DB 3 CH 7

Engineering Data Book III
Chapter 7
Condensation on External Surfaces
SUMMARY: The principles of film condensation on external surfaces are presented. Geometries that are
addressed are: vertical plates, horizontal tubes, horizontal tube arrays, tube bundles and enhanced
surfaces. The effects of vapor shear, interfacial waves, interfacial vapor shear, surface tension and noncondensable gases are addressed. The flow modes of laminar and turbulent films and intertube flow
modes between tubes are discussed. Numerous methods are presented for prediction of local condensation
heat transfer coefficients for plain surfaces and low finned tubes.
7.1 Modes of Condensation

Condensation is the heat transfer process by which a saturated vapor is converted into a liquid by means
of removing the latent heat of condensation. From a thermodynamic standpoint, condensation occurs
when the enthalpy of the vapor is reduced to the state of saturated liquid. In practice, the process is
dynamic and heat must be transferred in order to achieve condensation. Thus, condensation occurs when a
vapor contacts a solid surface or a fluid interface whose temperature is below the saturation temperature
of the vapor.
This chapter will deal with condensation on the outside of surfaces, such as on plates, horizontal or
vertical tubes and tube bundles. The following chapter provides a treatment of intube condensation, in
which the process takes place inside enclosed channels typically with forced flow conditions.
Four basic mechanisms of condensation are generally recognized: drop-wise, film-wise, direct contact and
homogeneous. The first three are categorized as heterogeneous processes. Figure 7.1 illustrates these
condensation modes. In drop-wise condensation, the drops of liquid form from the vapor at particular
nucleation sites on a solid surface, and the drops remain separate during growth until carried away by
gravity or vapor shear. In film-wise condensation, the drops initially formed quickly coalesce to produce a
continuous film of liquid on the surface through which heat must be transferred to condense more liquid.
In direct contact condensation, the vapor condenses directly on the (liquid) coolant surface, which is
sprayed into the vapor space. In homogeneous condensation, the liquid phase forms directly from supersaturated vapor, away from any macroscopic surface; it is however generally assumed, in practice, that
there are particles of dust or mist particles present in the vapor to serve as nucleation sites.
While drop-wise condensation is alluring because of the very high heat transfer coefficients that result, it
is not considered to be suitable for deliberate employment in process equipment. Generally, special
materials must be employed (of low thermal conductivity, low surface energy, low wetting or be highly
polished) to attain drop-wise condensation. Hence, the process is susceptible to any fouling or oxidation
of the surface that may bring the process back into the film-wise mode, with a corresponding reduction in
thermal performance. Drop-wise condensation is known to occur unexpectedly in experimental test
facilities since the surfaces may be new, polished and very clean, which gives much higher coefficients
than for the film-wise conditions under study (some researchers hence deliberately oxidize their test
surfaces to avoid the drop-wise mode). For more information on drop-wise condensation, refer for
instance to Griffith (1983).
7-1
Figure 7.1. Various modes of condensation. (a) Filmwise

condensation. (b) Homogenous condensation - fog formation.
(c) Dropwise condensation. (d) Direct Contact Condensation
Direct contact condensation is a very efficient process, but it results in mixing the condensate with the
coolant. Therefore, it is useful only in those cases where the condensate is easily separated, or where there
is not desire to reuse the condensate, or where the coolant and condensate are in fact the same substance.
Condensate forming as suspended droplets or mist in a subcooled vapor is called homogeneous
condensation, of which the most common example in nature is fog. Homogeneous condensation is
primarily of concern in fog formation in equipment, i.e. to be avoided, and is not a design mode.
Film-wise condensation is the only of the above processes of particular industrial interest. Thus, all the
subsequent treatment of condensation will implicitly mean film-wise condensation, in which the heat
transfer surface is covered by a thin film of condensate flowing under the influence of gravity, vapor
shear and/or surface tension forces. A review of recent literature is also available in Thome (1998).
7.2 Laminar Film Condensation on a Vertical Plate

7.2.1 Integral Analysis of Nusselt
Laminar film condensation of a pure single-component saturated vapor was among the first heat transfer
problems to be successfully analyzed from a fundamental point of view. The definitive work is by Nusselt
(1916) in two papers published two weeks apart in 1916 that has been widely described in numerous
books since. The original analysis applies specifically to laminar flow of a condensing film on a vertical
7-2
surface. However, it is possible to generalize this approach to a number of other geometric cases and for
this reason it is worthwhile to examine his analysis in some detail here. The Nusselt falling film analysis
closely represents experimental results on vertical plates if no ripples or non-condensable gases are
present and the film flow is laminar.
Figure 7.2. Film condensation on a vertical plate.

First of all, Figure 7.2 depicts the process of laminar film condensation on a vertical plate from a
quiescent vapor. The film of condensate begins at the top and flows downward under the force of gravity,
adding additional new condensate as it flows. The flow is laminar and the thermal profile in the liquid
film is assumed to be fully developed from the leading edge. Thus, the temperature profile across the film
is linear and heat transfer is by one-dimensional heat conduction across the film to the wall. Other
assumptions in the Nusselt analysis are as follows:
The vapor temperature is uniform and is at its saturation temperature;

Gravity is the only external force acting on the film (momentum is neglected so there is a static force
balance);
The adjoining vapor is stagnant and does not exert drag on the film;
Fluid properties are constant;
The sensible cooling of the film is negligible with respect to the latent heat;
The curvature of the interface is negligible so the saturation temperature of the interface is that of a
planar interface determinable from the vapor pressure curve of the fluid.
7-3
Figure 7.3. Integral representation of falling laminar film on a vertical plate.

The integral analysis of the process on a vertical plate is represented in Figure 7.3. At a distance z from
the top, the thickness of the film is . Ignoring inertia effects, i.e. no acceleration of the flow, a force
balance on the liquid element gives
duy
dz
dy
( y)dz(L G )g = L
[7.2.1]
In this expression gravity acts as a body force on the element of volume (-y)(dz)(1) where a unit width
of the plate is assumed. The viscous force is for the shear on the film at distance y from the wall over the
length dz. While this expression is for a vertical plate, it is applicable to an inclined plate as long as the
angle of inclination is sufficient for drainage of the condensate. For an inclined plate, the force of gravity
g on the film in the above expression is replaced with g sin , where is the angle of the plate relative to
horizontal. Now, rearranging and integrating this expression from the initial boundary condition of uy = 0
at y = 0, then the velocity profile at any location y in the film is obtained to be:
7-4
uy =
(L G )g y y 2
L
[7.2.2]
Integrating this velocity profile across the film, the mass flow rate of condensate per unit width of the
plate is
= L u y dy =
0
L (L G )g3
3 L
[7.2.3]
has the dimensions of kg/ms, which represents the flow rate in kg/s per unit width of the plate.
Differentiating this expression with respect to , where = 0 at z = 0, the rate of increase of the film flow
rate with film thickness is
d L ( L G )g 2
=
d
L
[7.2.4]
Taking the film surface temperature as Tsat and the wall temperature as Tw, the heat conducted across the
liquid film element of length dz with a thermal conductivity of kL is
dq =
kL
(Tsat Tw )dz
[7.2.5]
Applying an energy balance, this rate of heat transfer by conduction is equal to the rate of latent heat
removed from the vapor at the interface, which means dq = hLG d. The rate of condensation on this
element (d) is thus
d =
kL
(Tsat Tw )dz
h LG
[7.2.6]
Substituting [7.2.6] into [7.2.4], separating variables and then integrating from = 0 at z = 0 gives
4
L k L (Tsat Tw )z = L ( L G )gh LG
4
[7.2.7]
Rearranging this expression for the local film thickness, it is
4 k z(T Tw )
= L L sat
L ( L G )gh LG
1/ 4
[7.2.8]
The physical significance of the film thickness is that of the conduction length through a solid of the same
thickness, seeing that is laminar film condensation the thermal resistance is only conduction from the
interface to the wall. The film thickness is shown by the above expression to be directly proportional to
the temperature difference (Tsat-Tw), such that a larger temperature difference results in a higher
condensation rate.
7-5
From the thermal conductive resistance across the film, the local condensation heat transfer coefficient
f(z) at any point z from the top of the plate is
k L L (L G )gh LG k 3L
f (z ) =
=
4 L z(Tsat Tw )
1/ 4
[7.2.9]
Rearranging, the local Nusselt number for film condensation is obtained using z for the characteristic
dimension:
(z ) z L ( L G )gh LG z 3
Nu (z ) = f
=
k L 4 L k L (Tsat Tw )
1/ 4
[7.2.10]
Integrating [7.2.9] from z = 0 to z, the mean heat transfer coefficient for the plate up to point z is
( G )gh LG k 3L
1z
f = f (z)dz = 0.943 L L
z0
L z(Tsat Tw )
1/ 4
[7.2.11]
Comparing the above expressions, it is seen that the mean coefficient f on the plate from 0 to z is 4/3
times the value of the local coefficient f(z) at z. The mean heat transfer coefficient can also be obtained
from
f =
(z ) h LG
z(Tsat Tw )
[7.2.12]
where (z) is the condensate flow rate per unit width at a distance z from the top of the plate. Combining
[7.2.12] with [7.2.6] to eliminate (TsatTw), another expression for the thickness of the condensate at point
z from the top is
k L(z ) dz
f zd
[7.2.13]
Eliminating by combining [7.2.13] with [7.2.3] yields the differential expression
( G )g
L L L
3 L
1/ 3
dz f 1 / 3d
=
z
(z )
[7.2.14]
Integrating over z gives the mean heat transfer coefficient as
( G )gk 3L
f = 0.925 L L
L (z )
1/ 3
[7.2.15]
7-6
It is inconvenient to utilize an expression for the condensing heat transfer coefficient in terms of (Tsat-Tw)
as in [7.2.11] since the wall temperature is unknown beforehand in heat exchanger design and results in
an iterative solution procedure. For the present situation the heat transfer coefficient can also be expressed
in terms of the local film Reynolds number, which at a distance z below the top of the plate is defined as
Re =
4(z )
L
[7.2.16]
Substituting [7.2.16] into [7.2.15] and rearranging, the mean heat transfer coefficient up to point z is
f
kL
2L
L ( L G )g
1/ 3
= 1.47 Re 1 / 3
[7.2.17]
where the bracketed term to the left of the equal sign together with its exponent is the characteristic
length. The local condensing heat transfer coefficient in terms of film Reynolds number is
f (z )
2L
k L L ( L G )g
1/ 3
= 1.1 Re 1 / 3
[7.2.18]
The condensing heat transfer coefficient for a laminar film is thus seen to be inversely proportional to the
film Reynolds number to the 1/3 power. This can be compared, for instance, to fully developed laminar
flow inside a tube in which the laminar heat transfer coefficient is independent of the Reynolds number.
Bromley (1952) extended the Nusselt theory to include subcooling of the condensate in the heat balance.
Following in this line, Rohsenow (1956) showed that empirically adding a sensible heat term to the latent
heat of vaporization gave reasonable results, defining an effective latent heat of evaporation as:
c pL Tf
hLG = h LG 1 + 0.68
h LG
[7.2.19]
where Tf = (Tsat-Tw). The subcooling correction is typically negligible with respect to the latent heat
since condensing temperature differences tend to be small.
Regarding other physical properties, Drew (1954) proposed that they be evaluated at an effective film
temperature, which he gave as [Tw+ 0.25(Tsat-Tw)]. For small temperature differences, it is sufficiently
accurate to calculate the properties at the saturation temperature. The above analysis can also be applied
to the outside or inside of a vertical tube, as long as the tube diameter is much larger than and the effect
of vapor shear remains small.
7.2.2 Boundary Layer Analysis

