Deterrnination of Hoat franster Coofficients 667

pplicabletoturbulent conditions are more accurate than those for laminar conditions.
obtained from the correct use of cquations in the turbulent flow
coefiicients
be within +20 percent of the true experimental value; but
range
will
Filmordinarily values deter
for
viscousflow conditions or for condensation, boiling, natural convection, and
of heat exchangers may be in error by more than 100 percent. Be-
mined
for shellsides inaccuracies in the methods for estimating film coefficients, some
those
theinherent
cUISe engineers prefer to use
of overall coefficients based on past experience, while oth
includea large safety factor
design in the form of fouling factors or fouling coefficients.
ers
Film Coefficients for Flulds Flowing Inside of Pipes
Tubes(No Phase Change)
and
equations are based on the correlations presented by Sieder and Tate:
Thefollowing
flow (D;Gi/u < 2100),
&
Forviscous

(Pe"()"()
0.14
h; = 1.86
D;
1/3 0.14

= 1.86( Am, Cp (14-17)

IkL
For turbulent flow. above the transition region (D, Gi/p > 10,000, 0.7 < Pr < 160 and
L/D, > 10)
0.8 1/3 0.14
h; = 0.23
() w
(14-18)

For the transition region (2100<D;Gi/u <10,000), see Fig. 14-9.

In Egs. (14-17) and (14-18), D; is the inside diameter of the tube, G, the mass
velocity inside the tube, C, the heat capacity of the fluid at constant pressure, k the
thermal conductivity of the filuid, u the viscosity of the fluid (subscript w indicates
evaluation at the wall temperature), L the heated length of the straight tube, and rm the
mass rate of flow per tube. Physical properties k, Cp, and u are evaluated at the av
erage bulk temperature of the fluid.
Equation (14-18)is recommended for general use, bearing in mind that the stan
dard eviation of error, according to Hewittet al., is approximately 13 percent. Equa
bons (14-17) and (14-18) are plotted in Fig. 14-9 to facilitate their solution and to
indicate values to use in the transition zone.
ln some references a constant of 0.027 is used in place of the 0.023 value used in
E4- (14-18). The 0.023 constant is more appropriate for water, organic fluids, and gases
a noderate AT values, while the 0.027 constant is more applicable to heat transfer of
quids, particularly when the viscosities at the wall and in the bulk fluid
asignificant show
difference value.
in
E.N. Sieder and G. E. Tate, Ind. Eng. Chem., 28: 1429 (1936).
G.F. Hewit ,,G.L. Shires,
andT. R. Bott, Process Heat Transfer, CRC Press, Boca Raton, FL, 1994, Sec. 2.4.7.3.
638 CHAPTER 14 Heat-Transfer Equipment-Design and Costs

103

102

L/D = 60
10 100
200
400

1
10 102 10 10 105
10
Reynolds number

Flgure 14-9
Simple procedure for estimating film coefficients for fluids flowing in pipes 2n4
tubes [based on Eqs. (14-17) and (14-18)].

The use of computers has made it possible to develop a more accurate andDoe
widely applicable correlation. Such a correlation was developed by Guielinski au
summarized by Hewitt. This corelation incorporates a semitheoretical basis with te
Prandtl analogy to skin friction in terms of the Darcy friction factor fp. (The latter is
4 times the Fanning friction factor.) The correlation has beenshown to cover a rang
in Reynolds number from 2300 to 1,000,000 and a range in Prandtl number from 06
to 2000.

h; =
k(fo/8)(Re - 1000)Pr
D,[1+ 12.7(fp/8)/2JP-2/3 -(] (14-19

where
1 (14-191)
fo =
(1.82 log Re - 1.64)2
fluid conditions. or vs
All the properties in Eq. (14-19) are evaluated at the bulk bythe correction facor
liquids, the right-hand side of Eq. (14-19) is multiplied and wall temperaturemult-con-
(Pr,/Pryl, wherethe subscripts band wrefer to the bulk is
right-hand side of the equation.thebulk
ditions, respectively. Similarly, for gases, the temperatures of
plied by the correction factor (T,/T,)045, using the absolute
fuid and the wall.
Sec. 2.5.1.
Exchanger Design, Begell House, New York, 1992,
'G. F. Hewitt, ed., Handbook of Heat
 Determination of Heat-Transter Coefficients 669

Qrassonally
,asituationissencountered in whicha fluidis transmitted through acon-
crosssection. The heat-transfer coefficients for
noncircular turbulent flow
a determnedby using the same equations that applyto pipes and tubes
be
dtameter appearinginthese equations is replaced by an equivalent diameter
can,

this
case
rDobtainedifFthis equivalent diameter is taken as 4timesthe hydraulic
i ap i e
results are
s hydraulicradius is defined as the cross-sectional flow area divided by
ihe

2.
Best
the
where .Forexample,
perineter
ifheattisbeing transferredfrom afluid in acenter pipe
eheated througha.circular annulus, the film coefficient for the outside surface
would,be based on the:following equivalent diameter
fowing D,:
ppe
inner
afthe
[D;/4 nDË/4] D; D; (14-20)
hydraulicradius =
x D,
D,= 4
and outer diameters of the annulus, respectively.
theinner flow
and D, are
between the hydraulic radii for heat transfer and for fluid
D,
where difference
equivalent diameter for evaluatingfriction due
12,the correct
The Chap.
In annulus was
noted. cross-sectional flow area
shown to be 4 times the
be
should flowin the
4(7D;/4 - nD{/4)/(TD, +nD) D, - D.
wettedperimeter, or
o the
fuid

bythe
vided
Flowing Outside of Pipes
Goefficientsfor Fluids
Flm
(No Phase Change) dimensions such as
Tubes over bluff bodies of finite
and involves flow
often cylinders is particularly
transferflow airfoilshapes. Heat transfer to and from Since the heat
Heat
spheres, or application to cross-flow in heat exchangers. been neces-
cglinders,
imDortantbecause
of the dynamic behavior of the fluid, ithas
cross-flowislinkedto the on experimental data. Onesuch
Iansferin
suitable correlations basedessentially cylinder or sphere is that devel
yiodevelop cross-floW across a single
welation developed for
ChurchillandBernstein.0.62Re'/p/3 Re
5/8¬4/5
(14-21)
onedby 1+
h,D, = 0.3 + 28,200
(1+(0.4/Pr23ju4 approach velocity V
k Vp/u, based on the
D,
numberis defined as
where the Reynolds
D, of the cylinderorsphere. so that one of the fuids fiows
diameter exchangers are designed fiuid. The
andthe outside heat the cross-flowofanother single
plants,
Inprocess tubes andis heated orcooledby tubes is similarto that for a the
lrougha bankof coefficientfor the first row of subsequentrows oftubes,
in cre
QUtsdeheat-transfer notedin Eq. (14-21). However, and additional turbulence
flowing
elncross-flow,
as local velocity the fluid
bytheincreasedheat-transfer coefficient for
a ansteris
enhanced
average maximum velocity betweenthe
aled by the preceding tubes. The
correlate withthe properties, andthe geometry
of
found to been
between the tubes has been
approach velocity),thefluid experimental datahave
tubes Vmas (ratherthan empirical correlations based onofitsrelativesimplicityandthe
the:
the tube bank. Various recommendedbecause
piblished.
Equation (14-22)is s
300(1977).
SW. Churchil and J. Bermstein, J. Heat Transfer, 999:
660 CHAPTER 14 Heat-Transfer Equipment-Design and Costs

ease with which it can be incorporated into a computer code.

