Research project PEREGRINE
Parallelisation in energy system analysis: Added value for large-scale optimisation models by innovative solver development
Modelling large and complex energy systems is essential for the analysis and planning of future energy supply. This is very often done with the help of energy system optimisation models that are solved as linear programmes. However, the increasing complexity of these models is pushing existing software for solving linear programmes (solvers) to their limits. Supercomputers and special solver software (PIPS-IPM++) for so-called distributed memory computer architectures can help here. However, as this requires special expertise, the PEREGRINE project, funded by the German Federal Ministry for Economic Affairs and Climate Action, aims to improve the usability of these advanced methods for solving very large energy system optimisation models.
Research project PEREGRINE | |
|---|
Duration | April 2024 to March 2027 |
Funded by | Federal Ministry for Economic Affairs and Climate Action |
Project participants | - Institute of Networked Energy Systems
- GAMS Software GmbH
|
The project focuses on model annotation, i.e. intelligent and automated pre-processing. This is currently still a time-consuming step for using the open source solver PIPS-IPM++. This solver is also to be optimised in terms of robustness and performance. Other project priorities are aimed at adapting the solver for cloud-based applications on user-owned clusters and supercomputers. Data for modelling complex energy systems will also be created and made freely available as a model library for testing different solvers. In addition, workshops and other exchange formats for modelling experts from the energy industry and science will be offered during the course of the project.
The contribution of the Institute of Networked Energy Systems to the PEREGRINE project is focussed on two areas: Firstly, the project team is developing semi- and fully automated methods for generating structural information for the solver PIPS-lPM++ (model annotation). Background: Until now, modellers have defined the annotation during model formulation and specified application-related parameters such as days or weeks. Fully automatic, model-agnostic annotation, on the other hand, investigates approaches to determine a suitable structure on the restriction matrix of the optimisation problem after model translation, which makes parallel solving more efficient.
On the other hand, the knowledge base created by solving energy system optimisation models on supercomputers is being expanded and standardised. Scientists should be able to use PIPS-IPM++ directly, even if they use a modelling system other than GAMS ("General Algebraic Modelling System"; mathematical modelling program from GAMS Software GmbH, which is involved in the project). The project team also invites modelling experts to contribute their anonymised instances of high-resolution, complex energy system optimisation models as part of the model library. This serves to validate heuristics for reducing model complexity and to increase the efficiency of the solution methods.