Sparrow and Greg (1959) applied boundary layer analysis to the problem of laminar falling film
condensation. They removed the assumption of negligible momentum changes in the liquid film but, like
Nusselt, they assumed that the shear exerted by the stagnant vapor on the falling film is negligible. Their
solution obtains the local film heat transfer coefficient from the heat conducted from the film into the wall
7-7
f (z ) =
q
kL
=
Tsat Tw Tsat Tw
T

y y=0
[7.2.20]
by evaluating the calculated temperature profile at y=0. Their analysis gives the local Nusselt number as:
g ( G ) z 3 h LG
(z)z
= [ (0 )] Ja1L/ 4 L L
Nu (z ) = f
kL
4 L k L (Tsat Tw )
1/ 4
[7.2.21]
where the Jakob number JaL is defined as
Ja L =
c pL (Tsat Tw )
h LG
[7.2.22]
and is the first derivative of the dimensionless temperature at = 0, defined as
() =
Tsat T
Tsat Tw
[5.2.23]
and the dimensionless variable is
= c L y z 1 / 4
[5.2.24]
The boundary layer solution given by [7.2.21] includes an additional term compared to the Nusselt
solution given by [7.2.10], namely [(0) JaL1/4]. For common fluids with Prandtl numbers near unity,
their simulations showed that the momentum effects were negligible as assumed by Nusselt. Furthermore,
the error in utilizing [7.2.11] for JaL 0.1 and 1 PrL 100 was less than 3%. On the other hand, for
fluid with very low Prandtl numbers, e.g. liquid metals, the heat transfer coefficients are predicted with
the boundary layer analysis are much smaller than those given by the Nusselt solution.
7.3 Influence of Interfacial Phenomena on Laminar Film

Condensation
7.3.1. Ripples and Waves
The assumption that the interface of the laminar falling film is smooth is often not true. Observations
indicate that the interface becomes unstable and forms ripples or waves as shown in Figure 7.4 and Drew
(1954) has pointed to them to explain the positive difference between carefully measured experimental
values and those given by Nusselt theory, with up to 50% higher values in some cases. The ripples
increase heat transfer by enlarging the interfacial area and by reducing the mean thickness of the film.
Based on analysis of experimental observations, Brauer (1956) related the film Reynolds number Re for
the onset of waves to the Archimedes number ArL, which is defined as
7-8
ArL =
L2 3 / 2
L2 g 1 / 2 ( L G )3 / 2
[7.3.1]
The laminar falling film condensation heat transfer coefficient on

vertical plates is enhanced when
Re > 9.3 ArL1 / 5
[7.3.2]
This can be used as a criterion for the onset of ripples or waves;

an alternative criterion suggested elsewhere is when Re > 30. To
account for the effect of waves on heat transfer, Kutateladze
(1963) suggested multiplying the film Reynolds number in
[7.2.18] by an empirical correction of [0.8 (Re/4)0.11], such that
[7.2.18] becomes
f (z )
2L
k L L (L G )g
1/ 3
= 0.756 Re 0.22
[7.3.3]
Integrating over the film Reynolds number range from 0 to Re
Re
Re
dRe
(z )
0
[7.3.4]
Figure 7.4. Ripples and waves on a

falling film.
using expression [7.2.18] for values up to 30 and [7.3.3] for Re > 30, the resulting expression is:
f
kL
2L
L ( L G )g
1/ 3
Re
1.08 Re1.22 5.2
[7.3.5]
This expression is that of Butterworth (1983) and is valid up to the onset of turbulence. He noted that
most experimental studies give the onset of turbulence at a film Reynolds number of about 1600.
Expression [7.3.5] is solved in an iterative fashion up to the point of z in order to determine the
condensate flow rate and film Reynolds number from the heat transferred.
Local expressions for the condensation heat transfer coefficient in terms of the film Reynolds number are
helpful during thermal design. For example, a real plate with liquid cooling may have a non-trivial
variation in (Tsat-Tw). Similarly, a non-isothermal plate can be modeled by dividing it into successive
isothermal zones, including the variation in local physical properties. Finally, calculation of the local film
Reynolds number is an explicit way to determine when the laminar condensation equations are applicable
or if the turbulent film flow regime has been reached.
7.3.2 Cocurrent Interfacial Vapor Shear

