Fraunhofer Institute for Energy Economics and Energy System Technology
Agent-based investment decision model for building owners
AgentHomeID is a model developed at Fraunhofer IEE to describe the energy-relevant investment decisions of decentralized actors, in particular private households. The model is integrated into a detailed picture of the German building stock and aims to predict investments in energy-related measures in residential buildings up to 2050, taking socio-economic framework conditions into account.
The decisions of households regarding energy-efficient refurbishment of the building envelope, heating replacement, installation of photovoltaic roof systems (possibly in combination with batteries, heat pumps and electric vehicles) are modeled on the basis of empirically determined preferences and willingness to pay. This enables a projection of the future development of the building stock, heating requirements and the installed capacities of renewable energy sources and storage systems.
The combination of AgentHomeID with synthetic reconstruction (SR) and combinatorial optimization (CO) enables the generation of a realistic local building stock and an associated population of owners and tenants. Technical data of the buildings as well as socio-economic and demographic profiles of the residents are taken into account.
The model uses a variational autoencoder framework with regional boundary conditions to represent the local building stock including residents and owners. The joint probability distribution of the building stock as well as the socio-economic and demographic profiles is learned by unsupervised learning in the CVAE process from the survey data of the Microcensus Scientific Use Files 2014 (SUF).
The information generated serves various use cases, including the consideration of neighboring buildings and owners in decision-making processes, the planning of network infrastructure taking into account building development, the development of heating and heating network infrastructures and the support of municipal heating planning.
AgentHomeID thus enables realistic prediction and analysis of investment decisions in the building sector and contributes to the development of sustainable energy infrastructures.