Cryogenics 44 (2004) 823831

www.elsevier.com/locate/cryogenics

Study of ow distribution and its improvement on the header of

plate-n heat exchanger
Jian Wen *, Yanzhong Li

Department of Refrigeration and Cryogenics Engineering, School of Energy and Power Engineering, Xian Jiaotong University, Xian 710049, China
Received 25 November 2003; received in revised form 19 March 2004; accepted 30 April 2004

Abstract
In order to enhance the uniformity of ow distribution, an improved header conguration of plate-n heat exchanger is put
forward in this paper. Based on the analysis of the uid ow maldistribution for the conventional header used in industry, a bae
with small holes of three dierent kinds of diameters is recommended to install in the header. The ow maldistribution parameter S
is obtained under dierent header conguration. When the bae is properly installed with an optimum length, with stagger arranged
and suitably distributed holes from axial line to bae boundary, the ratio of the maximum ow velocity to the minimum ow
velocity drops from 3.443.04 to 1.571.68 for various Reynolds numbers. The numerical results indicate that the improved header
conguration can eectively improve the performance. The conclusion of this paper is of great signicance in the improvement of
plate-n heat exchanger.

2004 Elsevier Ltd. All rights reserved.
Keywords: Plate-n heat exchanger; Flow maldistribution; Optimum design

1. Introduction
Plate-n heat exchangers are widely used in process
industries such as air separation and petrochemical
industries to exchange heat energy among more than
two uids with dierent supply temperatures because of
their higher eciency, more compact structure and
lower costs than two-stream heat exchanger networks.
The use of multi-stream plate-n heat exchangers is
more cost-eective and can oer signicant advantages
over conventional two-stream heat exchangers in certain
applications, especially in cryogenic plants [13]. In the
design of plate-n heat exchanger, it is usually presumed
that the inlet ow and temperature distribution across
the exchanger core are uniform and steady. However,
the assumption is generally not realistic under actual
operating conditions due to various reasons. One of

*
Corresponding authors. Tel.: +86-29-8266-3725; fax: +86-29-82668725 (J. Wen); tel.: +86-29-8266-8725; fax: +86-29-8266-8789 (Y. Li).
E-mail addresses: wenjian@mailst.xjtu.edu.cn (J. Wen), yzli-epe@
mail.xjtu.edu.cn (Y. Li).

0011-2275/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cryogenics.2004.04.009

these reasons is mainly related with ow nonuniformity.

In fact, the ow nonuniformity is usually divided into
two kinds: gross maldistribution and passage-to-passage
maldistribution. The gross ow maldistribution in platen heat exchangers is mainly associated with improper
heat exchanger entrance conguration, such as poor
design of header and distributor conguration. The
passage-to-passage ow maldistribution occurs in a
highly compact heat exchanger caused by various
manufacturing tolerances, frosting of condensable
impurities, etc. [4].
The objective of the thermal design of a heat exchanger is to determine the most favorable size and
conguration of the exchanger core, which meets the
demands of the required heat transfer rate within the
specied uid pressure drops, space and cost limitations.
The design of the header signicantly aects the velocity
distribution approaching the face of exchanger core.
Hence, the present analysis has been carried out to
determine the eects of gross ow maldistribution of
crossow heat exchangers on thermal performance and
pressure drop variations. The ow maldistribution effects have been well recognized and presented for heat
exchangers. Fleming [5] set up a ow maldistribution

824

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

Nomenclature
A
A0
A1
A2
A3
C
D
f x
H
h
L
N
Ntu
Ntu; n
Ntux
R
Re
Rx

area of header outlet (m2 )

area ratio of medium-size holes
total area of small-size holes (m2 )
total area of medium-size holes (m2 )
total area of big-size holes (m2 )
heat capacity rate
diameter of the inlet tube (mm)
Beta distribution function
radius of the header (mm)
length of the inlet tube (mm)
length of the header (mm)
the number of channels
number of transfer unites
Theoretical value of Ntu of channel
actual value of Ntu of channel
radius of holes (mm)
Reynolds number
nondimensional velocity

