See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/349511146

Optimal setting parameters of cooling system under different climate zones

for data center energy efficiency

Article in International Journal of Energy Research · February 2021

DOI: 10.1002/er.6499

CITATIONS READS

6 275

6 authors, including:

Jing Liu Qin Sun

University of Nottingham Guangzhou Institute of energy conversion, China,Guangzhou
2 PUBLICATIONS 11 CITATIONS 18 PUBLICATIONS 210 CITATIONS

SEE PROFILE SEE PROFILE

Gongsheng Huang
City University of Hong Kong
179 PUBLICATIONS 4,129 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jing Liu on 16 March 2021.

The user has requested enhancement of the downloaded file.

Received: 4 August 2020 Revised: 12 January 2021 Accepted: 12 January 2021
DOI: 10.1002/er.6499

RESEARCH ARTICLE

Optimal setting parameters of cooling system under

different climate zones for data center energy efficiency

Jing Liu1,2,3,4 | Lin Su1,3,4 | Kaijun Dong1,3,4 | Qin Sun1,3,4 |

Zhenhua Shao1,3,4 | Gongsheng Huang5

1
Guangzhou Institute of Energy
Conversion, Chinese Academy of
Summary
Sciences, Guangzhou, China The high rates of electricity consumption in data centers become a worldwide
2
University of Chinese Academy of concern. Water-side free cooling is an effective solution to reduce the energy
Sciences, Beijing, China
consumed by the data center's cooling system. However, the water-side free
3
CAS Key Laboratory of Renewable
cooling system is not well planned, and the free cooling utilization rate is low
Energy
4
Guangdong Provincial Key Laboratory of in different climate zone. Moreover, the effects of local climatic conditions on
New and Renewable Energy Research and the optimal setting parameters have not been researched. In this study, the
Development, Guangzhou, China optimal setting in chilled water supply temperature, temperature difference,
5
Department of Architecture and Civil
and chilled water return temperature is analyzed to enhance the efficiency of
Engineering, City University of Hong
Kong, Hong Kong, China the free cooling system. We observed that the optimal setting could reduce
annual energy consumption by 23.5%–60.8% in various climate zones and has
Correspondence
substantial energy-saving potential in the subtropical highland climate zone.
Lin Su and Kaijun Dong, Guangzhou
Institute of Energy Conversion, Chinese The results show that the optimal setting parameters are highly affected by the
Academy of Sciences, Guangzhou, China. annual accumulative wet-bulb temperature distribution. Furthermore, consid-
Email: [email protected] (L. S.) and
[email protected] (K. D.)
ering a monthly analysis of these enhanced parameters, we proposed a dual-
working condition cooling system for the data center's future construction.
This research work guides the design of such cooling systems in various cli-
mate zones.

KEYWORDS
climate condition, data centers, energy-saving, optimal setting, water-side free cooling system

1 | INTRODUCTION electricity consumption in 2020.1 The data centers in

China consumed 160.9 terawatt-hours in 2018, which
The increasing scale and amount of data centers yield was 1.6 times of the whole generation of the Three Gor-
energy consumption, with the fast growth of cloud com- ges.2 The energy consumed by data centers in China may
puting, 5G communication, and data storage. The elec- reach 266.8 terawatt-hours by 2023.3 Average power
tricity using in data centers made up 1% of the global usage effectiveness (PUE)4 is employed to evaluate data
center power utilization efficiency. The lower the PUE is,
Abbreviations: PUE, average power usage effectiveness; CRAC, the higher the data center efficiency. The main idea is to
computer room of air conditioning; IT, information technology; RH, enhance the data center cooling system's energy effi-
relative humidity; DBT, dry-bulb temperature,
C; WBT, wet-bulb
ciency since the cooling system consumes 30%—50% of
temperature,
C; DP, dew-point temperature,
C; PLR, part load ratio;
FFLP, the power correction factor of full load power; TMY, typical
its energy consumption, leading to high PUE.5
meteorological year; VFD, variable-frequency drive; FCR, free cooling Free cooling technology can effectively reduce the
utilization rate. energy consumption of the data center cooling system.6

Int J Energy Res. 2021;1–14. wileyonlinelibrary.com/journal/er © 2021 John Wiley & Sons Ltd 1
2 LIU ET AL.