recognized that corelations of this type are not as accurate as
internal flow in tubes and have a standard deviation of at least 25
flowand 15percent for turbulent flow.
HotwhDeoesverecre,ndtevSetlohorpoenddarmhi o
The majority of cross-flowtube banks have geometries that are
as in-line square pitch arrays, in which tubes in
direction of fluid flow, or as staggered triangular
sive rows are staggered so that the tubes in any
successIve
pitch arrays, dpositionesignaed énte
rows are in line
in
given row oCcupy which tubes a iinin
the midpoint of the open space between the tubes in the
for details.) The empirical correlation that applies to bothprevious row.is(See IFi'gine wth
suC E,
h,D, geometries of the Tor mlA4
= aRe"P,04F, F
k
where a and m are correlation constants tabulated in Table 14-1, the (l4
Reynolds number is given by
Re = D,Vmax p cros-toN
the correction factor Fi, for wall-to-bulk physical property variations,is (14-0
0.26
evaluated fom
Fj = Prb for Pr < 600
Pru. (142
and the correction factor F2, which accounts for the effect of the number of
in either the in-line or the staggered configuration, is presented in Table tj4.)
ube roF.ws

Table 14-1 Values of a and m for use wth Eq. (14-22)t

In-line banks Staggered banks
Range of 1.2 <Pr/D, <4 1<Pr/D, <4
Re number
103x 10 0.742 0.431 1.309 0.360
3x10- 2x 10 0.211 0.651 0.273 0.635
2x10° -2x106 0.116 0.700 0.124 0.700

"ESDU, Convective heat transfer during crossflow of fluids over plain tube banks, ESDU
International plc, London, Data Item No. 73031, Nov. 1973 unamended. Reprinted by permisson.

Table 14-2 Correction factor Fa as a function of tube rows in

Eq. (14-22)t
Number of F, in-line banks Fz, staggered banks
tube rows (Re> 2x 10) (Re > 10)
3 0.86 0.85
4 0.90 0.90
6 0.94 0.95
8 0.985
0.98
10+ 0.99
0.99
tubebanks,
"ESDU, Convective heat transfer during crossflow of fluids over plainunamended.
ESDU International plc, London, Data Item No. 73031, Nov. 1973
Reprinted by permission.
 Determination of Heat-Transfer Coefficients 661
configuration Vmax iS given by
tube
either.
Vmax =
pNT(Pr - D,)L (14-22c)
wheremristhetotal mass flow rate of the fluid, p the fluid density, N, the number of
nubesinthetubebank, Pr the tube pitch (as defined in Fig. 14-4O), D, the outside tube
and Lthe tube length. Equation (14-22c) assumes that the sum of the
diameter,
betweentubesin agiven transverse rowis equal to the sum of the spaces
oW.Forexample, in an in-line square array, the gap at each edge oftube
the diameters
tube bundle
in isa
equaltoone-halfthe transverse tube spacing. Excessive gaps at the edge of tube bun-
should be avoided to minimize bypass flow.
dles
Film Coefficients and Overall Coefficients for Various
Beat-iTransfer Situations
Table 14-3briefly summarizes heat transfer coefficients and fouling resistances for sen-
sible heat transfer in tuubular exchangers, while Table 14-4 provides information on
thesesame two quantities for condensation and vaporization. Since there is consider-
these de between the lower and upper values of many of these
coefficients,
erable care must be exercised in their use. Accordingly, design engineers considto
often elect
use overall heat-transfer coefficients directly without attempting to evaluate the indi-
nal flm coefficients. When this 1s the case, the engineer must predict an overall
nficient on the basis of past experience with equipment and materials similar to those
involved in the new study. Design values of overall heat-transfer coefficients for many
of the situations commonly encountered by design engineers are listed in Table 14-5.

Table 14-3 General range of heat-transfer coefficients and fouling resistances for
sensible heat transfer in tubular exchangers
Fluid conditions h, W/m².Kb h,, m².K/W
Water, liquid 5x 10 lx10+ lx10-42.5x 10-4
Light organics,° liquid 1.5x 10 -2x103 1x10-- 2x10-4
Medium organics, liquid 7.5x 10²- 1.5x 10 1.5x 10-4 4x 10-4
Heavy organics, liquid, heating 2.5x 10²-7.5x102 2x10-4- 1x10-3
Heavy organics, liquid, cooling 1.5x 102-4x102 2x10-4- lx10-3
Very heavy organics liquid, heating 1x10-3x102g 4x 10-4-3x10-3
Very heavy organics liquid, cooling 6x10 1.5x 1028 4x 10-4-3x10-3
Gas," p = 100-200 kPa 8x10 - 1.2x10? 0- 1x10-4
Gas," p = 1MPa 2.5x 10 4x 102 0- l×10-4
Gas," p = 10MPa 5x10 -8×10 0- 1x 10-4
Data:aadapted from Heat
Coeffi cients based on Exchanger
clean DDesign Handbook, G.F. Hewitt, ed., Begell House, New York, 1998, Sec. 3.1.4.
surface area in contact with the fluid and an allowable pressure drop of about
50-100 KPa for the fAuid.
Hydrocarbons through C&, gasoline, light alcohols, ketones, etc.; u <0.5 x10-3 Pas
'ALubebsorberoils, oifuell, hot gas oil, kerosene, and light crudes; 0.5 ×10- <u<2.5 x10-* Pa-s
oils, cold gas oil, heavy and reduced crudes; 2.5 x 10-3 <p< 5.0 x 102 Pas
I"EstTars,im,atasphalts, greases, is approximate
"Air, ion of coefficientspolymer
melts, etc.,andu depends
> 5.0 × strongly
10- Pa·son the temperature difference.
nitrogen, OXygen, carbon dioxide, light hydrocarbon mixtures, etc.
64 CHAPTER 14 Heat-Transfer Equipment-Design and Costs

