Overview
Energy storage scheduling problems typically involve operating a storage system to maximize its profitability, driven by price signals. These problems are inherently framed as infinite-horizon optimization challenges, given the continuous operational nature of storage systems without a predefined end-point. Optimal solutions can be achieved through a rolling-horizon approach, provided that the planning horizon employed for problem resolution is sufficiently long. This adequate horizon is termed a forecast horizon.
Research Context
The selection of a planning horizon for energy storage scheduling applications is frequently arbitrary, despite its critical importance for accurate optimization. Existing methodologies have not provided practical means to evaluate or select these planning horizons systematically. The continuous operation of storage systems necessitates an optimization framework that accounts for an indefinite operational lifespan, leading to the formulation of infinite-horizon problems.
Approach
The research introduces a specific, easy-to-check condition designed to verify whether a given planning horizon qualifies as a forecast horizon. This condition also allows for the derivation of a suboptimality bound if the chosen horizon does not meet the criteria of a forecast horizon. Building upon these theoretical foundations, a lower bound for the minimum forecast horizon is derived. Subsequently, an algorithm is developed to precisely determine this minimum forecast horizon.
Findings
- An easy-to-check condition was developed that confirms if a planning horizon serves as a forecast horizon.
- This condition can be utilized to derive a bound on suboptimality when the planning horizon is not a forecast horizon.
- A practical means for evaluating and selecting planning horizons for energy storage scheduling problems was provided.
- A lower bound on the minimum forecast horizon was derived.
- An algorithm was developed to determine the minimum forecast horizon.
- This algorithm enables the identification, a posteriori, of the shortest planning horizon that guarantees optimal rolling-horizon decisions.
- The approach avoids unnecessary forecasting effort and computational cost.
- Numerical experiments illustrate the practical use of the proposed framework.
- Numerical experiments investigate how storage system characteristics influence the minimum forecast horizon.
- Numerical experiments investigate how electricity price patterns influence the minimum forecast horizon.
Why This Matters
The proposed framework enables practitioners to assess the sufficiency of a chosen planning horizon for energy storage scheduling. By providing a practical method for evaluating and selecting planning horizons, it addresses a gap in current practices where such horizons are often arbitrarily chosen. The ability to identify the shortest planning horizon that ensures optimal rolling-horizon decisions reduces forecasting effort and computational costs, offering an efficiency improvement in operations.
Potential Applications
The framework introduced can be applied by practitioners to evaluate whether the planning horizon they have chosen for energy storage scheduling problems is sufficient. It allows for the identification, retrospectively, of the shortest planning horizon that ensures optimal decisions within a rolling-horizon context, thereby optimizing forecasting and computational resource allocation for energy storage operation.