Decision‐Support Modeling for Building Retrofit Planning Across Individual Buildings and Building Stocks

The decarbonization of the building sector is one of the central challenges to meet climate goals and mitigate climate change. For the retrofit of existing buildings, well‐informed retrofit decisions are crucial, and several modeling approaches exist to support these decisions. In this paper, we reviewed more than 200 approaches that support this decision‐making and the energy design of existing buildings. Our analysis covers both individual buildings and building stock perspectives, providing a comprehensive overview of the strengths and limitations of existing modeling approaches. Approaches can generally be categorized into scenario‐based, meta‐heuristic, and deterministic models. At the individual building level, mixed‐integer linear programming (MILP) and combinations of simulations and genetic algorithms (GA) are widely used. At the building stock level, scenario‐based approaches and metaheuristic approaches are the most common. Most of the analyzed approaches consider both the building envelope and the energy supply system (ESS), which has also been recommended in many works. While individual building works focus on applications of their methods in specific buildings, building stock approaches mostly use typical buildings. However, important challenges remain in modeling the decision‐making of modernizing building energy systems. Especially, improved representations of long‐term investment decisions, uncertainty in environmental developments, and limited generalization of results due to a strong focus on specific energy systems and building types. A common limitation of all methods is the limited integration of resource limitations and uncertainty assessments. Future research should therefore focus on integrating dynamic, long‐term planning approaches with uncertainty modeling. Hybrid approaches combining heuristics, machine learning (ML), and optimization can improve computational efficiency and scalability, especially for building stock modeling. In addition, incorporating resource allocation into models can improve their practical relevance. Finally, improving model validation through empirical case studies and benchmarking with real‐world data will be critical to ensure the reliability and practical implementation of all of the methods mentioned.