DETERMINATION OF
IN HEAT EXCHANGERSPRESSURE DROP Correction factor B; to acC
It may seem strange to include pressure drop The expansion, and revers
changes. However, it has become
generally result in larger increasinglyconsiderations
heat-transfer apparent that
in the de
coefficients and thus increased
tion,
sudden
Bi = 1+
area and consequently lower heat exchanger costs for a less
the other hand, increased fluid velocities lead to
increased
exchanger, which can result in higher operating costs. In pressure
requireheatd
given rate of h heat
drop trtahnrsoiuex.0 of BË = 1
maximum pressure drop that can be tolerated in a heat fact, often
operating cost at minimum levels. Thus, most heat exchanger ihere
is a
to keep temnperature
require an evaluation to determine whether the exchangerdesign he
pressure drop limitation auad
speihea where
AT;is the

exceeded.
The major source of pressure drop in a heatexchanger is friction
calculaioNS Shell-Side Pressure D
dueto fricti
pressuredrop the norma
the fluid as it flows through either the shell or the
tubes of the
sure drops occur because of friction due to sudden expansion, exchanger.
reversal in the direction of flow of the fluids. Changes in
sudden olher encountered pe
b
The calculatedin
canbe
4times
the hydraulic
equaltocontraction

energy can influence the pressure drop, but these two effects are
vertical head Contrandacuonki,nehk due to usually
and expa
across t
shell side is result, no singl
smalland can be neglected in most design calculations. ordinarily relaivey he used. As a
shell sideof
thesetyp
onthe
evaluation is
presented i
Tube-Side Pressure Drop an of flow di
For the case to ap
used
It is convenient to express the pressure drop for heat exchangers in afom sinl. equation can be tubes
of the
the Fanning equation as presented in Chap. 12. Because the transfer of heat is in side diameter
volved, a factor must be included for the effect of temperature hange on the fhce
factor. Under these conditions, the pressure drop on the tube side of aheat exchena
may be expressed as where B, is a
correction
tube
28, G}Lnp flow, recrossing of
Ap; = (1423) unbaffled tubes, B, is
P,D, , estimated as equal to th
tube at the bulk temperature. Th: tion factor and N, the
where the subscript i refers to the inside of thebased on the arithmetic-average bul tion factor is a functie
Fanning friction factor f, for isothermal flow is turbulent flow in a smooth pipe, te configuration of the to
temperature of the ffuid. For fully developed ber D, G,/ho is betw
following correlations for f; are recommended:
(14-23a
f;=0.079 Re-0.2s for Re < 2100
(14-236)
t; =0.046 Re-0.2 for Re > 2100 'The friction due to sudd
1.1(u;/uyj025 whentheis den contraction F is give
factor ; for nonisothermal flow is equal to the Reynolds number to reversal of flow direct
The correction 1.02(u;/u)o when temperatureoftke friction term for design
less than 2100 and
Reynolds number is
viscosity at the arithmetic-average bulk oftheinside-tube *K, is obtained from (
tube croSs-sectional are
greater than 2100; u, is the fluid at the average
temperature involvedwitha
SFor flowS with higher
tluid, and uy the viscosity of the thanfone tube pass
is
recognizes if more book, 6th ed., McGraw
wall surface. The np term
length ofL.
 Deterrnination of Pressure Drop in Heat Exchangers 666

corrcctionfactorB, to account for the added friction due to sudden contrac-

The
sudaden
expansion,and reversal
of flow direction for the fluid is given by
F, + F + F,
pon, B = 1+
26,G}L/(p D,0;) (14-23c)'

B, =1
0.51 K1np, AT,iui/u,,j02%
(T,in - Tiou) Pr (14-23d)
4yi sthetemperature difference across the film located inside the tube.
where
Shell-Side Pressure Drop
pressuredropdue
tofriction when afluid is flowing parallel to and outside of tubes
The calculated lin the normal manner described in Chap. 12 by using a mean
diameter
canbe 4 timesthe hydraulic radius of the system and by including all friction effects
cqualcontractionand expansion. In heat exchangers, however, the fluid
t o
flow on the
dueto is usually across the tubes, and many types and arrangements of baffles may
side
shell result. nosingle explicit equation can be given for evaluating pressure drop
eUsed.
Asa
the
shelIsside of these types of heat texchangers. Adetailed procedure for making such
onevaluationis presentedin the shell-and-tube heat exchanger
an
analysis.
Forthecase off flow directly across tubes, either in-line or staggered, the following
equationcan be usedto approximate the pressure drop due to friction based on the out-
diameter off the tubes at the bulk temperature of the fluid:
side
Ap, =
2B,f N,G?
Po
(14-24)
where B, is a correction factor to account for friction due to reversal in the direction of
flow, recrossing of tubes, and variation in the cross section. When the flow is across
nbafied tubes, B, is assumed to be equal to unity, while across baffled tubes B, is
stimated as equal to the number of tube crossings. In Eq. (14-24), f' is amodified fric
tjon factor and Ny the number of tube rows over which the shell fluid flows. The fric
tion factor is a function of the shell-side Reynolds number and is influenced by the
configuration of the tube array. For the normal case where the external Reynolds num
ber D,G./u, is between 2000 and 40,000, the friction factor may be represented as
-0.15

(14-24a)

"The friction due to sudden expansion F, may be obtained from F, =(V - b)/2. The friction due to sud
en contraction F, is given by F= K V}/2, where V,, V.and K, are defined in Table 12-1. The friction due
Oreversal of flow direction F. depends on the details of the exchanger construction, but a good
fhction tern for design work is F, = 0.5V} (n, - estimate of the
"K is obtained 1)/(2n p).
from (11- S,/SH )² + K + 0.5(n, 1)/np where S;/ SH is the ratio of the total inside-
uDe CrOSS-sectional area per pass to the header
YFor flows with higher Reynolds numbers, cross-sectional area per pass.
see R. H. Perry and D. W. Green, Perry's Chemical Engineers' Hand-
OK, Oth ed., McGraw-Hill., New York.
1984, pp. 5-50 to 5-53.
 666 CHAPTER 14 Heat-Transfer Equipment-Design and Costs

where bo has been approximated by Grimson for in-line tubes ast

bo = 0.044 + 0.08xL
(IT- 1)(0.43-1.13/s)
and for staggered tubes as (424
0.11
b, = 0.23 +
(IT - 1)1.08

In Egs. (14-24b) and (14-24¢), xT is the dimensionless ratio of

center-to-center distance) transverse to fow to the
outside tube pitch (ie., the
diameter t
the dimensionless ratio of pitch parallel to flow to the outside tube
results are obtained if x is between 1.5 and 4.0. For
design purposes thdeiameter. Bes
extended down to 1.25.
range anb.
EXAMPLE 14-3 Estimation of Film Coefficient and Pressure DropInside
in a Shel-and-Tube Exchanger Tubes
Ahorizontal shell-and-tube heat exchanger with [wo tube passes and one shell pass is
heat 9 kg/s of 100%% ethanol from 15 to 65 Cat atmospheric pressure. The ethanol passesbeing ust
through te
inside of the tubes, and saturated steam at 110 Ccondenses on the shell side of the tubes. The tubes
are steel with an OD of 0.019 mand a Birmingham wire gauge (BWG) of 14. The exchanger om
atotal of 100tubes (50tubes per pass). The ratio of total inside-tube cross-sectional area per pasD
header cross-sectional area per pass S/Sy can be assumed to be 0.5. Estimate the filn coeficet g
the ethanol and the pressure drop due to friction through the tube side of the exchanger.
Solutlon
Equations (14-18) and (14-23) can be used to determine the heat-transfer coefficient and presug:
drop inside the tubes, respectively. Note that the average wall temperature will need to be evaluatd
todetermine the effect of viscosity on the beat-transfer coefficient. This can be accomplished wiba
with the ethanol and oar
simple estimate based on the thermal resistances in the films associated
densing steam.
Physical property data
Atan average temperature of (15 + 65)/2= 40°C
u for ethanol at 40°C= 0.0009 Pa-s
k for ethanol at 40°C= 1.63 x 10-+ kJ/s-mK
C, for ethanolat 40°C= 2.594 kIkg-K
pfor ethanol at 40°C=785 kg/m
k, for steel = 0.045 kJ/s-m-K

