Вы находитесь на странице: 1из 14

Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering


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

Stress characteristics of plate-fin structures in the cool-down process


of LNG heat exchanger
Hongqiang Ma a, Weihua Cai b, Wenke Zheng a, Jie Chen c, Yang Yao a, Yiqiang Jiang a, *
a
School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, China
b
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin, China
c
CNOOC Gas and Power Group, Beijing, China

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

Article history: In order to ensure the structural safety of plate-fin structures, a model is proposed to simulate the stress
Received 20 July 2014 characteristics of plate-fin structures in the cool-down process of LNG heat exchanger based on Finite
Received in revised form Element Method and thermal elastic theory. The simulation results show that the equivalent stress
13 October 2014
rapidly changes and consistently reaches the peak value in the brazed joint, and a crack initiates in this
Accepted 15 October 2014
Available online 14 November 2014
region in the cool-down process. The peak value of equivalent stress in plate-fin structures steadily in-
creases with the increase of temperature difference between natural gas (NG) and mixture refrigerant
(MR) and it is slightly impacted by the temperature drop rate as the temperature drop rate of NG is the
Keywords:
LNG
same as that of MR in the cool-down process. The simulation results are consistent with the experiment
Plate-fin structure results. These results will provide some constructive instructions in safe operation and improvement of
Stress the cool-down process for plate-fin heat exchanger in a large-scale LNG cold-box.
Safety © 2014 Elsevier B.V. All rights reserved.
Finite element method

1. Introduction the empirical data of actual operation in a small-scale LNG cold-


box. This may influence the safe operation of PFHE in a large-
Plate-fin heat exchanger (PFHE), as a type of compact ex- scale cold-box. Because of the complexity of stress phenomenon
changers, has been widely used in the liquid natural gas (LNG) plant and the lack of research work in this field, the stress phenomenon
(Kuznetsov and Shamirzaev, 2007; Liu and Winterton, 1991). in plate-fin structures is still far from well understood in the cool-
Basically, PFHE is comprised of layers of corrugated fins separated down process of PFHE in a large-scale LNG cold-box.
from one another by flat plates, and seals along the edges with bars At present, most of the investigations have been mainly focused
(Jiang et al., 2009b; Peng and Ling, 2008). Currently, the aluminum on residual stress in brazing process. For example, due to the
PFHE has been mainly used in small-scale LNG plants while rarely mismatching thermal expansion coefficients between the solder
in large-scale LNG plants as main cryogenic heat exchanger in the and base metal in the heating stage as well as the dissimilar
world. However, with the growing demand for natural gas around shrinkage in the cooling process, the largest residual stress would
the world, the key technologies should be developed for a large- occur in the brazed joint. Large residual stress would decrease
scale LNG cold-box (Ligterink et al., 2012; Picard et al., 2006). At greatly the strength of the structure, which is generated during the
present, for the PFHE that is used in a large-scale LNG cold-box, the brazing (Chang and Lee, 2007; Liu, 1997). And, the large residual
operating procedure in the cool-down process draws lessons from stress in brazed joint would induce stress cracking, greatly influ-
that in a small-scale LNG cold-box. The cool-down process is ence creep behavior and result in failure, eventually (Aiyangar et al.,
operated by virtue of controlling the temperature drop rate of NG or 2005; Galli et al., 2006). Therefore, in Chena's study (Chena et al.,
MR and temperature difference between NG and MR within PFHE. 2005), the residual stress based on the two-dimensional model
Plate-fin structures, as the key component of PFHE, will cause a was analyzed which was in the brazing process of three layer
severe stress concentration and thermal shocking if the operation interactive flat plates and plane-fin corrugations assemblies. The
on the temperature difference between NG and MR and the tem- result obtained by numerical simulation based on finite element
perature drop rate of NG or MR is unreasonable, which is based on method (FEM) indicate that the stress is very complex at the brazed
joint, and high residual stresses will have a significant influence on
the quality of PFHE. Similarly, Xie et al. (Xie and Ling, 2010) also
* Corresponding author. concluded that the maximum residual stress in brazing process of
E-mail address: jyq7245@sina.com (Y. Jiang).

