Energy Efficient Buildings

Cooling Towers
Introduction
A cooling tower is a counter-flow or cross-flow heat exchanger that removes heat from water
and transfers it to air. Cooling towers come in many configurations. An induced-draft cooling
tower, which is common in HVAC and industrial applications, is shown in Figure 1a. As warm
water from the process falls through the tower, some of it evaporates, which cools the
remaining water. The cooled water collects at the bottom of the cooling tower and is returned
to the plant where it is used for cooling. Figure 1b shows an evaporative condenser, which is
common in industrial refrigeration applications. Water, which is cooled by evaporation, falls
over a closed heat exchanger (usually carrying refrigerant) in the top part of the tower. It then
falls over more fill to enhance evaporation in the lower part of the tower. A small pump
circulates water from the bottom to the top of the tower.
HOT
WATER
IN

HOT
WATER
IN
WARM AIR OUT
Hot Water
Distribution

AIR IN
Fill

Fill

AIR IN

Air Inlet
Louvers
Sump

COLD WATER OUT

Figure 1a) open circuit cooling tower

Hot Water
Distribution

WARM AIR OUT

Refrigerant Liquid Out

Heat Exchange Coil

Refrigerant Vapour In

Fill
AIR IN

Secondary
Recirculating
Pump

Air Inlet
Louvers

1b) closed circuit evaporative cooling tower

The temperature difference of water through a tower, dT = Tw1-Tw2, is determined by the
load, Ql, and the mass flow rate of water, mw. Neither the size of the tower nor the state of
the outside air influences the temperature difference; however, larger towers or lower outdoor
air wet-bulb temperatures will decrease the exit water temperature, Tw2.

Sensible and Latent Cooling

Depending on the entering air and water temperatures, the water may be cooled by sensible
and latent cooling of the air, or simply by latent cooling of the air. In either case, latent, i.e.
evaporative, cooling is dominant. For example, consider the case in which the air enters at a
lower temperature than the water (Figure 3a). The air will leave completely saturated and the
cooling is part sensible and part latent. The sensible portion occurs as the air temperature
increases by absorbing heat from the water. The latent portion occurs as some of the water
evaporates, which draws energy out of the water.
If the air enters at the same wet bulb temperature as before, but at a higher dry-bulb
temperature than the water, then the air will cool as it saturates (Figure 3b). Thus, the sensible
2

cooling component is negative, and the all the cooling is due to evaporation. In general, cooling
is dominated by latent cooling.

A2

A 2
Qlat

Qtot

Qsen

A1

A1

A2

A 2
Qlat

Qtot

-Qsen

A1
A1

Figure 2. Psychrometric process lines for air through a cooling tower, if the entering air
temperature is a) less than the entering water temperature, and b) greater than the entering
water temperature.
The total cooling, ma (ha2 ha1) is the same for both cases since enthalpy is a function of wetbulb temperature alone. However, the dry-bulb temperature significantly influences the
evaporation rate, mwe = ma (wa2-wa1). The rate of evaporation increases as the dry-bulb
temperature increases for a given wet-bulb temperature.

Cooling Towers as Heat Exchangers

Based on the previous discussion, it is clear that cooling tower performance is a function of the
wet-bulb temperature of the entering air. In an infinite cooling tower, the leaving air wet-bulb
temperature would approach the entering water temperature, and the leaving water
temperature would approach the web-bulb temperature of the entering air. The difference
between the leaving water temperature and the entering air wet-bulb temperature is called the
approach. The relationship between air wet-bulb and water temperature is shown in the figure
below. In an infinite cooling tower, the approach would be zero.

Source: ASHRAE Handbook, HVAC Systems and Equipment, 2004.

