Extension of a Virtual Refrigerant Charge Sensor

Purdue University Purdue e-pubs International Refrigeration and Air Conditioning Conference School of Mechanical Engineering 2012 Extension of a Virtual Refrigerant Charge Sensor Woohyun i ki644@purdue.edu Jaes E. Braun Follow this and additional works at: http://docs.lib.purdue.edu/iracc i, Woohyun and Braun, Jaes E., "Extension of a Virtual Refrigerant Charge Sensor" (2012). International Refrigeration and Air Conditioning Conference. Paper 1243. http://docs.lib.purdue.edu/iracc/1243 This docuent has been ade available through Purdue e-pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu for additional inforation. Coplete proceedings ay be acquired in print and on CD-ROM directly fro the Ray W. Herrick Laboratories at https://engineering.purdue.edu/ Herrick/Events/orderlit.htl

2297, Page 1 Extension of a Virtual Refrigerant Charge Sensor Woohyun i, Jaes E. Braun* Herrick Laboratory, Purdue University, Mechanical Engineering, West Lafayette, IN, 47906, USA * Corresponding Author: jbraun@purdue.edu ABSTRACT The priary goal of the work deribed in this paper was to evaluate and extend a virtual refrigerant charge sensor (VRC) for deterining refrigerant charge for equipent having variable-speed copressors and fans. To evaluate the accuracy of the VRC, data were first collected fro previous laboratory tests for different systes and over a wide range of operating conditions. In addition, new laboratory tests were perfored to consider conditions not available within the existing data set. The systes for the new laboratory tests were two residential ductless split heat pup systes that eploy a variable-speed copressor and R-410a as the refrigerant. Based on the evaluations, the original virtual charge sensor (tered Model I) was found to work well in estiating the refrigerant charge for systes with a variable-speed copressor under any operating conditions. However, for extree test conditions such as low outdoor teperatures and low copressor speed, the VRC needed to be iproved. To overcoe the liitations, the odel associated with the VRC sensor was odified to include a ter involving the inlet quality to the evaporator (tered odel II). Both the odel I and II showed good perforance in ters of predicting charge levels for systes with a constant speed copressor, but odel II gave better perforance for systes with a variable-speed copressor. However, when the superheat of the copressor was zero, neither odel I nor II could accurately predict charge level. Therefore, a third approach (Model III) was developed that includes the diharge superheat of the copressor. This odel iproved perforance for a laboratory-tested syste that included a nuber of points with no superheat entering the copressor. 1. INTRODUCTION There have been laboratory studies that have docuented the ipact of refrigerant charge on the perforance of air conditioning equipent, including work by Rice (1987), Moshen (1990), Breuker et al. (1998), and Goswai (2002). Recently, i and Braun (2012) found that a refrigerant charge reduction of 25% led to an average energy efficiency reduction of about 15% and capacity degradation of about 20%. These studies showed that iproper refrigerant charge could significantly decrease energy efficiency and capacity and lead to operating conditions that decrease equipent lifespan. Furtherore, refrigerant charge leakage can contribute to global waring in the long ter. The leakage of refrigerant released to the atosphere contributes to the greenhouse effect. The other long-ter ipact is caused by the extra carbon dioxide eissions fro fossil fuel power plants due to lower energy efficiency. Packaged air conditioners are widely used in 46% of all coercial buildings, serving over 60% of the coercial building floor space in the U.S. (EIA, 2003). The survey data indicates that annual cooling energy consuption related to packaged air conditioner is about 160 trillion Btus. Therefore, sall iproveents in packaged air conditioner perforance can lead to significant reductions in overall energy use and environental ipact. Based on a survey and analysis of 215 rooftop units on 75 buildings in California, it has been shown that 46% of the units were not properly charged, which resulted in reductions in capacity and energy efficiency. The average energy ipact of refrigerant charge probles was about 5% of the annual cooling capacity. Based on research of ore than 4,000 residential cooling systes in California, only 38 % have correct charge (Downey, 2002) and the data fro Blasnik et al. (1996) have indicated that an undercharge of 15 % is coon.

