Enabling efficient sizing of hybrid power plants: a surrogate-based approach to energy management system modeling

Abstract

Optimal sizing of hybrid power plants (HPPs), which include wind power plants and battery energy systems, is essential to prevent financial losses from under- or over-sizing relative to grid connection capacities. Accurate sizing requires high-fidelity energy management systems (EMSs) to model bidding strategies and operations in electricity markets, resulting in precise operational revenues and costs. However, due to the computational burden of such models, sizing methodologies often resort to low-fidelity EMS models, leading to faulty sizing evaluations. To address the need for a computationally efficient and accurate model that enables quantitative assessment of HPPs, we evaluate the potential of surrogate models to replace a high-fidelity EMS participating in the day-ahead electricity market in Denmark with perfect forecasts. Given the limited literature on surrogates of EMS models for utility-scale, grid-connected HPPs with batteries, we develop and compare four different surrogate models to approximate a state-of-the-art EMS model. The best-performing surrogate employs singular value decomposition for dimensionality reduction and a feed-forward neural network for regression. This surrogate achieves a normalized root mean square error of 0.81 % in approximating yearly revenues. This method proves effective in accurately evaluating the operation of HPPs across various geographical locations and hence in multiple sizing problems. Furthermore, we utilize the surrogate to evaluate the profitability of several HPP sizes, achieving a root mean square error of 0.010 on the profitability index, with values ranging between -0.13 and 0.18. This demonstrates that the developed surrogate model is suitable for HPP sizing under the given cost and financial assumptions.