions and ISFC whereas NOx is reduced. • The interaction effects of CR and SOI were dominant for the responses of NOx and soot. • The interaction effects EGR and SOI were dominant for the responses ISFC and NOx. • The interaction effects CR and EGR were very less significant as compared to other two
combinations. • The optimum set of parameters was found out to be CR – 16.75, SOI – 21.9° bTDC, and
EGR – 10.94%.
Acknowledgment
This work was supported by Center of Excellence (COE), Center for sustainable en-ergy studies of Technical Education Quality Improvement Programme (TEQIP-II), India.
Nomenclature aTDC – after top dead center bTDC – before top dead center BSFC – brake specific fuel consumption CFD – computational fluid dynamics CI – compression ignition CR – compression ratio DOE – design of experiments EGR – exhaust gas recirculation
FIP – fuel injection pressure IC – internal combustion ISFC – indicated specific fuel consumption RPM – rotations per minute RSM – response surface methodology SOI – start of injection VCR – variable compression ratio
References [1] Dhananjaya, D. A., et al., Combustion Characteristics of diesel Engine Operating on Jatropha Oil Methyl
Ester, Thermal Science, 14 (2010), 4, pp. 965-977 [2] Knežević, D., et al., The Characteristics of Combustion Process of diesel Engine Using Vegetable Oil
Methyl Esters, Thermal Science, 19 (2015), 6, pp. 2255-2263 [3] Jain, S., Sharma, M. P., Prospects of Biodiesel from Jatropha in India: a Review, Renewable and Sus-
tainable Energy Reviews, 14 (2010), 2, pp. 763-771 [4] Dwivedi, G., Sharma, M. P., Prospects of Biodiesel from Pongamia in India, Renewable and Sustainable
Energy Reviews, 32 (2014), Apr., pp. 114-122 [5] Baiju, B., et al., A Comparative Evaluation of Compression Ignition Engine Characteristics Using Me-
thyl and Ethyl Esters of Karanja Oil, Renewable Energy, 34 (2009), 6, pp. 1616-1621 [6] An, H., et al., Detailed Physical Properties Prediction of Pure Methyl Esters for Biodiesel Combustion
Modeling, Applied Energy, 102 (2013), Feb., pp. 647-656 [7] Brakora, J., et al., Development and Validation of a Reduced Reaction Mechanism for Biodiesel-Fueled
Engine Simulations, SAE Int. J. Fuels Lubr., 1 (2009), 1, pp. 675-702 [8] Brakora, J., Reitz, R., A Comprehensive Combustion Model for Biodiesel-Fueled Engine Simulations,
SAE technical paper 2013-01-1099, 2013 [9] Murugesan, A., et al., Biodiesel as an Alternative Fuel for diesel Engines – a Review, Renewable and
Sustainable Energy Reviews, 13 (2009), 3, pp. 653-662
Ganji, P. R., et al.: Computational Optimization of Biodiesel Combustion … THERMAL SCIENCE, Year 2017, Vol. 21, No. 1B, pp. 465-473 473
[10] Qi, D., et al., Effect of EGR and Injection Timing on Combustion and Emission Characteristics of Split Injection Strategy DI-Diesel Engine Fueled with Biodiesel, Fuel, 90 (2011), 5, pp. 1884-1891
[11] Gopal, K. N., Karupparaj, R. T., Effect of Pongamia Biodiesel on Emission and Combustion Character-istics of DI Compression Ignition Engine, Ain Shams Engineering Journal, 6 (2015), 1, pp. 297-305
[12] Wilson, V. H., Optimization of diesel Engine Parameters Using Taguchi Method and Design of Evolu-tion, Journal of the Brazilian Society of Mechanical Sciences and Engineering, 34 (2012), 4, pp. 423-428
[13] Richards, K. J., et al., CONVERGE (Version 2.2. 0) Manual. Convergent Science, Inc., Middleton, WI., USA
[14] Ganji, P. R., et al., Parametric Study and Optimization Using RSM of DI diesel Engine for Lower Emis-sions, Journal of the Brazilian Society of Mechanical Sciences and Engineering, http://link.springer.com/10.1007/s40430-016-0600-0
[15] Box, G. E., Draper, N. R., Empirical Model-Building and Response Surfaces, Vol. 424, John Wiley and Sons, New York, USA, 1987
[16] Han, Z., Reitz, R. D., Turbulence Modeling of Internal Combustion Engines Using RNG κ-ε Models, Combustion Science and Technology, 106 (1995), 4-6, pp. 267-295
[17] Liu, A. B., et al., Modeling the Effects of Drop Drag and Breakup on Fuel Sprays, SAE paper 930072, 1993
[18] Schmidt, D. P., Rutland, C. J., A New Droplet Collision Algorithm, Journal of Computational Physics, 164 (2000), 1, pp. 62-80
[19] Post, S. L., Abraham, J., Modeling the Outcome of Drop-Drop Collisions in diesel Sprays, International Journal of Multiphase Flow, 28 (2002), 6, pp. 997-1019
[20] O'Rourke, P. J., Collective Drop Effects on Vaporizing Liquid Sprays (No. LA-9069-T). Los Alamos National Lab., N. Mex., USA, 1981
[21] Gonzalez, D., M. A., et al., Modeling diesel Engine Spray Vaporization and Combustion, SAE Transac-tions, 101 (1992), 3, pp. 1064-1076
[22] Chiang, C. H., et al., Numerical Analysis of Convecting, Vaporizing Fuel Droplet with Variable Proper-ties, International Journal of Heat and Mass Transfer, 35 (1992), 5, pp. 1307-1324
[23] Beale, J. C., Reitz, R. D., Modeling Spray Atomization with the Kelvin-Helmholtz/Rayleigh-Taylor Hy-brid Model, Atomization and Sprays, 9 (1999), 6, pp. 623-650
[24] Senecal, P. K., et al. Multi-Dimensional Modeling of Direct-Injection diesel Spray Liquid Length and Flame Lift-off Length Using CFD and Parallel Detailed Chemistry, SAE technical paper 2003-01-1043, 2003
[25] Montgomery, D. C., Design and Analysis of Experiments, John Wiley & Sons, 2008, ISBN 978-0-470-12866-4
[26] Ryan, T. A., et al., Minitab™, John Wiley and Sons, Inc., 2004, DOI: 10.1002/0471667196.ess1647
Paper submitted: December 29, 2016 © 2017 Society of Thermal Engineers of Serbia. Paper revised: February 12, 2017 Published by the Vinča Institute of Nuclear Sciences, Belgrade, Serbia. Paper accepted: February 14, 2017 This is an open access article distributed under the CC BY-NC-ND 4.0 terms and conditions.