2297, Page 2 The typical approach used to verify refrigerant charge for systes having variable-speed copressors was reviewed. Despite the fact that there are slight differences between anufacturers, the basic ethods are based on using easured pressure at the service valve deterined with a anifold gauge, when syste is operating at fixed speed in a test ode set by reote controller. A technician decides to add or reove refrigerant based on the difference between a pressure easureent and a target pressure specified by technical data provided by the anufacturer. These approaches can only deterine whether the charge is high or low, not the level of charge. In addition, the current charge verification protocols utilize pressure gauges or transducers installed at the service valve. The installation of these gauges or transducers can lead to refrigerant leakage. Because of these liitations, the current protocols for checking refrigerant charge ay be doing ore har than good in any situations. The original VRC sensor (Li and Braun: 2007, 2009) uses a correlation in ters of superheat and subcooling that are deterined using low-cost surface ounted teperature sensors. Paraeters of the ethod can be estiated using readily available anufacturers data. Furtherore, the charge estiates are relatively insensitive to the existence of other syste faults. Based on previous research (i and Braun: 2010), the VRC sensor was found to work well in estiating the refrigerant charge for a range of air conditioner and heat pups having fixed speed copressors. However, none of the tested equipent included variable-speed copressors. The current research extends the VRC sensor for systes with variable-speed copressors and fans. The VRC sensor is evaluated over a wide range of abient conditions for both heating and cooling. In particular, the odified VRC sensor represents an iproveent over the original ethod at extree conditions such as low copressor speed, and low outdoor teperature conditions. 2. EXTENSION of VRC SENSOR 2.1 Original VRC sensor for fixed speed copressor Li and Braun (2007) developed the VRC sensor (tered odel I) for correlating refrigerant charge level in ters of superheat and superheating. Deviations fro noinal charge can be obtained by using four easureents and four paraeters. The charge deviation relative to the charge is expressed as total total, total, 1 ch T T T T, sh/ sh sh, where total is the actual total charge, is the noinal total refrigerant charge, sh/ and ch are two constants that are characteristics of a given syste, and T, and T sh, are liquid line subcooling and suction line superheat at conditions with the noinal charge, respectively. The two constants T, and T sh, can be readily obtained fro technical data provided by anufacturers. As presented by Li and Braun (2009a), sh/ and ch can be estiated using the following equations. ch total, T, 1 o X hs, T sh T, (3) T T sh/ sh sh, total, o o total, hs, X hs, total, (4) (5) where total,o is the total refrigerant charge of the syste when subcooling and superheat are zero. X hs, is the ratio of high-side charge to the total refrigerant charge at the condition and α o is the ratio of refrigerant charge necessary to have satu liquid at the exit of the condenser to the refrigerant charge. In the current paper, three different approaches are considered for deterining the epirical paraeters: default paraeters, siulation paraeters, and tuned paraeters. Based on data available fro Hars (2002), Li and Braun (2009) found that a reasonable estiate for X hs, is 0.73 whereas a value of 0.75 was deterined for α o as a default paraeter. A reasonable estiate for sh/ for systes using a theral expansion valve (TXV) or fixed orifice (1) (2)

2297, Page 3 (FXO) as the expansion device is 1/2.5 based on previous test results. For a syste using an electronic expansion valve (EEV), superheat reains constant regardless of charge, and refrigerant inventory in the evaporator is relatively constant. In this case, a reasonable estiate for sh/ is 0. To iprove charge predictions, a siulation ethod for estiating ch was developed for extreely over and under refrigerant charge level (i and Braun: 2010). The siulation ethod deterines this epirical paraeter using a physical deription of the heat exchangers and piping along with easureents at conditions and the noinal charge level. ch should depend on three eleents of each syste: the liquid line length, the subcooling, and the charge. Different split and packaged systes can have very different liquid line lengths. The subcooling and the charge also vary as well, depending on each unit. Based on those findings, ch can be calculated fro the refrigerant ass distribution in the syste. For