ORNL/TM-2024/3585 CRADA/NFE-21-08858 Compressor is A Sensor - Final Report Bo Shen, Zhenning Li (ORNL), Drew Welch (Copeland LP) September 2024 Approved for Public Release ORNL IS MANAGED BY UT-BATTELLE LLC FOR THE US DEPARTMENT OF ENERGY DOCUMENT AVAILABILITY Reports produced after January 1, 1996, are generally available free via OSTI.GOV. Website www.osti.gov Reports produced before January 1, 1996, may be purchased by members of the public from the following source: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone 703-605-6000 (1-800-553-6847) TDD 703-487-4639 Fax 703-605-6900 E-mail info@ntis.gov Website http://classic.ntis.gov/ Reports are available to US Department of Energy (DOE) employees, DOE contractors, Energy Technology Data Exchange representatives, and International Nuclear Information System representatives from the following source: Office of Scientific and Technical Information PO Box 62 Oak Ridge, TN 37831 Telephone 865-576-8401 Fax 865-576-5728 CONTENTS LIST OF FIGURES ..................................................................................................................................... iv LIST OF TABLES ........................................................................................................................................ v 1. ABSTRACT ......................................................................................................................................... 6 2. STATEMENT OF OBJECTIVES ........................................................................................................ 7 2.1 SIMULATION of Pump-Down Operation ................................................................................. 8 2.2 CHARGE PREDICTION VALIDATION USING RESIDENTIAL HEAT PUMP .................. 8 2.3 CHARGE PREDICTION VALIDATION USING COMMERCIAL HEAT PUMP ................. 8 3. BENEFITS TO THE FUNDING DOE OFFICE'S MISSION ............................................................. 8 4. TECHNICAL DISCUSSION OF WORK PERFORMED BY ALL PARTIES ................................... 9 4.1 MODELING APPROACH OF PUMP-DOWN ......................................................................... 9 4.2 VALIDATION OF THE PUMP-DOWN BASED REFRIGERANT CHARGE FAULT Prediction on a residential Heat pump ...................................................................................... 12 4.2.1 TESTED RESIDENTIAL HEAT PUMP .................................................................... 12 4.2.2 TEST MATRIX ........................................................................................................... 14 4.2.3 VALIDATION OF COMPRESSOR PERFORMANCE ............................................. 14 4.2.4 VALIDATION OF METHOD BASED ON CHARGE MIGRATION IN COMPRESSOR ........................................................................................................... 15 4.2.5 METHOD BASED ON CHARGE MIGRATION IN COMPRESSOR ...................... 16 4.2.6 METHOD USING QUASI-STEADY STATE DATA AT THE END OF PUMP-DOWN ............................................................................................................. 19 4.2.7 RESIDENTIAL HEAT PUMP VALIDATION RESULTS ........................................ 19 4.3 VALIDATION OF THE PUMP-DOWN BASED REFRIGERANT CHARGE FAULT PREDICTION ON A LIGHT COMMERCIAL SYSTEM ...................................................... 20 4.3.1 PERFORMANCE TEST OF PACKAGED ROOFTOP HEAT PUMP ...................... 20 4.3.2 EXPERIMENT VALIDATION AND CALABORATION OF HEAT EXCHANGER MODEL .............................................................................................. 23 4.3.3 LIGHT-COMMERCIAL HEAT PUMP VALIDATION MATRIX AND RESULTS .................................................................................................................... 25 5. COMMUNICATIONS, COMMERCIALIZATION POSSIBILITIES, AND FUTURE COLLABORATION .......................................................................................................................... 27 6. CONCLUSIONS ................................................................................................................................ 27 7. REFERENCES ................................................................................................................................... 28 iii LIST OF FIGURES Figure 1: Schematic of Pump Down Operation of Heat Pump (https://highperformancehvac.com/)........... 7 Figure 2: Copeland CoreSenseTM Diagnostic, Communications and Protection Technology ...................... 8 Figure 3: (a) low-side half system; (b) high-side half system. .................................................................... 10 Figure 4: Transients of charge migration .................................................................................................... 11 Figure 5: Transients of suction pressure ..................................................................................................... 11 Figure 6: Transients of mass flow rate in compressor ................................................................................ 12 Figure 7: Outdoor unit of heat pump placed on scale to measure charge migration................................... 12 Figure 8: Indoor Unit of heat pump. ........................................................................................................... 13 Figure 9: Refrigerant side instrumentations. ............................................................................................... 14 Figure 10: Compressor map validation – compressor power...................................................................... 15 Figure 11: Compressor map validation – refrigerant mass flow rate .......................................................... 