Parameter estimation for empirical and semi-empirical models in a direct ethanol fuel cell

Abstract

Experimental data from a Direct Ethanol Fuel Cell (DEFC) provides a general perspective about its performance; nevertheless, it does not provide information about the cell’s physical characteristics nor information to improve its performance. On the other hand, numerical simulation can be used to test the cell’s design and boost its performance but requires a set of physical parameters. In this proposal, we introduce a novel modification to an empirical model for a Direct Methanol Fuel Cell to make it suitable for DEFC simulations at different temperatures by a new semi-empirical mathematical model. In addition, we introduce temperature-depending parametric forms of several terms to reduce the number of possible parameters to estimate from the DEFC. Then, we combined the models with an estimation of distribution algorithm to find the numerical simulation that best reproduces the experimental polarization curve. The method is validated by estimating the parameters to reproduce the experimental data at different temperatures reported in the literature, and with data obtained in an in-house open-cathode DEFC, recorded at a scan rate of 10 mVs−1, using as fuel CH3CH2OH 1 M at 25 °C and 60 °C. From the estimation results at temperature set T1→=(T1a,T1c)∘C, the same parameters are used for a simulation at T2→=(T2a,T2c)∘C, demonstrating that it reproduces the two experimental polarization curves. Hence, the models and methods presented here can be used to reduce physical experimentation and to test different designs and operation settings.