model in paired-channel heat exchangers and investigated the eect of ow maldistribution on the performance deterioration. Based on the experimental data
obtained from wind tunnel experiments, Chiou [6] set up
a continuous ow distribution model and studied the
thermal performance deterioration in cross-ow heat
exchangers. Ranganayakulu and Seetharamu [7] investigated the eect of inlet ow nonuniformity on thermal
performance and pressure drops in crossow plate-n
heat exchangers using nite element method.
It must be noted that all of the above works mainly
concentrated on the eect of ow nonuniformity on the
heat exchanger performance. The literature of improved conguration to enhance the ow uniformity in
plate-n heat exchanger is little in recently years. Lalot
and Florent [8] studied the gross ow maldistribution
in an electrical heater and found that the perforated
grid can improve the uid ow distribution. Jiao [9]
investigated experimentally and analyzed theoretically
the combined eects of distributors inlet angle, conguration parameter h=H and header conguration on
ow velocity distribution. Zhang [10] proposed a
structure of two-stage-distribution and the numerical
investigation shows the ow distribution in plate-n
heat exchanger is more uniform if the ratios of outlet
and inlet equivalent diameters for both headers are
equal. In this paper, a simple way is put forward to
homogenize the ow distribution. A bae with smallsize holes is installed in the traditional header to optimize the header conguration. The investigation on the
eect of the conguration of the bae on the ow
distribution is presented.

S
SI
T
v
d
e
ex

absolute maldistribution parameter

relative maldistribution parameter
temperature (K)
uid ow rate (m/s)
punched ratio
eectiveness of heat exchanger
eectiveness of channel

Subscripts
ave
average value
c
cold uid
ch
channel
h
heat uid
i
serial number of passage
i
inlet
max
maximum value
o
outlet

2. Basic conguration analysis

2.1. Physical model
A schematic view of conventional header (denoted as
conguration A) presented in this study is shown in Fig.
1. The test section, simulating the nitrogen passage in
the plate-n heat exchanger of 20,000 m3 /h oxygen plant
in a certain steel-making industry, has the geometrical
characteristics listed as follows. The diameter of the inlet
ow pipe is equal to 200 mm, the radius of the header is
154 mm and it is 905 mm in length, which are all the
same with actual dimensions. There are 43 micro-passages in the outlet of the header and whose arrangement
is shown in Fig. 2. Composite constructive grids are
used in the analog computation and the nest implemented grid involved about 245,817 cells. There are
selective rened grids in some local place where parametric variation is severe.

Z
Inlet tube

Header

Y
X

Fig. 1. Model of header construction.

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

y

825

Inlet tube

Channel 11
Channel 22
Channel 33

Small passages

3 5 7

9 11

13 15 17

19 21 23

25 27 29

31 33

35 37 39

41 43

Fig. 2. Denition of the channels at the outlet of header.

In this work, CFD software FLUENT was employed

to stimulate the uid ow distribution and pressure
drops in the header of plate-n heat exchanger. Continuity equation and momentum equation are discretized
using nite volume method and two-equation Ke ow
turbulent model is used in the calculation [11,12]. Semiimplicit SIMPLER algorithm is used in the velocity and
pressure conjugated problem and second order upwind
dierence scheme is used in convective terms [13].
Boundary conditions and convergent condition are as
follows: inlet uid Reynolds numbers and pressure are
given. The wall condition is adiabatic and no slip occurs
on the wall. Convergence criterion is specied to residuals 6 1.0 10
6 .
2.2. Numerical results of the conguration A
Two parameters are introduced in this paper to
evaluate the ow maldistribution, namely, relative ow
maldistribution parameter Si and absolute ow maldistribution S, which are dened as follows:
mchi
mave
;
1
Si
mave
v
u
N
u 1 X
2
mchi
mave 2 :
St
N
1 i1
They disclosure the dispersion degree of numerical
results and reect the ow maldistribution conditions
under dierent working conditions, dierent kinds of
header conguration parameters and so on. The smaller
the absolute value, the more uniform the ow distribution. Where N stands for the passage number (here is
43), Vchi stands for the velocity of each passage and Vave
stands for the average velocity of all the passages.
Figs. 3 and 4 present the numerical results of conventional conguration A, Fig. 3 shows the ow distribution in y direction of some passages chosen at random
from the conventional header. The distribution will
further result in the uid maldistribution in the distributor. It is seen in Fig. 4, the ow maldistribution of the
outlet along the x direction is very serious and the

