Skip to content

irm-codebase/RESTORE_energy_modelling_framework

Repository files navigation

The RESTORE model

RESTORE: RetrospEctive SecTor cOupled eneRgy toolsEt (tentative name)

A bit of history

RESTORE is based on D-EXPANSE, a stylized national-level nodal power system model used in hindcasting studies:

Important: D-EXPANSE is not the same model as EXPANSE, which is a spatially explicit electricity model with no inter year slicing! You can learn more about EXPANSE in these studies:

New features in RESTORE

RESTORE builds on D-EXPANSE by implementing:

  • Graph-based flows
  • Spatial disaggregation
  • Sector coupling functionality
  • Reworked architecture to improve readability and modularity
  • Generic, pre-made constraints and expressions that can be easily re-used in sector modules defined by developers

RESTORE also features a fully standardized prototyping workflow based on FAIR principles. Model components (called "entities") are defined in single files, where the user can specify parameter names, values, units and sources. These files are rapidly converted into a single configuration file that the model uses as input. Conversion of currencies, energy units and power units is also integrated into this process.

This lets model developers track the sources of their data, and gives users and other researchers full transparency into the model's operation and assumptions.

Features currently in development

  • Implement an option for imperfect foresight.
    • Variable foresight length.
    • Variable length of years saved in each run.
  • Cycle flow constraints in the energy transmission module.
  • Seasonal storage capacity expansion.
  • Improve representative day algorithm.
    • Ensure weather synchronicity (PV, Wind and Hydro run-off).
    • Create cnf file standard for hourly data series in representative days. Must be searchable by entity_id.
    • Add options for different types of clustering algorithm (k-means, spectral, etc).

IMPORTANT

Although hindcasting/retrospective studies are useful to test modeller assumptions, they are subject to a plethora of uncertainties that are difficult to avoid. Essentially, their usefulness is limited by the availability and fineness of historical energy system data, which worsens the further to the past you go and the more specific your data requirements are. Temporal and spatial resolution matter a lot when it comes to calculating prices, system resilience and the viability of renewable technologies.

Due to this, I would argue that RESTORE is not a validation tool, but rather a useful test-bench to evaluate features before they are added to more complex models.

For more on the topic of model evaluation and past uncertainty, see the following:

For examples of hindcasting studies, see:

About

RetrospEctive SecTor cOupled eneRgy toolsEt

Resources

License

Stars

Watchers

Forks

Packages

No packages published