AI-Based Metric for Building and Grid Reliability Assessment via Input Propagation and Stress Testing

Traditional energy planning approaches often treat building performance and grid reliability as separate challenges, resulting in fragmented analyses and suboptimal planning outcomes. This study presents an Data-driven framework that links urban building data with electrical system design through a unified Building+Grid Reliability Metric (BGRM). The approach uses machine learning to infer missing building attributes, predict neighborhood peak power, and propagate retrofit and energy efficiency scenarios to assess grid impacts. By comparing predicted peaks against grid capacity limitations, the framework identifies when upgrades enhance performance or risk overload. This integrated methodology provides a scalable and practical pathway to guide tailored investments in both building efficiency and grid infrastructure.