International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

www.elsevier.com/locate/ijhmt

Recirculation model of kettle reboiler

a,*
S. Kumar , A. Jain b, B. Mohanty c, S.C. Gupta c

a
Department of Mechanical Engineering, Faculty of Mechanical Engineering, R.I.T. Jamshedpur, Jamshedpur 831 014, India
b
Department of Applied Mechanics, M.N.R.E.C. Allahabad, Allahabad, India
c
Department of Chemical Engineering, I.I.T. Roorkee, Roorkee, India
Received 26 November 2001; received in revised form 29 October 2002

Abstract
The present paper deals with the simulation of a kettle reboiler. Considering rectangular tube sheet, concept of
internal recirculation developed in a kettle reboiler during boiling, changes in physico-thermal property of liquid and
liquid vapour mixture with temperature and pressure and using empirical correlations, a hydrodynamic model has been
developed to determine pressure drop, vapour quality, recirculation rate, boiling regime, and heat transfer coefficient at
various tube rows of the bundle.
Results show, recirculation rate in a reboiler has been found to vary with the heat flux and pressure. Further, at a
given value of heat flux and pressure vapour quality, mass flux, and heat transfer coefficient have been found to increase
gradually from bottom to top tube row of the bundle.

2003 Elsevier Science Ltd. All rights reserved.

Keyword: Interaction factor

1. Introduction away from the bundle are neglected, and (iii) the cas-
cading effect of tube interaction is negligible. Fair and
It is a known fact that in a tube bundle, vapour Klip [11] have proposed models for the design and
bubbles emerging from lower heating tube interact with analysis of kettle reboiler on the basis of following as-
upper heating tube vapour bubbles and enhances the sumptions: (i) recirculating liquid enters from bottom
heat transfer rate of upper tubes by contributing to- only, (ii) sensible heating zone at the lower part of the
wards enhancement in turbulence. The enhancement bundle is negligible, (iii) subcooled boiling at the lower
cascades from bottom to top of the bundle. As a con- part of the bundle is not considered, and (iv) the cas-
sequence of it liquid hold up and vapour fraction and cading effect of the tube interaction is negligible. Leong
other relevant variables continuously change. Besides, and Cornwell [16] have performed experimental study
recirculation of liquid also occurs across the tube bundle considering: (i) tubes are heated at constant heat flux
due to density difference. Some investigators [4– and atmospheric pressure, (ii) flow between the inter-
6,11,14,19] have develop models to predicts the occur- tube column is small, and (iii) the mass flow rate in each
rence of various flow regimes and thereby heat transfer column at any particular uniform heat flux is approxi-
in each regime inside a tube bundle. Brishbane et al. [4] mately constant. Burnside [5] has developed a 2D kettle
have developed a method to predict kettle reboiler per- reboiler model and obtained bundle average heat
formance assuming following: (i) liquid starts boiling at transfer coefficient using typical superposition and
the bottom of the bundle even though it is slightly asymptotic flow boiling correlations. Further, Burnside
subcooled, (ii) frictional losses in single-phase liquid et al. [6] have performed test on a thin slice rig with the
help of particle image velocimetry though a complete
solution of the problem could not be obtained. How-
*
Corresponding author. Tel.: +91-657-200-846. ever, none of the above models validate experimental
E-mail address: [email protected] (S. Kumar). observations correctly. This might be due to the absence
0017-9310/03/$ - see front matter
2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0017-9310(03)00015-2
2900 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

Nomenclature

cp specific heat at constant pressure, J/kg K Kt criteria for bubble break-off frequency,
ðqg kv Þ2
csf liquid surface combination factor
cpl Ts ql ½grðql qv Þ1=2
d diameter, m qffiffiffiffi 
h heat transfer coefficient, W/m2 K KSub KSub ¼ 1 þ qL TSat T
qV TSat
h0U heat transfer coefficient of upper tube when
both tubes heated simultaneously, W/m2 K Greek symbols
k thermal conductivity, W/m K l dynamic absolute viscosity, N s/m2
pt pitch of tube, m q density, kg/m3
q heat flux, W/m2 r surface tension, N/m
Ts surface temperature of tube k latent heat of vapourisation, J/kg
Subscripts
Dimensionless numbers b bulk, boiling, bundle
3
Gr Grashof number, gd b1ðTt2W Ts Þ B bubble
Nu Nusselt number, hd=k qffiffiffiffiffiffiffiffiffiffiffi L liquid
r
NuB Nusselt number for boiling, khl ðql q vÞ L lower heating tube
qffiffiffiffiffiffiffiffiffiffiffiffiffi
PeB q r
Peclet number of boiling, qv ka1 gðql qv Þ O outer
PL plain heating tube
Pr Prandtl number, lcp =kl
qffiffiffiffiffiffiffiffiffiffiffi PU plain upper heating tube
q r
ReB Reynold number, lk ðql qv Þ S saturation
q0U
qffiffiffiffiffiffiffiffiffiffiffi U upper
Re0B modified Reynold number, lk
r
ðql qv Þ
where
V vapour
q0U ¼ qU þ kqL and k ¼ ð13:77q 0:215
 1Þ ef effective