15 Figure 12. Validation of charge prediction method based on charge migration in compressor.................. 16 Figure 13. Compressor volumetric efficiency during pump-down at Toutdoor = 71 °F, charge = 7 lbm ................................................................................................................................................. 16 Figure 14. Refrigerant mass flow rate via compressor at Toutdoor = 71 °F, charge = 7 lb ............................ 17 Figure 15. Suction and discharge pressure during pump-down at Toutdoor = 71 °F, charge = 7 lb ............... 17 Figure 16. Refrigerant mass flow rate via four-way valve.......................................................................... 18 Figure 17. Refrigerant mass flow rate via four-way valve.......................................................................... 18 Figure 18. Validation of charge prediction method based on theoretical analysis for steady-state operation. ....................................................................................................................................... 20 Figure 19: 4-ton package roof-top heat pump system ................................................................................. 21 Figure 20: Copeland single-stage R410A compressor: ZP42K5E-TF5 ...................................................... 22 Figure 21: Roof-top unit cooling performance test results ......................................................................... 22 Figure 22: Roof-top unit heating performance test results.......................................................................... 23 Figure 23: Heating capacity loss at different ambient temperatures ........................................................... 23 Figure 24: Model validation for cooling mode: (a) EER; (b) Capacity ...................................................... 25 Figure 25: Model validation for heating mode: (a) COP; (b) Capacity ...................................................... 25 Figure 26: Roof-top unit charge prediction validation, pump-down from steady-state operation (a) calibration equation (b) predicted charge versus measure charge.................................................. 26 Figure 27: Roof-top unit charge prediction validation, pump-down from system standby (a) calibration equation (b) predicted charge versus measure charge.................................................. 27 iv LIST OF TABLES Table 1: Dimensions of indoor and outdoor microchannel heat exchangers. ............................................. 13 Table 2. Experiment test matrix for residential split system ....................................................................... 14 Table 3: Performance test matrix for the packaged roof-top unit ............................................................... 21 Table 4: Simulation Inputs for 4-ton Roof-top Unit ................................................................................... 23 Table 5: Summary of Model Validation Data and Deviation ..................................................................... 24 Table 6: Pump-down test matrix for 4-ton roof-top unit............................................................................. 26 v 1. ABSTRACT The performance of the heat pump system varies greatly depending on the refrigerant charge amount. Improving the refrigerant charge fault detection and diagnostics (FDD) method of vapor compression systems have the potential for increasing energy efficiency and reducing service cost. Previous studies to predict refrigerant charge amount are mostly empirical methods which require significant amount of experimental data for high accuracy. The primary goal of this research is to develop a charge fault detection method which requires only a few experimental data with high prediction accuracy. Currently, pump down operations are typical practices by HVAC technicians when they need to open the refrigerant circuit to make a repairment. When pumping down, a valve at the liquid line would close to stop the refrigerant flow, and the refrigerant is pumped by the compressor from the evaporator side to the condenser side. By monitoring the real time suction/discharge pressures and temperatures, the refrigerant mass at the evaporating and condensing sides can be accurately calculated. Therefore, it is important to conduct dynamic simulation of the pump-down operation. In this project, ORNL partnered with the Helix Innovation Center in Copeland LLC developed a model to simulate the pumping down process. The simulation is driven by DOE/ORNL Heat Pump Design Model (HPDM) [1]. The model successfully captured the dynamics of system during pump-down operation and the result is reviewed and validated by the test data. Secondly, the project team validated the pump-down based charge prediction method via experiment on a 3-ton residential R410A split heat pump system. The experiments were conducted at three different charge levels, i.e., 7 lbm, 8 lbm, and 9 lbm. At each charge level, the pump down operations were conducted for four different outdoor conditions. For the same charge level and the same outdoor temperature, the pump-down operations were repeated three times. The charge prediction method innovatively utilizes the compressor low-pressure cut-off protection and the compressor performance map. Two charge prediction methods are validated. The first method is based on the prediction of charge migration in the compressor during the pump-down process, and the second method is based on the usage of detailed heat exchanger model with inputs from quasi-steady-state operation at the end of pump-down. Thirdly, the charge prediction method was validated on a 4-ton light commercial packaged roof-top heat pump with by manually closing a valve on the liquid line at system steady-state operation mode and system stand-by mode. The experiments were conducted at four different charge levels, i.e., 8 lbm, 9 lbm, 10 lbm and 11 lbm. At each charge level, the pump-down operations were conducted for four different outdoor conditions. As a result of validation, the deviation of the charge prediction method based on refrigerant mass migration during the pump-down process is within 10% for residential split heat pump system, and the deviation of the charge prediction method using detailed heat exchanger model and quasi-steady-state data at the end of pump-down is within 6% for both residential split system and light commercial packaged roof-top heat pump system. 