these calculations, the void fraction correlations based on the slip ratio correlated equation fro Zivi (1964) was found to give the best results. Alternatively, the epirical paraeters within the VRC sensor algorith can be tuned to iprove accuracy if data are available at different refrigerant charge levels and operating conditions. The paraeter tuning ethod iniizes the errors between predicted and known refrigerant charge by using linear regression techniques. In the current study, linear regression techniques were applied to all of the available data points for each syste: which can include variations in charge level, outdoor flow rate, indoor flow rate, abient teperature, and indoor dry bulb teperature. 2.2 Modified VRC sensor for variable-speed copressors Based on previous researches (Li and Braun: 2009, i and Braun: 2010), the VRC sensor for equipent with fixed speed copressor worked well with tuned paraeters, unless the syste was extreely over or undercharged. To extend the VRC sensor for equipent with variable-speed copressors, odified VRC sensor (ter odel II) were developed in this research. For odel II, a correlation for refrigerant charge in ter of evaporator inlet quality was added to the odel I. Figure 1 Operating states of vapor copression cycle. Figure 1 shows the operating states of vapor copression cycle. The high-side refrigerant charge is related to subcooling using hs T hs, o (6) where hs is the charge in the high-pressure side of the syste, hs,0 is the high-side refrigerant ass for the case of zero subcooling and is a constant that depends on the condenser geoetry. hs,o is assued to be a constant, independent of operating conditions and total charge. The low-side charge is related to superheat and inlet quality for the evaporator based on

2297, Page 4 T x x ) (7) ls ls, o sh sh X (, o in where ls,o is low-side refrigerant ass for the case of zero superheat and zero subcooling, x in is evaporator inlet quality, x is a constant characteristic of a given syste, and sh is a constant that depends on the evaporator geoetry. ls,o is assued to be a constant, independent of operating conditions and total charge. The subript,o denotes that the case of zero subcooling is eployed. Equations 6 and 7 can be applied to all operating condition including the condition so that hs, T, hs, o ls, ls, o sh Tsh, X ( x, o xevap, (8) ) (9) where the subript denotes that the rating operating conditions are eployed. T, and T sh, are liquid line subcooling and suction line superheat at conditions with the noinal charge, respectively. Equations 6 to 9 can be cobined to give expressions for changes in subcooling, superheat, and inlet quality of evaporator fro conditions in ters of charge variation by eliniating hs,o and ls,o. (, hs hs, T T ) ( ) (10) T T ) ( x x ) ( ) (11) sh ( sh sh, x in ls ls, Equations 10 and 11 are cobined and the result anipulated to give a single expression that relates subcooling, inlet quality of evaporator, and superheat to the total refrigerant charge. ( total total ) ( T T, ) sh ( Tsh Tsh, ) x ( x in x, ) (12) Equation 12 is then anipulated to give charge deviation fro charge relative to charge as a function of three inputs deterined fro easureents (T, T sh, x in ) and seven constants (, sh/, x/, total,, T,, T sh,, x ). total total, total, 1 ch T T T T x x, sh/ sh sh, x / in where sh/, x/ and ch are three epirical constants that ust be estiated for a given syste and the other four paraeters are directly deterined fro easureents at the condition and noinal charge level. The quality entering the evaporator can be readily estiated using easureents exiting the condenser and assuing an isenthalpic expansion process, In Equations 13, it is necessary to know x/. Default estiates of this epirical paraeter can be deterined fro data at the condition. Using equations 4 and 5 along with equation 14 with the case of zero superheat and zero subcooling, an expression for estiating x/ is deterined using the following steps, ( total, total, o, sh sh, x in ( total, x / ) T T ( x x ) (14) (1 o )) T T X, hs,, sh T sh, x ( x in x ) (13) (15) sh x T, Tsh, ( x in x ) (16) 1 Xhs, T, Tsh, sh/ X x x x x hs, hs, o (17) hs, o Under low copressor speed conditions with low abient teperature, the laboratory test results for this study had zero subcooling and superheat. In these cases, neither the odel I or odel II approaches, which use subcooling and

