Pilot Implementation of Cost-Optimized Control for Collective Residential Heat Pumps with Thermal Buffers
High-rise residential buildings represent a key target for decarbonization, yet electrification of heating via heat pumps can exacerbate local grid congestion. Thermal buffers act as flexible energy storage, allowing heat pumps to shift electricity consumption to cost-effective or low-demand periods while maintaining indoor comfort. This study investigates two complementary approaches for managing thermal buffers in high-rise apartments, with both methods focused on optimized operation: a white-box, rule-based model using modelpredictive control (MPC) in a fully virtual simulation, and a solver-based, black-box optimization implemented at a pilot site via the FlexMeasures (FM) platform. The white-box model employs an Energy Flow-Based Model (EFBM) to simulate heat pump, thermal buffer, PV, electrical battery, and grid interactions under varying weather, occupancy, and electricity price conditions. A Breadth-First Search MPC explores hourly control trajectories over a 36-hour horizon, optimising buffer pre-charging and battery usage to minimize operational cost. At the pilot site, FM optimization focuses on the thermal buffer as the primary flexibility asset. The Foxtrot BEMS collects real-time measurements, including buffer state-of-charge (SoC), heat pump energy use, building load, PV generation, and ambient conditions, and communicates with FM every 15 minutes. FM returns optimized charging setpoints, which are translated into local heat pump operation while respecting operational constraints and comfort limits. The results presented here constitute the first experimental exploration that will support a subsequent comparison of the two optimization approaches. Both methods successfully shift buffer charging to costeffective periods, with the white-box model providing transparent, interpretable insights and FM demonstrating real-world adaptive scheduling under multi-objective constraints. This work lays the groundwork for a larger, ongoing study to systematically evaluate which approach (or hybrid combination) provides the most effective, scalable, and grid-aware control of thermal buffers in high-rise residential heating systems.