Optimization of performance and emissions of a CI engine using butanol-TiO ₂ nano-biodiesel blends: RSM-ANN methodology

The objective of this research is to evaluate the impact of butanol and TiO2 nanoparticles blended into biodiesel produced from mango seeds on the engine performance and exhaust emission of the diesel engine operating under different loads. Response surface methodology with the aid of central composite design was used for modeling and optimization, where engine load, butanol content, and TiO2 nanoparticle content were set as input variables, whereas brake thermal efficiency, specific fuel consumption, carbon monoxide, unburnt hydrocarbons, and nitrogen oxides emissions were chosen as response variables. According to the experimental results, an increase in the engine load and the ratio of additives increased premixing combustion, fuel atomization, and catalytic oxidation, improving engine performance and reducing emissions. Analysis of variance showed that the generated models had significant predictive value (p < 0.0001) with high accuracy (R2: BTE = 0.9953, SFC = 0.9987, CO = 0.9749, UHC = 0.9919, NOx = 0.9729). Multi-objective optimization using desirability functions provided optimum levels for 0.73 butanol, 0.00 TiO2, and 0.00 load, giving composite desirability of 0.658. At optimal conditions, the brake thermal efficiency, specific fuel consumption, carbon monoxide emission, unburnt hydrocarbons emission, and nitrogen oxides emission were estimated to be 25.89%, 0.329 kg/kWh, 0.0428%, 115.91, and 148.39 ppm, respectively. In order to improve the reliability of optimization process, an artificial neural network was designed based on experimental results. The artificial neural network predictions had high consistency with RSM results, with prediction error <5% and R2 equal to 0.93.