- Email: [email protected]
Experimental investigation of an integrated cooling system driven by both liquid refrigerant pump and vapor compressor
Energy and Buildings 154 (2017) 560–568
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Experimental investigation of an integrated cooling system driven by both liquid refrigerant pump and vapor compressor Jiankai Dong a,∗ , Yana Lin a , Shiming Deng b , Chao Shen a,∗ , Zhuo Zhang c a b c
School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, China Department of Building Services Engineering, The Hong Kong Polytechnic University, Kowloon, Hong Kong Special Administrative Region China Railway Siyuan Survey and Design Group CO., LTD., China
a r t i c l e
i n f o
Article history: Received 17 July 2017 Received in revised form 13 August 2017 Accepted 16 August 2017 Available online 24 August 2017 Keywords: Integrated cooling system Pumped refrigerant two-phase cycle Vapor compression cycle Liquid refrigerant pump Feasibility analysis
a b s t r a c t The energy consumption of air conditioning system accounts for a significant proportion of the total energy consumed in data center. To reduce the energy consumption of the cooling system, an integrated system, which combines pumped refrigerant two-phase cycle (PRTC) with vapor compression cycle (VCC), is proposed in this paper. The impacts of input frequency and outdoor air temperature on the performances of PRTC, and that of the outdoor air temperature on the performances of VCC have been investigated. It was found that PRTC had higher EER than VCC, but its cooling capacity was limited by outdoor air temperature. In PRTC, the designed cooling capacity of 2500 W could be obtained when the outdoor air temperature was −3.0 ◦ C. In addition, 80% and 50% of the designed cooling capacity can be obtained at ambient temperatures 3.2 ◦ C and 12.0 ◦ C, respectively. Besides, the feasibility of the system in different climate zones in China was demonstrated by analysis of the climate data in Harbin, Beijing, Wuhan, Guangzhou and Guiyang, respectively. Results indicated the proposed system had high feasibility in Harbin and Beijing. The results of this paper can provide valuable insights into the cooling system driven by both liquid refrigerant pump (LRP) and vapor compressor, targeting at reducing the cooling energy consumption and maximizing the economic efficiency of the cooling system in data center. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The rapid development of Information Technology (IT) industry has resulted in a significant increase of data centers [1]. To ensure the reliable operation of the equipment in the data center, it’s necessary to apply the cooling system [2]. In 2011, the energy consumption of data center was more than 100 billion kWh in US [3]. Among these, the energy consumed in cooling system accounted for 30% [4], which means that proper optimization in cooling system would make a vital contribution to energy saving. Vapor compression cooling system, as a conventional cooling system, has been widely applied in the data center no matter what the outdoor air temperature is. However, there is a considerable amount of free cooling energy that could be used when the outdoor air temperature is low. It’s no doubt that using free cooling technology can decrease energy consumption effectively. From the open literature available, several investigations in terms of direct fresh-air cooling were conducted [2,4–9]. Bulut [2]
∗ Corresponding authors. E-mail addresses: [email protected] (J. Dong), [email protected] (C. Shen). http://dx.doi.org/10.1016/j.enbuild.2017.08.038 0378-7788/© 2017 Elsevier B.V. All rights reserved.
carried out an investigation on the performance characteristics of the cooling system in Istanbul. It can be concluded that the temperatures of indoor supplied air and outdoor ambient air affected the amount of the energy saving in the cooling system. According to their research, free cooling technology was not an optimum choice in the period from June to August when outdoor air temperature was equal to or slightly lower than the supply air temperature. Ham [4] simulated the performances of three types of air-side economizers, including direct air-side economizer with ultrasonic humidifier, indirect air-side economizer with heat pipe and indirect air-side economizer with indirect evaporative cooler, under different supply air temperatures and heat exchanger effectiveness in South Korea. The results demonstrated that the optimum supply air temperature was within 18–23 ◦ C. Lee [5] used a dynamic building energy simulation program to examine the potential energy savings of the air-side free cooling technology with differential enthalpy control used in data centers in 17 climate zones. The study showed that the energy saving of free cooling technology decreased by 2.8–8.5% with 2 ◦ C decline of indoor temperature. Endo [6] experimentally investigated a cooling system introducing fresh air directly for a data center located in Tokyo under different climate conditions. The results indicated that the energy
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Nomenclature cp h m Q T
Specific heat, kJ/(kg K) Enthalpy, kJ/kg Mass flow rate of the fluid, kg/s Cooling capacity(W) Temperature (◦ C)
V W
Volumetric flow rate, m3 /s Power (W)
·
Subscripts a Air com Compressor Condenser con eva Evaporator f Fan in Inlet of indoor coil Outlet of indoor coil out PRTC Pumped refrigerant two-phase cycle LRP Liquid refrigerant pump VCC Vapor compression cycle Abbreviations Energy efficiency ratio EER IT Information technology LRP Liquid refrigerant pump Proton exchange membrane PEM TPTL Two-phase thermosyphon loop Greek symbols Fluid density (kg/m3 )
consumption of the cooling system was 20.8% less than that of a conventional air conditioning system. According to the meteorological data of the past sixteen years in Islamabad, Pakistan, Hassan [7] carried out an investigation on the potential application for free cooling technology. The results showed that significant energy savings through free cooling can be achieved during the months of December, January, and February. Siriwardana [8] made an investigation on using air-side economizers to introduce outside air based on analyzing the hourly temperature and humidity data over past 12 years in Australia, and determined that the outside air cooling energy could be used for 5500 h per year in Tasmania and Southern Australian cities, such as Melbourne and Adelaide. Liu [9] concluded that full fresh air ventilation can save 40% of the air conditioning energy conservation and make power usage effectiveness (PUE) lower than 1.4. However, most of these investigations mentioned above actually focused on the energy saving in direct fresh-air cooling, while some focused on air-side economizers to analyze optimum supply air temperature to evaluate the potential usage of direct fresh-air cooling. Although the direct fresh-air cooling system could achieve enormous energy conservation, the utilization of direct fresh-air cooling was limited due to its performance fluctuation, which was resulted from the uncontrollability of outdoor air. Besides, the indoor air quality was badly affected if the outdoor air was supplied directly without purification treatment. Furthermore, heating the supplied air was necessary to avoid fogging or dewing when the outdoor air temperature was low, which caused an increase in the initial cost of direct fresh-air cooling system. Therefore, indirect fresh-air cooling attracts more and more attention because of its relatively high cooling capacity and controllability. The studies on indirect fresh-air cooling were mainly divided into natural cooling and mechanical cooling. There were some researchers concentrating on the study of natural cooling.
561
Zhang [10] established a visual experimental setup to investigate the performances of a two-phase thermosiphon loop (TPTL), in terms of the variation of temperature difference, liquid charge, height difference, and circulation flow resistance, etc. The results showed that larger height difference was not always advantageous to the performance of TPTL. Garrity [11] developed a flow boiling micro-channel cooling plate which can be easily inserted into the stack of a proton exchange membrane (PEM) fuel cell for thermal control. Furthermore, some researchers paid special attention to heat pipe. Jouhara [12] studied the thermal management based on heat pipe. The results revealed that 75% of potential energy saving can be achieved. Ding [13] made a conclusion that the total entransy dissipation of the heat pipe system was 48.3% lower than that of a traditional air conditioning system by analyzing the heat transfer process. On the other hand, some researchers showed their interest in mechanical cooling. Choi [14] investigated an integrated system designed for telecommunication equipment rooms. It can be seen that the optimum outdoor air temperature was 8.3 ◦ C, which was the control parameter to convert the cycle using ethylene glycol as refrigerant into VCC. Considering the driving force into the indirect free cooling technology, some researchers showed a significant concern for applying pump into cooling system. Hannemann[15] applied pump into liquid multiphase cooling system. According to this study, the mass flow rate, size of condenser and power consumed by the pump in water-circulating single-phase cooling system were 4.6 times, 2 times and 10 times as those in liquid multiphase cooling system, respectively. Marcinichen [16] carried out a further investigation on the thermal performances of refrigerated cooling of microprocessors with micro-evaporation. The results suggested that the energy consumption of vapor compression cooling system was far higher than that of pump-driven cooling system. Yan [17] introduced a novel energy saving cooling system combining traditional VCC with pumped liquid two-phase cooling cycle, and evaluated the system performance under various operation conditions. The results showed that the proper shift temperature between these two operations was about −5 ◦ C according to the system EER and cooling capacity. Ma [18] proposed a novel integrated cooling system of vapor compression and pump-driven two-phase loop for energy saving in data centers. And the results showed that the maximum cooling EER of the integrated system reached up to 29.71, and the proper shift temperature of the system using 3 HP compressor and 1 HP compressor was 5 ◦ C and 15 ◦ C, respectively. As mentioned previously, a considerable amount of literature related to free cooling technology has been carried out. Some focused on direct fresh-air cooling, while other paid attention to indirect fresh-air cooling. Among the studies for the indirect freshair cooling, the integrated system, combining vapor compression cooling with pump-driven two-phase cooling, has obvious energy saving, controllability, and can be used in both micro-channel evaporator and data centers to effectively use the natural cooling energy and reduce the energy consumption. Although Yan [17] and Ma [18] introduced the integrated systems combining vapor compression with pump-driven two-phase cooling cycle, both did not indicate the correspondence of LRP and compressor for generating certain cooling capacity. Simultaneously, the used LRP in these two experiments were both constant frequency. However, the cooling capacity of the LRP varied obviously with the change of outdoor air temperature under fixed condenser and evaporator. And thus, changing the input frequency into the LRP according to the variation of outdoor environment would improve its cooling performance. In this paper, an integrated system driven by vapor compressor and LRP is firstly developed. Then the operation performances of the system are experimentally evaluated under two operation cycles. This is followed by presenting the experimental results and detailed
562
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 1. Experimental setup and arrangement of measuring points.