0
0

50

100

150

200

250

300

Location (mm)
Fig. 3. Outlet velocity of channels along with y direction.

1.0
4

Re=1.010
5
Re=1.010
6
Re=1.010

0.8
0.6
0.4

S i =(v ch(i) -vave )/vave

Velocity magnitude (m/s)

0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
0

10

20

30

40

50

Serial number of channels i

Fig. 4. Maldistribution parameter Si along with x direction at dierent
Re.

absolute ow maldistribution S is equal to 0.95, where

the average value for y direction is adopted. Because the
ow header has larger dimension comparing to the inlet
tube diameter, the uid tends to go preferentially into
the channels in the center (Fig. 4). However, the velocities of central passages straightly facing the inlet tube
are unexpectedly lower, it is perhaps the function of a
large isolation zone which rejects ow on the central
part of the header outlet (Fig. 2). Therefore, a reverse
ow occurs and the ow branches o to the two sides.
The velocities in the two ends of the header are also
lower for the uid has been distributed before it reaches
the end of the header.
As presented in Fig. 4, the range of relative maldistribution parameter Si is wider in some sort under larger
Reynolds number. The Si tendency along the length of

826

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

R
boundary. It is presumed that v dA is equal to a constant value under ideal condition. Thus the uid ow is
distributed uniformly before it reaches the header outlet
and the expected object of uniformly distribution is
achieved.
The bae congurations are demonstrated in Figs. 6
and 7. The baes have the dimension of 670 260 5
mm3 (length width height), on which three dierent
kinds of holes in diameter are punched. There are
spotted symmetrically small (2R 10 mm), medium
(2R 20 mm) and big (2R 30 mm) holes from the
axial line to the boundary on the bae. In Fig. 6 the
bae with holes distributed in line arrangement is denoted as conguration B while in Fig. 7 the one with
holes distributed in stagger arrangement is denoted as
conguration C.
The numerical results according to the bae congurations B and C are demonstrated in Figs. 8 and 9,
respectively, under dierent inlet Reynolds numbers.
The headers with baes have more uniform velocity
distribution at outlet. The relative maldistribution
parameter Si is )0.280.59 for conguration B with inline arrangement and )0.230.30 for conguration C
with staggered arrangement, which are much smaller
than those of conguration A ()0.430.81). For the
improved congurations, the velocities increase in the
zone of two ends of header and decrease in the zone near
the axial line. Thus, the uid ow is distributed more
uniformly. Unfortunately the pressure drop may increase and result in the decrease of mean velocities to
some extent, which is inevitable but not anticipated. So
it is obliged to get the suitable bae conguration, such

the header A is conforming and the values are close to

each other under three dierent Re, which demonstrates
that it is mainly the unsuitable header conguration
resulting in the ow maldistribution and the eect of
Reynolds number on ow distribution is not signicant.
So it is better to improve the header conguration to
reduce the ow maldistribution.
In the design of exchangers, it is usually presumed
that the inlet uid ow distributions across the exchanger core are uniform. But the numerical results
indicate that the maldistribution at the outlet of header
A is very serious, which will reduce the exchanger eciency and increases the pressure drops, thus heat exchanger may deviate the designing operating working
conditions. So it is obliged to improve the header conguration and make the uid distribute uniformly in
the header.