of interaction factor in the boiling heat transfer corre- bundle the temperature of the liquid increases, reaches
lation used in saturated boiling regime and other in- to a saturation temperature and then vapour bubble
herent assumptions as discussed above incorporated in formation on the tube surface takes place leading to
their model. Thus, a new hydrodynamic model incor- two-phase liquid and vapour mixture. This phenomenon
porating liquid recirculation from bottom as well as continues and vapour fraction in the mixture rises. Thus,
from the side of tube bundle and considering interaction recirculation of liquid from top to bottom sets in due to
factor is developed for detailed analysis of boiling heat difference in density. The recirculated liquid joins the
transfer phenomena occurring in multitubular reboilers. fresh liquid entering to the reboiler. The combined (fresh
+ recirculating) liquid attains a velocity depending upon
physico-thermal properties, and the quantity of liquid,
2. Simulation of kettle reboiler reboiler geometry and other parameters. Heat transfer in
this region is by convection due to velocity induced by
Kettle reboiler is basically a shell and tube type heat recirculation of liquid. Thus, many flow regimes coupled
exchanger consisting of a tube bundle arranged on a with convection and boiling is observed across the
square-in-line pitch enclosed in a shell for easy cleaning. tube bundle. Moreover, the magnitude of each of them
It also contains a vertical oriented weir to ensure suffi- depends on velocity of liquid, heat flux, operating
cient height of liquid pool in the shell. Heating medium, pressure, diameter of tubes, spacing between them and
usually steam, flows in the tubes whereas the liquid to be other pertinent parameters for a given liquid to be va-
partially vapourised remains in the shell side. Initially, pourised.
the liquid, usually below boiling temperature, in the
bottom-most portion of the bundle is heated by natural 2.1. Development of model
convection and then by subcooled followed by saturated
boiling regimes when it moves from bottom of the Heat transfer from a tube to fluid in a reboiler ba-
bundle to top. The extent of each regime depends upon sically occurs by two mechanism operating simulta-
composition of fluid as well as parameters affecting neously––boiling heat transfer and convective heat
performance of distillation column such as type and transfer due to velocity (turbulence due to vapour
volume of liquid, operating pressure, heat flux, geo- bubbles) induced in the fluid. However, contribution
metrical parameters etc. From the bottom to top of the of each depends upon heat flux, type of fluid, reboiler
 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909 2901

 0:25
geometry, operating pressure, etc. Hence, heat transfer Pr
coefficient of a tube-row can be written as follows: Nu ¼ 0:033ðReÞ0:8 ðPrÞ0:4
Prw
when Re > 2 105 ð4dÞ
h ¼ ahb þ bhc ð1Þ
Once boiling starts, fluid becomes a two-phase mix-
ture. Heat transfer coefficient for such a two-phase
where, a and b represent nucleate boiling and two-phase
mixture is computed by the use of above equations, Eqs.
convective correction factors, respectively. Since, two
(4a)–(4d) with the inclusion of a two-phase multiplier,
phase convective process dominates over most of the
which is as follows [21]:
tube length, b may be assumed to be unity [11] so as to
   0:6
get actual contribution of convective boiling in tube ql
bundle. It may be mentioned here that the values of hb /¼ 1þ 1 x ð5Þ
qv
and hc at a tube row depend upon the physico-thermal
properties of fluid, heat flux, operating pressure, velocity
of fluid, reboiler geometry and other pertinent variables. 2.3. Boiling(nucleate) heat transfer coefficient
Nucleate boiling heat transfer coefficient and convective
heat transfer coefficient on a tube row of the reboiler can Tube sheet of the bundle in reboiler is hypothesised
be computed as follows. to be of a rectangular shape having its cross sectional
area equal to that of an actual tube sheet so that oper-
2.2. Convective heat transfer coefficient ating conditions of the liquid at a tube row in each
column remain uniformly same. Schematic diagram of
The extent of natural and forced convection is de- the assumed rectangular tube sheet in the reboiler is
termined by the computation of Grashof number, Gr shown in Fig. 1. This portion pertains for identification
and the Reynolds number, Re of the fluid around the of flow-regime and computation of heat transfer coeffi-
tube row in question. As recommended by Kreith [12], cient associated with the regime. Following assumptions
following criteria for the determination of mode of are made for mathematical modelling of reboiler:
convective heat transfer are used.
Gr • Pressure, temperature and state of fluid entering the
When natural convection; P1 ð2aÞ reboiler are same as those of recirculating fluid.
Re2
• Disengagement of vapour completely occurs in the
and top space above the bundle and liquid only returns
Gr to the bottom portion of reboiler.
When forced convection occurs 61 ð2bÞ • Fluid moves upward from bottom to top of the tube
Re2
bundle as a single column.
For the case of natural convection as envisaged by
Eq. (2a), conventional method of computing heat
transfer coefficient during natural convection by
Vapour disengagement space
Nu ¼ cðGr PrÞn ð3Þ
E A
is employed.
liquid
For the case of forced convection as determined by ZDE
D
Eq. (2b), following forced convection correlations as
suggested by Zukauskas and Ulinskas [23] are used:
 0:25
Pr
Nu ¼ 0:9ðReÞ0:4 ðPrÞ0:36 Rectangular tube
Prw sheet
when 1 < Re < 100 ð4aÞ
 0:25
Pr C
Nu ¼ 0:52ðReÞ0:5 ðPrÞ0:36
Prw
B ZBC
when 100 < Re < 1000 ð4bÞ Bottom of the shell