Table 1 Specification of the split air-conditioner.
Table 2 Specification of measuring points.
Parameters
Value
Frequency conversion or not Refrigerant Cooling capacity Input power in cooling mode Volumetric flow rate Size of condenser Size of evaporator Copper tube diameter in condenser Copper tube diameter in evaporator
not R22 3300 955 600 780 × 250 × 650 795 × 194 × 273 8 12
Unit
Testing items
Notation
Measuring objects
Temperature
T1, T2 T3, T4 T5, T6 T7, T8 T9 T10 T11 T12 T13 T14 T15 P1 P2 P3 P4 P5 P6 P7
air at inlet of condenser air at outlet of condenser air at inlet of evaporator air at outlet of evaporator Refrigerant at inlet of condenser Refrigerant at outlet of condenser Refrigerant at inlet of LRP Refrigerant at outlet of LRP Refrigerant at inlet of evaporator Refrigerant at outlet of evaporator Refrigerant at outlet of compressor Refrigerant at inlet of condenser Refrigerant at outlet of condenser Refrigerant at inlet of LRP Refrigerant at outlet of LRP Refrigerant at inlet of evaporator Refrigerant at outlet of evaporator Refrigerant at outlet of compressor
W W m3 /h mm mm mm mm
discussion. Finally, conclusions are made following the feasibility analysis. The results of this paper can provide valuable insights into the cooling system driven by both LRP and vapor compressor, targeting at reducing the cooling energy consumption and maximizing the economic efficiency of the cooling system in data center.
Pressure
2. Experimental set-up and procedures 2.1. Experimental setup An experimental setup was firstly established, and it was made up of a psychrometric chamber and the integrated cooling system. The psychrometric chamber was of size of 3.2 m (L) × 3.0 m (W) × 2.4 m (H), in which the air could be maintained at temperature of 25.0 ± 1.0 ◦ C and relative humidity of 20%–30% by using a separated air conditioning system, a 6.5 kW electric heater, a humidifier of 3.0 kg/h water evaporation. At the same time, the outdoor coil was placed in the outdoor natural environment in Harbin, China during experiments. The air temperature in winter in Harbin was low enough for the experiments, and the lowest temperature could be as low as −35.0 ◦ C. As shown in Fig. 1, the integrated cooling system combining PRTC with VCC mainly consisted of an indoor coil, an outdoor coil, a compressor, a LRP, a reservoir, a capillary, and air fans, etc. The integrated cooling system were modified on an ordinary split air-conditioner with a nominal cooling capacity of 3300 W. The detailed specification of the split air-conditioner is given in Table 1. The reservoir with volume of 5.0 L was installed between the outdoor coil and LRP. The installation height of the reservoir was 0.5 m
higher than that of LRP to ensure the effective operation of the LRP. The volumetric flow rate of the indoor air fan was 600 m3 /h. The copper tube used between outdoor coil and indoor coil was of length of 3.0 m. Besides, R22 was chosen as the refrigerant. Table 2 gives the specification of measuring points in the system. During experiments, platinum resistance thermometers (PT1000, ±0.1 ◦ C accuracy) were used to measure the temperature at the inlet and outlet of evaporator, condenser and other components. Refrigeration pressures were measured by pressure transmitters (at ±0.1% accuracy). The input powers into LRP and compressor were measured by power transducer (at ±0.2% accuracy). All the measurement data was recorded and stored by Agilent 34980A series data acquisition system with an interval of 10 s.
2.2. Selection of liquid refrigerant pump LRP was a key component in the successful development of this integrated system. To select a matched pump, the values of mass flow rate and delivery head were both parameters needed to be consider.
J. Dong et al. / Energy and Buildings 154 (2017) 560–568 Table 3 Calculated parameters for choosing the LRP. refrigerant saturation pressure refrigerant state at inlet of evaporator refrigerant state at outlet of evaporator refrigerant state at outlet of condenser refrigerant state at inlet of condenser the diameter of the connected copper tube refrigerant state in the connected copper tube
563
The PRTC operated under two conditions. The performances of PRTC under the conditions of outdoor air temperature 0 ◦ C, input frequencies into the LRP 15, 20, 30 and 40 Hz, were firstly investigated. Then the performances of PRTC under the fixed input frequency into the LRP 20 Hz, outdoor air temperature ranging from −15.0 to 15.0 ◦ C were conducted. During operation of VCC, the inside condition in the psychrometric chamber were maintained at the same values with those in the operation of PRTC. but the outdoor air temperature ranged from 10.0 to 30.0 ◦ C.
497.59 kPa saturated-liquid two-phase mixture with 0.5 dryness saturated-liquid two-phase mixture with 0.5 dryness 10.0 mm 0.5 of dryness
2.4. Data reduction
Table 4 Specification of the LRP. Parameters
Value
Frequency converter or not Volumetric flow rate Delivery head Shaft power Power supply Measurement
Yes 0.04–0.2 22–100 0.2 380 V/3P&220 V/1N, 50 Hz 312 × 155 × 165
Unit m3 /h m kW mm
Firstly, the mass flow rate of the working fluid was evaluated by Eq. (1) under the calculated parameters given in Table 3. QPRTC = mPRTC (h2 − h1 )
(1)
where h2 is determined by the refrigerant pressure and temperature at the outlet of the evaporator. h1 by the refrigerant pressure and temperature at the inlet of the evaporator. The value of QPRTC is set as 3300W, equals to the nominal cooling capacity of the split air-conditioner. Therefore, the value of mass flow rate of the working refrigerant was 0.321 kg/s. Secondly, to choose the delivery head, the total flow resistance of PRTC should be calculated, which was made up of the resistance of evaporator, condenser and connected copper tube. The equations for calculating the flow resistance of single-phase working refrigerant can be found in Reference [19], while those for calculating the flow resistance of two-phase working refrigerant can be found in References [20–22]. Therefore, the total flow resistance of PRTC was 168.04 kPa including the flow resistance in evaporator of 78.45 kPa, condenser of 49.78 kPa, and copper tube of 39.81 kPa. Finally, based on the calculation results, the values of reasonable mass flow rate and delivery head were 0.321 kg/s (0.0901 m3 /h) and 168.04 kPa (13.4 m), respectively. However, there was no LRP satisfying simultaneously these two precise parameters. Therefore, a LRP with mass flow rate of 0.2 m3 /h and delivery head of 22.0 m was chosen. The detailed specification of the LRP is shown in Table 4.
During the experiments, the data of cooling capacity and EER were calculated. The cooling capacities of both VCC and PRTC can be calculated by Eq. (2):
·
Q = ma hout,a − hin,a = a V cp Tout,a − Tin,a
where Q is the cooling capacity, W; ma the air mass flow rate across indoor coil, kg/s; hout,a the enthalpy of air at outlet of indoor coil, kJ/kg; hin,a the enthalpy of air at inlet of indoor coil, kJ/kg; a the air ·
density, kg/m3 ; V the volumetric flow rate of air, m3 /s; cp the specific heat of air, kJ/(kg K); Tout,a air temperature at outlet of indoor coil, ◦ C; Tin,a air temperature at inlet of indoor coil, ◦ C. The EER of PRTC can be calculated by Eq. (3):
EERPRTC = QPRTC / WLRP + Weva,f + Wcon,f
(3)
Where EERPRTC is the energy efficiency ratio of PRTC; QPRTC the cooling capacity of PRTC, W; WLRP the power input of the liquid refrigerant pump, W; Weva,f the power input of the indoor coil fan, W; Wcon,f the power input of the outdoor coil fan, W. The EER of VCC can be calculated by Eq. (4):
EERVCC = QVCC / Wcom + Weva,f + Wcon,f
(4)
Where EERVCC is the energy efficiency ratio of VCC; QVCC the cooling capacity of VCC, W; Wcom the power input of the compressor, W. 3. Results and discussion The influences of outdoor air temperature on performances of PRTC and VCC, as well as the influences of input frequency on performances of PRTC, are shown in Figs. 2–10.
2.3. Experimental conditions and procedures For determining the cycle operation and improving the performance of proposed system combining PRTC with VCC, it was essential to investigate the performances of these two cycles. During the experiments, the cycle operation can be shifted by using solenoid valves 1 and 2, which depended on the cooling demand and outdoor air temperature. The system was shifted to the PRTC by opening solenoid valve 1 and shutting off solenoid valve 2. On the other hand, the system was shifted to the VCC by opening solenoid valve 2 and shutting off solenoid valve 1. Besides, the one-way valves 1, 2 and 3 in the loop could ensure the proper flow direction of working fluid in different cycles. The performances of PRTC under different the input frequencies and outdoor air temperatures were investigated, as well as those of VCC under different outdoor air temperatures.
(2)
Fig. 2. Variation of refrigerant pressures at inlet and outlet of LRP.
564
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 3. The air temperature difference between inlet and outlet of indoor coil versus input frequency.
