Explicit representation of n largest district heating systems

[cpschau]

District Heating Subnodes Feature

Summary

This PR introduces district heating (DH) subnode modeling to PyPSA-Eur, enabling finer spatial resolution for urban central heating systems. Key capabilities include:

• Selectable n largest DH systems: Configurable number of subnodes selected by demand ranking from ISI data
• Country filtering: Subset of modeled countries can be selected for subnode modeling via sector.district_heating.subnodes.countries
• Parent node linkage: Subnodes are linked to their parent cluster buses for:
  • Electricity (HV for CHP generation and distribution grid for P2H supply of DH)
  • Resource supply (Gas/coal/oil/H2/waste/biomass)
  • CO₂ storage & global atmosphere buses
• Subnode-specific data: Each subnode receives:
  • Own ambient temperature profiles (for COP calculations)
  • Heat pump COPs based on local conditions
  • Renewable heat potentials (solar thermal, geothermal, river_water etc.)
  • Network forward temperature timeseries
  • Heat demand profiles
• Demand compensation: Additional subnodal demands are compensated by reduction of demand in respective parent node.

Data Source

The feature is based on ISI/seenergies district heating area shapes derived from heating density data. These shapes provide:

• Geographic boundaries of DH systems
• Annual heat demand estimates (GWh/a)
• City/system labels

Structural Integration

New Scripts

Script	Purpose
scripts/identify_district_heating_subnodes.py	Identifies n largest DH systems, maps to parent clusters, extends onshore regions
scripts/prepare_district_heating_subnodes.py	Extends energy/heat totals, district heat shares, industrial demand, and time-series data for subnodes
scripts/map_dh_systems_to_cities.py	Maps DH system IDs to human-readable city names using GeoNames cities500 (see below)

Modified Scripts

Script	Changes
prepare_sector_network.py	Extended define_spatial() for district heating nodes; heat buses, loads, and technologies now iterate over heat_nodes (including subnodes); parent node mapping for resource bus lookups
rules/build_sector.smk	New rules identify_district_heating_subnodes and prepare_district_heating_subnodes; conditional inputs for subnode-extended data files
Various data pre-processing scripts	Extended to output subnode-specific COP, solar thermal, temperature, and PTES profiles

Configuration

sector:
  district_heating:
    subnodes:
      enable: false  # Toggle subnode modeling
      n_subnodes: 40  # Number of largest DH systems
      countries: []  # Empty = all countries; or specify subset
      demand_column: "Dem_GWh"  # ISI data column for demand
      label_column: "Label"  # ISI data column for city names

Testing

Test Configuration

Three scenarios are compared: master_branch (code before this PR), subnodes_off (PR code with subnodes disabled), and subnodes_on (PR code with subnodes enabled).

config/config_subnodes.yaml:

run:
  prefix: "test_subnodes"
  name:
    - subnodes_on
    - subnodes_off
  scenarios:
    enable: true
    file: config/scenarios_subnodes.yaml
scenario:
  clusters:
    - 4
clustering:
  temporal:
    resolution_sector: 25h
countries: [DK, DE, FR]
sector:
  district_heating:
    subnodes:
      enable: true
      n_subnodes: 10
      countries: [DK, DE]

config/scenarios_subnodes.yaml:

subnodes_on: {}

subnodes_off:
  sector:
    district_heating:
      subnodes:
        enable: false

master_branch: {}

Network Topology

Heat Demand Preservation

District Heating Energy Balance

Total System Costs

Scale-up

I also tested the feature with 39 clusters and 200 subnodes at 3h resolution including river water and geothermal as heat sources for district heating supply to ensure robustness. The network was solved in 6 h with max mem usage of 172 GB.

Here are some related results:

District heating balance

Composition of system costs:

District heating balance map

Notably, the potentials for limited heat sources (river water in this case) in the parent nodes, are overestimated due to spatial aggregation.

References

• PyPSA-DE Implementation: PyPSA/pypsa-de#58
• Preprint Paper: arXiv:2601.07647 - Uses this feature for PTES (Pit Thermal Energy Storage) analysis in Germany

Remaining TODOs

• Handle missing countries in ISI data: Add graceful handling/warnings when ISI shapes are unavailable for selected countries
• Map IDs to city names: Implement georeferenced mapping from ISI system IDs to human-readable city names
• Config parameter mapping: Map config parameters (planning horizon, market share, energy demand projections) to ISI data assumptions (market share assumptions, marginal DH expansion costs)
• Rebase with new DH features: Integrate with recent master branch additions:
  • Heat pump boosting
  • Lake water heat sources
  • Improved excess heat handling from industrial processes

[lkstrp]

All doctables are gone and this needs to be merged in the pydantic schema