 0:25
Pr
Nu ¼ 0:279ðReÞ0:63 ðPrÞ0:36
Prw Fig. 1. Approximate configuration of reboiler with assumed
when 103 < Re < 2 105 ð4cÞ rectangular tube sheet.
2902 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

• Two-phase mixture of liquid and vapour is consid- where

ered to be homogeneous. d0 G
Msl
• Loss of heat to surrounding is negligible. It amounts Re
¼ and G
¼
l
ðpt  do ÞLt Nt
to transfer of the entire quantity of heat to the fluid
from tube. and u is a parameter for two-phase pressure drop [21].
• All the tubes of the bundle are energised with the It is as follows:
same value of heat flux. However, in the practical sit-
uation, variable heat flux may occurs. ql
u¼
q

Fluid when moves from point A at liquid vapour

interface to point B at the bottom of the shell and then and
from point B to point C is essentially a single phase  1
x ð1  xÞ
liquid. As the fluid further moves up in the bundle its q
¼ þ
qv ql
temperature rises, vapour bubbles develop and a two-
phase mixture of liquid and vapour is formed. So, the Fluid is essentially a two-phase liquid–vapour mix-
liquid at somewhere between points C and D turns into ture between points D and E. Hence, pressure drop
a two-phase mixture and thereafter, continues to be so between these points is given by:
up-to point E. The exact location of tube row at which
boiling starts and formation of two-phase mixture be- DpDE ¼ q

gZDE ð7dÞ
gins depends upon heat flux, pressure and relevant where q

represents density of liquid–vapour mixture

physico-thermal properties of liquid and vapour. A at the conditions of point D.
pressure balance in the flow circuit A–B–C–D–E is made Eq. (6) along with Eqs. (7a)–(7d) can be used to
to determine the velocity of liquid recirculating in the calculate recirculation rate of fluid for the vapourisation
reboiler. It is as follows: of a given liquid in a reboiler of specified geometry. A
DpAB  DpBC  DpCD  DpDE ¼ 0 ð6Þ value of mass flow rate of liquid is assumed in the re-
boiler. Pressure drops in various legs of flow circuit are
where, DpAB , DpBC , DpCD , and DpDE represent pressure calculated from Eqs. (7a)–(7d) and substituted in Eq. (6)
drop of circuit legs AB, BC, CD, and DE, respectively. to examine its validity. The value of assumed mass flow
Pressure drop in leg AB is due to static head available rate of fluid that satisfies Eq. (6) is the recirculation rate
between points A and B. It is as follows: in reboiler.
X
N

 At this junction it is essential to identify the mode of
DpAB ¼ qlj g zðjÞ  zðj1Þ ð7aÞ boiling taking place for the prediction of actual heat
j¼1 transfer coefficient in the tube bundle. For the condition
when the temperature of tube wall surface is below the
Similarly, pressure drop in the leg BC is due to static
value of TwONB , nucleate boiling can not be sustained and
head available between points B and C and can be ex-
the mode of heat transfer in this region is termed as
pressed by following equation:
subcooled boiling which may be determined by the fol-
DpBC ¼ ql;o gZBC ð7bÞ lowing correlation of Smith [21]:
 0:5
where subscript, o refers to the condition at point B. 8rqTs
TwONB ¼ Ts þ ð8Þ
Pressure drop in the leg CD occurs due to static head, kl kqv
acceleration of the fluid and the frictional force acting
on the fluid between points C and D. It is as follows: Correlation due to Alam [1] has been selected here to
 determine the value of heat transfer coefficient in sub-
X N