Many researchers analyze different free cooling strategies by analyzing several variables. The variables involve chil-
under various climate conditions. Mahdi7 compares the led water supply temperature (Tchs), chilled water return
energy-savings of three kinds of free cooling economizers temperature (Tchr), and chilled water temperature differ-
in Iran and finds that the water-side economizer shows ence (Δt). To obtain the optimal setting parameters, we
the highest energy-saving effect in Tabriz, Iran. Diaz8 devise eight groups of improved settings and compare the
mainly compares the water-side economizer's perfor- energy consumption and free cooling utilization rate
mance with an indirect evaporative cooler under differ- (FCR) of various settings in a simulation study. The
ent climate conditions. Agrawal9 evaluates the effect of energy-saving rates of the optimal settings in five climate
the water-side economizer under different climate zones zones have been studied. Additionally, based on the
and compares the water consumption of the cooling sys- monthly analysis of each improved setting, we propose a
tem. Most researchers focus on applying the water-side new dual-working condition cooling system for the future
economizer in various climate zones rather than improv- construction of the data center.
ing the energy efficiency of the water-side economizer. In
hot climate zones, the cooling system energy consump-
tion is high due to the reduction of free cooling dura- 2 | METHODOLOG Y
tion.10,11 To extend the free cooling duration, Zhang12
and Kang et al13 realize that increasing the temperature 2.1 | Determining the upper value of
of supply chilled water and return chilled water can computer room of air conditioning unit's
enhance the free cooling duration. Few studies adopt supply air temperature and chilled water
high-temperature of chilled water in their researches. return temperature
Yang et al14 discuss the free cooling potential with 12
C
of chilled water supply temperature. Cheung15 utilizes The data center is operated in a strict environment of the
13
C as chilled water supply temperature in the cooling computer room of air conditioning (CRAC) unit's supply
system design. Ling16 proposes a genetic algorithm to air temperature and humidity to ensure the stability of
obtain the optimized chilled water supply temperature, the information technology (IT) equipment. ASHRAE23
which is 15
C. However, the chilled water return temper- stipulates the heat requirements at the data center envi-
ature on the cooling system's energy consumption has ronment, and the recommended range of supply air tem-
not been well addressed in their researches. perature is provided in Table 1. So, the upper value of the
The temperature difference of chilled water is another supply air temperature is 27
C. Gao24 analyses the char-
critical factor influencing the cooling system's efficiency. acteristic of cross-flow heat exchanger in CRAC, and
Researches17-20 in large temperature difference of chilled finds that the chilled water return temperature is mainly
water cooling systems confirm that it can improve the influenced by the CRAC unit's supply air temperature.
effectiveness of the cooling system because it reduces the Kang13 finds that if the Δt between the supply air temper-
energy consumption of pumps and cooling towers. The ature and the chilled water return temperature is less
traditional air-conditioning system with large tempera- than 2
C, it is hard to meet the ASHRAE requirement.
ture difference is achieved by increasing the chilled water Therefore, the upper value of the chilled water return
return temperature and reducing the chilled water supply temperature is considered as 25
C.
temperature.21 If it is applied in the water-side cooling
system of the data center, the free cooling duration and
chiller's efficiency will decrease with the reduced chilled 2.2 | Determining the setting
water supply temperature. Zhang et al22 improve the chil- parameters of the cooling system
led water supply temperature and temperature difference
simultaneously, which is 15
C and 10
C, respectively. The traditional setting parameters of the data center
This design achieves more energy saving than the tradi- cooling system are 10
C chilled water supply temperature
tional cooling system, but the optimal chilled water tem- (Tchs), 15
C chilled water return temperature (Tchr),
perature and temperature difference have not been and 5
C chilled water temperature difference. In the new
studied. settings of the cooling system, 25
C is set as the Tchr, but
Previous researches discuss the effects of the individ- we need to explore the optimal chilled water supply tem-
ual variables on the free cooling duration or the energy- perature and temperature difference. Thus, we set eight
savings of the cooling system, ignoring other factors and improved setting parameters to obtain the optimal setting
design parameters on the whole cooling system. There- parameters of the cooling system. Table 2 outlines the
fore, in this study, we explore the optimal setting parame- eight improved setting parameters in the cooling system.
ters of the cooling system with the water-side economizer The temperature difference (Δt) varies from 5
C to 12
C.
LIU ET AL. 3

2.3 | Simulation study temperature of supply chilled water (Tchs), the system
switches to full free cooling mode and the chiller is shut
To come up with the most efficient of the eight settings, down. Valve 1, valve 3, and valve 6 are turned on. The
we simulate a data center cooling system to analyze these cooling tower produces all the cooling energy required to
settings' performance. cool the IT equipment in CRAC, which in return saves a
In the simulation study, we establish a water-side free huge amount of energy.
cooling using the transient simulation software TRNSYS.
This simulation research is performed on a medium-sized
data center, real equipment algorithms, and a dynamic 2.3.2 | Simulation system and
control strategy. designation of the cooling system

