Garcinia indica (GI) feedstock poses high oil content (45.2%) and after transesterification resulted with 94.8% yield. The GI crude oil, biodiesel, and their blends were tested for fuel characteristics and run in a diesel engine. Response surface methodology based on central composite design experimental matrices was used to model and examine the input variables (engine load, injection timing, injection pressure, and blend type) on engine performance (brake thermal efficiency (BTE), brake specific fuel consumption (BSFC)) and emission characteristics (nitrogen oxide (NOx), unburnt hydrocarbon (UHC), carbon monoxide (CO)). All factors showed significant effects (except injection pressure and injection time for NOx) on all responses. The empirical equations predicted 27 experimental cases with 4.75% accuracy. Desirability function approach was applied to transform all output functions (maximize BTE and minimize BSFC, CO, NOx, and UHC) with different weight fractions (WF) to single composite desirability function for maximization. Teaching learning-based optimization (TLBO) determined optimal condition corresponding to case 4 (maximum WF to CO, minimum WF to BTE, BSFC, UHC, NOx) resulting in highest desirability function value (0.9432) with a percent deviation of 7.09%. The developed models assist novice users in predicting unknown parametric conditions and improving engine performance and emission characteristics without practical experiments.
Read full abstract