'E.D.Grimson, Trans. ASME, 59: 583 (1937); 60: 381 (1938).

 Determination of Pressure DropinHeat Exchages S67

configuration
Exchanger
Dube
outsidediameter D, = 0.019 m
Tube
inside
diameter D, = 0.015 m)
Ip = 0.002 m
Tubewallthickness
tube A, = 0.000177 m²
per
Insidesurfacearea/m
Flow
area A, = 0.047 m²/m tube length
mass velocity in each tube.
the
detemine

Fis,
Gi = 1017 kg/s-m²
Ac 0.000177(50)

Prandtl numbers are

Reynoldsand
The D, G: 0.015(1017)
Re = = 16.950
0.0009
2.594(0.0009)
Pr= p2
k 0.000163
= 14.32

average heat-transfer coeffñcient, applies under these condi-

(14-18),for evaluation of the
Equation remaining unknown in the relation is
Theonly
å, and this requires the temperature of the tube
tions. approximation, assume the following:
first
wal. Asa
ethanol flm = 70% of average total AT, or 0.7(110 40) = 49C
AT across
steam film =10% of average total AT, or 0.10(110 40) = 7°C
AT acroSS
40+ 49 = 89C
Wall temperature =
89°C= 0.0004 Pa-s
u of ethanol at
0.0009 =2.25
0.0004

From Eq. (14-18)

= (0.023) 0.015
0.163
6,950)0 (14.32)/(2.25)014 = 1650 W/m'-K
coefficient of 5000 W/m-K is
Now check on the AT assumptions. From Table l4-3a fouling
adequate for the ethanol. No fouling coefficient will be used for the steam.
1.0 x 10* Wim-K.
From Table 14-4, the steam film coefficient can be estimated conservatively as
lne overall heat-transfer coefficient can now be estimated from Eq. (144a):
D,ln(D./D:) D;
hid 2k h,D,
0.015 In(0.019/0.015) 0.015
+ 10,000(0.019)
1650 S000 2(45)
=0.00061 +0.0002 -+0.00004 +0.00008 = 0.00093 m-KW
668 CHAPTER 14 Heat-Transter Equipment-Design and Costs

Percent AT across ethanol film 0.00061

0.00093 -(100) = 65.6% of total AT
Percent AT across steam film=
0.00008
0.00093-(100) = 8.6% of total AT
The AT assumptions are sufficiently adequate to proceed.
In the determination of the pressure drop due to frtction in a new
constant over the length of the exchanger.
tube or pipe,
1
s ume heA U'%
U, = 0.00093 1075 W/m'.K
(110 15) (110 - 65)
Alo.log mean In[(110 - 15)/(110 - 65)] = 66.9°C
From Eq. (14-6)
!= (mCp AT),=9(2594)(65- 15) =1.17x 106 w
By Eq. (14-5)
1.17 x l0
A, = 4/U; Alo.log mean = 16.3 m²
1075(66.9)
Length per tube is then
16.3
L= = 3.47 m
0.047(100)
To obtainthe pressure drop inside the tube, use Eq. (14-23) and Eqs. (14-23d) and (14-23b) to evl
uate B, and f;, respectively.
A=1+[0s1Kin, aT,. u/uP"/Tia-IiPi)
+K, + 5n,-1)
K-(-)
where (from Table 12-1) K, = 0.4(1.25- S;/Su) = 0.4(1.25 -0.5) = 0.3 and
0.5(2- 1)
K, = (1-0.5) +0.3+ 2
=0.8

f; =0.046 Re-02 =0.046(16,950)-02 =0.0066

0.51(0.8)(2)(0.656)(110 40)(2.25)023 =1.16
B.=1+ (65- 15)(14.32)2/3
0.14

AP, = 281,G}Ln,
PD;O,
2(1.16)(0.0066)(1017'(3.47)(2)
785(0.015)(1.02)(2.25)0.4
= 8169 Pa 8kPa
slightlygreaterthan
ris
The pressure drop due to friction through the tube side of the exchanger dropis6kPa.
Asoftware program utilizing the given data determines that the pressure
706 CHAPTER 14 Heat-Transfer Equipment--Design and Costs

fluid leakages, the widely accepted Bell-Delaware method

factors based on experimental data. However, these
limitedtothe shell-side configurations from whichitheempirical
utilizes several coy
more generic method covering the full range of possibledatabase has
initiated by Tinker and developed for hand calculation by Wil s & been obtar
faco cor ect ion
shell-side ar angemer
approach is known as the stream analysis method in and
which
designated for each of the posible flow routes of the fluid inseparat
the e fuid strear
changer. To provide an appreciation for these three methods, shell side of tr
Johnston,.
the procedural steps in each of the methods. this section suroma
Kern Method The corelation for the shell-side
heat-transfer
exxchangercoefficiwientth adevelbafteo
by Kern, and based solely on the operation of an industrial
of 25 percent, is given by
0.55 0.33 0.14
h, = 0.36
D. (14
where G, is the shell-side mass velocity obtained from
mT
Sm (143
Here rmT is the total mass flow on the shell side, and Sm
the cross-flow area measurei
along the centerline of flow in the shell. The value for S, is obtained from
Sm = D, PpL
(1432)
Pr
where D, is the inside diameter of the shell, P, the tube pitch, Pp the tube
or distance between adjacent tubes, and L the distance between clearanc:
adjacent batties. Ihe
equivalent diameter D, for the square pitch arrangement is defined as
D, =
4(P; - 0.257 D;) (14-33)
I Do
where D, is the outside diameter of the tubes. For thee triangular puch. arTangement D,
is defined as

D, =
4(0.86 P? - 0.25r D) (14:34)

'ofDelawurbiy
"K. J. Bell,
Exp. Station "Cooperative Research Program on Shell and Tube Heat Exchangers," University
Bulletin 5 (June1963). and
AmecHI
*T. Tinker, Proceeding of General Mechanical Engineers
Discussion Heat Transfer, Institute of
Society for Mechanical Engineers, New on
York, 1951. Societytr
`M. J. N. Wills and D. American
Johnston, 22nd National Heat Transfer Conference, HTD, vol. 36,
Mechanical Engineers, New York, 1984.
 Design of Key Heat
Exchanger Types 707