Figure 2 exhibits the TRNSYS simulation model for the

2.3.1 | Water-side free cooling system in water-side free cooling system. The simulation system is
data center set up in regard to the schematic diagram in Figure 1.
The simulation system simulates a medium-sized data
Figure 1 exhibits a typical schematic diagram of the center with 2050 kW cooling load, including a 2050 kW
water-side free cooling system in the data center. The chiller, three water pumps, and two cooling towers.
water-side free cooling system is composed of a chiller, Table 3 exhibits the equipment specification of the
plate heat exchanger, cooling tower, pumps, and terminal cooling system in various designs. To guarantee the
CRAC in the data center. The main differences between equipment's high efficiency under different operating
this system and the conventional air-conditioning system conditions, the specifications of the equipment are
are as follows: an extra plate heat exchanger is installed decided according to the system's water flow rates. There-
to make use of the free cooling energy; When the outdoor fore, in different settings (shown in Table 2), the equip-
wet-bulb temperature is low, the cooling tower transfers ment specifications such as pumps, cooling towers, and
cold energy directly to the CRAC through the plate heat the heat exchanger are different. The rate at which water
exchanger; the cooling energy is desired the whole year; flows through each equipment in the design is calculated
the cooling tower operates throughout the year. There with regard to the heat balance formula, and the Δt var-
are three kinds of working modes in this system: ies from 5
C to 12
C according to Table 2.
(a) mechanical cooling mode, (b) part free cooling mode, Figure 3 shows the dynamic control strategy and the
(c) full free cooling mode. running mode of the simulation system. The running
Whenever the cooling tower's outlet temperature mode is determined by detecting the changes in water
(Tcws) is higher than the chilled water return tempera- temperature at the cooling tower and plate heat
ture (Tchr), the system runs in mechanical mode. Valve exchanger in real-time. Table 4 shows the operational
2, valve 4, and valve 5 are turned on. Chilled water is pro- state of the equipment under three running modes, and
duced from the chiller, which is used to cool the IT the equipment list corresponds to Figure 2.
(information technology) equipment of CRAC in the data
center. And the return chilled water goes back to the
chiller to complete a water loop. When the cooling 2.3.3 | Component models
tower's outlet temperature grows colder than the chilled
water return temperature, the system switches to part The simulation model should be validated to guarantee the
free cooling mode. Valve 1, valve 2, valve 3, and valve correctness of simulation results. However, the improved
5 are turned on. A part of the cooling water is used to settings of cooling systems are different from the parame-
pre-cool the return chilled water through the plate heat ters of the current data center cooling system, so it is wrong
exchanger. If the plate heat exchanger's outlet tempera- to use the data from the existing data center to confirm
ture (Thx) on the load side is lower than the set simulation results. Therefore, in this paper, the equipment
with high-energy consumption, such as the chiller, water
T A B L E 1 Thermal guidelines for data center's environment pump, cooling tower, will be validated by manufactures'
(ASHRAE 2011) specifications of the equipment. Besides, all the algorithms
of each equipment model in the simulation system should
DBT
be in line with the algorithms in real equipment.
Class (
C) Humidity range
1. Chiller
Recommended range of 18-27 5.5
C DP to RH 60% Chiller is modeled by TRNSYS model type 666, which
supply air temperature and 15
C DP
is a water-cooled chiller. Type 666 is a semi-experienced
4 LIU ET AL.

T A B L E 2 Setting parameters of
Setting parameter (temperature) Tchs Tchr Temperature difference
different settings in the data center
Conventional setting 10
C 15
C 5
C cooling system
Improving setting 1 13
C 25
C 12
C
Improving setting 2 14
C 25
C 11
C
Improving setting 3 15
C 25
C 10
C



Improving setting 4 16 C 25 C 9
C
Improving setting 5 17
C 25
C 8
C
Improving setting 6 18
C 25
C 7
C
Improving setting 7 19
C 25
C 6
C



Improving setting 8 20 C 25 C 5
C

F I G U R E 1 The schematic diagram

of the water-side free cooling system in
data center

Qload = MCpchw ðT chw,in −T chw,set Þ ð3Þ

model, which depends on a catalog data lookup and
linear interpolated method to predict the performance where M is the chilled water mass flowrate, Cpchw is the
of the chiller. Two external data files are: capacity ratio specific heat of fluid, Tchw, in is the fluid temperature
and COP ratio data file, and part load ratio and fraction entering the chiller, Tchw, set is the set temperature of sup-
of full load power data file. The first data file provides ply chilled water.
capacity ratio and COP ratio under different chilled The PLR (part load ratio) is calculated by Equation (4):
water set temperature and cooling water temperature.
The second data file gives the chiller's power correction Qload
PLP = ð4Þ
factor of full load power for varying part load ratio. The Capacitynom
interpolation data files provided by the manufacturer's The chiller's power draw is therefore:
equipment specifications have been validated in Tang's
study.25 Capacitynom
P= FFLP ð5Þ
The chiller's nominal COP is measured according to COPnom
Equation (1). The nominal capacity is measured
employing Equation (2): where FFLP is the power correction factor of full load
power.
COPnom = COPrated COPratio ð1Þ 2. Water pump
The water pump is modeled by TRNSYS model type
Capacitynom = Capacityrated Capacityratio ð2Þ 201. The power of the pump is calculated by Equation (6):

where subscript of “nom” is the value at current condi- ΔPM

P= ð6Þ
tion with full power, “rated" is the value at rated condi- ρfluid ηpumping ηmotor ηVFD
tion with full power, "ratio"is the value at current
condition divided by the rated value. where ΔP is the pressure drop of fluid; M is the mass
The chiller load (Qload) is calculated by Equation (3): flowrate of fluid; ρfluid is fluid density; ηpumping is the
LIU ET AL. 5