 X N
1 1 cooled boiling region, as this correlation is based on the
DpCD ¼ q
g zj  zðj1Þ  G2

j¼1 j¼1
q
ðjÞ q
ðj  1Þ subcooled boiling data of a large number of liquids over
XN
4f
G2j
a wide range of heat flux and degree of subcooling. The
 u ð7cÞ correlation due to them is as follows:
j¼1
2q

Nu ¼ 0:084ðPeB Þ0:6 ðksub Þ0:5 ðkt Þ0:37 ð9Þ

where subscript
represents single/two-phase fluid con-
dition occurring at a tube row. Once, the temperature of tube wall surface crosses
Friction factor f
is given by the following expression the value of TwONB , nucleate boiling starts and there is
[11]: plethora of correlations [2,3,18–20,22] available for nu-
cleate boiling regime in literature. But none of them
0:63 seems to be suitable for calculation of heat transfer co-
f
¼
Re0:22

efficient in nucleate boiling regime of a reboiler because
 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909 2903

they do not account for interaction caused by vapour investigation is due to Leong and Cornwell [16] for the
bubbles of lower heating tube on upper heating tubes vapourisation of R-113 for heat flux values of 10, 20,
which is an important aspect in multitubular bundle. and 50 kW/m2 at atmospheric pressure. Hence, these
Some investigators [11,19] have used available correla- data have been used to examine the validity of the
tions with modifications to account the interaction of model. Figs. 2 and 3 show a plot between heat transfer
vapour–bubbles amongst heating tubes but could not coefficient computed from present model and the
predict heat transfer rate satisfactorily in reboiler. In the
present investigation, correlation based on experimental
data [15] for boiling of liquids over a row of two hori-
zontal heating tubes arranged in a vertical grid has been 6
+30 %
employed which is of the form:
NuB ¼ c1 ðRe0B Þ0:7 ðPrÞ0:4 ð10Þ

Predicted heat transfer coefficient, kW/m2

It is important to mention here that Eq. (10) is dis-
tinctly different from single tube nucleate boiling corre- 4
lations available in literature as it contains the term, -30 %
modified Reynolds number, Re0B which considers inter-
action amongst tube rows of the bundle.
Following correlation due to Fair and Klip [11] for
nucleate boiling correction factor, a has been used to
determine flow regime in the reboiler: 2

 1:28
0:00735 1
a ¼ 1:1  G ð11Þ
4:88 Xtt

where Xtt is the flow parameter developed by Martinelli 0

et al. [11]. It is given by the following relationship: 0 2 4 6
 0:9  0:5  0:1 Experimental heat transfer coefficient, kW/m2K
1x qv ll
Xtt ¼ ð12Þ Fig. 2. Comparison of experimental data due to Leong and
x ql lv
Cornwell [16] with the present model.
The value of X in Eq. (12) represents vapour fraction
in the mixture. The value of a defines the flow regime.
Flow is bubbly for a ¼ 1, slug and frothy for 0 < a < 1, 18
and annular or mist for a < 0.
16
Thus, Eqs. (1)–(12) represent a heat transfer model of
a multitubular reboiler which permits the computation
14
of pressure, temperature, flow regime, fluid velocity,
void fraction, recirculation rate and heat transfer coef- 12
ficient at each tube row of the bundle in the reboiler
from the knowledge of heat flux, operating pressure,
Tube row

10
reboiler geometric parameter and physico-thermal
property of a given fluid. 8
To compute the parameters discussed above, a 2500-
lines computer code in FORTRAN-77 has been devel- 6
oped to simulate the above physical model of the kettle
reboiler. Programme has been run using 32 bit MS 4
FORTRAN POWERSTATION Compiler on a Pen- Leong & Cornwell[16]
2 Present model
tium-133 MHz machine. The algorithm of the pro-
gramme is given in Appendix A.
0
0 1 2 3 4 5 6
2.4. Testing of model 2
Heat transfer coefficient, h, kW/m

Literature contains only one experimental investiga- Fig. 3. Variation of heat transfer coefficient with tube row due
tion on heat transfer studies in a reboiler of 241 tubes to Leong and Cornwell [16] and modified model at heat flux of
arranged in a square-in-line pitch arrangement. This 10 kW/m2 .
2904 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

experimental values due to Leong and Cornwell [16] at X

N

a heat flux of 10 kW/m2 for the boiling of R-113 at Gi ¼ G ð13Þ

i¼0
atmospheric pressure. From these figures, it is seen that
the predictions due to model do not match well with Further, it is assumed that no frictional loss or ac-
experimental values. The reasons for deviation between celeration loss in pressure occurs when recirculating
model and experimental values may be attributed to the liquid joins the liquid flowing upward from lower tubes.
value of constant, c1 , employed in Eq. (10) and might be With this, pressure drop across ith tube row becomes:
due to the fact that in this model the recirculating liquid !2 
has been assumed to enter into the tube bundle only as a Xi
1 1
Dpi ¼ q
gfzðiÞ  zði1Þ g  Gk 
single column from the bottom of the bundle. As a k¼1
q

