1994. An upper bound on suboptimality for finite-horizon approximations of energy storage scheduling problems
Invited abstract in session TC-44: Approaches for handling computational complexity, stream Energy Economics & Management.
Tuesday, 12:30-14:00Room: Newlyn 1.01
Authors (first author is the speaker)
| 1. | Eléa Prat
|
| DTU | |
| 2. | Richard Lusby
|
| DTU Management, Technical University of Denmark |
Abstract
Energy storage systems play a crucial role in the energy transition, making it essential to study their optimal operation. Energy storage scheduling problems are inherently infinite-horizon optimization problems and are therefore challenging to solve directly. In practice, rolling-horizon approaches are widely used, by repeatedly solving finite horizon problems, each time fixing the solution for the first periods and sliding the horizon forward. This raises a fundamental question: how does the horizon length affect solution quality?
We are the first to propose an upper bound on the suboptimality introduced by finite-horizon approximations of energy storage scheduling problems. This allows decision-makers to assess the trade-off between computational cost, data forecasting needs, and solution quality. We also derive a necessary and sufficient condition to detect when the solution of the finite-horizon approximation matches that of the infinite horizon.
Through various case studies, we illustrate how this upper bound decreases as the length of the horizon is incrementally increased until it ensures the optimal solution of the infinite-horizon problem is obtained. We demonstrate the limitations of approaches that increase the horizon length until the optimal solution does not change.
Beyond energy storage, our insights are also directly relevant to capacitated inventory management and warehouse scheduling problems, where similar finite-horizon approximations are common.
Keywords
- OR in Energy
Status: accepted
Back to the list of papers