i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3

available at www.sciencedirect.com

w w w . i i fi i r . o r g

journal homepage: www.elsevier.com/locate/ijrefrig

A study on an optimal approach temperature control strategy

of condensing water temperature for energy saving

Chia-Wei Liu, Yew-Khoy Chuah*

Graduate Institute of Mechanical and Electrical Engineering, National Taipei University of Technology, Taipei, Taiwan, ROC

article info abstract

Article history: An optimal approach temperature (OAT) control strategy is proposed for resetting the
Received 12 June 2010 condensing water temperature hourly, so to maximize the performance of the combined
Received in revised form water chiller and cooling tower system. A system performance factor is introduced for
15 September 2010 evaluation of the system performance. A regression function is presented for the calcula-
Accepted 4 January 2011 tion of an optimal condensing water temperature at each hour of air-conditioning opera-
Available online 9 January 2011 tion. The parameters in the regression function include the ambient wet bulb temperature,
the chiller load ratio at the hour, and a dimensionless relative efficiency of chiller and
Keywords: cooling tower. The regression function has an R2 close to 1 compared to the computed
Cooling tower results. When applied to two cities in Taiwan, the OAT control strategy has a potential to
Energy saving save energy more than 4% on an annual basis. The OAT control strategy is most advan-
Temperature tageous when applied to regions with large seasonal variation of wet bulb temperature.
Wet bulb ª 2011 Elsevier Ltd and IIR. All rights reserved.

Etude sur une approche optimale de la stratégie de régulation

de la température de l’eau de condensation permettant
d’économiser de l’énergie
Motsclés : Tour de refroidissement ; Économie d’énergie ; Température ; Bulbe humide

1. Introduction control. However the performance of a cooling tower also

depends on the ambient wet bulb temperature. On this basis,
The control of condensing water temperature is an important an optimal control scheme has to consider the operation of
factor for the performance of an air-conditioning water the air-conditioning system at each hour of the year, either at
chiller. Lower condensing water temperature can be achieved full or partial load of the water chillers.
at the expense of higher cooling tower fan power. Conse- The importance of this subject has attracted the interest of
quently the combined power consumption of chillers and many researchers. Schwedler (1998) presented several exam-
cooling towers has to be considered in the energy saving ples to evaluate the chiller and cooling tower total energy

* Corresponding author. Department of Energy and Refrigerating Air-Conditioning Engineering, 1, Sec. 3, Chung-Hsiao E. Rd. Taipei 10608,
Taiwan, ROC. Tel.: þ886 2 27712171x3508; fax: þ886 2 87733713.
E-mail address: yhtsai@ntut.edu.tw (Y.-K. Chuah).
0140-7007/$ e see front matter ª 2011 Elsevier Ltd and IIR. All rights reserved.
doi:10.1016/j.ijrefrig.2011.01.001
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3 817

Nomenclature Qchr rated chiller capacity (kW)

Qctr rated heat rejection capacity of cooling tower (kW)
a, b, c, d, e, f, g, h, i, j regression coefficients
Qe chillercooling load (kW)
CLR chiller load ratio
SPF chiller-cooling tower system performance factor
CPF chiller performance factor
TLR tower load ratio
CPR chiller power ratio
TPF tower performance factor
CTF cooling tower factor
TPR tower power ratio
CWT condensing water temperature supplied ( C)
WB ambient wet bulb temperature ( C)
Pch chiller power (kW)
Pchr rated chiller power (kW) Subscripts
Pct cooling tower fan power (kW) opt optimal
Pctr rated cooling tower fan power (kW) r rated condition
Qc heat rejection load of cooling tower (kW)