ðiÞ q
ði  1Þ
matter of fact, this situation does not represent a true  Pi 2
picture of the phenomenon occurring in the tube bundle. 4f
k¼1 Gk
 u
Hence, an improvement of the model is required. 2q

So, pressure drop across the tube bundle, CD be-

2.5. Improved model
comes
The model developed above has a major drawback. It X
N


assumes the whole amount of recirculated liquid to enter DpCD ¼ q
g zðiÞ  zði1Þ
i¼1
into the bottom portion of reboiler as a single column !2 
X
N X
i
and then to pass through the tubes of the bundle. This 1 1
situation is far away from real situation of reboiler. The  Gk 
i¼1 k¼1
q
ðiÞ q
ði  1Þ
recirculated liquid while moving downward from top  Pi 2
and joins the liquid flowing upward at a tube-row de- X
N
4f
k¼1 Gk
 u ð14Þ
pending upon pressure available at that level. In other 2q

i¼1
words, each tube row receives liquids from recirculated
stream, in addition to that flowing upward from lower Thus, Eqs. ((1)–(7b) and (7d)–(14)) constitute an
tubes. This is clearly depicted in a schematic block dia- improved model of heat transfer in a kettle reboiler. It
gram of improved model in Fig. 4. This can be analysed can be used to predict the values of mass flux, recircu-
as given below: lation rate, void fraction, and heat transfer coefficient at
Let G0 , G1 , G2 , G3 , . . ., GN represent the amount of each tube row of the bundle at given temperature and
recirculated liquid entering from bottom, first, second, pressure.
third, . . ., nth tube row of the bundle, respectively. Ap- The heat transfer coefficient predicted by present
plication of law of conservation of mass yields: model have been validated with the experimental data of
Leong and Cornwell [16] for the boiling of R-113 at a
heat flux of 10 kW/m2 and atmospheric pressure in a
kettle reboiler. The physical property of R-113 has been
taken from the handbook [10]. Such a plot is given in
Fig. 5. The values of heat transfer coefficient in the
Vapour disengagement space subcooled regime are more whereas in saturated boiling
A regime they are less as compared to experimental values.
E
This might be due to interaction of vapour bubbles of
D lower tube rows on all tube rows lying above it in the
liquid
bundle. The present improved model has been formu-
lated on the basis of an equation of nucleate boiling
accounting for the interaction of vapour bubbles of
lower heating tube on the heating tube lying just above
Rectangular tube it. With such a scenario the value of interaction factor
bundle may not remain constant. In fact, it is likely to vary from
tube row to tube row and then model may trace the
curve representing the values experimentally observed
C by Leong and Cornwell [16].
Now, it has been thought proper to compare the
B improved model with that of Brisbane et al. [4] and Fair
Bottom of the shell and Klip [11] for the boiling of R-113 at a heat flux of 10
and 20 kW/m2 and atmospheric pressure. The compar-
Fig. 4. Schematic representation of the improved model. ison are shown through Figs. 6 and 7. Figs. 6 and 7
 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909 2905

8 18
+ 20 %

7 16
Predicted heat transfer coefficient, kW/m2

14
6
- 20 % 12
5

Tube row
10
4
8

3
6

2 Leong & Cornwell[16]

4 present improvd model
Fair & Klip[11]
1 Brisbane et al.[4]
2

0 0
0 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7
Experimental heat transfer coefficient, kW/m2 Heat transfer coefficient, h, kW/m2

Fig. 5. Comparison of heat transfer coefficient predicted by the Fig. 7. Variation of heat transfer coefficient with tube row at 20
improved model with the experimental values [16] for the kW/m2 heat flux and atmospheric pressure.
vapourisation of R-113 at 10 kW/m2 heat flux and atmospheric
pressure.
2.6. Effects of operating parameters

Effect of heat flux and operating pressure in a reboiler

18 on various quantities such as mass flux, recirculation
rate, vapour quality, and tube row wise heat transfer
16
coefficient can be analysed by present improved model.
The reboiler used for this purpose is of same geometrical
14
configuration as being used by Leong and Cornwell [16]
and vapourising fluid is R-113. Heat flux has been
12
considered in the range from 5 to 30 kW/m2 and oper-
ating pressure from 78.278 to 101.325 kPa.
Tube row