2. STATEMENT OF OBJECTIVES Currently, HVAC technicians are regularly performing refrigerant pump down operations when they need to open the refrigerant circuit to make a repairment. Figure 1 shows a typical residential heat pump system, there are two cut-off valves, one valve in the suction line and one valve in the liquid line. As the technician closes the liquid line valve and turns the air conditioning on. The compressor pushes all the refrigerant into the high-pressure side of the system. Usually, the compressor has low-pressure protection mechanism, e.g., 19 Psig low pressure limit for a typical R410A system, when the suction pressure gets to this predetermined lower limit, the compressor will be turned off by the unit control. Most of the vapor compression systems using Copeland scroll compressors have equipped with the low- pressure protection feature, thus, the termination of the pump down will be executed automatically at a constant low suction pressure without extra cost. At the end of the pump down, the suction pressure is extremely low. And the refrigerant mass flow rate is very small, because the compressor mass flow rate is dictated by the compressor displacement volume and low suction pressure setting (vapor density). At the termination, the vapor compression system reaches near steady-state balance points. Automatic pump down operation can be conducted by systems equipped with one solenoid cut-off valve at the condenser exit. After the pump down operation, the refrigerant can be released back to low-pressure side by opening the valve. The charge prediction method developed in this project is based on pump-down operation. Figure 1: Schematic of Pump Down Operation of Heat Pump (https://highperformancehvac.com/) Copeland is a world leader in making building equipment including innovative scroll compressors. For years, Copeland scroll compressors are flagship products setting the standard in commercial applications with proven reliability and superior performance. With the CoreSenseTM sensor (Figure 2), Copeland Scroll compressors can feedback real-time diagnostics information during the compressor operation. By using the compressor as a sensor, valuable data from inside the system can be accessed and monitored by the controller. This project will take advantage of measured data from CoreSense TM to achieve an automatic, fast, accurate, cost-effective solution for refrigerant charge fault detection. Figure 2: Copeland CoreSenseTM Diagnostic, Communications and Protection Technology 2.1 SIMULATION of Pump-Down Operation To make the innovative FDD method a success, accurate simulation of the compressor behavior during the pump down procedure is the key. And the extensive library of compressor maps [2] are an asset to achieve the project goal. Therefore, the first objective of project team it to develop a model to simulate the pump- down operation. it will contain detailed phase-to-phase heat exchanger models to simulate charge migration in the system and correlate how the outdoor and indoor conditions impact on the time span and system balance points when terminating the pump down. Suitable air side and refrigerant heat transfer correlations will be adopted from the DOE/ORNL Heat Pump Design Model. Standard compressor maps will be used to predict the compressor mass flow rate and power consumption. The model aims at revealing dynamic key factors impacting the pump down. 2.2 CHARGE PREDICTION VALIDATION USING RESIDENTIAL HEAT PUMP The project teams conducted laboratory validation of the charge fault detection method on residential heat pumps. The test system was modified to accommodate variances with/without charge buffers, i.e., liquid receiver and suction line accumulator. The project team tested a 3-ton split system under various indoor and outdoor temperatures as well as various refrigerant charge levels. The performance metrics corresponding to this objective is to evaluate whether the charge identification approach can capture the system charge with +/-10% accuracy. 2.3 CHARGE PREDICTION VALIDATION USING COMMERCIAL HEAT PUMP After the validation on residential heat pump, the project team conducted validation on light commercial systems under various ambient conditions and charge levels. A 4-ton packaged roof-top unit was investigated. Validation on the light commercial systems aims at showing how accurate the FDD technique performs in large systems and verify the method’s feasibility in commercial systems. The performance metric corresponding to this objective is to evaluate whether the charge identification approach can capture the system charge with +/-10% accuracy for light commercial heat pump. 3. BENEFITS TO THE FUNDING DOE OFFICE'S MISSION Reported by Kim and Braun [3], more than 50 percent of packaged air conditioning systems are improperly charged. Improper refrigerant charge can increase energy usage, reduce capacity, and decrease equipment lifespan. Based on other research covering more than 4,000 residential cooling systems in California, many systems were identified with incorrect refrigerant charge levels, of which about 34% are undercharged, 28% are overcharged, and only 38% of systems have correct charge. More commonly, the residential systems tend to be undercharged by 15%. The improper charge levels cause long term inefficient operations, reduced capacities, lack of dehumidification, etc. An effective and real-time charge identification technique can detect the fault and minimize the inefficient operations. The pump down based FDD is a low-cost technique, which does not need extensive laboratory data for model characterization. It can be universally applied, and thus, reduce energy consumption on average by up to 10% in the residential and commercial sectors, via minimizing inefficient, low charge operations. 