PRINDIP LES OF HEAT FLOST 11 PLUS DS

319

at both the reactor inlet and outlet are relatively small. Clearly the average
temperature drop is much greater than the drop at either end of the reactor
and cannot be found from the logarithmic mean of the terminal A T's.

VARIABLE OYERA ' orrFicIENT. When the overall coefficient varies regu-
larly, the rate of hea ra sfer may be predicted from Eq. (11.16), which is based
on the assumption that f7 varies linearly with the temperature drop over the
entire heating surface’ :

where U „ U, —— local overall coefficients at ends ot exchanger

k TI, *2 - temperature approaches at corresponding ends of exchanger
Equation (11.16) calls for use of a logarithmic mean value of the t/ fiP cross
product, where the overall coefficient at one end of the exchanger is multiplied
by the temperature approach at the other. The derivation of this equation
requires that assumptions 2 to 4 above be accepted,
In the completely general case, where none of the assumptions is valid and
U varies markedly from point to point, Eq (11.9) can be integrated by evaluating
local values of U, ö T, and q at several intermediate points in the exchanger.
Graphical or numerical evaluation of the area under a plot of I fU ö T vs.
q, between the limits of zero and g„ will then give the area Az of the heat-
transfer surface required.

MULTIPASS ESCH GERS. In multipass shell-and-tube exchangers the flow

pattern is complex i h arallel, countercurrent, and crossflow all present. Under
these conditions, even when the overall coefficient U is constant, the LMTD
cannot be used. Calculation procedures for multipass exchangers are given in
Chap. 15.

Individual Heat-Transfer Coefficients

The overall coefficient depends upon so many variables that it is necessary to
break it into its parts. The reason for this becomes apparent if a typical case is
examined. Consider the local overall coeficient at a specific point in the double-
pipe exchanger shown in Fig. 11.3. For definiteness, assume that the warm fluid
is flowing through the inside pipe and that the cold fluid is flowing through the
annular space. Assume also that the Reynolds numbers of the two fluids are
sufficiently large to ensure turbulent flow and that both surfaces of the inside tube
are clear of dirt or scale. If, now, a plot is prepared, as shown in Fig. 11.8, with
temperature as the ordinate and distance perpendicular to the wall as the
abscissa, several. important facts become evident. Tn the figure, the metal wall
of the tube separates the warm fluid on the right from the cold fluid on the left.
The change in temperature with distance is shown by the line T.TzT„,T„,T T .
The tempera- ture profile is thus divided into three separate parts, one through
each of the two
320 ii i TRANSF eR AND ITS APPLICATI OSS

Distance
FIGURE 13.8
Temperature gradients in forœd convection.

fluids and the other through the metal wall. The overall eifect, therefore, should
be studied in terms of these individual parts.
It was shown in Chap. 5 that in turbulent flow through conduits three zones
exist, even in a single fluid, so that the study of one fluid is, itself, complicated. In
each fluid shown in Fig. 11.8 there is a thin sublayer at the wall, a turbulent core
occupying most of the cross section of the stream, and a buffer zone between them.
The velocity gradients were described in Chap. S. The velocity gradient is large
near the wall, small in the turbulent core, and in rapid change in the buffer zone.
It has been found that the temperature gradient in a fluid being heated or cooled
when flowing in turbulent flow follows much the same course. The
temperature gradient is large at the wall and through the viscous sublayer, small in
the turbulent core, and in rapid change in the buffer zone. Basically, the reason
for this is that heat must flow through the viscous sublayer by conduction, which
calls for a steep temperature gradient in mos I fluids because of the low
thermal conductivtty, whereas the rapidly moving eddies in the core are
effective in equalizing the temperature in the turbulent zone. In Fig. 11.8 the dashed
lines F F and I ›* represent the boundaries of the viscous sublayers.
The average temperature of the warm stream is somewhat less than
the
maximum temperature T, and is represented by the horizontal line MM, which is
drawn at temperature T . Likewise, line NN, drawn at temperature T„ represents
the average temperature of the cold fluid.
The overall resistance to the flow of heat from the warm fluid to the cold
fiuid is a result of three separate resistances operating in series. Two
resistances
are those offered by the individual fluids, and the third is that of the solid wall.
In general, also, as shown in Fig. 11.8, the wall resistance is small in comparison
with that of the fluids. The overaI1 coefficient is best studied by analyzing it in
tetms of the separate resistances and treating each separately. The separate
resistances can then be combined to form the overall coefficient. This approach
requires the use of individual heat-transfer coefficients for the two fluid streams.
The individual, or surface, heat-transfer
dqfdA coefficient h is defined generally by
ii =° (11.17
the equation T— )
T„
where dqJdA -— local heat flux, based on the atea in contact with
fluid 'r = local average temperature of fluid
T„ -— temperature of wall in contact with fluid

