You are on page 1of 12

International Journal of Heat and Mass Transfer 54 (2011) 40784089

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


journal homepage: www.elsevier.com/locate/ijhmt

Condensation of vapor in the presence of non-condensable gas in condensers


Jun-De Li , Mohammad Saraireh, Graham Thorpe
School of Engineering and Science, Victoria University, PO Box 14428, Melbourne, Victoria 8001, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a set of differential and algebraic equations that model heat and mass transfer in con-
Received 13 July 2010 densers in which a mixture of water vapor and non-condensable gas is cooled. The model has been used
Received in revised form 24 March 2011 to predict the condensation rate, the bulk temperatures of the coolant and the gasvapor mixture, and the
Available online 5 May 2011
surface temperatures of the condenser wall. The predicted results for counter ow tube condensers are
compared with three sets of published experimental data for system in which air is the non-condensable
Keywords: gas. It is found that the predicted condensation rates and coolant bulk temperatures agree very well with
Film condensation
all the three sets of experimental data, the predicted wall temperatures agree reasonably well with the
Non-condensable gas
Mass transfer
experimental results, and the agreement between the predictions and the experimental results on the
Heat transfer bulk temperature of the airvapor mixture is excellent for one set of the experimental data, reasonable
for the second set of experimental data, but poor for the third set of experimental data. It is suggested
that the poor agreement between the predicted and measured bulk temperatures of the mixture for
the third set of experimental data arises from the experimental errors. The results from this study show
that when modeling vapor condensation in the presence of a non-condensable gas, a simple model for the
mixture channel alone may not be sufcient since neither the temperature nor the heat ux at the wall
can be assumed to be constant. The results also show that the wall temperature in the coolant channel
can be quite high, and careful modeling of the heat transfer in the coolant channel is needed in order
to achieve good agreement between the model predictions and the experimental results.
Crown Copyright 2011 Published by Elsevier Ltd. All rights reserved.

1. Introduction condensate and droplets slide downward after they attain a certain
size. It is difcult to sustain drop-wise condensation that may be
The rate of condensation of water vapor determines the opera- achieved initially on some surfaces and the process usually converts
tion of a number of engineering processes. For example, condensa- to lm condensation after some time. Therefore, it is customary to
tion plays a key role in applications such as air conditioning, assume lm condensation in the design of heat-transfer equipment
electricity generation, refrigeration, reactor safety, aerospace, [1].
desalination, and some heat exchangers. Environmental and eco- The pioneering work in the eld of condensation of pure vapor is
nomic pressures are imposing a need to design increasingly ef- due to Nusselt [2] who predicted, from a simplied theoretical anal-
cient systems. As a result, we need to enhance our understanding ysis, the heat transfer coefcient of stationary pure vapor in lm
on the physics of condensation in the presence of non-condensable condensation on a vertical at plate. Improvements and modica-
gases. tions to Nusselts theoretical solution have been made by a number
Condensation is initiated and sustained when the temperature of researchers. For example, Bromley [3] assumed a linear temper-
of a surface is maintained below the dew-point temperature of ature distribution in the liquid lm. Subsequently, Rohsenow [4]
the surrounding vapor. In the process a condensate is formed with and others considered the effect of interfacial shear stress arising
the liberation of latent heat. The process may take place in one of from the interaction of the condensate and vapor. The effects on
the two modes, namely lm condensation or drop-wise condensa- the condensation arising from the variation of longitudinal velocity
tion. In lm condensation the surface is completely wetted by the of the vapor were also accounted for.
condensate and is blanketed by a liquid lm of increasing thickness In many industrial operations, the condensing vapor is one com-
as it ows downward under the inuence of gravity. In drop-wise ponent of a gas mixture, and the other components are non-
condensation, however, the surface is not fully wetted by the condensable gases. The presence of non-condensable components
in gasvapor mixtures can greatly reduce the rate of condensation
heat transfer and degrade the performance of systems that rely
Corresponding author. Tel.: +61 3 9919 4105. on condensation heat transfer. Fundamental studies on the conden-
E-mail address: Jun-De.Li@vu.edu.au (J.D Li). sation of vapors from gasvapor mixtures were pioneered by

0017-9310/$ - see front matter Crown Copyright 2011 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2011.04.003
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4079

Nomenclature

Bm dimensionless driving force based on mass fractions x distance from the inlet
Cp specic heat (J kg1 K1) x0 initial entrance length (m)
C concentration (kg m3) y lateral position (m)
d diameter of the inner pipe (m)
D diffusivity (m2 s1) Greek symbols
f Fanning friction factor b thermal expansion coefcient
g acceleration due to gravity (m s2) d thickness of condensate lm (m)
G mixture mass ux (kg m2 s1) ds thickness of solid wall (m)
Gr Grashof number e roughness height (m)
H heat transfer coefcient (W m2 K1) l dynamic viscosity (kg m1 s1)
hv specic enthalpy of the vapor (J kg1) q density (kg m3)
hf specic enthalpy of the condensate (J kg1) - specic humidity
hfg,i latent heat of water vapor at the interface (J kg1) sg interfacial shear stress (N m2)
k thermal conductivity (W m1 K1)
K mass transfer coefcient (m s1) Subscripts
L length of the pipe (m) a air
m_ mass ow rate (kg s1) c condensate
m_ 00 mass ux (kg m2 s1) lm interface
Nu Nusselt number f lm
Pr Prandtl number g gas
P pressure (kPa) l liquid
q00 heat ux (W m2) o without suction
Qs sensible heat (W) r roughness
Ql latent heat (W) s solid, smooth
Ra Rayleigh number t developing
Re Reynolds number v vapor
ri tube inner radius (m) av airvapor
ro condenser tube outer radius (m) w water
Sc Schmidt number ww at the wall
Sh Sherwood number wb wet bulb
T temperature (C) v, i water vapor at the airliquid interface
U bulk velocity (m s1) v, b water vapor in the airvapor mixture
u velocity (m s1)
w mass fraction