10

8 2.7. Effect of heat flux on recirculation rate

6 To demonstrate the effect of heat flux and pressure

on the variation of recirculation rate, Fig. 8 has been
4 drawn. It is seen from this plot that at given pressure,
Leong & Cornwell[16]
Fair et al.[11]
recirculation rate increases with increase in heat flux,
2 Present improved model reaches to a maximum value at a particular value of heat
Brisbane et al.[4]
flux and then decreases with increase in heat flux. It may
0
be mentioned here that Fair and Klip [11], and Brisbane
0 1 2 3 4 5 6
Heat transfer coefficient, h, kW/m
2 et al. [4] have also obtained similar trend. This trend is
quite obvious and can be explained as follows: The re-
Fig. 6. Variation of heat transfer coefficient with tube row at 10 circulation rate is dependent on the trade off between
kW/m2 heat flux and atmospheric pressure. frictional, acceleration, and static heads in the bundle.
Further, the frictional and acceleration heads are power
law function of reciculation rate. At low heat flux values
clearly reveal that the present improved model predicts the quality is low, leading to a lower driving force for
the heat transfer coefficient of upper tube of the bundle recirculation. This results in a low recirculation rate in
more accurately as compared to the models of Brisbane the bundle. With the increase in heat flux (upto a certain
et al. [4] and Fair and Klip [11] at heat flux of 10 and 20 limit where recirculation rate is maximum) this driving
kW/m2 respectively. force increases (due to increase in quality) more rapidly
2906 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

28 values. These features are same as has been described by

Jensen [14]. Possible reason for this behaviour lies in the
fact that pressure affects considerably boiling pheno-
27 menon and physical properties of the fluid causing
hydrostatic, frictional and acceleration pressure drop
components to vary with pressure.
Recirculation rate, kg/s

26

2.8. Effect of heat flux on vapour quality

25
Fig. 9 shows a plot for tube row wise vapour quality in
the reboiler for the vapourisation of R-113 at atmo-
24 spheric pressure. In this plot, heat flux is a parameter. As
can be noted from this plot, vapour quality for a given
value of heat flux increases continuously from bottom to
23 78.278 top of the tube bundle. This is quite natural as vapour
89.895
101.325 bubble population in a tube bundle increases from bot-
tom to top of the bundle due to continuous upward
22 movement of vapour bubbles. Further, an increase in
0 5 10 15 20 25 30 35
heat flux increases the vapour quality at a tube row. This
Head Flux, kW/m2 is in view of the fact that an increase in heat flux multiplies
Fig. 8. Variation of recirculation rate with heat flux keeping
number of vapour bubbles on the tube. Similar behaviour
pressure as a parameter. has also been observed at subatmospheric pressures.

2.9. Effect of heat flux on mass flux

in comparison to friction pressure drop––responsible for
recirculation rate to decrease. The overall effect is an Fig. 10 shows a plot to represent the variation of
increase in recirculation rate with heat flux. When heat liquid mass flux with tube row in the reboiler for
flux is increased further the frictional pressure drop and vapourisation of R-113 at a pressure of 78.278 kPa. In
to some extent the acceleration pressure drop becomes this plot, heat flux is a parameter. From this plot, fol-
dominant leading to a decrease in recirculation rate. lowing important features are inferred:
However, increase in pressure, at a given value of
heat flux, increases recirculation rate. It also shifts the 1. At a given value of heat flux, liquid mass flux contin-
peak value of recirculation rate and the value of heat uously increases from first tube row at the bottom to
flux corresponding to peak recirculation rate to higher last tube row at the top of the bundle.

0.030
Heat flux,
kW/m2
5
0.025 10
20
30

0.020
vapour quality, x

0.015

0.010

0.005

0.000
0 2 4 6 8 10 12 14 16 18

Tube row

Fig. 9. Variation of vapour quality with tube row keeping heat flux as a parameter at atmospheric pressure.
 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909 2907

1400

1200

1000
Mass flux, kg/m2-s

800

600

400 Heat flux,

kW/m2
5
10
200 20
30

0
0 2 4 6 8 10 12 14 16 18
Tube row

Fig. 10. Variation of mass flux with tube row keeping heat flux
as a parameter at a pressure of 78.278 kPa.