2297, Page 5 superheat easureent as input paraeters, can accurately predict the charge level. Therefore, a odel III approach was developed to provide iproved perforance in these situations. Model III is a odification of the odel II equation that includes a correlation for refrigerant charge in ter of diharge superheat of copressor. total total, total, 1 ch T T T T x x T T, sh/ sh sh, x / in dsh/ dsh dsh, where dsh/ is a constant characteristic of a given syste, and T dsh, is diharge superheat of the copressor at conditions with the noinal charge. 3. EVALUATION of VRC SENSOR for COOLING EQUIPMENT 3.1 Syste deriptions and test conditions To evaluate the VRC sensor, data for air conditioners and chillers with variable-speed copressors were collected where the effects of refrigerant charge on perforance were considered. Table 1 gives specifications for three units where data were obtained by laboratory testing. This includes data for two water-to-water chiller units and a conventional split air conditioner unit. R-22 and R-410A were used as the refrigerant and roll and reciprocating type copressors and electronic expansion valves (EEV) were eployed. The syste test conditions considered for the systes are listed in Table 2. The test data were all obtained at different copressor speeds. Refrigerant charge levels were varied between about 70 % and 130 % of noinal charge levels. However, ost of the tests were perfored at a single indoor and abient teperature. Table 1 Syste deriptions for existing refrigerant charge level test data Syste Capacity (kw) Refrigerant Copressor Expansion device Syste I Choi (2001) 3.5 R-22 Scroll EEV Water to Water II i (2003) 3 R-22 Reciprocating EEV Water to Water III Cho (2005) 7.2 R-410A Scroll EEV Air Split Type Table 2 Test conditions for cooling equipent having a variable-speed copressor Syste Indoor Water / Air Indoor Air Outdoor Water/ Air Cop Refrigerant Inlet Teperature Wet Teperature Inlet Teperature Speed Charge C C C Hz % I Choi (2001) 25-30, 34, 38, 42 30 ~ 60 80 ~ 120 II i (2003) 26.7-35 20 ~ 60 70 ~ 120 III Cho (2005) 27 19.51 35 40 ~ 60 70 ~ 130 3.2 Evaluation of VRC sensor for cooling The VRC sensor was evaluated in ters of RMS deviation between predicted and actual charge levels relative to noinal charges for the cooling systes having variable-speed copressors. The test data did not provide inforation necessary to estiate paraeters using the siulation approach. Therefore, odels I and II were only evaluated based on the use of default and tuned paraeters. Figures 2 to 5 show the perforance of the VRC sensor odel I and II based on the default and tuned paraeters. Figures 2 shows the perforance of the VRC sensor based on odel I with default paraeters. Overall, the RMS errors of the VRC sensor algorith for odel I were 8% based on default paraeters. In any cases, the accuracy of the refrigerant charge predictions is good when using default paraeters. However, the use of the default paraeters led to soe significant errors greater than 10% in refrigerant charge estiates at both low and high charge levels with low copressor frequencies. When the VRC sensor odel II with default paraeters was applied, there was an iproveent copared to using odel I with RMS errors of 6 % in figure 3. Model II with default paraeters can also lead to significant iproveents in cases where odel I does not work well, such as at (17)

2297, Page 6 extreely low outdoor teperatures and high charge level. However, there were still soe points with significant refrigerant charge estiate errors at high charge level with low copressor frequencies. To increase the accuracy of the VRC sensor, the paraeters were tuned for each specific syste based on easureents obtained at different refrigerant charge levels. When tuned paraeters were applied to the odel I and II, the VRC sensor showed better perforance than when the default paraeters were applied, as shown in Figure 4 and 5. The RMS errors were reduced to 4 % for odel I and 3 % for odel II. The results verified that tuned paraeters significantly iprove the accuracy of the VRC sensor. It can be seen that when the syste is not over charged, odel I with tuned paraeters has good perforance under various copressor speeds. However, when the syste is extreely over charged, odel I ay have significant errors. Copared to odel I, odel II led to soe iproveents in cases where odel I did not work well, such as low copressor speed and high charge level. Overall, the VRC sensor using odel II with tuned paraeters can provide very accurate estiates of refrigerant charge levels for cooling systes having variable-speed copressors. Figure 2 Perforance of VRC sensor odel I based on default paraeters for cooling Figure 3 Perforance of VRC sensor odel II based on default paraeters for cooling Figure 4 Perforance of VRC sensor odel I based on tuned paraeters for cooling Figure 5 Perforance of VRC sensor odel II based on tuned paraeters for cooling 4. EVALUATION of VRC SENSOR for HEAT PUMPS

