Development, validation, and testing of advanced mathematical models for the optimization of BESS operation

Abstract

Current decarbonization strategies are driven by the fast-paced diffusion of non-programmable renewable energy sources (NP-RESs), mainly through solar and wind power generation. Energy storage technologies are emerging as key solutions for coping with the variability and low-inertia characteristics of NP-RESs. Particularly, battery energy storage systems (BESS) are diffusing more widely for both behind-the-meter (BTM) and utility-scale applications. In this context, we still lack a shared solution on how to proceed from the on-field data collected about the performance of BESS to reliable and fast mathematical formulations for operational optimization. This study provides a validated modeling framework that can be exploited during or after BESS commissioning to (i) identify and derive the useful parameters to characterize BESS performances, (ii) formalize them in a mathematical formulation while being aware of its specific trade-off between accuracy and computational effort, and (iii) exploit the selected BESS model within a multi-energy system optimization problem. We discuss three different modeling approaches that we developed for optimizing BESS operation, with each providing a different balance between modeling accuracy and computational effort. These three mathematical models were validated against a numerical simulation model based on on-field performance data, and they were eventually tested on a reference case study. The results indicate that it is possible to restrict the average error in estimating BESS efficiency while simultaneously limiting the computational effort of the model. Regarding the operation of BESS, the conducted simulations demonstrate that an approximate BESS model may result in an overestimation of the expected revenues.