Fig. 6. Variation of refrigerant temperature in indoor coil versus outdoor air temperature.
Fig. 4. Variation of the input power of LRP versus input frequency. Fig. 7. Variation of refrigerant pressure of LRP versus outdoor air temperature.
Fig. 5. Variation of cooling capacity and EER versus input frequency.
Fig. 2 shows the variation of refrigerant pressures at the inlet and outlet of LRP in PRTC. During the experiments, the inlet refrigerant pressure of LRP always maintained at about 380 kPa. However, the outlet refrigerant pressure increased with an increase in input fre-
Fig. 8. Variation of flow resistance of indoor coil, outdoor coil and pipeline versus outdoor air temperature.
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 9. Variation of air temperature in indoor coil versus outdoor air temperature.
Fig. 10. Variation of input power versus outdoor air temperature.
quency. It reached 508.1 kPa when the input frequency was 15 Hz, and 600.0 kPa at input frequency of 40 Hz, respectively. With the increase of input frequency, the input power into the LRP increased as well as its delivery head rose, which led to high outlet pressure and refrigerant flow rate. However, the inlet pressure was relatively stable due to that the resistance in both indoor and outdoor coils and connected copper tubes would also increase under high flow rate. By contrast, the system cooling capacity could be effectively improved under higher input frequency. Therefore, changing input frequency according to variation of indoor environment could be used to adjust system cooling capacity. Fig. 3 shows the air temperature difference between inlet and outlet of the indoor coil under different input frequencies into PRTC. At input frequency of 15 Hz, the air temperature differences reached 10.7 ◦ C, and it reached 10.6 ◦ C at 40 Hz. At the input frequency of 20 Hz, the air temperature difference reached the maximum value of 10.9 ◦ C. It can be seen that the air temperature difference between inlet and outlet of indoor coil increased with an increased input frequency at first, and then decreased after the input frequency increased to a certain value. At the beginning, the increase of input frequency caused the increase of refrigerant mass flow rate, which resulted in the original trend of increase when the input frequency increased from 15 Hz to 20 Hz. However, when the input frequency increased from 20 Hz to 40 Hz, the increase of input
565
frequency caused the increase of delivery head and refrigerant pressure in the system, which further rose the saturation temperature of the refrigerant in the cycle. Besides, the air volume of outdoor fan was fixed in this experiment, which cannot be changed according to the variation of input frequency. Therefore, the heat exchange of the outdoor coil was insufficient, which further caused the rising of refrigerant temperature in the evaporator and decrease of outlet air temperature difference. Fig. 4shows the variation of input power of LRP under different input frequency in PRTC. It can be seen that the input power into LRP was 56.0 W at input frequency of 15 Hz and increased to 110.0 W at 40 Hz. Under high input frequency and power, the system could provide high pressure and cooling capacity, which can satisfy the variation of indoor cooing demand. Fig. 5 shows the variation of cooling capacity and EER with the input frequency in PRTC. It can be clearly seen that the variation trend of cooling capacity was similar to that of air temperature difference. At input frequency of 15 Hz, the cooling capacity reached 2150.7 W, and then reached 2120.6 W at 40 Hz. At input frequency of 20 Hz, the cooling capacity reached its maximum value of 2193.9 W. The original increase of cooling capacity was due to the increase of refrigerant mass flow rate caused by the increase of input frequency. However, the input frequency played a significant role in influencing the refrigerant evaporating temperature and counteracted the positive effect of refrigerant mass flow rate. Therefore, the optimum input frequency was 20 Hz when the cooling capacity reached the maximum value in this experiment. And the later experiments of PRTC were designed to perform under input frequency of 20 Hz. The variation of EER in Fig. 5 shows the obvious tendency of decrease with the increase of input frequency. The EER of PRTC reached 19.9 under input frequency of 15 Hz, and dropped to 18.8 under 20 Hz, then decreased to about 13.1 under 40 Hz. It was interesting that the EER still decreased when the cooling capacity increased, which was because the cooling capacity increased slowly while the power consumption increased relative quickly with an increased input frequency. Besides, the EER always maintained at a relative high level of keeping above 10.0 when the frequency ranged from 15 Hz to 40 Hz. Fig. 6 indicates the variation of refrigerant temperatures at the inlet and outlet of indoor coil with the outdoor air temperature in PRTC under input frequency 20 Hz. When the outdoor air temperature was at −13.7 ◦ C, the refrigerant temperatures at inlet and outlet were −3.3 ◦ C and 3.0 ◦ C, respectively. And thus, the refrigerant temperature difference reached 6.3 ◦ C. Under this outdoor air temperature, the indoor coil had sufficient cooling capacity. With the increase of outdoor air temperature, the refrigerant temperatures at the inlet and outlet of indoor coil were both increased, while the temperature difference decreased, which meant that the sensible heat transfer dropped with the increase of outdoor air temperature. When the outdoor air temperature rose to 0.2 ◦ C and 6.6 ◦ C, the refrigerant temperature difference decreased to 3.9 ◦ C and 1.8 ◦ C, respectively. Finally, the refrigerant temperature difference decreased to 0.4 ◦ C when the outdoor air temperature rose to 14.7 ◦ C. In other words, the PRTC could not cool indoor environment under this outdoor air temperature. Therefore, it should be shifted to the VCC at an appropriate outdoor air temperature considering both the cooling capacity and EER. Fig. 7 shows the variation of refrigerant pressure of LRP under different outdoor air temperatures in PRTC at input frequency 20 Hz. It was seen that the refrigerant pressures at the inlet and outlet reached 324.0 and 546.7 kPa, respectively when the outdoor air temperature was −13.7 ◦ C. The refrigerant pressures at both inlet and outlet of LRP increased with an increase in outdoor air temperature. When the outdoor air temperature rose to 14.7 ◦ C, the inlet and out pressures increased to 760.0 kPa and 837.9 kPa, respectively.
566
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 11. Variation of EER and cooling capacity of the integrated system versus outdoor air temperature.
This was because that the refrigerant temperature was affected by outdoor air temperature, then the increase of refrigerant temperature caused the increase in refrigerant pressure. However, the pressure difference between the inlet and outlet of LRP decreased, which implied a decrease of the total system resistance. Fig. 8 shows the variation of flow resistance of indoor coil, outdoor coil and pipeline with outdoor air temperature in PRTC under input frequency 20 Hz. The total flow resistance reached 222.7 kPa when outdoor air temperature was −13.7 ◦ C, then it dropped to 77.9 kPa when outdoor air temperature was at 14.7 ◦ C. It was worth mentioning that the flow resistance of indoor coil, outdoor coil and pipeline all decreased with an increase in outdoor air temperature. The reason was that the outdoor air temperature greatly affected the refrigerant thermal dynamic parameter and fluid viscosity, and increased outdoor air temperature resulted in the increase of the temperature and pressure of static refrigerant. When outdoor air temperature was at −7.3 ◦ C, the flow resistance of indoor coil and outdoor coil accounted for 41% and 31% of the total, respectively, and the flow resistance of pipeline accounted for 28%. Furthermore, increasing diameter of pipeline between indoor and outdoor coils was favorable for reducing the flow resistance of pipeline and improving the cycle performance. Fig. 9 shows the variation of air temperature at the indoor coil under different outdoor air temperatures in PRTC under input frequency 20 Hz. The air temperature at the inlet of indoor coil maintained at about 25.0 ◦ C, while that at the outlet of indoor coil increased linearly from 9.2 to 19.8 ◦ C. During this process, the air temperature difference dropped linearly from 15.8 ◦ C to 4.9 ◦ C. The decline of the air temperature difference with increased outdoor air temperature meant that the cooling capacity decreased with the increased outdoor air temperature. Fig. 10 shows the variation of input power under different outdoor air temperatures in PRTC under input frequency 20 Hz. The input power into the LRP was 71.7 W when the outdoor air temperature was −13.7 ◦ C and dropped to 59.1 W when outdoor air temperature was 14.7 ◦ C. As analyzed in Fig. 9, the increased outdoor air temperature caused the increase of the temperature and pressure of static refrigerant, which further resulted in the decrease of flow resistance. Therefore, the input power into LRP decreased with the increased outdoor air temperature. Fig. 11 shows the variation of EER and cooling capacity of VCC and PRTC. In VCC, the cooling capacity reached 2668.4 W when outdoor air temperature was 10.6 ◦ C, and dropped to 2567.3 W when outdoor air temperature was 29.8 ◦ C. In addition, EER decreased with the increase of outdoor air temperature, which was 5.1 and
2.8 at outdoor air temperature of 10.6 and 29.8 ◦ C in VCC, respectively. And was 25.7 and 19.1 at outdoor air temperature of 10.6 and 29.8 ◦ C in PRTC, respectively. It can be concluded that EER in PRTC was generally higher than that in VCC, which implied the use of PRTC was significant in energy saving. Despite of the advantages of PRTC, the cooling capacity of PRTC dropped dramatically with the increased outdoor air temperature. The cooling capacity reached 2234.9 W when outdoor air temperature was −0.2 ◦ C, and dropped to 990.8 W when outdoor air temperature increased to 14.7 ◦ C. As a result, the integration of VCC and PRTC was a promising method in improving cooling system. It was seen that the targeted cooling capacity of 2500.0 W, the value of actual cooling capacity of VCC used in this experiment, in PRTC can be obtained when outdoor air temperature was at −3.0 ◦ C. In addition, the cooling capacity can reach 80% and 50% of targeted cooling capacity at outdoor air temperatures of 3.2 ◦ C and 12.0 ◦ C, respectively, and the relative EER of PRTC was higher than that of VCC. It can be noticed that outdoor air temperature can be used as a control parameter to convert system cycle. For example, to ensure the 100% cooling capacity of VCC, the PRTC should be converted to VCC at outdoor air temperature of −3.0 ◦ C. 4. Analysis of feasibility The outdoor air temperature was an important factor that affects the cooling performance and economics of the proposed integrated system. Therefore, the feasibility of this integrated system in different location and climate zones were studied. At the same time, Harbin, Beijing, Wuhan, Guangzhou and Guiyang were chosen as typical cities of severe cool zone, cool zone, hot summer and cool winter zone, hot summer and warm winter zone and temperate zone, respectively. The variation of dry-bulb temperature of five typical cities versus time is shown in Fig. 12. The data of dry-bulb temperature was from Jan 1st to Dec 31st. The outdoor air temperature of Guangzhou and Guiyang always kept being above −3.0 ◦ C. Furthermore, the total hours (Nbin ) that the outdoor air temperature was below −3.0 ◦ C was just 6, so that the integrated system was not suitable to be used in Wuhan. By contrast, Nbin of outdoor air temperature below −3.0 ◦ C in Harbin and Beijing were large enough. More detailed parameters about annual Nbin values for six temperature bins in Harbin and Beijing are given in Tables 5 and 6, respectively. The total Nbin value in Harbin was 3099 h below −3.0 ◦ C, and that in Beijing was 961 h. Therefore, the PRTC was proven to fulfill the targeted cooling capacity for 3099 h in Harbin and 961 h in Beijing. Table 5 Annual Nbin values for six temperature bins in Harbin. Temperature bin(◦ C)
Nbin (h)
Total Nbin (h)
≤−25 (−25, −20] (−20, −15] (−15, −10] (−10, −5] (−5, −3]
100 475 797 816 628 283
3099
Table 6 Annual Nbin values for six temperature bins in Beijing. Temperature bin(◦ C)
Nbin (h)
Total Nbin (h)
≤−15 (−15, −10] (−10, −5] (−5, −3]
0 81 510 370
961
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
567
Fig. 12. Variation of dry-bulb temperature of five typical cities versus time.