LFriction
factor
f

102 10 104 10 10 10
10 Reynolds number

Figure14-44
frictionfactorfas a fuunction of shell-side Reynolds number
Plotof

Ap, observed on the shell side of the heat exchanger, assuming no

loss
The
pressure simple relation
by the
uid leakage, is given 4fG}D,(Ng+ 1) (14-35)
Ap, = 2pD,(u/ w)Ql4
shell-side
obtained from Fig. 14-44 as a function of the
wbore the friction factor f is The
number of baffiles specified for the heat exchanger.
number of
Bavnolds number and Ng isthe fluid passes across the tube bundle. The
accounts for the
N +1value by
baffles in the shellis given (14-36)
L -1
Ng = Latb
the spacing between baffies, and t, the
of the exchanger, Lg
where L, is the length of the tube bundle
is
thickness of the bafe. the centerline
nunmber of tubes located across
The maximum
determined from (14-37)
D, - A, DoTL
Nb = Pr and
PT recommended by TEMA
diametral clearance
where Ap is the bundle-to-shell
diameter, designated as
DorL·
tothe ideal"
D, - A, is the tube bundle correctionfactors
appliesvarious for theleakage ofthe
account
Method This methoddropevaluations to
Bell-Delowaware
crOSs-fl heat-transfer and pressure
 CHAPTER 14 Heat-Transfer Equipment--Design and Costs

shell-side fluidthat occurs through gaps between the tubes and

and the shell, and bypassing of the flow between the tube
tion factors also account for the effect of baffle
bundlebafandflesthe, betwe n'
configuration complexitisehesl . Core
verse temperature gradient on heat-transfer during laminar flow.
factor is applied only in those rare cases where the shell-side The \at e
than 100. Reynolds COT eclion
The initial step in the Bell-Delaware method is to calculate the
heat-transfer coefficient. This requires obtaining a Reynolds number ideal cross-o
Re =
pVmax Do defined as
where p is the fluid density, uthe fluid viscosity, D, the outside tube
Vmax the maximum fluid velocity. The last term is defined as the
between the tubes near the centerline of flowand is given by maximdumiamerevelr, ociandty
mT
Vmar =
pSm (l439)
where Sm is the cross-flow area near the centerline and lis defined for square and tran.
gular pitch arrangements as
(DorL - D,)(Pr - D,)1
DotL t (1440)
P,
Definitions of S,, for rotated square pitch and rotated triangular pitch arrangemens e
earlier.
given in the Bell reference mentioned coefficient is obtained from
The actual shell-side heat-transfer
h, = h,JcJlo (144)
coefficient calculated from Eq, (14:2)
where h; isthe ideal cross-flow heat-transfer exponentsprovided in Tables l41
appropriate values for the constant and
byusing the factors Jc. JL, and J; are evaluated with
relationships
and 14-2. The correction
developed by Bell as described below.
accounts for baffie configuration is a
function of
The correction factor Jc that cross-flow. This fraction is calculated from the
assumed to be in
the fraction of tubes
relation
D, - 2L, (1442)

F= - +
2(D, - 2L) sin
cos-! D,DotL-2L-? 2cos-!
DorL
DoTL Here b, is u
from the product B,D, /100.shell
distance obtained inside diameter
where L is the baffle cut between 15and 45 percent the
of
percentage baffie cut, normally tube-to-bafileleakage
Figure 14-45 is used to obtain Jc.relatedtothe shell-to-bafle and
is
The correctionfactor J These two leakage areas are evaluated trom
respectively. (1440)
areas Ssb and Stb.
S,, =
 Design of Key Heat Exchar yA 709

13
B = 45% B, = 15%
Working range
12

J, 1.l
Correction
factor

1.0
Original curve
0.9

0.8 Linear approxirmation

J= 0.5 +0.72F
0.7

0.6

0.5 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.0 0.1
F

Flgure 14-45
Correctionfactor for the influence of baffie configuration. (FromR. H. Perry and
COGreen, Perry's Chemical Engineers' Handbook, 4th ed., McGraw-Hill,
ew York, 1963. Reproduced with permission of The McGraw-HillCompanies.)

and
1+ F (14-43b)

where A.+ and Ajs are the shell-to-baffie and tube-to baffe diametral clearances, re
spectively, and JL is related to the ratio (S,b + Srb)/ Sm in Fig. 14-46 with S,b/(S,b + S;e)
as the parameter. The cross-fiow area S is obtained directly from Eq. (14-40).
The correction factor Jg, which accounts for the bypass in the bundle-to-shell gap,
is expressed as a function of Fap. the fraction of the cross-fow area available for
bypass flow. Here Fbp is defined as

Fbp = SmLB(D, - DorL) (14-44)

To minimize the bypass flow in the bundle-to-shell gap, pairs of sealing strips are
added to the shell side. These affect the heat-transfer rate and need to be included when
obiaining the correction factor Ja. Figure 14-47 shows Ja as a function of the ratio
S3/S. with Ns/N, as the parameter. Here N, is the number of pairs of sealing strips,
and N, is the number of cross rows as estimated from
N =
P-) (14-45)

eTP 0s the distance between tube rows in the diretion of flow and is equal to Pr
quare tube arrangements and 0.866P, for triangular tube arrangements.
 CHAPTER 14 Heat-Transfer Equipment-Design and Costs

1.0

S, 0
0.8 (AIlleakage tube-to-baffle)
Correction
factor
J,

0.6
s/s/s/a
0.4
0.25
0.50
0.75
Sy =0
0.2
(AIl leakage baffle-to0-shell) 1.00

0.0
0.2 0.4
0.6
(S, + Sh)/S,m 02

Figure 14-46
Correction factor for the effect of tube-to-baffle and
for calculating heat-transfer coefficients in shell-and-tube
(From R. H. Perry and D. W. Green, Perry's Chemical heat leakagebaffle-to-shelexchanger
l s.
Tth ed. McGraw-Hill, 1999. Engineers'
Handbook,
Reproduced with permission of The McGraw.H;i1
Companies.)
1.0

0.9
Correction
factor
JR Recommended
0.8 limit

10
0.7 -

0.6 20

0.5

(No sealing strips)

