A multi-objective optimization framework for functional layer design in solid oxide electrolysis cells

초록

Solid Oxide Electrolysis Cells (SOECs) present a promising solution for high-efficiency hydrogen production; however, their large-scale adoption requires optimized performance and cost-effectiveness. This study develops a multi-objective optimization (MOO) framework targeting the design of functional layers (FLs), specifically their thickness and porosity. A detailed three-dimensional SOEC model in ANSYS Fluent was coupled with a Multi-Layer Perceptron (MLP) surrogate trained on Latin Hypercube Sampling data. The MLP model, integrated with Particle Swarm Optimization, achieved a minimum cell voltage (Vcell) of 1.288 V at 1.0 A/cm2 in single-objective optimization, closely matching CFD simulation results (1.2883 V). For MOO, the MLP was paired with the Non-dominated Sorting Genetic Algorithm II to minimize both Vcell and material cost of FLs (Cstack FL ) for a typical SOEC stack. The resulting Pareto front revealed trade-offs between electrochemical efficiency and economic feasibility, with optimal designs ranging from high-performance (Vcell = 1.289 V, CstackFL =$1.122/kW)tocost-effective(Vcell=1.349V,CstackFL=$0.126 /kW). An overpotential breakdown quantified losses from vapor and oxygen activation, ionic ohmic, and electronic ohmic components. Additionally, spatial analyses of current density distributions provided further insights into transport and reaction behavior. The proposed framework offers actionable guidance for balancing microstructural design, performance, and cost of SOECs.

키워드