Experimental and artificial neural network optimization of the effects of oil alcohols and biodiesel blends on combustion and performance parameters in a diesel engine

In this study, oil was extracted from safflower seeds (SSs) using different methods, and the extracted SSO was synthesized into B100 through the transesterification process. The performance, emissions, and combustion parameters of fuels obtained by blending B100, DF, and different heavy alcohols (HA) in specific proportions were investigated. The experimental results of combustion parameters were predicted using ANNs. Among the extraction methods, the soxhlet extraction method achieved a maximum oil extraction efficiency of 32.49%. The FA compositions of SSO are approximately 30.9% and 56.4% for C18:1 and C18:2, respectively. The alcohol/oil molar ratio for the synthesis of SSO biodiesel was calculated as 4.799/1. The synthesis of biodiesel was confirmed using spectroscopic methods. The conversion of fatty acids to FAMEs was determined to be 95.23% using 1H NMR spectroscopy. Based on the conducted analyses, the synthesized biodiesel was found to meet the ASTM-D6751 and EN-14214 standards. In the tests, the fuel blend DF50DE50 resulted in the lowest brake-specific fuel consumption (BSFC) and brake-specific energy consumption (BSEC) compared to DF, while it exhibited the highest brake thermal efficiency (BTE) among the fuels tested. The fuel blend DF50HE50 achieved the highest exhaust gas temperature (EGT), while the lowest EGT was obtained with the fuel blend DF50B25HE25. The NOx emissions produced by B100 are higher compared to DF. Low NOx emissions were produced by mixtures containing decanol. The CO emissions produced by B100 are higher compared to those generated by DF. The CO emissions produced by DF50B50 are lower compared to those generated by single and dual fuel blends. The particulate matter (PM) emissions produced by DF are higher compared to those generated by all tested fuels. DF50OC50 generates high cylinder pressures (CP) with increasing loads, while the highest CP is achieved with the DF50HE50 fuel blend. The net heat release rate (NHRR) of the DF50HE50 fuel blend is the highest. The pressure rise rate (PRR) of B100 is higher than that of DF, while the PRR of the DF50B50 fuel blend is lower compared to single fuels. It has been observed that the prediction of combustion data using ANNs is highly successful, as indicated by the low MSE values.