Fig. 13 shows the variation of Nbin in PRTC fulfilling 80% and 50% targeted cooling capacity. The total Nbin of outdoor air temperature fulfilling 80% targeted cooling capacity under 3.2 ◦ C in Guangzhou, Harbin, Beijing, Wuhan and Guiyang were 0 h, 4013 h, 2281, 673 h and 464 h, respectively. The total Nbin of outdoor air temperature fulfilling 50% targeted cooling capacity under 12.0 ◦ C was 5289 h in Harbin, 4063 h in Beijing, 3076 h in Wuhan, 779 h in Guangzhou and 3090 h in Guiyang. Therefore, the proposed cooling system was proven to have high feasibility in Harbin and Beijing when PRTC was designed to fulfill both 80% and 50% targeted cooling capacity. In other words, it was worth considering the application of the proposed system in severe cool zone and cool zone.
5. Conclusions An integrated system combining PRTC with VCC was presented in this paper, and the performances of the integrated system were experimentally investigated. In addition, the feasibility of the pro-
Fig. 13. Variation of Nbin in PRTC fulfilling 80% and 50% targeted cooling capacity versus time.
568
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
posed system in different climate zones was finally reported. The main results were summarized as follow: (1) It was obvious that the input frequency affected the cooling capacity and EER in the PRTC. When the outdoor air temperature was −0.2 ◦ C and input frequency 20 Hz, the cooling capacity and EER of the designed PRTC reached 2234.9 W and 19.1, respectively. (2) The increased outdoor air temperature caused a significant reduction of cooling capacity in PRTC. The cooling capacity reached 2234.9 W when outdoor air temperature was −0.2 ◦ C, and dropped to 990.8 W when outdoor air temperature increased to 14.7 ◦ C. (3) Outdoor air temperature can be used as a control parameter for converting system cycle. The targeted cooling capacity of 2500.0 W can be achieved when outdoor air temperature was −3.0 ◦ C in PRTC, and 80% and 50% of the targeted value can be achieved when outdoor air temperature was 3.2 ◦ C and 12.0 ◦ C, respectively. (4) The proposed cooling system was proven to have high feasibility in Harbin and Beijing. The total Nbin of outdoor air temperature below −3.0 ◦ C were 3099 h in Harbin and 961 h in Beijing. Acknowledgement This work has been supported by “National Natural Science Foundation of China (Grant No. 51608146)”. References [1] Y. Ma, G. Ma, S. Zhang, F. Zhou, Cooling performance of a pump-driven two phase cooling system for free cooling in data centers, Appl. Therm. Eng. 95 (2015) 143–149. [2] H. Bulut, M.A. Aktacir, Determination of free cooling potential: A case study for I˙ stanbul, Turkey, Appl. Energy 88 (3) (2011) 680–689. [3] S U. Environmental Protection Agency, Report to Congress on Server and Data Center Energy Efficiency Public Law, Lawrence Berkeley National Laboratory, 2007, 2007. [4] S.W. Ham, J.S. Park, J.W. Jeong, Optimum supply air temperature ranges of various air-side economizers in a modular data center, Appl. Therm. Eng. 77 (2015) 163–179. [5] K.P. Lee, H.L. Chen, Analysis of energy saving potential of air-side free cooling for data centers in worldwide climate zones, Energy Build. 64 (5) (2013) 103–112.
[6] H. Endo, H. Kodama, H. Fukuda, T. Sugimoto, T. Horie, M. Kondo, Effect of climatic conditions on energy consumption in direct fresh-air container data centers, Green Comput. Conf. 6 (2) (2013) 1–10. [7] S.F. Hassan, M. Ali, A. Sajid, U. Perwez, Free cooling investigation of SEECS data center, Energy Procedia 75 (2015) 1249–1254. [8] J. Siriwardana, S. Jayasekara, S.K. Halgamuge, Potential of air-side economizers for data center cooling: a case study for key Australian cities, Appl. Energy 104 (2) (2013) 207–219. [9] Y. Liu, R. Guan, J. Ma, K. Zhang, Research of data center fresh air ventilation cooling system, in: Proceedings of the 8th International Symposium on Heating, Ventilation and Air Conditioning, Springer Berlin Heidelberg, 2014, pp. 299–306. [10] P. Zhang, B. Wang, W. Shi, X. Li, Experimental investigation on two-phase thermosyphon loop with partially liquid-filled downcomer, Appl. Energy 160 (2015) 10–17. [11] P.T. Garrity, J.F. Klausner, R. Mei, A flow boiling microchannel evaporator plate for fuel cell thermal management, Heat Transfer Eng. 28 (10) (2007) 877–884. [12] H. Jouhara, R. Meskimmon, Heat pipe based thermal management systems for energy-efficient data centers, Energy 77 (C) (2014) 265–270. [13] T. Ding, Z. He, T. Hao, Z. Li, Application of separated heat pipe system in data center cooling, Appl. Therm. Eng. 109 (2016) 207–216. [14] J. Choi, J. Jeon, Y. Kim, Cooling performance of a hybrid refrigeration system designed for telecommunication equipment rooms, Appl. Therm. Eng. 27 (11–12) (2007) 2026–2032. [15] R. Hannemann, J. Marsala, M. Pitasi, Pumped liquid multiphase cooling, ASME 2004 International Mechanical Engineering Congress and Exposition (2004) 469–473. [16] J.B. Marcinichen, J.R. Thome, New Novel Green Computer Two-Phase Cooling Cycle: A Model for Its Steady-State Simulation. in International Conference on Efficiency, Cost, Optimization, Simulation and Environmental Impact of Energy Systems – Ecos, 2010. [17] G. Yan, Y. Feng, L. Peng, Experimental analysis of a novel cooling system driven by liquid refrigerant pump and vapor compressor, Int. J. Refrig. 49 (2015) 11–18. [18] Y. Ma, G. Ma, S. Zhang, S. Xu, Experimental investigation on a novel integrated system of vapor compression and pump-driven two phase loop for energy saving in data centers cooling, Energy Convers. Manag. 106 (2015) 194–200. [19] X. Fang, Y. Xu, Z. Zhou, New correlations of single-phase friction factor for turbulent pipe flow and evaluation of existing single-phase friction factor correlations, Nucl. Eng. Des. 241 (3) (2011) 897–902. [20] J.R. García-Cascales, F. Vera-García, J.M. Corberán-Salvador, J. Gonzálvez-Maciá, Assessment of boiling and condensation heat transfer correlations in the modelling of plate heat exchangers, Int. J. Refrig. 30 (6) (2007) 1029–1041. [21] X. Fang, H. Zhang, Y. Xu, X. Su, Evaluation of using two-phase frictional pressure drop correlations for normal gravity to microgravity and reduced gravity, Adv. Space Res. 49 (2) (2012) 351–364. [22] F. Vera-García, J.R. García-Cascales, J.M. Corberán-Salvador, J. Gonzálvez-Maciá, D. Fuentes-Díaz, Assessment of condensation heat transfer correlations in the modelling of fin and tube heat exchangers, Int. J. Refrig. 30 (6) (2007) 1018–1028.
