Evaluating control strategies directly in district-heated buildings is challenging because experimenting with control actions may compromise indoor comfort and operational reliability. This work develops and calibrates a dynamic grey-box simulator of a district-heated residential building using operational data. The simulator couples a thermal resistance–capacitance building model with heat-exchanger and hydronic radiator dynamics. Calibration and validation against historical data demonstrate the model’s ability to approximate space-heating demand and provide a computationally efficient environment for comparative controller evaluation. Its use is demonstrated by comparing rule-based and proportionalintegral control strategies and by establishing the feasibility of training and evaluating a reinforcement learning agent within the simulated environment, while providing a basis for future extension to building portfolios.