Neglecting fan power and assuming steady state operation, an energy balance on a cooling
tower gives:
mw1 cpw Tw1 mw2 cpw Tw2 + ma (ha1 ha2) = 0
Assuming steady state operation, a mass balance on water flow gives:
mw1 mw2 + ma (wa1 wa2) = 0
mw2 = mw1 + ma (wa1 wa2)
Substituting mw2 into the energy balance gives:
mw1 cpw Tw1 [mw1 + ma (wa1 wa2)] cpw Tw2 + ma (ha1 ha2) = 0
mw1 cpw Tw1 mw1 cpw Tw2 - ma (wa1 wa2) cpw Tw2 + ma (ha1 ha2) = 0

The fraction of incoming water that is evaporated, ma (wa2-wa1) / mw1, is typically less than
1%. Thus, ma (wa1 wa2) is much less than mw1, and the term ma (wa1 wa2) cpw Tw2 can
be neglected with negligible error to give:
mw1 cpw (Tw1 Tw2) = ma (ha2- ha1)
Both sides of this equation represent the total cooling capacity of the tower.
The effectiveness, E, of a heat exchanger is the ratio of the actual to maximum heat transfer.
E = Qactual / Qmax
For a heat exchanger, Qmax occurs if the air leaves the cooling tower completely saturated at
the temperature of the incoming water. Thus, effectiveness is
E = Qactual / Qmax = [mw1 cpw (Tw1 Tw2)] / [ ma (ha,sat,tw1- ha1)]

Energy Efficiency of Counterflow and Crossflow Towers

The two most common tower designs for HVAC applications are forced-air counterflow and
induced air cross-flow. Cooling tower energy use is a function of fan and pump power. To
generate the same quantity of cooling, forced-air counterflow towers require more fan and
more pump energy then induced-air crossflow towers. Thus, induced-air crossflow towers are
almost always more energy efficient.

Forced-air counterflow towers require more fan energy because centrifugal fans are made to
generate low flow against high pressure, but cooling towers generally need high flow at low
pressure. In comparison, induced air crossflow towers use propeller fans, which generate high
flow against low pressure, which is more suited to cooling towers.
Forced-air counterflow towers require more pump energy because these towers are taller in
order to facilitate the counterflow heat transfer as the water falls through the tower. This
height increases elevation head in the piping system. In addition, forced-air counterflow
towers spray water through nozzles, which increases pressure drop. In comparision, inducedair crossflow towers are shorter and wider since the path of the air through the water is
horizontal. In addition, the supply water simply drains from feeding pans into fill, which
eliminates the need for nozzles.
A comparison of cooling tower energy use for the same loads is shown below.

Comparison of F.D. Blower Tower vs I.D. Propeller Tower for 400 Tons
Cooling
Operating
Fan
Tower
Additional
Total
Tower
Fan Motor
Motor
Pump Head Pump Motor
Operating
Type
HP
KW (1)
FT. (2)
KW (3)
KW
Counterflow
with Blower
40
32.4
23
6.9
39.3
Crossflow
With Propeller
20
15.2
10
3.0
19.2
Source: Marley Technical Report H-001A, Cooling Tower Energy and Its Management,
October, 1982.

Cooling Tower Control

In HVAC applications, chiller evaporator loads vary depending on weather and building
occupancy, and the quantity of heat rejected by the condenser varies accordingly. The cooling
tower will always reject the all the heat from the condenser. However, the temperature of the
cold water return to the condenser will decline at lower loads.
Various methods are used to control cooling tower capacity to generate the desired cold water
return temperature. The two control points for cooling towers are water flow and air flow.
However, cooling tower manufacturers strongly recommend that water flow remain constant
at all times. Thus, primary control methods generally rely on varying air flow. The common
control methods are listed below.
Run Fans Continuously
This type of control results in the coldest possible return water temperature, which reduces
chiller energy use. However, it also results in the highest cooling tower fan energy use.
Because the improvement of chiller efficiency with lower condenser water temperature is
asymptotical at some minimum temperature, this method of control rarely results in the best
overall energy efficiency.
Cycle Fans On and Off
This type of control reduces excess fan energy use at cold outsider air temperatures, and is
widely used. At relatively cold temperatures, however, the fan may cycle on and off too
frequently. The maximum number of fan cycles is about 8 per hour. Thus, many cooling towers
are equipped with water bypass loops. In most applications, water bypass control is only used
at low temperatures when fan cycling could be a problem.
Use Two-Speed Fan
This method of control adds an intermediate level of cooling between full-on and full-off. This
results in considerable fan energy savings, since fan energy varies with the cube of flow. Thus,
fan energy at 50% air flow is only 12% of the fan energy at full air flow. This type of stepped
7

