Você está na página 1de 9

ESL-IE-81-04-97

DESIGNING A "NEAR OPTIMUM" COOLING-WATER SYSTEH


Ralph A. Crozier, Jr.
E. I. du Pont de Nemours & Co.
\?1lmington, Delaware
Reprinted by special permission from CHEMICAL ENGINEERING April Zl Copyright (c)
1980, by McGraw-Hill, Inc. New York, N.Y. 100ZO.

ABSTRACT
Cooling water is expensive to circulate.
'Reducing its flow --ie, hiking exchanger outlet
temperatures --can cut tower, pump and piping
investment as much as one-third and operating cost
almost in half.
Heat-exchanger-network optimization has been
accomplished in large integrated plants, such as

petroleum refineries. In many of the chemical

process industries, however, a plant contains sever

al individual processes, and network optimization,


except on a limited basis, is not feasible.
So far, no one has developed similar procedures
for designing and optimizing a cooling-water once
through-exchanger system.* This article attempts m
fill the void by presenting a design basis that will
produce a "near optimum" system.
A cooling-water system consists of four major
components: heat exchangers, cooling towers, cir
culation piping and pumps. To optimize such a sys

tem, one must define the system interactions and

apply these relationships to the simultaneous design


of the aforementioned equipment. This article de
velops criteria that for most applications allow
one to ignore system interactions, and still design
a "near optimum" system.
Cooling-water systems have long been designed
by "rules of thumb" that call for fixing the cool

ant temperature-rise across all heat exchangers

(usually ZOF) and setting the coolant inlet temper


ature to the heat exchanger at the site's wet-bulb
temperature plus 8F. These rules produce a work
able cooling system; but, by taking the same coolant
rise across all exchangers, regardless of the indi
vidual process outlet-temperatures, this cannot re
sult in an optimized design.
The design method presented in this article re

places the "rules of thumb" with criteria that are

easy to apply and that take into account the effect


that the individual exchanger process outlet-temper
atures have on cooling-system economics.
Economic analyses of actual process have shown
that cooling-system investment can be reduced by one
~hird, and cooling-system operating cost by one hali,
1f the proposed design criteria are used instead of

the "rules of thumb." It has been found that the

controlling economic factor for a cooling system is

the quantity of water being circulated. Reducing


the flow (raising the coolant outlet temperature cif
heat exchangers) significantly reduces cooling
tower, pump and piping investment, and operating
cost, and only moderately increases the heat-ex
changer investment. The overriding conclusion to.
be drawn is that cooling water is very expensive,
and its conservation can result in significant

savings.

Three System Interactions


What are the system interactions and their
relationships? When can they be ignored, and whe~
must they be considered?
From Fig. 1, it can be seen that there are
three coolant interactions: (1) between the outl~t
temperature of the heat exchanger and the inlettem
perature of the cooling tower, (Z) between the inT
let temperature of the exchanger and the outlet t~m
perature of the cooling tower, and (3) between tO$Il
coolant flowrate and the size of connecting piping
and circulating pumps.
.
First Interaction - In the Fig. 1 system, cool
ing water enters the shell side of the exchanger ~t
a fixed temperature of Tl and exits at a variablel
temperature of TZ; the process fluid passes throu~h
the tube side at fixed inlet and outlettemperatur~
of tl and t2' respectively. The cooling tower dis
Sipates heat to the ambient air through humidific~
t ion (evaporat ion) and sens::'ble heat ing.
'
Cooling-tower investment and operating cost are
reduced by returning the water to the tower as hot
as possible, because the hotter th~ water, the
,
larger the temperature difference (thermal driving
force) between it and the ambient air; hence. less
heat-transfer area (tower packing) is requir~d. :
Additionally, as the effluent air temperature is i
raised, so is the moisture content of the exiting:
air; hence, less air flow is necessary.
Upping the temperature of thp. water to the
tower also cuts do,~ the quantity of water circu-:
lated through the tower. From Eq. (1), it can be:
seen that for a fixed heat load and tower outlet
temperature (Tout), the quantity of cooling water:
reqUired to satisfy the heat balance is inversely'
proportional to the tower inlet temperature (Tin):
\'T=QT/[C(Tin-Tout)]
(1)

563
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

Boosting the tower inlet temperature means increas


ing the coolant temperature rise across the individ
ual exchangers; hence, to reduce cooling-tower in
vestment and ope~ating cost, one must widen the coo~
ant temperature rise across th:e individual exchangers.
T2

A significant discovery has been that in most


instances exchanger "lrea is relatively insensitive
to incremental changes in coolant outlet temperatuIE.
To understand this, one must recognize that a well
designed exchanger must have adequate heat-transfer
area, but it must also have sufficient fluid-flow
area to satisfy the shell-and-tube pressure-drop re
strictions.

T,

Exchanger
Pip;ng
Tower

design ----design--- design

Because coolant flowrate is inversely propor


tional to coolant temperature rise, a small coolant
temperature rise produces a large increase in cool
ant flow. This means that the exchanger diameter
must be increased, if the pressure-drop (maximum
velocity) restriction is to be satisfied.

