Speeding up energy system optimization models - lessons learned from heuristic approaches, parallel solvers and large scale models

Most state-of-the art optimizing energy system models are characterised by a high temporal and spatial resolution to include detailed information of local weather conditions. This became necessary through the integration of renewable energy sources such as photovoltaics and wind energy. Similarly the integration of cross-sectoral technologies for the decarbonisation of the energy, heating and transportation sector makes energy system models more complex and as a consequence the time required for solving the problem. To address this increasing computational demand the BEAM-ME project brought together experts from the fields of energy systems analysis, mathematics, operations research, and informatics to establish interdisciplinary solutions. The talk provides an overview of the final project results and more in-depth highlights from two stage heuristic approaches and the parallel interior point solver PIPS-IPM++. Depending on the problem at hand and available computation resources a speed-up factor of up to 26 was achieved. Taking up the results from the BEAM-ME project an outlook on the follow-up project UNSEEN shows how the significant reduction in time required for solving the problems can be used to generate a more holistic view on the near-optimal solution space. This allows providing decision makers with a wide range of alternatives showing the trade-offs between several decision criteria