4. TECHNICAL DISCUSSION OF WORK PERFORMED BY ALL PARTIES 4.1 MODELING APPROACH OF PUMP-DOWN At the termination of a pump down operation, the low-pressure side refrigerant mass is negligible. However, it can still be calculated to improve the overall charge identification accuracy. Upon the low-pressure protection, the low side volumes, including the evaporator, lines, and charge buffers downstream of the shut off valve, contain all vapor at the suction pressure and temperature. For example, a typical R410A system turns on a low-pressure protection at 30 Psig, -16.0°F saturation temperature. It has a 90 °R temperature difference to a typical room temperature. Therefore, all the liquid refrigerant at the low side should have been vaporized. Once the liquid refrigerant is gone, the suction process can be approximated by the ideal gas law. Equation (1) represents the moment that the liquid refrigerant dissipates, for example, approximately 20 Psi higher than the low-pressure protection. Equation (2) represents the moment of low- pressure protection. Equation (3) depicts the integration process from moment 1 to moment 2. With the suction pressures (𝑃1 and 𝑃2 ) and temperatures (𝑇1 and 𝑇2 ) measured, and the compressor mass flow rate 𝑚̇ predicted by the map and integrated with time, the low side charges ( 𝑚1 and 𝑚2 ) and the total low side volume 𝑉 can be solved. With the low side density known at the pump-down termination, the remaining low side charge can be calculated by 𝑚2 = 𝜌(𝑃2 , 𝑇2 ) ∗ 𝑉. 𝑃1 ∗ 𝑉 = 𝑚1 ∗ 𝑅 ∗ 𝑇1 (1) 𝑃2 ∗ 𝑉 = 𝑚2 ∗ 𝑅 ∗ 𝑇2 (2) 2 ∫1 𝑚̇ 𝑑𝑡 = 𝑚1 − 𝑚2 (3) Regarding a reversible heat pump, the low side charge identification method can be used to calculate the whole system charge by performing two consecutive pump-down operations in reversed cycles. During the first pump down operation, the remaining low side charge at the termination can be determined. Right after the first pump down, the heat pump will activate a reversed cycle to switch the low side and high side, and then pump the refrigerant back. During the second pump down, the refrigerant mass migrates from the low side to high side will be monitored and integrated with time. Sum of the remaining low side charge at the first pump down, the migrated charge during the second pump down, and the remaining low side charge at the second pump down will give an estimate of the total system charge with high accuracy. To develop and validate the charge identification approach, detailed quasi steady-state system modelling to simulate pump down operation is conducted. The quasi steady-state simulation model is developed using DOE/ORNL Heat Pump Design Model (HPDM). As an industry leading system simulation, HPDM integrates expertise in thermodynamics, heat transfer, fluid dynamics, numerical analysis, nonlinear equation solving, and object-oriented programming with optimization tools for the design and control of highly efficient vapor compression systems. Firstly, the system is dissembled into two half systems as shown in Figure 3. (a) (b) Figure 3: (a) low-side half system; (b) high-side half system. We assumed that the low-side refrigerant temperature is maintained at indoor temperature during the pump-down process and the refrigerant mass is distributed evenly per component inner volumes at start up. The simulation starts by specifying the low-side inlet enthalpy as the saturated liquid enthalpy under room temperature. Then the low-side and high-side half systems are simulated to obtain the charges in both half system. At each time step, by assuming a guessed evaporator inlet enthalpy, the new charges at low side and high side are calculated. The charge increase in high-side is compared to the charge reduction in low-side. And the difference of them are the criteria to adjust the guessed evaporator inlet enthalpy. The algorithm converges when the suction pressure reaches a predefined threshold which is determined by the compressor low-pressure protection mechanism. The quasi steady-state pump-down process simulation is conducted on a 5-ton residential heat pump system tested by Copeland. The total R410A charge is 13.28 lbm. Figure 4 shows the transients of charge migration. Pump-down operation takes 24 seconds until compressor cut-off initiates at 55 psi suction pressure. At the end of 24s, the remaining charge in low-side is 0.69 lbm, which is 5.2% of the total charge. This result matches with what the industrial partner observed during their pump-down experiment tests. Figure 4: Transients of charge migration Figure 5 shows the transients of suction pressure, from 0 s to 4 s, the evaporator inlet is still in two-phase, so the pressure does not decrease as dramatically as latter session when the entire low side is occupied with superheated vapor. Figure 5: Transients of suction pressure Figure 6 shows the mass flow rate transients. As indicated, the mass flow rate drops significantly fast after the low pressure side is superheated. Figure 6: Transients of mass flow