Development of an outdoor air cooling-type air-cooled package air conditioner for data centers.
In recent years, the amount of energy consumed by data centers has tended to increase, and the amount of energy consumed by air-conditioning equipment, not just the housed Information and Communication Technology (ICT) devices, must be reduced. Air conditioning in data centers must be operated year-round due to the extremely large amounts of heat emitted from the ICT devices, such as housed blade servers and network appliances. Therefore, in order to decrease year-round energy consumption, it is important to wisely use low-temperature outdoor air during cool seasons, such as mid-to-late fall and early spring, and cold seasons like winter.
As for outdoor-air cooling methods which use low-temperature outdoor air, there is the direct outdoor air cooling method (air-side economizer), which brings outdoor air directly into the room. However, as this method requires avoidance of electrical plate corrosion by impurities such as dust and particles of sea salt, as well as variable humidity adjustments corresponding to the volume of outdoor air, there are cases where the operating costs and energy consumption increase. Also, the large openings installed in the outer wall for bringing in air are large restricting condition in building design and become an obstacle for their introduction.
Therefore, in order to solve the problems with the air-side economizers, we examined air-conditioning systems that can use low-temperature outdoor air without directly introducing outdoor air into the data center. At data centers in Japan, there are numerous cases of air-cooled package air conditioners being used that allow for easy consecutive expansion along with an increase in ICT facilities. For this reason, we focused our attention on package-type air conditioners and worked on their development. The new air conditioner we developed combines a compression cycle and a cycle that stops the compressor and circulates coolant with a pump (henceforth, "pump cycle") and operates by switching the cycle based on numerous conditions, such as the outdoor air temperature. In this report, first, we provide characteristics of the new air conditioner we developed. Next, we discuss the tests performed in the test room and evaluate these tests. Finally, we discuss the results of the reduction in the amount of energy consumed by the air conditioner we developed.
Characteristics of an Outdoor Air Cooling-Type Air-Cooled Packaged Air Conditioner
Table 1 and Figure 1 show specifications of the outdoor air cooling-type air-cooled package air conditioner for data centers we developed. This air-conditioner system consists of three units: an indoor unit, an outdoor unit, and a pump unit. The greatest feature of this air conditioner is that, in addition to the original compression cycle, it takes in cold and heat from the outdoor air during the refrigerant cycle from the heat exchanger inside the outdoor unit and combines it with a pump cycle that transports it into the room with a pump. When the outdoor temperature is sufficiently lower than the indoor temperature, the air conditioner operates in the pump cycle, and at all other times, it automatically switches cycles so that it operates in the compression cycle. Depending on operating conditions, there are times when the energy consumed by the refrigerant pump is about one tenth of that consumed by the compressor so the amount of electrical power consumed can be greatly reduced. In the aforementioned direct outdoor air-conditioning method (air-side economizers), a separate seasonal air-conditioning system becomes necessary during the summer season when outdoor air cannot be drawn in. However, in the method we have developed, this can be addressed just by switching the routing of the pipes.
Table 1. Specifications of the New Air Conditioner Rated cooling capacity 45.0 kW * (1) * (2) (Sensible heat capacity) Rated power 16.2 kW * (1) / 5.0 kW * consumption (2) Air volume of indoor 240 [m.sup.3]/min unit Refrigerant R410A Size Indoor unit W1295 x D900 x H1950 mm (W51.0 x D35.4 x H76.8 in.) Outdoor unit W1350 x D900 x H1980 mm (W53.1 x D35.4 x H78.0 in.) Pump unit W1100 x D580 x H550 mm (W43.3 x D22.8 x H21.7 in.) * (1) Values of the following conditions at compression cycle: Indoor unit intake air temperature is 27 (80.6) (CDB [[degrees]FDB])/19 (66.2) ([degrees]CWB [[degrees]FWB]), outdoor unit intake air temperature is 35 (95) (CDB [[degrees]FDB]), and refrigerant pipe length is 7.5 (24.6) m (ft). * (2) Values of the following conditions at pump cycle: Indoor unit intake air temperature is 27 (80.6)(CDB[[degrees]FDB])/19 (66.2) ([degrees]CWB[[degrees]FWB]), outdoor unit intake air temperature is - 5(23)(CDB[[degrees]FDB]), and refrigerant pipe length is 7.5(24.6) m (ft).
