Experimental and machine-learning based investigation of second generation juliflora biodiesel blended with dimethyl carbonate and CoCr 2 O 4 nanoparticles in a compression ignition engine

Improving the efficiency and emission characteristics of biodiesel fueled CI engine remains challenging due to limitations in fuel reactivity and combustion behavior. This study investigates the combined effect of dimethyl carbonate (DMC) and cobalt chromite nanoparticles on engine performance and emissions, supported by machine learning-based prediction. Experiments were carried out on a single-cylinder diesel engine using five fuel blends including diesel, J20, J20D10, J20D10C50, and J20D10C100 across the full load range. The J20D10C100 blend delivered the highest BTE of 29.67%, showing improvements of 13.57% over diesel and 20.78% over the J20 blend. Brake specific fuel consumption decreased to 0.375 kg per kWh, corresponding to reductions of 5.44% and 6.38% compared with diesel and the J20 blend. At full load, CO decreased from 0.53% to 0.454%, HC reduced by 14.6%, and nitrogen oxides decreased from 1070 ppm to 990 ppm. Smoke opacity showed a reduction from 67.21% to 65.19%. Among the predictive models, gradient boosting achieved the highest accuracy with an R 2 of 0.965, followed closely by random forest with 0.964. The results confirm that the proposed fuel formulation enhances engine performance while reducing emissions, which supports reliable predictive modeling.