rate in compressor 4.2 VALIDATION OF THE PUMP-DOWN BASED REFRIGERANT CHARGE FAULT PREDICTION ON A RESIDENTIAL HEAT PUMP 4.2.1 TESTED RESIDENTIAL HEAT PUMP To validate the charge prediction method on a residential system, a 3-ton R410A split heat pump system using microchannel heat exchangers as both indoor and outdoor coils was extensively investigated in laboratory. The outdoor unit is shown in Figure 7 and the indoor unit is shown in Figure 8. The heat pump uses microchannel heat exchangers in both outdoor and indoor units. Detailed heat exchanger parameters are shown in Table 1. The system used adjustable thermo-expansion valves (TXVs) for cooling and heating superheat control. In the experiments, both the indoor and the outdoor units were located in two environmental chambers, where temperature and humidity were controlled. The indoor return and supply air temperatures were measured via averaging six thermocouples. The return air relative humidity was measured using one probe sensor. The outdoor air temperature was sensed via six thermocouples uniformly placed at the inlet of each outdoor coil. Figure 7: Outdoor unit of heat pump placed on scale to measure charge migration. Figure 8: Indoor Unit of heat pump. Table 1: Dimensions of indoor and outdoor microchannel heat exchangers. Parameter Indoor MCHX Outdoor MCHX Face area (m2) 0.58 2.09 Total Tube Number 54 229 Number of rows 1 1 Fin density (fins/m) 787 630 Tube width (m) 0.254 0.206 Tube Height (m) 0.013 0.013 Number of Microchannels 26 20 Microchannel hydronic diameter (m) 0.0068 0.0071 Figure 9 depicts the refrigerant side instrumentations on each system, including a refrigerant mass flow meter at the liquid line; pressure transducer and inserted probe thermocouple at the compressor suction and discharge to monitor the compressor operation; pressure transducer and probe thermocouple at the vapor line and liquid line connecting to the indoor heat exchanger. The compressor power, outdoor fan power and indoor blower power were measured separately using watt transducers. The pressure measurements at the compressor suction and discharge correspond to the suction and discharge saturation (dew point) temperatures. When postprocessing the data, the suction superheat degree was calculated using the suction temperature minus the suction saturation temperature. The subcooling degree was calculated using the saturation temperature (bubble point) corresponding to the pressure measured at the liquid line entering/exiting the indoor unit, minus the temperature measured at the same location. The discharge temperature was measured by an inserted thermocouple at the compressor discharge. The pressure at the vapor line of the indoor heat exchanger was used to indicate the indoor heat exchanger saturation vapor temperature. Figure 9: Refrigerant side instrumentations. Regarding the uncertainty of the sensors, the refrigerant mass flowrate measurements were measured using Coriolis mass flow meter (±0.1% reading). The temperature was measured using calibrated thermocouples (±0.3 ◦C), and the pressure was measured with precision pressure sensors (±0.08%). The electric powers were measured using high precision power meters (±0.2% reading). 4.2.2 TEST MATRIX The pump-down operations were conducted under 3 different charge levels, i.e., 7.0 lbm, 8.08 lbm and 9.11 lbm. For each charge level, the pump-down were repeated three times. Pump-down starts from system stand-by mode by closing the cur-off valve at liquid line and start the compressor and outdoor fan. When the charge level is increased to 9.11 lbm and the outdoor temperature was higher than 79 °F, the compressor low-pressure protection cannot be activated due to the leakage of refrigerant from high-pressure side to low-pressure side via the 4-way valve. Therefore, for 9.11 lbm, only test data at 71 °F and 79 °F were obtained. Table 2. Experiment test matrix for residential split system Charge 7 lb 8.08 lb 9.11 lb Outdoor 71 °F (Cooling) 71 °F (Cooling) 71 °F (Cooling) Temperature 79 °F (Cooling) 79 °F (Cooling) 79 °F (Cooling) 87 °F (Cooling) 87 °F (Cooling) NA 95 °F (Cooling) 95 °F (Cooling) NA 4.2.3 VALIDATION OF COMPRESSOR PERFORMANCE To guarantee the compressor has predictable performance according to compressor map provided by the manufacturer, the compressor power and mass flow rate were validated against compressor map under steady-state heat pump operations. This is important because the charge prediction methods highly depend on the refrigerant mass flow rate and the volumetric efficiency of compressor prediction of compressor map. Figure 10 and Figure 11 show the power and mass flow rate validation of the compressor map, respectively. The power and mass flow rate both are within 3% deviation. The compressor map has reliable prediction of the compressor performance. Compressor Map Validation - Power Map Predicted Power [W] 2940 2935 2930 2925 2920 2915 2785 2790 2795 2800 2805 2810 Measured Power [W] Figure 10: Compressor map validation – compressor power Compressor Map Verification - Mass flow rate 0.15 Map Mass Flow Rate [lb/s] 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 SuctionDensity * DispVol [lb/s] Figure 11: Compressor map validation – refrigerant mass flow rate 4.2.4 VALIDATION OF METHOD BASED ON CHARGE MIGRATION IN COMPRESSOR Figure 12 shows the validation result for the charge prediction method based on charge migration in compressor and 4-way valve. Since the stationary charge in condenser before pump-down and stationary charge in evaporator after pump-down are in vapor phase and occupies small portion of total charge, assuming the migrated charge is 95% of the total charge, the total charge prediction is within 6% deviation compared with measurements. Figure 12. Validation of charge prediction method based on charge migration in compressor. 