Semantic Digital Twinning for Cost-Optimal HVAC Operation: Real-Time Application to a House with Smart Thermostats and PV/Battery under a Time-of-Use Tariff
Matin Abtahi, Luis Rueda, Benoit Delcroix, Andreas Athienitis
Abstract
Semantic digital twinning has traditionally supported design coordination, documentation, and planning during the early stages of building projects. However, its application in building operation and maintenance—particularly in real time—remains limited. This study proposes a methodology for cost-optimal HVAC load management using an operational digital twin, and demonstrates its real-time application under a residential time-of-use pricing scheme. The framework is implemented in a grid-connected single-family house equipped with smart thermostats, rooftop photovoltaic panels, and battery storage, and is evaluated under two progressive layers of control and system integration: predictive thermostat control alone, and combined coordination of thermostats, on-site generation, and battery systems. Each configuration is assessed against a static reference derived from two baseline weeks without energy flexibility. Results show that predictive thermostat control reduced electricity costs by an average of 34.7 %, with a total increase in energy import of approximately 84 kWh, while maintaining average indoor temperature deviations below 0.3 °C. Coordinated control achieved 78.4 % average cost savings, reduced net grid import by 115 kWh, and enabled 25.3 kWh of energy export. Relative demand shift analysis confirmed effective load advancement and midday demand reduction, delivering both economic and grid-responsive outcomes. These findings highlight the feasibility of deploying real-time predictive control in operational residential buildings to enhance load flexibility and improve alignment with dynamic electricity pricing.