Optimizing the homogeneity and efficiency of a solid oxide electrolysis cell based on multiphysics simulation and data-driven surrogate model
Optimizing the homogeneity and efficiency of a solid oxide electrolysis cell based on multiphysics simulation and data-driven surrogate model
复制标题
基于多物理场模拟和数据驱动替代模型优化固体氧化物电解池的均匀性和效率
DOI：
10.1016/j.jpowsour.2023.232760
复制
发表时间：
2023-04
影响因子：
9.2
通讯作者：
Ying-wei Chi;Ken Yokoo;H. Nakajima;Kohei Ito;Jin Lin;Yonghua Song
中科院分区：
文献类型：
--
作者：
Ying-wei Chi;Ken Yokoo;H. Nakajima;Kohei Ito;Jin Lin;Yonghua Song
电流和温度分布的不均匀性对固体氧化物电解槽（SOEC）的耐久性是有害的。高的蒸汽利用率有利于系统效率，但也增强了不均匀性。本研究结合分段SOEC实验，多物理场模拟，神经网络，以优化的不均匀性和效率联合。建立了三维单元模型，并通过实验验证了模型的正确性。快速代理模型与仿真数据进行训练，并集成到一个多目标优化问题的数值求解。它的解决方案形成了一个帕累托前沿量化蒸汽利用率，不均匀性，电压，氢气产量和工作温度之间的冲突关系，从中选择最佳的解决方案，以实现权衡。在1.11 W cm-2的功率密度下，当蒸汽利用率从0.72增加到0.82时，下游电流与上游电流之比从63.1%下降到55.2%。帕累托前沿可以增强电池组制造商和系统运营商之间的合作，使后者能够优化系统效率和不均匀性之间的平衡的操作点。
Inhomogeneous current and temperature distributions are harmful to the durability of solid oxide electrolysis cells (SOECs). A high steam utilization is favorable for system efficiency, but also enhances the inhomogeneity. This study combines segmented SOEC experiments, multiphysics simulation, and neural network to optimize the inhomogeneity and efficiency jointly. A three-dimensional (3D) cell model is built and experimental validation shows that the model correctly predicts the decreased down-stream current after the steam utilization exceeds 0.8. Fast surrogate models are trained with the simulation data and integrated into a multi-objective optimization problem for numerical solution. Its solutions form a Pareto front quantifying the conflicting relationship between the steam utilization, inhomogeneity, voltage, hydrogen production and working temperature, from which optimal solutions are chosen to achieve a trade-off. Under a power density of 1.11 W cm−2, the ratio between the down-stream and up-stream currents drops from 63.1% to 55.2% when the steam utilization increases from 0.72 to 0.82. The Pareto fronts can enhance the collaboration between stack manufacturers and system operators by enabling the latter to optimize the operating point for a balance between system efficiency and inhomogeneity.