http://dx.doi.org/10.1016/j.jngse.2014.10.019
1875-5100/© 2014 Elsevier B.V. All rights reserved.
1114 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

plate-fin structures occurs at the brazed joint by FEM. Simulta- cool-down process, a criterion is needed to evaluate whether the
neously, Jiang (Jiang et al., 2007, 2008a, 2009a, 2011b, 2008b) also stress of plate-fin structures is more than the yield limit or not. The
analyzed the brazed residual stress based on FEM, and discussed Von Mises yield criterion (Lee, 1988) is the fundamental criterion to
the effects of geometric parameters and material properties on judge whether the plastic yield failure occurs in plate-fin structures
residual stress for a 304 stainless steel PFHE. It had been found that or not. According to the Von Mises yield criterion, the yield failure
the filler metal thickness has a great effect on residual stress. At the of plate-fin structures is mainly caused by the deviator strain en-
same time, it was also found that the brazed residual stress has a ergy. In other words, no matter how the stress state is for plate-fin
great effect on creep for a stainless steel plate-fin structures. In structures, the plastic yield failure may occur when the deviator
order to optimize the brazing technology for 304 stainless steel strain energy uf is more than the limit value u0f of uniaxial tension
plate-fin structures (Jiang et al., 2011a, 2010, 2011c, 2011d, 2012), yield. The Von Mises yield criterion, in which the safety factor is
the brazed residual stress and its influencing factors were also considered, can be expressed as follows:
analyzed. The results show that the quick cooling method (Chen
et al., 2006) should not be used because thermal cracks will be sr  ½s (1)
created and decrease the strength, however the slow cooling
method can greatly improve the brazing performance. The brazed where [s] is the allowable stress; sr is the equivalent stress (von
processing parameters (including applied load, face sheet thick- Mises stress), which can be expressed as follows:
ness, truss thickness and truss length) have a significant effect on rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1h i
residual stress. These above investigations show that FEM is an sr ¼ ðs1  s2 Þ2 þ ðs2  s3 Þ2 þ ðs3  s1 Þ2 (2)
effective way and is widely used to predict stress characteristics of 2
plate-fin structures (Brown et al., 2006; Celik et al., 2007; Civalek,
where s1 is the first principle stress; s2 is the second principle
2004), and most investigations for plate-fin structures have been
stress; s3 is the third principle stress.
focused on residual stress based on the thermal elasticeplastic
Based on the above yield criterion, the plastic deformation
theory. However, the stress characteristics of plate-fin structures
which induces the plastic yield failure of plate-fin structures may
based on the thermal elastic theory are less investigated in the cool-
occur in the cool-down process, when the equivalent stress is more
down process.
than the allowable stress, whereas the elastic deformation will
The investigation on stress characteristics of plate-fin structures
occur. In other words, the plastic yield failure may occur when the
in the cool-down process of PFHE based on the thermal elastic
equivalent stress obtained by thermal elastic theory is more than
theory is needed to aid the guidance for safe operation and improve
the allowable stress. At the same time, the initial cracking location
the cool-down process of PFHE in a large-scale LNG cold-box.
in plate-fin structures is corresponding with the peak value of
However, it is lack for the study on stress characteristics of plate-
equivalent stress in the cool-down process of heat exchanger ac-
fin structures in the cool-down process. For predicting stress
cording to this evaluation standard on joint cracking. Therefore, the
characteristics of plate-fin structures in the cool-down process,
