Demo Abstract: FlexHologram: Generating Flexible Building HVAC Operation Pareto Front in Seconds

The growing penetration of renewable energy poses challenges to a resilient power grid, especially during extreme weather events. While building HVAC systems are a primary source of demand-side flexibility, exploiting their thermal mass as "virtual storage" across massive portfolios requires challenging and computationally expensive counterfactual inferences. Furthermore, traditional optimizations often rigidly predefine shifting or shedding regimes, ignore dynamic occupant preferences, and they obscure the trade-offs among flexibility, energy efficiency, and human comfort. We present FlexHologram, an interactive flexibility assessment tool for large building portfolios. Powered by a GPU-accelerated Parallel Physics-Consistent Neural Network (PCNN) trained on real-world operational records, FlexHologram enables Load Serving Entities (LSEs) to evaluate long-horizon (e.g, 168-hour) multi-objective trajectories (Comfort vs. Peak Energy vs. Off-peak Energy). By generating dense 3D Pareto fronts in under 5 seconds per building, the tool allows users to visually explore how shifting/shedding strategies of space heating & cooling naturally emerge without heuristic rules. This tool paves the way to unlocking demand-side flexibility in a scalable, transparent, and human-centric manner.