Professional Documents
Culture Documents
www.elsevier.com/locate/apenergy
Received 15 July 2003; received in revised form 21 August 2003; accepted 6 September 2003
Abstract
A simplified model for indirect cooling towers behaviour is presented. The model is devoted
to building simulation tools and fulfils several criteria such as simplicity of parameterisation,
accuracy, possibility to model the equipment under various operation conditions and short
computation time. On the basis of Merkel’s theory, the model is described by using the
Effectiveness-NTU method. The model introduces only two parameters, air-side and water-
side heat-transfer coefficients which can be identified from only two rating points, data easily
available in manufacturers’ catalogues. Thus, the model allows one to estimate energy and
water consumptions under different operating conditions such as variable wet-bulb tempera-
tures or variable airflow rates.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Closed cooling tower; Heat and mass transfer; Modelling; Air conditioning; Variable air-flow
rate; Water consumption
1. Introduction
Indirect cooling towers are commonly used in industrial processes and in air-
conditioning systems to reject heat to the atmosphere. An indirect cooling tower is a
device which uses heat and mass transfer to cool water from a process. The hot
water goes through serpentine tubes arranged in rows, whereas the air passes on the
external side of this coil. An additional circuit sprays water to cool further the coil
by evaporation (Fig. 1). The main advantage of this system compared with an open
cooling-tower is that the contamination risks with airborne dusts and corrosion are
0306-2619/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2003.09.004
434 P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451
Nomenclature
A surface (m2)
c:p specific heat (J kg1 K1)
C capacity rate (W K1)
d tube diameter (m)
e: thickness (m)
G: a air mass velocity based on minimum section (kg m2 s1)
Gspray flow rate of spray water per unit breadth (kg m1 s1)
h enthalpy (J kg1)
hc heat-transfer coefficient (W K1 m2)
hfg water vaporisation heat at 0 C (J kg1)
hm mass-transfer coefficient (kg m2 s1)
:
m mass flow rate (kg s1)
NTU Number of Transfer Units
r radius (m)
Re Reynolds number
T dry-bulb temperature (K)
T0 wet-bulb temperature (K)
U heat-transfer coefficient (W/K)
w humidity ratio (kg kg1)
effectiveness
l conductivity (W K1 m1)
air density (kg m3)
relative humidity (%)
Indices
a humid air
ext external side
f water film (spray)
i inlet
int internal side
o outlet
v water vapour
w water
limited since the process water never contacts the outside air. The main drawback
compared to an open cooling-tower is that the cost and the size are increased
since a larger heat-exchange surface is required to reach the same heat transfer.
In order to reduce water consumption, some closed cooling towers can operate in
a dry regime when the outdoor conditions are favourable and the cooling demand
is low.
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 435
In the literature, several models fulfil the requirement of accuracy and running
time. Facão [1] compared simplified models [2,3] with detailed models such as
numerical models [4,5] and noticed that simplified models based on an overall
approach provide as good or even better results as those based on finite differences.
Notice that other simplified models have been developed such as the unified model
for evaporative cooling devices [6]. Most of the simplified models are built on the
basis of Merkel’s theory assuming a Lewis number equal to unity and neglecting
the losses due to water evaporation. Additional hypothesis are often taken into
account such as:
The aim of this paper is to adapt a simplified model for analysing the combined
heat-and-mass transfers in indirect cooling towers which can (i) be easily para-
meterised through performance data available in manufacturer’s catalogues, (ii)
evaluate the tower performance under different operating conditions, notably vari-
able airflow rate and (iii) assess the water consumption.
2. Theoretical model
In this section of the paper, the basic equations of the mass and heat transfer
which occurs in the systems using evaporative cooling are stated. The main
assumptions are:
The heat exchange between the cooling tower and the surroundings is
negligible.
The specific heats of the fluids are assumed to be constant.
The mass and heat transfers take place only in the direction normal to the
flow.
The fluids at entry are uniformly distributed in the plane perpendicular to the
flow.
The water film covers all the wall separating the air from the water.
The air and water flows are uniformly distributed in the plane perpendicular
to the flow.
The interface temperature between water film and air is assumed to be equal
to the water-film temperature.
Fig. 2. Scheme of the heat and mass transfers in the indirect cooling tower.
water spray, there is a film of saturated air, in close contact with the water. This film
of saturated air is at the average temperature of the water spray film in this small
element of volume. Since the water-vapour partial-pressure at this interface is higher
than the water-vapour partial pressure in the air, there is a transfer of water-vapour
towards the air. This mass transfer brings a heat transfer related to the water-
vaporisation, called transfer of latent heat. At the same time, because of the differ-
ence in temperature between surface of the water and the air, there is transfer of heat
by convection.