Sparrow and Lin [5], Minkowycz and Sparrow [6] and Sparrow et al. The analysis by heat and mass transfer analogy in situations
[7]. They explored analytically the condensation from mixtures with condensation has been described by many researchers. Col-
containing non-condensable gases in forced convection ow along burn and Hougen [12] were the rst ones who developed a theory
a horizontal at plate. Following their studies a plethora of publica- for condensation mass transfer which was controlled by the mass
tions has been devoted to the condensation of gasvapor mixtures concentration gradient through the non-condensable layer. They
in laminar or turbulent ows along horizontal, vertical or inclined described the heat transfer process as the sum of sensible heat
plates, inside or outside tubes and in tube bundles. An example of and latent heat ows. Dehbi and Guentay [13] derived a theoretical
this genre is the work of Wang and Tu [8] who presented the solu- prediction of the heat and mass transfer in a vertical tube con-
tions which take into account the effects of a non-condensable gas denser from steam and non-condensable gas mixture. An algebraic
on the lm-wise condensation of a gasvapor mixture with turbu- equation for the lm thickness was derived. The mass and heat
lent ow in a vertical tube using the heat and mass transfer analogy. transfer analogy was invoked to deduce the condensation rate.
They found that the reduction in heat transfer due to the non-con- Che et al. [14] used the method of Colburn and Hougen [12] to ana-
densable gas was more signicant at low pressures and in mixtures lyze the heat and mass transfer process for the condensation of
owing with low Reynolds numbers. water vapor from moist air on a tube. They conducted experiments
Condensation from both pure vapor and from gasvapor mix- and found that the convective condensation heat transfer coef-
tures has been widely studied experimentally. Lebedev et al. [9] cient is 1.52 times higher than that of forced convection without
performed an experimental study of combined heat and mass condensation. They have performed several experiments to study
transfer in the condensation of vapor from humid air. They found condensation from gasvapor mixtures in vertical tubes. Munoz-
that the condensation heat transfer increases with the relative Cobo et al. [15] used an approximate method to calculate the con-
humidity and the velocity of the air. Also, Mamyoda and Asano densate lm thickness without the need to iterate to solve the
[10] studied the condensation of vapor in the presence of a non- transcendental equations. No and Park [16] and De la Rosa et al.
condensable gas on a short horizontal tube. Satisfactory agreement [17] proposed the non-iterative mechanistic models solving explic-
was obtained between their experimental results and a mathemat- itly the real interface temperature by means of algebraic equations.
ical model. Oh and Revankar [11] found from experiments that the Groff et al. [18] presented numerical solution of lm condensation
condensation heat transfer coefcient decreases with the concen- from gasvapor mixtures in vertical tubes, and solved the complete
tration of a non-condensable gas. They also developed a theoretical parabolic governing equations in both the liquid and gas phases
model for the annular lmwise condensation with non-condens- with no need for additional correlation equations for interfacial
able gas using the analogy between heat and mass transfer. heat and mass transfer. Revankar et al. [19] presented a
4080 J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089

mechanistic correlation for condensation of vapor and air mixture The airvapor mixture is characterized by an inlet temperature
in a vertical tube using data obtained from experiments and a heat Ta,j, an inlet mass ow rate of water vapor m _ v ;j , an outlet tempera-
and mass transfer analogy model. Siddique [20], Kuhn [21], and ture Ta,j+1, an outlet mass ow rate of water vapor m _ v ;j1 , and a
Tanrikut and Yesin [22] studied the condensation of steam in the mass ow rate of dry air m _ a . The condensate lm is characterized
presence of air owing downward in vertical tubes and cold water by an inlet mass ow rate m _ f ;j , an outlet mass ow rate m _ f ;j1 , a lm
owing upward inside a cooling jacket. thickness d, and an interface temperature T. The condenser wall
So far, theoretical predictions of vapor condensation and heat separating the coolant and the condensate lm is characterized
transfer in the presence of non-condensable gas have been focused by the surface wall temperatures Tfw and Tww. The cooling water
on the gas and vapor mixture. The cooling of the gasvapor mix- is characterized by an inlet temperature Tw,j, an outlet temperature
ture is usually calculated by assuming a constant wall temperature Tw,j+1, and mass ow rate m _w.
or a constant heat ux at the wall. In condensers, this wall temper-
ature or heat ux at the wall is in general not known a priori, and 2.1. Heat and mass transfer analysis
the temperature of the cooling uid (e.g. water) has normally been
used as an approximation for the wall temperature. This may be a 2.1.1. Heat balance on the airvapor mixture
valid approximation when the mass ow rate of the cooling water
is much larger than that of the gasvapor mixture or when the For steady ows, the heat balance of the airvapor mixture can
mass fraction of the water vapor in the gasvapor mixture is low. be expressed as:
However, a better approach is to solve the heat and mass balance
m _ v ;j hfg;0 C pv T a;j m
_ a C pa T a;j m _ a C pa T a;j1 m
_ v ;j1 hfg;0 C pv T a;j1
on the cooling water and the gasvapor steams simultaneously.
In this paper we will derive equations for the heat and mass bal- 2pri  ddxHav T a  T f ;i
ance of condensers involving water vapor condensation in gasva- 1
por mixture ows with water as cooling uid. The equations, in
where hfg,0 is the latent heat of the water at T = 0 C which is the ref-
combination with many theoretical models for the heat and mass
erence temperature for calculating the enthalpy. In the above equa-
transfer, will be solved numerically and the predictions will be
tion, hfg,0 + CpvTa,j is the specic enthalpy of the water vapor [23]
compared with available experimental results where experimental
and the temperature Ta is in Celsius. By re-arranging the terms,
details can be easily accessed from the literature.
dividing both sides of the equation by dx, and taking the limit, Eq.
(1) can be expressed as:
2. Mathematical model dT a dm_v
_ a C pa C pv m
m _ v  hfg;0 C pv T a
dx dx
The condensers studied in this work consist of two uid streams 2pr i  dHav T a  T f ;i 2
separated by a solid wall. For the vertical tube case the cooling
water ows vertically between the inner and outer cylinders, while The total heat transfer from the airvapor mixture to the conden-
a mixture of air (we only consider air as the non-condensable gas) sate lm consists of the sensible heat Qs and the latent heat Ql, i.e.
and water vapor ows vertically in the inside tube. The water va-
Q Q s Q l 2pr i  ddxHa T a  T f ;i m
_ v ;j  m
_ v ;j1 hv
por condenses on the inner surface of the inside tube and forms
a liquid lm that ows downwards under the inuence of gravity. 2pr i  ddxHav T a  T f ;i 3
The system is analyzed by dividing the tube length longitudi-
After simplication, we obtain
nally into a number of discrete elements of length dx as shown
in Fig 1. Four domains exist within each element, namely the hv dm_v
H av H a  4
airvapor mixture (1), the condensate lm (2), the solid wall (3) 2pr i  dT a  T f ;i dx
and the cooling water (4).
If the discrete element shown in Fig. 1 is the jth from the airva- where Hav is the overall heat transfer coefcient including both the
por inlet then we may designate the temperature of the mixture sensible and latent heat, and hv is the heat transported by the water
entering the cell as Ta,j (we are assuming that the temperature of vapor to the condensate lm as water vapor is condensed into water
the air and that of the water vapor in the mixture are the same). liquid.
The rate of condensation can be calculated from the denition
r for mass transfer coefcient, which can be expressed as
_ v m
dm _ v ;j1 2pr i  ddxKC v  C v ;i
_ v ;j  m 5
x . . . .
Ta,j, ma, mv,j mf,j Tw,j, mw where K is the mass transfer coefcient of the water vapor, Cv is the
average concentration of water vapor in the airvapor mixture, Cv,i
is the water vapor concentration at the interface between the mix-
ture and the condensation, and depends on interface temperature
T. By using the bulk density q as the representative density across
(1) (2) (3) (4) the mixture channel, the above equation can be written as
Tf,w
dx dm_v
2pr i  dK qwv  wv ;i 6
dx
Tf,i
where wv is the mass fraction of the water vapor in the airvapor
mixture, and it is related to the specic humidity, -, by
_v
m _v
m 1
wv 1 7
. . . . m_ _ am
m _v 1-
Ta,j+1, ma, mv,j+1 mf,j+1 Tw,j+1, mw
_ is the total mass ow rate of the airvapor mixture. We as-
and m
Fig. 1. Heat and mass balance of a parallel-ow condenser. sume that the air, the water vapor and their mixture are all ideal
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4081