consumption. It was found that the lowest possible condensing annual wide range of wet bulb temperature as typical in
water temperature might not always save the most energy due subtropical regions would provide high potential to save
to high cooling tower fan power. Ahn and Mitchell (2001) used energy by the control of the condensing water temperature.
a quadratic regression scheme to predict the total energy Therefore, it is appropriate to consider chiller and cooling
consumption of an air-conditioning system. In the study they tower as a system for the evaluation of the effects of
found that the ambient wet bulb temperature is an important condensing water temperature control. This study introduces
factor on the optimal supervisory control of the condensing a new optimal control so to maximize the system perfor-
water temperature. Briley (2003) suggested that the operating mance of the chiller and the cooling tower. The optimal
efficiency of a refrigeration system could be increased by approach temperature (OAT) control as proposed would
operating the system at lowest possible condensing water minimize the total power of chiller and cooling tower at each
temperature. Briley suggested condensing water temperature hour using the hourly operating conditions.
can be controlled according to ambient wet bulb temperature.
Crowther and Furlong (2004) used adjustment of condensing
water temperature to measure the total power of chillers and 2. OAT control scheme
cooling tower fans, and used the adjustment to seek an optimal
temperature setting. The results showed for three different The approach temperature is the difference between the
cities in the United States that optimization of chiller-cooling ambient wet bulb temperature and the condensing water
tower was possible to save energy significantly. temperature. The past literature often used a simple fixed
Lu et al. (2004) used a modified genetic algorithm to mini- approach temperature (FAT) control (Crowther and Furlong,
mize the total energy consumption of the condenser water 2004; Ma et al., 2008; Yu and Chan, 2008) that used only
loop. The model considered the interaction constraints of ambient wet bulb temperature as the parameter to control the
cooling tower air flow rate, water flow rate and condensing condensing water temperature. In difference to the FAT
water temperature to optimize all equipment energy. Ma et al. control, the OAT control considers four parameters, namely,
(2008) used the simplified semi-physical models of chiller and ambient wet bulb temperature, chiller load ratio, chiller and
cooling tower to predict the system energy performance under cooling tower fan performance characteristics.
various conditions and condensing water temperature control
strategies. Then Ma et al. developed a model based supervisory 2.1. Chiller and cooling tower energy performance
control strategy using performance map and exhaustive search
(PMES) based technique for online control and operation the Fig. 1 shows the performance data of two types of chillers
system. Morris and Blaine (2008) presented a chiller plant (screw and centrifugal) and also two types of cooling towers
spreadsheet model to predict the energy consumption at any (cross flow and counter flow), provided by international brand
operating conditions. The spreadsheet included optimizer tool suppliers in Taiwan. Power ratios (CPR, TPR) with respect to
to determine the minimum chiller plant energy consumption. the rated power of chillers and cooling towers are shown in
The results showed that lower energy control for the condenser Fig. 1 under two different wet bulb temperatures, with chilled
system (pumps and cooling towers) might have a consequence water setting at 6.7  C.
of much higher energy use for the chillers. Yu and Chan (2008) The performance data at different load ratios and con-
used thermodynamic chiller and cooling tower models to densing water temperatures can be obtained from Fig. 1.
evaluate chiller system COP under different condensing water Variable speed fan modulation is used in Fig. 1 for the
temperature control logics. They also considered the range calculation of the cooling tower power ratio (TPR). In Fig. 1,
temperature of the condensing water kept constant with CLR and TLR are respectively the chiller load ratio and the
varying condensing water flow rate. tower load ratio (with respect to the rated capacity) as
The above discussion indicates that the ambient wet bulb described in Equation (1). Qe is the chiller load, Qchr is the
temperature is an important factor for the optimal control of rated chiller capacity, Qc is the heat rejection load and Qctr is
the condensing water temperature. It is then obvious that the the rated heat rejection capacity of cooling tower. The CPR
818 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3

Screw chiller Centrifugal chiller Cross flow tower Counter flow tower
a WB=15°C b WB=25°C
1 1

T LR=1

TLR=0

T LR= 0
T LR=0

TLR=0

T LR = 0

T LR=1
T LR=0

T LR=0

T LR=0
CLR=1

.6
.2

.8
.2

.4

.6

.8

.4
0.8 0.8 CLR=1
CLR=0.8

CLR=0.8
C PR o r TPR

C PR o r TPR
0.6 0.6
CLR=0.6
CLR=0.6

0.4 CLR=0.4 0.4

CLR=0.4
CLR=0.2
0.2 0.2 CLR=0.2

0 0
16 18 20 22 24 26 28 30 24 26 28 30
CWT (°C) CWT (°C)

Fig. 1 e Performance of chiller and cooling tower at different condensing water temperatures, (a) at wet bulb [ 15  C, (b) at
wet bulb [ 25  C.

and TPR are respectively the ratios of chiller and cooling TLR are all operating conditions, and the controllable factor is
tower power against the rated powers (Pchr and Pctr) as des- CWT. Therefore, the maximum SPF can be obtained by
cribed in Equation (2) resetting CWT according to the operating conditions. Using
the performance data of chillers and cooling towers such that
CLR ¼ Qe =Qchr ; TLR ¼ Qc =Qctr (1)
in Fig. 1, an OAT control scheme is presented to seek for the
maximum SPF as follows:
CPR ¼ Pch =Pchr ; TPR ¼ Pct =Pctr (2)