3. Improvement of header conguration

It is demonstrated from the above numerical results
that, at the 1/2 height of the header, the uid velocities
near the axial line are high while they are low far away
from the center. It has been found that the best position
for the perforated grid is midway between the inlet tube
and the core of the header [8]. So a bae with small
holes is put forward to install at the 1/2 height of the
header asymmetrically, which is demonstrated in Fig. 5.
The small holes are spotted in the bae according to the
velocity distribution and the punched ratio is gradually
increasing in symmetry from the axial line to the

Baffle

Fig. 5. Denition of the bae position in the header.

Fig. 6. Schematic drawing of bae construction of conguration B.

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

827

Fig. 7. Schematic drawing of bae construction of conguration C.

4. Optimization of bae conguration

1.0

Re=1.0 104
Re=1.0 105
Re=1.0 106

0.8

4.1. The eect of holes distribution

0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
0

10

20

30

40

50

Serial number of channels i

Fig. 8. Maldistribution parameter Si along with x direction (with bae
conguration B).

1.0

Re=1.0 104
Re=1.0 105
Re=1.0 106

0.8
0.6

Si=(vch(i)-vave)/vave

The velocity and pressure drop distribution of three

header congurations are shown in Figs. 10 and 11 in
order to compare the eects of dierent holes distribution. The curves in Fig. 10 illustrates the distribution
characteristic of ow velocity and their dierences for
three header congurations at similar working conditions. The inlet conditions are the same for dierent
headers at Re 1:0  105 and p 27 kPa.
It is indicated that the average velocity is 1.94 m/s and
the absolute maldistribution parameter is 0.36 for conguration B, while they are 2.67 m/s and 0.32 for conguration C, respectively. It shows that the average
velocity of conguration C is much larger than that of
conguration B. It is easily understood that when the
hole distribution in bae is changed from in-line
arrangement to staggered arrangement, the punched
ratio on bae will increase from 47% to 53%, and the

0.4
0.2
0.0

-0.2
-0.4
-0.6
-0.8
-1.0
0

10

20

30

40

50

Serial number of channels i

Fig. 9. Maldistribution parameter Si along with x direction (with bae
conguration C).

Velocity of outlet (m/s)

Si=(vch(i)-vave)/vave

0.6

Type A, Average V=2.42 m/s

Type B, Average V=1.94 m/s
Type C, Average V=2.67 m/s

0
0

as the dimension, the area ratio of dierent holes and so

on, for getting the optimum point of ow uniform distribution and pressure drops.

10

20

30

40

50

Serial number of channels i

Fig. 10. Outlet velocity versus dierent types of headers at
Re 1:0  105 .

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

800

Absolute maldistribution S

600

Pressure drop (Pa)

0.5

400
200
0

2
0.3

0.2

Flow maldistribution
Average velocity

-400

0.0
0.45

10

20

30

40

0.1

Type A
Type B
Type C

-200

0.4

50

Outlet average velocity (m/s)

828

0
0.50

0.55

0.60

0.65

0.70

0.75

0.80

Relative length of baffle(%)

Serial number of channels i

Fig. 12. S and average velocity versus bae length at Re 1:0  105 .
Fig. 11. Pressure drop versus dierent types of headers at
Re 1:0  105 .

ow resistance brought about by bae necessarily decreases. Moreover, the increase of punched ratio leads to
the increase of ow area on the bae and further results
in the decrease of absolute maldistribution parameter S.
Fig. 11 shows the pressure drop distribution for dierent
headers. It can be seen from the gure that the improved
conguration has better pressure distribution, the average pressure drop of conguration C is smaller than that
of conguration B for its smaller ow resistance. Negative pressure drops happen for conguration A in the
center part, which is due to the ow stagnation in the
isolation zone of the header outlet and the ow dynamic
head compensates for the hydrostatic head.
It can be concluded from Figs. 10 and 11 that the
bae with stagger arranged holes is suitable to improve
the ow distribution in header. The ow distribution is
more uniform and the pressure drops are smaller. The
improvement of header conguration with a stagger
arranged bae should be selected rstly.

decreases and the ow is distributed more uniformly.

