Development of degradation cost models for electrolyzers and fuel cells considering megawatt-scale applications

To accelerate the large-scale integration of renewables for transition to a low-carbon energy system, the deployment of megawatt-scale electrolyzers and fuel cells systems is important. For their widespread integration and effective operations, the appropriate consideration of electrolyzer and fuel cell degradation is necessary. To this effect, this study presents two novel comprehensive degradation cost models, which are developed based on their dynamic degradation rates and life-cycle operating hours. The proposed degradation cost models are state transition-based method and ramp rate-based method , designed to quantify degradation as a function of input/output current density and ramp rates, respectively. To ensure their practical application in megawatt-scale power system operations, wherein their dispatch is typically based on electric power rather than electrochemical current density, these models were reformulated in terms of the power consumption/production. To estimate the parameters of the proposed degradation cost models, three regression approaches — linear, quadratic and third-degree polynomial — were carried out. Comparative analysis demonstrates that higher-degree polynomial models offer improved accuracy by capturing the nonlinear degradation behavior during dynamic operations, with state transition-based models consistently showing superior performance. Furthermore, the prediction performance of the estimation models are compared with the actual costs, followed by a sensitivity analysis of the estimated cost parameters. • State transition and ramp rate degradation cost models are developed to represent electrolyzers and fuel cells degradation. • Regression approaches were used to quantify degradation performance of grid-scale electrolyzers and fuel cells. • The STDC model demonstrated better performance compared to the RRDC model. • Higher-degree models predict degradation costs accurately, while linear models offer simplicity and ease of integration. • Degradation cost coefficients demonstrated proportional variation with the change in input degradation rates.