Switching from the compression cycle to the pump cycle, and from the pump cycle to the compression cycle, is performed automatically under the conditions shown in Table 2. Also, at times when there is insufficient cooling capacity, or the pump is not working, it has an auto-switch function, which raises the reliability of the air conditioner. Moreover, the amount of refrigerant circulation is controlled with the compressor volume and expansion valve in the compression cycle and with the pump frequency and expansion valve in the pump cycle.
Table 2. Condition of Changing Cycle From the compression A The difference between the indoor cycle to the pump cycle temperature and the outdoor (A and B and C) temperature is more than the defined value. B The presumed cooling capacity at the pump cycle is more than the air-conditioning load. C The pump has not failed. From the pump cycle to D The difference between the indoor the compression cycle temperature and the outdoor (D or E or F or G) temperature is less than defined value. E The difference between a balloon temperature or an inhalation temperature and each set value is more than the defined value. F The degree of subcooling is less than defined value. G The pump has failed.
Features of the Air Conditioner
The features of the pump cycle will be described below. Since the pump consumes less energy compared to the compressor, the amount of energy consumed by the air conditioner can be reduced. However, lowered pump performance or the occurrence of vibrations and noise due to cavitation, as well as damage to the device from erosion, are problems that occur when it is used as a refrigerant transport method. In this air-conditioning system, in addition to using a pump highly resistant to cavitation on the pump suction side there are controls which ensure the supercooling of the refrigerant. The features of the compression cycle will be described below. During the pump cycle, highly efficient operation is possible because refrigerant heat transfer is performed with the pump. During the compression cycle as well, the technology we developed for data centers, such as high-efficiency compressors and low-pressure ratio operating areas, are used. Specifically, the compressor uses a scroll compressor that employs a release valve (Figure 2), and has an asymmetric tooth profile scroll, which reduces leak loss and operates a DC inverter motor, which does not require excitation current in the rotor. Also, for the expansion of the low-pressure ratio operating area, the minimum allowable compression ratio is expanded to 1.2 (Figure 3). In previous technology, since the minimum value of the operable compression ratio (minimum operable compression ratio) was about 1.5, even if the outdoor air temperature was low, it was necessary to maintain a high condensing pressure, consuming an excessive amount of compressor power. In contrast to this, by expanding the allowable minimum compression ratio to 1.2, it is possible to reduce the amount of compressor power used by reducing the condensing pressure. Furthermore, as a result of restricting the condensing pressure, it is also possible to make wide use of the difference in the refrigerant's specific enthalpy (Figure 4), which contributes to the air conditioner's increased efficiency.
Verification of Performance of the Pump Cycle in the Test Chamber
In developing this air conditioning system, we performed tests in order to verify its performance in the pump cycle (stable operation, cooling capacity, and operating efficiency). We installed the indoor and outdoor units for this air-conditioning unit in both a test room simulating an actual data center and in an artificial climate chamber (Figure 5, Table 3), and conducted a performance verification in the pump cycle. In the data center test room, heat sources resembling ICT devices (henceforth, "model heat source") were placed on the racks as the cooling load. The main measurement points are shown in Table 4. The cooling capacity during the pump cycle was computed using the differences in the inlet and outlet air temperatures for the indoor unit. Also, the inlet air temperature for the indoor unit was set as the indoor air temperature, and the inlet air temperature for the outdoor unit was set as the outdoor air temperature. Furthermore, the total amount of heat emitted by the model heat source was changed with each experiment so that the indoor temperature would be 27[degrees]C (80.6[degrees]F).
