Predictive Modeling of Energy and Exergy Effects from Injector Placement in HCCI Engines Running on Diethyl Ether and Biogas Using Machine Learning Techniques

Abstract

The global emphasis on sustainable energy has highlighted biogas as a viable alternative fuel for internal combustion engines. Among advanced engine technologies, Homogeneous Charge Compression Ignition (HCCI) engines stand out due to their notable efficiency and low emissions, making them well-suited for clean energy applications. A key aspect in maximizing the benefits of biogas in HCCI engines, especially when paired with diethyl ether (DEE) as a pilot fuel, involves determining the optimal injector location. This study employed machine learning techniques to identify the most effective injector position by analyzing variables such as engine load, biogas flow rate, methane content, and intake air temperature. Three injector configurations were tested: one at the intake port and two further upstream at 6 cm (Manifold 1) and 10 cm (Manifold 2) from the intake. To solve this complex optimization problem, five advanced machine learning models SVM, Random Forest, AdaBoost, CATBoost, and XGBoost were utilized to predict crucial performance metrics, including brake-specific energy use, exergy losses, thermal efficiency, fuel consumption, and exergy-based efficiency. XGBoost emerged as the most accurate model, consistently delivering the highest R² values. The findings underscore the potential of machine learning to enhance injector placement strategies, thereby improving the energy efficiency and overall performance of HCCI engines operating on biogas and DEE.