Dynamic Techno-Economic Sizing and Dispatch Optimization of Co-Located Solar-Wind-Green Hydrogen Plants

Abstract

Co-located photovoltaic (PV) and wind plants producing off-grid green hydrogen face operational inefficiencies driven by resource intermittency. Rapid power fluctuations and prolonged low-load idling accelerate degradation in polymer electrolyte membrane (PEM) and alkaline electrolyzer stacks, reducing membrane lifetime and inflating the levelized cost of hydrogen (LCOH). LCOH outcomes are themselves sensitive to the underlying electricity- and fuel-price trajectories, and general-purpose price-forecasting architectures, such as the sentiment-guided transformer of Gurrib et al. developed for daily oil prices, illustrate one route toward incorporating such price uncertainty, albeit in a commodity-market rather than a plant-design context. Current capacity expansion models typically assume static electrolyzer efficiency and treat stack replacement as an exogenous maintenance cost rather than an endogenous consequence of the dispatch decisions being optimized. This paper presents a two-stage stochastic co-optimization framework that jointly determines capital sizing and sub-hourly dispatch while embedding empirically calibrated degradation stress penalties, for cold starts, low-load idling, and ramping, directly in the objective function. Applied to a 100-MW hybrid hydrogen facility sited in West Texas, the degradation-aware framework reduces LCOH by 14.7% to $3.82/kg and extends PEM electrolyzer stack life from 4.1 to 7.5 years relative to an unconstrained maximum-capture dispatch strategy.