Typical systam demonstrates thr.. interactions

Fig. 1

~175r--"'--"""'---,-----,---,
~' 150

If the heat-transfer equation Q=UA(LMTD) were


solved for the small coolant temperature-rise geome
try, it would show that the area needed for heat
transfer is less than required by the pressure-drop
limitation. In other words, for small coolant tem
perature rises, the exchanger's heat-transfer area
would be excessive if determined by fluid-flow con
siderat ions.

Process outlet temoer.turl 158"F

~ 125

I limitillg
OUA(LMTD)

S 75
j

v Tutlnidl

I hllHranshr

~100

50

I
I

I
I
:,.

ll5
o

AA"

~
1,500 ;

1.250
.c
1.000]

g.

AAA

O=-_---::::: __~:.....;:......,...,--.....l.::_-.....J750 a:
10

If the coolant outlet temperature is sequen~


tially increased and the required exchanger area is
calculated for the different coolant temperature
rises, a plot similar to that in Fig. 2 would be gen
erated. Notice that the required exchanger area
initially lessens, reaches a maximum, and then gains
as the coolant temperature rise is increased.

100

120

140

Coollnt outlet temperature,

'SO

lao

Exchanger overall coefficient and araa


fall into three distinct design regions

Fig. 2

Specifying the coolant outlet temperature is


difficult because it is bound on the lower side by
the wet-bulb temperature and on the upper side by
the minimum process outlet temperature. To minimize
cooling-tower investment and operating cost. one
needs to maximize the approach to wet-bulb tempera
ture. However, to minimize heat-exchanger invest
ment, one needs to maximize the approach to the min~
mum process outlet temperature. Therefore, the tow
er outlet temperature has to be a compromise between
these two boundaries.

The required exchanger area, as depicted in


Fig. 2, passes through three distinct design regions.
In the first (80F to 115F), the required area is
controlled by fluid-flow phenomena. In the second
(115F to 158F), the fluid-flow and heat transfer
considerations are the same order of magnitude; that
is, neither requirement is controlling. In the third
(above 158F), the heat-transfer requirement becomes
controlling (reduced LMTD). Therefore, to minimize
exchanger area, the design should be in the second
(transition) region.
Fig. 2 was developed on the basis of full tube
bundle. At low coolant temperatures, the better de
sign would be to have the tube bundle partially fill
ed, so as to maintain a constant tubeside velocity.
In the transition region, however, the shell velocity
was kept at a constant rate by reducing the baffle
spacing as the coolant flow decreased.

The importance of the cooling-tower and heat-ex


changer designers' discussing the temperature limi
tations imposed on each before any calculations are
started cannot be overemphasized. Usually, the 101.1
est cooling temperature is required for a refrigera
tion system or a steam turbine condenser. Consider
able savings can be realized if alternative coolin?
methods are provided for these services, rAth~r than
penalizing the entire cooling system for one, or
both, of them.

Second Interaction -- The tower outlet temper


ature (approach plus wet bulb) is the variable with
which the cooling-tower designer has the least lati
tude, yet it is the most significant determinant of
cooling-tower investment and operating cost. The
tower outlet temperature is the "Achilles heel" of
this study and the reason why only a "near optimum,"
rather than a truly optimized, cooling system is
attained.

Many cooling towers are designed by a rule of


thumb that sets the tower outlet temperature equal to
the site wet-bulb temperature plus 8F. This can re
sult in significant cooling-tower investment and op
erating cost penalties if process requirements do not
demand cooling water at a temperature that low. As
previously stated, cooling-tower investment and oper
ating cost are minimized by maximizing the approach
to wet-bulb temperature.

564
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

Third Interaction--Via Eq. (1), it can be shown


that the system flowrate decreases as the tower in
let temperature is hiked. Therefore, as the coolant
rise across the individual exchangers is increased,
the supply and return piping, as well as the circu
lating pumps, can be reduced in size.

presented here.
Basis of the Economic Evaluations
The investment and operating cost for the cool
ing- system components were calculated as follows:,
heat exchangers, by a method similar to that describ
ed by Woods, Anderson and Norman [lJ; cooling-tower
pumps, by a proprietary correlation developed for,
electrically driven, vertical turbine water pumps;
operating at 880, 1,200 and 1,800 rpm; piping, by a
proprietary correlation based on material and man
hours per linear foot (1- to l2-in. dia., "Cast Iron
Pressure Pipe--Cement Lines," and 14- to 96-in. dia.,
"Concrete Pressure Pipe, American Water Works Assn.,
Spec. C-30l Prestressed"); and cooling t.owers, by
t.wo booklets of The Marley Co. [2,3J.

In summary, the investment and operating cost of


the tower, pumps and piping will be reduced when the
cooling-tower inlet temperature is increased. The
exchanger investment mayor may not be reduced as
the coolant temperature rise is increased, depending
on the relationship of the coolant outlet temperature
and the exchanger design region (Fig. 2).
If the exchanger's optimum coolant outlet tem
perature is in the transition region, its area will
be minimized, and system interactions can be ignore~
If the exchanger's coolant outlet temperature is not
in the transition region, however, system interac
tions must be considered.

