Operations Research 2025
Abstract Submission

2263. Prototypical warm-starts for demand-robust LP-based energy system optimization

Invited abstract in session TC-9: Operational planning in energy systems, stream Energy and Sustainability.

Thursday, 11:45-13:15
Room: H15

Authors (first author is the speaker)

1. Niels Lindner
Optimization, Zuse Institute Berlin
2. Lukas Mehl
Zuse Institut Berlin
3. Karolina Bartoszuk
Zuse Institut Berlin
4. Janina Zittel
Applied Algorithmic Intelligence Methods Department, Zuse Institute Berlin

Abstract

The expressiveness of energy system optimization models (ESOMs) depends on a multitude of exogenous parameters. For example, sound estimates of the future energy demand are essential to enable qualified decisions on long-term investments. As these are inherently uncertain, ESOMs are formulated in a robust or stochastic setting. The drawback is however that the resulting large-scale problems are computationally challenging, even if they are merely based on linear programming (LP). A thorough analysis of demand sensitivity and the resulting consequences on technology investments is therefore hardly accessible.

One way to cope with demand uncertainties is to solve a number of independent demand scenarios. For LP models, this offers the potential for warm-starting the dual simplex method, as demand variations translate into changes of right-hand sides of linear constraints. However, depending on time horizon and resolution, these modifications might be too numerous, so that solving from scratch is more effective.

We therefore propose a decomposition approach to facilitate and accelerate warm-starting procedures for LP-based ESOMs with uncertain demand. The key idea is to single out prototypical and easily solvable demand slices, limited to very few time steps. We then combine the resulting optimal simplex bases to a larger one, matching energy demands with prototypes, and thereby producing more meaningful warm-starts. This principle naturally extends to ESOMs that integrate multiple scenarios into a single one, as in stochastic programming.

We evaluate the feasibility of our approach on a real-world case study, using a sector-coupled ESOM with hourly resolution for the Berlin-Brandenburg area in Germany, based on the oemof framework.

Keywords

Status: accepted

Back to the list of papers