The reciprocal of this coefficient, 1/li, is called a thermal resistance. For

conduction through a solid, such as a metal wall of thickness x„, and thermal
conductivity k, the thermal resistance equals x„/k. Appropriately corrected for
changes in area, the individual resistances may be added to give the overall
resistance 1/é/.
A second expression for /i is derived from the assumption that there are no
velocity fluctuations normal to the wall at the surface of the wall itself. The
mechanism of heat transfer at the wall is then by conduction, and the heat flux
is given by Eq. (10.2), noting that the normal distance n may be replaced by y,
the normal distance measured into the fluid from the wall in the direction of the
flow of heat. Thus

(11.18)
322 H EAT TRANSFER A ND ITS A
PPLICANlONS

The subscript iv calls attention to the fact that the gradient must be evaluated
at the wall. Eliminating dqJdA from Eqs. (11.17) and (11.18) gives

(11.19)

Note that /i must always be positive. Equation (11.19) can be put into a
dimensionless form by multiplying by the ratio of an arbitrary length to the
thermal conductivity. The choice of length depends on the situation. For heat
transfer at the inner surface of a tube, the tube diameter D is the usual choice.
Multiplying Eq. (11.19) by D/ k gives

(I
k T— 1.20)
T„
On the cold-fluid side of the tube wall T < T„, and the denominator in Eqs.
(11.19) and (I 1.20) becomes T’„ — T. The dimensionless group liD/l is called
a Nusselt number N N. . That shown in Eq. (11.20) is a local Nusselt number
based on diameter. The physical meaning of the Nusselt number can be
seen by inspection of the right-hand side of Eq. (11.20). The numerator (dTld)
j„ is, of course, the gradient at the wall. The factor (T — T j/D can be
considered the average temperature gradient across the entire pipe, and the
Nusselt number is the ratio of these two gradients.
Another interpretation of the Nusselt number can be obtained by
considering the gradient that would exist if all the resistance to heat transfer were
in a laminar layer of thickness x in which heat transfer was only by
conduction. The heat- transfer rate and coefficient follow from Eqs. (10.1) and
(11.17):
(11.21
dA x )

(11.22
)
From the definition of the Nusselt number,
hD kD D
(t1.23
)
The Nusselt number is the ratio of the tube diameter to the equivalent
thickness of the laminar layer. Sometimes x is called the film thickness, and it is
generally slightly greater than the thickness of the laminar boundary layer because
there is some resistance to heat transfer in the buffer zone.
Equation (11.17), when applied to the two fluids of Fig. 11.8, becomes, for
the inside of the tube (the warm side in Fig. 11.8),

(11.24)
 eeinciPies oF risAi Flow in Fcuins

323

dq/dA
and for the outside of the tube (the cold side) (11.25)
T„ — T
where A; and A are the inside and outside
areas The
of thecold fluid
tube, could, of course, be inside the tubes and the warm fluid
respectively.
outside. Coefficients ñi and h refer to the inside and the outside of the tube,
respectively, and not to a specific fluid.

CALCULATION OF OV CALL COEFFICIENTS FROM INDIVIDUAL

COEFFI-
CIENTS. The overall coefficient is constructed from the individual coefficients and
the resistance of the tube wall in the following mannet.
The rate of heat transfer through the tube wall is given by the differential
form of Eq. (10.13), dq k„( —
T„) (I1.26)

where Tpz — T„ -— temperature dAz differencex„through tube waI1

kp -— thermal conductivity of
wall x„ = tube-\vall thickness
dq(dA z —— local heat flux, based on logarithmic mean of inside and
outside areas of tube
If Eqs. (11.24) to (11.26) are solved for the temperature differences and the
temperature differences added, the result is