Economic evaluations are based on cost plus re


t.urn on invest.ment. Using cost plus return, one can
combine t.he investment and operating cost fact.ors
into a single variable.

It has been empirically determined that if the


exchanger LMTD is greater than 30F when the coolant
outlet temperature is set at the process outlet tem
perature or some maximum temperature (scaling limit~
tion) , the exchanger is in the transition region. If
the LMTD is less than 30F, the exchanger is in the
heat-transfer-limiting region and the exchanger area
will be very sensitive to coolant temperature
changes. Therefore, if the LMTD is less than 30F,
system interactions must be considered; and a com
promise between higher exchanger investment and lower
cooling-tower investment must be determined.

Cost plus return is the total cost of owning


and operating a cooling syst.em for one year. Tot~l
investment (allocated, permanent, working capital;
etc.) is included as a cost--that is, as an interest
penalty. Operating cost (depreciation, maintenane,
electrical, etc.) is added to the interest penalty.
This total is defined as cost plus return:
CT=C+ {xI/[(l-a)(l-i)J}

(2)

In Eq. (2), C= annual operating cost; x = re-,


quired net return on investment; I = total invest
ment; a = state and local taxes; and i = federal
income tax.

Constraints on Outlet Temperature


Two temperature constraints limit the exchanger
coolant outlet temperature.
First, the wood fill
of the conventionally designed cooling tower limits
the maximum water temperature to 150F. Second, the
temperature from individual exchangers is set by pre
cess outlet temperatures or by tube fouling or mate
rial of construction.

Hereafter, cost means cost plus 10% net return


on investment. The lowest-cost alternative proposal
is the economic choice. All investment and operat
ing costs represent the January 1976 Engineering ,
News Record Construction Cost Index of 215
(1967 = 100).
COMPONENT DESIGN CRITERIA

Waterside deposits of suspended solids, mineral


salts, biological growths, and corrosion products
interfere with heat transfer. In the design calcu
lations for sizing heat exchangers, the magnitude of
the interference is the fouling factor. Attempts to
quantitatively relate the factor to specific fouling
conditions have been only partially successful;
therefore, the numerical value assigned to different
degrees of fouling by specific deposits has been
based largely on experience.

Heat-exchanger criteria are divided into two,


categories: LMTD greater and less than 30F. Codl
ing tower criteria are similarly divided: inlet
temperature greater and less than 110F.
Exchanger LMTD greater than 30F
Fig. 3a shows system cost as a function of tre
coolant temperature rise for an exchanger whose p~o
cess outlet temperature is 195F. Two concepts ih
Fig. 3a are worth discussing:

Because there is no theoretical basis for select


ing fouling factors at different coolant temperatures,
the same factor was used at all temperatures in this
study. To analyze the effect of this assumption on
the conclusions, a cooling system for which the
fouling factor was arbitrarily doubled was evaluated.
It was found that the cooling tower, pump and piping
investment and operating cost savings far outweighed
the higher exchanger investment and operating cost.
Therefore, the conclusions reached in the increased
fouling-factor evaluation were the same as those ob
tained by assuming a constant fouling fac tor. For
simplification, only the latter evaluation is

1. Because this exchanger's LMTD is large, its


area is relatively insensitive to changes in the
coolant temperature rise. This phenomenon can be
explained by a simple example.
It was shown previously that the area and ov~r
all heat-transfer coefficient of an exchanger in the
transition region are fairly constant; hence, the
required area is only a function of the L~ITD (Fig 2).

565
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

Assume two extremes, a large and a small LMTD.


If the LMTD is large, narrowing it from 150F to
140F augments the exchanger area by 7%. On the
other hand, if the L~ITD is small, lessening it from
50F to 40F boosts the exchanger area by 20%.
Hence, as the LMTD is reduced, the exchanger area
becomes more sensitive to the coolant outlet temper
ature.

From this analysis, one can see that by consid


ering the system cost, a temperature cross is eco
nomically justified. But before deliberately design
ing a heat exchanger for a cross, one must realize
that an error in the predicted outlet temperature of
either stream will result in a thermodynamically
unstable exchanger.
In summary, the design criteria for heat ex
changers that have an ~ITD greater than 30F (cal
culated at the coolant outlet temperature specified
by the criteria) are:

2. Recall that as the coolant outlet tempera


ture is raised, the coolant flow is reduced. Assume
coolant investment to be $50/gpm (Fig. 3a). Hiking
the coolant temperature rise from 20F to 60F re
duces the coolant flow by 660 gpm, with a resultant
lowering of the coolant investment by $33,000. The~
fore, a significant investment saving can be real
ized for exchangers in this category if the coolant
outlet temperature i~ raised to 150F.

Single-pass geometries-- The coolant outlet


temperature should be set at the process inlet tem
perature minus 10F, or at 150F if the process llliet
temperature is greater than 160F.
Multiple-pass geometries-- The coolant outlet
temperature should be set at the process outlet tem
perature, or at l50F if the process outlet tempera
ture is greater than 150F.

As can be seen in Fig. 3a, the system cost could


be further reduced by increasing the coolant outlet
temperature above 150F. However, there are practi
cal reasons for not going this high, although this
limit is arbitrary and should be evaluated for a
particular site.