It can be seen in Fig. 1(a and b) that the power ratios are Step 1: Obtain the cooling load ratio (CLR) of the chillers from
significantly affected by the condensing water temperature the real time operation using Equation (1), usually
(CWT), under different ambient wet bulb (WB) conditions. obtained from measured operating data.
Moreover, similar trends are observed for the different types Step 2: The real time ambient wet bulb temperature is used to
of chillers and cooling towers. It can be seen that the chiller determine the low limit constraint of CWT, as shown
power ratio decreases with lower condensing water temper- in Fig. 1.
ature. In reverse, the cooling tower power ratio increases with Step 3: From the low limit of CWT, obtain CPR for every 0.1  C
lower condensing water temperature. Therefore, these increase in CWT.
different trends of energy performance for the chillers and the Step 4: CPR obtained in step 3 is used with Equation (2) to
cooling towers necessitate an optimal control to achieve calculate chiller power (Pch).
minimum energy consumption. Step 5: Energy balance constraint ðQc ¼ Qe þ Pch Þ is used with
It can also be seen in Fig. 1 that there is a limit for the Equation (1) to calculate TLR, and then tower fan
condensing water temperature. As shown in cross flow cool- power (Pct) with Equation (2).
ing tower for WB at 15  C, the lowest possible condensing
water temperature is 22  C for TLR equals to 1.0. However,
when TLR equals to 0.2, the lowest possible condensing water
5.2
temperature is 17.2  C.

2.2. Computation scheme for OAT control

5
S PF

A system performance factor (SPF) is introduced as Equation

(3). The system performance factor is computed from the
chiller performance factor (CPF) and the cooling tower
performance factor (TPF) as in Equation (4). 4.8 A cross flow tower + A screw chiller
1 A cross flow tower + A centrifugal chiller
SPF ¼ (3)
1=CPF þ 1=TPF A counter flow tower + A screw chiller
A counter flow tower + A centrifugal chiller
CPF ¼ Qe =ðCPR  Pchr Þ; TPF ¼ Qe =ðTPR  Pctr Þ (4) 4.6
25 26 27 28 29 30
It can be noted from Fig. 1 that SPF is a function of wet bulb
CWT (°C)
temperature (WB), condensing water temperature (CWT),
chiller load ratio (CLR), and tower load ratio (TLR), with chilled Fig. 2 e Computation results of SPF at different condensing
water temperature unchanged. In all these factors, WB, CLR, water temperatures.
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3 819

10
Table 1 e Office building composition.
Building type Office
8 Floor area (m )2
18,000
Floor height (m) 4
Predicted OAT (oC )

Floor dimensions (L  W) (m) 30  60

6 Number of floor 10
Air-conditioning schedule Weekdays 8:00 am
R2 = 0.997 to 6:00 pm
4 Construction
U-value of exterior 1.25/1.02/3.2
wall/roof/glazing (W m2 K1)
2 Shading coefficient of glass 0.7
Fenestration ratio 30%
Internal cooling load
Occupancy (m2 person1) 5
0
0 2 4 6 8 10 Lighting power density (W m2) 20
Equipment load density (W m2) 10
Computed OAT (°C) Ventilation rate (L s1 -person1) 8.5

