Performance and emission characteristics of novel biodiesel-diesel blends: an RSM and ANN approach

Abstract

Abstract In this paper, the impact of different input variables on the performance and emission features of a pongamia pinnata and rapeseed oil biodiesel with n-Butanol additive were investigated, statistically analyzed, and optimized by employing the powerful response surface methodology (RSM) based design of experiment (DOE) techniques. The vegetable oils (pongamia pinnata and rapeseed oils) were transesterified and their corresponding methyl esters were blended with diesel and n-Butanol at blend ratios 10:84:6, 10:78:12, 20:74:6 and 20:68:12. The samples were tested on a direct injection CI engine at a rated speed of 1500 rpm and standard CR of 17.5:1 at different loads. In each test, performance and emission parameters were measured. Expert machine learning (ML) methods were used to forecast these features. In addition, polynomial equations were developed for each blend using regression techniques and compared with an artificial intelligence technique. It was observed that the engine performance increased as biodiesel and additive weight percentage increased. Regardless of the loads placed on the engine and the blend ratios, the use of PPME and RSME combined with n-Butanol blends demonstrated a clear decrease in NOx compared to diesel (7.07% for P20B12 and 6.58% for R20B12). As per the trend, it is seen that the percentage reduction in CO2 emissions is greater with high percentage increase of n-Butanol in the tested sample irrespective of loads applied on the engine (2.95% more P20B12 for as compared to P20B6). For the emission characteristics, ANN demonstrated a range of 87.92% to 98.83% prediction accuracy while that of regression varies from 81.4% to 98.8% for all the samples of PPME blended biodiesel.