peak value of equivalent stress is of concern in this paper.
pure experimental approaches are sometimes not practical because
of the difficulties of measurement. With the development of
computer technology, numerical simulation method has been 3. Analysis of the cool-down process
widely used for the analysis and design of metal structures (Kim
and Kuwamura, 2007; Kong et al., 2008; Sen et al., 2007). There- In order to relieve the thermal shock and stress concentration of
fore, numerical simulation method will be a useful tool to analyze plate-fin structures in PFHE of a large-scale LNG cold-box, the cool-
stress characteristics of plate-fin structures for LNG heat exchanger down process has to be implemented to generate uniform tem-
in the cool-down process. perature gradient in plate-fin structures before the heat exchanger
In this work, according to FEM and thermal elastic theory, a reaches the normal operation condition. At the same time, plate-fin
model was proposed to simulate the stress characteristics of plate- structures would also seriously cause the thermal shock and stress
fin structures in the cool-down process of LNG heat exchanger. The concentration, and the crack failure would occur if the cool-down
stress distributions were analyzed in plate-fin structures at process is unreasonable. Therefore, it is necessary to analyze the
different times of the cool-down process. The influence of tem- cool-down process in order to investigate the influence of tem-
perature drop rate of natural gas (NG) or mixture refrigerant (MR) perature drop rate of NG or MR and temperature difference be-
and temperature difference between NG and MR were investigated tween NG and MR for the stress characteristics of plate-fin
for the peak value of equivalent stress (sr) in the stress concen- structures in the cool-down process of PFHE in a large-scale LNG
tration region of the total LNG plate-fin structures in the cool-down cold-box.
process of heat exchanger. And a cool-down experiment, in which According to the literature (Chen Yongdong, 2010), in the cool-
the liquid nitrogen was used as the cryogenic medium of heat down process of PFHE, the NG and MR temperatures decrease
exchanger, was implemented to confirm the simulation results. and their pressures increase gradually with time. For the actual
cool-down process of PFHE, the temperature drop rate of the NG
2. Analysis of strength theory may be different from that of the MR. This would lead to the change
of temperature difference between NG and MR with time in the
In the cool-down process of heat exchanger in a large-scale LNG cool-down process. Thus, the influence of temperature difference
cold-box, the temperature drop rate of MR or NG and the temper- between NG and MR and temperature drop rate of the NG or MR are
ature difference between MR and NG may cause the structural very complicated for stress characteristics of plate-fin structures in
damage of plate-fin structures. At present, many investigations PFHE. In order to analyze the influence of temperature drop rate of
show that the structural damage is induced by the two types: NG or MR and temperature difference between NG and MR for the
brittle fracture and plastic yield. As for plate-fin structures in stress characteristics of plate-fin structures in the cool-down pro-
complex stress state, the plastic yield is mainly reason of structural cess of PFHE, the simplification is necessary: (1) the influence of
damage. In other words, the plastic yield is not allowed for plate-fin temperature difference between NG and MR is analyzed for stress
structures in the cool-down process of LNG heat exchangers. In of plate-fin structures as the temperature drop rate of NG is the
order to ensure the structural safety of plate-fin structures in the same as that of MR and remains constant in the cool-down process;
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1115