Fig. 11. Variation of heat transfer coefficient with tube row at

2. An increase in heat flux, at a given tube row, shifts different heat flux keeping pressure as a parameter.
the curve of liquid mass flux in upward direction in-
dicating an increase in mass flux. However, it is more
drawn for different values of pressure so as to determine
evident in the curves of low values of heat flux rang-
the effect of pressure on tube row heat transfer coefficient.
ing from 5 to 10 kW/m2 . For 20 and 30 kW/m2 values
As can be inferred from this plot, heat transfer coefficient
of heat flux, liquid mass flux, at a given tube row, are
at subatmospheric pressures for a given value of heat
found to be lower than that of 10 and 5 kW/m2 , re-
flux continuously rises from bottom to top of the re-
spectively, especially in the upper region of the re-
boiler in the same manner as observed at atmospheric
boiler.
pressure. However, pressure decreases heat transfer co-
efficient at a tube row but its effect is only marginal.
Heat transfer from a heating surface to boiling liquid
Based on above, it can be concluded that heat flux
occurs by two mechanisms convective and nucleate
and pressure are the two important operating parame-
boiling operating simultaneously. Consequently, con-
ters, which affect recirculation rate, vapour quality, mass
vective heat transfer coefficient continuously increases
flux, and heat transfer coefficient of a tube row in a re-
from bottom to top due to interaction of vapour bubbles
boiler significantly.
of various tubes and causes corresponding decrease in
the temperature of heating surface. This in turn leads to
decrease the rate of increase of nucleate boiling com- 3. Conclusions
ponent. Consequently liquid flow rate undergoing
vapourisation decreases with tube row from bottom to 1. Based on physical phenomenon like recirculation of
top. Besides, the recirculating liquid joining the stream liquid in the tube bundle and the cascading effect of
at a tube row also decreases as it move from top to interaction due to the vapour bubbles generated at
bottom of the bundle. The net result of all this is that different tubes, a mathematical model has been devel-
mass flow rate of liquid at a tube row is found to in- oped to predict tube row-wise heat transfer coeffi-
crease from bottom to top of the tube bundle. Further, it cient, vapour quality, mass flux in the bundle.
has also been observed that with the increase in pressure 2. Tube wise heat transfer coefficient predicted from the
mass flux also increases. model has been found to agree well with the experi-
mental data due to Leong and Cornwall [16] for the
2.10. Effect of heat flux on heat transfer coefficient vapourisation of R-113 within a maximum error of
20%.
Fig. 11 represents a plot showing variation of heat 3. Recirculation rate in a reboiler has been found to
transfer coefficient with tube row of the reboiler at vari- vary with the heat flux and pressure. Further, at a
ous values of heat flux. In this plot, curves have also been given value of heat flux and pressure, vapour quality,
2908 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909

mass flux, and heat transfer coefficient have been Step 14: Mass flow rate of liquid at current tube row,
found to increase gradually from bottom to top tube MSL ðjÞ ¼ Mass flow rate of liquid at previous
row of the bundle. tube row, MSV ðj  1Þ  Liquid vapourisation
4. The model is capable of predicting the parametric ef- rate at this tube row, MLV . Mass flow rate of
fect of heat flux and pressure on recirculation rate, vapour at current tube row, MSV ðjÞ ¼ Mass
vapour quality, mass flux, and heat transfer coeffi- flow rate of vapour at previous tube row,
cient in a multitubular bundle. MSV ðiÞ þ mass flow rate of liquid vapourised,
at this tube row, MLv .
quality of liquid; xM ðjÞ
Appendix A. Algorithm of the kettle reboiler
mass flow rate of vapour; MSV ðiÞ
¼
Following algorithm has been used to simulate kettle mass flow rate of of total fluids; MS ðiÞ
reboiler through modified model. Step 15: Compute heat transfer coefficient by Eqs. (3)–
(4d) and (9), (10).
Step 1: The geometric parameters, Dto, Dtb, Dsi, Ct, Step 16: Compute pressure at liquid vapour interface
Cb, Lt, Nt, pt , are collected. level.
The tube lay out, types of tubes, number of Step 17: Compare the computed pressure at liquid va-
tube pass are identified. pour interface and given Pr.
The process data, tube side heating media, If the above pressure difference is greater than
shell side fluid are identified. 0.01 %, modify the recirculation and Go to
The physical properties, Ts, ql , qv , ll , lv , cpl , Step 6.
cpv , k, r, kl , are obtained, and the station points Step 18: If the above computed pressure (Step 16) does
as given in Fig. 1 are identified. not vary with given pressure greater than
Step 2: Compute saturation temperature of R-113 for 0.09%, print the values of heat transfer coeffi-
the given pressure at liquid vapour interface cient, pressure drop, temperature of fluid, qual-
by correlation: ity, recirculation, mass flux, etc. for each tube
T ¼ 78:38 þ 7:817ðpÞ0:249 þ 273:15 row.
Step 19: Stop.
Obtain density of liquid at this temperature, T .
Step 3: Compute bottom pressure at station point ÔBÕ
from Eq. (7a). References
Step 4: Assuming rectangular tube layout obtain no of
tubes in each tube row. [1] S.S. Alam, Nucleate pool boiling of liquid mixtures, Ph.D.
Step 5: Assume flow rate and temperature of inlet fluid Thesis, University of Roorkee, Roorkee, 1972.
and recirculation. [2] C.F. Bonilla, C.W. Perry, Heat transmission to boiling
Step 6: Carry out Step 7 to Step 18 for each tube row. binary liquid mixture, Trans. AIChE 37 (1941) 685–
Step 7: Assuming fluid to be saturated compute pres- 705.
sure for first tube row. [3] V.M. Borishanski, Correlation of the effect of pressure on
the critical heat flux and heat transfer rates using the theory
Step 8: Obtain physical properties at temperature ob-
of thermodynamics on similarity, in: S.S. Kutateladze
tained in Step 2.
(Ed.), Problems of Heat Transfer and Hydraulics of Two-
Step 9: Compute sensible heat. Phase Media, Pergamon, Oxford, 1969, pp. 16–37.
Step 10: Check Q  Qsensible . [4] T.W.C. Brisbane, I.D.R. Grant, P.B. Whalley, A predic-
If it is positive, go to Step 13. tion method for kettle reboiler performance, ASME paper
If it is equals to zero, go to Step 12. no. 80-HT-42, 1980.
If it is negative, go to next Step 11. [5] B.M. Burnside, 2-D kettle reboiler circulation model, Int.
Step 11: Sensible heating is not completed. Obtain rise J. Heat Fluid Flow 20 (4) (1999) 437–445.
in temperature of liquid. Go to Step 8. [6] B.M. Burnside, K.M. Miller, D.A. Mcneil, Conditions at
Step 12: Sensible heating is just completed. Liquid tem- the outside of a thin slice reboiler bundle. Available from
<195-15-67-226/ext/heattransfer/website/pdffiles/1102.pdf>.
perature is saturation temperature. Go to Step
[10] J.G. Collier, Heat Exchanger Hand Book, Hemisphere
14.
Publishing Corporation, Washington DC, 1983.
Step 13: Boiling continues, Liquid is at saturation tem- [11] J.R. Fair, A. Klip, Thermal design of horizontal reboiler,
perature. CEP 79 (3) (1983) 86.
Qvap ¼ Qsupplied  Qsensible [12] F. Kreith, Principles of Heat Transfer, International Text
Book Company, Scranton, PA, 1965.
Liquid vapourisation rate, Mlv ¼ Qvap =latent [14] M.K. Jensen, A model for the recirculating flow in a kettle
heat. reboiler, AIChE Symp. Ser. 84 (263) (1988) 114.
 S. Kumar et al. / International Journal of Heat and Mass Transfer 46 (2003) 2899–2909 2909