2297, Page 7 4.1 Syste deription and test conditions The priary liitations of the previously available data for systes with variable-speed copressors are that the test conditions were liited to 1) cooling ode only, 2) with 70% as the lowest refrigerant charge level, 3) one abient teperature condition, and 4) for systes that do not incorporate ulti-speed fans. To better assess the accuracy and broaden the application of the VRC sensor, new test plans were established to consider the following key issues: 1) heating ode operation for heat pups, 2) various abient teperature conditions, 4) lower levels of refrigerant charge, and 5) systes with ulti speed fans. Two heat pup systes having a variable-speed copressor and ulti-speed fans were selected for testing and installed within the psychroetric chabers at Herrick Laboratories. Tables 3 provide an overview of the two systes that were tested. R-410A was used as the refrigerant for both systes and a heretic type copressor and EEV were eployed. The ranges of test conditions in cooling and heating ode are given in Tables 17. The test atrix was designed to consider both cooling and heating ode operation with low levels of refrigerant charge and low abient teperatures. Data for relatively low abient teperatures in cooling were necessary to test the validity of the algorith during off-season when regular aintenance procedures are often perfored. Refrigerant charge levels were varied between 50% and 130% of noinal charge levels with outdoor teperatures between about 67 F and 110 F for cooling ode and 17 F and 47 F for heating ode. The effects of reduced indoor air flow were also considered for syste III, with airflow rates fro 260 to 430 [CFM] for heating and cooling ode. The variable copressor speeds were considered fro 18 to 65 Hz in cooling ode and fro 18 to 130 Hz in heating ode. Table 3 Syste deription for heat pup systes having a variable-speed copressor Syste Size (kw) Refrigerant Type Copressor Expansion Device Accuulator Assebling Type III 3.5 R-410A Heretic type EEV Yes Split IV 3.5 R-410A Heretic type EEV Yes Split Syste III IV Table 4 Test conditions for heat pup systes having a variable-speed copressor Indoor Tep. Outdoor Tep. Indoor Refrigerant Copressor Speed Mode Dry Wet Dry Fan Speed Charge Level (F) (F) (F) [Hz] [CFM] (%) Cooling 80 67 110 / 95 / 67 21 ~ 65 430 / 260 50 ~ 130 Heating 70-47 / 37 / 17 21 ~ 130 430 /260 50 ~ 130 Cooling 80 67 105 / 95 / 67 18 ~ 49 410 50 ~ 130 Heating 70-47 / 27 / 17 18 ~105 410 50 ~ 150 4.2 Evaluation of VRC sensor for heat pup systes Figures 6 to 9 show the accuracy of the VRC sensor for the heat pup systes in cooling and heating ode. The perforance was evaluated in ters of RMS deviation fro the actual charge levels presented on a percentage basis for odels II and III. Figure 6 shows perforance of the VRC sensor based on odel II and default paraeters in cooling and heating ode. Based on the RMS errors of 16 % for cooling ode and 22 % for heating ode, the VRC sensor did not perfor well in predicting the charge level. As the fault level of refrigerant charge increased or decreased, there was bigger difference between estiated and actual charge aounts. For exaple, the odel II with default paraeters predicts 20 % undercharge when the syste is charged at 50% of noinal charge. When the abient teperature and copressor speed were low, the refrigerant charge error increased copared to other test conditions. Figure 7 shows results based on the use of paraeters that were deterined using the siulation approach. The odel II based on siulation paraeters showed RMS errors of 15 % for cooling ode and 20 % for heating ode. The use of siulation paraeters led to significant errors in refrigerant charge estiates at low and high charge level. The errors were relatively large at low copressor speed conditions at overcharge conditions. Overall, odel

2297, Page 8 II with siulation paraeters did not iprove the perforance of the VRC sensor copared to the default paraeters for the heat pup syste. Figure 6 Perforance of VRC sensor odel II based on default paraeters for heat pups Figure 7 Perforance of VRC sensor odel II based on siulation paraeters for heat pups Figure 8 shows perforance based on tuned paraeters. The RMS errors were reduced to 13 % for cooling ode and 12 % for heating ode. When tuned paraeters were applied in heating ode, there was a significant iproveent copared to using the default and siulation paraeters. Although the RMS error is reduced, the errors at high charge levels are greater with ore variability in the predictions. The errors were still large at high charge levels because the superheat exiting the copressor was nearly zero for various operating conditions. The VRC sensor II with