You are on page 1of 7

Chem~

Engineering
and
Processing
ELSEVIER Chemical Engineering and Processing 35 (1996) 479-485

Direct heat and mass transfer in structured packings 1


L. Spiegel, P. Bomio , R. Hunkeler
Sulzer Chemtech Ltd.• Winterthur. S witzerland
Received 11 March 1996; accepted 29 April 1996

Abstract

Direct heat transfer is an important method in the exchangeof heat betweentwo countercurrent process streams within a column.
The processcan be simulated using either the theoretical stage or the rate based concept. With both concepts. a reliable heat transfer
coefficient is needed. Additionally, the rate of the heat transfer coefficient is influenced by the simultaneous mass transfer.
A number of application-dependent methods to estimate the heat transfer coefficient have been developed. mainly for random
packings. It is the purpose of this paper to extend this work to structured packings.
A number of experiments with air/water have been performed in a column of300 mm inner diameter with Mellapak 250.Y.250.X
and 12S,X at ambient conditions. A second group of measurements were done using an oil/air system where only sensible heat
was transferred.
Based on these experimental results a method was developed to predict the heat transfer coefficient for structured packings.
The method is applied to examples of industrial importance, like a gas quench, a gas saturator and a pump-around zone in an
atmospheric tower.
Keywords: Direct heat transfer; Mass transfer; Process streams; Heat transfer coefficient

1. Introduction formance under heat transfer conditions was lacking.


Therefore, a test program was started to gain a better
Direct heat transfer is an important method in the understanding of packing behaviour under such condi-
exchange of heat between two countercurrent process tions.
streams within a column. Well known examples are, . For most experiments the air/water system was used .
among others, pump-around zones in refinery columns This system has well known properties, a large enthalpy
and ethylene quenchers. The heat transfer takes place of vaporization and a Lewis number of I, which greatly
between the gas and the liquid phase that are in close simplifies the evaluation of the test results. The presence
contact. Because no additional heat exchanger equip- of the transfer of both latent and sensible heat is
ment is needed, investment costs are saved. characteristic.
The process can be simulated using either the theo- In order to investigate the purely sensible heat trans-
retical stage or the rate based concept. With both fer (conduction), the system air/dibutylphthalate (DBP)
concepts, a reliable heat transfer coefficient is needed. was selected. DBP has a reasonably low viscosity and a
Additionally, the rate of the heat transfer coefficient is very low vapour pressure in the temperature range
influenced by the simultaneous mass transfer. investigated. The data can be easily evaluated with the
A number of application dependent methods to esti- methods used for tubular heat exchangers, yielding
mate the heat transfer coefficient have been developed, overall heat transfer coefficients.
mainly for random packings [I]. Based on the experimental results a method was
The performance of the structured packing Mellapak developed to predict the heat transfer coefficient for
under distillation and absorption conditions is well structured packings.
known . However, a similar knowledge of packing per-
2. Pilot plant
I Prepared for presentation at the AIChE Spring National Meeting,
March 19-23, 1995 in Houston, Texas (Distillation Column Design The pilot plant consisted of a well-insulated pyrex
and Operation-II). glass column of 400 mm internal diameter, as well as

0255-2701/95/$15.00 © 1996 - Elsevier Science S.A. All rights reserved


PI] S0255 -2701(96)04162-1
480 L. Spiegel et al. / Chemical Engineering and Processing 35 (1996) 479-485

Exhaust The pressure was 96 kPa (atmospheric conditions).


For a test series, the liquid load was kept constant
(between 1 and 25 m" m -2 h -I), the gas load F-factor
Llquld Fv varied between 0.8 and 2.2 Pao. s.
The first group of experiments was done with the
air/water system with exchange of both latent and
sensible heat, and the second group with the air/DBP
system where only sensible heat was transferred.