gases and ideal gas law can be applied to determine the respective where Hw is the convective heat transfer coefcient of the water in
densities. the coolant channel. An energy balance on the cooling water shown
in Fig. 1 results in the following expression
2.1.2. Force and momentum balance on the condensate lm
By considering the balance between the weight of the uid ele- dT w
_w
C pw m 2pr o Hw T w;w  T w 14
ment, the buoyancy force and the viscous shear force in the small dx
region 2 shown in Fig. 1, the velocity gradient of the condensate in
i.e. the heat transfer into the coolant will cause the water tempera-
the lm can be derived as
ture to change. In deriving Eq. (14), we have assumed that the outer
du ql  qg g sg wall of the cooling channel is well insulated and there is no heat
d  y 8 loss from this outer wall.
dy ll ll
The above equations are derived for condensers of co-current
where ql is the density of the condensate, qg is the density of the ows. For condensers of counter ows, these equations remain va-
airvapor mixture at the interface, lf is the dynamic viscosity of lid but the mass ow rate of the cooling water assumes a negative
the condensate, and sg is the shear stress at the interface between value. In this case, we are assuming that the cooling water in Fig. 1
the airvapor mixture and the condensate lm. Here we have as- ows upwards which is in the opposite direction to that of the air
sumed that the condensate lm is very thin and the direction of vapor mixture.
the shear stress is the same as that of the buoyancy force. Nusselts
theory neglects interfacial shear. The velocity distribution in the 2.2. Modeling of heat and mass transfer
condensate lm can then be derived as
 
ql  qv g 1 sg 2.2.1. Airvapor mixture ow
u dy  r i  y2  r2i y  r i 9 The heat transfer coefcients Ha and Hw for the airvapor mix-
ll 2 ll
ture can be obtained from the correlations for Nusselt number in
Here we have assumed that the velocity of the condensate is zero at channel ows
the inner surface of the condenser wall (ri). The mass ow rate in-
crease in the condensate lm as a result of water vapor condensing Hd
Nu f Red; Pr 15
from the airvapor mixture can be calculated from k
2  3 where Pr is the Prantle number and
dm_v ql  qg g  2 5 3  sg  dd
2pql 4 ri d  d r d  d2 5 10 qUd
dx ll 6 ll i dx Red
l
Combining Eqs. (6) and (8) enables the condensate lm thickness d
where d can be the diameter of the inner pipe for airvapor mixture.
to be determined. The heat transfer balance in the condensate lm
The mass transfer coefcient K is obtained by using correlation for
can be calculated after the manner of Nusselt [2], i.e.
Sherwood number
kf T f ;i  T f ;w 1 _ f hf
dm
Hav T a  T f ;i Kd
r i  d lnri =r i  d 2pri  d dx 11 Sh f Red ; Sc 16
D
T f ;i T f ;w if d 0
Here Sc is the Schmidt number and D is the molecular diffusivity
where hf is the specic enthalpy of the condensate. The last term in that can be calculated as a function of temperature and pressure
the above equation takes into account the energy change of the con- as that recommended by Rao et al. [24]
densate in the small element dx as shown in Fig. 1. This change is
due to the change in m_ f and the change of the condensate temper- 1:87  1010  T 2:072
D 17
ature. This term has been neglected in previous studies and is in- P
cluded in this work to form part of the differential and algebraic Here P is the total pressure in the airvapor mixture. The saturation
equations. vapor pressure of water is a function of temperature and can be
determined from the expression given by Rao et al. [24], namely
2.1.3. Heat transfer at the solid wall
Under the assumption that the temperature in the condensate Pv exp77:3450 0:0057T  7235=T=T 8:2 18
varies linearly across the lm, the heat conduction at the interface where T is in Kelvin. The specic humidity of airvapor mixture can
between the solid wall and the condensate lm can be calculated be calculated as
as kf(T  Tfw)/ln[ri/(ri  d)]. This heat conduction through the con-
densate lm and that through the solid wall are equal to each mv 0:622Pv
- 19
other, hence ma P  Pv

kf T f ;i  T f ;w ks T f ;w  T w;w At high mass fractions of water vapor in the mixture, it is important


12
lnr i =ri  d lnro =r i to consider the suction effect on heat and mass transfer, which
where ks and kf are the thermal conductivities of the solid wall and arises as a result of steep temperature and concentration gradients
the condensate lm, respectively. near the interface. Kays and Moffat [25] obtained the correlations
for the suction effects, which are given as following:
  00  1  1
2.1.4. Heat transfer in the cooling channel m_ v Re Pr Gmix
Nux exp 1 20
The heat owing across the solid wall enters the stream of cool- Gmix Nuo;x m_ 00v Re Pr
ing water, and this is balanced by the heat transfer in the coolant
channel. Hence where m _ 00v is the condensate mass ux at the interface between the
airvapor mixture and the condensate lm and Gmix is the mass ux
ks T f ;w  T w;w
Hw T w:w  T w 13 of the mixture. Using the analogy between the heat and mass trans-
ro lnr o =r i
fer, the Sherwood number is expressed as
4082 J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089

  00  1  1
m_ v Re Sc Gmix For rough walls, the friction factor is given by Haaland [30] as
Shx exp 1 21 "
Gmix Sho;x m_ v Re Sc
00   #
1 6:9 e=d 1:11
After taking into account the suction effect on the rate of mass
1:8log10 31
fr1=2 Re 3:7
transfer at the gas and condensate lm interface, the mass ux at
the interface can be calculated according to Bucci et al. [26] where Re is the Reynolds number for the airvapor mixture ow.
_ 00v qK ln1 Bm
m 22 The friction factor fr increases as the surface roughness is increased.
A conservative assumption of roughness height e being equal to 0.5
where d is chosen, as suggested by Siddique [20]. In this paper, we apply
wv ;i  wv ;b the roughness corrections to the condensate lm starting from
Bm 23 x = 0 and we assuming that all the ows are turbulent.
1  wv ;i
The interfacial shear stress is calculated as
Here wv,i is the mass fraction of water vapor at the gasliquid inter-
face and wv,b is the bulk mass fraction of water vapor in the airva- U  ufb 2
por mixture. Eqs. (22) and (23) are more general than Eq. (6) since
sg f qg 32
2
they account not only the suction effect but also the density varia-
where U is the bulk velocity of the airvapor mixture, ufb is the
tion of the mixture due to the concentration increase of non-con-
mean velocity of the water liquid in the condensate lm, and qg is
densable gas near the interface between the condensate lm and
the density of the airvapor mixture.
the airvapor mixture. In Eqs. (20) and (21), Nuo,x and Sho,x denote
the respective local Nusselt and Sherwood numbers after taking
into account the developing ow in the thermal entrance region 2.2.2. Coolant ow
as suggested by Reynolds et al. [27] Eqs. (15), (24), and (26) are also applicable to model heat trans-
" # fer for the cooling channel.
0:81 7  104 Re3=2 Many experimental studies of condensation from gasvapor
Nuo;x Nuo 1 24
x x0 =d mixtures are based on water owing as a coolant in an annulus
in tube condensers. The Reynolds number for the coolant is calcu-
Using the heat and mass transfer analogy, the Sherwood number lated using the hydraulic diameter for the coolant owing between
including the effect from the thermal entrance region can be writ- the two cylinders. It has been found experimentally that the tem-
ten as: perature increase of the coolant may be small but the wall temper-
" # ature of the inner pipe can be high. This could be the results of low
0:81 7  104 Re3=2
Sho;x Sho 1 25 heat transfer on the cooling side. In this case, the heat transfer in
x x0 =d
the coolant channel will be affected by buoyancy. The buoyancy ef-
where x0 is an initial entrance length. fect is conned within an enclosed space and can be modeled fol-
The Nusselt and Sherwood numbers for fully developed ows lowing the suggestion by Kays et al. [34] as
without accounting for the suction effect can be calculated accord-
Nuw 0:046 Ra1=3 33
ing to Holman [28]
where Ra is the Rayleigh number and can be calculated as
Nuo 1:04  0:0395 Re0:75 Pr 1=3 26
gbq2 C p d3o
Sho 1:04  0:0395 Re 0:75
Sc 1=3
27 Ra Gr Pr T w;w  T w 34
lk
The subscript o represents fully developed ows. The above equa-
and do is the channel width of the coolant ow. The combined heat
tions without the factor 1.04 are those for constant wall tempera-
transfer coefcient in the cooling ow is then calculated as a com-
ture. For constant heat ux at the wall, the factor should be
bination of the forced and natural convection
included. In this paper, this factor is included even though it is close
to unity. Nucombined Nunforced  Nunnatural 1=n 35
As condensation proceeds along the tube, the annular ow of
the condensate changes from laminar to turbulent ow, and the with n = 3 as recommended by Churchill [35] for vertical surfaces.
condensate lm surface becomes rough and wavy. This roughness The plus sign is invoked when the forced and natural convection
effect is modeled using the correlations suggested by Norris [29] assisting each other and the minus sign is used when opposing
for the heat transfer over rough surfaces ows caused by forced convective and natural convective. The mod-
 nh els using Eqs. (33)(35) assume that the coolant stays as liquid
fr phase. In some applications, the temperature in the coolant reaches
Nuo;r Nuo;s 28
fs high enough so that boiling exists. In these cases, different models
Using the heat and mass transfer analogy, the Sherwood number is should be used to model the heat transfer in the coolant.
given by An important step in the calculation is the denition of physical
properties of the uids. The thermal conductivity, density, viscos-
 nm
fr ity, and specic heat for the gas mixture, condensate and the cool-
Sho;r Sho;s 29 ant are required for the calculation. Given the large temperature
fs
variation that can be expected in the condensation of water vapor
where the subscripts r and s refer to the rough and smooth walls, in the presence of non-condensable gas, the assumption of con-
nh = 0.68 Pr0.215, nm = 0.68 Sc0.215 and fr/fs represents the ratio of stant uid properties may give rise to errors, because the uid
the Fanning friction factor for a rough wall to that for a smooth wall properties vary with temperature. These variations of uid proper-
as suggested by Norris [29]. For smooth walls the friction factor is ties will then cause a variation of velocity and temperature
calculated from throughout the boundary layer or over the ow cross section of
the duct or tube. In this paper, the physical properties for liquid
fs 0:316  Re0:25 30 water, condensation lm, dry air, and water vapor were calculated
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4083

