Desirability-based optimization of dual-fuel diesel engine using acetylene as an alternative fuel

Abstract

The study examined the dual-fuel engine performance employing acetylene gas as primary fuel and diesel as pilot fuel. The engine's operational parameters were adjusted using the Box-Behnken design, and the results were recorded. The best operating settings were yielded as 81.25 % engine load, 4.48 lpm acetylene gas flow rate and the compression ratio were 18. At this optimized setting the BTE was 27.1 % and the engine emitted 360 ppm of NOx, 56.2 ppm of HC, 104 ppm of CO. The experimental data at optimized setting was compared to the optimized results, and the percentage of errors was within 7 %. Two advanced machine learning methods, LightGBM and Tweedie, were used to predict engine efficiency and emissions. Tweedie-based models had an R2 value of 0.89–1, while LightGBM-based models had an R2 value of 0.38–1. The mean squared error was 0.24–45.04 for Tweedie-based models and 8.5 to 153.89 for LightGBM-based models. On the basis of R2 and MSE, it was observed that Tweedie was superior at making predictions than LightGBM. The study demonstrated the efficient functioning of a dual-fuel engine using acetylene as an alternative fuel for increased performance and lower emissions.

Keywords

• Dual-fuel engine
• Optimization
• Acetylene gas
• Alternative fuel
• Sustainable energy
• Machine learning