0.4 Re > 100
- 1 Re s 100
0.3
0.1 0.2 0.3 0.5 0.6 0.7
0.4
S,/Sm = Fop
Figure 14-47
Correction factor for the effect of bypass on the heat-transfer coefficientin
shell-and-tube heat exchangers. N is the number of pairs of sealingsis
N, is the number of cross-flow rows. (From R. H. Perry and D. W. Green,
Repro-
Perry's Chemical Engineers' Handbook, 7th ed., McGraw-Hill, 1999.
duced with permission of The McGraw-HillCompanies.)
 Design of Key Heat Exchanger Types 711
shell-sidepressure drop utilizing the
Calculationsfor the for
to those
made theheat-transfer calculations. Bell-Delaware
First the
method
similar
cross-flow is determined, andthen correction factors are pressure drop for
are
and bypass. Separate calculations are made for the
applied to account for
ideal
leakage the configuration correction. cross-flow and win-
thereby eliminating
o
baftie
cross-flow pressure drop is
zones,
dow ideal calculated from
The
Ap, = (Ka + Nk)(PpV2 max
2 (14-46)
constantK, is related to the
pressure losses of the fluidat the inlet and exit
wherethe approximated
generally with a value of 1.5. (See Table 12-1
tube bundle and is accounts for
additionaldetails.) The constant Kf
ofthe
for the friction and momentum losses
passes between each successive row of tubes in the bundle. Values for this
fiuid
asthe are given in Table 14-10 as a function of the Reynolds number for the shell-
constant values for Vmax and are obtained from Eqs. (14-39) and (14-45).
side fluid. The
respectively.
drop for the ideal'" window zone between baffles is given by Bell as
The pressure
26rmT L New Le
+ + for Re < 100 (14-47a)
Apw p(S,Su)1/2 Pr - D, D" Sm Sup

(2 +0.6Ncw)m for Re> 100 (14-47b)

Apw 2S, Swp

Table 14-10 Values of K for square and triangular tube configurations

K, (Square pitch) K, (Triangular pitch)
1.25 1.50 2.0 1.25 1.50 2.:0
Re
9.4 4.0 29.0 12.0 6.0
10 21.7
102 1.2 0.65 3.20 1.80 1.25
2.35
0.38 0.24 0.99 0.70 0.60
10 0.48
0.38
10 0.42 0.34 0.24 0.51 0.43
10 0.29 0.22 0.20
0.29 0.25 0.18
0.23 0.18 0.17
106 0.27 0.25 0.18
Ihe 1.25, 1.50, and 2.0 values in Table 14-10 are the pitch-to-diameter ratios for the specific tube arrangement.
ierpolation relationships for a tube pitch of 1.25 have been developed by Zukauskas and Ulinskas as follows:
For in-line square
arrays
0.927 x 107 0.100 >x 10!!
Kf =0.267 + 0.249 x 10+ for 2 x 10 < Re < 2 x 106
Re Re2 Re3
For staggered arrays
0.599 x 1013
Kf =0.245 + 0.339 >x 10 0.984 x 107 0.133 x 10!1 for 10 < Re < 106
Ret
Re Re2 Re3
Hemisphere Publishing. Washington, 1983,
A.Zukauskas and
Sec. 2.2.4. R. Ulinskas, in Heat Exchanger Design Handbook,
 cMAPTR14 Heat-Transfer Equlpment-Design and Coste

Here New is the number of effective cross-flow rows in

the
New = 0.8Le
PrP
window zone.
The window flow area S,, is defined as

S.=cos-' Dp - Da(1-
4
D;)"]--FN
where D is defined as (D, -2L)/ D, in radians and N is the
in the tube bundle. The equivalent diameter in the window zone
Dis
Dy =
4S, gven by
4[(r/2) N(1 - F)D, + 2cos-'Dal
The pressure drop across the shell side is then given as (l4g
Ap, = [(Ns - 1) Ap, Ra+Ny Apu]R +24p:Ra i+
N. (145
where the correction factors Rg and R are obtained from Figs. 14-48 and
spectively, and Ng is the number of baffles required as evaluated
from Eq, (14:A
Wills and Johnston Method The stream analysis method has formed the basis i
many modern computer programs. Asimplified version offthis method can serve zs1
means of verifying the results obtained with various commercial computer programs

1.0

0.9
Correction
factor
RB
0.8

0.7

0.6 -
-
0.5
(No sealing strips)
0,4 Re > 100
Re < 100 0
0.6
0.3 0.4 0.5
0 0.1 0.2 0.3
Fbp
Peray
(FronmR.H
Flguro 14-48 drop. McGrav-Hill
Correction factor Rp for the infiluence of bypass on pressure ed,
Engineers'Handbook, 7th
Companies,)
and D. W. Green, Perry's Chemical The McGraw-Hill
1999. Reproduced with permission of
 Design of Key Heat Exchnger Types 717

solutto
three methods have been outlined briefly in he shell-and tube
for all section.
following data for water:
Dp
e rocedures

providesthe
30 C 35 C 40 C
Phstkaly
propertydata
conductivity.k, kJ/s-m-K
0.000616 0.000623
4.179 0.000632
Thermal kJkgK
capacityCG. 0.000803
4.179 4.179
0.000724 0.000657
VisosiY
JP , as 995
Heat 995 995
pensiy
p,.
kgm
schanger
contiguration
chell
internaldiameter D, = 0.584 m
Dheoutsidediameter D, = 0019 m
P;=0.0254 m
TIbepitch(triangular) N=384
Nurmberof
tubes L¡=0.1524 m
L,=3.66m
Baflespacing
A, =0.035 m
Bundle-to-shell
Shell
length diametral clearancet Ash =0.005 m
Chell-t0-bafilediametralclearancet
A,, =0.0008 m
Duhe-to-bafiediametralclearancet Ih,=0.005 m
Thicknessofbafilet Ns./Ne = 0.2
crOSs-flow rowt
strips per
by J. Taborek, in Heat Exchanger Design Handbook, Hemisphere
Sealing recommendations
with
consistent Sec.3.3.5
Tems Washington, 1983,
Publishing,
0.00635m. The
between tubes Po is given as
KernMethod centerline. The gap
area at the shell given by Eq. (14-32).
Determinethe
flow
centerline of flow in the shell is
ornss-flowarea along the 0.584(0.00635)(0.1524) = 0.02225 m²
D, PpLB 0.0254
S, = Pr

Eq. (14-33).
Determine D, from 4[(0.0254)²- (n/4)(0.019)?1 0.02423 m
4(P - nD;/4) I(0.019)
D, = IDo

is 494.4kg/m's
The mass flow rate G, mT
0.02225 evaluationof the
G, S, temperaturerequires
water-film
average
coefficient at an
heat-transfer
To obtain the
Reynolds and Prandtl numbers. 0.02423(494.4) =16,550
D,G, 0.000724
Re =
4.179(0.000724) 4.86
0.000623
r=()
718 cHAPTER 14 Heat-Transfer Equipment-Design and Costs

From Eq. (14-30)