Combinative consideration of the relationship between
the average velocity and the absolute maldistribution
parameter S, the bae length is selected as 670 mm in
length, which is just 3/4 of the length of header.
4.3. The eect of total area ratio
The dimension of the bae is 670 260 5 mm3 , and
the diameters of the holes on the bae are the same as
above mentioned. The velocities near the center are large
and that near the boundary are small. So the punched
zone of the small-size holes (2R 10 mm) is the center
where it is corresponding to the cross section of the inlet
ow pipe. It is concentrated on the relationship between
the total area of medium-size holes (2R 20 mm) and
that of big-size holes (2R 30 mm) in the following. A1 ,
A2 and A3 are denoted as the total area of small, medium
and big-size holes, respectively. The area ration A0 is
dened as follows:
A0

4.2. The eect of bae dimension

In the calculation, the location where the bae is
installed has been determined to the 1/2 height of the
header, and the thickness of the bae is determined to 5
mm, so the bae is 260 mm in width. The diameters of
the three kinds of holes are the same as described above.
Fig. 12 shows the ow distribution performance along
with the change of relative length of bae to header
at Re 1:0  105 .
Fig. 12 illustrates that the average velocity decreases
when the bae length increases, and also the ow
resistance increases and brings about the increase of
pressure drops. The inuence of bae on ow distribution is signicant along with the increase of bae
length since the absolute maldistribution parameter S

A2
:
A1 A2 A3

The numerical computation has been conducted and

the results of parameters A0 are illustrated in Fig. 13
where the average inlet Reynolds number is equal to
1.0 105 . All the absolute maldistribution parameters
are among 0.300.36, which are close to each other. The
pressure drops decrease as the area ratio A0 increase,
while the average velocities increase at rst and then
decrease. As the area ratio increases, the total area of
medium-size holes increases while that of big-size decreases, so the number of holes increases and the inuence of the bae on ow distribution increases. On the
other hand, the increase of area ratio results in the decrease of punched ratio and the ow resistance increases,
so a combinative consideration of distribution uniformity and ow resistance should be done together. The

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

3

500

400
300
1
Pressure drop
Average velocity

200

Actual curve
Expected curve
Punched ratio (%)

Pressure drop (Pa)

600

70

Outlet average velocity (m/s)

700

829

60

50

100
0

0
0.25

0.30

0.35

0.40

0.45

0.50

Area ratio of medium-size holes

Fig. 13. Pressure drops and average velocity versus area ratio of
medium-size holes at Re 1:0  105 .

average velocity reaches an extreme point at the adjacent of area ratio A0 0:33. For the header with conguration C, it is suitable when area ratio is equal to
0.33, namely the total area of medium-size holes takes
up 1/3 of the whole area of all the holes.
4.4. The eect of punched ratio
When the bae conguration is determined, the
average velocity of conguration C is larger than that of
conguration A due to the decrease of the highest uid
velocity. For conguration C, the velocity distribution is
mostly concentrated in the range of 2.54.0 m/s. The
numbers of passages with the ow velocity between 2.5
and 4.0 m/s takes about 72% of the whole passages at
Re 1:0  105 . While for conguration A, the velocity
distribution is concentrated between 1.0 and 2.5 m/s,
which takes about 65% of the numbers of whole passages. The ow velocity ratio of the maximum to the
minimum drops from 3.443.04 of conguration A to
1.571.68 of conguration C, which reects a more
uniform ow for the improved header. From the above
discussion, the eect of header congurations on ow
maldistribution is prominent. The ow velocity of the
passages near the boundary can be enhanced eectively
by changing the header conguration from A to C. It is
also possible to modify the velocity ratio by depending
on the pressure loss coecient of the bae. The ow
velocity distribution of conguration C gives the most
uniform result among the cases considered in this paper.
From the above discussion, it can be concluded that
the determination of bae conguration has the relationship with the diameter of inlet tube, the length and
diameter of header, the diameters and distribution of
holes when the bae installation location has been dened. Fig. 14 shows the distribution of punched ratio, in
which the staircase curve is drawn according to the