F I G U R E 2 The water-side free

cooling system of data center modeled in
TRNSYS [Colour figure can be viewed at
wileyonlinelibrary.com]

efficiency of pumping the fluid; ηmotor is the motor effi- VFD (variable-frequency drive) are fitted by experiment
ciency; ηVFD is the efficiency of the variable-frequency results of a real pump.26
drive.
• b = [0.94187, −9.04], which is the coefficient of motor
ηpumping = 0:77 under different speed.
• c = [0.5842, −1.42, 1.283, 0.5067], which is the coeffi-

cient of VFD under different speed.
ηmotor = b1 1 −eb2 I
• I = MMrated
, which is current flow rate divided by rated
flow rate.
ηVFD = c1 I 3 + c2 I 2 + c3 I + c4
3. Cooling tower
The pumping coefficient remains constant since there The cooling tower is modeled by TRNSYS model type
is a stable cooling load, and the pump always runs under 51 b. It is a counter-flow wet cooling tower, which has
similar working conditions. The coefficient of motor and hot water in direct contact with the cold air supplied by
6 LIU ET AL.

the cooling tower fan. Then, the hot water is cooled due where γ is relative fan speed, Pmax is maximum power of
to evaporation and sensible heat transfer with the air. the fan.
The colder the weather, the lower the outlet cooling The mathematical principle of the cooling tower has
water temperature. been derived in regard to the Merkel enthalpy method.27
The fan laws have been considered to derive the power Tower effectiveness calculation follows the ASHRAE
of the cooling tower fan. Given the maximum fan power, Equipment Guide.28 The algorithms inside the cooling
the cooling tower's power is calculated by Equation (7): tower relate to the algorithms in real equipment. The
accuracy of the cooling tower model has been verified
Ptower = γ3 Pmax ð7Þ using measure data by Tang.25

T A B L E 3 Equipment specification of the cooling system in 2.3.4 | Operating condition of the

various settings simulation system
Equipment list Rated parameters Quantity
Figure 4 outlines a procedural simulation in TRNSYS. The
Chiller Rated cooling capacity (Q): 1
inputs include cooling load, setting parameters, and
2050 kW
weather files. To examine the settings' performance in a
Rated COP: 6.46
more realistic situation, the simulation system performs
Chilled water flow rate: q1 with a typical meteorological year (TMY) weather file15
Cooling water flow rate: q2 under five typical climate zones. The climate zone is
Heat exchanger Heat transfer rate: 2150 kW 1 selected based on the global temperature zone. Consider-
Water flow rate: q1 ing the region in polar climate is too cold to use water-side
Chilled water Flow rate: q1 1
pump 1 Design pressure head:
343 kPa T A B L E 4 The operation state of the equipment under different
running modes
Cooling water Flow rate: q2 1
pump 2 Design pressure head: Mode Operational state
314 kPa Mechanical (1) Turn on: chiller, pump 1, pump 2, and
Cooling water Flow rate: q1 1 cooling mode cooling tower 1
pump 3 Design pressure head: (2) Turn off: pump 3, cooling tower 2, and
314 kPa heat exchanger
Variable-speed Water flow rate: q2 2 Part free (1) Turn on: chiller, pump 1, pump 2, pump
cooling tower cooling mode 3, cooling tower 1, cooling tower2, and
heat exchanger
Note: q1 = Q × 1.1/(c × Δt), where Q is cooling load (which is 2050 kW), c is
the specific heat of fluid (which is 4.195 kJ/ (kg
C)), Δt is the temperature Full free (1) Turn on: pump 1, pump 3, cooling
difference of chilled water (which varies from 5
C–12
C in different cooling mode tower1, cooling tower 2, and heat
settings). q1 = 161 263.41–387 032.18 kg/s. q2 = 1.1 × q1, the q2 (cooling exchanger
water flow rate) is 1.1 times of q1 (chilled water flow rate).
(2) Turn off: chiller and pump 2
q2 = 177 389.75–425 735.40 kg/s.

F I G U R E 3 Operation mode and

dynamic control strategy
LIU ET AL. 7

FIGURE 4 The flow chart of the

simulation

free cooling, we select the typical climates of the temperate The FCR is used to assess the energy-saving potential
climate zone, subtropical climate zone, and tropical cli- of free cooling. FCR is calculated by Equation (8):
mate zone. Table 5 shows the sample cities from typical
five climate zones. Table 6 shows the vital weather param- Qfc
FCR = ð8Þ
eters of selected cities, such as the description of the cli- Qtot
mate and wet-bulb temperature. Figure 5 shows the
annual wet-bulb temperature distribution of typical cities. where Qfc refers to the cold energy provided by full free
cooling or part free cooling; Qtot is the cooling load of the
cooling system.
3 | R E S U L T A N D DI S C U S S I O N Figure 6A–10A show the energy consumption of
cooling systems in eight improved settings in five cli-
The simulation results are investigated in three aspects: mate zones. With the increases in chilled water supply
the analysis of optimal settings under five climate zones, temperature (Tchs), the energy consumption of the
the rate at which energy is saved at optimal setting chiller decreases, while the energy consumed by
parameters, and the amount of energy consumed in every the pumps and cooling towers increases. Therefore, in
month at different settings. Figure 6A–9A (Temperate continent climate zone—
Subtropical monsoon), the energy consumption of the
cooling system decreases till the valley is reached. Then,
3.1 | Analysis of optimal settings under there is a steady rise despite the increase of total FCR
five climate zones and total free cooling hours in Figure 6B–9B. However,
in Figure 10A (Tropical climate zone), the total energy
This section mainly analyses the cooling system's energy consumption rises with the rise in temperature of supply
consumption and free cooling utilization ratio in eight chilled water. The total FCR and free cooling hours in
improved settings. Bangkok (Figure 10B) remain steady as the increase of
8 LIU ET AL.