3. Air/water tests

The measured temperatures on the top and bottom


of the packing in both phases, together with the flow
Back up rates and humidities, were used to calculate the operat-
water
ing line in the enthalpy-temperature diagram. The
, equilibrium line for air/water was calculated based on
Hl ••,. H•• ter
published data [2].
The method described by [3] was used to evaluate the
experimental data. The number of overall transfer units
NTU oo was calculated based on the enthalpy differ-
z ence on the gas side assuming no mass transfer resis-
tance on the liquid side:
dho
I
t op
NTU oo =
Fig. I. Sketch of the pilot plant. bottom ho - hi

where bo is the enthalpy of the bulk gas phase and hi is


liquid tanks and pumps, gas ventilators and auxillary the enthalpy of the gas phase at the interface.
equipment. A sketch is given in Fig. 1. The plant was The number of overall transfer units per meter
operated in a once-through mode for air (open loop), NTUMoo was calculated:
whereas the liquid was recirculated. Provision was
made for air and liquid heating and/or cooling. NTUM oo = NTUoo/Z
The gas flow rate was measured with an U-tube with Z being the packing height.
manometer and propeller, and the liquid flow rate with In Figs. 2 and 3, the NTUM oo is plotted against the
an inertial meter. For humidity measurements, on-line gas load F-factor F v for Mellapak 250.X and 250.Y for
sensors were available. Special attention was devoted to different liquid loads.
the measurement of the temperature in the column. The NTUM oo depend on the gas and the liquid
Twenty high precision Pt100 thermoresistors were used, load. For a system that is gas side controlled, a depen-
together with an on-line hybrid recorder for data aqui- dence on the gas load is expected. The influence of the
sition. Painstaking calibration of the Ptl00 sensors liquid load may be attributed to the effective interfacial
allowed reproducible temperature measurements to a area, which depends on the liquid load for the air/water
real accuracy of 0.1 K. system. This system does not wet a metallic packing
The following packings were tested: Mellapak 250.X surface well at low liquid loads. The effect has been
and 250.Y The geometric packing data are listed in described by de Brito et at. [4] and others.
Table 1. McNulty et at. [5] published measurements of Flexi-
The packing height corresponded to three layers ro- pac 2Y with the same system which agree with our
tated to each other by 90°. More information on the measurements of Mellapak 250.Y quite well (see Fig.
packings is given in [9]. 4). His data demonstrate a strong dependence on liquid
load also.
Table 1 The influence of the liquid load on the effective
Geometric packing data interfacial area is found by a regression analysis to obey
the following power law:
Type Height (m)

Mellapak 250.x 250 30 0.675


Mellapak 250.Y 250 45 0.63 where al,efT is the effective interfacial area and VL is the
superficial velocity of the liquid phase.
L. Spiegel et al. / Chemical Engineering and Processing 35 (1996) 479-485 481

4. Air/DBP tests
MELLAPAK 250.x
airlwater heat and mass transfer Fig. 5 shows the overall heat transfer coefficient U of
10 Mellapak 250X for the system air/DBP in function of
I the gas load F v . Because of the low volatility of DBP,
9 liquid load there is only sensible heat transferred. The temperature
m3/m 2 h of the DBP at the inlet was around 50°C.
The overall heat transfer coefficient was calculated
8
• 5.9
using the theory of tubular heat exchangers:
A 12.5
7
• 25.1

.... 6

I
where Q is the heat flow, A is the heat transfer area of

~
E
e0 5
A
- the packing, and ATLM is the mean logarithmic temper-
ature difference.
:::>
I- 4
A
• Analysis of the data yields the following dependence
of the overall heat transfer coefficient
Z
A
A
• A
.. A
b A
3
•• • A•
••
A

2
• - which is completely in accordance with the air/water
tests. There seems to be no dependence on liquid load.
This is typical for systems wetting the packing surface
well.