The theoretical model is based on heat-and-mass transfer equations and on heat
and mass balance equations which are established for the steady state.
The mass transfer between the water film and the air flow can be written as:
:
m a dwa ¼ h m wsat Tf wa dA ð1Þ
438 P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451
The mass balance which occurs between the water film and the air flow gives:
: :
m a dwa ¼ dmf ð2Þ
The heat transfer between the water film and the air flow can be written as:
:
m a dha ¼ h m wsat Tf wa dA hfg þ cpv Tf þ h c Tf Ta dA ð3Þ
The enthalpy balance on the overall elementary volume can be expressed as:
: : : : :
m a dha þ mw dhw ¼ mf dmf dhf þ dmf hf ð4Þ
The numerical solution of these equations requires input parameters such as the
heat-and-mass transfer coefficients, the diameter of the tubes, the number of tubes,
which are generally not accessible for simulation users. In addition, the computation
time of one rating point makes this type of resolution completely unsuited to the
constraints of energy simulation programming. The models founded on the simpli-
fication of heat-and-mass transfer equations seem more suitable to fulfil the fixed
requirements.
On the water side, the heat transfer between the water and the water film can be
aggregated through an electric analogy by introducing a overall heat transfer coef-
ficient (Fig. 3). It results in:
:
dQ ¼ Uint dAint Tw Tf ð6Þ
In order to have the same potential in Eqs. (5) and (6), one can express enthalpy as
a function of temperatures [8] in Eq. (5) or introduce enthalpy instead of temper-
atures [9] in Eq. (6). Since the characteristics of the air side are generally limited to
the wet-bulb temperature in the performance catalogues of manufacturers, one
chooses to modify Eq. (5) in order to get the wet-bulb temperature.
Assuming that iso wet bulb temperature curves intermingle with isenthapic curves,
the enthalpy can be rewritten as:
ha ¼ cpa Ta þ wa hfg þ cpv Ta ffi h T 0a ¼ cpa T 0a þ wsat T 0a hfg ð7Þ
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 439
Likewise, the enthalpy of the saturated air film can be written as follows:
hsat Tf ¼ cpa Tf þ wsat Tf hfg ð8Þ
Then, if one assumes that the enthalpy is linear over small ranges of temperature:
ha ¼ ho þ cpsat T0a ð9Þ
The Eq. (5) can be rewritten as:
: h c;ext dAext
dQ ¼ cpsat Tf T0a ð10Þ
cpa
with
1 1 1 cpa 1
¼ þ ¼ þ ð12Þ
Ut dAt Uext dAext Uint dAint hc;ext cpsat dAext Uint dAint
The heat exchange between water and air is now represented by an overall heat-
transfer coefficient. One can apply the effectiveness-NTU relations of classical heat
exchangers.
Table 1
Effectiveness-NTU relations in wet and dry regimes
1 cpa
¼ hwet þ Uwet1Aint 1
¼ 1
þ 1
Ut A t c;ext cpsat Aext int
Ut At hdry
c;ext Aext Udry A
int int
The effectiveness of closed cooling towers is written as the ratio of the actual heat
transfer rate to the maximum possible heat transfer rate:
:
Q
"¼ : ð15Þ
Qmax
The closed cooling tower coils are generally staggered rows of finned or unfinned
circular tubes. The finned tubes are used for cooling towers which can operate in a
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 441
dry regime. First, one considers staggered tubes without fins. In the literature, many
correlations have been established [1,10–12] for mass transfer coefficient. Misushina
and Miyashita [12] proposed a correlation for 12004Rea 414,000, 504Respray
4240 and 124dext 440 mm :
h m a ¼ 5:028 108 Re0:9 0:15 2:6
a Respray dext ð18Þ
:
Gspray
When dext > 0:7, Niitsu et al. [10] noticed that the mass-transfer coefficient was
independent of the spray flow rate. Most of the correlations [1,10,11,13] express the
mass-transfer coefficient in a more simple form than Eq. (18):
: n
h m ¼ C Ga ð19Þ
The parameter, C, lies between 0.049 and 0.076 and the exponent, n, lies between
0.773 and 0.905. The mass-transfer coefficient varies essentially with the air mass
flow rate. By considering the correlations, a mean exponent can be set to 0.8. So, one
can reduce the air-side heat-transfer coefficient as:
:
UAext ¼ ext cpsat m0;8
a ð20Þ
where ext is a constant which depends on the thermal properties of air and on the
coil’s geometry.
