The article proposes a new method to model storage behavior in linear optimization long term energy planning models. The proposal allows eliminating the State of Charge equation in exchange for a data-driven approach, constraining storage discharge based on a dome-shaped constraint associated with storage capacity and power.

The proposal’s performance is tested against other commonly used methods across different data sequences. Model validation against a complete resource sequence demonstrates more than 90\% shortage coverage. All cases surpass the standard methodology with similar computational times. Sequential clustered methods are also outperformed when the number of clusters is low. Compared with complete sequential descriptions, the proposed method achieved 10\% lower deficit coverage, but with decreased computational time. The method showed robustness in system design, maintaining efficient computational times and reducing problem complexity.

The method has limitations to model several storage technologies or the resource-demand interaction. Nevertheless, the present approach is effective for highly flexible energy supply scenarios with only one type of energy storage (e.g.: solar or wind farm optimization).