Interfacial shear of cocurrent vapor flow on a laminar falling film tends to thin the film and increase heat
transfer. On the other hand, counter-current vapor flow would counteract the body force on the
condensate and thicken the film. Nusselt analyzed condensation under the influence of cocurrent
7-9
interfacial vapor shear at high vapor velocities for a laminar liquid film and the corresponding equation to
[7.2.1] for cocurrent vapor shear conditions is
du y
dz
dy
( y ) dz L g sin dp + idz = L
dz
[7.3.6]
The inclination of the plate is included by replacing g with g sin where is the inclination angle of the
surface relative to horizontal. The cocurrent vapor produces an interfacial shear stress i. For convenience
sake, a fictitious vapor density G* is defined:
dp
*
= G g sin
dz
[7.3.7]
If the static head of the vapor is taken as the pressure gradient, then G* = G. Substituting [7.3.7] into
[7.3.6] and integrating, the following relationships are obtained:
uy =
y2 i y
G* g sin
y
L
2 L
[7.3.8]
L ( L *G ) g sin 3 2
=
+ i L
3
2 L
L
[7.3.9]
d L ( L G* ) g sin 2 i L
=
+
L
d
L
[7.3.10]
Repeating the Nusselt analysis that was used to obtain [7.2.8], the corresponding expression is
4 L k L z(Tsat Tw )
4
i 3
4
=
*
*
g sin hLG L ( L G )
3 L G g sin
[7.3.11]
Rohsenow, Webber and Ling (1956) defined the three dimensionless variables (film thickness *, distance
from the leading edge z*, and interfacial shear stress i*) to resolve this expression:
( * ) g sin
= L L 2G
1/ 3
[7.3.12]
4k z(Tsat Tw ) *
z* = L
hLG L
[7.3.13]
7-10
*
i
i* =
*
( L G ) g sin
[7.3.14]
and transformed [7.3.11] to
4
z * = *4 + *3 i*
3
[7.3.15]
Integrating, this gives
Nu = f
kL
L2
*
L ( L G )g sin
1/ 3
( )
4 *z
*i (*z ) 2
2
=
+
3 z*
z*
[7.3.16]
where the corresponding film Reynolds number is
4 ( z )
4 *3
(
z ) + 2 i* ( z* ) 2
3
[7.3.17]
By substituting [7.3.17] into [7.3.16], z* can be eliminated. Figure 7.5 shows this expression evaluated
for cocurrent downflow by Rohsenow, Webber and Ling (1956) where the dashed lines depict the onset
of turbulence in the film. As an example of the potential effect of vapor shear, at a film Reynolds number
of 40 the Nusselt number is increased by about a factor of 1.5 to 6 times for the range of shear stresses
simulated!
Butterworth (1981) analyzed an analogous case for the situation where the gravitational forces are
negligible with respect to the interfacial shear imposed by the co-current vapor flow and laminar film
flow. His dimensionless local condensation heat transfer coefficient for these conditions is:
+f (z ) = 1.41 Re1 / 2 ( i+ )
1/ 2
[7.3.18]
where the dimensionless local heat transfer coefficient is
(z )
2L
(z ) = f
k L L ( L G )g
1/ 3
+
f
[7.3.19]
and the dimensionless interfacial shear stress is
i+ =
L i
[ L ( L G ) L g ]2 / 3
[7.3.20]
7-11
Figure 7.5. The influence of co-current vapor shear on the mean laminar falling film
condensation heat transfer rate according to Rohsenow et al. (1956) [from Collier and
Thome (1994)].
In his reworking of the Rohsenow, Webber and Ling (1956) analysis, low and high vapor shear effects are
distinguished by use of a critical liquid film Reynolds number Recrit for the onset of turbulence. An
important effect of the downward interfacial shear created by the vapor is that it reduces the critical
Reynolds number at which the film flow becomes turbulent. For i+ 9.04, the critical film Reynolds
number is
( )
Re crit = 1600 226i+ + 0.667 i+
[7.3.21]
For i+ > 9.04, Recrit is a fixed value, i.e.
Re crit = 50
[7.3.22]
The interfacial shear stress i is
i =
1
i G u LG
2
[7.3.23]
where uLG is the superficial velocity of the vapor with respect to the velocity of the film. The interfacial
friction factor i is assumed to be that for laminar flow on a flat plate:
i = 0.664 Re G1 / 2
[7.3.24]
7-12
where the local vapor Reynolds number is
Re G =
uGz
G
[7.3.25]
and z is the distance from the leading edge. Thus, depending on the local value of Re with respect to
Recrit, the appropriate expression is used to calculate the local heat transfer coefficient. To implement this
method, [7.3.18] must be integrated from the leading edge and an energy balance performed to determine
the film flow rate and hence the value of Re.
7.3.3 Combined Effects of Gravity and Interfacial Vapor Shear

When both the gravitational and interfacial vapor shear forces are significant, Butterworth (1981)
recommends using a simple asymptotic expression to incorporate their respective influences:
2
2
= sh
+ grav
1/ 2
[7.3.26]
where grav is the heat transfer coefficient for gravity-dominated flow from one of the Nusselt
expressions, such as [7.3.5], and sh is the heat transfer coefficient for shear-dominated flow, such as
[7.3.16].
7.4 Turbulent Film Condensation on a Vertical Plate without

Vapor Shear
The critical film Reynolds number at which a falling film becomes turbulent is still in dispute. Colburn
(1934) set the transition at a film Reynolds number of 2000 when comparing his experimental data to the
Nusselt theory. Applying an analogy to turbulent liquid flow in pipes, Colburn proposed the following
correlation for the local condensing coefficient for turbulent film condensation on a vertical plate
f (z)
2L
k L L ( L G )g
1/ 3
= 0.056 Re 0.2 PrL1 / 3
[7.4.1]
where z is measured from the top of the isothermal plate. In applying this expression, the Nusselt
expression [7.2.9] is used up to a local film Reynolds number of 2000 and then [7.4.1] is used above the
value of 2000.
Labuntsov (1957) proposed a similar expression for the local condensation heat transfer coefficient when
PrL 10:
f (z )
2L
k L L ( L G )g
1/ 3
= 0.023 Re 0.25 PrL0.5
[7.4.2]
There is a notable disagreement between the empirical values in these expressions.
7-13
Figure 7.6 depicts the local Nusselt number for condensation on a vertical plate without vapor shear
obtained with [7.2.18], [7.3.3] and [7.4.2] for the laminar, wavy laminar and turbulent regimes,
respectively. The heat transfer coefficient decreases with increasing film Reynolds number in the laminar
regime but increases with increasing Reynolds number in the turbulent regime. The large jump in the heat
transfer coefficient when passing from the laminar wavy regime to turbulent film flow is due to the fact
that Prandtl number effects were ignored in the Nusselt laminar film theory.
Figure 7.6. Local Nusselt numbers on a vertical plate.
Figure 7.7. Mean Nusselt numbers predicted by various methods [from Collier and
Thome (1994)].
Butterworth (1983) obtained the mean heat transfer coefficient over the laminar wave-free, the wavy
laminar and the turbulent zones by combining [7.2.18], [7.3.3], and [7.4.2] with the help of [7.3.4]:
7-14
f
kL
2L
L ( L G )g
1/ 3
Re
8750 + 58 PrL0.5 Re 0.75 253
[7.4.3]
This expression uses a turbulence threshold of 2000 and compares well to experimental data for
condensation of steam over the range 1 < Re < 7200.
The film Reynolds number threshold to turbulent film flow remains to this day in dispute. Some
publications cite this occurring at a film Reynolds number as low as 1200 while others propose values of
1800 and 2000.
As an alternative to the empirical approaches above, turbulent boundary layer theory has been applied to
falling film condensation by Seban (1954), Dukler (1960) and then Lee (1964) using universal velocity
profiles to evaluate the eddy diffusivities in the governing boundary layer equations. A comparison of the
Nusselt laminar flow theory to the Colburn, Seban and Dukler methods is depicted in Figure 7.7. Mills
and Chung (1973) argued that the interface has a damping effect on large turbulent eddies in the film,
such that there is a viscous sublayer at both the wall and the interface. Kutateladze (1982) arrived at the
same intuitive conclusion and proposed an eddy viscosity variation that goes to zero at the wall and the
interface.
7.5 Laminar Film Condensation on a Horizontal Tube

Applying the Nusselt integral approach to laminar film condensation on a vertical isothermal plate, the
similar process on the outside of a single, horizontal isothermal tube can be analyzed. Condensation on
the outside of horizontal tube bundles is often used for shell-and-tube heat exchanger applications and the
first step is the analysis of a single tube. The flow is nearly always laminar on a single tube because of the
short cooling length around the perimeter and is illustrated in Figure 7.8. Taking the same assumptions
for the vertical isothermal plate and an integral approach for the flow, an energy balance between the onedimensional heat conduction across the liquid film of thickness and the latent heat absorbed by the
liquid from the condensing vapor at the interface gives
h LG
d k L [Tsat (p G ) Tw ]
=
dz
[7.5.1]
Gravitation (g sin) is applied around the circumference of the tube, where is the angle around the
perimeter from the top. Similar to [7.2.2] for a vertical plate, a momentum balance on this element and
integration from the initial boundary condition of u = 0 at y = 0 results in the following velocity profile in
the film as a function of :
uy =
( L G )g sin y
L
y2
2
[7.5.2]
Integrating the velocity profile from the wall to the film interface gives the condensate mass flow rate per
unit length of tube, at the angle :
= L u y dy =
0
L ( L G )g sin 3
3 L
[7.5.3]
7-15
Figure 7.8. Condensation model for film condensation on a horizontal tube.

