Multi-objective optimization of proton exchange membrane fuel cells by RSM and NSGA-II

This study optimized the performance of a proton exchange membrane fuel cell by combining the response surface methodology and non-dominated ranking genetic algorithm. Firstly, the design variables are determined, including operating pressure (p), operating temperature (T), Anode stoichiometry ratio (lambda(a)), thickness of the proton exchange membrane (H-mem) and gas diffusion layer (GDL) porosity (epsilon(GDL)). The objective functions are also identified, including power density (P), system efficiency (eta) and exergy efficiency. Then, the Box-Behnken design is employed to arrange the numerical investigations. Analysis of variance is used to verify the appropriateness and reliability of the constructed regression models. Response surface analysis is used to show the interaction between each pair of design parameters. Finally, the Pareto optimal frontier is obtained by non-dominated ranking genetic algorithm II and the regression models constructed by response surface methodology. The Pareto optimal solution offers a power density of 0.6327 W center dot cm(-2), a system efficiency of 26.16% and an exergy efficiency of 43.94 %, which is 13.18 %, 7.06 % and 20.29 % better than the initial direct current channel, respectively. The corresponding design variables is p = 2.6498 atm, T = 341.621 K, lambda(a) = 1.1808, H-mem = 0.0577 mm and epsilon(GDL) = 0.4908. This work provides a new multi-objective optimization method for designing more efficient proton exchange membrane fuel cells.

Automated summary

This study optimized the performance of a proton exchange membrane fuel cell by using response surface methodology and non-dominated ranking genetic algorithm. The Pareto optimal solution obtained through this method improves the power density, system efficiency, and exergy efficiency of the fuel cell. This work provides a new multi-objective optimization method for designing more efficient fuel cells.