tuned paraeters underestiates charge when the syste is highly overcharged with errors up to 30% at a charge level of 130 %. In cooling ode, the large deviations still reained at conditions having zero subcooling. Figure 9 shows perforance of odel III based on tuned paraeters. In this case, the RMS errors were reduced to 10% for cooling and 7 % for heating ode. In heating ode, odel III can lead to significant iproveents in cases where odels I and II do not work well, such as at overcharge conditions with extreely low outdoor teperatures and low speed copressor. Overall, the VRC odel III is better than the other two odels for characterizing refrigerant charge levels for heat pups with variable-speed copressor. However, there were still soe significant errors (over 5%) at low abient and low speeds when subcooling was zero. 5. COMPARISONS WITH MANUFACTURERS CHARGING METHOD The charging ethod specified by the anufacturer for syste IV was applied and copared with the VRC sensor based on odel III for cooling ode. The approach used to verify refrigerant charge in the field involves the use of pressure at the service valve. Suction pressure for cooling ode and diharge pressure for heating ode are used to indicate the charge level with the copressor operating at a fixed speed in a test ode. The technicians can evaluate whether to add or reove refrigerant based on a difference between the pressure easureent and a target pressure. Figure 10 shows easureents associated with applying the copany refrigerant charge protocol for syste IV in cooling ode at three different abient teperatures. The three horizontal lines correspond to the target suction pressures at the three teperatures. Although the suction pressure increases with charge level, it doesn t achieve the target even at 130% of noral charge. The deviation between the easured and target pressure is greatest at the lowest outdoor teperature. It appears that current approaches would have difficulty in identifying the proper charge aount during off-season aintenance. Figure 11 shows perforance of odel IV based on tuned paraeters using the data at axiu copressor speed in cooling ode. The VRC sensor provides accurate refrigerant charge estiates in cooling ode regardless of the abient teperature.

2297, Page 9 Figure 8 Perforance of VRC sensor odel II based on tuned paraeters for heat pups Figure 9 Perforance of VRC sensor odel III based on tuned paraeters for heat pups Figure 10 Refrigerant charge ethod based on copany ethod for cooling ode (Syste IV) Figure 11 Perforance of VRC sensor odel III based on tuned paraeters for the cooling ode (Syste IV) 6. CONCLUSIONS The original VRC sensor (tered odel I) using tuned paraeters worked well for different systes at any operating conditions but the perforance was significantly worse for low copressor speed and at low abient teperatures in both cooling and heating ode. Iproved perforance was achieved with a odification that accounts for variations in the quality of refrigerant entering the evaporator (tered odel II) but tended to fail under conditions with zero condenser subcooling and evaporator superheat for variable-speed heat pups. Better perforance was achieved for those conditions when copressor diharge superheat was included (tered odel III). For cooling equipent with variable-speed copressors, the odel I and II approaches were evaluated based on the use of default and tuned paraeters. The RMS errors based on default paraeters were 8% and 6% for odel I and

2297, Page 10 odel II, respectively. When tuned paraeters were used, the RMS errors were 4% and 3% for odel I and odel II, respectively. The odel II showed better perforance when the charge levels were sall and large. For the laboratory testing results fro heat pup systes with both variable-speed copressors and fans, the odel I and II ethods did not work well so the ethod was extended to include an additional input (odel III). When the odel III algorith was tuned using all available data, the overall RMS errors were 10% for cooling ode and 7% for heating ode, copared to over 10% for both cooling and heating ode when odel I and II were used. The cases where the VRC sensor with odel III had difficulty were when the syste was ope with zero subcooling at low copressor speed. The VRC sensor could be used as part of a peranently installed control or onitoring syste to indicate charge level and/or to autoatically detect and diagnose low or high levels of refrigerant charge. Continuous or frequent onitoring of charge level should lead to early detection of refrigerant leakage and avoidance of under or overcharging. It could also be used as a standalone tool by technicians in order to deterine existing charge