Green Valorization of Cassia Fistula Biodiesel: Experimental Insights, RSM-Machine Learning Optimization, and Sustainable Engine Applications

Abstract

The urgent global transition toward sustainable energy sources necessitates green alternative fuels that mitigate environmental burdens while ensuring compatibility with compression ignition engines. Biodiesel represents a renewable and biodegradable fuel, but variability in performance and emissions across feedstocks requires advanced green engineering strategies. In this study, a non-conventional seed oil (Cassia fistula), rarely explored in biodiesel research, was converted via transesterification using a wastederived CaO catalyst (eggshells), demonstrating circular economy principles. Biodiesel blends (B10, B20, B30) were tested in a single-cylinder diesel engine under varying loads and injection timings. Experimental responses (BTE, BSFC, NOx, CO, HC, CO2, cylinder pressure) were optimized using Response Surface Methodology (RSM, CCD design). Machine learning (Random Forest, Support Vector Regression, and Artificial Neural Networks) enabled predictive modeling of performance and emissions. Results confirmed B20 as the optimal trade-off between efficiency and emissions, with substantial potential for reducing lifecycle carbon intensity. RSM analysis revealed significant interactions between blend ratio and injection timing, validated by ANOVA (p < 0.001). Machine learning predictions achieved high accuracy (Random Forest R² = 0.97 for BTE). This integrated green framework-combining waste-derived catalyst utilization, experimental validation, statistical optimization, and computational intelligence-advances biodiesel research toward scalable, sustainable, and low-carbon energy systems.