Exchanger LMTD Less Than 30F


Fig. 3c shows cost as a function of the coolant
temperature rise for a process outlet temperature of
115F and an LMTD of less than 30F (calculated at a
coolant outlet temperature of 1l5F). Once again
the coolant cost decreases as the coolant temperature
rise increases; however, in this example, the higher
cost of the heat exchanger as the coolant outlet tem
perature is raised has a significant effect on sy~em
economics.

The evaluation should consider: for the heat


exchanger, such effects ~s (1) the use of exotic
tube materials of construction, and (2) increased
fouling stemming from the lower coolant velocity; for
the cooling tower, such effects as (1) increased
blowdown to remove mineral salts, and (2) costlier
water treatment because of the higher temperatures.
Fig. 3b shows system cost as a function of cool
ant temperature rise for an exchanger whose process
outlet temperature is 115F. Again, the coolant cost
drops as the coolant temperature rise increases. In
this case, however, the effect of the L~ITD on ex
changer area is becoming evident. At a coolant out
let temperature of 125F, multiple exchangers in
series are required because of a temperature cross
(coolant outlet temperature exceeds the process out
let temperature). Although the system cost for two
exchangers in series is less, the savings are margin
al and one exchanger is preferred, with the coolant
outlet and process outlet temperatures set equal to
one another.

This example differs from the previous two in


yet another respect. For the first time, the opti
mum coolant outlet temperature depends on coolant
economics. In the previous examples, coolant costs
so predominated that an incremental increase or de
crease in the coolant cost did not change the design
criteria. In this case, however, if the coolant cost
increases incrementally, one wants to reduce the
quantity of coolant; therefore, the optimum coolant
outlet temperature increases (or if the coolant cost
decreases, the optimum coolant outlet temperature be
comes lower).
In Fig. 3c, the total cost curve has a minimum
that is a function of both the heat exchanger and
coolant economics. Therefore, for an exchanger with
an ~ITD less than 30F, the exchanger and coolant
interaction precludes setting the coolant outlet te~
perature a priori by using the design criteria. For
an exchanger in this category, the optimum coolant
outlet temperature must be found by constructing a
graph similar to Fig. 3c.

An economic evaluation of Fig.

3b at three dif
ferent coolant outlet temperatures provides the first
insight into the effect of coolant flowrate on system
economics. It also establishes the basis for the
practice of setting the process and coolant outlet
temperatures equal to each other.
In Table I, the three cases represent three dif
ferent coolant outlet temperatures: (1) 15F coolant
temperature rise, (2) equal to process outlet temper
ature, and (3) 5F higher than the process outlet
temperature.

Such a graph is generated by (1) using plant


coolant cost data, or estimating the coolant econom
ics via a Fig. 4a or 4b type graph; (2) determining
the area requirements of the exchanger at different
coolant temperature rises; (3) calculating the ex
changer and coolant cost at the different temperature
levels; (4) plotting the exchanger and coolant cost;
and (5), at a given coolant temperature rise, sum
ming the individual costs to obtain the system cost.
The optimum coolant outlet temperature is that at

Using Case 3 as the basis, the incremental in


vestment penalties for Cases 1 and 2 are $33,700 and
$6,100, respectively. The incremental cost penalties
for Cases 1 and 2 are $15,OOO/yr and $3,000/yr,
respectively.

566
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

70

which the total system cost is minimized. In Fig.


3c, this coolant outlet temperature happens to coin
cide with the design criteria temperature of 115F.

....t

The Step 1 graphical estimate of coolant eco


nomics is made as follows: Using Fig. 4a and assum
ing a tower having a 20F range and a 91F outlet
temperature, one arrives at a total cost of $255,000/
yr, or $3l.25/h (8,000 h/yr). For the tower, duty of
100 million Btu/h, one then calculates, using Eq.
(2), a tower flowrate of 600,000 gal/h. Dividing
$3l.25/h by 600,000 gal/h, one obtains a coolant
operating cost of 5.2/1,000 gal, which can be used
to calculate the coolant economics at this tempera
ture level.

dt.rtv: 10 rnill'on ItWh

Matllrial of conatruction: 304 ftIIifll....


Proens.tempernur. inlet: 340.;0 F
PToc:nI-hmperature OCJn.t: 195"F

60

tt~

CooIM1tumper.tunl inlet: go-F

SO

:ac:

.......Total COlt

,:I~

T~

30
'if
u

CfmCM,ature
limit

Coolant Cott .. '

20

10

20

80

60

Coolant rise across exchanger, -F


HighLMTD

Coolan t Piping and Pumps


70

The connecting piping (cooling tower to heat e~


changer) in the Fig. 5 cooling system is composed of
large underground distribution headers and smaller
takeoffs from central distribution points to indivi~
ual heat exchangers. It was originally thought that
these two piping configurations (one of straight
runs, the other of many fittings and valves) would
have different optimum design velocities. It was
found, however, that the optimum velocities coincida

He,t duty: 10 millien 8tu/h


Mltllriaj of construction: 304 .t.inlllSs ItHI
0
PrQCHI-ttmper.ture inl't; 260 F

60

PToeus-ttmpcratur. outl.t: 115" F


Cool.lnt-tllmperltur. inlet:

SO

:>

...~

40
30

u 20

From Fig. 6, it can be seen that the cost is in


sensitive to coolant velocity changes, and that the
optimum velocity falls between 5 and 10 ft/s.
(The
cost-pIus-return calculation of the Fig. 6 pipe and
pump cost was based on the Fig. 5 cooling-system pip
ing configuration.)

