Predicting Biodiesel Ignition Delay Using a Skeletal Multistep Chemical Reaction Scheme

From an environmental standpoint, biodiesel fuels have been under consideration for use in compression ignition engines as a viable alternative to fossil diesel. Though there are several simple to detailed kinetics investigations concerning fossil diesel, the studies on biodiesel ignition kinetics are scarce. Among engine modelers, the Shell ignition model is a widely accepted multistep reaction scheme for predicting diesel ignition. This work optimizes the multistep reaction shell model for predicting ignition delay for major ester constituents of biodiesel fuel, namely, methyl palmitate, methyl stearate, methyl oleate, methyl linoleate, and methyl linolenate. The model and its optimized parameters obtained using a genetic algorithm are useful in predicting the ignition delay of biodiesel from its composition details. The values of ignition delay of biodiesel constituents predicted from the optimized model over the range of equivalence ratios and pressures compare well with those obtained from a detailed reaction mechanism. The optimized Shell model is validated with experimental ignition delay data at an equivalence ratio of 0.5-1 with an average error of 32%. The optimized Shell model predicts the ignition delay under typical compression engine conditions (p = 40 bar, T < 1000 K) with an average error of 28% and the least computational time of less than a minute compared to over 2 h for the detailed reaction mechanism. However, the ignition delay prediction for methyl esters and biodiesel fuels at an equivalence ratio of 0.25 remains qualitative, with the model overpredicting ignition delay values by a factor of 3. The sensitivity analysis reveals that the ignition delay is most sensitive toward the main propagation reaction at all temperature regimes. On modifying the rate of the main propagation reaction, the model predictions for methyl esters and biodiesel fuels at a low equivalence ratio of 0.25 are improved considerably with the average error reduced to 40%.