control can be further extended with two cell towers with one fan in each cell. This leads to
four possible steps of control. A typical relationship between cold water temperature and fan
flow is shown below.
Continuously Control Fan Speed with VSD
This method results in the lowest fan energy use by continuously achieving savings, due to the
fan law that fan energy varies with the cube of flow.
Vary Air Flow Using Inlet Air Dampers
Before VSDs, cooling towers were sometimes controlled by running the fan at full speed while
varying the inlet air dampers to modulate air flow. This method of control results in
intermediate energy savings between fan cycling and continuous VSD control. However, is
rarely used now that the VSD control is now commonplace.

Comparison of Energy Use with Various Methods of Cooling Tower Control

Total chiller and cooling tower energy use for these control methods for a typical HVAC
application are shown below.
Comparative Energy Usage with Various Methods of Control
Propeller
Blower
Operating
Hours of
Average KW Fan Energy
Fan Energy
Situation
Operation
Usage
(kW hr)
(kW hr)
Constant Operation
P = 16.2
at Full Capacity
1202.2
B = 32.4
19475.6
38951.2
Single-Speed
P = 765.3
P = 16.2
Fan Cycling
B = 852.7
B = 32.4
12397.9
27627.5
Two-Speed
P = 1132
P = 4.3
Fan Cycling
B = 1146
B = 8.55
4867.6
9798.3
Variable Control
P = 2.72
at Constant Speed
1202.2
B = 5.44
3270
6540
Variable Speed
P = 1.99
Control
1202.2
B = 3.98
2392.4
4784.8
Source: Marley Technical Report H-001A, Cooling Tower Energy and Its Management,
October, 1982.
Variable Cold Water Set-Point Temperature
The energy efficiency of all the control discussed above can be improved by varying the cold
water set-point temperature with the outdoor air wet bulb temperature. This type of control
takes into account the fact that towers can only produce water at a few degrees above the wetbulb temperature (this temperature difference is called the approach); hence fan energy can

be reduced when that temperature is achieved, since continued fan operation results in
minimal further reductions in cold water temperature.

Fan Motor Power with Fan Speed and Air Volume Flow Rate
The figure below shows fan motor power draw as a function of input frequency for a cooling
tower fan equipped with a VFD. The fan affinity laws would predict a relationship between
fraction power (FP) and fraction speed (FS) of:
FP = FS3
Regression of the data show a slightly better fit using the exponent 2.8:
FP = FS2.8
Since fan speed is proportional to volume flow rate, this relation also hold for fraction volume
flow rate, FV.
FP = FV2.8
The slightly reduced exponent is caused by declining VFD, motor and fan efficiencies at reduced
speed.

ASD Performance
Input Power and Fan Speed vs Frequency
2000

35

1800

30

25

1400
1200

20

1000
15

800

Fan Speed (RPM)

Input Power (kW)

1600

600

10

I np

ut

Po

r
we

400
200

peed
Fan S

0
5

10

15

20

25

30

35

40

45

50

55

60

65

ASD Frequency (Hz)

Source data: An Application of Adjustable Speed Drives for Cooling Tower Capacity Control,
Welch, W. and Beckman, J.

Variable-speed cooling tower fans generate the least savings compared to constant-speed fans
during warm weather and when the cooling tower set point temperature is low because the fan
runs more frequently at these times. Alternately, variable-speed fans generate the greatest
savings during cool weather and when the cooling tower set point temperature is high because
the fan runs less frequently at these times. The CoolSim output screens shown below
demonstrate these concepts. Thus, variable frequency drives on cooling tower fans will
generate the greatest savings on year-round cooling applications with relatively high set-point
temperatures characteristic of industrial process applications.

Cooling Tower Bypass Plumbing

Bypass control is typically used only at low outdoor air wetbulb temperatures in order to
reduce fan cycling. Bypass should not be used in sub-freezing temperatures since this can lead
to tower freeze up. The preferred tower bypass plumbing is shown below. The preferred valve
is a single two-way butterfly valve placed in the bypass line.
Basic Tower Bypass Methods

10

Cooling Tower Pumping Pressure Drop

Typical cooling tower pressure drops are shown below. The Estimated Head Loss column is for
a standard condenser and 15 year old piping. The Actual Head Loss column is for a lowpressure loss condenser and new piping.
The same tower system will be evaluated; estimated head will be compared with actual head loss.

