Optimisation of a solid oxide fuel cell reformer using surrogate modelling, design of experiments and computational fluid dynamics

Abstract

The present paper proposes a trio approach including surrogate modelling, design of experiments and computational fluid dynamics. An experimentally validated 3D continuum model has been used in the aid to optimise the performance, weight and the associated manufacturing costs of a plate type pre-reformer. The effect of plate number, plate porosity, wire mesh porosity and alternative materials on the pre-reformer performance has been studied in detail using a D-optimal experimental design plan and computational fluid dynamics. Multi-regression analyses depict that the number of reformer plates has the greatest potential for optimisation. A surrogate model has been derived and employed to perform rapid process and design optimisations. Results show that the methane reforming can be 40% increased by reducing the pre-reformer plate number to half of the actual value. A mass reduction of 50%, associated with a saving of 50% may be enabled. The surrogate model is proven to be a powerful tool to aid in reduced physical prototype costs and product development time.