You are on page 1of 6

Mean Temperature Differences in Multipass

Heat Exchangers
W. M. NAGLE,Massachusetts Institute of Technology, Cambridge, Mass.

I
K CALCULATING heat transfer coefficients from ex- hy Nusselt (S), hut the problem of a tiieaii temperature
perimental data, the logarithmic mean temperature difference for multipass heat exchangers has received almost
difference, no attention in the literature. It is widely realized that the
logarithmic mean temperature difference applies only when
the assumptions made in its derivation are fulfilled; never-
theless, the logarithmic mean is generally used in heat trans-
fer calculations involving multipass exchangers. I n many
is generally employed. The logarithmic mean tempersture multipass heat exchangers tlie flow is neither parallel nor
difference is usually derived for parallel flow or counterflow, countercurrent. I n spite of the importance of multipass heat
assuming steady state conditions, a constant over-all coeffi- exchangen, no derivation of a suitable mean temperature
cient of heat transfer, constant specific heats, and negligible difference is known to the writer. I n 1931 Davis (1) pub-
heat losses. The derivation is available in numerous books lislied curves of correction factors to use with the logarithmic
and papers on heat transfer. Most laboratory data on heat mean temperature difference in heat exchanger calculations.
transfer have been obtained in single-tube heat exchangers Eowevcr, tlie method by which these curves were obtained
in which the flow is either parallel or countercurrent. In was not given. I n this paper, equations are derived for the
the case of such laboratory test mean temperature d i f f e r e n c e
data the use of the logarithmic The mean lerriperalure difference in multipass i n multipass h e a t exchangers.
mean temperature difference for Figures 2 and 5, prepared from
c a l c u l a t i n g a cocficicnt of heat exchangers may be calculated by means of these e q u a t i o n s , give the cor-
heat transfer seeiiis reasonable. charts presented in this paper. These charts rection factors by which the log
However, w h e n coefficients gioe correction factors by which the logarithmic mean t e m p e r a t u r e daerence
of heat t r a n s f e r o b t a i n e d in mean temperature difference for counlrrflow for counterflow should be multi-
such single-tube excliangers are may be multiplied to give the viean temperature plied to g i v e t h e m e a n tem-
used in tile des+ of multipass perature difference in multipass
exchangers, the irnres t r i c t e d differencefor multipass exchangers. These cor- exchangers.
use in design of blie losarithisiic rection factors depend only on the number of
mean t e m per a t u r e difference passes in the exchanger and on certain di- ASSWXiPrIoNs
may iiitroduce serious e r r o r . mensionless ratios irivoloirig the inlet and outlet In most multipass b e a t ex-
It seems more logical to derive temperatures .f the two fluids. The derivation changers there is simultaneous
R mean temperature difference lieat t r a n s f e r from the fluid in
for multipass h a t e xc 11a n ge rs of equations .for mran temperature difference are one shell-sidr pass to the fluid
based on simplifying a s s u m p - given in detail for the exchanger haaing one pass in two or more tube-side passes,
tions similar to those used for shellhide and two passes tube-side, and for the each pass being a t a different
the case of parallel and counter- exchanger haeing two passes shell-side and four temperature. I n the a b s e n c e
flow in the d e r i v a t i o n of tlrc pusses tube-side. The assumptions involved of experimental d a t a o n t h e
l o g a r i t h m i c mean. A mean temperature relations in multi-
temperature difEerence suitable are those made in the derivation of the logarithmic pass exchangers, two alternat-
for problerns of heat t r a n s f e r meun temperature difference plus the additional ing assumptions seem reason-
in crossflow Ins been derived assumption of good mixing of the shell-side fluid. able: (1) It may hc assumed
604
June, 1933 INDUSTRIAL AND ENGINEERING CHEMISTRY 603

that the fluid outside the tubes mixes perfectly, resulting in


a uniform temperature of the shell-side fluid over an entire dT
t1 +~ I +
I 2111 + tIv + t~ + VI (3b)
cross section of the exchanger; or (2) it may be assumed that T -
no mixing occurs, and that the h i d outside the tubes of the 6
first pass on the tube side is not affected by the temperature The problem consists in evaluating UAIWC from Equation
3 and substituting that expression for UA/WC in Equation 1.
In order to evaluate the integral in Equation 3, two rela-
tions between T I t', and t" must be obtained. One of these
relations is given by a heat balance around the right end of
the exchanger (Figure 1):
WC(T1 - T) = zoc(t1' - t') (4)
Another relation may be obtained from the dzerential heat
transfer equations for each pass of the fluid inside the tubes:

wcdt' = u dA
- (T - t')
2 (5)
dA
wCdt'I = - u7 (T - t") (6)