Table 3. Test Conditions Mock heat devices 0~3 kW (variable) x 20 units Number of racks 20 units Refrigerant pipe Gas [phi] 22.22 mm (0.866 in.), length:15m (49.2 ft) Liquid [phi] 15.88 mm (0.625 in.), length:15 m (49.2 ft) Table 4. Measurement Items Measurement point Remarks column Pump frequency Power consumption Refrigerant temperature (Inlet Thermocouple of type T and outlet of each unit) (class 1) Refrigerant pressure (Inlet and Pressure transmitter outlet of each unit) Air temperature (Indoor and Thermocouple of type T outdoor unit intake air) (class 1)
Initially, the test was conducted with the outdoor air temperature as the variable. The test conditions are shown in Table 5.
Table 5. Test conditions Indoor temperature Constant (27[degrees]C [80.6[degrees]F]) Outdoor - 10, -5, 0, 5, 10, 15[degrees]C (14, temperature 23, 32,41, 50, 59[degrees]F) Pump frequency 48 Hz * sss * The pump frequency was set to 40 Hz only when the outside temperature was 15[degrees]C(59[degrees]F), in order to ensure overcooling in pump intake.
Figure 6 shows the P-h line diagram for conditions where the indoor temperature is 27[degrees]C (80.6[degrees]F), and the outdoor temperatures are 15[degrees]C (59[degrees]F) and - 10[degrees]C (14[degrees]F) in the pump cycle. Figure 7 shows the positions of measurements points 1-6. From Figure 6, even at outdoor temperatures of either 15[degrees]C (59[degrees]F) or - 10[degrees]C (14[degrees]F), it was confirmed that the refrigerant which flows up to Inlet 1 of the indoor unit via Outlet 4 of the outdoor unit was in a state of supercooled liquid, and the refrigerant flowing from Outlet 2 of the inside unit to Inlet 3 of the outdoor unit was in a state of superheated vapor. From the above, it is inferred that this air-conditioning system operates in a stable manner in the pump cycle. Regarding the above, results were similar in conditions when the outdoor temperature was - 10[degrees]C (14[degrees]F) and above or 15[degrees]C (59[degrees]F) or less.
The results of experiments on cooling capacity, the amount of electricity consumed, and the operational efficiency when the outdoor temperature is the variable are shown in Figure 8. Here, operating efficiency was set to a value obtained by dividing the cooling capability by all of the electricity consumed by the air conditioner, including the indoor fan. From Figure 8, it can be seen that cooling capacity drops with a rise in outdoor temperature. The cooling capacity at each degree of outdoor air temperature in contrast to this air conditioner's rated cooling capacity is approximately 100% at an outdoor air temperature of - 10[degrees]C (14[degrees]F), but it was approximately 87% at an outdoor air temperature of 0[degrees]C (32[degrees]F), approximately 46% at 10[degrees]C (50[degrees]F), and approximately 27% at an outdoor air temperature of 15[degrees]C (59[degrees]F). As shown in Figure 6, the reason for this can be inferred to be the main factor in the vapor temperature climbing when the outdoor air temperature rises and the amount of heat exchanged with the air in the evaporator drops. The amount of energy consumed is fixed regardless of the outdoor air temperature. Therefore, the operating efficiency has the same tendencies as the cooling performance. In this test, it is 7 at an outdoor air temperature of 0[degrees]C (32[degrees]F) and 8 at an outdoor temperature of - 5[degrees]C (23[degrees]F).
Next, tests were conducted with the pump frequency as the variable. The experiment conditions are shown in Table 6.
Table 6. Test Conditions Indoor temperature Constant (27[degrees] C [80.6[degrees] F]) Outdoor temperature Constant (- 5[degrees] C [23[degrees] F]) Pump frequency 20, 30, 40, 48 Hz
The experimental results of the cooling capacity, electrical power consumption, and operating efficiency with the pump frequency as a variable are shown in Figure 9. The operating efficiency is similar to when the outdoor temperature is set as the variable. From Figure 9, it can be seen that the amount of refrigerant circulated increases and the cooling capacity rises with an upturn in the pump frequency. In other words, we could confirm that cooling capacity can be controlled by pump frequency. Also, the electrical energy consumption of an air-conditioning unit is almost fixed regardless of pump frequency. This is due to less electrical power being consumed by the pump than the electrical power consumed by other air-conditioning components, such as indoor fans, so the effect upon the electrical power consumption of air-conditioning units that change pump frequency is small. Therefore, the operational efficiency displays a similar tendency to cooling efficiency. From the above, by increasing the amount of refrigerant circulated in the pump cycle and increasing the cooling capacity, improvements in efficiency can be expected.