Fig. 3 e Comparison between predicted OAT and computed

OAT.
a range of CWT for four combination cases of different types of
chillers and cooling towers. It can be seen in Fig. 2 that a peak
Step 6: Using Equations (3) and (4) to calculate SPF. value of SPF is obtained for an optimal setting of CWTopt.
Step 7: Iterative calculation from Step 3 to Step 6 for every For example when CWT is controlled at rated temperature
0.1  C increase in CWT is used to seek the maximum 29.4  C, the SPF obtained for the four combination cases are
SPF with respect to the optimal condensing water significantly lower than the peak SPF. When operating at
temperature (CWTopt). TPR ¼ 1.0 for lowest CWT, a combination of a cross flow cooling
tower and a screw chiller, SPF is calculated to be 5.08. However
The above calculation scheme from Step 1 to Step 7 would for the same case the results of the calculation scheme as
then result in maximum SPF with optimal setting of CWT, presented in Fig. 2, the maximum SPF is 5.22 at CWTopt of
using the real time CLR and wet bulb temperature. In the 25.9  C. The optimal approach temperature (OAT) is defined in
calculation scheme, a prescribed low limit of condensing using Equation (5). Therefore this OAT approach is different
water temperature is to be predetermined (or stated by chiller from the common fixed approach control scheme. Optimal
manufacturer). Also, the cooling tower power ratio (TPR) energy performance based on real time operation of chillers
cannot exceed 1.0. Therefore SPF is computed using WB, CLR, and cooling towers are considered in the control scheme.
TLR and CWT. A stepwise increase of 0.1  C is used in the
OAT ¼ CWTopt  WB (5)
computation. When the peak of SPF is reached, it is taken to be
the maximum SPF, with the optimal setting of condensing The above illustration using actual performance data of
water temperature (CWTopt). commercial chillers and cooling towers is to confirm the bene-
An illustration of the above calculation scheme is shown in fits of this control in practice. In practice, a regression function of
Fig. 2 using a case study described later. For operation condi- OAT would be more convenient to incorporate the performance
tions at 20  C WB and 892 kW cooling load, SPF is calculated for data of chillers and cooling towers for commercial application.

a CPF =4.45 b CPF =5.5

7 7
WB=15 C, CTF=1.047
WB=15 C, CTF=1.094
6 WB=15 C, CTF=1.141 6

5 5
SPF

SPF

WB=15 C, CTF=1.058
WB=15 C, CTF=1.116
4 4
WB=15 C, CTF=1.174
WB=25 C, CTF=1.047 WB=25 C, CTF=1.058
WB=25 C, CTF=1.094 WB=25 C, CTF=1.116
3 WB=25 C, CTF=1.141 3 WB=25 C, CTF=1.174

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1

CLR CLR

Fig. 4 e Variation of OAT control SPF with CLR, (a) at CPFr [ 4.45, (b) at CPFr [ 5.5.
820 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3

As mentioned earlier the optimization also rests on the CT,3 CT,2 CT,1
rated power consumption of the cooling tower fan against the
rated chiller power. Therefore, another factor, CTF (cooling
tower factor) is introduced in the computation of OAT. The
definition of CTF is shown in Equations (6) and (7).


CPFr
CTF ¼ 1þ (6)
TPFr
Condenser water loop
CPFr ¼ Qchr =Pchr ; TPFr ¼ Qchr =Pctr (7)
CH,1
In the OAT control scheme, as the heat rejection load (Qc) is
related to the chiller load (Qc ¼ Qe þ Pch ), CLR and TLR are
related. So the optimal condensing water temperature is reset at Chilled water loop
each hour according to the operating WB and CLR. Then the
optimal approach temperature can be predicted using a regres- CH,2
sion function of WB, CLR and CTF as shown in Equation (8).

OAT ¼a þ b  WB þ c  CLR þ d  CTF þ e  WB2 þ f  CLR2 þ g

 CTF2 þ h  WB  CLR þ i  WB  CTF þ j  CLR  CTF ð8Þ
CH,3
The discussion later is for the case of the combination of
screw chiller and cross flow cooling tower as shown in Fig. 1.
Equation (8) was obtained for wide ranges of the variables,
ambient wet bulb from 14 to 30  C, CLR from 0.2 to 1, CTF from Cooling load
1.021 to 1.2134, that include most operation conditions for air-
conditioning. The regression coefficients a, b, c, d, e, f, g, h, i, j are Fig. 6 e Schematic of the water chiller plant system.
respectively 67.5059, 0.0384, 4.3095, 119.6209, 0.0032, 0.3813,
46.5253, 0.1085, 0.3171 and 1.4231. The computed results of
OAT are compared with the predicted results using the
of SPF. In comparison, higher CPFr (COP) would result in better
regression function in Equation (8). The comparison is shown in
SPF. CTF as described in Equation (6) is a parameter that
Fig. 3 with an R2 of 0.997. Therefore it is shown that OAT can be
considers the relative performance of the chiller and the cool-
predicted with Equation (8) so to determine the optimal setting
ing tower. Lower WB would result in higher SPF. Lower CTF that
of condensing water temperature, as CWTopt ¼ OAT þ WB.
corresponds to higher cooling tower fan efficiency would result
Fig. 4 shows the computed results of the maximum system
in better SPF. In general lower SPF is obtained at lower CLR.
performance factor (SPF) for a chiller-cooling tower system
using the OAT control scheme. Equipment efficiency would
naturally affect the maximum possible system performance
3. Case study in subtropical cities
factor as seen in Fig. 4. For a given equipment efficiency (CPFr
and TPFr), a maximum SPF can be obtained with the OAT
A case study is used to compare three other control strategies
control. Ambient wet bulb (WB), chiller load ratio (CLR) and
of condensing water temperature with OAT, namely, fixed
cooling tower factor (CTF) are seen to be the important factors
temperature target at 29.4  C, 18.3  C, and fixed approach
800 temperature of 3.9  C as adopted by Crowther and Furlong
(2004), Ma et al. (2008) and Yu and Chan (2008). A commer-
1 chiller operating 2 chillers operating 3 chillers operating
cial building in Taipei city is used for the case study.

