Advances in Automobile Engineering
Open Access

Optimization of Diesel Engine Parameters using Genetic Algorithm

Author info »

Abbreviations

NOH: Number of Nozzle Holes; PHC: Piston Head Clearance; IP: Injection Pressure; DOC: Direct Oxidization Catalyst; SCR: Selective Catalytic Reduction; FC: Time Taken for 50 ml Fuel Consumption; BTE: Brake Thermal Efficiency; HC: Hydro Carbons; GA: Genetic Algorithm; PM: Particulate Matter

Introduction

Conventionally all immobile engines are designed for a general optimal set up, but almost all engines are run at specific load which is constant in most of the cases. Hence it is highly required to optimize the engine parameters at different loads. This optimal set is likely to have comparatively lower emission and higher efficiency. To achieve this Genetic Algorithms (GA) technique is used and the results were verified by experiments.

GA was introduced in United States in 1970s with the motive of improving the performance / price of computational systems by John Holland, University of Michigan. Considering the workability of GA in continuous and discrete, combinatorial and multi-parameter optimization problems the usage GA in optimization enhanced exponentially. The key advantage of GA is that they do not get stuck at local optima like gradient search method. Owing to the levels of optimization required in engines and considering the advantages of GA, the same is used as an optimization tool in this work.

Literature Review

Based on the need for performance improvement and reduction of emission the details of works/studies conducted earlier are collected and are discussed in this section. The particulate matter emissions were strongly dependent on the variation in the number of holes in the nozzle up to four. In the range tested the number of holes in the nozzle had a strong influence on particulate matter emissions when compared to injection pressure. On the other hand specific fuel consumption was strongly influenced by injection pressure when compared to number of holes in the nozzle. Interestingly the influence of numbers of holes in the nozzle on particulate matter was fundamentally different, as it was distinct [1]. The speed range 1500 to 2500 rpm more number of holes in nozzle will increase the efficiency of the engine. When the number of holes is increased to seven, the engine was at its best performance, due to finer atomization, higher dispersion of fuel and better mixing rate, however increasing the number of holes in the nozzle is not widely undertaken as it involves manufacturing complications [2]. Numerical analysis shows that increasing the number of nozzles increases the performance of IC engines, in terms of efficiency [3]. From the above it is clear that the increase in number of nozzles increases the combustion phenomenon with the stroke volume as a constraint.

In another study has affirmed that, increase in injection pressure is in-line with increase in power. This conclusion was made after experimenting the consumption of fuel for different loads with different speeds by varying either load or speed at a time. This study also inferred that an increase in injection pressure leads to increase in both emissions and consumption of fuel in diesel engines [4]. Kannan asserted that an improvement in brake thermal efficiency and reduction in brake specific fuel consumption was evident above 150 bar injection pressure. However, only the peak enhancement in performance was about 1% only, but while considering the running hours of the engine this could lead to phenomenal saving in fuel [5]. From the above it is clear that the increase in injection pressure increases the performance of the engine.

It stated that at elevated temperatures, the use of reducing catalysts will reduce oxides of nitrogen to atmospheric nitrogen [6-9]. Investigations affirm the conversion of carbon dioxide from hydrocarbons when catalyzed [10]. The above studies show appropriateness of catalytic conversions in diesel engine exhaust.

The details of the experimental setup and the conduct of experiment are discussed in the next section.

Engine setup

The 5.968 kW Kirloskar engine investigated, has a 0.242 m wide brake drum, a three hole nozzle, piston head clearance of 1.4 mm and an injection pressure of 170 bar.

Based on the literature surveyed six parameters are chosen and the variations of these parameters are shown in Table 1. The variation of parameters was done in accordance to the operating range of the engine, cost and their availability. The reactivity of the catalyst depends on its molecular weight hence this numerical value is used for analysis and representation.

Table 1: Variation of the selected parameters.

Experimentation

The output such as oxides of nitrogen, hydrocarbons, carbon dioxide, carbon mono-oxide and brake thermal efficiency were obtained experimentally for all 486 combinations [11].

Model generation technique used, using the experimentally obtained data is discussed in the following section.

Optimization

Model generation using MINITAB 14: Mathematical models were generated using MINITAB 14 the same are shown below in equations 1 to 5. Since the number of variations in the parameter number of nozzles is two, only linear models can be generated (i.e. NOH*NOH cannot be obtained) where as a quadratic models are generated for all other parameters.

