Buildings account for around 40% of EU energy consumption, but most existing buildings — especially multi-residential blocks, schools, hotels and smaller commercial assets — do not have dense submetering or a modern BMS and are unlikely to receive one at scale. Owners and operators therefore rely on static EPCs, annual bills, main meters and manual audits, which do not show how the building actually performs, where losses occur, or which intervention should be implemented first.
This creates a practical bottleneck for decarbonisation: energy savings are difficult to verify, renovation budgets are allocated with limited evidence, and portfolio owners cannot prioritise buildings or measures consistently. Current energy-management platforms mainly serve sensor-rich buildings, while IPMVP measurement and verification remains largely consultancy-driven and hard to scale across the existing European building stock.
The project will develop an inference engine that fuses public building data, operational consumption data and weather information to estimate actual building performance rather than relying on static asset ratings. It will generate weather-normalised IPMVP baselines, disaggregate consumption into key end uses such as heating, cooling, domestic hot water and base load, and identify likely sources of losses linked to the envelope, controls, schedules or ventilation.
A recommendation-first optimisation engine will translate each diagnosis into a prioritised action plan, starting with no- or low-investment operational measures and then proposing capital measures with expected impact and payback. For buildings with little or incomplete historical data, typology priors based on TABULA/EPISCOPE, GIS embeddings and ARVIO’s existing energy-efficiency modelling will support cold-start estimation.
The innovation lies in solving the inverse problem of deriving actionable and verifiable energy performance insights from sparse, heterogeneous data, and in closing the prediction-to-measurement loop. Verified savings from implemented measures will be fed back into the models, improving future baselines, recommendations and portfolio screening; the same evidence can also support carbon accounting and, secondarily, green finance and valuation use cases.
The consortium will develop the data ingestion and harmonisation layer for sparse building data, including bills, main meters, smart-meter feeds, weather data, EPC registers, GIS/cadastre data and other public datasets. It will implement typology-based priors and portfolio screening methods to support cold-start assessments where consumption history is incomplete.
Core R&D will focus on the inference and M&V engine: weather-normalised IPMVP baselines, consumption disaggregation from limited data, localisation of likely losses, causal attribution of savings to measures, and an explainable audit trail for verified savings. The team will also build the recommendation engine, multi-tenant portfolio manager, carbon accounting functions and interfaces for utility and financial exploitation channels.
Pilots will validate the platform across 3–4 building typologies and three climate zones, including multi-residential, commercial, public and hospitality use cases where applicable. The pilots will compare predicted and measured savings, test operational measures before capital retrofits, and feed verified IPMVP results back into the models to improve future recommendations.
.png)