T A B L E 5 Selected cities from

Climate main classification Representative city Latitude Longitude
typical climate zone
Temperate continent Chita, Russia 52.31 113.22
Temperate monsoon Beijing, China 39.91 116.41
Subtropical highland Kunming, China 25.05 102.73
Subtropical monsoon Guangzhou, China 23.16 113.23
Tropical monsoon Bangkok, Thailand 13.72 100.64

TABLE 6 Important weather statistics of selected cities from five typical climate zones

Representative Average Maximum Minimum

city Description WBT (
C) WBT (
C) WBT (
C)
Chita Very cold, very cold and dry in winter −3.03 20.80 −37.70
Beijing Cold and wet, cold winter and hot summer 8.08 28.80 −15.79
Kunming Temperate, dry winter and mild summer, small 11.91 21.19 −2.52
annual temperature difference
Guangzhou Hot and wet, warm winter and hot summer 18.73 29.91 1.98
Bangkok Extremely hot and wet, small annual temperature 24.09 28.55 11.61
difference

between the chiller's energy savings and the pumps and

cooling towers' energy consumption. Therefore, the
analysis of optimal setting parameters in different
climate zones is vital in minimizing the cooling system's
energy consumption. From the results in Figures 6 to
10, the most energy-saving settings in various climate
zones are as follows: improved setting parameters of
14
C to 25
C in Chita (Figure 6A—Temperate conti-
nent); improved setting parameters of 15
C to 25
C in
Beijing (Figure 7A-Temperate monsoon); improved set-
ting parameters of 18
C to 25
C in Kunming
(Figure 8A-Subtropical highland); improved setting of
16
C to 25
C parameters in Guangzhou (Figure 9A-
Subtropical monsoon); improved setting parameters of
13-25
C in Bangkok (Figure 10A-Tropical monsoon). It
is worth noting that the optimal setting parameters vary
F I G U R E 5 Annual wet-bulb temperature distribution of
with each climatic zone. As presented in Figures 6 to 10,
typical cities [Colour figure can be viewed at
the effects of increasing Tchs on free cooling time, FCR,
wileyonlinelibrary.com]
and chiller's energy consumption are different in five
climate zones. The annual wet–bulb temperature distri-
chilled water supply temperature, which results in a bution highly affects it. Figure 11 shows the relation
small reduction of the chiller's energy consumption in between optimal chilled water supply temperatures with
Figure 10A. Furthermore, Figure 6B–10B indicate that annual accumulative wet–bulb temperature distribution
the highest FCR and free cooling hours do not corre- of typical cities. Without considering the approach tem-
spond to the lowest energy consumption of the cooling perature, in free cooling mode, the outdoor wet-bulb
system. temperature is equal to the cooling tower outlet water
The results indicate that higher Tchs enhances the temperature, which is also equal to supply chilled water
chiller's energy saving and FCR but reduces Δt. The set temperature. The higher the Tchs, the longer the free
pump and cooling tower will consume more energy. cooling duration. Therefore, the region with a cumula-
The free cooling continuously modifies the balance tive number of wet-bulb temperatures increasing rapidly
LIU ET AL. 9

F I G U R E 6 A, Cooling system's energy consumption of eight improved settings in Chita (Temperate continent); B, Free cooling
utilization rate and cooling hours of eight improved settings in Chita (Temperate continent) [Colour figure can be viewed at
wileyonlinelibrary.com]

once the Tchs increases can save more energy by raising consumed in a year by 23.5%–60.8%. It verifies that the
the Tchs. Thus, Kunming, where the cumulative wet- high-temperature chilled water and large temperature
bulb temperature curve is the steepest, has the highest difference cooling system can drastically enhance the
optimal chilled water supply temperature. Bangkok, energy efficiency of data centers.
where the cumulative wet-bulb temperature curve is
gentlest, has the lowest optimal chilled water supply
temperature. 3.3 | Cooling system's monthly energy
In summary, the optimal setting parameters in vari- consumption in different settings
ous climate zones have been obtained in this section; the
highest total free cooling utilization ratio and free cooling The annual-basis analysis of optimal setting parameters
hours do not correspond to the most energy-saving has been studied in Section 4.1 and 4.2 above. We find
cooling system, so the combined mutual influence of that the optimal setting of different months varies from
other design parameters on the whole cooling system is the annual optimal setting, due to the changes in outdoor
necessary; the optimal setting parameters are affected by wet-bulb temperature. Owing to the excellent energy-
the local climate, especially the annual accumulative saving effect, we chose the setting in Kunming as an
wet-bulb temperature distribution. example. Figure 13 shows the monthly energy consump-
tion of eight improved settings in Kunming. In cold
months (such as January, February, December), the
3.2 | The energy-saving rate of optimal lower Tchs settings, the lower energy consumed of
setting under five typical climate zones cooling systems. In the cold season, the low outdoor wet-
bulb temperature causes a low outlet temperature of
Figure 12 shows the performance of the optimal setting cooling towers. The cooling tower can produce enough
and conventional setting in five climate zones. From the cold energy even with the chiller off. If the system
severe cold climate zone to the hot climate zone (Chita to operates under the setting of the high Tchs in the cold
Bangkok), the colder the climate the lower the energy season, the overall energy consumption will increase due
consumed and the higher the amount of energy saved. to the rise in energy consumption by the pump. In the
However, the energy-saving rate of Kunming is the hot months (such as May, June, July, August), higher
highest among the five cities, which is 60.8%, showing Tchs settings correspond to lower energy consumption of
that improving Tchs and the Δt is significant in the area cooling systems. With the outdoor wet-bulb temperature
with a subtropical highland climate. The optimal setting increases, the cooling tower's outlet temperature
in various climate zones reduces in the amount of energy increases. If the system runs in a higher Tchs setting, the
10 LIU ET AL.

