CO2 dual-source (air/solar) heat pump on-field testing: results and performance maps development for optimal mode selection

This work investigates the performance of a direct-expansion dual-source (air/solar) heat pump operating with CO 2 as refrigerant, with the aim of identifying optimal operating conditions and supporting advanced control strategies. An experimental campaign was carried out under real-field conditions on a 10 kW prototype installed at the ENEA Casaccia Research Centre, considering both Air-Mode and Sun-Mode operation for domestic hot water production and producing a total amount of 307 tests. The results show that solar-assisted operation can significantly increase the system performance when sufficient irradiance is available. In particular, increasing the global irradiance from 322 to 904 W/m 2 led to a 21.1% increase in heating capacity and an 18.7% improvement in performance, primarily due to the increase in evaporation temperature. The higher evaporation level reduces the specific compression work and increases the refrigerant density at the compressor suction, resulting in a more efficient thermodynamic cycle. However, under low irradiance conditions, Air-Mode operation remains more efficient due to lower thermal resistance at the evaporator. To overcome the limitations of non-uniform experimental data, Artificial Neural Networks (ANNs) were developed to predict system performance over a wider operating domain. A robust model selection methodology based on bootstrap and K-fold resampling techniques was adopted to evaluate both interpolation and extrapolation capabilities. The trained ANNs were used to generate performance and convenience maps, allowing the identification of irradiance thresholds (from approximately 600 to 1000 W/m 2 ) at which Sun-Mode becomes more efficient than Air-Mode. A final analysis was also performed considering different climatic conditions and domestic hot water demand profiles according to UNI EN 16147. The results show that, although the electrical energy savings associated with the source-switching strategy remain limited (approximately 100–120 kWh/year under the most favorable conditions), the simultaneous electricity generation of the PV/T field significantly affects the overall economic performance of the system. A complementary thermo-economic assessment highlighted a strong dependence on climatic conditions, with simple payback periods ranging from approximately 6–10 years in locations characterized by high solar availability to over 20 years in less favorable climates.