Condenser
Valves, Strainer, etc.
150' Piping (15 year Old)
Total Flow-Friction
Static or Open
Total Pump Head

Estimated Head Loss

25'
7'
6'
38'
12'
50'

Actual Head Loss

8'
7'
3'
18'
12'
30'

Cooling Tower Selection

In HVAC applications, the starting place for cooling towers selection is typically to match the
nominal cooling tower tons, as supplied by the tower manufacturer, to the cooling capacity of
the chiller or chiller plant. The water flow rate through the cooling tower is initially set at 3 gpm
per nominal cooling tower ton. Subsequent design optimization may occur from this starting
point. Engineering data for a typical model of induced-air crossflow cooling towers are shown
below. Based on these data, fan motor hp is about 0.1 hp/ton and air flow rates are about
2,000 cfm/hp.
A nominal cooling tower ton is defined as cooling 3 gpm of water from 95 F to 85 F at an air
wetbulb temperature of 78 F. Thus, the actual cooling associated with a nominal cooling
tower ton is:
Qact = 3 gpm x 8.33 lb/gal x 60 min/hr x 1 Btu/lb F x (95 85) F = 15,000 Btu/hr
This strange convention exists to make it easy for users to select cooling towers by matching
the nominal cooling capacity of the chiller with the chiller cooling capacity. The convention
works because most chillers have a COP of about 3, and total heat rejected by the condenser to
the cooling tower is about 15,000 Btu/hr for every 12,000 Btu/hr through the evaporator.

Model
Number
3240A
3272A
3299A
3473A
3501A
3985A
31056A

Norminal
Tonnage
240
272
299
473
501
985
1056

Motor
HP
10
15
20
25
30
60
75

Weights (lbs)
Fan
Heaviest
(CFM) Operating Shipping Section
62,790
14,770
6,790
6,790
71,340
14,900
6,920
6,920
78,110
14,960
6,980
6,980
118,870
23,090
10,190
10,190
125,900
23,140
10,240
10,240
229,950
40,240
15,560
9,460
246,700
40,330
15,650
9,550

Dimensions
L

8'-5 3/4"

18'-0 1/2"

9'-3 5/8"

8'-7 3/4"

11'-9 3/4" 21'-6 1/2" 10'-10 1/8" 9'-11 3/4"

11'-9 3/4" 21'-6 1/2" 21'-8 7/8" 20'- 9 1/2"

11

Source data: BAC Product and Application Handbook, Volume 1, 2005.

Model
8601
8603
8605
8607
8608
8610
8611
8613
8614
Nominal Tons
100
150
200
250
300
350
400
450
500
Length
5'-4 1/2" 7'-3 1/2" 7'-3 1/2" 7'-10 1/2" 7'-10 1/2" 9'-4 1/2" 9'-4 1/2" 9'-4 1/2" 9'-4 1/2"
Width
12'-0"
12'-0 1/4" 13'-7 5/8" 15'-3 1/4" 16'-7 3/4" 16'-7 3/4" 17'-7 3/4" 19'-3 3/4" 19'-3 3/4"
Height
8'-9 5/8" 9'-8 3/8" 9'-8 3/8" 10'-7 3/8" 10'-7 3/8" 10'-7 3/8" 11'-6 1/8" 11'-6 1/4" 13'-3 7/8"
Shipping Wt.
3430
4610
5130
6640
7800
8690
9860
10700
11990
Operating Wt. PVC
5500
7120
8260
10880
12200
13980
15520
17280
19260
Motor HP
5
7 1/2
10
10
15
20
20
25
25
RPM
666
547
547
547
427
427
427
427
427
CFM
29820
50000
54820
58610
69050
82630
91920
94760 105270
GPM (min.)
100
120
175
190
200
210
240
250
280
GPM (max.)
500
750
960
1250
1440
1715
1880
2160
2500

Source data: Marley Cooling Towers, 2000.

