A Piece-Wise Model-Based MILP Approach for Optimal Dispatch of Hybrid Energy Storage Systems with Practical Validation

This work presents a piece-wise model-based mixedinteger linear programming (MILP) approach for optimal dispatch of a hybrid energy storage system comprising a lithiumion battery and a vanadium redox flow battery. By employing uni-variate and bi-variate piecewise-linear models, the method approximates the system's non-linear behavior more closely than conventional linear models, while retaining computational efficiency. Additionally, lithium ion battery's calendric aging is integrated as a stress factor, influencing the cost function to minimize degradation. The proposed algorithm is validated under real-world operating conditions, comparing two variants: Algorithm A with linear piecewise models and Algorithm B with singular linear models. Results from intra-day tests showed that Algorithm A improved self-sufficiency by 3.9% without increasing operational costs. Also, the improved model fidelity enable more foresighted operation, resulting in better alignment with renewable generation and a strategic energy reserve. The findings demonstrate the practical value of integrating higherfidelity system approximations into MILP-based optimization. ... mehrThe approach offers a scalable and implementable tool for managing hybrid energy systems in renewable-powered buildings.