600
3.1. Description of the building case
Op e ra tin g h ou rs

The tropic of cancer cuts across central Taiwan. The Taipei

400 city is situated in northern Taiwan, and exhibits typical
subtropical weather characteristics. An office building was
used in the analysis and the hourly building cooling load was
200 calculated using the DOE-2.1E building energy program
(Winkelmann et al., 1993). Weather data of 2007 was used in
the analysis. The office building is described in Table 1, with
ten floors and a total floor area of 18,000 m2. The building
0
0 500 1000 1500 2000 2500 3000 schedule was weekdays from 8:00 am to 6:00 pm. The indoor
temperature was 26  C (required by the Taiwanese govern-
Building cooling load (kW)
ment for energy saving). The design maximum cooling load
Fig. 5 e Case building cooling load profile and operating was referred to ASHRAE Handbook (2005) using 0.4% annual
hours for Taipei city. cumulative frequency of occurrence. The resulting design dry
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3 821

corresponds to the energy saving baseline of ASHRAE 90.1

Table 2 e Energy efficiency cases for chiller-cooling tower
system. (2004), with the cooling tower fan power at 0.0167 kW per
kW of heat rejection. Chiller COP varied from 4.45 to 5.5 in the
Case Cooling tower fan Chiller CTF
cases. The definition of CTF is described in Equation (6), is
power (kW kW1 heat COP
rejection capacity) a relative power of cooling tower fan.
Fig. 7(a and b) shows the computed results using OAT
Case A 0.00835 4.45 1.047
control strategy for the five cases. Computed results for
Case B 0.00835 5.5 1.058
a week in April and July are presented with wet bulb
Case C 0.0167 4.9 (middle 1.103
level) temperature. SPF is affected by the chiller and cooling tower
Case D 0.0251 4.45 1.141 performance, with Case B having the highest SPF and Case D
Case E 0.0251 5.5 1.174 having the lowest SPF. It can be noted that the variation of SPF
is related to the variation of the ambient wet bulb. Lower wet
bulb and wet bulb temperatures are respectively 34.8  C and bulb generally would result in higher SPF. The observed trend
27.7  C for Taipei city. is due to lower CWTopt setting when the wet bulb is lower. It is
Fig. 5 shows the operating hours for different building also seen in Fig. 7(b) that due to higher wet bulb temperature,
cooling load, and the operation hours of the three chillers. The SPF is generally lower in July compared to that in April.
total operating hours are 2816, with the peak at 2865.8 kW.
Three chillers with cooling capacity of 1000 kW each were 3.3. Annual energy consumption analysis with different
used in the analysis, and each has a heat rejection rate of control strategies
1263 kW. Fig. 6 shows a schematic of the water chiller plant
system. Fixed flow rates of chilled water and condensing Annual energy consumption analysis is necessary when OAT
water were used according to that of ARI (2003) standards. is compared to other control strategies due to annual wet bulb
For the OAT control strategy, the regression function as variation. Table 3 shows the comparison of annual energy
given in Equation (8) was used to seek the optimal hourly reset consumption of chiller-cooling tower system using different
of CWTopt. The lower limit of the condensing water temper- CWT control strategies for the commercial building in Taipei
ature was taken to be 18.3  C, which corresponds to the low city. Fixed target of condensing water temperature at 29.4  C
limit of condensing water temperature as given by ARI (2003) (design temperature for chiller) is used as the baseline. It is
rated conditions. Therefore with the regression function as observed in Table 3 that OAT control has the best energy
given in Equation (8) the calculation of maximum SPF is given saving potential, at about 4.6e5.9%. For OAT control, the
as in Equation (9). difference in energy use between Case B (best case) and Case D