° 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 MELLAPAK 250.Y
Gas load FV1 mls 3
(kg/m )0.5 air/water heat and mass transfer
10
Fig. 2. Heat and mass transfer of Mellapak 2S0X measured with the
air/water system. 9 liquid load
m3/m2h
8
• 5.9
A 8.5
7
This has to be compared to the random packings
where a much stronger dependence on V L is found. This
• 12.5
<> 25.1
may be due to the fact that aI.efT of random packings .... 6


I
depends much more on the liquid load than structured E <>
packings. 0 5 A <>
0
• •~
Because the Lewis number of the air/water system is
close to one, the overall heat transfer coefficient U is
calculated from NTUM oo as follows:
~
:::>
I-
Z
4 •
A
A

• • ~
..
<() <>

U= NTUMooPGvG
al.errCP,G
3

2
-
."

where VG is the superficial velocity of the gas phase, PG


is the gas density, and cP,G is the specific heat capacity
of the gas.
Combining these results together we find the follow-
ing correlation for U of Mellapak 250.X: °0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2

UOCV~8
Gas load FVt mls {kg/m 3 )0.5
The exponent 0.8 of the gas load F-factor is typical Fig. 3. Heat and mass transfer of Mellapak 2S0.Y measured with the
for gas side controlled systems. air/water system.
482 L. Spiegel et al.; Chemical Engineering and Processing 35 (1996) 479-485

FLEXIPAC2Y 6. Industrial heat transfer data


airlwater heat and mass transfer
10 We analyzed industrial heat transfer data of two
different applications.
liquid load
9
m3/m2h 6.1. Atmospheric crude oil tower
8 ~"'-12.2
r---e-24.4 The first example is an atmospheric crude oil tower
~.-48.8 in the USA. It has been described by Roza et al. [7] and
7
McNulty et al. [8]. Only the bottom pump-around bed
was equipped with structured packing: 0.91 m Flexipac
6
.-I 3Y.
E

---------.-
.1"--.... The published overall volumetric heat transfer coeffi-
8 5 ............ cient of 17.9 kW m - 3 K -1 (see Table 2) is based on the
0
:E -I--- external heat duty of 8.4 MW that includes latent and
I---
...
::>
Z
4
r-.. I - - sensible heat transport [7]. An analysis analogous to the
... '--- procedure given above based on the internal sensible
3

2
I--
-- ...
heat transfer only shows that the overall volumetric
heat transfer coefficient is 19.2 kW m - 3 K - 1. If we
consider the complete heat transfer including the latent
heat, we find a value of 62.7 kW m - 3 K - 1. This is in
agreement with the analysis of McNulty et al. [8] who
found a heat transfer coefficient of 61.8 kW m - 3 K - 1
simulating the tower with RATEFRAC (ASPEN-
°0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 TECH).
Gas load Fv• m/s (kg/m 3)0.5
MELLAPAK 250.X
Fig. 4. Heat and mass transfer of Flexipac 2Y measured with the air/DBP heat transfer
air/water system [5]. 70

Additional tests with cold DBP showed that U does Liquid load
65
not depend on the direction of the heat flux. m 3/m2h
60 • 3
...
Q' • 6
5. Correlation of the data
C'l
... 10
555
~
., 19
In Fig. 6, the overall heat transfer coefficient U is
plotted against the gas side Reynolds ( - Reo) number:
-
::i 50
c<I)
• 25
.. ~

~ 45
... •
Reo = PoVo dtJ(p.o cos(y» ~.
<I)
.. •
where dh is the hydraulic diameter of packing, Jlo is the
dynamic viscosity of the gas phase, and y is the corru-
...8 40
~
II) •
. • t

gation angle of packing. The cos(y)-term in the
Reynolds number is used to account for the effective
c 35
~
...gbo -
velocity of the gas within the packing. Then one must • •
not distinguish between packings of different corruga- :c •
tion angles (X- or V-types).
25 ~
•A •

The experimental data of the two tests fit together .;


well. A regression analysis yields the following correla-
20
tion:
o~ 1~ 1~ 1A 1~ 1~ 2~ 2~ 2A

U = 0.0925Re~8 Gas load Fv• m/s (kg/m 3)O.5

This theoretical U is also shown in Fig. 6 (labelled Fig. 5. Direct heat transfer of Mellapak 250.x measured with the
"model") for comparison. system air/DPB.
L. Spiegel et al. / Chemical Engineering and Processing 35 (/996) 479-485 483

