6 - Predicting Lausanne thermal network’s future – understand Lausanne’s buildings decarbonization via district heating

As Lausanne advances its transition toward a low-carbon energy system, district heating networks play a central role in reducing fossil fuel dependence. This project aims to develop a predictive model to estimate future district heating network demand under uncertainties related to climate change and building retrofits.

A reliable proxy for thermal demand is the total production of heating plants, as distribution and production losses can be quantified with reasonable accuracy. Using this approach allows for stable, system-level demand estimation without relying solely on highly variable building-level consumption data.

The shall model integrate two key uncertainties. First, climate variability: while average temperatures are expected to rise, heating demand remains sensitive to seasonal fluctuations and extreme events. Second, building retrofits: Lausanne’s diverse building stock will progressively improve in energy efficiency through insulation upgrades and modern standards such as MINERGIE, significantly affecting heat demand over time.

This predictive capability has direct operational and strategic value. In the short term, it enables optimized dispatch of production assets, prioritizing renewable but less flexible sources such as waste heat, biomass, and heat pumps. In the long term, it supports infrastructure planning, investment decisions, and network expansion while avoiding over- or under-capacity risks.

Ultimately, the model shall serve as a decision-support tool for operators and policymakers, helping align urban energy planning with decarbonization goals. By improving demand forecasting, Lausanne can enhance energy efficiency, reduce emissions, and ensure a reliable and sustainable heat supply.

This prediction will help the district heating network operator to plan the production beforehand as well as to make a long-term planning of its assets. A good prediction can also help diminishing fossil fuels usage by carefully planning the activation of the renewable but less flexible plants.

• Hourly production (regressor)
• Hourly external temperature (predictions to be found)
• Annual degree-days (previous and planned) ; dhn installed power (previous and planned)
• District heating network price proxy (previous and planned)
• Annual building retrofit percentage (previous and planned)

• Machine learning
• Regression
• Data analysis
• General knowledge of the energy sector