4.2.5 METHOD BASED ON CHARGE MIGRATION IN COMPRESSOR The first validated charge prediction method is based on the charge migration in the compressor and 4-way valve during the pump-down process. Compressor map was used to predict the volumetric efficiency degradation during pump-down process. Figure 13 shows the compressor volumetric efficiency during pump-down when the Toutdoor is 71 °F and the total charge 7 lb. When the volumetric efficiency drops significantly with the increase of pressure difference between suction and discharge points. Compressor Volumetric Efficiency during Pump-down 1 Volumetric Efficiency [-] Eta_V 0.8 0.6 0.4 0.2 0 0 5 10 15 20 25 30 35 40 45 50 Time [s] Figure 13. Compressor volumetric efficiency during pump-down at Toutdoor = 71 °F, charge = 7 lbm With the volumetric efficiency predicted, the refrigerant mass flow rate via compressor is calculated using Equation (4) as shown in Figure 14. Mrcompressor = DisplacementVolume  suction volumetric (4) Refrigerant Mass Flow Rate via Compressor 0.50 0.45 Refrigerant Mass Flow Rate [lb/s] Mass Flow Rate [lb/s] 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 0 5 10 15 20 25 30 35 40 45 50 Time [s] Figure 14. Refrigerant mass flow rate via compressor at Toutdoor = 71 °F, charge = 7 lb On the other hand, there is leakage in the 4-way valve. Therefore, it is important to calculate the refrigerant migration via the 4-way valve driven by the pressure difference between discharge and suction. The volumetric flow rate of refrigerant via 4-way valve is calculated by Equation (5) where Cv is a constant to correlate pressure difference with refrigerant volumetric flow rate. The mass flow rate of the refrigerant via 4-way valve is obtained using Equation (6). VFR4WayValve = Cv fourWayValve  ( Pdisch arg e − Psuction )2 (5) Mr4−wayValve = disch arg e *VFR4−WayValve (6) Figure 15 shows the suction and discharge pressure during pump-down. The pressure difference increases, therefore, the refrigerant migration from high-pressure side to low-pressure side also increases. Suction and Discharge Pressure 360 320 280 Pressure [psi] 240 P_discharge [psi] 200 P_suction [psi] 160 120 80 40 0 0 5 10 15 20 25 30 35 40 45 50 Time [s] Figure 15. Suction and discharge pressure during pump-down at Toutdoor = 71 °F, charge = 7 lb Figure 16 shows the refrigerant mass flow rate via 4-way valve. During the pump-down process, the refrigerant leakage in 4-way valve is increasingly significant. Refrigerant Mass Flow Rate via Four-way Valve 0.0004 Refrigerant Mass Flow Rate 0.0003 Mass Flow Rate [lb/s] [lb/s] 0.0003 0.0002 0.0002 0.0001 0.0001 0.0000 0 5 10 15 20 25 30 35 40 45 50 Time [s] Figure 16. Refrigerant mass flow rate via four-way valve Figure 17 shows an overview for the pump-down experiment at different charge levels and at different outdoor temperatures. The more refrigerant charged in the system, the pump-down takes longer time to reach the low-pressure protection. The higher the outdoor temperature, the longer the pump-down takes. Pump-down Duration at Different outdoor T and Charge 120 Pump-down Duration [s] 100 80 60 40 Charge = 9.109 lb 20 Charge = 8.078 lb Charge = 7 lb 0 60 70 80 90 100 110 Outdoor Temperature [° F] Figure 17. Refrigerant mass flow rate via four-way valve For the pump-down process, the total charge in the system consists of four components as shown in Equation (7). On the right hand side of the Equation (4), Charge ViaCompressor is the charge migration from low-side to high-side through compressor. Chargevia4WayValve is the charge migration from high-side to low-side through 4-way valve. The last terms refer to two stationary charges. One stationary charge is the vapor refrigerant in the condenser side before pump-down and the other stationary charge is the vapor refrigerant in the evaporator side after pump-down. These stationary charges are in vapor phase and thus occupies a very small portion of the total charge. It is crucial to predict the charge migration via compressor and the charge migration via 4-way valve accurately. Charge=ChargeViaCompressor -Chargevia4WayValve +StationaryChargeCondenserBeforePumpDown +StationaryChargeEvaporatorAfterPumpDown (7) 4.2.6 METHOD USING QUASI-STEADY STATE DATA AT THE END OF PUMP-DOWN The 2nd method is enlightened by a theoretical analysis based charge prediction method for steady-state operation of heat pump developed by Hong et al (2019) [4]. At the end of the pump down, the suction pressure is extremely low. And the refrigerant mass flow rate is very small, because the compressor mass flow rate is dictated by the compressor displacement volume and low suction pressure setting (vapor density). At the termination, the vapor compression system reaches quasi-steady-state. The derivation of the prediction equation is not included for brevity. Reader can refer to the original publication for more details of this method. Equation (8) is the final charge prediction equation. • SuperHeatdisch arg e (8) m mch arg e = C1  h fg + C2  + C3 (Tcondensation − Toutdoor ) (Tdisch arg ePo int − Toutdoor ) + (Tcondensation − Toutdoor ) In Equation (5), there are three constants C1, C2 and C3. Those constants are specific values for each system. They can be obtained through a few runs of pump-down operation. Only the last data point when the pump-down reaches quasi-steady-state is used. Since there are three unknowns, at least three data sets are required to assign the value of them. In next section, it will be shown that it is more desirable to derive the coefficients through more experimental data points. The more data points used to regress the constant; the equation can capture the charge more accurately. The information