Table 1
Geometric parameters (From the bottom plate to top is from No. 1 to No. 5).

Layer number Plate thickness d (mm) Fin number in 100 mm Fin thickness f (mm) Fin height h (mm) Effective length L (mm) Brazing seam thickness (mm)

No. 1 1.6 100 0.4 6 2.5 0.1


No. 2 1.6 100 0.4 6 2.5 0.1
No. 3 1.6 100 0.4 6 2.5 0.1
No. 4 1.6 100 0.4 6 2.5 0.1
No. 5 1.6 e e e 2.5 0.1

(2) the influence of temperature drop rate of the NG or MR is dependent. In this work, only one unit of plate-fin structures (as
analyzed for stress of plate-fin structures as the temperature dif- dotted lines shown in Fig. 1) which consists of four layers plate and
ference between NG and MR remains constant in the cool-down fin, is analyzed in a very short length (L) due to the periodicity and
process. In other words, as the temperature drop rate of the NG is symmetry of plate-fin structures. Fig. 2(a) shows the simplified
the same as that of the MR in the cool-down process, the influence model of plate-fin structures. Fig. 2(b) and (c) display the local
of temperature drop rate of NG or MR is analyzed for the stress of structure of regions 2, 4, 5 and 6 in Fig. 2(a), respectively.
plate-fin structures; (3) the NG and MR pressures are assumed to
increase nearly linearly with time in the cool-down process. Based 4.2. Material properties
on the above assumptions, the influence of temperature drop rate
of NG or MR and the temperature difference between NG and MR is In this analysis, the fin and plate are assumed to be AL3003 and
analyzed for the stress of plate-fin structures in the cool-down the brazing material is AL4004. In order to get fine results, the
process of liquefied heat exchanger in a large-scale LNG cold-box expansion coefficient and elastic modulus are assumed to be tem-
in which the yield is 2.6  106 tons in per year. For the liquefied perature dependent for AL3003 and AL4004. Table 2 lists all the
heat exchanger in operational condition, the flows of NG and MR required material properties used in the simulations.
are counterflow, and the NG is cooled from 210 K to 150 K and
becomes liquid by condensation heat transfer at the NG pressure
PNG ¼ 71 bar and the MR is heated from 155 K to 215 K and 4.3. Analysis strategy
vaporized by the boiling heat transfer at the MR pressure
PMR ¼ 4 bar. In ANSYS, the thermal-mechanical coupling analysis for stress
mainly includes the sequentially coupled and directly coupled
(Khan et al., 2011; Zhao et al., 2012). In this work, the directly
4. Finite element analysis coupling method based on thermal elastic theory is used since the
interaction between temperature field and strain field needs to be
4.1. The model description considered because of the relatively large deformation of plate-fin
structures in the cool-down process of PFHE. For the thermal
Plate-fin structures, as the core component of PFHE in LNG, analysis, the convective heat transfer between NG (or MR) and the
mainly consist of fins and plate (Carey and Mandrusiak, 1986; wall of plate-fin structures is applied, because the heat transfer
Manglik and Bergles, 1995), as shown in Fig. 1. The geometric pa- process with phase change between NG and MR is implemented by
rameters of plate-fin structures are listed in Table 1. Due to the the convective heat transfer between NG (or MR) and the wall of
cyclical repeatability and complexity of plate-fin structures, some plate-fin structures and the heat conduction in plate-fin structures.
issues would be assumed and simplified in the numerical simula- It is expressed as:
tions: (1) the impact of layers number is ignored for the stress of  
plate-fin structures; (2) the flow of NG and MR is assumed to be q ¼ h Tf  Tw (3)
alternating and counterflow, as shown in Fig. 1; (3) both the ma-
terials are treated as isotropic and linear elastic; the mechanical
where q is heat flux; h is the convective heat transfer coefficient; Tw
and physical property parameters are assumed to be temperature
is wall temperature; Tf is the fluid temperature, such as NG or MR.
For the stress analysis, cyclical symmetry boundary on the left
most and right most surfaces is considered due to the periodicity
and symmetry of plate-fin structures. The nodes on the bottom
surface are constrained in vertical direction. At the same time, the
operating pressures of NG and MR are applied to simulate the
interaction between NG or MR and the wall of plate-fin structures.
The impact of external load is ignored in the investigation on stress
of plate-fin structures because the liquefied heat exchanger can free
expansion or shrinkage by means of the free slide of sliding guide
frame under the support structure.

5. Results and discussion

5.1. Stress distribution

In this paper, the stress distribution of plate-fin structures is


analyzed in a very short length near the NG entrance of the lique-
Fig. 1. The diagram of Plate-fin structures d is Plate Thickness; f, Fin Thickness; h, Fin fied heat exchanger in the cool-down process. For the liquefied heat
Height; L, Effective Length; and g, Fin width (g ¼ 100/Fin Number  Fin Thickness). exchanger, the heat transfer coefficient at the NG entrance changes
1116 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Fig. 2. The simplified model of plate-fin structures and its local structures.

with time in the cool-down process and is also distinct for different fin structures when the temperature difference between NG and
cool-down processes. Therefore, it is very difficult to determine the MR is less than 10 K in the cool-down process. So it is reasonable to
heat transfer coefficient in the cool-down process. However, the assume the heat transfer coefficient between NG or MR and the
influence of heat transfer coefficient is slight for the stress of plate- wall of plate-fin structures to be constant as the temperature

Table 2
Material parameters used in FEM simulation.

