Multi-objective optimizations of solid oxide co-electrolysis with intermittent renewable power supply via multi-physics simulation and deep learning strategy

Solid oxide electrolysis cell (SOEC) is a novel approach to utilize excess renewable power to produce fuels and chemicals. However, the intermittence and fluctuation of renewable energy requires more advanced optimization strategy to make sure its performance in safety and cost-effectiveness. Here, we propose a hybrid model for the precise and quick optimization of the co-electrolysis process in the SOEC for syngas production, based on the multi-physics simulation (MPS) and deep learning algorithm. The hybrid model fully considers electrochemical/chemical reactions, mass/momentum transport and heat transfer, and presents a small relative error (<1%) in most the cases (>96%). Various targets including the single-objective, dual-objective and multi-objective optimizations are evaluated with particular attentions on the reactant conversion rate and energy efficiency at different temperatures. The electrolysis efficiency is negatively correlated with the power supply in all strategies and thermal neutral condition (TNC) can be achieved at different temperatures, where 1023 K, 1053 K, 1083 K and 1113 K are corresponded to the TNC power range of 10-16 W, 14-23 W, 18-29 W and 22-37 W, respectively. This theory can be flexibly applied in the sustainable manufacturing and circular economy sectors and energy according to the optimization targets.

Automated summary

Solid oxide electrolysis cell (SOEC) is a novel approach to utilize excess renewable power to produce fuels and chemicals. This study proposes a hybrid model based on multi-physics simulation and deep learning algorithm for the optimization of the co-electrolysis process in the SOEC.