Let us now consider a finned coil. The air-side heat-transfer coefficient is modified
and can be expressed as:
UAext ¼ hc;ext ðAtube þ
fin Afin Þ ð21Þ
The fins are usually circular and their effectiveness is:
sffiffiffiffiffiffiffiffiffiffiffiffiffi
tanhðLÞ hc;ext
fin ¼ with ¼ 2 ð22Þ
L lfin efin
rext
ln
1 1 rint 1
¼ þ þ film ð23Þ
UAint hwc Aint l hc Aext
Aint
rint
One can neglect the conductivity through the tubes compared to convection in the
film and in the tube. So Eq. (23) becomes:
In the tubes, one considers the correlation of Dittus Boetler [15] which is valid for
Re > 104.
lw
hwc ¼ 0:023 Re0:8 Pr0:3 ð25Þ
dint
At smaller Reynolds numbers, the convection coefficient will be overpredicted by
this correlation.
Correlations of heat-transfer coefficient between tube surface and water film have
been established [1,10,12,13]. Most of them are expressed as follows:
: !n
G spray
hfilm
c ¼C ð26Þ
dext
Furthermore, on the water temperature range (15–50 C), the water viscosity
varies a lot whereas the conductivity, the density and the specific heat of water are
almost constant (Table 2). The variation of viscosity cannot be neglected a priori.
Finally, the water-side heat-transfer coefficient is expressed as follows:
0:8 1
UAint ¼ int mw ð29Þ
0:5
w
where int is a constant which is influenced by the coil’s geometry and constant
water-properties.
with:
Two T 0ai Twi T 0ao
DTlm ¼ DTLM ¼ ð31Þ
Two T 0ai
ln
Twi T0ao
Once both overall heat-transfer coefficients are calculated, the parameters repre-
senting the internal and the external heat-transfer coefficient are extracted using the
following equation:
1 1 e0;5
¼ : þ : ð32Þ
UAt ext cpsat m 0;8
a int m0;8
e
Table 2
Liquid water properties [16]
Table 3
Tower characteristics and studied domain
Thanks to two rating points with different water flow rates, one determines the
parameters int and ext. In manufacturers’ catalogues, one finds performance data
for different water-flow rates but it is rare to find data for variable air-flow rates.
Once the model is parameterised, the performance of the closed cooling-tower can
be calculated under different inlet conditions and variable air-flow rates.
4. Validation
The model has been compared to manufacturers’ data [17,18]. Three closed cool-
ing towers from two manufacturers have been studied (Table 3). The three coils are
unfinned and with staggered tube ranks. The first one is equipped with a centrifugal
fan placed under the coil, the second one with a centrifugal fan located on the side of
the coil. In this last case, the arrangement aims to limit the height of the cooling tower.
The air is still extracted at the top of the cooling tower and one assumes a counter-
flow exchange. The last one is equipped with an axial fan with data for two speeds.
The model is tested on the studied domain of inlet conditions, at different water-
flow rates and at a variable air-flow rate for the third cooling tower.
The developed model has been compared on 18 rating points at different wet-bulb
temperatures, inlet-water temperatures and water-flow rates for the first cooling-
tower. Twelve rating points have been used for the second cooling-tower and 25 for
the last one. The model has been parameterised for each tower using two of these
rating points. The choice of these parameterisation points plays a role on the accu-
racy of the model. However, the accuracy of the model is stable provided that the
water-flow rate ratio between both rating points is lower than 0.6. The results
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 445
between manufacturers’ data and the model are given in Figs. 4–7. The model leads
to maximum errors in heat-transfer rate of 4.1, 6.1 and 18.7% respectively for the
first, second and third cooling towers. The standard deviations of the outlet-water
temperature are respectively, 0.21, 0.26 and 0.32 C. For the third cooling tower, the
heat-transfer rate deviation exceeds 10% for a few points which correspond to low
Fig. 4. Error between the simplified and the detailed water-side heat-transfer coefficients.
Fig. 5. Comparison between the manufacturers’ data and the model for the first cooling tower.
446 P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451
Fig. 6. Comparison between the manufacturers’ data and the model for the second cooling-tower.
Fig. 7. Comparison between the manufacturers’ data and the model for the third cooling-tower.
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 447
Data with a fan rotation speed at 3=4 of nominal speed are provided for the third
cooling-tower. The results of the comparison with the model are presented in Fig. 8.
The standard deviations are 9.3% for the heat-transfer rate and 0.32 C for the
outlet-water temperature. The model provides results in good agreement with
manufacturers’ data at variable air-flow rates.