Material Temperature, K E, GPa CET (106), 1/K Poisson's ratio Specific heat, J/(kg K) Density, kg/m3 Conductivity, W/(m K)

AL3003 305 68.9 22.4 0.33 962 2740 159


250 70.6 19.7
195 72.4 16.9
175 73.2 15.9
145 74.5 14.4
AL4004 305 94.6 15.1 0.35 864 2710 155
250 96.4 14.9
195 98.2 14.7
175 98.8 14.6
145 99.6 14.5
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1117

difference is less than 10 K in the cool-down process. According to terminals of the brazing seam and is much larger at the time
the design condition, the heat transfer coefficient between NG and T ¼ 30 min than that at other times. The direct stress in z direction
the wall of plate-fin structures is 1.5 kW/(m2 K) at the NG tem- also reaches the peak value at the brazed joint. The stress gradient
perature TNG ¼ 155 K and the NG pressure PNG ¼ 71 bar; the heat is much larger at the brazed joint than that at the adjacent region in
transfer coefficient between MR and the wall of plate-fin structures the cool-down process. In other words, when approaching the
is 1 kW/(m2 K) at the MR temperature TMR ¼ 150 K and the MR brazed joint, the direct stresses are presented with a remarkable
pressure PMR ¼ 4 bar. At the same time, in order to ensure the elevating trend and reach the highest value at the brazed joint. At
reliability of heat transfer coefficient, the heat transfer coefficient the same time, the direct stresses in x and y directions are promi-
between MR and the wall of plate-fin structures is also calculated nent while the direct stresses in z direction and shear stresses on x-
by Liu & Winterton's correlation (Liu and Winterton, 1991). The y plane and y-z plane are relatively small. So, it is the direct stresses
calculation result is also close to that in design condition. Therefore, in x and y directions that have the main influence for the strength of
the heat transfer coefficient in design condition is adopted to the brazed joint.
investigate the stress distribution of plate-fin structures. In Fig. 4, the direct stress in z direction reaches the peak value at
In the stress distribution field, there is a distinct stress gradient the brazed joint (regions 1 and 2) and its gradient is much larger
in plate-fin structures. For the sake of a better understanding, four around the brazed joint. The maximum absolute value of direct
reference paths are selected, as marked in Fig. 2. Figs. 3e11 present stresses in x and y directions locates on the middle of path 2 and is
the stress distribution of the four reference paths at the different much larger at the time T ¼ 30 min than that at other times. The
times in the cool-down process. The plots are drawn based on the shear stress on x-y plane is around zero. The direct stresses are
above parameter setting and the temperature drop rate Tr ¼ 2.5 K/ prominent while the shear stresses on x-y plane and y-z plane are
min. Figs. 3 and 4 depict the direct stress and shear stress distri- relatively small at the brazed joint. Therefore, the direct stresses
butions of paths 1 and 2 at different times in the cool-down process, have the main influence for the strength of the brazed joint near the
respectively (sx, sy and sz are direct stress along x, y and z di- fin side while the shear stress on x-y plane has slight influence.
rections, respectively; txy and tyz are shear stress on x-y plane and Fig. 5 depicts the stress distribution of the local position in region 2.
y-z plane, respectively). Paths 1 and 2 are located at the interface Also, it can be drawn the same conclusions as the above suggested
between the brazing filler metal and the plate and between the ones.
brazing filler metal and the fin, respectively. The three direct The direct stresses and shear stresses can merely show the
stresses are symmetrically distributed while the shear stress is stress state while they cannot be used to judge whether the plastic
unsymmetrical. yield failure occurs in plate-fin structures or not. According to the
From Fig. 3, it can be also seen that the absolute value of direct Von Mises yield criterion, the von Mises stress is an equivalent
stresses in x and y directions reaches the peak value at the brazed stress which can be used to evaluate the plastic yield failure.
joint (regions 1 and 2) which is the small transitional arc on the two Therefore, it is necessary to analyze the equivalent stress

Fig. 3. The stress distribution of Path 1.