by using curve ttings which are functions of respective bulk hv hfg;i 36


temperatures.
where hfg,i is the latent heat of vaporization at the temperature Tf,i.
This assumes that the energy transported by the water vapor to the
3. Numerical solution procedures
condensate lm is that due to the condensation of saturated water
vapor into saturated liquid. The second method is given by Baehr
The full system of seven differential and algebraic equations
and Stephan [32]
describing the heat and mass transfer that occur in condensation
from airvapor mixture in a tube condenser is given by Eqs. (2), hv C pv T a  T f ;i hgh;i 0:68C pL T f ;i  T f ;w 37
(6), (10)(14). These equations can be used to solve for the seven
_ v , d, T, Tfw, Tww, Tw. The differential equations were where Cpv is the specic heat of water vapor and CpL is the specic
unknowns, Ta, m
heat of liquid water. In Eq. (37), the rst term on the right hand side
discretized using the rst order forward Euler method, and the cal-
is that due to superheating and the third term is that due to sub-
culations were performed using the fourth-order RungeKutta
cooling.
method to solve the differential and algebraic equations to obtain
By examining Fig. 1, it can be seen that as the water vapor is
the mixture temperature, water temperature, wall temperatures,
transported from the airvapor mixture to the condensate lm,
and condensation rate. The boundary conditions are: the inlet
the vapor carries with it the energy which includes the latent heat
airvapor temperature, inlet watervapor mass ow rate, inlet
and the sensible heat of the liquid water as it crosses the interface
cooling water temperature, cooling water mass ow rate, air mass
(because of the mass transfer). One way to represent this is
ow rate and inlet pressure. A computer program has been written
in MATLAB to solve these equations. hv hfg;0 C v T f ;i 38
In this paper, the numerical solutions were performed for vapor
condensation in cylindrical systems in order to compare the pre- However, it was found from numerical experiments that Eqs. (37)
dictions with available experimental results from the literature. and (38) can produce temperature jumps near the airvapor inlet
All the experimental results to be discussed later applied to coun- for some situations (which we think is physically unrealistic) and
ter ows between the coolant and the airvapor mixture, and a more appropriate representation of this energy can be expressed
therefore the mass ow rate for the coolant m _ w was taken as neg- as
ative in order to apply the differential equations presented above. hv hfg;0 C pv T a 39
In solving the differential equations, the location of the airva-
por mixture inlet was taken as x = 0 and that of the exits was taken This shows that all the energy from the water vapor in the bulk of
as x = L. The x0 in Eqs. (24) and (25) was taken as 0.5d to avoid the airvapor mixture has been transported into the condensate
divergence. A marching method was used starting from x = 0. We as the water vapor is condensed while Eq. (38) shows that some
assumed that at x = 0, the thickness of the condensate lm d = 0, of the energy (C pv T a  T f ;i ) is left behind in the airvapor mixture.
and this results in T = Tfw. When d = 0 at x = 0, Eq. (10) cannot be Also, the enthalpy value calculated from Eq. (39) is very close to the
used to solve for the lm thickness at x = dx. To overcome this dif- enthalpy of the superheated steam as that given in http://
culty, Eq. (10) was integrated and an algebraic equation was www.engineeringtoolbox.com/steam-vapor-enthalpy-d_160.html
solved for calculating the lm thickness at x = dx. At x > dx, Eq. (access date: 6/2010). One consequence of Eq. (39) is that when it is
(10) was used. This equation is easier to solve than the integrated combined with Eqs. (2) and (4), it shows that the energy balance for
algebraic equation. the airvapor mixture involves only the sensible heat, the latent
For the counter ow condensers considered here, boundary heat is released at the interface when the phase change occurs. In
conditions at both x = 0 (for the airvapor mixture) and x = L (for the results presented later, only results using Eq. (39) will be
the coolant) are given. Because of this, a straight forward marching considered.
method cannot be used. Instead, a shooting scheme in combination In solving Eq. (11) for the energy balance in the condensate lm,
with the marching method was used. At x = 0, the exit temperature the average temperature across the lm is required. This was cal-
of the coolant, Tw,1, was estimated to initiate the solution proce- culated by assuming that the temperature varies linearly across
dure and the solutions were marched to x = L. The predicted Tw,N the lm as that used by Bromley [3].
was compared with the boundary condition Tw at x = L. If the rela-
tive error between the predicted Tw,N and Tw at x = L (a boundary 4. Comparison of model predictions with existing experimental
condition) is larger than 107, a newly estimated Tw,1 was used results
and the solution was repeated. The convergence of the solutions
was established by comparing the predicted total condensation The present theoretical work is validated against the experi-
rates from using two discetizations with N = 1000 and N = 10,000, mental data of Siddique [20], Kuhn [21] and Tanrikut and Yesin
respectively. It was found that the difference was less than 0.01%. [22]. The test sections of all these experiments were circular, ver-
In the results presented in this work, only the results from tical, metallic tubes, surrounded by annular jackets through which
N = 10,000 are given. a liquid coolant (liquid water) owed. The gasvapor mixture o-
In solving the differential and algebraic equations, the physical wed downward in the tube while the coolant in the jacket owed
properties such as specic heat, dynamic viscosity, density, and upward. Each of these experimental studies includes a large num-
diffusivity for water liquid, dry air and water vapor were all al- ber of tests, and several representative tests of a wild range of con-
lowed to vary with temperature and calculated from the curve t- ditions were selected from each experimental study, as
ted correlations using the local mean temperature at each xj. The summarized in Table 1.
thermal conductivity and dynamic viscosity of the airvapor mix- The experiments of Siddique [20] were conducted using air
ture were calculated according to Bird et al. [31], and the specic steam mixture with inlet temperatures of 100, 120 and 140 C,
heat of the airvapor mixture was calculated using the mass respectively, inlet air mass fractions from 8% to 35%, and inlet mix-
weighted average. ture pressure from 0.11 MPa to 0.48 MPa. Siddique [20] measured
In solving for the airvapor mixture temperature Ta, the energy the respective temperatures of the cooling water, the gasvapor
transported by the water vapor into the condensate lm, hv, needs mixture on the centerline and the inside and outside walls of the
to be calculated. In the literature, several methods of calculating tube at nine stations spaced 30.5 cm apart along the length of
this energy are available. One is the condenser (L = 2.54 m). To measure the bulk temperature of
4084 J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089