required to use this equation is the mass flowrate, the condensing pressure (or temperature), the compressor discharge temperature of the refrigerant, and the inlet temperature of the air at the condenser. The compressor isentropic and volumetric efficiencies are also obtained from compressor map. 4.2.7 RESIDENTIAL HEAT PUMP VALIDATION RESULTS To validate the prediction method based on theoretical analysis of steady-state operation, the test data was separated into the training group to regress the three constants C1, C2 and C3 in Equation (5) and a validation group to validate the accuracy of the fitted equation. There are 30 pump-down data sets as shown in Table 2, and different partition of training-validation data set were used as below. Step-1: Use 4 data at 4 different Toutdoor for 7 lbm charge for training, the remaining 26 data for validation, Figure 18(a) Step-2: Use 4 data at 4 different Toutdoor for 8 lbm charge for training, the remaining 26 data for validation, Figure 18(b) Step-3: Use 10 data at all Toutdoor and charge levels for training, the remaining 20 data for validation, Figure 18(c) Figure 18 (a) and Figure 18 (b) show that if the training data are from the test using the same charge level under different Toutdoor, it shows superior prediction for that specific charge level and less accurate for other charge levels. The overall prediction accuracy is within 15%, if the training data are from tests with only one charge level. Figure 18 (c) shows that the prediction accuracy is significantly improved to be within 10% if the training data covers three charge levels. The charge prediction method based on theoretical analysis for steady-state operation is capable to predict total charge with 10% deviation. Despite there are only 3 constants in Equation (8), 10 data sets at different charge levels are necessary to make the prediction result reliable. (a) (b) (c) Figure 18. Validation of charge prediction method based on theoretical analysis for steady-state operation. 4.3 VALIDATION OF THE PUMP-DOWN BASED REFRIGERANT CHARGE FAULT PREDICTION ON A LIGHT COMMERCIAL SYSTEM 4.3.1 PERFORMANCE TEST OF PACKAGED ROOFTOP HEAT PUMP A 4-ton commercial packaged roof-top unit (Figure 19) was tested in the laboratory. The pump-down is manually initiated by close the valve between the EXV and evaporator inlet. Figure 19: 4-ton package roof-top heat pump system To validate the RTU is installed properly, a performance test under AHRI 210/240 test standards were performed. Table 3 shows the test matrix. Table 3: Performance test matrix for the packaged roof-top unit Outdoor Indoor room Mode room Dry bulb, °F Wet bulb, °F Dry bulb, °F Wet bulb, °F Cooling 80 67 115 × Cooling 80 67 105 × Cooling (A) 80 67 95 × Cooling 80 67 75 × Cooling 80 67 65 × Cooling 80 75 95 × Cooling 80 63 95 × Cooling 80 60 95 × Heating 80 × 47 43 Heating (H1) 70 × 47 43 Heating 60 × 47 43 Heating 70 × 60 55 Heating 70 × 52 42 Heating 70 × 25 21 Heating 70 × 12 10 Heating 70 × 5 3 The system is equipped with Copeland single-stage compressor as shown in Figure 20. Figure 20: Copeland single-stage R410A compressor: ZP42K5E-TF5 The cooling test results are shown in Figure 21Figure 11. Figure 21 (a) shows the EER and Figure 21 (b) shows the cooling capacity. (a) (b) Figure 21: Roof-top unit cooling performance test results The heating test results are shown in Figure 22. Figure 22 (a) shows the COP at different ambient conditions and Figure 22 (b) shows the cooling capacity. The COP and EER are calculated using the delivered capacity, i.e., air side capacity. (a) (b) Figure 22: Roof-top unit heating performance test results During the performance test, it is observed that there is significant COP degradation under cold climate attributes to the capacity loss between evaporator and condenser boxes via the package thermal insulation. Figure 23 shows that the refrigerant side capacity is much larger than air side capacity at low ambient temperature. At 5 °F, the delivered refrigerant side capacity is 70% of the air side capacity. Figure 23: Heating capacity loss at different ambient temperatures 4.3.2 EXPERIMENT VALIDATION AND CALABORATION OF HEAT EXCHANGER MODEL A detailed system model is established using DOE/ORNL Heat Pump Design Model, The geometry inputs for each component used in the model are listed in Table 4Table 4. Table 4: Simulation Inputs for 4-ton Roof-top Unit Component Parameter Unit Value Number of Tube Banks - 1 Indoor Microchannel Finned tube length mm 743 Heat Unfinned tube length mm 10 Exchanger Tube depth mm 32 Number of tubes - 88 Tube height mm 1.3 Tube vertical spacing mm 9.4 Number of ports per tube - 32 Port cross-section - rectangular Port width mm 0.66 Port height mm 0.74 V-shape Fin type - louver Fin density Fins per inch 20 Number of louvers 20 Number of Tube Banks - 1 Finned tube length mm 802 Unfinned tube length mm 5 Tube depth mm 32 Number of tubes - 176 Tube height mm 1.3 Outdoor Tube vertical spacing mm 9.4 Microchannel Heat Number of ports per tube - 32 Exchanger Port cross-section - Rectangular Port width mm 0.66 Port height mm 0.74 V-shape Fin type - louver Fin density Fins per inch 16 Number of louvers 20 Emerson Compressor Compressor map - Copeland ZP51K5E Length m 4.33 Vapor line Diameter mm 22.22 Liquid line Length m 2.83 Diameter mm 12.7 The model is validated under a wide range of operating conditions from the product manual. The range of cooling and heating test data available in the product manual and the validation results are summarized in Table 5. The deviation of EER, COP and capacity is within 6%. Table 5: Summary of Model Validation Data and Deviation Flow Rate Indoor EER (or COP) Absolute Capacity Absolute Outdoor Number Operation of Indoor Return Air Deviation Deviation