Multiphysics distribution characteristics and performance study of kWclass crossflow SOFC stack

Solid Oxide Fuel Cell (SOFC) usually operates in a high temperature environment above 700 ℃. The experimental research of a largescale stack is expensive, and it is difficult to evaluate the internal state of the stack. Modeling the SOFC stack has become an important research method to explore its internal gas thermalelectricity distribution characteristics. Due to the mutual coupling of various physical fields such as chemical reaction, electrochemical reaction, heat transfer, and mass transfer in the fuel cell, it is difficult to simulate and calculate the largescale stacks with conventional multi physics fully coupled models. Therefore, a multi physics decoupling model based on the BP neural network is proposed, and the flow field, temperature field distribution characteristics and I-V performance of the kilowatt class crossflow stack are simulated and calculated. The research results show that increasing the intake flow rate and operating temperature can directly enhance the output power of the stack. However, these operations reduce stack fuel utilization and power generation efficiency, and lead to increased thermal stress inside the stack, with the risk of leakage and structural damage. The height of the anode PEN and the volume of the gas distribution cavity should be carefully chosen since they can significantly affect the gas flow uniformity of the stack. The results show that the gas flow uniformity of the stack is improved by 2% when the height of PEN is changed from 0.5 mm to 0.3 mm, and the maximum temperature difference is reduced by 17%. Therefore, a stack with better gas flow uniformity can significantly improve its temperature field distribution and reduce its maximum temperature and internal thermal stress. The verification with the experimental results of the kilowattclass crossflow stack shows that the multiphysics decoupling model can be well applied to the simulation of largescale SOFC stacks and lay a foundation for the further optimization of largescale stacks.