Fuel adaptive analysis and methane conversion rate prediction based on Gaussian process regression for an SR-SOFC system

Abstract

Solid oxide fuel cell (SOFC) has strong fuel adaptability, but different fuel components after reformer reforming will make SOFC have different performance characteristics. The reformer performance will directly affect the stack performance of the external steam-reforming SOFC system fed by methane (SR-SOFC), such as stack voltage and temperature distribution, and stack electrical efficiency. The existing SOFC system researches focus on the factors affecting the reforming performance, but ignore the influence of the reforming performance on the stack thermoelectric characteristics. An effective indicator of reforming performance is methane conversion rate. In this paper, the influence of methane conversion rate on the stack thermoelectric characteristics is analyzed through a high-fidelity physical SR-SOFC system model, which is verified by 200-hour experimental data obtained from a 1kW external steam-reforming SOFC independent power generation system. Then, a Gaussian process regression (GPR) model is constructed to predict the methane conversion rate, since it is difficult to directly measure the methane conversion rate without damaging the thermal insulation performance of the hot zone. The results show that, different methane conversion rates in the SR-SOFC system can shift the stack temperature distribution significantly, and properly increasing the methane conversion rate can effectively increase the stack voltage and electrical efficiency. Furthermore, the GPR model-based estimator can accurately predict the methane conversion rate with MAPE of 0.030%. This work lays a foundation for SR-SOFC stack thermal management, ensuring the safety of the stack temperature and improving the stack performance.