Impact of sizing and system design on the real-world performance of residential heat pumps

This thesis investigated how heat pump sizing and system design affect the real-world performance of residential heat pumps. The main motivation was to evaluate system performance using real operational data and to understand the factors that influence efficiency in practice. To do this, a dataset of 282 residential heat pump systems was used, obtained from HeatpumpMonitor.org. The database consists of real-world monitored operational data, including variables such as heat output, electrical consumption, and system supply and outdoor temperatures, along with metadata describing system characteristics. The data are recorded at a time resolution of 5 minutes. The selected data covered the heating season from September 2024 to February 2025. A data pre-processing method was applied, to ensure good data quality. This included removing inconsistent data, selecting a common time period, and converting high-resolution data into daily averages for easier analysis. The building heat demand was estimated using two approaches: a regression-based method and a maximum daily heat output method. Based on these estimates, key performance indicators were calculated for each installation, including Oversizing Factor, Full Load Hours, and Seasonal Coefficient of Performance. The results showed the median Seasonal Coefficient of Performance of 3.80, which is higher than the value of 2.65 reported in earlier UK field trials [1]. The analysis revealed median oversizing factor of 1.65 from regression-based heat loss and 1.56 from maximum daily heat output, which shows that in most installation practices, systems being typically 56% to 65% larger than the estimated design heat load. The main objectives of this work were to analyses how these performance indicators vary across different types of heat pump installations, to identify key performance trends, and to assess how system design factors such as hydraulic configuration, space heating control type, DHW configuration and terminal unit type influence the relationship between oversizing and performance. Directly coupled systems achieved the highest median SCOP of 3.86, compared to buffer tank systems and systems with low-loss headers, representing an approximately 11% gap between the directly coupled and low loss header hydraulic configurations. Systems with pure underfloor heating achieved the highest median SCOP deviation of +0.09 above the population median. Systems operating under pure weather compensation without room influence achieved the highest median SCOP of 3.97, compared to 3.48 for systems with significant room influence, representing a 12% performance difference driven by control strategy alone. The study also compared the measured performance of systems with expected benchmark values in order to identify the conditions that lead to good or poor performance, the results showed that the mean absolute percentage difference of 8% between the measured and benchmark performance In addition, the effect of oversizing on SCOP was analyzed across different system designs and utilization levels. The results showed that oversizing is most impactful in low-demand buildings (heat loss <4 kW), where highly oversized systems (OSF >1.8) exhibit negative SCOP error of (-0.04) indicating performance below the benchmark, while the same degree of oversizing in high-demand buildings (heat loss >6 kW) produces positive SCOP error of (+0.19). High system utilization can reduce the negative effect of oversizing as at low utilisation (<800 FLH), SCOP error is negative across all oversizing levels, whereas at moderate to high utilisation (>800 FLH) it becomes consistently positive. Directly coupled systems maintain positive SCOP error across all oversizing levels, buffer tank systems turn negative at high oversizing (>1.8), and low-loss header systems show negative error even at moderate oversizing (1.3–1.8).