The length from to the top of the tube is z, which is related to the angle as = z/r, where r is the radius
of the tube. Substituting into [7.5.3] and then substituting [7.5.3] into [7.5.1] yields:
rk (T Tw ) ( L G )g
d = L sat
3
h LG
L
1/ 3
1/ 3
sin 1 / 3 d
[7.5.4]
Integrating from the top of the tube where = 0 at = 0 to the bottom where = at = gives the
condensate flow rate on one side of the tube per unit axial length of tube
r 3k 3L (Tsat Tw )3 ( L G )g
= 1.924
h 3LG v L
1/ 4
[7.5.5]
where is the condensate flow rate for one side of the tube. Here, L is the kinematic viscosity of the
liquid. An energy balance on the circumference of the tube gives the mean heat transfer coefficient for the
perimeter of the tube as
7-16
2hLG = 2r f (Tsat Tw )
[7.5.6]
Substituting for from [7.5.5], the mean heat transfer coefficient is
( G )ghLG k 3L
f = 0.728 L L
D L (Tsat Tw )
1/ 4
[7.5.7]
where D is the outside diameter of the tube. The original value of 0.725 was obtained from a numerical
solution while the correct analytical value is 0.728. Heat transfer is inversely proportional to the tube
diameter to the power. The mean condensate heat transfer coefficient may also be written in terms of
the film Reynolds number as
f
2L
k L L ( L G )g
1/ 3
= 1.51 Re 1 / 3
[7.5.8]
where the film Reynolds number is defined by [7.2.16]. (z) is the mass flow rate of condensate draining
from the bottom of one side of the tube per unit axial length (i.e. the condensate is given by [7.5.5]).
This expression is valid for Re up to about 1600, which is unlikely to be exceeded on a single tube. The
influence of subcooling may be incorporated by introducing [7.2.19]. The mean condensation heat
transfer coefficient written in terms of the Nusselt, Rayleigh and Jakob numbers is
1/ 4
Ra
D
Nu D = f = 0.728 L
kL
Ja L
[7.5.9]
The Nusselt number is written with respect to the tube diameter D. The Rayleigh number represents the
ratio of the buoyancy force acting on the liquid in the film to the viscous force opposing flow while the
Jakob number represents the ratio of the liquid sensible heat to the latent heat. The Jakob number JaL is
given by [7.2.22] and the Rayleigh number RaL is defined as
Ra L =
g ( L G ) PrL D 3
[7.5.10]
L L2
The local heat transfer coefficient around the perimeter of the tube as a function of the angle is:
( G )g sink 3L
f () = 0.693 L L
() L
1/ 3
[7.5.11]
Applying instead a uniform heat flux as the boundary condition around the tube rather than an isothermal
temperature, Fujii, Vehara and Kurata (1972) obtained identical expressions to [7.5.7] and [7.5.8] except
that the lead constants become 0.70 and 1.47, respectively. The isothermal boundary condition solution is
nearly always used for thermal design.
7-17
7.6 Condensation on Horizontal Tube Bundles

In order to thermally design a shell-side condenser, the previous single tube film condensation analysis
must be extended to model the process on horizontal tube bundles, which is the most widespread
application of film condensation. Condensation on tube bundles raises several important considerations:
In what manner does the condensate flow from one tube to the next?
Is subcooling of the film important?
Is the influence of vapor shear significant and, if so, how can this be accounted for?
At which point does the film go through the transition from laminar to turbulent flow?
To date, these have only been partially resolved and the current knowledge will be presented below.
7.6.1 Tube Row Effect
Figure 7.9. Condensation flow modes on horizontal tube arrays.

During condensation on a tube bundle, the condensate from the above tubes drains onto the tubes below,
increasing the amount of condensate flowing on each tube in addition to the new condensate formed on
that particular tube. This inundation of condensate from tube row to tube row is often referred to as the
tube row effect. In fact, it is more complex than just being a factor of how many tubes are located one
above the other, as analyzed by Nusselt (1916) in his landmark study in which the original vertical plate
Nusselt theory was presented. The tube row effect is not only dependent on how much condensate flows
from tube to tube but also in what physical mode the condensate achieves this. The flow regimes formed
by the condensate as it flows from one tube to that directly below it in an array of horizontal tubes are
depicted schematically in Figure 7.9. The regimes encountered are described as follows for increasing
film mass flow rate:
Droplet mode. The liquid flows from tube to tube as individual droplets, often in rapid succession at
uniform intervals along the bottom of the upper tube. The droplets range in size depending on the
7-18
properties of the fluid, primarily density difference, surface tension and liquid viscosity. The droplets
essentially form a jet that impinges on the lower tube. The droplets form because a continuous film on
the bottom of a tube becomes unstable as a result of the Kelvin-Helmholtz instability, which causes
liquid droplets to be formed at characteristic wavelengths such that droplet jets are formed at nearly
equal intervals along the bottom of the tube.
Column mode. At higher flow rates, the jets of droplets coalesce to form individual liquid columns
that extend from the bottom of the upper tube to the top of the lower tube. The liquid columns hit the
top of the lower tube as impinging jets and spread along the tube while flowing on around towards the
bottom as shown in Figure 7.10 adapted from Mitrovic (1986), in which the columns are staggered
from one tube row to the next. The columns may also be inline, one above the other in each
successive tube row.
Figure 7.10. Condensate spreading over the tube for column type
of condensate drainage.
Sheet mode. At even higher flow rates, the columns become unstable and form short patches of
liquid sheets that flow from one tube to the next. Eventually, with enough liquid, these unstable sheets
join together to form more or less one continuous sheet of liquid falling from the upper tube to the
lower one. This occurs only if the intertube spacing is small enough to prevent breakup of the sheet.
For staggered tube arrays, if the column sheet is unstable, it may attach to the side of the next lower
out-of-line tube without reaching the top of the next inline tube. Waves or ripples are typically
noticeable on the flowing sheets.
Spray mode. Under vapor shear conditions with its complex flow field around tubes, the above flow
modes may be interrupted and the liquid carried away by the vapor. In this case, there is significant
entrainment of liquid droplets into the vapor flowing between the tubes and hence a spray flow is
formed. Any liquid flowing from one tube to another does so in a very chaotic manner. For staggered
tube arrays, the vertical tube spacing is quite large and these liquid columns and droplet jets are easily
influenced by vapor buffeting, especially where the vapors boundary layer detaches from the sides of
the tubes in vortices and in the recirculating flow regions behind the tubes. At high vapor flow rates, a
large fraction of the liquid is entrainment as small liquid droplets, resulting in very thin liquid films
on the tubes.
7-19
For an actual view of these intertube flow modes, which may be thought of as shell-side falling film twophase flow regimes, Figure 7.11 depicts the three modes (droplet, column and sheet) and their two
transitions. The images were taken with a high-speed digital video camera. The columns in this case are
inline but staggered columns also occur as mentioned above.
Figure 7.11. Video images of three flow modes and their

transitions [images courtesy of Laboratory of Heat and Mass
Transfer, Swiss Federal Institute of Technology Lausanne].
The Nusselt equation for a single tube may be extended to a vertical array of horizontal tubes, i.e.
horizontal tubes situated one above the other, assuming that the sheet mode exits in between all the tubes.
Essentially, this is equivalent to assuming that there is an embossed vertical plate of half-tubes with no
gap between the tubes with the condensate flowing from one to another. Starting with [7.5.8], the analysis
is applied to each individual tube utilizing the summation of the condensate flowing from the above tubes
onto the top of the Nth tube, counting the tubes from the top row towards the bottom. Applying an energy
balance to the entire surface area of the array of tubes yields
2rN(Tsat Tw ) = 2h LG [bottom (N )]
[7.6.1]
where is the mean condensation heat transfer coefficient for all the tubes, N is the number of tubes in
the vertical array, hLG is the latent heat and bottom(N) is the total flow rate of condensate off the bottom
tube from one side per unit length of tube. The mean condensation heat transfer coefficient for N tubes of
diameter D is
g(L G )(ND )3 h LG
(ND )
= 0.728
kL
k L L (Tsat Tw )
1/ 4
[7.6.2]
The mean heat transfer coefficient in terms of the film Reynolds number for the condensate leaving the
bottom of the Nth tube is
7-20
L2
k L L ( L G )g
1/ 3
4(N )
= 1.92
L
1 / 3
[7.6.3]
where the flow rate of condensate draining from one side of the Nth tube is
(N ) =
ND(Tsat Tw )
h LG
[7.6.4]
The mean condensation heat transfer coefficient , for an entire array of tubes is:
(N = 1)
= N 1 / 4
[7.6.5]
where (N=1) is the heat transfer coefficient for the top tube row, i.e. the original Nusselt equation for a
single tube given by [7.5.7] and [7.5.8]. This method is applicable if the film flow remains laminar all the
way to the bottom of the Nth row. The heat transfer coefficient on the Nth tube row in the bundle (N) is
(N )
(N = 1)
= N 3 / 4 (N 1)
3/ 4
[7.6.6]
Kern (1958) concluded from his practice experience in designing condensers that the above tube row
expression was too conservative and that this resulted in condensers that were consistently over-surfaced.
To improve his thermal designs, he replaced the exponent of (-1/4) in expression [7.6.5] with a value of (1/6) so that corresponding equations become
(N = 1)
(N )
(N = 1)
= N 1 / 6
[7.6.7]
= N 5 / 6 ( N 1) 5 / 6
[7.6.8]
These equations are presently widely used in the thermal design of condensers. Figure 7.12 presents a
comparison of experimental data from different sources compiled by Marto (1986) with respect to the
Nusselt and Kern tube row methods. The wide bandwidth of the data compared to the two curves may be
some vapor shear effects in the data but is more likely the influence of intertube flow mode types
encountered during the experiments, i.e. the modes described above. Experiments have shown that most
data are equal to, or higher than, those given by Nusselts sheet flow analysis. The discrepancy arises
because the type of flow has an effect on the analysis. For example, if the flow mode is column-wise, then
a portion of the lower tubes will perform without condensate inundation, while the rest of the tube surface
will suffer more inundation than in the sheet mode. Hence, the performance will change accordingly.
Overall, the column mode is thought to have a positive influence on heat transfer. Therefore, the Nusselts
sheet mode solution is the lower limiting case for the tube row effect. Heat transfer in the other two
modes (column and droplet) is expected to be higher.
7-21
Figure 7.12. Competing tube row methods for condensation on a vertical array of tubes
compared to experimental results from Marto (1986).
Figure 7.13. Illustration of realistic flow of liquid on tube bundle.