8
< Calculate OAT at each operating hour using Equation ð8Þ
max SPF ¼ CWTopt ¼ WB þ OAT (9)
:
If CWTopt < 18:3  C then CWTopt ¼ 18:3  C

(worst case) is about 215,114 kWh yr1. Therefore, chiller and

3.2. SPF analysis for OAT control strategy cooling tower performance is generally a key factor in energy
use.
All together five cases of different chiller and cooling tower Although fixed temperature target at 18.3  C would provide
performances were studied, as shown in Table 2. Case C lowest possible condensing water temperature, but the

OAT C ase A OAT C ase C OAT C ase E

OAT C ase B OAT C ase D WB

a April b July
7 32 6

28 32
6
5

24
Temp. ( C)

Temp. ( C)

5
SPF

SPF

28
o

4
20
4

16
3
3 24
12
Mon. Tues. Wed. Thur. Fri. Mon. Tues. Wed. Thur. Fri.

Fig. 7 e Weekly variation of SPF in April and July, (a) April, (b) July.
822 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3

Table 3 e Comparison of annual energy consumption of chiller-cooling tower system using different CWT control
strategies in Taipei city.
Case CWT control Tower Chiller Total Energy saving
strategy (kWh yr1) (kWh yr1) (kWh yr1) (%)

Case A Fixed29.4 19,784 922,333 942,117 Baseline

Fixed18.3 56,356 835,869 892,225 5.30
FAT 36,657 854,149 890,806 5.45
OAT 37,466 849,420 886,886 5.86
Case B Fixed29.4 18,495 746,113 764,608 Baseline
Fixed18.3 56,306 673,985 730,291 4.49
FAT 34,833 690,465 725,298 5.14
OAT 31,642 689,796 721,438 5.65
Case C Fixed29.4 38,329 837,548 875,877 Baseline
Fixed18.3 112,667 757,858 870,525 0.61
FAT 71,558 775,363 846,921 3.31
OAT 43,814 790,418 834,232 4.75
Case D Fixed29.4 59,470 922,333 981,803 Baseline
Fixed18.3 169,406 835,869 1,005,275 2.39
FAT 110,192 854,149 964,341 1.78
OAT 55,168 881,384 936,552 4.61
Case E Fixed29.4 55,595 746,113 801,708 Baseline
Fixed18.3 169,254 673,985 843,239 5.18
FAT 104,708 690,465 795,173 0.82
OAT 48,230 715,934 764,164 4.68

cooling tower has to operate at high TPR (tower power ratio).

6 Higher TPR may more than offset higher chiller efficiency at
low setting of condensing water temperature. The cooling
4 tower fan power is higher in Case D and Case E. Therefore
fixed temperature target at 18.3  C would result in higher
Ene rgy sa vi ng (%)

2 energy use in Case D and Case E.

Fixed approach has been described in the literatures as an
0 effective energy saving control. It is observed in Table 3 that
Baseline: Fixed29.4 control at each case
FAT is superior to fixed temperature target of 29.4  C.
-2 However, FAT would not perform well at higher cooling tower
Fixed18.3 fan power. This result is due to no consideration of the cooling
-4 FAT tower fan power consumption.
OAT It is clear that OAT control is superior to the above control
-6 strategies as significant energy saving is achieved in all the
cases. The basic principle of OAT is to achieve the best SPF.
A B C D E
Therefore, condensing water temperature is reset in the hourly
Case
operation of the chiller-cooling tower system. As described
Fig. 8 e Comparison of annual energy saving potential above, the chiller power and cooling power at different chiller
using different CWT control strategies in Kaohsiung city. loads are considered together with the ambient wet bulb, so to
seek for the maximum system performance factor (SPF).

Table 4 e Average monthly wet bulb of four subtropical cities.

City Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Avg. wet bulb temperature ( C)
a
Hong Kong 13.5 14.2 17.0 20.6 23.5 25.7 25.9 25.8 24.8 21.5 17.7 13.4
(Typical year)
Miamia 17.0 17.2 17.6 19.4 22.4 23.2 24.4 24.0 23.8 21.9 19.2 16.9
(Typical year)
Taipeib 14.6 15.4 16.9 17.8 21.9 24.7 25.3 24.9 24.2 21.3 17.8 16.0
(2007 measured)
Kaohsiungb 16.0 17.7 19.8 21.1 23.9 25.2 26.2 25.2 25.0 23.3 19.8 18.1
(2007 measured)

a Typical year weather data for Hong Kong and Miami obtained from http://www1.eere.energy.gov/buildings.
b 2007 hourly measured weather data of Taipei and Kaohsiung.
 i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 8 1 6 e8 2 3 823

3.4. Apply OAT control strategy to another subtropical strategy of cooling systems in climatic regions with large
city variation of ambient wet bulb.

Kaohsiung city is in the southern Taiwan, with weather

generally warmer than Taipei. The results of an attempt to
Acknowledgment
apply OAT control for the same building case with 2007
weather data of Kaohsiung are presented in Fig. 8. The
This research was supported by the National Science Council
regression function in Equations (8) and (9) was applied in the
of Taiwan and is grateful acknowledged.
analysis. As seen in Fig. 8, OAT control is superior compared to
the other control strategies. Similarly it is observed in Fig. 8
that OAT control has significant energy saving potential for references
all cases, lies between 4.1 and4.8％. The lower energy saving
compared to Taipei is due to annual average higher wet bulb
temperature in Kaohsiung. Ahn, B.C., Mitchell, J.W., 2001. Optimal control development for
Therefore, OAT is best use in subtropical regions with chilled water plants using a quadratic representation. Energy
annual wide variation of ambient wet bulb temperature. Build. 33, 371e378.
There are many typical subtropical cities outside Taiwan, ARI Standard 550/590, 2003. Performance Rating of Water Chilling
Packages Using the Vapor Compression Cycle. Air-
such as Hong Kong and Miami. Table 4 presents typical
Conditioning and Refrigeration Institute, Arlington, Virginia.
average monthly wet bulb temperature for four subtropical
ASHRAE Standard 90.1, 2004. Energy Standard for Buildings
cities. It is observed in Table 4 that the four cities have similar except Low-Rise Residential Buildings. American Society of
trends of wide variation of ambient wet bulb temperature. The Heating, Refrigerating and Air-Conditioning Engineers, Inc.,
wet bulb temperature is generally higher in summer and Atlanta, USA.
lower in other seasons. Therefore the OAT scheme as pre- ASHRAE Handbook, 2005. Fundamentals. American Society of
sented can be widely applied in energy saving control of air- Heating, Refrigerating and Air-Conditioning Engineers, Inc.,
Atlanta, USA.
conditioning systems.
Briley, G.C., 2003. Increasing operating efficiency. ASHRAE J. 45,
73.
Crowther, H., Furlong, J., 2004. Optimizing chillers and towers.
4. Conclusion ASHRAE J. 46, 34e40.
Lu, L., Cai, W.J., Soh, Y.C., Xie, L.H., Li, S.J., 2004. HVAC system
An OAT control strategy which maximizes the system optimizationecondenser water loop. Energ. Convers. Manag.
performance factor of the chiller-cooling tower system is 45, 613e630.
Ma, Z.J., Wang, S.W., Xu, X.H., Xiao, F., 2008. A supervisory control
presented. It has been found that OAT control strategy is
strategy for building cooling water systems for practical and
advantageous compared to presently used strategies for real time applications. Energ. Convers. Manag. 49, 2324e2336.
condensing water temperature control. A regression function Morris, T., Blaine, S., 2008. Chiller plant optimization tool.
is presented for the OAT control scheme, which can be used ASHRAE J. 50, 54e60.
to reset condensing water temperature at each hour accord- Schwedler, M., 1998. Take it to the limit or just halfway. ASHRAE J.
ing to the operating conditions of the air-conditioning 40, 32e39.
Winkelmann, F.C., Birdsall, B.E., Buhl, W.F., Ellington, K.L.,
system. The energy saving potential of the OAT control has
Erden, A.E., 1993. DOE-2 Supplement Version 2.1E. Lawrence
been analyzed for a commercial building for two typical
Berkeley Laboratory, Berkeley, California, USA.
subtropical cities in Taiwan. The results show that the OAT Yu, F.W., Chan, K.T., 2008. Optimization of water-cooled chiller
control has a potential to save energy more than 4% annually. system with load-based speed control. Appl. Energy 85,
The OAT control can be generally applied in the operating 931e950.