F I G U R E 7 A, Cooling system's energy consumption of eight improved settings in Beijing (Temperate monsoon); B, Free cooling
utilization rate and cooling hours of eight improved settings in Beijing (Temperate monsoon) [Colour figure can be viewed at
wileyonlinelibrary.com]

F I G U R E 8 A, Cooling system's energy consumption of eight improved settings in Kunming (Subtropical highland); B, Free cooling
utilization rate and cooling hours of eight improved settings in Kunming (Subtropical highland) [Colour figure can be viewed at
wileyonlinelibrary.com]

energy consumption of the system goes down owing to system runs in different seasons. Figure 14 exhibits the
the rise in free cooling potential. Besides, the energy monthly energy consumed by the new system in two
saved by free cooling is more than the increased energy groups of settings parameters. From November to April,
consumption of the pump. Although the annual optimal the system runs in the setting of 16
C to 25
C. From May
setting parameters in Kunming is the improved setting of to October, the system operates at the setting of 20
C to
18
C to 25
C, it is not favorable in some of the months. 25
C. Consequently, the annual energy consumption of
Therefore, we propose a new dual-working condition the dual-working condition system is 937 MWh, which is
cooling system. Based on the optimal setting of 18
C to an increment in saving energy consumption by 10.9%.
25
C, we adjust the Tchs in a reasonable range when the Table 7 shows the energy-saving performance of the
LIU ET AL. 11

F I G U R E 9 A, Cooling system's energy consumption of eight improved settings in Guangzhou (Subtropical monsoon); B, Free cooling
utilization rate and cooling hours of eight improved settings in Guangzhou (Subtropical monsoon) [Colour figure can be viewed at
wileyonlinelibrary.com]

F I G U R E 1 0 A, Cooling system's energy consumption of eight improved settings in Bangkok (Tropical monsoon); B, Free cooling
utilization rate and cooling hours of eight improved settings in Bangkok (Tropical monsoon) [Colour figure can be viewed at
wileyonlinelibrary.com]

dual-working condition cooling systems in five climate at a small flow rate. Therefore, a new dual-working
zones. Compared with the optimal settings, the energy- condition chiller should be studied in further work.
saving rate is increased ranges between 2.4% and 10.9%.
When the system runs in two different settings, the
efficiency of the equipment varies with the water flow 4 | CONCLUSIONS
rate. Smaller multi-pumps and multi-cooling towers
should be used to maintain higher efficiency when This paper proposes a high-temperature chilled water and
operating on low flow rate conditions.15 However, the large temperature difference cooling system to enhance
chiller will consume more energy when the system runs the free cooling duration and in turn reduce the energy
12 LIU ET AL.

F I G U R E 1 3 The monthly energy consumption of eight

F I G U R E 1 1 Optimal chilled water supply temperatures and improved settings in Kunming (Subtropical highland) [Colour
annual accumulative wet-bulb temperature distribution of typical figure can be viewed at wileyonlinelibrary.com]
cities [Colour figure can be viewed at wileyonlinelibrary.com]

F I G U R E 1 2 The performance of optimal setting and

conventional setting [Colour figure can be viewed at F I G U R E 1 4 The monthly energy consumption of dual-
wileyonlinelibrary.com] working condition system in two settings [Colour figure can be
viewed at wileyonlinelibrary.com]

consumption of the cooling system. We investigate the

optimal setting parameters of the cooling system by con- conventional cooling systems, their performance yields total
sidering multiple setting parameters on the whole cooling energy-savings of 23.5%-60.8% in the annual energy con-
system. The optimal settings in five sample cities from five sumption. Among five climate zones, the region in the sub-
different climate zones are as follows: improved setting tropical highland climate shows the highest optimal chilled
parameters of 14
C to 25
C in Chita (Temperate conti- water supply temperature and energy-saving rate. The
nent); improved setting parameters of 15
C to 25
C in Bei- results of this research are helpful as design-making guid-
jing (Temperate monsoon); improved setting parameters ance for the energy efficiency of data center cooling systems.
of 18
C to 25
C in Kunming (Subtropical highland); The results show that the highest FCR and free
improved setting of 16
C to 25
C parameters in Guang- cooling duration do not correspond to the lowest energy
zhou (Subtropical monsoon); improved setting parameters consumption of the cooling system. Therefore, mutual
of 13
C to 25
C in Bangkok (Tropical monsoon). combined effect of chiller, pumps, and cooling towers is
We find that the optimal setting parameters are vital to the cooling system's energy consumption.
affected by the local climate, especially the annual accumu- Furthermore, through analyzing each improved set-
lative wet-bulb temperature distribution. Compared to ting's monthly performance, we propose a dual-working
LIU ET AL. 13