Contents lists available at ScienceDirect
Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild
Experimental investigation of an integrated cooling system driven by both liquid refrigerant pump and vapor compressor Jiankai Dong a,∗ , Yana Lin a , Shiming Deng b , Chao Shen a,∗ , Zhuo Zhang c a b c
School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, China Department of Building Services Engineering, The Hong Kong Polytechnic University, Kowloon, Hong Kong Special Administrative Region China Railway Siyuan Survey and Design Group CO., LTD., China
a r t i c l e
i n f o
Article history: Received 17 July 2017 Received in revised form 13 August 2017 Accepted 16 August 2017 Available online 24 August 2017 Keywords: Integrated cooling system Pumped refrigerant two-phase cycle Vapor compression cycle Liquid refrigerant pump Feasibility analysis
a b s t r a c t The energy consumption of air conditioning system accounts for a significant proportion of the total energy consumed in data center. To reduce the energy consumption of the cooling system, an integrated system, which combines pumped refrigerant two-phase cycle (PRTC) with vapor compression cycle (VCC), is proposed in this paper. The impacts of input frequency and outdoor air temperature on the performances of PRTC, and that of the outdoor air temperature on the performances of VCC have been investigated. It was found that PRTC had higher EER than VCC, but its cooling capacity was limited by outdoor air temperature. In PRTC, the designed cooling capacity of 2500 W could be obtained when the outdoor air temperature was −3.0 ◦ C. In addition, 80% and 50% of the designed cooling capacity can be obtained at ambient temperatures 3.2 ◦ C and 12.0 ◦ C, respectively. Besides, the feasibility of the system in different climate zones in China was demonstrated by analysis of the climate data in Harbin, Beijing, Wuhan, Guangzhou and Guiyang, respectively. Results indicated the proposed system had high feasibility in Harbin and Beijing. The results of this paper can provide valuable insights into the cooling system driven by both liquid refrigerant pump (LRP) and vapor compressor, targeting at reducing the cooling energy consumption and maximizing the economic efficiency of the cooling system in data center. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The rapid development of Information Technology (IT) industry has resulted in a significant increase of data centers [1]. To ensure the reliable operation of the equipment in the data center, it’s necessary to apply the cooling system [2]. In 2011, the energy consumption of data center was more than 100 billion kWh in US [3]. Among these, the energy consumed in cooling system accounted for 30% [4], which means that proper optimization in cooling system would make a vital contribution to energy saving. Vapor compression cooling system, as a conventional cooling system, has been widely applied in the data center no matter what the outdoor air temperature is. However, there is a considerable amount of free cooling energy that could be used when the outdoor air temperature is low. It’s no doubt that using free cooling technology can decrease energy consumption effectively. From the open literature available, several investigations in terms of direct fresh-air cooling were conducted [2,4–9]. Bulut [2]
∗ Corresponding authors. E-mail addresses: [email protected] (J. Dong), [email protected] (C. Shen). http://dx.doi.org/10.1016/j.enbuild.2017.08.038 0378-7788/© 2017 Elsevier B.V. All rights reserved.
carried out an investigation on the performance characteristics of the cooling system in Istanbul. It can be concluded that the temperatures of indoor supplied air and outdoor ambient air affected the amount of the energy saving in the cooling system. According to their research, free cooling technology was not an optimum choice in the period from June to August when outdoor air temperature was equal to or slightly lower than the supply air temperature. Ham [4] simulated the performances of three types of air-side economizers, including direct air-side economizer with ultrasonic humidifier, indirect air-side economizer with heat pipe and indirect air-side economizer with indirect evaporative cooler, under different supply air temperatures and heat exchanger effectiveness in South Korea. The results demonstrated that the optimum supply air temperature was within 18–23 ◦ C. Lee [5] used a dynamic building energy simulation program to examine the potential energy savings of the air-side free cooling technology with differential enthalpy control used in data centers in 17 climate zones. The study showed that the energy saving of free cooling technology decreased by 2.8–8.5% with 2 ◦ C decline of indoor temperature. Endo [6] experimentally investigated a cooling system introducing fresh air directly for a data center located in Tokyo under different climate conditions. The results indicated that the energy
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Nomenclature cp h m Q T
Specific heat, kJ/(kg K) Enthalpy, kJ/kg Mass flow rate of the fluid, kg/s Cooling capacity(W) Temperature (◦ C)
V W
Volumetric flow rate, m3 /s Power (W)
·
Subscripts a Air com Compressor Condenser con eva Evaporator f Fan in Inlet of indoor coil Outlet of indoor coil out PRTC Pumped refrigerant two-phase cycle LRP Liquid refrigerant pump VCC Vapor compression cycle Abbreviations Energy efficiency ratio EER IT Information technology LRP Liquid refrigerant pump Proton exchange membrane PEM TPTL Two-phase thermosyphon loop Greek symbols Fluid density (kg/m3 )
consumption of the cooling system was 20.8% less than that of a conventional air conditioning system. According to the meteorological data of the past sixteen years in Islamabad, Pakistan, Hassan [7] carried out an investigation on the potential application for free cooling technology. The results showed that significant energy savings through free cooling can be achieved during the months of December, January, and February. Siriwardana [8] made an investigation on using air-side economizers to introduce outside air based on analyzing the hourly temperature and humidity data over past 12 years in Australia, and determined that the outside air cooling energy could be used for 5500 h per year in Tasmania and Southern Australian cities, such as Melbourne and Adelaide. Liu [9] concluded that full fresh air ventilation can save 40% of the air conditioning energy conservation and make power usage effectiveness (PUE) lower than 1.4. However, most of these investigations mentioned above actually focused on the energy saving in direct fresh-air cooling, while some focused on air-side economizers to analyze optimum supply air temperature to evaluate the potential usage of direct fresh-air cooling. Although the direct fresh-air cooling system could achieve enormous energy conservation, the utilization of direct fresh-air cooling was limited due to its performance fluctuation, which was resulted from the uncontrollability of outdoor air. Besides, the indoor air quality was badly affected if the outdoor air was supplied directly without purification treatment. Furthermore, heating the supplied air was necessary to avoid fogging or dewing when the outdoor air temperature was low, which caused an increase in the initial cost of direct fresh-air cooling system. Therefore, indirect fresh-air cooling attracts more and more attention because of its relatively high cooling capacity and controllability. The studies on indirect fresh-air cooling were mainly divided into natural cooling and mechanical cooling. There were some researchers concentrating on the study of natural cooling.
561
Zhang [10] established a visual experimental setup to investigate the performances of a two-phase thermosiphon loop (TPTL), in terms of the variation of temperature difference, liquid charge, height difference, and circulation flow resistance, etc. The results showed that larger height difference was not always advantageous to the performance of TPTL. Garrity [11] developed a flow boiling micro-channel cooling plate which can be easily inserted into the stack of a proton exchange membrane (PEM) fuel cell for thermal control. Furthermore, some researchers paid special attention to heat pipe. Jouhara [12] studied the thermal management based on heat pipe. The results revealed that 75% of potential energy saving can be achieved. Ding [13] made a conclusion that the total entransy dissipation of the heat pipe system was 48.3% lower than that of a traditional air conditioning system by analyzing the heat transfer process. On the other hand, some researchers showed their interest in mechanical cooling. Choi [14] investigated an integrated system designed for telecommunication equipment rooms. It can be seen that the optimum outdoor air temperature was 8.3 ◦ C, which was the control parameter to convert the cycle using ethylene glycol as refrigerant into VCC. Considering the driving force into the indirect free cooling technology, some researchers showed a significant concern for applying pump into cooling system. Hannemann[15] applied pump into liquid multiphase cooling system. According to this study, the mass flow rate, size of condenser and power consumed by the pump in water-circulating single-phase cooling system were 4.6 times, 2 times and 10 times as those in liquid multiphase cooling system, respectively. Marcinichen [16] carried out a further investigation on the thermal performances of refrigerated cooling of microprocessors with micro-evaporation. The results suggested that the energy consumption of vapor compression cooling system was far higher than that of pump-driven cooling system. Yan [17] introduced a novel energy saving cooling system combining traditional VCC with pumped liquid two-phase cooling cycle, and evaluated the system performance under various operation conditions. The results showed that the proper shift temperature between these two operations was about −5 ◦ C according to the system EER and cooling capacity. Ma [18] proposed a novel integrated cooling system of vapor compression and pump-driven two-phase loop for energy saving in data centers. And the results showed that the maximum cooling EER of the integrated system reached up to 29.71, and the proper shift temperature of the system using 3 HP compressor and 1 HP compressor was 5 ◦ C and 15 ◦ C, respectively. As mentioned previously, a considerable amount of literature related to free cooling technology has been carried out. Some focused on direct fresh-air cooling, while other paid attention to indirect fresh-air cooling. Among the studies for the indirect freshair cooling, the integrated system, combining vapor compression cooling with pump-driven two-phase cooling, has obvious energy saving, controllability, and can be used in both micro-channel evaporator and data centers to effectively use the natural cooling energy and reduce the energy consumption. Although Yan [17] and Ma [18] introduced the integrated systems combining vapor compression with pump-driven two-phase cooling cycle, both did not indicate the correspondence of LRP and compressor for generating certain cooling capacity. Simultaneously, the used LRP in these two experiments were both constant frequency. However, the cooling capacity of the LRP varied obviously with the change of outdoor air temperature under fixed condenser and evaporator. And thus, changing the input frequency into the LRP according to the variation of outdoor environment would improve its cooling performance. In this paper, an integrated system driven by vapor compressor and LRP is firstly developed. Then the operation performances of the system are experimentally evaluated under two operation cycles. This is followed by presenting the experimental results and detailed
562
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 1. Experimental setup and arrangement of measuring points.