1118 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Fig. 4. The stress distribution of Path 2.

distribution based on the Von Mises yield criterion. Figs. 6 and 7 (regions 3 and 4). The equivalent stress is very complex at the
show the equivalent stress distributions of paths 1 and 2, respec- brazed joint and much larger at the time T ¼ 30 min than that at
tively. It can be seen clearly that the equivalent stress (Eq. (2)) are other times.
also symmetrically distributed and is also much larger at the time In fact, the main factors to induce the stress of plate-fin struc-
T ¼ 30 min than that at other times. In Fig. 6, the equivalent stress tures may have the following aspects: (1) the thermal distortion is
has its peak value at the brazed joint. The same conclusion can be restricted by external constraints; (2) the temperature is uneven in
also drawn for path 2 in Fig. 7, i.e., the peak value of equivalent plate-fin structures; (3) the thermal expansion coefficient is mis-
stress occurs in the region of brazed joint and the gradient of matched between AL4004 filler metal and AL3003 base metal.
equivalent stress at the brazed joint is much larger than that at the Figs. 12e16 depict the temperature distribution of four reference
adjacent region. paths at the different times in the cool-down process. The results
Fig. 8 depicts the equivalent stress distribution of path 3. The show that the temperature is slightly changed along paths 1, 2 and
results show that the equivalent stress distribution is very complex 4 and at the local positions in region 2 and 6. The influence of
along path 3. The stress gradients of brazed joint (region 2), brazing temperature maldistribution is slight for the stress of plate-fin
seam (region 6) and rectangular region (region 5) are much larger, structures. Hence, the stress of plate-fin structures in the cool-
and the equivalent stress reaches the peak value in these regions. down process is mainly induced by thermal distortion restricted
But the peak value of equivalent stress in region 2 is larger than that by external constraints and the mismatching thermal expansion
in regions 5 and 6. In order to further analyze the equivalent stress coefficients between AL4004 filler metal and AL3003 base metal.
distribution in regions 2 and 6, the equivalent stress distribution of However, when the temperature is lower for plate-fin structures,
the local positions in regions 2 and 6 are presented in Figs. 9 and 10. the influence of mismatching thermal expansion coefficients is
Fig. 9 depicts the equivalent stress distribution of the local position more slightly for the stress of that according to Table 2 while the
in region 6. The results show that the equivalent stress is much influence on which thermal distortion is restricted by external
larger at the brazing seam between plate and fin than that at the constraints is more obviously. Therefore, it is the comprehensive
adjacent region. Fig. 10 presents the equivalent stress distribution effect of thermal distortion restricted by external constraints and
of the local position in region 2. The results indicate that the peak mismatching thermal expansion coefficients between AL4004 filler
value of equivalent stress occurs at the brazed joint near the fin metal and AL3003 base metal that the stress at the brazed joint
side. At the same time, Figs. 9 and 10 also show that the equivalent (regions 1, 2, 3 and 4) and the brazing seam between the fin and the
stress in regions 2 and 6 is much larger at the time T ¼ 30 min than plate is much larger at the time T ¼ 30 min than that at other times.
that at other times. At the same time, the stress concentration is also the main reason
Fig. 11 exhibits the equivalent stress distribution of path 4. These on the stress to reach the peak value at the brazed joint.
results are also obtained that the equivalent stress reaches the peak To sum up, in the cool-down processes of PFHE, the strength of
value, and the stress gradient is much larger at the brazed joint the plate-fin structures is mainly impacted by the direct stress
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1119

Fig. 5. The stress distribution of the local position in region 2.