Table 1
Experimental runs used for comparison with predictions.

Reference Run number T w;in  C T mixt;in  C umixt (m/s) wa,in Pressure (kPa)

Siddique [20] 7 7 120 1.378 0.080 208.6


11 7.9 120 1.388 0.33 259.6
13 12.7 140 0.695 0.11 389.2
17 7.7 139.9 0.694 0.34 474.5
26 8 119.9 3.054 0.22 221.2
Kuhn [21] 2.112 26.9 133.4 4.77 0.34 408.4
2.28 25.6 129.1 2.54 0.146 115.7
3.54 22.7 132 6 0.35 400.5
4.55 22.7 140.3 14 0.38 503.4
Tanrikut and Yesin [22] 2.3.1 16 160.9 18.22 0.098 296.9
3.3.1 14.4 171.7 20.46 0.189 290.1
3.4.1 14.4 163.4 18.30 0.193 390
4.3.1 14.6 158.9 23.07 0.279 316
5.4.1 14.4 152.7 17.04 0.369 394

the coolant in the annulus, Siddique [20] used small air bubbles to 6
mix with the coolant to achieve a uniform temperature prole Predicted
across the coolant channel. 5

Condensation rate (g/s)


The experiments of Kuhn [21] were conducted for pure steam, Experimental
steamair mixture and steamhelium mixture. In this paper, we 4
compare the data from Kuhn [21] only for the steamair mixture
experiments. In Kuhn [21], the coolant bulk temperature was not 3
directly measured rather it was estimated by measuring the tem-
peratures at the inner and outer walls of the annulus, and by calcu- 2
lating the temperature difference ratio (dened as a shape factor)
numerically. Kuhn [21] also investigated the effect of turbulent 1
condensate lm on the heat transfer by injecting liquid water using
lm distributors near the mixture inlet. 0
7 11 13 17 26
The experimental results of Tanrikut and Yesin [22] were ob- Run
tained for the condensation in the presence of air for the pressure
range of 1.85.5 bars, Rev = 45,00094,000, and non-condensable Fig. 2. Comparison of the predicted condensation rates with the experimental
results of Siddique [20], run number 11, 13, 17, and 26.
gas mass fraction X = 052%. They found that inlet superheating
of the steam has little effect on the heat ux, and the presence of
air has a large effect on reducing the heat transfer in comparison
7
with the condensation of pure vapor (X = 0).
Predicted
In all the three experiments [2022], the local heat ux was 6
estimated using Experimental
Condensation rate (g/s)

5
_ c C p dT c
m
q00 x 40
pd dL 4
where Tc is the measured or estimated bulk temperature of the
3
coolant, and the gradient dTc/dL was estimated from a curve t as
a function of condenser length. The condensation rates were then
2
estimated using
1
Q_
_ cond
m 41
hfg 0
2.1-12 2.2-8 3.5-4 4.5-5
Here m_ cond is the total condensation rate (or the total condensation Run
rate as collected), Q_ is the total heat transfer rate across the con-
Fig. 3. Comparison of the predicted condensation rates with the experimental
denser wall, and hfg is the latent heat of condensation and was cal- results of Kuhn [21], run number 2.112, 2.28, 3.54, 4.55.
culated using the averaged wall temperature of the condenser. In
using Eq. (41) to calculate the total condensation rate, the contribu-
tion of the sensible heat transfer in Q_ was neglected. rates of the water vapor in the airvapor mixture tube. Because of
this, the predicted condensation rates are the true condensation
4.1. Condensation rate and are due to the mass transfer across the boundary layers
in the channel only. This is in contrast to the condensation rates
The overall condensation rates computed from the present from the three experiments given in Table 1, which were estimated
model for the different experiments as listed in Table 1 are shown from the overall heat transfer rates.
in Figs. 24, respectively. The predicted overall condensation rates It can be seen from Figs. 24, the predicted and experimental
were calculated by summing dm _ v along the condenser tube and results obtained from different sources are in excellent agreement,
these can also be calculated by using the inlet and outlet mass ow with the maximum relative error being less than 6% between the
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4085

12 120
Predicted
10
Experimental
Condensation rate (g/s)

100

8
Predicted Ta
80

Temperature (C)
6 Experiment Ta
Predicted Tw
4 60 Experiment Tw
Predicted Tfw
2
40 Experiment Tfw
Predicted Tww
0
2.3.1 3.3.1 3.4.1 4.3.1 5.4.1
20
Run

Fig. 4. Comparison of the predicted condensation rates with the experimental


0
results of Tanrikut and Yesin [22], run number 2.3.1, 3.3.1, 3.4.1, 4.3.1, 5.4.1. 0 0.5 1 1.5 2 2.5 3
Distance from inlet (m)
predictions and the experimental results of Siddique [20], and less
Fig. 6. Comparison of the predicted temperatures of cooling water, wall and air
than 4% between the predictions and the experimental results of
vapor mixture with those measured by Siddique [20], run number 11.
Tanrikut and Yesin [22], and less than 8.5% between the predic-
tions and the experimental results of Kuhn [21]. The results in Figs.
24 also support the assumption made in deriving the condensa- 140
tion rates in the experiments, that the sensible heat transfer is Predicted T a
small in comparison with the latent heat transfer. Experimental Ta
120
Figs. 24 show that the errors between the predicted condensa-
tion rates and those from the experiments are random and no con- Predicted T w
sistent trend can be detected. It is expected that a low mass 100 Experimental Tw
Temperature (C)

fraction of water vapor in the airvapor mixture would result in Predicted T fw


larger errors since the contribution from the sensible heat transfer 80
Experimental Tfw
to the overall heat transfer would be higher.
Predicted T ww
60
4.2. Temperature
40
Predicted axial variations of the temperatures for airvapor
mixture, inside and outside condenser walls, and cooling water
20
temperatures are compared with the experimental data of Siddi-
que [20] and are shown in Figs. 57. These gures show that the
predicted temperatures agree very well with the experimental data 0
0 0.5 1 1.5 2 2.5 3
for all the three runs. In the experimental results of Siddique [20], Distance from inlet (m)
although both the inner and outer wall temperatures were mea-
sured, only results for inner wall temperature were presented in Fig. 7. Comparison of the predicted temperatures of cooling water, water and air
vapor mixture with those measured by Siddique [20], run number 17.