This air-conditioning system was installed at a data center in Sapporo, and field evaluations were conducted from December 2010 to January 2012. This indoor unit for this air conditioner was installed on the fourth floor, the outdoor unit and the pump unit were installed on the balcony of the floor directly above, and the length of the refrigerant piping between the indoor and outdoor units was approximately 50 meters. During the evaluation period, the operation of the pump cycle was stable. As an example, the operating status of a particular day is shown in Figure 10 (a) and (b). In Figure 10 (a), the cooling capacity fluctuated around 15 kW in relation to a rated cooling capacity of 45 kW. Also, the average load rate during the measurement period was about 28%, and it was a low-load operating environment for the capacity of the air conditioner. For the temperature conditions in the server room, Figure 10 (b) shows the indoor unit's intake air temperature and the temperature inside the double floor. The temperature fluctuations for the air intake for the indoor unit 24[degrees]C [+ or -] 1[degrees]C (75.2[degrees]F [+ or -] 1.8[degrees]F) and inside the raised floor 22[degrees]C [+ or -] 1[degrees]C (71.6[degrees]F [+ or -] 1.8[degrees]F) were stable.
Improvements for Commercialization
The aforementioned test-chamber tests and field evaluations were conducted at the development stage and the stability of the pump cycle's operation was verified. In the commercialization stage, improvements were made with the goal of further increasing cooling capacity and improving efficiency during operation. The major points of improvement are as follows.:
* Increase the cross-section area of the gas-side pipe between the indoor and outdoor units.
* Optimization of the pump (Improvements in the tooth shape of the pump that match the characteristics of the refrigerant).
From the results of the pump frequency (Figure 9), we determined that if the amount of circulated refrigerant is increased, cooling capacity also increases, and along with this, improvements in efficiency can be expected. Reducing refrigerant pressure loss in the cycle was effective for increasing the amount of circulated refrigerant. Therefore, with the aim of reducing refrigerant pressure loss, there is a great impact on pressure loss in the cycle and the cross-section area was increased for the gas-side pipes between the indoor and outdoor units, which are long, compared to the pipes in the indoor and outdoor units. Also, improving the performance of the pump that transports the refrigerant was thought to be necessary in order to improve efficiency, so the pump was also optimized by improving the tooth shape to match it to the characteristics of the refrigerant. Increasing the surface area of the condenser installed in the outdoor unit in order to bring in low-temperature outdoor air was also considered, but this has not been implemented due to the constraints of the dimensions of the outdoor unit's casing. In commercially-available units, improvements in cooling ability and operational efficiency exceeding the results of this test due to the aforementioned improvements can be expected. Next, we will discuss the characteristics of commercially-available units.
Reduction of energy consumption
Figure 11 shows the operating efficiency of commercially available outdoor air cooling-type air-cooled packaged air conditioners and air conditioners compared to the outdoor temperature of conventional computer-room air conditioners. Also, the time in which the outdoor air temperature occurred according to standard meteorological data in Sapporo and New York are also recorded. The plots in the figure are the measured values and the lines are the simulated values. Also, Figure 11 shows the results after the improvements for commercialization were made. The operating efficiency of the air conditioner is, as previously stated, set to a value obtained by dividing the cooling capability by all of the electricity consumed by the air conditioner, including the indoor fan, etc. It can be seen from the figure that the efficiency of conventional computer-room air conditioning does not improve regardless of the outdoor temperature, but in the package air conditioner we developed, the efficiency improves as the outdoor temperature drops. Also, it can be seen that efficiency improves even more by switching from the compression cycle to the pump cycle.