It is important to understand that these tube row expressions are applied in practice by counting the
number of vertical inline tube rows. Hence for an inline or square tube layout, each tube row from the top
to the bottom of the bundle is counted in applying these equations. Instead, for staggered tube layouts the
condensate is normally assumed to flow to the next inline tube row since it cannot flow onto the top of the
out-of-line tube in the next lower row. Thus, the total number of tube rows to use is one-half the actual
number, which naturally means that staggered layouts are more advantageous for heat transfer. Since
some condensate in a staggered bundle ends up on the sides of the next out-of-line tubes, this assumption
is a little optimistic as shown in Figure 7.13 adapted from Marto (1986). Hence, the total number of
7-22
vertical tube rows in a staggered tube bundle are somewhere between the number of inline rows and the
total number of staggered rows, and an average of these two values may be used as a reasonable
approximation.
7.6.2 Falling Film Flow Regime Transitions on Tube Arrays

At this date, no generalized falling film mode transition map is apparently available for condensation
conditions. Honda, Nozo and Takeda (1987) notably, however, has proposed several transition
expressions for several refrigerants condensing on low finned tubes. This process is very similar to that of
an adiabatic film of liquid falling on a tube array, which has been studied by Hu and Jacobi (1996a) for a
variety of fluids, tube diameters, tube pitches and flow rates and with/without cocurrent gas flow and by
Mitrovic (1986) as a function of tube spacing, all for plain tubes. More recently by Roques, Dupont and
Thome (2002) and Roques and Thome (2001, 2002) have run tests with water, glycol and one waterglycol mixture for one plain tube, then 19 fpi, 26 fpi and 40 fpi low fin tubes, and finally two enhanced
tubes (Turbo-Bii and Thermoexcel-E). The 26 fpi tube studied was a Wolverine Turbo-Chil tube.
Hu and Jacobi (1996a) proposed flow mode transition equations with Re versus GaL (film Reynolds
number vs. the Galileo number) for the following principal flow modes: sheet flow, column flow and
droplet flow. The mixed mode transition zones of column-sheet and droplet-column were also considered
as regimes, bringing the total to five. Hence, they presented four flow transition expressions (valid for
passing through the transitions in either direction and hence the symbol ):
DropletDroplet-Column:
Re = 0.148 Ga L0.302
[7.6.9]
Droplet-ColumnColumn:
Re = 0.192 Ga L0.301
[7.6.10]
ColumnColumn-Sheet:
Re = 2.828 Ga L0.233
[7.6.11]
Column-SheetSheet:
Re = 2.896 Ga L0.236
[7.6.12]
The film Reynolds number Re in their original publication was for flow down one side of the tube;
hence, the above expressions are evaluated using the film Reynolds number definition given by [7.2.16]
with flow down one side of the tube to conform to the situation here. The Galileo number GaL is defined
as
Ga L =
L 3
L4 g
[7.6.13]
7-23
Figure 7.14 depicts these expressions as a flow transition map, applicable only to plain tubes and air
velocities less than 15 m/s.
Figure 7.14. Flow mode map for falling film flow on a vertical array of horizontal tubes.
Honda, Nuzo and Takeda (1987) have looked at some of the transitions in flow modes on a low finned
tube for one refrigerant. More recently, Roques, Dupont and Thome (2002) and Roques and Thome
(2001) have reported transition curves similar to those above for plain tubes, low finned tubes (19, 26 and
40 fpi tubes), Thermoexcel-C and Turbo-Bii tubes. They observed that the low finned tubes and the
enhanced condensing Thermoexcel-C tube had some significantly different transition thresholds,
primarily enlarging the range of conditions at which the column mode exists, which is positive from a
heat transfer perspective. These types of studies will eventually lead to condensation design methods in
which the tube row effect includes the flow mode effect.
7.6.3 Vapor Shear Effects on Tube Bundles

Vapor shear manifests its influence on the condensation process in tube bundles primarily in three ways.
First of all, it has an effect of the film flow, tending to increase heat transfer by thinning the film, similar
to its effect on vertical plates described earlier. Secondly, the interfacial shear may remove the film from
the tube wall, similar to the transition from annular flow to mist flow inside a tube at high vapor core
velocities. Thirdly, the vapor shear modifies the intertube flow modes. For instance, the sheet mode is not
stable in presence of high velocity vapor and will be broken up into droplets, creating a mist flow. In a
condenser, high vapor velocities may exist near the inlet. Deeper into the bundle, however, the vapor
velocity decreases rapidly as the vapor condenses.
McNaught (1982) has estimated the shear-controlled heat transfer coefficient sh for the situation where
the vapor shear force is dominant and the gravity force on the film is negligible as
sh = 1.26 (1 / X tt ) 0.78 L
[7.6.14]
In this expression the liquid film heat transfer coefficient L is determined assuming that the liquid phase
occupies the entire flow channel, i.e. the liquid fraction of the total flow at the minimum cross-sectional
7-24
area of the bundle, and flows as a turbulent liquid across the bundle. The value of L can be determined
from correlations for crossflow heat transfer over tube bundles found in textbooks. The Martinelli
parameter Xtt he used is
1 x
X tt =
0.9
G

L
0.5
L

G
0.1
[7.6.15]
where x is the vapor quality. To find the combined effect of gravity controlled condensation and vaporshear controlled condensation, the gravity-controlled heat transfer coefficient grav is calculated with the
Nusselt single tube expression [7.5.7] together with the inundation tube row effect. For staggered and
inline layouts, he proposed the following expression
grav
( N 1) + N
= (N = 1)
[7.6.16]
where (N-1) is the total condensate flow rate from one side of the above tubes onto the Nth tube and N
is the condensation rate on one side of the Nth tube itself, whose values may be obtained with [7.6.4]. The
heat transfer coefficient for the top tube row is (N=1). The empirical exponent is 0.13 for triangular
tube layouts and 0.22 for square tube layouts, respecting the larger tube row effect anticipated for the
inline tube rows in the square layout, based on a relatively small database. Asymptotic approaches are
often used for modeling the combined effects of two different mechanisms on heat transfer on heat
transfer and in this case McNaught uses the expression
2
2
= sh
+ grav
1/ 2
[7.6.17]
where the above two expressions are used to determine the vapor-shear controlled and the gravitycontrolled coefficients.
A more detailed method has been proposed by Honda et al. (1989) for condensation on inline and
staggered tube bundles of plain tubes using an asymptotic model to the 4th power. Refer to that paper, or a
summary of it in Collier and Thome (1994), for more information on that method.
7.6.4 Onset of Turbulence and Turbulent Film Heat Transfer

Turbulent flow of the condensate film may be reached in a condenser, which significantly increases heat
transfer. Comparatively little has been published on turbulent film condensation on tube bundles
compared to the information available for laminar films. Butterworth (1983) recommends adapting the
Labuntsov expression [7.4.2] for turbulent film condensation on a vertical plate to horizontal tubes for
predicting local turbulent film condensation on the Nth tube row in horizontal tube bundles:
(N )
2L
k L L ( L G )g
1/ 3
= 0.023(Re / 2 )
0.25
PrL0.5
[7.6.18]
The film Reynolds number is obtained using [7.2.16] with the condensate flow rate on one side of the Nth
tube in order to conform to the same definition used for turbulent condensation on a vertical plate. The
transition film Reynolds number for the tube bundle is adapted from a vertical plate turbulent transition
7-25
criterion of 1600 (but also values of 1200, 1800 and 2000 have been proposed) for a film on a plate to one
side of a tube. Thus, the film will become turbulent on the tube bundle at Re equal to 1600 and thus for
values larger than 1600 the above expression should be used.
7.7 Condensation on Low Finned Tubes and Tube Bundles
Figure 7.15. Diagram of an integral low finned tube.