0.623
= 0.36
0.02423)16.550)°*" (4.86)0 9
(0.000803 4
= 3369 W/m?.K

Calculate the pressure drop on the shell side, assuming no effectfor

O.000657
number of baffles on the shellside is obtained from Eq. (14-36). any type of f
L, 3.66
Ng =
L + b 0.1524 +0.005 -l= 22.2 or 22
For a shell-side Reynolds number of l6,550, Fig. 14-44 provides a value of
tor. The pressure drop is obtained from Eq. (14-35) as
0.062for the frdn
4fG?D,(NB + 1)
Ap, =
2pD, (u/w)0.l4
4(0.062) (494.4) (0.584) (22+ 1)
2(995)(0.02423)(0.000803/0.000657)0.4 = l6,420 Pa
Bell-Delaware Method
The first step in this method is to calculate the ideal cross-flow
heat-transfer
coeticient. Cal
Vno from Eq. (14-39) and obtain Sm from Eq. (14-40) to substitute into Eq. (14-2)
(DorL - D.)(Pr - D,)
Sm =L,]D,- Doru + Pr where DoTL = D, -A, =03#

=o152/0035 +(0.549 0.019) (0.0254

0.0254 -0.0255 m'
11
Vmax = =0.4335 m/s
pSm 995(0.0255)

Re =
pVmax LD 995(0.4335) (0.019) =10,205
0.000803

Pr = Cpu 4.179(0.000803) =5.449

0.000616

The ideal heat-transfer coefficient is given by

h, = aRe" PrFA
D,
i
Staggeredtube aray,
where constants a and mare obtained from Table 14-1 for a
Eq. (14-22b), and F, from Table 14-2.
0.26
S.449 (0.99)
h, = 0616
0.019
(0.273)(10,205)065(5.449)0.4 4.345

= 5807 W/m².K
 Design ot Key Hont
obtained from Eq. (14-41). This Excharger Types
heat-transfer coefficient is
shell-sida

719
ond Ja using the appropriate
aual

1e
bypass.
Equation (14-42) correction factof oF.rs to account for bafflrequie res cbtaining
permits calculation
configuration,
and
slgt,
2(D, - 2L)
sin cos D, -2L, |-2cog-
DotL D,-2L.
DorL.
of25 percent
e cut
batle
Dott.
ira
L = 0.25D, = 0.25(0.584) =0.146 m

0.584 - 2(0.146)
D,- 2L, 0.549 =0.5318
DorL

r--I+ 2(0.5318) sin(cos0.5318) - 2cos0.53181 =06437

Frm
Fig.14-45
Jr
=0.55+0.72F, =0.55 +0.72(0.6437) = 1013
calculatethe leakage areas S,b and S,, from
J, Egs. (14-43a) and (14-43b),
obtain
To respectively.

2(0.146)
= (0.584) =0.003058 m
0.584

1+ F
2

0.0008
=n(0.019) (384) = 0.007535m
2

14-46, utilizing S,, and S.,.

The coection factor J is obtained from Fig.
S,, + Sib 0.003058+0.007535
=0.4154
Sm 0.0255
0.003058 =0.2887
0.003058+0,007535
S-b + Sib

Figure 14-46 provides a value of 0.56 for J. in the bundle-shell gap, obtain Fp, the fraction of
factor JR for bypass
lo obtain the correction
flow, with Eq. (14-44).
the cross-flow area available for bypass
0.1524 (0.035) = 0.2092
LB,
Fp =(D, - DorL) = 0.0255
Sm when N/N, = 0.2.
Jg valueof 0.935
14-47 can be usedto obtain a
Note that Fy =Sp/Sm, and Fig. then
coefficient from Eq. (14-41) is
The corrected heat-ttransfer
h=hJJL JB W/mK
5807(1.013 (0.56)(0.935)= 3080
=
 CHAPTER 14 Heat-Transfer Equipment--Design and Costa

Evaluation of the pressure drop using the Bell-Delaware method is

ing the heat-transfer coefficient. The ideal cross-flow
obtainedwith the use of Eq. (14-46). pressure dropsimiltahrroughtothe
Ymax
Ap, = (K, +N,K,) assume K, = 1.5
D,
N, = Prp =0.866P,.,for
triangular array
0.584 2(0.146)1
= 13.27
0.866(0.0254) 0.584

A value of 0.495 for K, is obtained by using the following relation given in

Table 14-10:
he iothee.
0.339 x 10 0.984 x 107 0.133 x 10!1
K, = 0.245 +
Re Re? Re' 0.599 x103
Re
(0.4335)²
Ap, = [1.5 + 13.27(0.495)](995) 2
= 754 Pa

Calculate the window zone pressure loss from Eq. (14-47b). First, determine the
S,, from Eq. (14-49). window foN
S, =cos' D, - Ds (1-D;) "
where
D, - 2L, =l
4

2L =l 2(0.146)
N-F)D,
Dg = =0.5
D, D, 0.584

(0.584)? 384
S,, =
4
(cos'0.5 0.5[1 (0.5))"}--0.6437)x(0.019
= 0.03298 m²

Next, calculate the number of effective cross-flow rows in the window zone from Eg, (1448)
0.8Le 0.8(0.146)
New = = 5.31
Pre 0.866(0.0254)
Nowcalculate the window zone pressure drop for Re> 100.

Ap, =
(2+0.6N.u)m
25,, S,up
(1)? = 375 Pa
= (2+0.6(5.31)b0 0255)(0.03298)(995)
the caku
obtain Rg. Use
Finally, estimate the leakage and bypass correction factors Rg and R- To For
of 0.82for R.-.
lated values of Fop and Ns/N, =0.2 with Fig. l4-48. This gives a value
14-49to obtain a
valueof.k
the area ratio values of (S,, + Sih)/ S,m, and S,,/(S«b +S;;) with Fig.
for R..
 Design of Key Hoat
Thepressuredrop across the shell is given by Eq. (14-S1). Exchanger Types 721

Ap, = 1(NB -
1)Ap, ,Ra +Ng ap.lR +2ap.
ka(1+
=[(22- 1)(754)(0.82) +
9481
-22(375)|
Pa
(0.365) + 5.31
13.27 )
= 7750 + 1731=

andJohnston
Method
The
heat-trarnster coefficient
Wlls calculated in this method is similar to that used in the Bell-Delaware
methodexceptthatthe value of tthe Reynolds number is estimated from m, =
sevaluatingtheflow stream resistance coefficients in Fig. 14-50 as
F,m,. To determine
f,requirese defined in Eqs. (14-55a)
(14-5Sc),(14-56), (14-58), (14-60), and (14-61).
through
Calculate the shell-to-baffle resistance coefficient n,, using Eqs. (14-56) and (14-57).
0.005\ /0.005\
s-r(a-))-(4 2
=0.004567 m?

