New Paper Published in Energy and Buildings – Forecasting Global Building Internal Heat Gains to 2100

We are pleased to announce the first publication of CURA Lab“Forecasting global building internal heat gains to 2100: an applied XGBoost model driven by Shared Socioeconomic Pathways” in Energy & Buildings (DOI: 10.1016/j.enbuild.2025.116642).

What the Study Is About

Buildings generate heat internally—from occupants, lighting, appliances, and other everyday activities. These internal heat gains strongly influence heating and cooling needs, yet most global energy models assume they remain constant over time. Our study challenges this assumption.

We developed a high-accuracy machine learning framework using XGBoost, trained on historical socioeconomic data and future projections from the Shared Socioeconomic Pathways (SSPs)—the same global scenarios used by the IPCC. This model forecasts household sizes and energy use for lighting, appliances, cooking, and water heating for nearly every country through the year 2100.

Key Findings

  • Dynamic change matters: Internal heat gains will not remain static. Our projections show substantial shifts as societies evolve, economies grow, and lifestyles change.
  • Global disparities: Using fixed assumptions can significantly misrepresent energy demand—by up to 30% higher in developing countries and 27% lower in developed ones.
  • Energy balance redefined: These changes alter the balance between heating and cooling, with implications for energy efficiency, HVAC design, and emissions mitigation.
  • Open dataset: The full dataset is openly available to support improved building simulations and climate modeling worldwide.

Why It Matters

This research highlights the limitations of current static assumptions in building codes and energy models. As internal heat gains evolve with demographic and technological shifts, accounting for these dynamics becomes crucial for designing resilient, low-carbon, and future-proof buildings.

The open dataset enables policymakers, researchers, and designers to integrate realistic, time-varying heat gain projections into long-term planning—improving forecasts of building energy demand and informing global strategies for climate adaptation and mitigation.