Equations 1: NOx= +25489.2 –972.826 *NOH –30243.8 *PHC +6.04599 *IP +8.24669 *LOAD –0.01185 *SCR –0.01151 *DOC –0.24172 *FC +9075.55 *PHC *PHC –0.00122 *IP *IP –0.8353 *LOAD *LOAD +1.74E –07 *SCR *SCR +1.85E –06 *DOC *DOC –6.19E –06 *FC *FC +409.342 *NOH *PHC +0.175607 *NOH *IP –5.90069 *NOH *LOAD +6.27E –05 *NOH *SCR +0.005154 *NOH *DOC –0.00557 *NOH *FC – 2.98637 *PHC *IP +3.72013 *PHC *LOAD –1.28E –04 *PHC *SCR –0.01611 *PHC *DOC +0.119764 *PHC *FC +0.061613 *IP *LOAD –1.11E –06 *IP *SCR +3.10E –05 *IP *DOC +0.000162 *IP *FC +1.58E –05 *LOAD *SCR –6.55E –04 *LOAD *DOC –0.00399 *LOAD *FC +1.51E –08 *SCR *DOC +6.14E –08 *SCR *FC –2.71E –06 *DOC *FC.

Equations 2: HC= –57.6508 +8.77475 *NOH –197.312 *PHC +0.043036 *IP +7.35443 *LOAD +0.002631 *SCR +0.029866 *DOC +0.007783 *FC +118.573 *PHC *PHC +0.000214 *IP *IP –8.88E –04 *LOAD *LOAD –3.89E –08 *SCR *SCR –1.03E –05 *DOC *DOC +5.08E –07 *FC *FC +15.2647 *NOH *PHC –0.02969 *NOH *IP +0.196934 *NOH *LOAD +1.32E –05 *NOH *SCR –0.00149 *NOH *DOC +0.001084 *NOH *FC –0.21581 *PHC *IP –0.51028 *PHC *LOAD +0.00014 *PHC *SCR +0.008923 *PHC *DOC +0.004582 *PHC *FC –0.004 *IP *LOAD –1.13E – 07 *IP *SCR +4.07E –05 *IP *DOC –1.89E –05 *IP *FC – 1.77E –06 *LOAD *SCR –0.00181 *LOAD *DOC +0.000302 *LOAD *FC +1.97E –08 *SCR *DOC +2.38E – 10 *SCR *FC –3.64E –06 *DOC *FC.

Equations 3: CO= +1.44933 –2.27E –01 *NOH –8.04E –01 *PHC +1.96E –03 *IP –0.05105 *LOAD –1.21E –06 *SCR – 8.43E –04 *DOC –1.33E –04 *FC +4.53E –01 *PHC *PHC +2.71E –07 *IP *IP +2.43E –03 *LOAD *LOAD +2.29E –11 *SCR *SCR –9.77E –08 *DOC *DOC +8.04E –09 *FC *FC +1.71E –02 *NOH *PHC –7.19E –05 *NOH *IP +1.04E –03 *NOH *LOAD –1.11E –07 *NOH *SCR +1.91E –04 *NOH *DOC +5.35E –06 *NOH *FC –3.00E –05 *PHC *IP –2.23E – 02 *PHC *LOAD –2.78E –07 *PHC *SCR +7.86E –05 *PHC *DOC –2.92E –05 *PHC *FC –6.66E –05 *IP *LOAD +2.58E – 10 *IP *SCR +3.18E –07 *IP *DOC –1.16E –07 *IP *FC +1.52E –08 *LOAD *SCR –1.33E –05 *LOAD *DOC +9.22E –06 *LOAD *FC –1.28E –10 *SCR *DOC +8.30E –12 *SCR *FC – 2.81E –08 *DOC *FC.