Table 1 Specification of the split air-conditioner.
Table 2 Specification of measuring points.
Parameters
Value
Frequency conversion or not Refrigerant Cooling capacity Input power in cooling mode Volumetric flow rate Size of condenser Size of evaporator Copper tube diameter in condenser Copper tube diameter in evaporator
not R22 3300 955 600 780 × 250 × 650 795 × 194 × 273 8 12
Unit
Testing items
Notation
Measuring objects
Temperature
T1, T2 T3, T4 T5, T6 T7, T8 T9 T10 T11 T12 T13 T14 T15 P1 P2 P3 P4 P5 P6 P7
air at inlet of condenser air at outlet of condenser air at inlet of evaporator air at outlet of evaporator Refrigerant at inlet of condenser Refrigerant at outlet of condenser Refrigerant at inlet of LRP Refrigerant at outlet of LRP Refrigerant at inlet of evaporator Refrigerant at outlet of evaporator Refrigerant at outlet of compressor Refrigerant at inlet of condenser Refrigerant at outlet of condenser Refrigerant at inlet of LRP Refrigerant at outlet of LRP Refrigerant at inlet of evaporator Refrigerant at outlet of evaporator Refrigerant at outlet of compressor
W W m3 /h mm mm mm mm
discussion. Finally, conclusions are made following the feasibility analysis. The results of this paper can provide valuable insights into the cooling system driven by both LRP and vapor compressor, targeting at reducing the cooling energy consumption and maximizing the economic efficiency of the cooling system in data center.
Pressure
2. Experimental set-up and procedures 2.1. Experimental setup An experimental setup was firstly established, and it was made up of a psychrometric chamber and the integrated cooling system. The psychrometric chamber was of size of 3.2 m (L) × 3.0 m (W) × 2.4 m (H), in which the air could be maintained at temperature of 25.0 ± 1.0 ◦ C and relative humidity of 20%–30% by using a separated air conditioning system, a 6.5 kW electric heater, a humidifier of 3.0 kg/h water evaporation. At the same time, the outdoor coil was placed in the outdoor natural environment in Harbin, China during experiments. The air temperature in winter in Harbin was low enough for the experiments, and the lowest temperature could be as low as −35.0 ◦ C. As shown in Fig. 1, the integrated cooling system combining PRTC with VCC mainly consisted of an indoor coil, an outdoor coil, a compressor, a LRP, a reservoir, a capillary, and air fans, etc. The integrated cooling system were modified on an ordinary split air-conditioner with a nominal cooling capacity of 3300 W. The detailed specification of the split air-conditioner is given in Table 1. The reservoir with volume of 5.0 L was installed between the outdoor coil and LRP. The installation height of the reservoir was 0.5 m
higher than that of LRP to ensure the effective operation of the LRP. The volumetric flow rate of the indoor air fan was 600 m3 /h. The copper tube used between outdoor coil and indoor coil was of length of 3.0 m. Besides, R22 was chosen as the refrigerant. Table 2 gives the specification of measuring points in the system. During experiments, platinum resistance thermometers (PT1000, ±0.1 ◦ C accuracy) were used to measure the temperature at the inlet and outlet of evaporator, condenser and other components. Refrigeration pressures were measured by pressure transmitters (at ±0.1% accuracy). The input powers into LRP and compressor were measured by power transducer (at ±0.2% accuracy). All the measurement data was recorded and stored by Agilent 34980A series data acquisition system with an interval of 10 s.
2.2. Selection of liquid refrigerant pump LRP was a key component in the successful development of this integrated system. To select a matched pump, the values of mass flow rate and delivery head were both parameters needed to be consider.
J. Dong et al. / Energy and Buildings 154 (2017) 560–568 Table 3 Calculated parameters for choosing the LRP. refrigerant saturation pressure refrigerant state at inlet of evaporator refrigerant state at outlet of evaporator refrigerant state at outlet of condenser refrigerant state at inlet of condenser the diameter of the connected copper tube refrigerant state in the connected copper tube
563
The PRTC operated under two conditions. The performances of PRTC under the conditions of outdoor air temperature 0 ◦ C, input frequencies into the LRP 15, 20, 30 and 40 Hz, were firstly investigated. Then the performances of PRTC under the fixed input frequency into the LRP 20 Hz, outdoor air temperature ranging from −15.0 to 15.0 ◦ C were conducted. During operation of VCC, the inside condition in the psychrometric chamber were maintained at the same values with those in the operation of PRTC. but the outdoor air temperature ranged from 10.0 to 30.0 ◦ C.
497.59 kPa saturated-liquid two-phase mixture with 0.5 dryness saturated-liquid two-phase mixture with 0.5 dryness 10.0 mm 0.5 of dryness
2.4. Data reduction
Table 4 Specification of the LRP. Parameters
Value
Frequency converter or not Volumetric flow rate Delivery head Shaft power Power supply Measurement
Yes 0.04–0.2 22–100 0.2 380 V/3P&220 V/1N, 50 Hz 312 × 155 × 165
Unit m3 /h m kW mm
Firstly, the mass flow rate of the working fluid was evaluated by Eq. (1) under the calculated parameters given in Table 3. QPRTC = mPRTC (h2 − h1 )
(1)
where h2 is determined by the refrigerant pressure and temperature at the outlet of the evaporator. h1 by the refrigerant pressure and temperature at the inlet of the evaporator. The value of QPRTC is set as 3300W, equals to the nominal cooling capacity of the split air-conditioner. Therefore, the value of mass flow rate of the working refrigerant was 0.321 kg/s. Secondly, to choose the delivery head, the total flow resistance of PRTC should be calculated, which was made up of the resistance of evaporator, condenser and connected copper tube. The equations for calculating the flow resistance of single-phase working refrigerant can be found in Reference [19], while those for calculating the flow resistance of two-phase working refrigerant can be found in References [20–22]. Therefore, the total flow resistance of PRTC was 168.04 kPa including the flow resistance in evaporator of 78.45 kPa, condenser of 49.78 kPa, and copper tube of 39.81 kPa. Finally, based on the calculation results, the values of reasonable mass flow rate and delivery head were 0.321 kg/s (0.0901 m3 /h) and 168.04 kPa (13.4 m), respectively. However, there was no LRP satisfying simultaneously these two precise parameters. Therefore, a LRP with mass flow rate of 0.2 m3 /h and delivery head of 22.0 m was chosen. The detailed specification of the LRP is shown in Table 4.
During the experiments, the data of cooling capacity and EER were calculated. The cooling capacities of both VCC and PRTC can be calculated by Eq. (2):
·
Q = ma hout,a − hin,a = a V cp Tout,a − Tin,a
where Q is the cooling capacity, W; ma the air mass flow rate across indoor coil, kg/s; hout,a the enthalpy of air at outlet of indoor coil, kJ/kg; hin,a the enthalpy of air at inlet of indoor coil, kJ/kg; a the air ·
density, kg/m3 ; V the volumetric flow rate of air, m3 /s; cp the specific heat of air, kJ/(kg K); Tout,a air temperature at outlet of indoor coil, ◦ C; Tin,a air temperature at inlet of indoor coil, ◦ C. The EER of PRTC can be calculated by Eq. (3):
EERPRTC = QPRTC / WLRP + Weva,f + Wcon,f
(3)
Where EERPRTC is the energy efficiency ratio of PRTC; QPRTC the cooling capacity of PRTC, W; WLRP the power input of the liquid refrigerant pump, W; Weva,f the power input of the indoor coil fan, W; Wcon,f the power input of the outdoor coil fan, W. The EER of VCC can be calculated by Eq. (4):
EERVCC = QVCC / Wcom + Weva,f + Wcon,f
(4)
Where EERVCC is the energy efficiency ratio of VCC; QVCC the cooling capacity of VCC, W; Wcom the power input of the compressor, W. 3. Results and discussion The influences of outdoor air temperature on performances of PRTC and VCC, as well as the influences of input frequency on performances of PRTC, are shown in Figs. 2–10.
2.3. Experimental conditions and procedures For determining the cycle operation and improving the performance of proposed system combining PRTC with VCC, it was essential to investigate the performances of these two cycles. During the experiments, the cycle operation can be shifted by using solenoid valves 1 and 2, which depended on the cooling demand and outdoor air temperature. The system was shifted to the PRTC by opening solenoid valve 1 and shutting off solenoid valve 2. On the other hand, the system was shifted to the VCC by opening solenoid valve 2 and shutting off solenoid valve 1. Besides, the one-way valves 1, 2 and 3 in the loop could ensure the proper flow direction of working fluid in different cycles. The performances of PRTC under different the input frequencies and outdoor air temperatures were investigated, as well as those of VCC under different outdoor air temperatures.
(2)
Fig. 2. Variation of refrigerant pressures at inlet and outlet of LRP.
564
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 3. The air temperature difference between inlet and outlet of indoor coil versus input frequency.
Fig. 6. Variation of refrigerant temperature in indoor coil versus outdoor air temperature.
