A novel on-board method for quantifying refrigerant charge in residential heat pumps

Heat pump performances may be dramatically impacted by refrigerant leakages, but only after a substantial loss of refrigerant. It is thus crucial to detect and quantify leakages during the initial phase of slight performance decline, to provide accurate information to maintainers and adapt the maintenance accordingly. Most existing refrigerant charge assessment methods rely on subcooling as main charge-sensitive parameter. However, in modern, well-optimized heat pumps, subcooling is minimal and quickly drops to zero after a small loss of refrigerant, rendering the methods ineffective. A new method for refrigerant charge assessment, well-suited for these contemporary heat pumps operating in heating mode, is thus proposed in this study, leveraging the evaporator inlet vapor quality as key charge-sensitive feature. Unlike conventional methods which directly predict the refrigerant charge, this new method is based on an indirect scheme where the developed correlation uses the refrigerant charge as input and the charge-sensitive parameter as output. This method was validated in climatic chambers, on a variable-speed air-to-water heat pump filled with R32. It was trained on stationary data with imposed operating conditions and then tested using a semivirtual setup where the heat pump operated autonomously and dynamically based on a heat curve to compensate for the heat losses of a virtual building. The method demonstrated high accuracy, with a 1.5–2.5 % error on charge assessment for charge levels between 70 % and 110 %. The calibration effort was minimized to 40 operating conditions distributed across three charge levels, including an overcharge one.