Dilruba Rukhsana

Friday, March 13, 2026, 9:00 - 10:00

online

If you want to join the defense, please send an email to Dilruba Rukhsana/Jakob Harzer for the Link.

Abstract: In recent years, heat pumps have become the preferred technology for reducing building emissions. Existing multi-family houses with centralised heating systems often have high heating loads, which could be met with a single large heat pump. However, the installation space is often limited. This thesis investigates the control and optimisation of a cascade of three parallel air-source propane (R290) heat pumps as a flexible alternative to a single large- capacity unit. A representative building in Potsdam with a design heat load of 70 kW was simulated over a one-year period. Four hydraulic system topologies ranging from rigid single-mode operation to fully flexible layouts, were evaluated alongside different combinations of heat pump thermal capacities. The control system features a priority-based logic for scheduling the heat pump units and a hybrid staging algorithm that combines temperature hysteresis with temporization., with optimal parameter tuning via Bayesian optimisation. To maximize Seasonal Performance Factor (SPF) and minimize heat pump on/off cycling, Bayesian optimization was employed to tune the threshold parameters of the control logic. Results indicate that the single-mode configuration, which forces all heat pumps to operate in the same mode (either space heating or domestic hot water mode), resulted in the lowest SPF with excessive cycling. In contrast, the dedicated-II configuration, where two units are dedicated to space heating and one flexible unit between domestic hot water and space heating modes, achieved an SPF of 3.16 with reduced % less cycling than a single large reference case. Overall, the findings suggest that combining heterogenous heat pump sizing with parameter-optimised staging control can be a reliable approach for managing these retrofit systems.