Performance optimization of RCCI engines running on landfill gas, propane and hydrogen through the deep neural network and genetic algorithm

Abstract

The present study is to optimize the performance of the reactivity controlled compression ignition engines taking landfill gas plus propane or hydrogen or a combination of both as low reactivity fuels. A heavy duty, 2.44 L caterpillar 3401E single-cylinder diesel engine is employed for the numerical simulation, and a number of the most effective variables on the engine performance are determined as inputs. The research purposes are to procure the maximum energy efficiency (i.e. the maximum thermodynamics first and second law efficiencies, and the minimum indicated specific fuel consumption), and to fulfill the Euro VI engine emissions standards (i.e. CO, UHC, and NOX maximum emissions 1.5, 0.13, and 0.4 gr/kW-hr, respectively). The maximum peak pressure and peak pressure rise rate are restricted to 150 bar and 10 bar/crank angle deg., respectively. The deep neural networks are utilized to obtain the engine mathematical model and the genetic algorithms are used to attain the intended optimization. Based on the outcomes, injecting a combination of propane and hydrogen to the premixed charge by 3.4 and 6.6 % by volume, respectively, and adopting an engine equivalence ratio of 0.31, an intake pressure of 1.983 bar, and a diesel fuel injection timing of about 75 deg. before the top dead center provide the maximum performance for the considered engine. In the optimized state, a 6.6 % hydrogen by volume takes 2.5 % of the total fuels energy. This research has proven that numerical modeling in combination with deep neural networks and genetic algorithms is an effective methodology to optimize the performance of the reactivity controlled compression ignition engines. Moreover, a dimensional analysis made shows the dimensionless peak pressure is function of the dimensionless engine heat release and the ratio of the engine cumulative heat release to the primary inlet mixture flow work. In the optimized case, the dimensionless ratio of cumulative heat release to the inlet flow work of fluid is 11.5.