Abstract

Space heating accounts for 30% of the total energy use in the EU, and to reduce costs and meet climate goals, efficient operation that maintains comfort must be developed. Hydronic heating is popular and attractive, as it is compatible with efficient heat production and distribution. The operation of hydronic systems typically provides good comfort, but high peak loads and overheating are common, particularly in older, poorly insulated buildings.

Recent deployments of sensors have improved dynamic control of the supply temperature via air temperature feedback, as used in model predictive control (MPC). Although this has led to reduced overheating, static configurations still limit performance. This thesis explores modeling of building thermal dynamics to improve two such configurations—district heating price models and flow rate balancing—enabling cheaper and more efficient operation.

District heating price models incentivize desirable operation of hydronic systems, and with cost-optimizing control (economic MPC), these incentives have an immediate effect. While price models often target total energy consumption, limiting peak loads also benefits district heating companies. Penalizing peak demand encourages shifting loads in time by exploiting building thermal inertia. By simulating optimal control, we show that strong peak-demand penalties can reduce overall peak load in a district heating network by 10–20% compared to models without such incentives.

Within a hydronic system, all radiators share a centrally controlled supply temperature. To ensure uniform temperatures and heat supply across zones, flow rates must be balanced through statically configured valves. The effect of adjusting those valves is often not obvious, as the zonal temperature variations depend on the weather. With a thermal dynamics model, we show how zonal variations relate to the angle between two parameter vectors. Using operational data, we demonstrate weather-independent evaluation of balancing by comparing this angle before and after adjustments.

The proposed methods are compatible with existing equipment, enabling immediate real-world implementation.

For further information and to join via Zoom, click here.

Friday 17 October at 13:15.