40
-400

-300

-200

-100

100

200

300

400

X position (mm)

Fig. 14. Punched ratio versus x position along with the bae length
for conguration C.

realistic condition and the smooth curve is simulated

from the former. The correlation of punched ratio along
with the position in the direction of bae length under
the ideal situation should be established as follows:

2
X
d 47:75
;
4
100
where d stands for the punched ratio (%) and X stands
for the X position along with the bae length (mm). In
practical condition, if the holes on bae can be punched
according
to the simulated smooth curve, we can get
R
v dA const and the ow can be distributed more
uniformly than the conguration C used here.
In a word, the improved header installed with bae
can eectively enhance the ow uniformity. It is
noticeable that the bae conguration has the relationship with header conguration and uid ow condition, which must be in consideration in the
optimization of bae conguration. It can be foreseen
that uid ow can be distributed uniformly when the
bae is suitable in length, the holes punched on it are
distributed in stagger arrangement and the punched
ratio increases gradually from the axial line to the
boundary.

5. The eect of improved header conguration on thermal

performance
To improve the performance of plate-n heat exchanger is the objective of the research. So a simple
computation is done to show the improvement of the
performance, which is discussed with the eNtu method
extensively used in the design of heat exchanger at
present [14]. It is assumed that: (1) a two-stream counterow exists in the heat exchange, in which one side is

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

uniformly distributed while the other is not, (2) the

uniformly distributed uid is not inuenced by the other
side uid, (3) based on the mathematical theory of statistics, the dimensionless velocities Rxj of channels at
the outlet of header accord with beta distribution and
(4) Ch Cl const. From above assumptions, it is
analyzed that the nonuniformity of the uid is as follows:
(
m
m
Rxi mchi
; Ntux Ntu; n  mchi
; when mchi < mave
ave
ave
mave
Rxi mchi ; Ntux Ntu; n;
when mchi > mave :

1.0

Type A
Type B
Type C
0.9

Effectiveness,e

830

0.8

0.7

So, the eectiveness of the nonuniformly distributed

uid can be written as,
ex

1
e
Ntux1
Rxi 
1
Rxi e
Ntux1
Rxi 

0 6 x 6 1; Rxi < 1;

Th;o x Th;i
Th;i
Tc;i ex

Th;o x Th;i
Th;i
Tc;i ex mave
mi

when 0 6 x 6 x
when x 6 x 6 1 :
7

The average outlet temperature of total channels can

be calculated in Eq. (8) by the velocity and outlet temperature of every channel. The discrete representation
is shown in Eq. (9).
R1
Th;o xmi f x dx
T h;o 0 R 1
;
8
m f x dx
0 i
P
T h;o

mi Th;o x
P
:
mi

From Eqs. (7) and (8), the eectiveness of counterow

heat exchanger can be written as
Ch T h;i
Th;o
Cmin Th;i
Tc;i
Z x
Z 1
mchi

ex
f x dx
exf x dx;
mave
x
0

10

where
f x

xa
1 1
xb
1
Ba; b

Ba; b

0 6 x 6 1;

11

xa
1 1
xb
1 dx;

12

mchi
mmax

and x

mave
:
mmax

40

50

60

70

80

90

100

Ntu

6
(

0.6

13

When the Ntu of heat exchanger is given, the theoretical value Ntu; n of every channel is equal to Ntu from
its concept. The velocity mchi of every channel and the
average velocity mave can be obtained from CFD simu-

Fig. 15. Eectiveness of heat exchanger (e) versus number of transfer

unites (Ntu).

lation. Then a and b can be determined from the

condition of mi distribution.
As shown in Fig. 15, the heat exchanger eectiveness
e is greatly enhanced. And the eect is more obvious as
the increasing of Ntu. The result of calculation quantitatively shows that the improved header conguration
can prominently improve the performance of plate-n
heat exchangers.