10
Two

uenang.", in s,ries'

-'

,
I

20

40
60
CooI.nt rise across exchanger. -F

60

b. Im,rmed;." LMTD
100

The insensitivity of system cost to the coolant


velocity is the result of counterbalancing cost
factors. Two extremes are possible in piping system
design: (1) large pipe diameters (small friction
losses) and small pumps with low horsepower require
ments, or (2) smaller pipe diameters (large friction
losses) and large pumps with high horsepower require
ments.

He,t duty: 10 million Btuth


Mattr~1 of conttruc:tion: 304 ft.linltu nftll
p,.ocess-tlimper.tur. inl.t: 140~ F
Procen.tt1moef.tl.Hf outlet: 115'" F
Cooianttemperatu,.. inlet: 9tf F

80
~

';3 60
c:

,~

...
-5

In the first, the piping investment is high, and


the pump investment and operating cost low. In the
second, the reverse is true. ~hen these two extremes
are evaluated on a cost-pIus-return basis, the total
system costs are the same. This analysis will be
true for all liquid piping systems that do not re
quire exotic materials of construction.

40

B
u
20

20
~
Coolant rise across exchanger, F
Co

The skill in designing coolant piping systems is


in balancing flows through the various loops. To
adequately analyze the flow distribution problem, one
is forced to computer simulations of the piping sys
tems (the cooling-water supply and return headers,
and the piping to and from heat exchangers), choosing
by trial and error the pipe diameter that matches the
loop and system pressure drop. Such a piping config
uration forms a series of parallel flow loops, in
which, by definition, the pressure drops must be the
same.

60

Low LMTD

Imp.ct of coolant temperature rise on


IotaI system cost at three different LMTDs

Fig. 3

supply the pressure drop needed to achieve bala ced


pressure distirbution. This means excessive wear
and control-valve maintenance. Therefore, aproperly
designed cooling piping system will specify the cor
rect pipe diameter for each loop that will balaqce
the pressures without the need to resort to acortrol
valve.
Cooling-Tower Design

The system pressure-drop requirement can be sat~


isfied by increasing or decreasing the flowrate in
the individual loops. If the piping in a particular
loop is oversized, the flow through it will be
excessive. To reduce the flow, a control valve must

As previously mentioned, a cooling tower dis


sipates process heat to the atmosphere by humidif~
the ambienr air and, to a lesser degree, by s,ensibly

567
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

heating the incoming air. Therefore, cooling-tower


design depends very much on the ambient wet-bulb tem
perature. The tower outlet temperature (exchanger
inlet temperature) is calculated from the site wet
bulb temperature and an assumed approach to it
(typically 8F):
T2

assuming a reasonable approach to the wet-bulb tem


perature based on both the minimum refrigeration and
process outlet temperatures, (2) following the heat
exchanger guidelines to establish the coolant outlet
temperatures, (3) calculating the tower inlet tem
perature based on these outlet temperatures, (4) us
ing Fig. 4a or 4b to ascertain whether the assumed
approach is the optimum, and (5) considering putting
the low-temperature-coolant users on a separate cool
ing-tower cell, if the approach assumed is not
optimum.

= wet-bulb

temperature +
approach to wet-bulb temperature

(3)

Eq. (3) is the first step in designing any coo~


ing system. Often, the approach to wet-bulb temper
ature is set arbitrarily, and this can result in con
siderable investment and operating cost penalties.

The advantages of this procedure are that it is


quick, does not require heat-exchanger or cooling
tower sizing, and makes both the exchanger and tower
designers aware of the system limitations.

To analyze the magnitude of these cost differen


tials, consider the following example:
W
gpm

Q/[500 C(T

out

- T. )J
1n

Fig. 4a and 4b are for cooling-tower heat illties


of 100 million and 300 million Btu/h for a site wet
bulb temperature of 80F. The tower range (tower
inlet minus outlet temperature) is plotted against
total tower cost. Note that for a fixed tower outlet
temperature total cost decreases as the range widem.
Increasing the range requires higher coolant temper
ature rises across the individual heat exchangers.
Also notice that for a tower inlet temperature above
110F, the different approach to wet-bulb (or tower
outlet) temperature for a particular tower inlet
temperature has little effect on the total cost.

(4)

Here, Q = 300 X 10 6 Btu/h; C = 1 Btu/(lb)(OF);

Tin = 105F and 125F; and T


= 85F (80F wet bulb

+ 5F approach) and 95F (80Q~twet bulb + 15F


approach).
Using Fig. 4b and 6, and subtracting, one can
find the tower cost (Table II). There are several
things in this example that should be particularly
noted.
Although the tower heat duty in all four cases
is the same and each represents a valid design, the
cooling-system costs vary significantly. One can see
that, for a fixed tower outlet temperature, boosting
the tower inlet temperature materially reduces cool
ing-tower cost. At a fixed tower inlet temperature
mayor may not lower cooling system cost.