FIGURE1. DIAGRAM OF HEAT Ex- Dividing Equation 6 by Equation 5 gives:


CHANGER HAVINGONE PASSON SHELL
SIDE,Two PASSESON TUBESIDE
(7)
of the fluid outside the tubes of the second pass on the tube
side. Because of the baffles in the heat exchanger, the Expressions for t" and dt" may be obtained from Equation 4
assumption of complete mixing is probably the more accurate and substituted in Equation 7, giving:
and will be used in the derivations of this paper.
The assumptions used in these derivations are:
dt' - wc
- dT T
wc ( T I - T )
- t' - -
wc -
- - wc
(8)
dt1 T - t'
Uniform temperature of the shell-side fluid over any cross
section Rearranging and setting wc/WC = R:
Steady flow conditions
Uniform over-all coefficient of heat transfer _
dT - (1 + 8R)T - 2Rt' - 2'1 (9)
Constant specific heat dt' - T - t'
Negligible heat losses
No change of state
Equal heating surface in each pass

DERIVATIONSOF EQUATION FOR MEAN TEMPERATURE


DIFFERENCE IN EXCHANGERS
HAVING ONE PASSON SHELL
SIDE, Two PASSESON TUBESIDE. In Figure 1 a one-two
pass heat exchanger and the temperature relations involved
are shown diagrammatically. The heat transfer equation
over the entire area is:

WC(T1- Tz)= U A Atav.; or Atav. Ti - Tz


= __-
UAIWC ('I
The heat transfer equation over a differential length of the
exchanger is:
wo 01 02 03 a4 05 a6 a7 OB 09
WC dT = U
dA
y (T - t')+2
UdA
( T - t") (2)
= UdA (T - T
t' +) t11 FIGURE
2. CORRECTION
FACTORPLOTFOR HEATEXCHANGERS
OXE PASSON SHELLSIDE
HAVING

On integration, Equation 2 givw: Equation 9 may be integrated after a proper change of

g = LgT1 T--
t'
dT + t"
2
(3)
variables. The integrated expression may be written as:
(TI - t')(r2- (2
z - (1 + R ) -
[x - (1 + R) + d
+ 2R)z +
4 - 2

m ]
2 ~ ) 0 . 5

RD-2
= a constant (10)

Similarly, for a one-four pass heat exchanger, where


z = ( T I - T ) / ( T i - t')

dT The constant in Equation 10 is determined by the boundary


T -
t' + t" + t"' + t'V
(3a) conditions:
4 where
T I - T = Ti - Tz
and for a one-six pass exchanger, Ti - t' = T 1 - ti
606 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 23, No. 6
Solving Equation 4 for t" gives: At&\. =
Let (15)
t" t' + (l/R)(Ti - T) Atlog mean

Hence F is the correction factor by which the logarithmic


On substituting this value for t" in Equation 3 , there results: mean temperature difference must be multiplied to give the
mean temperature difference defined by Equations_,lO and
11. After dividing Equation 14 by Equation 13, we see that
1,

f0
t tw @ PASS TUBES
t' t'U Or*PASS TUBES

UAIWC can now be calculated from Equations 10 and 11


when T I , T z ,ti, and tz are known. For any value of TI - T ,
T I - t' can be calculated by trial and error from Equation 10.
The value of the integral,
fT1 - Tz
I I
FIGURE4. DIAGRAM OF HEAT EX-
CHANGER HAVING
Two PASSES
ON SHELL
SIDE,FOURPASSESON TUBESIDE
can then be determined graphically, and Atav. calculated
from Equation 1. The logarithmic mean temperature the correction factor, F , depends only on ratios of tempera-
difference for countercurrent flow is: ture differences, not on their absolute values:

which may be rearranged t o

The two ratios, R and (tz - t l ) / ( T , - tl), fix the value of


all ratios of temperature differences. Therefore, the cor-
I n a similar manner, the right-hand side of Equation 11 may rection factor, F, has been calculated for various values of R
be rearranged and substituted for UAIWC in Equation 1, and (tz - tl)/(T1- tJ and has been plotted as Figure 2.
giving: The coordinates of Figure 2 are given in Table I.
TABLEI. COORDINATES
FOR FIGURE
2
(One pass shell-side, t w o or more passes tube-side)
R p - F s 1-
Ti - 11 T I - tl TI - t l
4 0,975 0.125 11/2 0.975 0.227 */g 0.975 0.379
0.95 0.155 0.95 0.285 0.95 0.475
0.90 0.181 0.90 0.348 0.90 0.576
0.85 0.195 0.85 0.384 0.85 0.637
0.80 0.203 0.80 0.407 0.80 0.672
0.70 0.211 0.70 0.436 0.70 0.716
0.60 0.215 0.60 0.450 0.60 0.740
0.50 0.217 0.50 0.458 0.50 0.754
0 0,2192 0 0.465 0 0.764
3 0.975 0.146 1 0.975 0.275 0.975
0.95 0.185 0.95 0.350 0.95
0.90 0.224 0.90 0.428 0.90
0.85 0.245 0.85 0.470 0.85
0.80 0.257 0.80 0.501 0.80
0.70 0.270 0.70 0.539 0.70
0.60 0.274 0.60 0.561 0.60
0.50 0.276 0.50 0.575 0.50
0 0.2793 0 0.5858 0
2 0.975 0.182 8/4 0.975 0.328 1/1 0.975 0.513
0.95 0.234 0.95 0.411 0.95 0.615
-CURVE T A W FROM FIG 2 0.90 0.288 0.90 0.504 0.90 0.721
0 ONE-FOUR EXCHANGER 0.85 0.318 0.85 0.559 0.85 0.775
A ONE-SIX EXCHANGER 0.80 0.336 0.80 0.592 0.80 0.807
05 0.70 0.357 0.70 0.630 0.70 0.843
0.60 0.369 0.60 0.649 0.60 0.861
0.50 0.376 0.50 0.659 0.50 0.869
0 0.382 0 0.6667 0 0.8773
"0 OJ aZ 0.3 W+--A 0.5 06 0.7 0s 69
w Similar equations have been derived by the author (9)
FIGURE3. COMPARISON
OF CORRECTIONFACTORS
FOR ONE- for the mean temperature difference in exchangers with one
Two, ONE-FOUR,AND ONE-SIXHEATEXCHANGERS pass on the shell side, and four and six passes on the tube
June. 1933 INDUSTRIAL AND ENGINEERING CHEMISTRY 607

side. As is shown in Figure 3, the correction factors cal- Adding Equations 17 and 18,
culated from equations derived for one-four and one-six
exchangers fall close to the curves for one-two exchangers.
VC(T1 - Tz) = U A Atlav. + Atzsv. = U A At,,. (19)
Therefore, Figure 2 may be used in calculations involving Dividing Equation 17 by Equation 18,
TI - Tr - ALlav.
(20)
Tr - TZ Aha".
WC(T1 - T,) = wc(t2 - t t ) ; T1 - T, = R(t2 - t t ) (21)
WC(T, - T2) = WC(tL - tl); T , - T2 = R(tL - tl) (22)
Dividing Equation 21 by Equation 22,

From the above equations, At,,. for the entire exchanger


can be calculated by a trial-and-error procedure. However,
in computing values for Figure 5 and Table 11, a more rapid
method was used, involving no trial-and-error work, because
it was observed that the above equations were satisfied only
when:

TABLE FOR FIGURE


11. COORDINATES 5
(Two passes shell-side, four or more passes tube-side)
e 1 f l - e 1
TI - tl TI - tl Ti - ti
0.975 0.1836 1111 0.975 0.3525 0.975 0.591
0.95 0.2102 0.95 0.418 0.95 0.689
0.90 0.231 0.90 0.4808 0.90 0.784
0.80 0.2395 0.80 0.5325 0.80 0.861
0.50 0.2402 0.50 0.571 0.50 0.915
0 0.2407 0 0.5766 0 0.923
0.975 0.221 1 0.975 0.4312 0.975 0.678
0.95 0.2606 0.95 0.5185 0.95 0.771
0.90 0.2914 0.90 0.6995 0.90 0.859
0.80 0.3117 0.80 0.6675 0.80 0.918
0.50 0.3209 0.50 0.730 0.50 0.954
0 0.322 0 0.7385 0 0.956
0.975 0.2833 v 4 0.975 0.509 0.975 0.746
0.95 0.341 0.95 0.603 0.95 0.8365
0.90 0.3925 0.90 0.697 0.90 0.911
0.80 0.4303 0.80 0.774 0.80 0.955
0.50 0.457 0.50 0.828 0.50 0.979
0 0.4605 0 0.834 0 0.982

Courtesy of M . W . Kellogg Co. Since the correction factor is a function only of R and
CROSS SECTIONOF HORIZONTAL
HEAT EXCHANGER HAVINGONE ( t t - tJ/(Tr - t 1 ) or (12 - tt)/(TL - t t ) , Equation 24
P A S S THROUGH SHELL, FOUR requires that F1, the correction factor necessary for com-
PASSESTHROUGH TUBES
one-four and one-six exchangers as well as one-two exchangers.
1 Moreover, it was found, for the case of a one-two exchanger,
' that the correction factor curves are essentially the same
, whether the shell-side fluid enters near the fixed head or near
the floating head.
I Two PASSES o x SIXELL SIDE AND FOUR PASSES ON TUBE
SIDE. For the case of two passes on the shell side and four
passes on the tube side, the mean temperature difference is
readily derived from the curves (Figure 2) for exchangers
having one pass on the shell side and two passes on the tube
side, provided the heat flow through the longitudinal baffle
is negligible. The same temperature difference applies to two
identical exchangers, connected in series, each having one
pass on the shell side and two passes on the tube side. For
two such exchangers in series the temperatures between the
exchangers need not be known.
In Figure 4 the direction of the fluid streams and the FIGURE5. CORRECTIONFACTOR PLOTFOR HEAT Ex-
temperature relations in a two-four exchanger are shown CHANGERS HAVINGTwo PASSES ON SHELLSIDE
diagrammatically.
The heat transferred in the first pass of the shell-side fluid puting Ah,,. from the logarithmic mean, be equal to Fz,
(the third and fourth passes of the tube-side fluid), Q1,is the correction factor necessary for computing A h . The
equal to: total amount of heat transferred is equal to the sum of the
heat transferred in the two shell-side passes-that is,
From Equations 24 and 25, and Solut.im. (1) For a e o u n t e r -
the equation P, = Fa, it can he flow excliarrger, the logarithmic
&own algebraically that 1x1c a n t e m p e r a t u r e difference
applies. Therefore:
p = p , = I?, (26)
whero F = correction iactor 1,s.
which logarithmic mean temp. dif- IL, = IiiiaiiiYlil =

ference far entire exchanger is multi- ( T , - In)- (Tr -~1 3 ) =


plied to give Ala". Ti - I1
2.303 IaL: ,?~.
I i - 1,
Since F = F,, it was necessary 187 -
3& = 92.1' 12.
only toread tlievaluesof (13 - I&)/ '2.303 log
(T, - lr) on Figure 2, corre- 37
sponding to any elinsen F , and R,
and substitute ill tile equation: ( 2 ) For the one-two exchanger,
the correction factor is found in
Figure 2:

t? = (7'1 -T ~ / (-I t ~l ) =
20150 = 4
(12 - & ) / ( T I- t ~ ) =
50/237 = 0.211
Hence,
F = 0.70
AI,. = (0.70)(92.1) = 64.5' 17,
The values of F and the corre-
s p o n d i n g v a l u e s of ( I 2 - tJ/-
-
(T, & ) w e r et h e n p l o t t e d as (3) F o r t h e t w o - f o u r e x -
changer, Figure 5 is used, giving
Figure 5, the c o r r e c t i o n factor
plot f o r exchangers having P = 0.95 when I?. = 4 a n d
two passes on the shell side and (h - I,)/(Z'l - I ,) = 0.211.
f o u r passes on t h e t u b e side. Eence:
By a s i m i l a r p r o c e d u r e , cor- . ...', ., . .. -.. ...
r e c t i o n f a c t o r c u r v e s o r ex- At,,. = (0.95)(92.1) = 87.5" F.
HEAT EXCHANGEUS ON LARGEVACUUM UNIT
changers having two passes on Exohangers are single-p888 through shell and four-
the shell side and eight or twelve i m e through tubes. The mean temperature differ-
.
"asses on the tube side could be ences for these three tvoes of
constructed from curYes for the one-four or one-six exchangers. heat exchangers, operating with the same inlet and outlet
I.