MELLAPAK 250.X correlation for heat transfer for forced convection (see
direct heat transfer Bird et al. [6]):
100 Nuo/PrM3 a: Re~8

90
where Nuo is the Nusselt number (= 4U/(aIko U is »,
the overall heat transfer coefficient, ko is the thermal
• airlDBP
conductivity, and Pro is the Prandtl number (= f.loCp,o/
Q'
80 • airlwater
ko)·
N f----- model The abscissa in the Nusselt diagram are NUo/PrM 3,
E 70
~
and the ordinate is Reo.
The following procedure was used to obtain the data

-
~
c(1)
'0 50
60

• -:
points for the Nusselt diagram. In the first step, the
column was simulated using PRO/2 (SimSci). From the
profiles, the mean logarithmic temperature difference

-. ~ -~
IE
tiTL M and the amount of sensible heat Qs (difference of
8... 40 •• ':L total heat minus latent heat) was calculated. Then U,
~
II)
c
• •• Nu, Pr and Re were calculated. The physical properties
30 were taken from the simulation. The correction factor
g
-
'(1)
eu
20
/ ~ according to Ackermann [10] for high mass fluxes was
considered also. The data are compiled in the Table 2.
J:

10 /
V In Fig. 7, the Nusselt correlation is tested by this
industrial heat transfer data. We see that the data agree

o 1/ fairly well in the Nusselt diagram. The deviation be-


tween correlation and data is between 10 and 25%.
o 500 1000 1500 2000 2500 3000
Reynolds number ReG
7. Designing a direct heat transfer section
Fig. 6. Comparison of experimental and model data of sensible heat
transfer of Mellapak 250.x. The column was simulated with a rigorous column
design software. With this information, the amount of
The second example is the Kerosin pump-around latent heat transferred (mass transferred times heat of
zone of an atmospheric tower in Japan. The diameter vaporization) for the considered direct heat transfer
was 3.05 m. A mixed bed consisting of Mellapak l25.x section was calculated. This was subtracted from the
and 250.x was installed. Three different operating cases total heat flow to get the sensible heat transferred.
were measured. The external heat duty was 6.15, 6.93 From the temperatures of both phases at the ends of
and 7.39 MW. the bed, the mean logarithmtic temperature difference
was be calculated. U was taken from Fig. 7. The
necessary area for the heat transfer was then calculated:
6.2. Water quench tower in an Ethylene plant
A = Qs/(UtiTL M )
The first example has already been described in [7]. It was assumed that U depends on the gas velocity
The tower is located in Japan and has two sections. The (and on the liquid velocity for non wetting systems).
top section has a diameter of 3.5 m filled with a mixed For a column of given diameter d, the necessary pack-
bed of Mellapak 250.x and l25.Y. The bottom section ing height was then calculated.
has a diameter of 4.5 m filled with Mellapak l25.Y. The Z = A/(rrd 2/4aI )
external heat duty is 14.5 MW in the top PA and 26.7
MW in the bottom section. For operational reasons, the packed height installed
The second example is a German tower operated at a should comprise at least four packing layers.
top pressure of 12.4 kPa. Only the top section was
analysed. The diameter was 4.3 m filled with Mellapak
l25.x. Three different operating cases were analyzed, 8. Conclusion
according to a heat duty of 10.2, 14.7 and 25.6 MW.
The physical properties of the fluids in the industrial A Nusselt type correlation of direct heat transfer
applications described above are very different. To based on the sensible heat flow was developed. The
make possible a comparison of the heat transfer data, correlation was tested with measurements using the
we used the Nusselt diagram which is based on the air/water system and the air/DBP system. The correla-
484 L. Spiegel et al. / Chemical Engineering and Processing 35 (1996) 479-485