In summary, if, after the application of the


heat-exchanger guidelines, the tower inlet tempera
ture is less than 110F, the cooling-tower and heat
exchanger interactions are important and the apprmch
to wet-bulb temperature must be found by trial and
error. On the other hand, if the tower inlet tem
perature is greater than 110F, system interactions
can be ignored, and a first approximation to the
tower outlet temperature should be the site wet-bulb
temperature plus 10F.

At a 125F tower inlet temperature, varying the


tower outlet temperature has a minor effect on the
quantity of water circulated. At 105F, however, the
effect is quite significant.

APPLYING THE GUIDELINES

Therefore, even though cooling-tower cost can be


reduced by raising the tower outlet temperature, for
low tower inlet temperatures, the increase in piping
cost can exceed the reduction in cooling-tower cost.
Above a tower inlet temperature of 110F, the inter
action between the tower inlet and outlet tempera
tures and the quantity of water circulated has mini
mal effect on system cost.

It has been shown that cooling-system optimiza


tion requires reducing the coolant flowrate through
the system. How the application of the proposed de
sign guidelines to a cooling system can result in
considerable investment and operating cost savings
will now be illustrated (Table III).
To simplify the analysis, a cooling system com
posed of five heat exchangers will be considered.
Each exchanger has a heat duty of 18 million Btu/h,
water as the shell-side coolant, and ethylene glycol
as the orocess fluid. This cooling system is evalu
ated under two different circumstances: (1) at a
fixed coolant temoerature rise of 20F, or a 10F
approach to the p~ocess outlet temperature if a 20F
coolant rise results in a temperature cross, and
(2) by means of the design guidelines that have been
presented.

Tower outlet temperatures are usually chosen


arbitrarily because of lack of communication between
the cooling-tower and heat-exchanger designers. Be
fore heat exchangers can be sized, the coolant inlet
temperature must be known. The cooling-tower design
er will usually calculate the tower outlet tempera
ture based on either refrigeration or turbine con
denser requirements because these exchangers take the
lowest-temoerature coolant. The tower designer then
gives this' coolant outlet temperature (exchanger in
let temperature) to the exchanger designer, who or
dinarily does not question the low coolant tempera
ture (although it may not be needed) because it ma'es
the job easier.

First Application Example


The first exchanger in this application example
has a process outlet temperature greater than 150F.
Therefore, the design guidelines call for a coolant
outlet temperature of 150F. A check of the LMTD at
the specified temperature levels reveals that the

A better procedure would be for the tower and


exchanger designers to simultaneously develop the
optimum approach to wet-bulb temperature by: (1)

568
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

LMTD is greater than 30F; therefore, system inter


actions can be ignored and the exchanger coolant out
let temperature set at 150F.
The next three exchangers have process outlet
temperatures less than 150F; therefore, the design
guidelines require that the coolant outlet tempera
tures of these exchangers be set at their process
outlet temperatures. A check of the LMTDs at the
specified temperatures shows that they are greater
than 30F for these three exchangers. Again, system
interactions can be ignored, and the coolant outlet
temperatures of the second, third and fourth ex
changers can be set at 113F, 131F and 113F,
respectively.

&ie
~

2~

~ i
w.=

...

~ o~8ggl8

- ilS.""'l"'.'1
_
..... N ,...,

(3

The process outlet temperature of the fifth ex


changer is 104F. If this exchanger's coolant ouUet
temperature is set at 104F, a check at this temper
ature shows that the LMTD is less than 30F. This
means that, for the fifth exchanger, system inter
actions are important and must be considered. The
cooling-tower inlet temperature, based on the first
four exchangers, is calculated to be 121F; there
fore, as a first approximation, the tower outlet tem
perature (exchanger inlet temperature) should be set
at the site wet-bulb temperature plus 10F.

II
c:

.~

"ii

..,
~
,

11

il,

;,

.~

-0

-g l:l8
...

,'9

'0
c:

a:

o
.;:

"'
()

iQ.

"'
Finding the optimum outlet temperature of the
coolant from the fifth exchanger requires generating
a graph similar to Fig. 3c. The procedure is: (l)ffi
sume a coolant outlet temperature for the fifth ex
changer; (2) calculate the tower inlet temperature
for the five exchangers; (3) determine the coolant
cost from Fig. 4a, using the calculated inlet temper
ature of site wet-bulb temperature plus 10F; (4)
divide the total coolant cost by the tower heat duty
and multiply this result by the heat duty of the
fifth exchanger; (5) calculate the required heat-ex
changer area, using the assumed cooling-tower temper
atures; (6) plot the exchanger and coolant costs;
(7) repeat Steps 1 through 6 at two different coolant
outlet temperatures.

...