120 Siddique [20]. Siddique [20] measured the inner wall temperature
by using thermocouples inserted in drilled holes and the physical
locations of the thermocouples were less than 0.2 mm from the in-
100 ner surface of the condenser.
Figs. 57 show that the predicted centreline temperature of the
Predicted Ta
airvapor mixture and the coolant bulk temperature agree very
80
Temperature (C)

Experimental Ta well with the experimental data of Siddique [20]. The inner wall
Predicted Tw temperature measured by Siddique [20] in general falls between
60 the predicted Tfw and Tww as would be expected. This is remarkable
Experimental Tw
given that the wall temperature falls sharply from the airvapor
Predicted T fw mixture inlet to the coolant inlet.
40 Experimental Tfw Figs. 57 show that the wall temperatures in the experiments of
Siddique [20] are far from constant and very different from the
Predicted Tww
temperatures of the coolant. This shows that in modeling vapor
20
condensation in the presence of non-condensable gas, the wall
temperature cannot in general be assumed as constant. Because
0 of this large difference between the wall temperature and the bulk
0 0.5 1 1.5 2 2.5 3
temperature of the coolant, it is expected that buoyancy effects in
Distance from inlet (m)
the coolant channel will be large and their effect on the convective
Fig. 5. Comparison of the predicted temperatures of cooling water, wall and air heat transfer needs to be taken into account in order to achieve
vapor mixture with the experimental results of Siddique [20], run number 7. good agreement between the predictions and experimental results.
4086 J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089

90 140
Predicted Ta with buoyancy
Calculation
Predicted Ta without buoyancy
80 Siddique Exp. Run # 26
Experimental Ta
120
70 Predicted Tfw with buoyancy
Predicted Tfw without buoyancy
Heat Flux (kW/m )
2

60 100 Experimental T

Temperature (C)
fw
Predicted Tww with buoyancy
50 Predicted Tww without buoyancy
80 Experimental Tww
40 Predicted Tw with buoyancy
Predicted Tw without buoyancy
30 60 Experimental T
w

20
40
10

0 20
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5
Distance from inlet (m) Distance from inlet (m)

Fig. 8. Comparison of the predicted heat ux with the experimental results of Fig. 10. Comparison of the predicted temperatures of cooling water, airvapor
Siddique [20], run number 26. mixture, inner wall and outer wall with the experimental results of Kuhn [21], run
2.28.

Fig. 8 shows the comparison between the heat ux estimated by


the theoretical model and that from the experimental data of Sid- 140
Predicted Ta with buoyancy
dique [20] (Run #26). The predicted heat ux was obtained using Predicted Ta without buoyancy
Eq. (40). Instead of using differentiation of a curve t from the pre- Experimental Ta
120
dicted Tw, nite differences were used to approximate dTw/dL using Predicted Tfw with buoyancy
Tw obtained from the numerical solutions. Fig. 8 shows that the Predicted Tfw without buoyancy
agreement between the model predictions and the experimental 100 Experimental Tfw
Temperature (C)

data is very good, given the fact that the experimental data were Predicted Tww with buoyancy
obtained by the differentiation of a curve t. This close agreement Predicted Tww without buoyancy
in the heat ux is a reection of the remarkable agreement of the 80 Experimental Tww
coolant bulk temperature between the predictions and the experi- Predicted Tw with buoyancy
mental results. The predicted heat ux shows a sharp decrease at Predicted Tw without buoyancy
60 Experimental Tw
x = 0. This is because we have included the effect from the thermal
entrance region in our model (Eqs. (22) and (23)). This sharp de-
crease in the heat ux near x = 0 is not shown in the experimental 40
data. This is because, to capture this sharp decrease in the heat ux
experimentally, many more experimental points need to be mea-
sured near the inlet of the airvapor mixture. 20
0 0.5 1 1.5 2 2.5
Figs. 912 show the comparison of the predictions and the Distance from inlet (m)
experimental results of Kuhn [21] for the temperatures of cooling
Fig. 11. Comparison of the predicted temperatures of cooling water, airvapor
mixture, inner wall and outer wall with the experimental results of Kuhn [21], run
3.54.
140
Predicted Ta with buoyancy
Predicted Ta without buoyancy
Experimental Ta water, airvapor mixture, and the inner and outer walls. The con-
120
Predicted Tfw with buoyancy denser tube in the experiments of Kuhn [21] was 2.418 m long but
Predicted T without buoyancy
fw
the experimental results were presented only for distances up to
100 Experimental T 1.5 m from the inlet of the airvapor mixture. It can be seen from
Temperature (C)

fw
Predicted Tww with buoyancy these gures that the predicted temperatures of the cooling water
Predicted Tww without buoyancy agree very well with the experimental results, the predicted wall
80 Experimental T
ww temperatures of the inner and outer surfaces agree with experi-
Predicted T with buoyancy
w mental results reasonably well, and the predicted centreline tem-
Predicted Tw without buoyancy
60 peratures for the airvapor mixture are in general less than the
Experimental Tw
experimental results. Overall, it can be seen from these gures that
the model predictions and the experimental data are in good
40 agreement. Figs. 912 again show that the difference between
the wall temperature and the bulk temperature of the coolant is
20
large for all the experimental data of Kuhn [21]. This large differ-
0 0.5 1 1.5 2 2.5 ence is due to the large thermal resistance to the heat transfer from
Distance from inlet (m) the cooling channel. The Rayleigh numbers calculated using Eq.
(34) for the experimental data shown in Figs. 912 are in general
Fig. 9. Comparison of the predicted temperatures of cooling water, airvapor
mixture, inner wall and outer wall with the experimental results of Kuhn [21], run above 3  106. This shows that the ow induced by the buoyancy
2.112. effect in the cooling channel is fully turbulent [34], and will greatly
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4087

160 180
Predicted Ta with buoyancy
Predicted Ta without buoyancy
140 Experimental Ta 160
Predicted Tfw with buoyancy
Predicted Tfw without buoyancy
120 140
Experimental Tfw
Temperature (C)

Predicted Tww with buoyancy


100 Predicted Tww without buoyancy 120

Temperature (C)
Experimental Tww Predicted T a
Predicted Tw with buoyancy
80 100 Experimental Ta
Predicted Tw without buoyancy
Experimental Tw Predicted T wb
60 80 Predicted T w
Experimental Tw
40 60
Predicted T fw
20 40
Experimental Tfw
0 0.5 1 1.5 2 2.5
Distance from inlet (m) Predicted T ww
20
Fig. 12. Comparison of the predicted temperatures of cooling water, mixture, inner
wall and outer wall with the experimental results of Kuhn [21], run 4.55.
0
0 0.5 1 1.5 2 2.5
Distance from inlet (m)
enhance or dominate the heat transfer in the coolant channel as
can be seen from Eq. (35). Figs. 912 also show the results without Fig. 14. Comparison of the predicted temperatures of cooling water, wall and air
including the buoyancy effect in the cooling channel. The gures vapor mixture with the experimental results of Tanrikut and Yesin [22], run 2.3.1.