Temperature of data Mode Blower Temperature [°F] points Max. Min. Avg. Max. Min. Avg. [CFM] [°F] Cooling 1750-2250 75-125 59-71 165 5.65% 1.07% 3.46% 5.03% 1.07% 3.25% Heating 1750-2250 0-50 60-80 99 5.86% 1.14% 3.46% 5.45% 1.13% 3.14% Figure 24 shows the EER and capacity validation under cooling mode. (a) (b) Figure 24: Model validation for cooling mode: (a) EER; (b) Capacity Figure 25 shows the COP and capacity validation under heating mode. (a) (b) Figure 25: Model validation for heating mode: (a) COP; (b) Capacity 4.3.3 LIGHT-COMMERCIAL HEAT PUMP VALIDATION MATRIX AND RESULTS Pump-down operations are performed by manually closing a valve between the EXV and the indoor coil inlet. The pump-down operations were performed at 4 different charge level. At each charge level, the pump-downs are conducted at four different conditions. All the pump-downs are conducted under cooling mode. Table 6: Pump-down test matrix for 4-ton roof-top unit Test Matrix Charge 8 lbm 9 lbm 10 lbm 11 lbm Level [lb] Pump- Pump- Pump- Pump- Pump- Pump- Pump- Pump- down from down from down from down down down down from down steady- standby steady-state from from from steady- from state standby steady- standby state standby state 71 F √ √ √ √ √ √ √ √ 79 F √ √ √ √ √ √ √ √ 87 F √ √ √ √ √ √ √ √ 95 F √ √ √ √ √ √ √ √ At the end of the pump-down, it can be assumed that all the refrigerants were pumped to condenser. Therefore, at the end of pump down, the system reaches quasi steady state, and detailed HX model is used to predict the subcooled liquid length and the charge in condenser. Shen et al. (2009) [5] shows the errors between actual charge and simulated charge is a linear function of liquid length in the condenser. For the pump-downs initiated from system steady-state operation, Figure 26 (a) shows the calibration equation to tune the simulated charge with the measure charge. Using this collaborated charge and detailed heat exchanger model, the charge prediction is within 5% of the measured charge value. Using pump-down data from system steady-state operation mode (a) (b) Figure 26: Roof-top unit charge prediction validation, pump-down from steady-state operation (a) calibration equation (b) predicted charge versus measure charge. Similarly, for the pump-downs initiated from system stand-by mode, Figure 27 (a) shows the calibration equation to tube the simulated charge with the measured charge. And Figure 27 (b) shows the charge prediction is within 5% accuracy. In field application, this calibrating equation can be fitted using limited test data by OEM, and the equation can be used to predict charge after the system is installed on site. (a) (b) Figure 27: Roof-top unit charge prediction validation, pump-down from system standby (a) calibration equation (b) predicted charge versus measure charge. 5. COMMUNICATIONS, COMMERCIALIZATION POSSIBILITIES, AND FUTURE COLLABORATION The following conference paper was published and presented to disseminate the findings of this work: Li, Z., Welch, D., Shen, B., Gluesenkamp, K., Butler, B. and Morgan, S., 2022. A Universal Refrigerant Charge Fault Detection and Diagnostics Method Based on Pump Down Operation. 19th International Refrigeration and Air Conditioning Conference Purdue. Paper # 2429. 6. CONCLUSIONS The performance of the heat pump system varies greatly depending on the refrigerant charge amount. Improving the refrigerant charge fault detection and diagnostics (FDD) method of vapor compression systems have the potential for increasing energy efficiency and reducing service cost. The primary goal of this research is to develop a charge prediction method which requires only a few experimental data with high prediction accuracy. When pumping down, a valve at the liquid line would close to stop the refrigerant flow, and the refrigerant is pumped by the compressor from the evaporator side to the condenser side. By monitoring the real time suction/discharge pressures and temperatures, the refrigerant mass at the evaporating and condensing sides can be accurately calculated. In the report, the project team develops a model to simulate the pumping down process driven by DOE/ORNL Heat Pump Design Model (HPDM). The model captured the transients of system during pump-down operation and the result was validated by the test data provided by industrial partner. Two charge prediction methods are validated. The first method is based on the prediction of charge migration in the compressor during the pump-down process, and the second method is based on the heat pump quasi-steady-state operation at the end of pump-down. The pump-down experiments were conducted on a R410A 3-ton residential heat pump and a R410A 4-ton packaged roof-top unit. The experiments were conducted at three different charge levels. At each charge level, the pump down operations were conducted for different outdoor conditions. As a result, the deviation of the charge prediction method using both residential system and light commercial system is within 10%. 7. REFERENCES 1. HPDM, DOE/ORNL Heat Pump Design Model, https://hpdmflex.ornl.gov. 2. Copeland online product information: https://webapps.emerson.com/online-product-information/ 3. Kim, W. and J.E. Braun, Evaluation of the impacts of refrigerant charge on air conditioner and heat pump performance. International journal of refrigeration, 2012. 35(7): p. 1805-1814. 4. Hong, S.B., J.W. Yoo, and M.S. Kim, A theoretical refrigerant charge prediction equation for air source heat pump system based on sensor information. International Journal of Refrigeration, 2019. 104: p. 335-343. 5. Shen, B., Braun, J.E. and Groll, E.A., 2009. Improved methodologies for simulating unitary air conditioners at off-design conditions. International Journal of Refrigeration, 32(7), pp.1837-1849.