Fig. 4. Variation of the input power of LRP versus input frequency. Fig. 7. Variation of refrigerant pressure of LRP versus outdoor air temperature.
Fig. 5. Variation of cooling capacity and EER versus input frequency.
Fig. 2 shows the variation of refrigerant pressures at the inlet and outlet of LRP in PRTC. During the experiments, the inlet refrigerant pressure of LRP always maintained at about 380 kPa. However, the outlet refrigerant pressure increased with an increase in input fre-
Fig. 8. Variation of flow resistance of indoor coil, outdoor coil and pipeline versus outdoor air temperature.
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 9. Variation of air temperature in indoor coil versus outdoor air temperature.
Fig. 10. Variation of input power versus outdoor air temperature.
quency. It reached 508.1 kPa when the input frequency was 15 Hz, and 600.0 kPa at input frequency of 40 Hz, respectively. With the increase of input frequency, the input power into the LRP increased as well as its delivery head rose, which led to high outlet pressure and refrigerant flow rate. However, the inlet pressure was relatively stable due to that the resistance in both indoor and outdoor coils and connected copper tubes would also increase under high flow rate. By contrast, the system cooling capacity could be effectively improved under higher input frequency. Therefore, changing input frequency according to variation of indoor environment could be used to adjust system cooling capacity. Fig. 3 shows the air temperature difference between inlet and outlet of the indoor coil under different input frequencies into PRTC. At input frequency of 15 Hz, the air temperature differences reached 10.7 ◦ C, and it reached 10.6 ◦ C at 40 Hz. At the input frequency of 20 Hz, the air temperature difference reached the maximum value of 10.9 ◦ C. It can be seen that the air temperature difference between inlet and outlet of indoor coil increased with an increased input frequency at first, and then decreased after the input frequency increased to a certain value. At the beginning, the increase of input frequency caused the increase of refrigerant mass flow rate, which resulted in the original trend of increase when the input frequency increased from 15 Hz to 20 Hz. However, when the input frequency increased from 20 Hz to 40 Hz, the increase of input
565
frequency caused the increase of delivery head and refrigerant pressure in the system, which further rose the saturation temperature of the refrigerant in the cycle. Besides, the air volume of outdoor fan was fixed in this experiment, which cannot be changed according to the variation of input frequency. Therefore, the heat exchange of the outdoor coil was insufficient, which further caused the rising of refrigerant temperature in the evaporator and decrease of outlet air temperature difference. Fig. 4shows the variation of input power of LRP under different input frequency in PRTC. It can be seen that the input power into LRP was 56.0 W at input frequency of 15 Hz and increased to 110.0 W at 40 Hz. Under high input frequency and power, the system could provide high pressure and cooling capacity, which can satisfy the variation of indoor cooing demand. Fig. 5 shows the variation of cooling capacity and EER with the input frequency in PRTC. It can be clearly seen that the variation trend of cooling capacity was similar to that of air temperature difference. At input frequency of 15 Hz, the cooling capacity reached 2150.7 W, and then reached 2120.6 W at 40 Hz. At input frequency of 20 Hz, the cooling capacity reached its maximum value of 2193.9 W. The original increase of cooling capacity was due to the increase of refrigerant mass flow rate caused by the increase of input frequency. However, the input frequency played a significant role in influencing the refrigerant evaporating temperature and counteracted the positive effect of refrigerant mass flow rate. Therefore, the optimum input frequency was 20 Hz when the cooling capacity reached the maximum value in this experiment. And the later experiments of PRTC were designed to perform under input frequency of 20 Hz. The variation of EER in Fig. 5 shows the obvious tendency of decrease with the increase of input frequency. The EER of PRTC reached 19.9 under input frequency of 15 Hz, and dropped to 18.8 under 20 Hz, then decreased to about 13.1 under 40 Hz. It was interesting that the EER still decreased when the cooling capacity increased, which was because the cooling capacity increased slowly while the power consumption increased relative quickly with an increased input frequency. Besides, the EER always maintained at a relative high level of keeping above 10.0 when the frequency ranged from 15 Hz to 40 Hz. Fig. 6 indicates the variation of refrigerant temperatures at the inlet and outlet of indoor coil with the outdoor air temperature in PRTC under input frequency 20 Hz. When the outdoor air temperature was at −13.7 ◦ C, the refrigerant temperatures at inlet and outlet were −3.3 ◦ C and 3.0 ◦ C, respectively. And thus, the refrigerant temperature difference reached 6.3 ◦ C. Under this outdoor air temperature, the indoor coil had sufficient cooling capacity. With the increase of outdoor air temperature, the refrigerant temperatures at the inlet and outlet of indoor coil were both increased, while the temperature difference decreased, which meant that the sensible heat transfer dropped with the increase of outdoor air temperature. When the outdoor air temperature rose to 0.2 ◦ C and 6.6 ◦ C, the refrigerant temperature difference decreased to 3.9 ◦ C and 1.8 ◦ C, respectively. Finally, the refrigerant temperature difference decreased to 0.4 ◦ C when the outdoor air temperature rose to 14.7 ◦ C. In other words, the PRTC could not cool indoor environment under this outdoor air temperature. Therefore, it should be shifted to the VCC at an appropriate outdoor air temperature considering both the cooling capacity and EER. Fig. 7 shows the variation of refrigerant pressure of LRP under different outdoor air temperatures in PRTC at input frequency 20 Hz. It was seen that the refrigerant pressures at the inlet and outlet reached 324.0 and 546.7 kPa, respectively when the outdoor air temperature was −13.7 ◦ C. The refrigerant pressures at both inlet and outlet of LRP increased with an increase in outdoor air temperature. When the outdoor air temperature rose to 14.7 ◦ C, the inlet and out pressures increased to 760.0 kPa and 837.9 kPa, respectively.
566
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
Fig. 11. Variation of EER and cooling capacity of the integrated system versus outdoor air temperature.
This was because that the refrigerant temperature was affected by outdoor air temperature, then the increase of refrigerant temperature caused the increase in refrigerant pressure. However, the pressure difference between the inlet and outlet of LRP decreased, which implied a decrease of the total system resistance. Fig. 8 shows the variation of flow resistance of indoor coil, outdoor coil and pipeline with outdoor air temperature in PRTC under input frequency 20 Hz. The total flow resistance reached 222.7 kPa when outdoor air temperature was −13.7 ◦ C, then it dropped to 77.9 kPa when outdoor air temperature was at 14.7 ◦ C. It was worth mentioning that the flow resistance of indoor coil, outdoor coil and pipeline all decreased with an increase in outdoor air temperature. The reason was that the outdoor air temperature greatly affected the refrigerant thermal dynamic parameter and fluid viscosity, and increased outdoor air temperature resulted in the increase of the temperature and pressure of static refrigerant. When outdoor air temperature was at −7.3 ◦ C, the flow resistance of indoor coil and outdoor coil accounted for 41% and 31% of the total, respectively, and the flow resistance of pipeline accounted for 28%. Furthermore, increasing diameter of pipeline between indoor and outdoor coils was favorable for reducing the flow resistance of pipeline and improving the cycle performance. Fig. 9 shows the variation of air temperature at the indoor coil under different outdoor air temperatures in PRTC under input frequency 20 Hz. The air temperature at the inlet of indoor coil maintained at about 25.0 ◦ C, while that at the outlet of indoor coil increased linearly from 9.2 to 19.8 ◦ C. During this process, the air temperature difference dropped linearly from 15.8 ◦ C to 4.9 ◦ C. The decline of the air temperature difference with increased outdoor air temperature meant that the cooling capacity decreased with the increased outdoor air temperature. Fig. 10 shows the variation of input power under different outdoor air temperatures in PRTC under input frequency 20 Hz. The input power into the LRP was 71.7 W when the outdoor air temperature was −13.7 ◦ C and dropped to 59.1 W when outdoor air temperature was 14.7 ◦ C. As analyzed in Fig. 9, the increased outdoor air temperature caused the increase of the temperature and pressure of static refrigerant, which further resulted in the decrease of flow resistance. Therefore, the input power into LRP decreased with the increased outdoor air temperature. Fig. 11 shows the variation of EER and cooling capacity of VCC and PRTC. In VCC, the cooling capacity reached 2668.4 W when outdoor air temperature was 10.6 ◦ C, and dropped to 2567.3 W when outdoor air temperature was 29.8 ◦ C. In addition, EER decreased with the increase of outdoor air temperature, which was 5.1 and
2.8 at outdoor air temperature of 10.6 and 29.8 ◦ C in VCC, respectively. And was 25.7 and 19.1 at outdoor air temperature of 10.6 and 29.8 ◦ C in PRTC, respectively. It can be concluded that EER in PRTC was generally higher than that in VCC, which implied the use of PRTC was significant in energy saving. Despite of the advantages of PRTC, the cooling capacity of PRTC dropped dramatically with the increased outdoor air temperature. The cooling capacity reached 2234.9 W when outdoor air temperature was −0.2 ◦ C, and dropped to 990.8 W when outdoor air temperature increased to 14.7 ◦ C. As a result, the integration of VCC and PRTC was a promising method in improving cooling system. It was seen that the targeted cooling capacity of 2500.0 W, the value of actual cooling capacity of VCC used in this experiment, in PRTC can be obtained when outdoor air temperature was at −3.0 ◦ C. In addition, the cooling capacity can reach 80% and 50% of targeted cooling capacity at outdoor air temperatures of 3.2 ◦ C and 12.0 ◦ C, respectively, and the relative EER of PRTC was higher than that of VCC. It can be noticed that outdoor air temperature can be used as a control parameter to convert system cycle. For example, to ensure the 100% cooling capacity of VCC, the PRTC should be converted to VCC at outdoor air temperature of −3.0 ◦ C. 4. Analysis of feasibility The outdoor air temperature was an important factor that affects the cooling performance and economics of the proposed integrated system. Therefore, the feasibility of this integrated system in different location and climate zones were studied. At the same time, Harbin, Beijing, Wuhan, Guangzhou and Guiyang were chosen as typical cities of severe cool zone, cool zone, hot summer and cool winter zone, hot summer and warm winter zone and temperate zone, respectively. The variation of dry-bulb temperature of five typical cities versus time is shown in Fig. 12. The data of dry-bulb temperature was from Jan 1st to Dec 31st. The outdoor air temperature of Guangzhou and Guiyang always kept being above −3.0 ◦ C. Furthermore, the total hours (Nbin ) that the outdoor air temperature was below −3.0 ◦ C was just 6, so that the integrated system was not suitable to be used in Wuhan. By contrast, Nbin of outdoor air temperature below −3.0 ◦ C in Harbin and Beijing were large enough. More detailed parameters about annual Nbin values for six temperature bins in Harbin and Beijing are given in Tables 5 and 6, respectively. The total Nbin value in Harbin was 3099 h below −3.0 ◦ C, and that in Beijing was 961 h. Therefore, the PRTC was proven to fulfill the targeted cooling capacity for 3099 h in Harbin and 961 h in Beijing. Table 5 Annual Nbin values for six temperature bins in Harbin. Temperature bin(◦ C)
Nbin (h)
Total Nbin (h)
≤−25 (−25, −20] (−20, −15] (−15, −10] (−10, −5] (−5, −3]
100 475 797 816 628 283
3099
Table 6 Annual Nbin values for six temperature bins in Beijing. Temperature bin(◦ C)
Nbin (h)
Total Nbin (h)
≤−15 (−15, −10] (−10, −5] (−5, −3]
0 81 510 370
961
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
567
Fig. 12. Variation of dry-bulb temperature of five typical cities versus time.
