Multi-objective optimization of a natural aspirated three-cylinder spark ignition engine using modified non-dominated sorting genetic algorithm and multicriteria decision making

The main purpose of this study is to find a special set of filling and emptying parameters of a three-cylinder engine, including geometrical design of intake manifold, intake and exhaust valve timing as design variables of the multi-objective optimization problem, which lead to the best values of BSFC (Brake Specific Fuel Consumption), and torque of the engine at all working speeds as four separate objective functions. The modified Non-dominated Sorting Genetic Algorithm (NSGA-II), which is an evolutionary Pareto-based method, is used as the optimization algorithm, and the technique for order of preference by similarity to ideal solution is used as a Multi-Criteria Decision Making Method to select the trade-off design based on different design strategies through the design vectors which are proposed. Group Method of Data Handling-type of Artificial Neural Networks is used to predict the relation between the design variables and the objective functions based on a 243-sized set of samples, which are chosen by Factorial method, and the corresponding engine designs are simulated using GT-SUITE as an Engine Simulation Software. The method of simulation is verified by comparison of experiment and simulation results. The verification process shows that the engine simulation code could predict the engine's performance by 4.78% error. Convergence and start point independence of the optimization algorithm (modified NSGA-II) is also investigated. The final results show remarkable improvement in BSFC and torque at 3500 RPM and mean value of BSFC at all working speeds along with a small reduction in mean value of the engine's torque at all working speeds compared to corresponding characteristics of a similar three-cylinder engine. (C) 2016 AIP Publishing LLC.