Equations 4: CO2= +976.597 +5.74711 *NOH –1120.89 *PHC –0.06138 *IP –1.19812 *LOAD +0.000137 *SCR –7.77E –04 *DOC +0.000257 *FC +325.048 *PHC *PHC +2.28E –05 *IP *IP +9.09E –04 *LOAD *LOAD –1.28E –09 *SCR *SCR – 8.63E –08 *DOC *DOC +5.31E –08 *FC *FC –2.03247 *NOH *PHC –0.00191 *NOH *IP –0.01551 *NOH *LOAD +1.21E – 06 *NOH *SCR +8.52E –05 *NOH *DOC +5.74E –05 *NOH *FC +0.015539 *PHC *IP +0.972406 *PHC *LOAD –2.68E – 05 *PHC *SCR +0.000301 *PHC *DOC +0.000677 *PHC *FC –2.42E –04 *IP *LOAD –8.97E –09 *IP *SCR –1.79E –06 *IP *DOC –1.96E –06 *IP *FC –2.32E –08 *LOAD *SCR +6.84E – 05 *LOAD *DOC +1.75E –05 *LOAD *FC –7.26E –10 *SCR *DOC +3.52E –10 *SCR *FC +1.48E –07 *DOC *FC.

Equations 5: BTE= +0.14946 –6.16E –08 *NOH +3.06E –07 *PHC –5.58E –10 *IP –0.02491 *LOAD +3.46E –13 *SCR +1.17E –11 *DOC –8.12E –04 *FC –1.18E –07 *PHC *PHC – 2.14E –13 *IP *IP +6.77E –11 *LOAD *LOAD +2.28E –19 *SCR *SCR +7.44E –16 *DOC *DOC –1.38E –15 *FC *FC +9.00E –09 *NOH *PHC +6.83E –11 *NOH *IP –1.13E –09 *NOH *LOAD –7.69E –14 *NOH *SCR –7.56E –13 *NOH *DOC –3.78E –12 *NOH *FC +2.54E –11 *PHC *IP –5.89E – 11 *PHC *LOAD +7.02E –14 *PHC *SCR –4.09E –12 *PHC *DOC +1.79E –15 *PHC *FC +1.22E –11 *IP *LOAD –4.65E – 16 *IP *SCR –1.73E –15 *IP *DOC +5.31E –14 *IP *FC +1.24E –14 *LOAD *SCR –1.24E –14 *LOAD *DOC +0.000135 *LOAD *FC.

These equations were used as functions in MATLAB editor to be accessed by the optimization tool of MATLAB and the same is discussed in the following section.

Discussion

Optimization using genetic algorithm

Optimization was done using optimization tool of MATLAB, optimization technique in genetic algorithm was used for this purpose. The optimum sets of parameters are obtained separately for the standalone engine [12].

Regression coefficients obtained from optimization tool of MATLAB are tabulated in Table 2.

Table 2: Regression coefficients obtained from MINITAB 14.

Since the regression coefficients are above 90% the equations formed are acceptable, hence used as mathematical models in genetic algorithm.

The optimum values in number of nozzle holes, piston head clearance, injection pressure, load, direct oxidization catalyst and selective reduction catalyst are obtained for the stand alone engine and the same are shown in Table 3 [13].

Table 3: Optimum set of parameters for standalone engine.

From the above results a four holed nozzle proved to be more effective than a three holed nozzle. Piston head clearance and injection pressure increases with load. Direct oxidization catalyst magnesium oxide was efficient whereas anhydrous acetic acid and cupric oxide are not suitable for the selected loads. Selective reduction catalyst vanadium pent oxide is more efficient than the oxides of tungsten for the selected loads [14].

Performance and emission for optimum sets

The experimentally investigated performance and emission values for optimum sets when imposed on the engine are given in Table 4.

Table 4: Experimentally investigated values of the optimum set.

The performance and emission values predicted by GA for optimum sets when imposed on the engine are given in Table 5.

Table 5: Performance and emission for optimum sets predicted by GA.

It is apparent that the values forecasted by GA show close relationship with the experimental results hence GA can be used as a tool for predicting the operating parameters of an engine.

The performance and emission for the load applications selected at factory settings (three whole nozzle, piston head clearance of 1.4 mm and an injection pressure of 170 bars) are shown in Table 6.

Table 6: Performance and Emission at Factory Settings.

The above results show 7.68% decrease in NOX, 8.33% decrease in HC, 50% decrease in CO, 2.25% increase in CO2 and 0.7% increase in brake thermal efficiency. Hence changing the engine settings has enhanced combustion which led to the increase in in efficiency and carbon di oxide, while catalytic reaction reduced carbon mono oxide and oxides of nitrogen.