Figure 12 is the yearly percentage of the amount of consumed energy by the unit we developed compared to that of conventional air conditioners for computer rooms. The conditions for calculation are as shown in Table 7. Also, regarding the cooling load, in the compression cycle, the rated cooling capacity was assumed to be 100%. However, it was set as the air conditioner's maximum cooling capacity in the pump cycle operating range. As a result of our trial calculations, the unit we developed reduced energy consumption by approximately 54% in Sapporo and approximately 50% in New York compared to conventional computer-room air conditioners.
Annual Electrical Power Consumption Ratio [-] Conventional A/C 1.00 NEWA/C 0.46 Annual Electrical Power Consumption Ratio [-] Conventional A/C 1.00 NEWA/C 0.50 Figure 12 Annual electrical power reduction results: (a) Sapporo and (b) New York. Note: Table made from bar graph. Table 7. Operating Efficiency Test Calculation Conditions Indoor Unit Intake 27[degrees] C (80.6[degrees] F) DB/ Air Temperature 19[degrees] C (66.2[degrees] F) WB Outdoor Temperature Variable Refrigerant Pipe 7.5 m (24.6 ft) Length Air volume of indoor Constant (240 [m.sup.3]/min) unit
We developed an outdoor air cooling-type air-cooled package air conditioner for data centers that has both a compression cycle and pump cycle. Pump cycle performance tests in a test room during development demonstrated the characteristics of air conditioners in relation to outdoor temperature and pump frequency. Furthermore, we confirmed the stability of the pump cycle's operation through actual field tests. We made improvements in light of the tests, and in the commercially-available unit, it was possible to reduce the amount of energy consumed annually by approximately 54% in Sapporo and approximately 50% in New York. This airconditioning system can contribute to the reduction of energy consumption by air conditioners in data centers.
ASHRAE. 2009. Particulate and Gaseous Contamination in Datacom Environments. Atlanta: ASHRAE.
K. Sekiguchi, S. Waragai, T. Uekusa, K. Yamasaki. 2010. Development of a high-efficiency air cooled packaged air-conditioner for data centers. ASHRAE Transactions 116(1):330-335.
T. Uekusa, M. Yanagi, and S. Waragai. 2003. Study on efficiency improvement of year round cooling packaged air conditioners with a refrigerant pump. International Congress of Refrigeration, August 17-22, Washington, DC.
T. Uekusa, M. Nakao, and K. Ohshima. 1990. Control method of a cooling apparatus in low outdoor air temperatures. ASHRAE Transactions 96(1):200-4.
Y. Udagawa, S. Waragai, M. Yanagi, W. Fukumitsu. 2010. Study on free cooling systems for data centers in Japan. INTELEC, June 6-10, Orlando, Florida.
M. Yanagi, S. Waragai, K. Sekiguchi. 2011. Energy saving effects of free cooling air conditioners for data centers. Annual Meeting of SHASE of Japan, September 14-16, Nagoya, Japan: 445-448.
Keisuke Sekiguchi, DrEng
Masahide Yanagi, DrEng
Tsuneo Uekusa, DrEng
Yosuke Udagawa is a research engineer, Keisuke Sekiguchi is a senior research engineer, Masahide Yanagi is an executive manager, and Tsuneo Uekusa is a senior executive manager at the Research and Development Headquarters of NTT Facilities, Inc., Tokyo, Japan. Yasuhiro Naito is a senior engineer in the Air Conditioning Design Department of Hitachi Appliances, Inc., Shizuoka, Japan.
|Printer friendly Cite/link Email Feedback|
|Author:||Udagawa, Yosuke; Sekiguchi, Keisuke; Yanagi, Masahide; Uekusa, Tsuneo; Naito, Yasuhiro|
|Date:||Jan 1, 2013|
|Previous Article:||Modeling of proposed changes to the SIUC central heating, air conditioning, and power plant incorporating variable-frequency drives and...|
|Next Article:||The impact of barometric pressure, temperature, and operating temperature difference on air-cooled data centers.|