Skip to content

Harmonising capacities

A PyPSA model typically starts from its own existing power-plant database (e.g. powerplantmatching for PyPSA-Eur). The IAM, independently, reports capacity targets per (region, technology, year) via iampypsa.transforms.capacities.build_capacity_targets. These need to be harmonised.

The general approach

Rather than discarding the model's existing plant-level detail (e.g. location and decommissioning year) and running the PyPSA model greenfield, the recommended approach is as follows:

  1. Filter the power-plant database for the model year, dropping power plants that aren't built yet or are already decommissioned by then.
  2. Group the filtered power-plant database by (region, technology) — using the technology mapping.
  3. Compare the IAM's target against that filtered power-plant database's aggregate capacity, per (region, technology) group. Two cases need to be distinguished:
    • IAM target ≥ PyPSA capacity: the power-plant database is left exactly as filtered; any additional capacity needed to reach the target is left to the solver, which is free to decide where to build it — this needs to be enforced through an additional global constraint across all nodes within the region, see Enforcing the target during optimisation.
    • IAM target < PyPSA capacity — a scaling factor (target / existing, < 1) is applied to every matched power plant in the group, proportionally shrinking their capacity so the group's aggregate now equals the target. This is effectively early retirement: individual plants keep their location and other attributes, just at reduced capacity.

This keeps the model's spatial/technical detail while still being consistent with the IAM's capacity pathway.

Enforcing the target during optimisation

If IAM target ≥ PyPSA capacity, an additional global constraint on total capacity per (region, technology) group is required during optimisation to ensure that all capacity set by the IAM is actually built in PyPSA. Two constraints are possible:

  • A mandatory lower bound, which ensures that across all modelled PyPSA nodes within the IAM region the required capacity is actually installed (exceeding the capacity already set through the filtered power-plant database).
  • An optional upper bound, which additionally prevents PyPSA from building more capacity than the IAM specifies — effectively an adequacy check on the IAM's own capacity assumptions.

Worked example: PyPSA-Eur

pypsa-eur-iam/scripts/remind/adjust_powerplants_REMIND.py implements steps 1-3 above. pypsa-eur-iam/scripts/remind/installed_capacity_constraints_REMIND.py implements the global constraints.

Next