Conclusion

This study deals with the conduct of experiment on a typical engine by varying a set of parameters and assessing the engine with regard to power and emission. Using the data, Genetic Algorithm in MINITAB and MATLAB software, the optimum settings for minimum emission and maximum efficiency is instituted. As a result of implementing the optimum settings the harmful emissions (Carbon monoxide and Oxides of Nitrogen) of the engine are reduced. Carbon monoxide is reduced remarkably, Hydrocarbons are brought down significantly and Oxides of Nitrogen are condensed notably. This largely reduces the environmental effect caused by the engine. Apart from emission reduction the implementation of optimum parameters also induces a marginal increase in efficiency due to complete and enhanced combustion, which has been verified through experiments. Additionally the optimal settings will provide economic benefits to users due to increased efficiency. More parameters like type of combustion including pre-chamber type, retardation in injection timing, and number of injectors, induction swirl, and shape of combustion chamber can be chosen and better parameter settings, reducing emissions and improving performance may be obtained.

References

1. Salasoo L, Primus R, Pintgen F, Topinka J, Dean A. 21st Century Locomotive Technology: Quarterly Technical Status Report 20 DOE/AL68284-TSR20. GE Global Research. 2008.

   [Google Scholar]

2. Bakar RA, Semin AR, Ali I. Computational simulation of fuel nozzle multi holes geometries effect on direct injection diesel engine performance using GT-POWER. Am J Appl Sci. 2008;5(2): 110-116.

   [Crossref] [Google Scholar]

3. Semin RB, Ismail R. Investigation of fuel nozzle performance on direct injection diesel engines using computational simulation. Appl Therm Eng. 2006;5(6): 610-617.
4. Zheng M, Reader GT, Hawley JG. Diesel engine exhausts gas recirculation-A review on advanced and novel concepts. Energy Convers Manag. 2004;45(6): 883-900.

   [Crossref] [Google Scholar]

5. Bakar RA, Ismail S, Ismail AR. Fuel injection pressure effect on performance of direct injection diesel engines based on experiment. Am J Appl Sci. 2008;5(3): 197-202.

   [Crossref] [Google Scholar]

6. Kannan K, Udayakumar M. Experimental study of the effect of fuel injection pressure on diesel engine performance and emission. ARPN J Eng Appl Sci. 2010;5(5): 42-45.

   [Crossref] [Google Scholar]

7. Kuleshov AS. Multi-zone DI diesel spray combustion model for thermodynamic simulation of engine with PCCI and high EGR level. SAE Int J Engines. 2009;2(1): 1811-1834.

   [Google Scholar]

8. Atkinson CM, Long TW, Hanzevack EL. Virtual sensing: A neural network-based intelligent performance and emissions prediction system for on-board diagnostics and engine control. Progress in Technology. 1998;73(301-314): 2-4.

   [Google Scholar]

9. Kesgin U. Genetic algorithm and artificial neural network for engine optimisation of efficiency and NOx emission. Fuel. 2004;83(7-8): 885-895.

   [Crossref] [Google Scholar]

10. Adelman BJ, Beutel T, Lei GD, Sachtler WM. Mechanistic cause of hydrocarbon specificity over Cu/ZSM-5 and Co/ZSM-5 catalysts in the selective catalytic reduction of NOx. J Catal. 1996;158(1): 327-335.

    [Crossref] [Google Scholar]

11. Chen G. New Advances in Catalytic Systems for Conversion of CH4 and CO2. J Nat Gas Chem. 2002;11(3): 109-116.

    [Google Scholar]

12. Howlett RJ, Walters SD, Howson PA, Park I. Air-fuel ratio measurement in an internal combustion engine using a neural network. Advances in Vehicle Control and Safety. 1998;98(1): 1-10.

    [Google Scholar]

13. Manjunatha R, Narayana PB, Reddy KH. Application of Artificial Neural Networks for emission modelling of biodiesels for a CI engine under varying operating conditions. Mod Appl Sci. 2010;4(3): 77.

    [Crossref] [Google Scholar]

14. Ghobadian B, Rahimi H, Nikbakht AM, Najafi G, Yusaf TF. Diesel engine performance and exhaust emission analysis using waste cooking biodiesel fuel with an artificial neural network. Renew Energy. 2009;34(4): 976-982.

    [Crossref] [Google Scholar]