Integral low finned tubes have been utilized for
enhancing condensation for more than half a
century. The geometry of a low finned tube is
illustrated in Figure 7.15 and a photograph is shown
in Figure 7.16. These tubes have their fins formed
by typically three sets of rings of increasing
diameter that are forced into the tube wall as the
tube is pulled through the ring assemblies. The fins
form in the spaces between the rings and hence the
term integral. The low finned tubes used for
condensation typically have a fin tip diameter nearly
identical to the diameter of plain ends of the tube
Figure 7.16. Photograph of a low finned tube.
such that they are easily inserted through tube sheets
into tube bundles. The fins are helical around the tube with a small axial pitch. For condensation, the
optimum fin density depends on the particular fluid, primarily the surface tension, and vary from 19
fins/in. (19 fpi or 748 fins/m) up to 42 fpi (1653 fins/m). Fin heights depend on the fin density and the
particular tube metal, ranging from about 0.66 to 1.50 mm (0.026 to 0.059 in.). The most typical fin
thickness is 0.305 mm (0.012 in.).
7-26
7.7.1 Role of Surface Tension on Film Condensation on Low Finned

Tubes
The influence of surface tension on two-phase flow and heat transfer is quite complex. Surface tension
itself is a cohesive interfacial force created by the difference between the intermolecular forces at the
phase transition interface created by the respective molecules, which are closely packed in a liquid and
relatively loosely packed in a vapor. Surface tension is what holds a liquid droplet together and tends to
form a geometry that minimizes the interfacial surface area, being it a droplet in a vapor or a small film of
liquid on a surface. The pressure difference across a liquid-vapor interface of principal radii r1 and r2 is
given by the Laplace equation, whose general form is
1 1
p = +
r1 r2
[7.7.1]
where is the surface tension and the pressure difference is
p = p L p G
[7.7.2]
where pL is the local liquid pressure and pG is the local vapor pressure and r is the radius of the interface.
For a cylindrical interface, one of the radii goes to infinity and the pressure difference becomes
1
p =
r
[7.7.3]
Instead, for a spherical surface the two radii are equal and thus
2
p =
r
[7.7.4]
For a constant vapor pressure at isothermal conditions, if the curvature of the interface is not constant,
such as for a liquid film on a fin or a corrugation on the surface of a tube, then the local pressure in the
liquid will not be constant. This induces flow of the liquid from the region of small curvature towards that
with larger curvature. That is, this will create a surface tension driven flow in the film. On the other hand,
if the interface is of uniform radius but is subjected to a temperature gradient, again a pressure gradient
will be produced in the liquid and cause liquid to flow from the high temperature region (low surface
tension) towards the low temperature region (high surface tension). This latter process is referred to as the
Marangoni effect. This effect is also created if the fluid is a mixture and a concentration gradient exists in
the liquid phase along the interface, which creates a surface tension gradient. Of these two surface tension
flows, the pressure difference created by curvature of the interface is typically much stronger in film
condensation than that of the Marangoni effect for the small temperature differences involved. Hence,
condensate flow from the tip of a fin or corrugation towards its base is primarily from the effect of
curvature of the liquid-vapor interface.
For condensate formed on a horizontal low fin tube, the two forces acting on the liquid film in quiescent
vapor conditions are gravity and surface tension. Here, surface tension draws the liquid from the fin tip
towards its root and dominates the gravity force on the liquid with its downward influence. Instead, in the
root area between the fins, the radius of the film around the circumference of the tube is relatively large
7-27
and uniform; thus, gravity dominates in this direction and governs the drainage of condensate from the
tube while surface tension tends to promote retention of the condensate between the fins.
The first to exploit this surface tension induced flow in film condensation was Gregorig (1954) and this
has become to be known as the Gregorig effect, illustrated in Figure 7.17. He proposed a vertical,
longitudinally fluted surface on which the condensate is drawn to the troughs by surface tension forces
and then drains downward by gravity forces, enhancing heat transfer relative to a planar surface. His work
stimulated numerous developments of enhanced condensation surfaces. For more on this topic, refer to
the recent review of Shah, Zhou and Tagavi (1999).
Figure 7.17. Gregorig effect on a liquid film on a fluted surface.
7.7.2 Beatty and Katz (1948) Model of Condensation on a Single

Horizontal Low Finned Tube
Beatty and Katz (1948) proposed the first model for condensation heat transfer on a horizontal low finned
tube. They applied vertical surface Nusselt analysis to condensation on the fin assuming the condensate
drains downwards and Nusselt theory for a horizontal tube to condensation on the root area between
adjacent fins i.e. a two-zone approach. The condensation heat transfer coefficient is thus an area weighed
average of the condensation coefficient on the root area per unit length Aroot, referred to as root, and the
condensation coefficient on the fin area per unit length Afin , referred to as fin. Taking into account the fin
efficiency, the condensing coefficient on the low fin tube is
surface = root
A
Aroot
+ fin fin fin
Atotal
Atotal
[7.7.5]
where is based on the effective external surface area Aeff of the low finned tube. The total surface area
per unit length Atotal is
Atotal = Aroot + A fin
[7.7.6]
7-28
where Afin includes the fin tip area. The surface efficiency surface is
surface = 1
A fin
Atotal
(1 )
[7.7.7]
fin
The effective external surface area Aeff is equal to surfaceAtotal. The fin efficiency fin is normally obtained
from the fin efficiency expression for a circumferential fin of uniform thickness with an insulated tip,
assuming a uniform heat transfer coefficient over its surface. The fin height is corrected for the latter
assumption by adding one-half of the fin tip thickness to the height of the fin when calculating the fin
efficiency. Thus, its value can be estimated from the following expression applying the mean heat transfer
coefficient on the fin fin, while also introducing the fouling factor R, as follows:
fin
t

tanh m e +
2

=
t
m e +
2
1
1

2
+ R
fin

m=
k fin t
[7.7.8]
1/ 2
[7.7.9]
Note that the fouling factor R is often neglected in fin efficiency expressions in textbooks. The fouling
factor tends to increase the fin efficiency with increasing values of R, very significantly for tubes of low
thermal conductivity. Note that the thermal resistances of the condensing coefficient on the fin and the
fouling factor are in series in the above expression since the heat first passes through the condensate film
and then through the fouling layer into the fin. In the above expressions e is the fin height, t is the mean
fin thickness and kfin is the thermal conductivity of the fin.
For the root area between the fins, the applicable diameter is the root diameter Droot at the base of the fins
and this diameter is substituted directly into [7.5.7] for condensation on a horizontal plain tube so that
root
( G )g k 3L h LG
= 0.725 L L
L D root (Tsat Tw )
1/ 4
[7.7.10]
The equivalent vertical length is then obtained by taking the surface area for one face of the fin and
dividing it by the fin tip diameter D, such that the equivalent length of the fin Lfin is
L fin =
2
)
(D 2 Droot
4D
[7.7.11]
Substituting into [7.2.11], the heat transfer coefficient on the side of the fins is
7-29
fin
( G )g k 3L h LG
= 0.943 L L
L L fin (Tsat Tw )
1/ 4
[7.7.12]
Thus, one first assumes a wall temperature at the root of the fins and then calculates the respective surface
areas and heat transfer coefficients. Then the mean coefficient for the entire tube is obtained. The above
equations can be incorporated into one single expression as
surface
L ( L G )g k 3L h LG
= 0.689
L (Tsat Tw )
1/ 4
A root 0.25
A
D root + 1.3 fin fin Lfin0.25
A total
A total
[7.7.13]
Beatty and Katz replaced the original value of 0.725 in [7.7.10] with an empirical value of 0.689 that fit
their experimental data better for four different fluids they tested. Expression [7.2.19] can be used in place
of hLG to include subcooling effects of the condensate. The heat transfer coefficient given by [7.7.13] is
defined relative to the effective surface area, i.e. Aeff=Aroot+finAfin. The nominal heat transfer coefficient
nom, based on DL where L is the length of the tube, is obtained by multiplying by the ratio of Aeff
/DL, which allows the low finned tube to be compared to a plain tube on a tube replacement basis. Note
that their method does not include the effect of condensate retention (see next section) and also the flow
on the fins is actually controlled by surface tension and not gravity. Hence their method is reasonably
accurate but is unreliable for fluids with large surface tensions or for large fin densities.
7.7.3 Condensate Retention Models for a Horizontal Low Finned Tube

As mentioned above, the Beatty-Katz equation does not
account for the condensate retained between the fins on
the bottom side of the tube by surface tension forces. The
Beatty-Katz approach assumes that all the condensate
formed drains off the bottom without any holdup between
the fins while a notable perimeter can be affected,
typically on the order of 10 to 40% of the circumference.
Figure 7.18 depicts the condensate retention half-angle ,
which refers to the angle from the bottom of the tube up to
the highest point at which condensate is held up to the tip
of the fins. Based on a balance of forces, a liquid
condensate retention half-angle equation was derived by
Honda, Nozu and Mitsumori (1983) as
= cos 1{1 (4 cos / L g b D )}
[7.7.14]
where D is the fin tip diameter and b is the interfin spacing

at the fin tips (e.g. fin pitch minus the fin thickness), g is
the acceleration due to gravity and is the surface tension.
is one-half the apex angle of the fin in radians (typically Figure 7.18. Condensate retention angle on
about 3-7 for standard trapezoidal low fins and 0 for a a low fin tube.
rectangular fin).
Rudy and Webb (1985) subsequently observed experimentally the retention ange on a variety of low
finned tubes and arrived at the following expression for the half-angle of the tube that retains condensate:
7-30
2 (2e / cos ) + t tip t root

