Economic model predictive control for energy-efficient residential heating under dynamic electricity pricing

• EMPC cuts heating cost by 19.5% under dynamic electricity pricing. • Energy consumption is reduced by about 10% without sacrificing comfort. • Peak electrical demand drops from 3.8 kW to 3.4 kW with EMPC. • Thermal comfort is maintained within 18-22°C in all scenarios. • Real-time feasibility is ensured with low and stable solve times. This study develops and evaluates an Economic Model Predictive Control (EMPC) for residential heating under dynamic electricity pricing, benchmarked against a standard Comfort-Oriented MPC (C-MPC). An RC thermal model is driven by real price traces and weather data. EMPC embeds time-varying tariffs directly in the objective and includes a Δ u penalty to smooth actuation. The evaluation reports cost, energy, peak demand, and comfort, with bootstrap 95% confidence intervals. Robustness is tested under forecast errors in prices and temperature (MAPE 5–20%, bias scenarios). Computational feasibility is quantified through problem size, solver configuration, and per-step solve time relative to the sampling period. Results show −19.5% total heating cost, −10% energy use, and peak demand reduced from 3.8 to 3.4 kW with indoor temperature held in 18–22°C. Savings persist under moderate forecast errors, and the Δ u term lowers total variation in control actions. Solve times remain within the sampling budget, indicating real-time viability. These findings establish that price-aware MPC yields measurable economic and energy benefits without compromising comfort, while remaining implementable on typical building automation stacks. The approach supports demand-response readiness and provides a transparent, reproducible baseline for establishing the theoretical upper bound of economic efficiency in residential heating.