TABLE 7 Energy-saving performance of the dual-working condition cooling system in five climate zones

Representative cities Chita Beijing Kunming Guangzhou Bangkok

Climate zone Temperate Temperate Subtropical Subtropical Tropical
continent monsoon highland monsoon monsoon
Optimal setting 14–25
C 15–25
C 18–25
C 16–25
C 13–25
C





Dual-working condition setting 13-25 C 13–25 C 16-25 C 14-25 C -
16-25
C 17-25
C 20-25
C 18-25
C -
Energy consumption of optimal setting 612 1144 1051 1978 -
(MWh)
Energy consumption of dual-working 549 1087 937 1595 -
condition setting (MWh)
Energy-saving rate (%) 10.3 5.0 10.9 2.4 -

condition cooling system. The new system provides a bet- Capacityrated capacity at rated condition with full power
ter energy-saving way than the annual optimal setting of Capacityratio capacity at current condition with full
the system, which energy-saving rate varies from 2.4% to power divide by the rated capacity
10.9%. However, the performance of the chiller will Capacitynom capacity of current condition with full
decline when it runs in a dual-working condition. There- power
fore, a new dual-working condition chiller should be Qload chiller load at current condition, kW
researched in the future. M mass flowrate of fluid, kg/s
Cpchw specific heat of chilled water, kJ/ (kg
C)
ACK NO WLE DGE MEN TS Tchw, in fluid temperature entering the chiller,
C
This work was supported by the Science and Technology Tchw, set the desired outlet temperature of fluid in
Planning Project of Guangzhou, China [No.201902010031], the chiller,
C
International Cooperation Project of Guangzhou Develop- ΔP the pressure drop of fluid, Pa
ment District [No.2018GH05], and Guangdong Basic and ρfluid fluid density, kg/m3
Applied Basic Research Foundation [2019A1515111177]. ηpumping efficiency of pumping the fluid
ηmotor efficiency of the pump motor
ηVFD efficiency of the variable-frequency drive
DATA AVAILABILITY STATEMENT γ relative fan speed
The data that support the findings of this study are available Pmax maximum power of the cooling tower, kW
from the corresponding author upon reasonable request. Qfc the cold energy provided by full free
cooling or part free cooling, kW
NO MEN CLATU RE Qtot the cooling load of the data center cooling
system, kW
Symbols
Δt temperature difference,
C
Tcws cooling water supply temperature,
C ORCID
Tchs chilled water supply temperature,
C Jing Liu https://orcid.org/0000-0002-0866-8836
Tchr chilled water return temperature,
C
Thx outlet temperature of heat exchanger on RE FER EN CES
the load side,
C 1. Masanet E, Shehabi A, Lei N, Smith S, Koomey J. Recalibrating
Q cooling load, kW global data center energy-use estimates. Science. 2020;367:984-
q1 chilled water flow rate, kg/s 986. https://doi.org/10.1126/science.aba3758.
2. Liu J, Su L, Dong K, Liu X. Critical temperature of free cooling
q2 cooling water flow rate, kg/s
using indirect opening cooling tower in data centres: a review.
c specific heat of fluid, kJ/ (kg
C) Int J Ambient Energy. 2019;1:1-20. https://doi.org/10.1080/
COP coefficient of performance 01430750.2019.1694069.
COPrated COP at rated condition with full power 3. Gao gong li dian wang. GGII: When annual energy consump-
COPratio COP at current condition with full power tion exceeds 150 billion kilowatt-hours, lithium batteries will
divide by the rated COP help data centers achieve significant cost savings. https://www.
COPnom COP of current condition with full power gg-lb.com/art-38900.html.
14 LIU ET AL.