= cos 1 1
DL g (s e A p )

[7.7.15]
In the above expression, the half-angle is in radians, e is the fin height, ttip is the fin thickness at the tip,
troot is the fin thickness at the root, s is the fin pitch, Ap is the cross-sectional area of the fin, and is onehalf the apex angle of the fin in radians. The expression worked quite well for their measured retention
angles. These two expression give the same retention angle for rectangular fins. For trapezoidal fins, they
also give the same result if b is set to the mean interfin spacing.
7.7.4 Webb, Rudy and Kedzierski (1985) Model of Condensation on a

Single Horizontal Low Finned Tube
Utilizing their condensate retention equation, Webb, Rudy and Kedzierski (1985) improved on the
Beatty-Katz model to obtain
surface = (1 cb ) root
A fin
Aroot
+ fin fin
+ cb b
Atotal
Atotal
[7.7.16]
Here fin is determined from surface tension driven flow on the fins as
fin
k3 h
= 0.943 L L LG
L (Tsat Tw )
1/ 4
2 1 1
e 2 s + t
1/ 4
[7.7.17]
and root is determined based on gravity controlled flow as
root
2L Re root
= 1.514
3 2
gk L L
1 / 3
[7.7.18]
where the film Reynolds number of the condensate flowing in the root area is
Re root =
4root
L (s t root )
[7.7.19]
root is obtained from
root = fin + [ root A root (Tsat Tw ) h LG ]
[7.7.20]
fin = fin fin A fin (1 c b )(Tsat Tw ) / h LG
[7.7.21]
and
The condensate flooded fraction of the tube perimeter cb is equal to / while the unflooded fraction
completely available for film condensation is (1-cb). They also included the heat transfer on the flooded
7-31
perimeter of the tube b, estimating its value numerically. However, it is simpler to assume laminar, fully
developed flow in a rectangular channel of twice the size of the interfin half-channel to obtain b.
7.7.5 Other Recent Low Finned Tube Condensation Models

More complex models for condensation on two-dimensional low finned tubes have been developed, such
as those of Adamek (1981), Honda and Nozu (1986), and Adamek and Webb (1990). Rose (1994)
proposed a simplified empirical approach with a correlation that captures most of the features of these
analytical models. For a more detailed review, refer to Webb (1994) or Shah, Zhou and Tagavi (1999).
7.7.6 Effects of Tube Row and Vapor Shear on Low Finned Tube
Arrays
Low finned tubes promote the Helmholz-Kelvin instability in the condensate flowing off the bottom of
the tube such that the column mode is encouraged as shown in Figure 7.19. The low fins also inhibit
lateral spreading of the impinging film along the lower tube and the majority of the condensate flows
around a low finned tube with minimal lateral spreading. As a consequence, large zones of the finned
surface between these columns are unaffected by condensate from the tubes above and hence these zones
function as if they were in the top row. Therefore, applying the Nusselt and Kern tube row factors, in
[7.6.5] and [7.6.7], to low finned tube bundles is extremely conservative, resulting in significant underpredicting of the bundles condensing heat transfer coefficient. Some experimental studies found tube row
exponents as low as -1/25 but it is more prudent value to use a tube row exponent of -1/10 or -1/12.
Figure 7.19. Condensate flow as a column jet between low finned tubes.
Cavallini et al. (1995) investigated the effect of interfacial vapor shear on condensation on a single low
finned tube. According to their experiments, a very large vapor Reynolds number threshold must be
surpassed before a significant influence is seen on low finned tube performance. Videos described in
Cavallini et al. (1994), with several exemplary photographs, also prove that the vapor shear has little
effect on the condensate retention angle, at least up to the point where the condensate is then literally
blown off the tube and a mist flow is formed. Thus, low finned tube bundles can be designed using only a
7-32
single-tube condensation model together with the appropriate tube row equation, as long as the film flow
remains in the laminar regime and the vapor Reynolds number remains below 100,000.
Cavallini et al. (1996) proposed an asymptotic method for the combined effect of gravity and vapor shear
on a single-tube:
2
2
= sv
+ fc
1/ 2
[7.7.22]
where sv is the heat transfer coefficient on the finned tube under stationary vapor conditions and fc is
the heat transfer coefficient under forced convection (vapor shear controlled) conditions.
k
fc = Cfc L Re 0eq.8 PrL1 / 3
D
[7.7.23]
and the empircal lead constant is determined from
Cfc = 0.03 + 0.116 t tip / s + 0.07(e / s )
[7.7.24]
Reeq is the equivalent Reynolds number and PrL is the liquid Prandtl number. To extend this method to
tube bundles, sv could be obtained by applying the tube row correction using an exponent of -1/10 or 1/12.
7.7.7 Three-dimensional Enhanced Condensing Tubes

Integral low finned tubes with three-dimensional fins are made by notching of conventional twodimensional low fins to form saw-toothed fins, or by crosscutting the fins. This greatly reduces
condensate retention and enhances heat transfer coefficients on the fins. While several comprehensive
experimental studies have been published for these tubes, no general analytical method is available at this
time for predicting their performance. Several experimental studies are described below.
Briggs, Yang and Rose (1995) completed a very accurate, comprehensive experimental comparison of 17
commercially available low finned tube geometries, including both 2-d trapezoidal fins and notched 3-d
fins, for R-113. One 2-d trapezoidal fin gave the same performance as the best of the 3-d notched fins, but
the 3-d fins had not been optimized for use with R-113.
Rewert, Huber and Pate (1996a, 1996b) measured condensing coefficients for R-123 and R-134a in a
comparison of a 1024 fpm (26 fpi) low finned tube, a 1575 fpm (40 fpi) low finned tube, a Turbo-CII tube
and a Gewa-SC tube (the latter have 3-d enhancements). The Turbo-CII tube significantly outperformed
the 40 fpi low finned tube (the better of the two low fin tubes) while the Gewa-SC only provided
moderate improvement with respect to the 40 fpi tube. In tests on a five-by-five staggered tube bundle and
using liquid overfeed to simulate up to 30 tube rows, they reported data by tube row for these fluids and
tubes. Figure 7.20 depicts four graphs of performance. For both fluids, the Turbo-CII shows a faster
falloff in tube row performance than the other tubes; since it starts from a much larger top tube row value,
its overall condensing performance is still higher than the other tubes, however. The tube row effect of the
Gewa-SC tube in the lower left graph is nearly negligible, with the lower tubes performing like the top
tube.
7-33
Figure 7.20. Condensing data of Rewerts, Huber and Pate (1996a, 1996b). Top left: low finned
tubes with R-134a; Top right: 40 fpi tube with R-123; Bottom left: Turbo-CII and Gewa-SC for R134a; Bottom right: Turbo-CII for R-123. [their figs. 4& 8 and 4 & 5 reprinted by permission of
ASHRAE Trans., 1996 copyright by American Society of Heating, Refrigeration and AirConditioning Engineers].
Chen and Tuzla (1996) ran extensive tests on a 19 mm (3/4 in.) Turbo-CII of Wolverine Tube at a
saturation temperature of 37.7C (100F) for R-22, R-134a, R-410a and several zeotropic refrigerant
mixtures. The R-134a performance was slightly above that of R-22 while R-410a outperformed those
fluids by about 15%. Performances of the zeotropic mixtures were much lower as they suffered from the
mass diffusion resistance in the vapor phase.
7.8 Condensation with Non-Condensable Gases