Table 2
Data points for the Nusselt diagram

Application U (W (m2 K)-I) Reo Pro (Nu/Pr l / 3)o

Atmospheric tower USA, bottom PA 167 23969 I.10 21I.S5


Atmospheric tower Japan, case I 7I.S 6479 I.10 76.16
Atmospheric tower Japan, case 2 78.0 7381 I.10 82.31
Atmospheric tower Japan, case 3 75.1 6817 I.10 79.25
Water quench Japan, bottom PA 64.5 11194 0.75 85.70
Water quench Japan, top PA 87.9 4739 0.75 58.47
Water quench Germany, case 1 29.7 6371 0.75 42.53
Water quench Germany, case 2 47.3 11054 0.75 67.73
Water quench Germany, case 3 100.6 13755 0.75 142.66

cp specific heat capacity (J (kg K)-I)


Directheat transfercorrelation d column diameter (m)
220 , , dh
Fy
hydraulic diameter of packing ( = 4/alo m)
gas F-factor ( = Va' plfl, Pao. S)
• h specific enthalpy (J kg")
200 •
atm. tower 1
!1h LO latent heat of vaporization (J kg-I)

atm. tower 2
/ k thermal conductivity (W (m K)-I)
180 •
Ethylene quench 1 NTUoo number of overall transfer units based on

l:2
160
Ethylene quench2
'"
- - model V
V NTU-
gas phase
number of overall transfer units based on
.....oS' 140 w / Moo gas phase per meter (m- I)
~ /
V Nu
Pr
Nusselt number ( = Udh/k)
Prandtl number (= pCp/k)
:J 120
...
Z
~ 100
E /
/ Qs
Re
!1TL M
sensible heat flow (W)
Reynolds number (= vpdh/(p cos(y»
mean logarithmic temperature difference
~
:J
c:
80 ./ • (K)

Z
CI>
II)
II)
:J 60 -II
/' '"
U overall heat transfer coefficient based on
gas phase (W (m2 K)-I)
v superficial phase velocity (m S-I)
40 I w
Z packing height (m)

20 I
/ Greek letters
y corrugation angle of packing (0)
o p dynamic viscosity (N s m -2)
o 5000 10000 15000 20000 25000
p gas density (kg m")
Reynolds number ReG
Subscripts
Fig. 7. Test of the heat transfer model by operating data of industrial
applications.
G gas phase
I interface
L liquid phase

tion is confirmed by heat transfer data from various


industrial applications.
References

[I] R.F. Strigle Jr., Random Packings and Packed Towers, Gulf,
Houston, 1987.
9. Nomenclature
[2] VDI-Wiirmeatlas, 5th edn, VOl-Verlag, 1988.
[3] H.S. Mickley, Design of forced draft air conditioning equipment,
al geometric interfacial area of packing (m" Chern. Eng. Prog., 45(12) (1945) 739-745.
m ") [4] M.H. de Brito, U. von Stoekar, A.M. Bangerter, P. Bomio and
M. Laso, Effective mass-transfer area in a pilot plant column
al.cIT effective interfacial area of packing (m' equipped with structured packings and with ceramic rings, Ind.
m ") Eng. Chern. Res., 33(3) (1994) 647-656.
L. Spiegel et al. / Chemical Engineering and Processing 35 (1996) 479-485 485

[5] K.J. McNulty and C.H. Hsieh, Hydraulic performance and crude towers including packed bed pump-around zones using a
efficiency of Koch F1exipak structured packings, AIChE An- rate-based simulator, I. Chem. E. Symp, Ser., 128 (1992)
nual Meeting, Los Angeles, November, 1982. A329-A344.
[6] Bird, Stewart and Lightfoot, Transport Phenomena, Wiley, [9] Sulzer Brothers, Winterthur, Switzerland, Separation Columns
New York, 1960. for Distillation and Absorption, Publication No. 22.13.06,
[7] M. Roza, R. Hunkeler, O.J. Berven and S. Ide, MeUapak in 1991.
refineries and the petrochemical industry, I. Chem. E. Symp. [10] G. Ackermann, Wiinneiibergang und molekulare StofTiibertra-
Ser., 104 (1987) BI65-BI78. gung im gleichen Feld bei grossen Temperatur- und Partial-
[8] K.J. McNulty and S.G. Chatterjee, Simulation of atmospheric druckdifTerenzen, VDI-Forschungsheft 382, VOl-Verlag, 1937.

You might also like