0.036(2rt, /Asb)+ 2.3(2t,/A,,)-0.17

n, = 2pS?
0.036(2)(0.005)/0.005 + 2.3[2(0.005)/0.005]-0.177
2(995)(0.004567)?
= 50.75
resistance coefficient n, from Eqs. (14-58) and (14-59).
Calculate the tube-to-baffile clearance

s=N=(D.+))
m'
3847(0.019 + 0.0004)(0.0004) = 0.00936
=
2.3(21,/A,,)-0.177
0.036(2t, /A,b) +
n, = 2pS? 2.3[2(0.005/0.0008)]-0.177
(0.005/0.0008)+
0.036(2) 2(995)(0.00936)2

= 11,02
(14-60).
coefficient n,, from Eq.
Calculate the window flow resistance 1.9p0.6856S/S

n, = 2pS
calculations.
Bell-Delaware
=0.03298 m² from the
where S =0.0255 m² and S, 1.9exp[0.6856(0.03298/0.0255)] = 2.13

2(995)(0.03298)2
 Separation by Distilation 771
liguig,equilibriaarefor binary
datafor mixtures, and not for
touse the binary
multicomponent behavior. The
pairs and
Wilson,
multicomponent
combine these data with amimodel
xtures,to
NRTL, and UNIOUAC models' use
topredict
yapor/liquid
beIsed for the equilibria. When no binary data are
available, the
mostcommercial simulatorsprediction
can
case,model
ither
NFAC- basedof only
are capable on functional
providing reasonably accurate
groups. In
hemodynamicbases to predict phase equilibria.
eternminationof Number of. Equilibrium Stages Even though
shortcut
py-stagemethodsfor calculating the number of equilibrium stages andstage-in
are available
COmmercialcomputer software programs, itis
necessary to understand the fun-
bamentals
MLY involved to determine whether the computer results are realistic. Of the
shortcut,methods,the Fenske-Underwood-Gilliland method is the most widely used.
etermine the
number of
equilibrium stages, first the minimum number of stages
fo ninimum
reflux must be evaluated. The minimum number of stages Nmin 1S
ADd fromthe Fenskerelation
the.

ohrained
In[(*LK/*HK) D(XHK/xLK)&] (15-1)
Nmin
In(aLK/HK)av
fraction of the light key, xr the mol fraction of the heavy
isthe mol of the light key tothe heavy
the average geometric relative volatility
wherexLK
(aLKJHK)av and bottom products, respec-
key, whilethe subscripts Dand B refer to the distillate
relative volatility is calculated by using the
key, geometric average value for the the bubble point temper-
ively.Thetemperatureof the assumed overhead product and
dewpoint assumed bottomsproduct. Thus.
(15-la)
sthure ofthe )plaLK/HK)e]"
(aLK,HK)av = (aLK/HK (15-3) developed by
(15-2) and
minimum refluX ratio requires Eqs.
To determine the
Underwood,' namely, (15-2)
ajXFi
=l-4
rel
i=l
volatility ofcomponent i in the mixturemols
average geometric relative component i in the feed, and g the by
Where a; is the XE; the mol fraction of
The value of O is determined
key, offeed. components. The
ative to the heavy the feed trayper molvolatilities ofthe two key
liquid on
0fsaturated
and lies between the relative
ialand error
obtained from (15-3)
minimum reflux Rmin is ajXD.i
Rmia +l =)
(1968),
Jourmal, 14: 135
Prausnitz, AIChE
M. Liquid-Liquid Equilibria,
and J. Vapor-Liquidand
H. Renon
(1964), Multicomponent
86: 127
G.M. Wilson, J..Am. Chem. Soc.,Calculations for Netherlands,1977.
J. M. Prausnitz
., Compuer Amsterdam,
1980. Elsevier,
Prenice-Hall, Englewood I Clitfs, NJ,EquilibriaUsingUNIEAC,
'A. Fredenslund et tal., Vapor-Liquid.
(1932).
'M.R. Ind Eng. Chem.. 24: 482
Fenske. Prog., 44: 603
(1948).
'AJ.V. Underwood, Chem. Eng.
 whe is the anbe tintividual mpnents in the
omNN in tthe stillate, Gilliland related the feed and on, he
number of
as afuNiNN the mmtr of quilibium stages and the mol
p that was ranstini hy btuljee into the relation
eulbm
minimum retlux sItraagesclon N
wher is the oqeating ettux selevted by the designer.
Nutins o any mnkey ompnents in the feed mav be Distillate nd
mumber ot stageshas en caleulated from evaluated afterbotheoms dhstn.
(a, )av minimum
where x, is the mol fration of component in the bottoms and (159
(a,
geoneti relative volatility of component i relative to the heavy )aN iS he
averag by
Eq. (lS-le). The Kirkbride method is used to determine the ratio ofkey as gven
below the feed point. trays above and
log
(|5-61
where B and D are the mol tlow rates of the bottoms and distillate, respctively m
Np and Ng are the number of equilibrium stages above and below the tel
respectively.
Selection of Column Internals At this point the designer nmust make asclectim
hased on performance and cost whether atray, random packing. or structural packing
is bestfor the separationprocess being considered. Typically, trays are favored when
the operating pressure and liquid tlow rate are high and when the column dianmeter is
large. Random packings are more often recommended when the column diameter i
small, comOsion and foaming are present, or batch columns are to be used. Siructunl
packings, on the other hand, are considered for low-pressure and vacuum operaton.
Additionally,. they are often selected when low pressure drop across the column s
required or low liquid holdup is desired. trays.
Sieve, valve, and bubble-cap trays are examples of traditional crosstlowpertor-
The development of newer trays over the past decade to provide improvedincludingthe
mance has been significant. Deseriptions of some of these newer trays, ECMB
Irays,suchas
Nye, Max-Frac, Msome of the improved multiple-downcomer presentedby
and EEMD, as well as the Ultra-Frac,TM pP-K.M and Trutna trayS, ar
Humphrey and Keller.! distillationcolumain
Figure 15-2 provides a cross-sectional view of atraditional Ofthese tray(yfes
tundamentak
operation showing an example of asieve, vvalve, and bubble-cap tray.
tray
the sieve tray is the choice in many distillation separations since its
'ER. Gillhland, Ind Eng. Chem., 32: 1220 (1940).
HE. Eduljee, Hdnxarbn Prox. S4: 120 (1975). 1997
C G Kirkbrnide. Petrol. Refiner, 23: 32 (1944).
 Separation by Distillation 773

Gas out

Shell

Liquidin
Flooding Overflow weir
check line
Downcomer
Traysupportingring
Traystiffeningbar
Tray Sample or
sidestream
withdrawal
Liquid

Fecd in

Bubble-capcontactor

Valve-tray contactor
(Open valves)
(Example of closed
Foam, froth, or
valve T liquid droplets
Sieve-tray contacto

Vapor risers
Gas in

Liquid out

contactor column in operation

Flgure 15-2 finite-stage bubble-cap tray
Cross-sectional view of asieve tray, a valve tray, anda
of a relativeto
Showing an example trays are lowincost effi-
low risk. In addition,the in fiow rates. Overalltraythese
entailing
te well established. variations Disadvantages of
handling wide percent.
while
Many other ray typesgenerally between 60and85 relativetosome ofthe newer
are
ciencies of sieve trays drops and lower capacities
trays are their higher pressure