and during the process of adjusting the refrigerant charge. The current charge protocols that are based on low pressure can only indicate whether refrigerant charge is high or low, whereas the VRC sensor provides a easure of the quantity of charge. The technician in the field could easily use the tool to deterine the correct aount of charge to add to the unit. NOMENCLATURE EEV Electronic expansion valve (-) Subripts FXO Fixed orifice (-) dsh Diharge superheat of copressor dsh/ Constant characteristic of a given syste related to diharge superheat of copressor Diharge superheat of copressor at (-) dsh, condition ch Epirical constant (-) in Inlet of evaporator sh Constant related to condenser subcooling and depending on the condenser geoetry Constant related to evaporator superheat and depending on the evaporator geoetry (-) hs High side (-) hs,o High side for zero-subcooling sh/ Epirical constant (-) ls Low side x/ Constant characteristic of a given syste related to inlet quality of evaporator (-) ls,o Low side for zero-superheat Refrigerant charge (-) Rating operating conditions total Total refrigerant charge (kg) Subcooling total, Total refrigerant charge at condition (kg), Rated subcooling T Teperature (kg) sh Superheat T Liquid line subcooling (C) sh, Rated superheat T, Liquid line subcooling at condition (C) tot Total T sh Evaporator superheat (C) tot,o T sh, Evaporator superheat at condition (C) Greek TXV Therostatic expansion valve (C) α o X hs, Ratio of high side charge to the total (-) Total for zero-subcooling and zerosuperheat Ratio of refrigerant charge necessary to have satu liquid existing the condenser at rating conditions to the refrigerant charge

2297, Page 11 refrigerant charge at rating conditions x Refrigerant quality (-) REFERENCES Blasnik, M., T. Downey, J. Proctor, and G. Peterson., 1996, Assessent of HVAC Installations in New Hoes in APS Service Territory, Proctor Engineering Group Report for Arizona Public Service Copany Breuker, M.S. and Braun, J.E., 1998, Coon faults and their ipacts for rooftop air conditioner, HVAC&R Research, Vol. 4, part 3, pp. 303-318. Choi, J.M., i, Y.C., 2001, The effect of iproper refrigerant charge on the perforance of a heat pup with an electronic expansion valve and capillary tube, Energy, Vol. 27, p. 391-404. EIA. Energy Inforation Adinistration., 2003, Coercial Buildings Energy Consuption Suvrvey 2003., U.S Survey 2003. U.S. Departent of Energy, Washington, D.C. Last accessed in July 2011 Downey, T. and Proctor, J., 2002, What Can 13,000 Air Conditioners Tell Us? In Proceedings of the ACEEE 2002 Suer Study on Energy Efficiency in Buildings, Vol. 1: p. 53-68. Washington D.C.: Aerican Council for an Energy-Efficient Econoy. Goswai D. Y., July 2002, Effect of refrigerant charge on the perforance of air conditioning systes, Fuel and Energy Abstracts, Volue 43, Issue 4, pp 284. Hars, T.M., 2002, Charge Inventory Syste Modeling and Validation for Unitary Air Conditioners, Ph.D. Thesis, Herrick Labs 2002-13, Report No. 5288-2, Purdue University, West Lafayette, IN. i,.c. and Cho, D.G., 2005., An experiental study on the perforance of inverter heat pup with a variation of frequency and refrigerant charging rate, orea Journal of Refrigerant, Vo15, p. 187-205. i, M. and i, M.S., 2003, Perforance investigation of a variable-speed vapor copression syste for fault detection and diagnosis, International Journal of Refrigeration, Vo28, p.481-488. i, W.H and Braun, J.E., 2010, Evaluation of a Virtual Refrigerant Charge sensor, Copressor, Refrigeration and Air Conditioning Conferences, IN, USA. i, W.H and Braun, J.E., In Press 2012, Evaluation of the Ipact of Refrigerant Charge on Air conditioner and Heat pup perforance, International Journal of Refrigeration. Li, H. and Braun, J.E., 2007, Evaluation of a Virtual Refrigerant Charge Level Gauge for Vapor Copression Equipent, IIR Congress of Refrigeration, Beijing, China. Li, H. and Braun, J.E., 2009a, Developent, Evaluation and Deonstration of a Virtual Refrigerant Charge Sensor, HVAC&R Research, Vol.15, No.1, Pages 117-136. Mohsen Farzad, 1990, Modeling the Effects of Refrigerant Charging on Air Conditioner Perforance Characteristics for Three Expansion Devices, Ph.D, Departent of Mechanical Engineering, Texas A&M University, TX. Rice, C.A, 1987. The Effect of Void Fraction Correlation and Heat Flux Assuption on Refrigerant Charge Inventory Predictions, ASHRAE Transactions, Vol. 93, Part 1, pp.341-367. Zivi, S.M., 1964, Estiation of steady-state stea void-fraction by eans of the principle of iniu entropy production. Journal of Heat Transfer, Vol. 86, pp. 247-252. ACNOWLEDGEMENT This work was supported by the National Institute of Standards and Technology.