Condensation in the presence of a non-condensable gas occurs in numerous heat exchanger applications.
The most common example is dehumidification in an air-conditioning system, in which water vapor is
7-34
partially condensed out as the humid air passes through the evaporator. Another prime example is shellside condensation in surface condensers used in central power plants. Here, a small fraction of air that is
dissolved in the feedwater and it eventually arrives at the condenser together with the steam, where of
course it cannot condense. Hence, it will build up its concentration near the exit region of the condenser if
it is not removed with a steam ejector or some other similar device. In chemical processing plants and
refineries, non-condensable gases may be present in the process vapor leaving a distillation column that
goes to the overhead condenser or be produced in a reactor prior to a feed effluent heater. A noncondensable gas creates a significant mass transfer resistance to heat transfer depending on the
circumstances.
Figure 7.21 depicts this process in the presence of air under quiescent conditions. The total pressure of the
bulk vapor-air mixture is p while partial pressures of the vapor (G) and non-condensable gas (a) are pG
and pa, respectively. The bulk concentrations of vapor and air are WG and Wa, respectively (note that
WG + Wa = 1 at any location). TG is the bulk temperature of the vapor-air mixture and Tw is the wall
temperature where Tw is less than the dew point temperature of the vapor. As heat is transferred from the
vapor-air mixture to the cold wall, a condensate layer is formed on the wall and then condensation
proceeds at the interface (denoted by the subscript i). The non-condensable gas arriving at the interface
does not condense and, assuming it is not absorbed by the condensate (typically only a small fraction is),
its local concentration builds up at the interface to Wai. Since WGi = 1 - Wai, then the local concentration
of the vapor at the interface is WGi. This reduces the partial pressure of the vapor at the interface to pGi
and the saturation temperature at the interface is TGi. Condensation of the vapor at the interface creates a
mass diffusion process and its associated mass diffusion layer where the non-condensable gas diffuses
away from the interface towards the bulk where its concentration is lower. The vapor arrives at the
interface by (i) bulk motion of the vapor towards the interface (where it is removed as condensate) and (ii)
by mass diffusion down its concentration gradient from the bulk to the interface.
Figure 7.21. Condensation in the presence of a non-condensable gas.
7-35
The thermal resistances encountered by the process are thus three rather than one for condensation of a
pure vapor. First of all, there is the thermal resistance across the condensate film itself, whose value is
determined using a pure vapor heat transfer expression for the cold wall (plate or tube) where the film
may be laminar, laminar-wavy or turbulent. Then there is either natural convection heat transfer or forced
convective heat transfer across the temperature gradient from TG to TGi, depending on whether there is
forced flow of the gas-vapor mixture or not. Finally, there exists a mass transfer resistance to the diffusion
of vapor to the interface. Depending on the local circumstances, one or two of these resistances may be
negligible with respect to the others.
Various boundary layer treatments of condensation with non-condensable gases have been presented in
the literature over the years. For example, Minkowycz and Sparrow (1966) analyzed this process for
stagnant vapor conditions while Sparrow, Minkowycz and Saddy (1967) studied forced convention
conditions. Figure 7.22 depicts the effect of small fractions of air in saturated steam for falling film
condensation under stagnant and forced convection conditions. The vertical axis shows the condensing
heat flux with air present relative to that of pure steam for otherwise identical conditions, e.g. the ratio is
equal to 1.0 for no air present. For quiescent vapor conditions, even a small fraction of air reduces thermal
performance significantly. For forced convection conditions, the fall in thermal performance is less but is
still of particular magnitude. This comparison shows the importance of attaining high flow rates of the
vapor-air mixture to minimize the mass diffusion resistance. A simpler approach by Colburn is described
in Collier and Thome (1994).
Figure 7.22. Condensation of steam in the presence of air [taken

from Collier and Thome (1994)].
7-36
------------------------------------------------------------------------------------------------------------------------------Example Calculation: Assume R-134a is condensing on the outside of a 25.4 mm diameter, horizontal
low finned tube, where the diameter at the fin tips is 25.4 mm, at its saturation temperature of 40C. The
tube wall temperature is 35C. The tube has the following external dimensions: 1000 fins/meter, fin
height of 1.5 mm, fin thickness of 0.3 mm and fin tip diameter of 25.4 mm. The fins can be considered to
be rectangular in profile and without fouling. The thermal conductivity of the tube is 390 W/m K. Using
the Beatty-Katz method, determine the heat transfer coefficient for the tube based on the effective area,
the nominal heat transfer coefficient and the condensate retention angle.
Solution: The necessary physical properties at 40C are:
L = 1147 kg/m3; G = 50.0 kg/m3; L = 0.000183 Ns/m2; hLG = 162900 J/kg; L = 0.0776 W/mK;
cpL = 1.508 kJ/kgK; = 0.0061 N/m.
Using [7.2.19] to include subcooling of the condensate in the heat balance:
1508(40 35)
hLG = 162900 1 + 0.68
= 162900[1 + 0.0315] = 168027 J / kg
162900
D = 25.4 mm or 0.0254m and the root diameter Droot = 22.4 mm = 25.4 2(1.5) or 0.0224 m. Per meter of
tube, Aroot is the surface between the fins, such that for 1000 fins per meter (fpm):
A root = fpm (s t root )(D root ) = 1000(0.001 0.0003)(0.0224) = 0.04926 m 2

Afin is the fin surface area per meter of tube including the tip area:
A fin
D 2 D 2root
+ fpm D t tip
= A sides + A tip = 2 fpm
2
2
(0.0254) (0.0224)
+ 1000 (0.0254)(0.0003)
= 2(1000)
= 0.22525 + 0.02394 = 0.24919 m 2
Atotal is the total area per unit length of finned tube:
A total = A root + A fin = 0.04926 + 0.24919 = 0.29845 m 2

The equivalent length of the fin is obtained with [7.7.11]:
L fin =
(0.0254) 2 (0.0224) 2
= 0.004434 m
4(0.0254)
The heat transfer coefficient on the sides of the fin is obtained with [7.7.12]:
7-37
1/ 4
fin
1147(1147 50)(9.81)(0.0776)3 (168027)

= 0.943
0.000183(0.004434)(40 35)
= 3707.3 W / m 2 K
On the root area, the heat transfer coefficient is obtained with [7.7.10]:
1/ 4
root
1147(1147 50)(9.81)(0.0776)3 (168027)

= 0.725
0.000183(0.0224)(40 35)
= 1901.2 W / m 2 K
The parameter m in the fin efficiency is obtained with [7.7.9]:

1/ 2
1
1

+ 0
2
3707.3
m=
390(0.0003)
= 251.74
The fin efficiency fin is obtained with [7.7.8]:
fin
0.0003
tanh 251.74 0.0015 +
= 0.946
=
0.0003
251.74 0.0015 +
The surface efficiency surface is then obtained with [7.7.7]:
surface = 1
0.24919
(1 0.946 ) = 0.955
0.29845
The effective heat transfer coefficient for the finned tube is obtained from [7.7.5]:
0.04926
0.946(0.24919)
(0.955) = 1901.2
+ 3707.3
= 313.8 + 2928.2 = 3242.0
0.29845
0.29845
2
= 3242 / 0.955 = 3395 W / m K
The effective surface area is:
A eff = surfaceA total = A root + fin A fin = 0.04926 + 0.946(0.24919) = 0.285m 2

The nominal heat transfer coefficient is:
0.285
A
= 12126 W / m 2 K
nom = eff = 3395
DL
(
0
.
0254
)(
1
.
0
)
7-38
The comparable value for a plain tube of the same diameter was 1850 W/m2K. The condensate retention
angle is obtained with [7.7.14], where is equal to 0 for a rectangular fin, and the intertube fin spacing b
= 0.001 - 0.0003= 0.0007 m, so:
4(0.0061) cos(0)
= 0.499 radians or 28.6
= cos1 1
1147(9.81)(0.0007)(0.0254)
Thus about 16% of the circumference of the finned tube retains condensate between the fins.
7-39

DB 3 CH 7

Uploaded by

Copyright:

Available Formats

DB 3 CH 7

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DB 3 CH 7

Uploaded by

Copyright:

Available Formats

Engineering Data Book III

7.1 Modes of Condensation

Condensation on External Surfaces

Engineering Data Book III

Figure 7.1. Various modes of condensation. (a) Filmwise

7.2 Laminar Film Condensation on a Vertical Plate

Condensation on External Surfaces

Engineering Data Book III

Figure 7.2. Film condensation on a vertical plate.

The vapor temperature is uniform and is at its saturation temperature;

Condensation on External Surfaces

Engineering Data Book III

Figure 7.3. Integral representation of falling laminar film on a vertical plate.

Condensation on External Surfaces

Engineering Data Book III

Rearranging this expression for the local film thickness, it is

Condensation on External Surfaces

Engineering Data Book III

Eliminating by combining [7.2.13] with [7.2.3] yields the differential expression

Integrating over z gives the mean heat transfer coefficient as

Condensation on External Surfaces

Engineering Data Book III

7.2.2 Boundary Layer Analysis

Condensation on External Surfaces

Engineering Data Book III

where the Jakob number JaL is defined as

and is the first derivative of the dimensionless temperature at = 0, defined as

and the dimensionless variable is

7.3 Influence of Interfacial Phenomena on Laminar Film

Condensation on External Surfaces

Engineering Data Book III

The laminar falling film condensation heat transfer coefficient on

Re > 9.3 ArL1 / 5

This can be used as a criterion for the onset of ripples or waves;

Integrating over the film Reynolds number range from 0 to Re

Figure 7.4. Ripples and waves on a

7.3.2 Cocurrent Interfacial Vapor Shear

Condensation on External Surfaces

Engineering Data Book III

Condensation on External Surfaces

Engineering Data Book III

and transformed [7.3.11] to

Integrating, this gives

where the corresponding film Reynolds number is

where the dimensionless local heat transfer coefficient is

and the dimensionless interfacial shear stress is

Condensation on External Surfaces

Engineering Data Book III

Re crit = 1600 226i+ + 0.667 i+

For i+ > 9.04, Recrit is a fixed value, i.e.

The interfacial shear stress i is

Condensation on External Surfaces

Engineering Data Book III

where the local vapor Reynolds number is

7.3.3 Combined Effects of Gravity and Interfacial Vapor Shear

7.4 Turbulent Film Condensation on a Vertical Plate without

= 0.056 Re 0.2 PrL1 / 3

= 0.023 Re 0.25 PrL0.5