Fig. 13 shows the variation of Nbin in PRTC fulfilling 80% and 50% targeted cooling capacity. The total Nbin of outdoor air temperature fulfilling 80% targeted cooling capacity under 3.2 ◦ C in Guangzhou, Harbin, Beijing, Wuhan and Guiyang were 0 h, 4013 h, 2281, 673 h and 464 h, respectively. The total Nbin of outdoor air temperature fulfilling 50% targeted cooling capacity under 12.0 ◦ C was 5289 h in Harbin, 4063 h in Beijing, 3076 h in Wuhan, 779 h in Guangzhou and 3090 h in Guiyang. Therefore, the proposed cooling system was proven to have high feasibility in Harbin and Beijing when PRTC was designed to fulfill both 80% and 50% targeted cooling capacity. In other words, it was worth considering the application of the proposed system in severe cool zone and cool zone.
5. Conclusions An integrated system combining PRTC with VCC was presented in this paper, and the performances of the integrated system were experimentally investigated. In addition, the feasibility of the pro-
Fig. 13. Variation of Nbin in PRTC fulfilling 80% and 50% targeted cooling capacity versus time.
568
J. Dong et al. / Energy and Buildings 154 (2017) 560–568
posed system in different climate zones was finally reported. The main results were summarized as follow: (1) It was obvious that the input frequency affected the cooling capacity and EER in the PRTC. When the outdoor air temperature was −0.2 ◦ C and input frequency 20 Hz, the cooling capacity and EER of the designed PRTC reached 2234.9 W and 19.1, respectively. (2) The increased outdoor air temperature caused a significant reduction of cooling capacity in PRTC. The cooling capacity reached 2234.9 W when outdoor air temperature was −0.2 ◦ C, and dropped to 990.8 W when outdoor air temperature increased to 14.7 ◦ C. (3) Outdoor air temperature can be used as a control parameter for converting system cycle. The targeted cooling capacity of 2500.0 W can be achieved when outdoor air temperature was −3.0 ◦ C in PRTC, and 80% and 50% of the targeted value can be achieved when outdoor air temperature was 3.2 ◦ C and 12.0 ◦ C, respectively. (4) The proposed cooling system was proven to have high feasibility in Harbin and Beijing. The total Nbin of outdoor air temperature below −3.0 ◦ C were 3099 h in Harbin and 961 h in Beijing. Acknowledgement This work has been supported by “National Natural Science Foundation of China (Grant No. 51608146)”. References [1] Y. Ma, G. Ma, S. Zhang, F. Zhou, Cooling performance of a pump-driven two phase cooling system for free cooling in data centers, Appl. Therm. Eng. 95 (2015) 143–149. [2] H. Bulut, M.A. Aktacir, Determination of free cooling potential: A case study for I˙ stanbul, Turkey, Appl. Energy 88 (3) (2011) 680–689. [3] S U. Environmental Protection Agency, Report to Congress on Server and Data Center Energy Efficiency Public Law, Lawrence Berkeley National Laboratory, 2007, 2007. [4] S.W. Ham, J.S. Park, J.W. Jeong, Optimum supply air temperature ranges of various air-side economizers in a modular data center, Appl. Therm. Eng. 77 (2015) 163–179. [5] K.P. Lee, H.L. Chen, Analysis of energy saving potential of air-side free cooling for data centers in worldwide climate zones, Energy Build. 64 (5) (2013) 103–112.
[6] H. Endo, H. Kodama, H. Fukuda, T. Sugimoto, T. Horie, M. Kondo, Effect of climatic conditions on energy consumption in direct fresh-air container data centers, Green Comput. Conf. 6 (2) (2013) 1–10. [7] S.F. Hassan, M. Ali, A. Sajid, U. Perwez, Free cooling investigation of SEECS data center, Energy Procedia 75 (2015) 1249–1254. [8] J. Siriwardana, S. Jayasekara, S.K. Halgamuge, Potential of air-side economizers for data center cooling: a case study for key Australian cities, Appl. Energy 104 (2) (2013) 207–219. [9] Y. Liu, R. Guan, J. Ma, K. Zhang, Research of data center fresh air ventilation cooling system, in: Proceedings of the 8th International Symposium on Heating, Ventilation and Air Conditioning, Springer Berlin Heidelberg, 2014, pp. 299–306. [10] P. Zhang, B. Wang, W. Shi, X. Li, Experimental investigation on two-phase thermosyphon loop with partially liquid-filled downcomer, Appl. Energy 160 (2015) 10–17. [11] P.T. Garrity, J.F. Klausner, R. Mei, A flow boiling microchannel evaporator plate for fuel cell thermal management, Heat Transfer Eng. 28 (10) (2007) 877–884. [12] H. Jouhara, R. Meskimmon, Heat pipe based thermal management systems for energy-efficient data centers, Energy 77 (C) (2014) 265–270. [13] T. Ding, Z. He, T. Hao, Z. Li, Application of separated heat pipe system in data center cooling, Appl. Therm. Eng. 109 (2016) 207–216. [14] J. Choi, J. Jeon, Y. Kim, Cooling performance of a hybrid refrigeration system designed for telecommunication equipment rooms, Appl. Therm. Eng. 27 (11–12) (2007) 2026–2032. [15] R. Hannemann, J. Marsala, M. Pitasi, Pumped liquid multiphase cooling, ASME 2004 International Mechanical Engineering Congress and Exposition (2004) 469–473. [16] J.B. Marcinichen, J.R. Thome, New Novel Green Computer Two-Phase Cooling Cycle: A Model for Its Steady-State Simulation. in International Conference on Efficiency, Cost, Optimization, Simulation and Environmental Impact of Energy Systems – Ecos, 2010. [17] G. Yan, Y. Feng, L. Peng, Experimental analysis of a novel cooling system driven by liquid refrigerant pump and vapor compressor, Int. J. Refrig. 49 (2015) 11–18. [18] Y. Ma, G. Ma, S. Zhang, S. Xu, Experimental investigation on a novel integrated system of vapor compression and pump-driven two phase loop for energy saving in data centers cooling, Energy Convers. Manag. 106 (2015) 194–200. [19] X. Fang, Y. Xu, Z. Zhou, New correlations of single-phase friction factor for turbulent pipe flow and evaluation of existing single-phase friction factor correlations, Nucl. Eng. Des. 241 (3) (2011) 897–902. [20] J.R. García-Cascales, F. Vera-García, J.M. Corberán-Salvador, J. Gonzálvez-Maciá, Assessment of boiling and condensation heat transfer correlations in the modelling of plate heat exchangers, Int. J. Refrig. 30 (6) (2007) 1029–1041. [21] X. Fang, H. Zhang, Y. Xu, X. Su, Evaluation of using two-phase frictional pressure drop correlations for normal gravity to microgravity and reduced gravity, Adv. Space Res. 49 (2) (2012) 351–364. [22] F. Vera-García, J.R. García-Cascales, J.M. Corberán-Salvador, J. Gonzálvez-Maciá, D. Fuentes-Díaz, Assessment of condensation heat transfer correlations in the modelling of fin and tube heat exchangers, Int. J. Refrig. 30 (6) (2007) 1018–1028.