4. Garimella SV, Persoons T, Weibel J, Yeh LT. Technological 16. Ling L, Zhang Q, Yu Y, Ma X, Liao S. Energy-saving analysis of the
drivers in data centers and telecom systems: multiscale ther- cooling plant using lake water source base on the optimized control
mal, electrical and energy management. Appl Energy. 2013;107 strategy with set points change. Appl Therm Eng. 2018;130:1440-
(3):66-80. https://doi.org/10.1016/j.apenergy.2013.02.047. 1449. https://doi.org/10.1016/j.applthermaleng.2017.10.152.
5. Wang J, Zhang Q, Yoon S, Yu Y. Impact of uncertainties on the 17. Shu ZC, Zhang G. Research on cold air system with large tem-
supervisory control performance of a hybrid cooling system in perature difference and humidity. Build Energy Environ. 2020;
data center. Build Environ. 2019;148:361-371. https://doi.org/ 39:28-32. https://doi.org/10.3969/j.issn.1003-0344.2020.07.006.
10.1016/j.buildenv.2018.11.026. 18. Zhang M. Technical and economic analysis of chilled water
6. Zhang H, Shao S, Tian C, Zhang K. A review on thermosyphon temperature selection in data centers. CC&AC. 2019;3:53-57.
and its integrated system with vapor compression for free 19. Chen X, Zhang G. Terminal of chilled water with large temper-
cooling of data centers. Renew Sustain Energy Rev. 2018;81:789- ature difference solution of series connection fan coil unit. Ref-
798. https://doi.org/10.1016/j.rser.2017.08.011. rig Air-Cond. 2019;10:65-67.
7. Mahdi DD, Namanlo SV. Potentiometric and economic analy- 20. Kang ZB. Research on cooling energy-saving of data center
sis of using air and water-side economizers for data Centre based on high temperature cooling technology with large tem-
cooling based on various weather conditions. Int J Refrig-Rev perature difference. Northeast Electric Power University 2020.
Int Froid. 2019;99:213-225. https://doi.org/10.1016/j.ijrefrig. 21. Lee C, Chang Y. Energy-saving analysis of chilled water system
2019.01.011. for large temperature difference air conditioning. IOP Conf Ser
8. Diaz AJ, Caceres R, Torres R, Cardemil JM, Silva-Llanca L. Earth Environ Sci. 2019;252:1–6. https://doi.org/10.1088/1755-
Effect of climate conditions on the thermodynamic perfor- 1315/252/3/032112.
mance of a data center cooling system under water-side 22. Zhang T, Liu X, Li Z, Jiang J, Tong Z, Jiang Y. On-site mea-
economization. Energy Build. 2020;208:1-12. https://doi.org/10. surement and performance optimization of the air-
1016/j.enbuild.2019.109634. conditioning system for a datacenter in Beijing. Energy Build.
9. Agrawal A, Khichar M, Jain S. Transient simulation of wet 2014;71:104-114. https://doi.org/10.1016/j.enbuild.2013.12.014.
cooling strategies for a data center in worldwide climate zones. 23. TC 9.9 (2011) ASHRAE, Thermal guidelines for data processing
Energy Build. 2016;127:352-359. https://doi.org/10.1016/j. environments-expanded data center classes and usage guid-
enbuild.2016.06.011. ance, 9, Whitepaper prepared by ASHRAE technical committee
10. Cho J, Kim Y. Improving energy efficiency of dedicated cooling (TC), 2011.
system and its contribution towards meeting an energy- 24. Gao T, Geer J, Sammakia B. Review and analysis of cross flow
optimized data center. Appl Energy. 2016;165:967-982. https:// heat exchanger transient modeling for flow rate and tempera-
doi.org/10.1016/j.apenergy.2015.12.099. ture variations. J Therm Sci Eng Appl. 2015;7(041017):1–10.
11. Cho J, Yang J, Lee C, Lee J. Development of an energy evaluation https://doi.org/10.1115/1.4031222.
and design tool for dedicated cooling systems of data centers: 25. Tang H. Simulation and Parameter Analysis of Water-Side Free
sensing data center cooling energy efficiency. Energ Buildings. Cooling System in Data Center [D]. Hu Nan University, 2018.
2015;96:357-372. https://doi.org/10.1016/j.enbuild.2015.03.040. 26. Yang S, Li H, Yin J. Development and application of variable fre-
12. Zhang BJ. Application and research of tower cooling system in quency pump module based on TRNSYS. HV & AC. 2015;8:36-41.
energy-saving reconstruction of electronics factory's air-conditioning 27. Braun JE. Methodologies for the Design and Control of Chilled
system [D]. Shanghai, China: Dong Hua University; 2016. Water Systems. Ph. D. Thesis, University of Wisconsin—Madi-
13. Kang Z, Shao Z, Su L, Dong K, Liu H. Experimental study and son, 1988.
energy-saving analysis on cooling effect with large temperature 28. ASHRAE. Equipment Guide. Atlanta, American: American Society
difference and high temperature of chilled water system in data of Heating, Refrigerating, and Air Conditioning Engineers; 1983.
center. ISHAVC. 2019;2:933–942. https://doi.org/10.1007/978-
981-13-9524-6_97.
14. Yang Y, Wang B, Zhou Q. Air conditioning system design using
free cooling technology and running mode of a data centre in How to cite this article: Liu J, Su L, Dong K,
Jinan. Proc Eng. 2017;205:3545-3549. https://doi.org/10.1016/j. Sun Q, Shao Z, Huang G. Optimal setting
proeng.2017.09.924. parameters of cooling system under different
15. Cheung H, Wang S. Optimal design of data center cooling sys- climate zones for data center energy efficiency. Int
tems concerning multi-chiller system configuration and com-
J Energy Res. 2021;1–14. https://doi.org/10.1002/
ponent selection for energy-efficient operation and maximized
free-cooling. Renew Energy. 2019;143:1717-1731. https://doi.
er.6499
org/10.1016/j.renene.2019.05.127.

View publication stats