Background
Driven by decarbonisation and high cost of infrastructure expansion, the energy sector increasingly looks at harnessing flexibility in consumers’ activities. It is essential to accurately and realistically characterise and simulate demand at the level of heterogeneously-responsive agents: individual households, firms, buildings, etc., under a variety of policy and socio-environmental contexts. Mobility and Energy Demand Suite (MEDUSA) comprises a suite of models of synthetic population and preferences, agent activities, mobility and energy demand prediction tools that follow the agent- and activity-based paradigm. By linking agents and their attributes alongside activities they perform, the suite can produce energy demand profiles as a function of demographic, energy and transport policy scenarios.
Our Contribution
The goal of this project was to create a central controller/hub that could be used to run multiple models in various configurations. This system would allow models to be plugged-in to it and configured to be run so that they are coupled together. The configuration system was developed using Pydantic and accepts TOML files as config files. In these files, users can specify the models they want to run, the order they should run in and the frequency with which they are updated. They can also specify top-level parameters that are shared across all models.
Outcomes
The research team is now implementing a production version of the software and writing more models to plug in to the system.