I
.S

1:

Water treatment economics for higher


coolant outlet temperature

The coolant outlet temperature at which the sy~


tern cost is lowest should be chosen. In this exam
ple, the optimum coolant outlet temperature is l22~.
For a 122F coolant outlet temperature, an 18F cross
occurs. This requires multiple shells in series;
therefore, the coolant outlet temoerature was set at
104F. For an exchanger in this ~ategory, one shoilld
consider going to a single-pass tube geometry or to
a spiral exchanger, so that true countercurrent flow
will be achieved and the temperature cross eliminat
ed. By eliminating the temperature cross, the cool
ant outlet temperature of 122F can be specified.
Wider approach temperatu re cuts tower cost

ea.. l

Case 2

Tower inl.t ~mp.ratu~. OF

120
Circulation nJ'tB, gpm
To~r

outlet temperature, OF
TOlNer inlet temperature, OF
Makeup rate, gpm
Slowdown, gpm
Concentration cycles
InYe'S1JT'l8n t, S/gpm

6,000
90
140
450
150

50

3
64

16

30

Suspended-solids
removal, S/gpm

InYeStment cost, S

TOlM!r investment
Wastewater treatment
of blowdown
Suspended-solids removal

Table II
Case 3

140

10,000
90
120
400
100

384,000

.116,000

120,000
160,000

135,000
180,000

15,000
20,000
-81,000

Net incremental investment, $

Case 4

Inerem.nt.1
investment, $

500,000

Incremental

Tower inlet temperature, of


Tower outlet temperature, OF
Coolant flowrate, gpm
Cost of tOlIVer, pumps and
piping (from Fig. 4B), S/yr
COSt of pumps and piping.

Ifrom Fig. 61, Slyr

105
85
30,000

105
95
60,000

125
85
15,000

Opef'ating cost, $/yr

125
95
20,000

Tower costt
Water for makeup
at 1O /1,000 gal
Chemicals

700,000 810,000 500,000 430,000

150,000

90,000

20,000
48,000

22,500
54,000

pumps and piping.

tJlk.n .. $0.03/1.000 g.el ot

Coolln~to....,..r InclucMil

.,,

",

COst of tOlNer. S/yr

360,000 250,000 280,000 160,000

tO~,...tin;

CO$t i.

c.~itY.

".

569

2,500
6,000
51,500

Net incremental oPe'fating cost, S/vr

340,000 560,000 220,000 270.000

cost, S/yr
-<30,000

Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

This example shows why the cooling-tower outlet


temperature is the "Achilles heel" of this study.
To be rigorous, the fifth heat exchanger should be
resized, using the optimum coolant outlet tempera
ture calculated but with the coolant inlet tempera
ture reduced by 4F. If exchanger-area requirements
are greatly reduced by the lower coolant inlet tem
perature, the tower and exchanger interaction is
significant, and the foregoing Steps 1 through 6
should be repeated at the lower coolant inlet tem
perature.
Recalculating this exchanger for three differ
ent coolant outlet temperatures is, however, time
consuming and may not be necessary. Because the
four exchangers with the LMTD greater than 30F will
be unaffected by a slight reduction in coolant inlet
temperature, what is really being sought is how sig
nificant is the cost of the fifth exchanger to the
total system cost.
Through engineering judgment and application of
the proposed guidelines, one can design a partially,
nearly or wholly optimized cooling system, depending
on how much effort one is willing to put out.
1. If one applies the guidelines only to ex
changers that have an LMTD greater than 30F, a par
tial optimization will be attained.

Water-Treatment Costs
It is usually at about this point that skeptics
ask, "\';hat about the higher water-treatment cost?"
The following answers this question:
Water-treatment economics--Water-treatment re
quirements do rise if the cooling-tower inlet tempe~
ature is increased. The actual cost of water treat
ment at any operating temperature is related to the
quality of the makeup water and its particular treat
ment requirements. Obsiously, waters having high
concentrations of sludge or dissolved mineral salts
will require more treatment. Usually, the magnftude
of the increased treatment cost will be minimal for
the same source of makeup water.
Mineral salts--The commonly present mineralsa14
calcium carbonate, has a limited and inverse solubil
ity (less soluble at higher temperatures), and thus
requires more treatment to prevent scale deposits as
the water temperature is raised. Increased addition
of sulfuric acid should resolve most problems. How
ever, if the volume of blowdown to control the con
centration of mineral salts resulting from increased
evaporation in the cooling tower must be augmented,
the added cost of waste-water treatment for the ~ea~
er volume of blowdown could represent an appreciable
water-treatment cost increase.

2. If one is willing to generate curves simiVn


to Fig. 3c for exchangers with an LMTD less than
30F, a nearly optimum design will be gained.

Suspended solids--The tendency of suspended sol


ids (turbidity) in the cooling water to deposit on
tube surfaces increases as the coolant temperature is
raised. Such deposition can be prevented by the use
of a sludge dispersant.

3. If one is willing to recalculate the ex


changers with an LMTD of less than 30F at differ
ent coolant inlet temperatures, then a wholly opti
mized system will be designed.

Biological growth--Organic growth speeds up as


the coolant-water temperature increases. The resul~
ing slime deposits that develop on tube surfaces can
reduce heat transfer and tend to collect more of the
suspended solids than might otherwise deposit from a
cooling-temperature increase alone. This type of
fouling can be controlled by addition of free chlo
rine or a nonoxidizing biocide.