The cooling in the tower is mainly due to the evaporation of the water spray on
the tubes. The water losses represent about 1–4% of the spray-water-flow rate. The
drift losses can be neglected since they represent generally less than 0.2%. Since one
part of the water evaporates, it is necessary to bleed the tower in order to limit the
concentrations of impurities. The waste of a small percentage of circulating water is
expressed by the concentration ratio:
ðevaporation þ blowdownÞ
concentration ratio ¼ ð33Þ
blowdown
Fig. 8. Comparison between the manufacturers’ data and the model for the third cooling-tower at low
air-flow rates.
448 P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451
The water consumption is not negligible and it is necessary to estimate it. In the
calculation of water consumption, the evaporation rate and the concentration ratio
are required. The evaporation rate depends on the air and water inlet conditions.
In order to determine the evaporation rate, it is required to know the outlet-air
conditions. The effectiveness models provide only either the outlet wet-bulb tem-
perature or the outlet enthalpy. However, the outlet humidity-ratio is needed too.
An approximate method would consist in assuming that the outlet air is saturated
knowing that the air is often close to saturation. A second method consists in inte-
grating Eqs. (1) and (3). The outlet humidity ratio is given by:
!
h m Aext
wao ¼ wms þ ðwai wms Þexp : ð34Þ
ma
UAext
wao ¼ wms þ ðwai wms Þexp : ð35Þ
ma cpsat
Fig. 9. Determination of the mean humidity-ratio at the interface between the air and water.
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 449
h hai
hms ¼ hai þ ao : ð36Þ
1 exp UAext =ma cpsat
The point (hms, wms) corresponds to a fictitious point on the saturation line repre-
senting the mean conditions of air film at the interface between the air and the water.
The mean humidity ratio of the saturated air-film is determined by crossing the
mean air-film enthalpy with the saturation line (see Fig. 9).
A comparison between the approximate method and the integration method for
different inlet-conditions shows that the simplified method overestimates the eva-
poration rate by 7% on average and 17% at maximum. The comparison has been
carried out on 13 rating points for a cooling tower with inlet-water temperatures
lying between 25 and 45 C by steps of 5 C, wet-bulb temperatures from 16 to 24 C
by steps of 4 C and a dry-bulb temperature of 30 C. The case with an inlet water
temperature of 25 C is calculated with a 16 C wet-bulb temperature only. The air-
flow rate in the tower was 2.5 m3/s. The outlet relative humidity ranges between 92
and 96% except for two points below 90%.
In consideration of the uncertainty of the concentration ratio, the approximate
method can be adequate to estimate water consumption.
Fig. 10. Comparison between manufacturers’ data and the model for the second cooling-tower in the dry
regime.
450 P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451
side). Eqs. (20) and (29) will be changed through the coefficients int and ext and the
replacement of cpsat by cpa in Eq. (20). One can expect that the internal heat-transfer
coefficient will increase whereas the external coefficient will decrease.
The study of water film resistance shows that its order of magnitude is similar to
the convection resistance in the tube. In the dry regime, the water-film resistance is
suppressed and one can assume that internal heat-transfer coefficient is multiplied by
2. By studying many rating points from manufacturers’ catalogues for the dry
regime, one notices that ext is reduced by 65% such as one can propose:
dry humid
int ¼ 2 int and dry humid
ext ¼ 0:35 ext ð37Þ
The model has been parameterised with two rating points in the wet regime. Then,
the model has been validated for the dry regime by comparison with manufacturers’
data (Fig. 10). The relations (37) have been used for calculating the performance of
the tower in the dry regime. The standard deviation in heat-transfer rate is 3.4%
and the standard deviation in temperature is 0.17 C. The results are in good
agreement with manufacturer’s data however the corrective factor should be vali-
dated on other cooling towers to be generalised. Unfortunately, data are scarce in
this domain.
6. Conclusions
References
[1] Facão J, Oliveira AC. Thermal behaviour of closed wet cooling-towers for use with chilled ceilings.
Applied Thermal Engineering 1999.
[2] Peterson JL. An effectiveness model for indirect evaporative coolers. ASHRAE Transactions 1993;
99(2).
[3] Mizushina RIT, Miyashita H. Experimental study of an evaporative cooler. International Chemical
Engineering 1967;7(4).
[4] Mizushina RIT, Miyashita H. Characteristics and methods of thermal design of evaporative coolers.
International Chemical Engineering 1968;8(3).
P. Stabat, D. Marchio / Applied Energy 78 (2004) 433–451 451