6. Conclusions
The performance of uid ow distribution in the
header of plat-n heat exchanger has been calculated
and analyzed. It is found that the uid ow maldistribution is very serious in the direction of header length
for the conventional header used in industry due to poor
header conguration. Based on the analysis of the uid
ow maldistribution, a bae with small-size holes is
recommended to install in the header in order to improve the performance of ow distribution. The results
of calculation indicate that the improved header conguration can eectively enhance the uid ow uniformity. The ow absolute maldistribution parameter S in
plate-n heat exchanger has been reduced from 0.95 to
0.32 by installing the bae. When the bae is proper in
length, the holes are distributed in staggered arrangement, and the punched ratio gradually increases from
axes along with the dam board length, the ratio of the
maximum ow velocity to the minimum ow velocity
may drop from 3.443.04 to 1.571.68 for various
Reynolds numbers. The uid ow distribution in platen heat exchanger is more uniform by the optimum
design of the header conguration. And through a
rough calculation, it is quantitatively shown that the
improved header conguration can prominently improve the performance of plate-n heat exchangers.

J. Wen, Y. Li / Cryogenics 44 (2004) 823831

The bae is lower in cost and convenient in assembly, while the eect of the uid ow distribution uniformly by the improved conguration is obvious. The
conclusion of this paper is of great signicance in the
improvement of plate-n heat exchanger.
Acknowledgement
The authors would like to thank the Ministry of
Education, PR China for its nancial support.
References
[1] Yee TF, Grossmann IE, Kravanja Z. Simultaneous optimization
models for heat integrationI. Area and energy targeting and
modeling of multi-stream exchangers. Comput Chem Eng
1990;14(10):115164.
[2] Chen B-D, Lu Z-W, Liu B-Y, Lu S-Z. A study on heat exchanger
network with multi-uid heat exchanger. In: Chen K, Sarlos G,
Tong TW, editors. Energy and environment. Beijing, China:
China Machine Press; 1998. p. 3139.
[3] Wang L, Sunden B. Design of heat exchanger networks using
multi-stream plate-n heat exchangers. In: Grochal B, Mikielewicz
J, Sunden B, editors. Progress in engineering heat transfer.
Institute of Fluid-Flow Machinery Publishers; 1999. p. 597605.
[4] Changqing C, Yuhao S. Cryogenic heat exchanger. Beijing:
Mechanical Industry Press; 1993 [in Chinese].

831

[5] Fleming RB. The eect of ow distribution in parallel channels of

counter-ow heat exchangers. Adv Cryogenic Eng 1967;December:352.
[6] Chiou JP, The combined eects of maldistributions of inlet air
temperature and the induced ow nonuniformity on the performance of radiator. In: Heat and Oil Cooler International
Congress, 1985. SAE paper 850037.
[7] Ranganayakulu Ch, Seetharamu KN. The combined eects of
wall longitudinal heat conduction, inlet uid ow nonuniformity
and temperature nonuniformity in compact tube-.n heat exchangers: a nite element method. Int J Heat Mass Transfer 1999;42:
26373.
[8] Lalot S, Florent P, Lang SK, et al. Flow maldistribution in heat
exchangers. Appl Therm Eng 1999;26:84763.
[9] Jiao AJ, Li YZ, Zhang R, et al. A study of the conguration
and performance of distributor in plate-n heat exchanger.
In: Proceedings of the 18th International Cryogenic Engineering
Conference, Mumbai, India, 2125 February, 2000. p. 299
302.
[10] Zhang Z, Li YZ. CFD simulation on inlet conguration of platen heat exchangers. Cryogenics 2003;43:6738.
[11] John D, Anderson Jr A. Computational uid dynamics, the basic
with applications. McGraw-Hill Companies, Inc; 1995.
[12] Huang X, Van Sciver SW. Pressure drop and void fraction of twophase helium owing in horizontal tubes. Cryogenics 1995;35:467
74.
[13] Wenquan T. Numerical heat transfer. Xian: Xian Jiaotong
University Press; 2002 [in Chinese].
[14] Fleming RB. The eect of ow distribution in parallel channels
of counterow heat exchangers. Adv Cryogenics Eng 1967;12:
352.