[15] S. Kumar, Boiling heat transfer from a vertical row of [23] A. Zukauskas, R. Ulinskas, Efficiency parameters for
horizontal tubes and its applications to the reboiler simu- heat transfer in tube banks, Heat Transfer Eng. 6 (1)
lation, Ph.D. Thesis, University of Roorkee, Roorkee, 1998. (1985) 19.
[16] L.S. Leong, K. Cornwell, Heat transfer coefficients in a
reboiler tube bundle, The Chemical Engineer (April) (1979)
219.
[18] I.L. Mostinski, Calculation of heat transfer and internal For further reading
heat flux in boiling liquids based on law of corresponding
states, Brit. Chem. Eng. 8 (8) (1963) 580. [1] B.M. Burnside, K.M. Miller, D.A. Mcneil, T. Bruce,
[19] J.W. Palen, C.C. Yang, Circulation boiling model for Heat transfer coefficient distribution in an experimental
analysis of kettle and internal reboiler performance, in: kettle reboiler thin slice, Trans. I. Chem. E. 79 (Part A)
Heat Exchanger for Two Phase Application, 21st National (2001).
Heat Transfer Conference, Seattle, WA, 1983, pp. 55– [2] B.M. Burnside, K.M. Miller, D.A. Mcneil, T. Bruce,
61. Pressure drop measurements in a section of kettle reboiler,
[20] W.M. Rohsenow, A method of correlating heat transfer in: 34th National Heat Transfer Conference, Pittsburgh, PA,
data for surface boiling of liquids, ASME Trans. 74 (1952) ASME, NY, 2000, paper no. NHTC 2000-12306, CD.
969–975. [3] A.M.C. Chan, M. Shoukri, Boiling characteristics of a small
[21] R.A. Smith, in: D. Browning (Ed.), Vaporisers––Selection, multitube bundles, Trans. ASME, J. Heat Transfer 109
Design and Operation, John Wiley & Sons, New York, (August) (1987) 753.
1986. [4] W. Fritz, W. Ende, Physik Z. 36 (1935) 379.
[22] K. Stephan, M. Abdelsalam, Heat transfer correlations for [5] R.C. Martinelli, D.B. Nelson, Prediction of pressure drop
natural convection boiling, Int. J. Heat Transfer 23 (1980) during forced circulation boiling of water, Trans. ASME 70
73–87. (1948) 695–702.