show that the differences in the predicted coolant bulk tempera-


tures and the mixture bulk temperatures with and without
buoyancy effects are small, the differences in predicted wall tem- temperature of the cooling water. The gures show that the agree-
peratures can be more than 20% (relative to the coolant tempera- ment between the predicted bulk coolant temperatures and that of
ture), and the results including the buoyancy effect agree with the experiments are from good to reasonable with the experimen-
experimental results much better than those without including tal results exhibiting a large degree of scatter.
the buoyancy effect. Tanrikut and Yesin [22] measured the inner wall temperatures
Fig. 13 shows the comparison of the predicted heat ux with the using thermocouples embedded in the solid wall and the tips of
experimental results of Kuhn for the test 2.112. The gure shows the thermocouples were about 0.5 mm from the inner surface of
that the predictions agree with the experimental data very well. the inside tube. Figs. 1417 show that, in general, the measured
Again, the very close agreement in the heat ux as shown in wall temperatures fall between Tfw and Tww as expected except
Fig. 13 is a reection of the remarkable agreement in the coolant those shown in Fig. 14 which are slightly above Tfw.
bulk temperature between the predictions and the experimental However, Figs. 1417 show that the agreement in the bulk tem-
results. perature of the airvapor mixture between the predictions and the
Figs. 1417 show the comparison of the model predictions with experimental results of Tanrikut and Yesin [22] is poor with the
the experimental results of Tanrikut and Yesin [22] for the bulk experimental results being well below the predicted results.
temperatures of the airvapor mixture, the temperatures of the
inner and outer surfaces of the condenser wall, and the bulk

180

110
Calculation 160

100 Kuhn [21] Exp. Run 2.1-12


140

90
120
Temperature (C)
Heat Flux (kW/m2)

80
100 Predicted Ta

70 Experimental Ta
80 Predicted Tw
60 Experimental Tw
60 Predicte Tfw
50 Experimental Tfw
40 Predicted Tww
40 Predicted Twb
20

30
0 0.5 1 1.5 2 2.5 0
Distance from inlet (m) 0 0.5 1 1.5 2 2.5
Distance from inlet (m)
Fig. 13. Comparison of predicted heat ux with the experiment results of Kuhn
[21], run 2.112. Fig. 15. Comparison of the predicted temperatures of cooling water, wall and air
vapor mixture with the experimental results of Tanrikut and Yesin [22], run 3.3.1.
4088 J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089

 1n
180 T 1  T wb Pr hg
42
ln1 Bm Sc Cp
160
where T1 is the dry bulb temperature (Ta), Twb is the wet bulb tem-
140 perature, n = 1/3, hg is the specic enthalpy carried by the water
when it is evaporated at the wet bulb temperature, and
120 w0  w1
Temperature (C)

Bm 43
1  w0
100 Predicted Ta
Experimental Ta Here w0 is the mass fraction of the water vapor at the interface and
80 Predicted Twb w1 is the mass fraction of the water vapor at the far eld. Eq. (43) is
Predicted Tw similar to Eq. (23). The values for Pr and Sc are taken as 0.74 and
60 Experimental Tw 0.58 as suggested by Bird et al. [31]. In Figs. 1417, the predicted
Predicted Tfw wet bulb temperatures are also shown. It can be seen from the g-
40 Experimental Tfw ures, only for run 2.3.1 of Tanrikut and Yesin [22], that the predicted
Predicted Tww
wet bulb temperatures using Eq. (42) agree reasonably well with
20 the estimated centreline temperatures from the experiments.

0 5. Discussion and conclusions


0 0.5 1 1.5 2 2.5
Distance from inlet (m)
A mathematical model has been developed to study the con-
Fig. 16. Comparison of the predicted temperatures of cooling water, wall and air densation of water vapor in the presence of non-condensable gas
vapor mixture with the experimental results of Tanrikut and Yesin [22], run 3.4.1.
in condensers. The model includes the heat and mass transfer in
the mixture channel, the heat transfer in the condensate lm, con-
According to Tanrikut and Yesin [33], their central temperatures denser wall and the coolant channel. Previous models include
along the condenser tube were predicted from the GibbsDalton mainly the mixture channel and the condensate lm. The model
law by using calculated local air mass fraction from the energy bal- developed is quite general and can be used to predict the heat
ance since the measured central temperatures were lower than and mass transfers for both the co-current and counter ow con-
those predicted. They suggested that this can be attributed to the densers. A numerical scheme is presented to solve the differential
sub-cooling of the centreline possibly due to the detachment of li- and algebraic equations. In this work only results for the vapor
quid droplets from the condensate lm towards the centreline of mixture ows downward and the cold water owing upward are
the tube as reported in Tanrikut and Yesin [22]. Tanrikut and Yesin compared with the experimental data since most of the experi-
[22] also found that the predicted central temperatures inside the mental data available in the literature are for counter ow con-
condenser tube using the GibbsDalton law were in general less densers. The predictions from the model are compared with the
than the estimated saturation temperatures. experimental results of Siddique [20], Kuhn [21], and Tanrikut
By assuming that the tips of the thermocouples in the centreline and Yesin [22], all conducted in cylindrical condensers. It is found
of the condenser tube are covered by liquid water, it is possible that the predicted condensation rates and the bulk temperatures of
that the measured centreline temperatures are close to the wet the coolant agree very well with the experimental results obtained
bulb temperature rather than the dry bulb temperature. Following by all three groups of researchers, the predicted wall temperatures
Bird et al. [31], it can be shown that the following relationship ex- of the condenser agree reasonably well with the experimental re-
ists between the wet bulb temperature and the dry bulb sults, and the predicted bulk temperatures of the airvapor mix-
temperature, ture agree very well with the experimental results of Siddique
[20], reasonably well with the experimental results of Kuhn [21]
and poorly with the experimental results of Tanrikut and Yesin
160 [22]. Tanrikut and Yesin [22] presented the estimated centreline
temperature for the airvapor mixture because they found that
their measured temperature was not only below that estimated
140
from GibbsDalton law but also the saturation temperature. By
assuming that the thermocouples were covered by liquid water
120
in the experiments of Tanrikut and Yesin [22], it is possible that
the temperatures measured by Tanrikut and Yesin [22] are close
Temperature (C)

100
Predicted T to the wet bulb temperature. A relationship between the dry bulb
a
Experimental T temperature and the wet bulb temperature (Eq. (42)) was derived
80 a
Predicted T
and used to estimate the wet bulb temperature using the predicted
wb
Predicted T
bulk temperature of the airvapor mixture and the mass fraction of
60 w
Experimental T
the water vapor. It was found that only for one set of the experi-
w
Predicted T
mental results (run 2.3.1), did the predicted wet bulb temperature
fw
40
Experimental T
agree reasonably well with the estimated centerline temperature
w
of Tanrikut and Yesin [22].
Predicted T
ww
20 The model predictions and all the experimental results pre-
sented here show that the wall temperature of the condenser can
0 be much higher than the bulk temperature of the coolant. This
0 0.5 1 1.5 2 2.5
shows that it will be very difcult to select a constant wall temper-
Distance from inlet (m)
ature (or constant heat ux) in modeling condensation from heat
Fig. 17. Comparison of the predicted temperatures of cooling water, wall and air and mass transfer in the mixture channel alone, and the condenser
vapor mixture with the experimental results of Tanrikut and Yesin [22], run 5.4.1. (including the gasvapor mixture channel, the condenser wall and
J.D. Li et al. / International Journal of Heat and Mass Transfer 54 (2011) 40784089 4089

