Open Energy Benchmark | Payload Website Template

Website: https://openenergybenchmark.org/

Energy planning and the quest to decarbonize our energy systems are being underpinned by energy system modelling, which in turn depends critically on optimization solvers. These are algorithms that can find least-cost solutions to linear programming (LP) and mixed-integer linear programming (MILP) problems, mathematical models that represent the energy system and the scenarios under study. In turn, advances in optimization solvers rely on public benchmarks of realistic problems, which solver developers use to tune their code and develop new algorithms.

Unfortunately, energy planning is an underrepresented application domain in existing benchmarks of LP and MILP problems, such as MIPLIB. This gap creates challenges for energy planners trying to evaluate and compare different solvers for their use cases, and means that solver developers are not necessarily improving their solver performance on typical energy problems. Moreover, existing benchmark platforms are not interactive, or very intuitive to use, making it hard for non-experts in optimization to interpret and use effectively.

This project develops an open, automated, and interactive benchmark platform, Open Energy Benchmark, that will evaluate open-source solvers on key metrics like speed, memory consumption, and solvability across various widely-used energy planning models. Open Energy Benchmark will accelerate the energy transition by helping energy system modellers select the best solvers for their particular application, fostering innovation in solvers by providing them with realistic benchmark problems, and enhancing decision-making by giving stakeholders trustworthy progress metrics.