Inspection of Fig. 4a shows, however, that the


last procedure is not necessary if the calculated
tower inlet temperature is greater than 110F, or if
the exchangers that require the low-temperature cool
ant do not contribute significantly to the system
heat duty. If the calculated tower inlet temperarure
is less than 110F, the last procedure is necessary.

Accounting for Water-Treatment Costs

Returning to the example (Table III), if one


used the fixed coolant temperature-rise case as the
economic basis, it can be seen that the guideline de
si~a would reduce the total coolant flowrate by 49%
(14,400 gpm vs. 7,300 gpm), the operating cost by43%
(S90,000/yr), and the total investment by 27%
($364,000). This example again demonstrates that a
higher exchanger investment can reduce cooling-sys~m
investment and operating cost.

Table IV compares system cost between a cooling


tower receiving water at temperatures of 120F and
140F. Using the 120F temperature case as the eco
nomic basis, it can be shown that the tower invest
ment is reduced by $116,000 by going to 140F, but
the water-treatment investment is boosted $35,000
($15,000 for water-treatment facilities and $20,000
for augmented blowdown facilities).
Therefore, at the 140F cooling-tower return
water remoerature, the net investment is reduced by
$81,000 . . Additionally, an operating cost savings of
$60,OOO/yr can also be realized. However, this
$60.000/yr savings is reduced by the increase in
water-treatment chemicals and makeup-water cost of
S8,SOO/yr. Therefore, the net-operating-cost re
duction is $Sl,SOO/yr.

It is important to maintain a water velocity


that mlnlmlzes the deposition of suspended solids.
In the initial design of an exchanger, it is desir
able to select the baffle spacing or number of tube
passes to get a water velocity above 5 ft/s. This
velocity should never be less than 2 ft/s. One
should recognize that raising the coolant temper&ure
for an existing exchanger results in a lower coolant
velocity, which, if below 2 ft/s, intensifies the
tendency for suspended solids to deposit. Besides
increasing the fouling factor, the sludge deposits
accelerate the pitting of tube surfaces.

570
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

ESL-IE-81-04-97

References

r-~

1. Woods, D. R., Anderson, J. J., and Norman, S. L.,


Evaluation of Capital Cost Data: Heat Exchangers,
Can. J. of Chern. Eng., Vol. 54, December 1976.

2. The Marley Co., "Managing lo/aste Heat with the


~Iater Cooling Tower," 1970.
3.

~j

>~

, ~~

4O.000'ilPff\

'".,;
g
0

:.lE5.

~~~
';;" i

"
.JL

E 300
~

0.

E
:>
0

rt><>

..,
;

10.000 ggm

0.

'0.

.
...
~

100
4

14

10
12
8
Pipe velocity, ftls

16

.D
~

Pump end piping cost is reletively insensitive


to coolent velocity

"

>

TO\Nef duty: 100 million 8tu/h

To......, wetbulb tlmperlltUr,: 8OF

Fig. 6

350

:II
~

300

'".c,;

lp
...

ii 250
0.
E

5'
~

200

:>

'"c:

~i

&

0
u

t:

b"
,5 I
0'

Ii:

~.I:'

i""

:!

~2

, ,
,, '
,'

0.

';;
;;::

I----.J

82

'0

e
E
~ s

oi

'0.
'6.

..

-i><r

il:

;~

'"d
,

~.

0.

'1

a
~
:>
}

"

200

!:

20.000 gpm

G:~

'0

'"
~~

:;

'5 400

ii

';;

-<x1

.~.e

d
J'

..

J,

iH

'-

500

ii

~n

--.:lH
!'._

600 , . - - , - - - . , . - - , - - - . . . . . , - - - - . , , - - - . . . ,

'"d,

"~

The Marley Co., "Cooling Tower Fundamentals and


Application Principles," 1969.

To......routlet
t,moer.turt, -F

<;

5
u

Table t

Economic .dvanlage of a temperature cross


150
10

20

30
40
Cooling-tower range, -F

50
Proceu inlat:

900

250 OF

b. Tower dutY: 300 million: Bluth


TOMr ......lbulb temperatur,: aO-F

>-

'tJ
c

Process outlet: 115F

::l 800

'".,;
~

700

C_l
Cao2
No Croa.

C_3
CrOll

1.650

120
39
1,680

1.330

,500
820

0
700

34.500

6.600

33,700

6,100

32,000

20,000

17,000

15,000

3,000

Water outlet

0.

temperature, OF
Corrected LMTO, OF
Exchanger area. ft2

105
63
1,600

Incremental exchanger
investmen t. $

..sao

'0

ii
Q.

,E

600

0.

500

:a" 400
0>
~

,
,, ,,
' ,

f/

Coolant flow, gpm


Incremental coolant
investment. $

Tower outlet
t,mperature, ~F

Net inYe1tment

penaltY. S
Coolant operating
COst, S/yr

c
<3

Net C$erating

300
10

COSt

penalty. S/vr
20

115
54

30

40

50

Cooling-to'Ner range, -F

Coolent economics may be estimeted


from tower-range reletionship

Fig. 4

571
Proceedings from the Third Industrial Energy Technology Conference Houston, TX, April 26-29, 1981

i:

Você também pode gostar