Urbin4hd

urbin4hd generate-models --geometry buildings.geojson --templates prototype_models.json --output models/

| Problem | Likely Cause | Solution | |---------|--------------|----------| | Simulation crashes during shading calculation | Too many buildings (n>2000) | Reduce to representative clusters (e.g., archetype-based) | | Unrealistically low cooling loads | Overestimated shading | Check sun angles and surrounding building heights | | Out of memory | Storing full hourly results for all zones | Use --output-format parquet and filter to thermal zones | | Microclimate temperature jumps >5°C | Missing anthropogenic heat | Add --waste-heat-factor 0.15 (15% of elec load) |


High-density modeling requires more detailed inputs than low-rise areas. URBiN4HD

High-density models often show bias due to microclimate and shading errors.

Recommended calibration steps:

  • Use Bayesian calibration or Morris method for sensitivity analysis (Python library SALib).
  • Typical accuracy target after calibration:
    NMBE ≤ ±5%, CV(RMSE) ≤ 30% (hourly district aggregate).


    URBiN4HD produces:

  • District-level summaries:

  • Visualization-ready files:

  • Example output snippet (district): | Time (hour) | Cooling (MW) | Heating (MW) | Total Elec (MW) | |-------------|--------------|--------------|-----------------| | 0 | 12.3 | 45.6 | 88.2 | | 1 | 11.9 | 46.1 | 87.6 |