But the curves for the one-four and one-six exchangers temperatures, are tabulated below. The mean temperature
nearly coincide with those for the one-two exchanger. Hence difference RS calculated from the Ross cliart ( I ) is also in-
the curves derived therefrom for two-eight and t.rvo-twelve cluded:
exchangers would differ only slightly from Figure 5 .

UBE08 CHARTS
~ -
(TI 5w"F.; TI 300' F.; f t - ----
263' P.;tt
T r ~ mor E a c ~ m o e n M c ~ e o o01 Car.0~.
313' F.)
%v. -
c.e
Tile mean temperature difference in a multipass lieat
exchanger may be calculated by the aid of Figures 2 or 5,
when the terminal ternperahires of both fluids are known.
I chart 82
In these figures, P, the correction factor, is plotted against
the dimensionless ratio (h - t J / ( T , - h ) , for different values
of another dimensionless ratio, R. In this exaiirple the check between the Ross cliart and the
The mean temperature difference for a multipass exchanger curves presented in this paper is good. However, the
is then obtained by multiplying tlie logarithmic mean tem- Ross curves differ in shape from those presented herein.
perature difference for counterflow by tlie correction factor, Some examples may be selected where the agreement is
F: better than in the above example, others where it is worse.
TI - C ) - ( T 2 - la)
Al:,., = P (--______ T, - 1%
2.303 log
ACKNOWLEDGMENT
T* - I ,
Tho writer is deeply grateful to TV. H. McAdams for
EXAWLE.If the inlet and outlet temperatures of the enrwuragement and help during the course of this work and
slieil-side fluid in a heat exchanger are 500" and 300" 1'. to T. B. Drew for his careful review of this article.
(260° and 148.9O C.), respectively, and the inlet and outlet
temperatures of t,lie tube-side fluid are 203' sild 313" F.
NOMENCLATURE
(128.3' and I5ti.l" C.), calculate:
A = area of heating surface
(I) The mean temperature difference if a strictly aounteiflow
exchanger is used.
(2) The mean temperature difforence if a. niultipws exchanger,
having one pass shell-side and t w o passes tube-side, is
used.
(3) The mean temperature difference ii a multipass exchanger, dimensionless
having two pssses shell-side and four pasae stube-side, is T = temp. of shellhide fluid
used. T, = inlet temp., shell-side fluid, same scale as for k
June, 1933 INDUSTRIAL AND ENGINEERING CHEMISTRY 609

TZ = outlet temp., shell-side fluid, same scale as for t~ Superscript 11 11, 11x8 etc., refer to the pass in question
t = temp. of tube-side fluid Subscripts 1 and refer to inlet and outlet temperatures
tl = inlet temp., tube-side fluid, any consistent temp. Subscripts L and, refer to temperatures in the heads between
scale passes
t p = outlet temp., tube-side fluid, same temp. scale
as for 11 LITERATURE
CITED
At = temp. difference
(1) Davis, F. K., Ross Heater and Mfg. Go., Bull. 350, 72 (1931).
St,", = mean temp. difference (2) Nagle, S. M., Thesis, Mass. Inst. Tech., 1932.
r J = over-all coefficient of heat transfer (3) Nusselt, Tech. Mech. Thermodynamik, 1, 417 (1930).
TY = weight rate of flow of shell-side fluid
w = weight rate of flow of tube-side fluid RECEIVED
November 4, 1932.