1 x„
-— I1.27)
d dA,li,“ dS zk + dA h
'
Assume that the heat-transfer rate is arbitrarily based on the outside area. If Eq.
(11.27) is solved for dq, and if both sides of the resulting equation are divided by
&„ the result is
(11.28)
dA. I dA x dA. 1

Now

where D , D;, and Dz are the outside, inside, and logarithmic mean diameters of
the tube, respectively. Therefore

(11.29)
324 HEAT TRANSFER AND ITS APPLICATIONS

Comparing Eq. (11.9) with Eq. (t1.29) shows that

'
U - (11.30)
° 1 D. x„ D 1

If the inside area A; is chosen as the base area, division of Eq. (11.27) by dA;
gives for the overall coefficient
1
U; =
(11.31)
1 x„ D; l
D,

RESISTANCE F RM OF OVERALL COEFFICIENT. A comparison of Eqs.

(10.9)
1 D of an overall coefficient can be considered
and (11.30) sugge ts that the reciprocal
D (11.32)
to be an overall resistance composed of three resistances in series. The total, or
overall, resistance is given by the equation
The individual terms on the right-hand side of Eq. (11.32) represent the individual
resistances of the two fluids and of the metal wall. The overall temperature drop is
proportional to 1/U, and the temperature drops in the two fluids and the wall are
proportional to the individual resistances, or, for the case of Eq. (11.32),

(11.33)

where fiT —— overall temperature drop

b T —— temperature drop through inside fluid
k j, —— temperature drop through metal wall
k T —— temperature drop through outside fiuid

FOULING FACTORS. In actual service, heat-transfer surfaces do not remain clean.

Scale, dirt, and other solid deposits form on one or both sides of the tubes, provide
additional resistances to heat flow, and reduce the overall coeficient. The effect
of such deposits is taken into account by adding a term ! JdA liz to the term in
parentheses in Eq. (11.27) for each scale deposit. Thus, assuming that scale is
deposited on both the inside and the outside surface of the tubes, Eq. (11.27)
becomes, after correction for the effects of scale,

k T —— dq
dA,li „ dA,li; dAzk„
(11.3^)
dA li. dA,hdo

where hz¡ and are the /ou/inq /ac/ore for the scale deposits on the inside
and outside tube surfaces, respectively. The following equations for the overall
coeffi-
 PRINCI PLES OF HOT FLOW IN FXU US

325

cients based on outside and inside areas, respectively,

1 follow from Eq. (11.34):
(11.35)

and

The actual thicknesses of the deposits are neglected in Eqs. (11.35) and (11.36).
Numerical values of fouling factors are given in Ref. 3 corresponding to
satisfactory performance in normal operation, with reasonable service time be-
tween cleanings. They cover a range of approximately 600 to 11,000 W/m°-°C
(100 to 2000 Btu/ft’-h-°F). Fouling factors for ordinary industrial liquids fall
in the range 1700 to 6000 W/m°-°C (300 to 1000 Btu/ft’-h-°F). Fouling
factors are usually set at values that also provide a safety factor for design.

Example 11.1. Methyl alcohol fio wing in the inner pipe of a double-pipe exchanger
is cooled with water flowing in the jacket. The inner pipe is made from l-in. (25-mm)
Schedule 40 steel pipe. The thermal conductivity of steel is 26 Btu/ft-h-°F (45 W/m-
°C). The individual coefficients and fouling factors are given in Table 11.1. What is
the overall coefficient, based on the outside area of the inner pipe?

Solution
The diameters and wall thickness of l-in. Schedule 40 pipe, from Appendix 5, are
1.049 1.315 0.133
D; - — 0.0874 ft D, —— = 0.1096 ft x = = 0.0111 ft
12 2 1

The logarithmic mean diameter Dz is calculated as in Eq. (10.15) using

diameter in place of radius:
D, — D;
D =0.0983
0.1096 —
t
0.0871 “ In (D fD;j
In
TADLEI1.1 (0.1096/0.0874)
Data for Example 11.1

Btu/ft’-h-°F W/m*-
°C
Alcohol coefficient h; 180 1020
Water coefficient £, 300 1700
Inside fouling factor Ju; 1000 5680
Outside fouling factor li,. 500 2840