Optimization of next-generation alcohols and fishoil methyl ester blends in a single cylinder DI-CI engine using response surface methodology

Abstract

The present investigation delivers a comprehensive viewpoint on the current artificial intelligence (AI) meta-modelling in diesel engine system, particularly in the domains of multi-objective optimization. The relevance to the benefits of AI built modelling stratagems, and the probable Response Surface Methodology has been efficient with the consecutive growth in the current domains of the compression ignition technology. The study establishes the fundamental significance of inspecting certain multi-objective optimization stratagems matching with the accumulating need to deal with emission-performance trade-off trials of the compression ignition technology. To achieve the reliability and versatility of such model centered multi-objective optimization strategies, the current study delivers a unique case-study presenting the creditably accurate model-based standardization in an existing diesel engine. Therefore, n-Octanol produced in renewable ways, methyl esters of fish oil (MEFO) blended with diesel is used as fuel and the experiments were designed by Design of Experiments (DoE) based on response surface methodology (RSM) architecture. The results depicted that the tailor-made fuels proved their ability in terms of both performance and emissions when compared to mineral diesel. The model is further tested on a statistical platform with some special error matrices like Mean Squared Relative Error, MSRE, and Nash–Sutcliffe Coefficient of Efficiency, NCE along with conventional model testing metrics like MSE, RMSE and R which proved that the proposed model is robust and efficient in predicting the input-output paradigm. The ranges of correlation coefficients R, R2 and NCE are 0.99786-0.999992, 0.99786-0.999992 and 0.9957 – 0.999984 respectively. And the ranges of the error metrics Theil U2 and MSRE are 0.004048 0.065246 and 6.07E-07 to 0.000158 respectively. Optimization of input parameters was performed using the desirability approach of the response surface methodology for better performance and lower NOx and CO emission at a desirability index of 0.986. Experimental validation suggested a blend of 20% MEFO and 10% n-Octanol with petrodiesel at full loads were found to be optimal values for the test engine.