Cooling Tower Performance

The performance of typical cooling towers at water flow rates of 3 gpm/ton and 5 gpm/ton is
shown below. Similar performance data for specific cooling towers can usually be obtained
from the manufacturer. These curves predict the temperature of the cold water leaving the
cooling tower as a function of the water temperature range (Th-Tc) and entering air web bulb
temperature. Temperature range is generally known and can be used as an input value in these
charts, since the temperature range is set by the water flow rate and heat rejection rate of the
condenser.

12

Source: ASHRAE Handbook, HVAC Systems and Equipment, 2004.

Relations for the temperature of cooling water leaving the tower, Tc, can be derived from
regressing data from the 3 gpm/ton and 5 gpm/ton curves shown above. The relation and
regression coefficients are shown below. The R2 for these relations exceeds 0.995 and the
average error, [abs(Tc Tc,pred)], is less than 0.8 F.
Tc = a + b Twb + c Tr + d Twb2 + e Tr2 + f Tr Twb
Coef
a
b
c
d
e
f

3 gpm/ton
16.790751
0.6464308
2.2221763
0.0016061
-0.0159268
-0.015954

5 gpm/ton
24.6299229
0.45007792
3.32229591
0.00261818
-0.0324886
-0.0190476

These equations can be incorporated into software to predict cooling tower performance with
varying ambient conditions. For example, CoolSim (Kissock, 1997) calculates exit water
temperatures, and the fraction of time that a cooling tower can deliver water at a target
temperature, based on water temperature range Tr and TMY2 weather data. This information
is useful in determining how often a cooling tower can replace a chiller in cooling applications.
Cooling Tower Performance at Reduced Air Flow Rates
Comparison of the 3 gpm/ton and 5 gpm/ton performance maps can be used to predict cooling
tower performance at reduced air flow rates. For example, for a cooling tower operating with a
water flow rate of 3 gpm/ton, the 3 gpm/ton performance map shows tower performance at a
set water-to-air flow rate ratio. The 5 gpm/ton chart shows tower performance for a higher
water-to-air flow ratio, or, inversely, at a lower air-to-water flow rate ratio. Thus, the 5 gpm/ton
performance map indicates tower performance if water flow rate is held steady while the air
flow rate is reduced to 3/5 = 60% of maximum airflow.
Regressing the data from the 3 gpm/ton and 5 gpm/ton performance curves, with fraction of air
flow, FV, set to 1.0 for the 3 gpm/ton data and 0.6 for the 5 gpm/ton data gives the following
relation for the temperature of cooling water leaving the tower, Tc, at reduced air flow. The R 2
for this relation is R2 = 0.978 and the average error [abs(Tc Tc,pred)] is 1.9 F. Theoretically,
the fraction of air flow, FV, could vary between 0 and 1.0. However, this relation was
generated using data that represent peak air flow at 0.6 and 1.0. Thus, it is not recommended
that this relationship be used outside of this range.
Tc = a + b Twb + c Tr + d Twb2 + e Tr2 + f Tr Twb + g FV
Coef

Value

13

a
b
c
d
e
f
g

39.24367
0.548254
2.772236
0.002112
-0.02421
-0.0175
-23.1667

Evaporation Rate
As discussed in the previous section, cooling in cooling towers is dominated by evaporation.
The evaporation rate can be calculated from the psychrometric relations in the previous
section, if the inlet and exit conditions of the air are known. For example, consider the case in
which the cooling load, Ql, mass flow rate of air, ma, (which can be calculated based on the fan
cfm and specific volume of the inlet air), and inlet conditions of air are known. The enthalpy of
the exit air, ha2, can be calculated from an energy balance.
Ql = ma (ha2 ha1)
ha2 = ha1+ Ql / ma
The state of the exit air can be fixed by assuming that it is 100% saturated with an enthalpy ha2.
The evaporation rate, mwe, can be determined by a water mass balance on the air.
mwe = ma (wa2- wa1)
The fraction of water evaporated is:
mwe / mw
Using this method for entering air temperatures from 50 F to 90 F, we determined that the
fraction of water evaporated typically ranges from about 0.5% to 1%, with an average value of
about 0.75%.
Another way to estimate the fraction of water evaporated is to assume that all cooling, Ql, is
from evaporation, Qevap. The cooling load Ql, is the product of the water flow rate, mw,
specific heat, cp, and temperature difference, dT. The evaporative cooling rate is the product
of the water evaporated, mwe, and the latent heat of cooling, hfg.
Ql = Qevap
mw cp dT = mwe hfg
Assuming the latent heat of evaporation of water, hfg, is 1,000 Btu/lb, and the temperature
difference of water through the tower, dT, is 10 F, the fraction of water evaporated is:

14

mwe / mw = cp dT / hfg = 1 (Btu/lb-F) x 10 (F) / 1000 (Btu/lb) = 1%

If on average, 75% of the cooling were from evaporation and 25% from sensible cooling, then
the evaporation rate would be:
75% x 1% = 0.75%
Thus, both methods suggest that 0.75% is a good estimate of the rate of evaporation; however,
we have seen manufacturer data indicating average evaporation rates as low as 0.30%. Water
lost to evaporation should not be subjected to sewer charges. Typical sewer charges are about
$2.20 per hundred cubic feet.
Some water may be lost as water droplets are blown from the tower by oversized fans or wind.
This type of water loss is called drift. Drift rates are typically about 0.2% of flow (ASHRAE
Handbook, HVAC Systems and Equipment, 2000); however, we generally assume that drift
losses are included in the 0.75% evaporation rate.

Water Treatment and Blow Down Rate

Cooling tower water must be treated to prevent bacterial growth and maintain the
concentration of dissolved solids at acceptable levels to prevent scale and corrosion.
Bacterial Growth
The typical method of controlling bacterial growth is to add biocides at prescribed intervals and
to keep the cooling tower water circulating. If the tower will not be operated for a sustained
period of time, then the cooling water should be drained.
Dissolved Solids
Water evaporated from a cooling tower does not contain dissolved solids. Thus, the
concentration of dissolved solids will increase over time if only enough water is added to the
tower to compensate for evaporation. To maintain the dissolved solids at acceptable levels,
most towers periodically discharge some water and replace it with fresh water. This process is
called blow down. It the level of dissolve solids increases too high, scale will be begin to form,
and/or the water may become corrosive and damage piping, pumps, cooling tower surfaces
and heat exchangers. Usually, the primary dissolved solid to control is calcium carbonate
CaCO3.
Blow down can be accomplished by continuously adding and removing a small quantity of
water, periodically draining and refilling the cooling tower reservoir, or by metering the
conductivity of water and adding fresh water only when needed. By far the most efficient
method is to meter the conductivity of water, which increases in proportion to the level of
dissolved solids, and add fresh water only when needed.

15

The required quantity of blow down water depends on the acceptable quantity of dissolved
solids in the tower water, PPMtarget, the quantity of dissolved solids in the makeup water,
PPMmu, and the evaporation rate, mwe. The target level of dissolved solids is typically
quantified in cycles, where:
Cycles = PPMtarget / PPMmu
For example, if the quantity of dissolved CaCO3 in the makeup water, PPMmu, is 77 ppm and
the maximum level to prevent scaling, PPMtarget, is 231, then the cooling tower water must be
maintained at three cycles:
Cycles = PPMtarget / PPMmu = 231 ppm / 77 ppm = 3
By applying mass balances, it can be shown that the blow down water required to maintain a
certain number of cycles is
mwbd = mwe / (Cycles 1)
The total makeup water required mwmu, is the sum of the water added for evaporation and
blow down:
mwmu = mwe + mwbd
For example for a 1,000 gpm tower with a 0.75% evaporation rate and CaCO3 concentration at 3
Cycles, the quantity of makeup water required would be about:
mwe = (mwe/mw) x mw = 0.75% x 1,000 gpm = 7.5 gpm
mwbd = mwe / (Cycles 1) = 7.5 gpm / (3 1) = 3.75 gpm
mwmu = mwe + mwbd = 7.5 gpm + 3.75 gpm = 11.25 gpm
The overall makeup water rate would be about:
11.25 gpm / 1,000 gpm = 1.1%

16