Umeå University Research Shows AI Can Deliver More Effective and Cost Efficient Building Energy Renovations

Image: Magnus Mikaelsson

(IN BRIEF) Research from Umeå University demonstrates how AI-driven analysis can transform energy renovation strategies by providing highly tailored recommendations based on the unique characteristics of individual buildings. By combining machine learning, explainable AI, and data fusion, the study identifies the most effective interventions for reducing energy consumption and emissions, highlighting that optimal solutions vary widely depending on building type, location, and climate. The research also incorporates real occupant behavior, showing that traditional assumptions can significantly distort energy estimates and that behavioral changes alone can meaningfully reduce electricity demand. Through case studies across multiple Swedish cities, the findings emphasize the importance of moving beyond generic renovation models toward more precise, data-driven approaches. The development of an interactive 3D platform further enhances the practical value of the research, enabling homeowners and stakeholders to explore tailored scenarios and make informed decisions. Overall, the work represents an important step toward improving building energy efficiency and supporting climate targets through smarter, more localized strategies.

(PRESS RELEASE) UMEÁ, 23-Apr-2026 — /EuropaWire/ — Umeå University has presented new research demonstrating how artificial intelligence can significantly improve the effectiveness of energy renovation decisions in buildings, reducing both emissions and costs through more precise, locally tailored strategies. The findings highlight the limitations of traditional models, which often rely on generalized assumptions and fail to account for the unique characteristics of individual buildings.

The research, conducted by Santhan Reddy Penaka at the university’s Intelligent Human-Building Interaction lab, introduces advanced data-driven methods that combine machine learning, explainable AI, and data fusion techniques. These approaches integrate multiple data sources to provide a more detailed understanding of how different building components—such as walls, windows, roofs, and floors—affect overall energy consumption.

The study reveals that the most effective renovation measures can vary significantly depending on factors such as building type, climate conditions, and geographic location. Analysis of 81 building clusters across Swedish cities including Linköping, Lund, and Umeå showed that interventions like additional wall insulation may be highly beneficial in some cases, while offering limited impact in others. This variability underscores the need for more targeted approaches rather than uniform renovation strategies.

A key innovation of the research is the incorporation of real occupant behavior into energy models. Instead of relying on average assumptions, the system considers how residents actually use their homes, including patterns such as window usage and appliance operation. The findings indicate that simplified behavioral assumptions can lead to inaccuracies of up to 15% in energy consumption estimates. Furthermore, when applied to Sweden’s planned power-based electricity tariff system for 2027, the model suggests that behavioral adjustments alone could reduce peak electricity demand by between 6% and 17%, depending on building characteristics.

To bridge the gap between research and practical application, Penaka has also developed an interactive 3D visualization platform. This tool allows homeowners and stakeholders to compare their building’s energy performance with similar properties in the local area and simulate the impact of different renovation measures and behavioral changes. The platform is designed to support more informed, evidence-based decision-making for both individuals and policymakers.

The research contributes to broader efforts to improve building energy efficiency, a critical factor given that buildings account for around 30% of global energy use and more than a quarter of carbon dioxide emissions. By enabling more precise and context-specific recommendations, the study supports more efficient allocation of resources and more effective progress toward climate goals.

The doctoral dissertation, titled Heterogeneity-Aware Building Stock Modelling for Urban Energy Transitions, will be presented on April 29 at Umeå University under the supervision of Weizhuo Lu, with Joakim Widén serving as opponent.

About the doctoral defence

Dissertation title: Heterogeneity-Aware Building Stock Modelling for Urban Energy Transitions

Date and time: Wednesday, 29 April, 9:00 a.m. in NAT.D.300 (lecture hall), Naturvetarhuset, Umeå University

Supervisor: Weizhuo Lu, Professor, Department of Applied Physics and Electronics, Umeå University

Opponent: Joakim Widén, Professor, Department of Civil and Industrial Engineering, Uppsala University

Download the doctoral dissertation here.

Media Contact:

email: press@umu.se
phone: +46 70 610 08 05

SOURCE: Umeå University