the coolant channel) needs to be modeled as a whole system. Also, [13] A. Dehbi, S. Guentay, A model for the performance of a vertical tube condenser
in the presence of non-condensable gases, Nucl. Eng. Des. 177 (1997) 4152.
because of the large temperature difference between the con-
[14] D. Che, Y. Da, Z. Zhuang, Heat and mass transfer characteristics of simulated
denser wall and the coolant, it is found that the buoyancy effect high moisture ue gases, Heat Mass Transfer (Waerme- und
on heat transfer in the coolant channel is in general large and can- Stoffuebertragung) 41 (2005) 250256.
not be neglected. This buoyancy effect on enhancing the heat [15] J.L. Munoz-Cobo, L. Herranz, J. Sancho, I. Tkachenko, G. Verdu, Turbulent vapor
condensation with noncondensable gases in vertical tubes, Int. J. Heat Mass
transfer in the coolant channel has been included in the present Transfer 39 (1996) 32493260.
model, and was found to be an important phenomenon in govern- [16] H.C. No, H.S. Park, Non-iterative condensation modelling for steam
ing the performance of condensers. Further work is needed in com- condensation with non-condensable gas in a vertical tube, Int. J. Heat Mass
Transfer 45 (2002) 845854.
paring the present model with those assuming constant wall [17] J.C. De la Rosa, J.L. Munoz-Cobo, A. Escriva, Non-iterative model for
temperature or heat ux. condensation heat transfer in the presence of non-condensable gases inside
passive containment cooling vertical tubes, Nucl. Eng. Des. 238 (2008) 143
155.
Acknowledgments [18] M.K. Groff, S.J. Ormiston, H.M. Soliman, Numerical solution of lm
condensation from turbulent ow of vaporgas mixture in vertical tubes,
Int. J. Heat Mass Transfer 50 (2007) 38993912.
We would like to thank Professor Yesin and Dr. Ranrikut for [19] S.T. Revankar, S. Oh, W. Zhou, G. Henderson, Condensation correlation for a
kindly making their experimental data available. vertical passive condenser system, Nucl. Technol. 170 (2010) 2839.
[20] M. Siddique, The effects of noncondensable gases on steam condensation
under forced convection conditions, Ph.D. Thesis, Cambridge, MA, 1992.
References [21] S.Z. Kuhn, Investigation of heat transfer from condensation steamgas mixture
and turbulent lms owing downward inside a vertical tube, Ph.D. Thesis,
University of California, Berkeley, 1995.
[1] C. Bum-Jin, K. Sin, K. Min Chan, M. Ahmadinejad, Experimental comparison of
[22] A. Tanrikut, O. Yesin, Experimental research on in-tube condensation in the
lm-wise and drop-wise condensations of steam on vertical at plates with
presence of air, in: Proceedings of a Technical Committee Meeting, IAEA
the presence of air, Int. Commun. Heat Mass Transfer 31 (2004) 10671074.
TECDOC-1149, Switzerland, 1998, pp. 1417.
[2] W. Nusselt, Des oberachenkondensation des wasserdamfes, Z. Vereines
[23] Y.A. Cengel, M.A. Boles, Thermodynamics An Engineering Approach, sixth ed.,
Deutsch. Ing. 60 (1916) 541564. 569575.
McGraw-Hill, 2008.
[3] L.R.A. Bromley, Heat transfer in condensation effect of heat capacity of
[24] V.D. Rao, V.M. Krishna, K.V. Sharma, P.V.J.M. Rao, Convective condensation of
condensate, Ind. Eng. Chem. 44 (1952) 29662969.
vapor in the presence of a non-condensable gas of high concentration in
[4] W.J. Rohsenow, Condensation, in: Handbook of Heat Transfer, McGraw-Hill,
laminar ow in a vertical pipe, Int. J. Heat Mass Transfer 51 (2008) 60906101.
Berlin, 1973, pp. 77104.
[25] W. Kays, R. Moffat, The behavior of transpired turbulent boundary layers,
[5] E.M. Sparrow, S.H. Lin, Condensation heat transfer in the presence of a
NASA STI/Recon Technical Report N, 32383, 1975.
noncondensable gas, J. Heat Transfer (1964) 430436.
[26] M. Bucci, M. Sharabi, W. Ambrosini, N. Forgione, F. Oriolo, S. He, Prediction of
[6] W.J. Minkowycz, E.M. Sparrow, Condensation heat transfer in the presence of
transpiration effects on heat and mass transfer by different turbulence models,
noncondensables interfacial resistance superheating variable properties and
Nucl. Eng. Des. 238 (2008) 958974.
diffusion, Int. J. Heat Mass Transfer 9 (1966) 11251144.
[27] H.C. Reynolds, T.B. Swearngen, D.M. McEligot, Thermal entry for low Reynolds
[7] E.M. Sparrow, W.J. Minkowycz, M. Saddy, Forced convection condensation in
number turbulent ow, J. Basic Eng. 91 (1969) 8794.
the presence of noncondensables and interfacial resistance, Int. J. Heat Mass
[28] J.P. Holman, Heat Transfer, seventh ed., McGraw-Hill Inc., UK, 1992.
Transfer 10 (1967) 18291845.
[29] R.H. Norris, Some simple approximate heat transfer correlations for turbulent
[8] C.Y. Wang, C.-J. Tu, Effects of non-condensable gas on laminar lm
ow in ducts with rough surfaces, in: Augmentation of Convective Heat and
condensation in a vertical tube, Int. J. Heat Mass Transfer 31 (1988) 2339
Mass Transfer, ASME, New York, 1970, pp. 1626.
2345.
[30] S.E. Haaland, Simple and explicit formulas for the friction factor in turbulent
[9] P.D. Lebedev, A.M. Baklastov, Z.F. Sergazin, Aerodynamics heat and mass
pipe ow, J. Fluid Eng. (1983) 8990.
transfer in vapor condensation from humid air on a at plate in a longitudinal
[31] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport phenomena, second ed., John
ow in asymmetrically cooled slot, Int. J. Heat Mass Transfer 12 (1969) 833
Wiley & Sons, 2007.
841.
[32] H.D. Baehr, K. Stephan, Heat and Mass Transfer, second ed., Springer, Berlin,
[10] T. Mamyoda, K. Asano, Experimental study of condensation of vapors in the
Heidelberg, 2006.
presence of non-condensable gas on a short horizontal tube, J. Chem. Eng. 27
[33] A. Tanrikut, O. Yesin, Experimental research on in-tube condensation under
(1994) 485490.
steady-state and transient conditions, Nucl. Technol. 149 (2005) 88100.
[11] S. Oh, S.T. Revankar, Experimental and theoretical investigation of lm
[34] W.M. Kays, M.E. Crawford, B. Weigand, Convective Heat and Mass Transfer,
condensation with noncondensable gas, Int. J. Heat Mass Transfer 49 (2006)
fourth ed., McGraw-Hill, 2005.
25232534.
[35] S.W. Churchill, A comprehensive correlating equations for laminar assisting
[12] A.P. Colburn, O.A. Hougen, Design of cooler condensers for mixtures of vapors
forced and free convection, AIChE J. 23 (1977) 1016.
with noncondensing gases, Ind. Eng. Chem. 26 (1934) 11781182.

You might also like