Anode Process for Rubber Articles


and Coatings
C. L. BEAL,American Anode, Inc., Akron, Ohio

T HE term "anode process" has been chosen and used electrodeposition process; and (2) electrical-chemical depo-
widely in the trade to designate a fundamental method sition, the anode ionic deposition process. They are dis-
for the production, directly from rubber latex, rapidly, cussed in four basic U. S. patents.'
and incone application, of articles and coatings of the highest
grade of unmasticated rubber. ANODEELECTRODEPOSITION ON METALS
Rubber latex, a milky exudation from the bark of rubber A latex mix adjusted for anode electrodeposition may, for
trees, is composed chiefly of tiny particles of rubber suspended example, contain 35 per cent by weight of rubber and com-
in a water phase or serum, not unlike globules of butter f a t pounding ingredients; about 20 grams per liter of ammonia;
in milk. Rubber latex contains ,small amounts of many and about 30 grams per liter of ammonium, potassium, and
organic compounds and inorganic salts. Some of these non- sodium salts, the majority of which is ammonium chloride.
rubber materials, such as the proteins and resins, are con- The latex mixture is placed in the anode compartment of
sidered to be adsorbed on the surfaces of the rubber particles a diaphragm cell (Figure 1). Slightly alkaline water is
and to be responsible for many of the colloidal characteristics placed in the cathode compartment. An anode of zinc or
of latex. As it comes from the tree, the latex is unstable and galvanized iron is inserted into the latex, and a cathode is
coagulates easily, but, when stabilized with ammonia, it can placed in the cathode compartment. When a unidirectional
be safely shipped and stored for long periods. electromotive force is impressed across the electrodes, a
Like most colloidal particles in suspension in an alkaline number of phenomena take place.
medium, the rubber particles of ammoniated latex are nega- Electrolysis of the dilute ammonia and ammonium chloride
tively charged through the adsorption of hydroxyl anions. solutions produces a t the cathode bubbles of hydrogen, and
The particles, many as small as 1 / 2 5 , ~ inch in diameter, are in the cathode compartment, lower hydrogen-ion concentra-
in constant oscillation (Brownian movement) and are kept tion. It also produces a t the anode electrochemical solution
- from hitting one another and sticking of zinc, and in the immediate neighbor-
together (coagulating) by the repulsion hood of the anode, an increased concen-
of their e l e c t r i c charges. When the l f l R
tration of hydrogen ions and some zinc
hydroxyl ions are neutralized or other- ions. The hydrogen and zinc ions which
wise removed from the particles, the are moving away from the anode react
electric repulsion b e t w e e n p a r t i c l e s with the hydroxyl ions adsorbed on
disappears and coagulation results. the nearest latex particles, thus neutral-
Since the advent, about twelve years izing the electric charges of those parti-
ago, of stable ammoniated latex as an cles. The discharged particles, through
article of commerce, intensive research their rapid Brownian movement, are
and d e v e l o p m e n t , a p p l i e d to the immediately brought into intimate con-
problem of producing a method suit- tact with one another and stick together,
able for the commercial production of thus forming a compact deposit which
rubber articles and coatings, has cul- A-Annde with depoxf of latex rubber adheres to the anode.
m i n a t e d i n t h e development of t h e 6 -Cafmdes The negatively charged particles of
B,-Cathode diaphragm
a n o d e d e p o s i t i o n process. T h i s &-Cathode compartment r u b b e r and compounding ingredients
consists in the d e p o s i t i o n of rubber C-Lotex m i x exhibit electrophoretic m o v e m e n t in
D- DepoJited rwbber
globules and other particles suspended the d i r e c t i o n of t h e anode, bringing
in an aaueous medium (com-oounded FIGURE
~I
1. CELLFOR ANODEELECTRO-about a higher concentration of eolids
latex, disiersed rubber, or reclaim) upon DEPOSITION ON METAL in the immediate neighborhood of that
articles and forms having the capacity electrode.
of liberating a t their surfaces ions which neutralize the anionic During continued deposition the electromotive force im-
charges of the particles. pressed across the wet coagulated deposit on the anode re-
The two principal embodiments of the anode process have moves water from it by electroendosmotic action, and com-
been developed by the anode process pioneers a t the Hun- pacts - the deposit until it contains about 40 Der cent water.
garian Rubber Goods Factory- a t Budapest, Hungary, and
1 Sheppard and Eberlin, 1,476,374 (Dec. 4, 1923); Klein, 1,548,689 (Aug.
the Company Of Rochester, y' They may 4, 1925) ; Sheppard and Beal, 1,589,325 (June 15, 1926) ; Klein and Szegvari,
be classified as: (1) electrochemical deposition, the anode 1,825,736 (act. 6, 1931).

You might also like