Our group has recently proposed a large scale optimization technique for computing the optimum distribution of platinum particles in proton exchange membrane fuel cells (PEMFCs) [1]. The basic idea of the technique was to use the adjoint method to compute the catalyst sensitivity functions and to introduce these functions in a gradient-based algorithm to estimate the optimum catalyst distribution in the catalyst layers [2]. In our previous work, constraints were taken into consideration using the formalism of Lagrange multipliers on the functional space of possible catalyst distributions [3]. Although robust, this approach was restricted to equality constraints and could not be applied to the more general case in which constraints enter as inequalities in this optimization function.

In this work we will present a new optimization method for PEFMCs that includes inequality constraints. The method is still based on the computation of the catalyst sensitivity functions using the adjoint method but allows the optimizer to search for solutions away from the boundaries imposed by the constraints. By using this method, we are able to compute the global optimum platinum distribution inside the catalyst layer. In addition, we extend the method to the calculation of the optimum porosity and Nafion distribution in the gas diffusion and catalyst layers, respectively. Details about the numerical implementation of the method and sample optimization results will be presented and compared to previous results at the meeting.

References

[1] J. Lamb, G. Mixon and P. Andrei, “Adjoint Method for the Optimization of the Catalyst Distribution in Proton Exchange Membrane Fuel Cells”, Journal of the Electrochemical Society, 164, E3232 (2017).

[2] J. Lamb, G. Mixon, and P. Andrei, “Mathematical Optimization of the Spatial Distribution of Platinum Particles in the Catalyst Layer of PEMFCs”, ECS Transactions 2017 77(11): 1179-1196.

[3] J. Lamb and P. Andrei, “Analysis of Sensitivity of PEMFC Parameters to Non-Uniform Platinum Deposition Diagnostics and Characterization Methods”, ECS Transactions 2017 80(8): 291-300.