while the influence of shear stress on x-y plane is slight. At the same of NG and MR are different at different regions. This should result in
time, the equivalent stress along four paths of plate-fin structures stress concentration and impact the structural safety of plate-fin
reaches the peak value at the brazed joint (regions 1, 2, 3 and 4), structures in the cool-down process. Therefore, Fig. 17 shows the
and the equivalent stress is lager at the brazed joint near the fin side peak value of equivalent stress vs. the temperature difference be-
than that in the adjacent region. The brazing seam between the fin tween NG and MR at different temperature drops in per meter
and the plate also suffers the larger stress than that in the adjacent along the flow direction of NG for temperature drop rate
region. This is because the stress is concentrated at the brazed joint, DTr ¼ 2.5 K/min, heat transfer coefficient hMR ¼ 1.0 kW/(m2 K),
the thermal distortion is restricted by external constraints and the hNG ¼ 1.5 kW/(m2 K) and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar.
thermal expansion coefficient is mismatched between AL4004 filler The results show that the peak value of equivalent stress increases
metal and AL3003 base metal. This means that the crack and fatigue with the increase of temperature difference between NG and MR.
failure would occur in the area of brazed joint next to the fin side At the same time, the peak value of the equivalent stress is slightly
firstly when the cool-down process is frequently implemented in influenced by the temperature drop in per meter along the flow
heat exchanger or the operation is unreasonable in the cool-down direction of NG in the cool-down process.
process. At the same time, the relation between the peak value of
equivalent stress and the temperature difference between NG and
5.2. The influence of temperature difference for stress MR is also analyzed at different heat transfer coefficients between
MR or NG and the wall of plate-fin structures, temperature drop
In Section 5.1, it is stated that the equivalent stress reaches the rates and the MR temperature for the temperature drop in per
peak value at the brazed joint of plate-fin structures next to the fin meter which is ignored in a very short length along the flow di-
side due to the thermal distortion restricted by external constraints, rection of NG because its influence is slight for the peak value of
the stress concentration and the mismatching thermal expansion equivalent stress and it is obviously impacted by the temperature
coefficients between AL4004 filler metal and AL3003 base metal. difference between NG and MR in the cool-down process. Fig. 18
This is likely to lead to a crack failure in these regions. So it is depicts the relation between the peak value of equivalent stress
necessary to study the influence of temperature drop rate and and the temperature difference between NG and MR at different
temperature difference between NG and MR on the peak value of heat transfer coefficients between MR and the wall of plate-fin
the equivalent stress at the stress concentration region of plate-fin structures for temperature drop rate DTr ¼ 2.5 K/min, heat trans-
structures in the cool-down process. In the following content, the fer coefficient hNG ¼ 1.5 kW/(m2 K) between NG and the wall of
influence of temperature difference between NG and MR will be plate-fin structures and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar. The
analyzed for the peak value of equivalent stress. peak value of equivalent stress increases with the increase of
In the practical engineering, temperature drop (DTl) in per meter temperature difference between NG and MR. Furthermore, the
is different along the flow direction of NG because the temperatures peak value of the equivalent stress is slightly influenced by the heat
1120 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Fig. 6. The equivalent stress distribution of Path 1. Fig. 9. The equivalent stress distribution of the local position in region 6.

Fig. 7. The equivalent stress distribution of Path 2. Fig. 10. The equivalent stress distribution of the local position in region 2.

transfer coefficient between MR and the wall of plate-fin structures and the wall of plate-fin structures as the temperature difference
as the temperature difference between NG and MR is less than 10 K between NG and MR is more than 10 K in the cool-down process.
in the cool-down process. The peak value of equivalent stress is Fig. 19 shows the peak value of equivalent stress vs. the tem-
obviously impacted by the heat transfer coefficient between MR perature difference between NG and MR at different heat transfer

Fig. 8. The equivalent stress distribution of Path 3. Fig. 11. The equivalent stress distribution of Path 4.
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1121

Fig. 12. The temperature distribution of Path 1.

Fig. 13. The temperature distribution of Path 2.

Fig. 14. The temperature distribution of the local position in region 2.

coefficients between NG and the wall of plate-fin structures for heat transfer coefficient between NG and the wall of plate-fin
temperature drop rate DTr ¼ 2.5 K/min, heat transfer coefficient structures is, the more slight the influence of that is.
hMR ¼ 1.3 kW/(m2 K) between MR and the wall of plate-fin struc- Fig. 20 shows the peak value of equivalent stress vs. the tem-
tures and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar. Similarly, the perature difference between NG and MR at different temperature
peak value of equivalent stress also increases with the increase of drop rates for heat transfer coefficient hMR ¼ 1.3 kW/(m2 K),
the temperature difference between NG and MR, and is obviously hNG ¼ 1.5 kW/(m2 K) and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar.
impacted by the heat transfer coefficient between NG and the wall The same results are also obtained from Fig. 20. The peak value of
of plate-fin structures as the temperature difference between NG equivalent stress is obviously impacted by the temperature differ-
and MR is more than 10 K in the cool-down process. But the greater ence between NG and MR and increases as the temperature
1122 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Fig. 15. The temperature distribution of the local position in region 6.

Fig. 16. The temperature distribution of Path 4.

Fig. 17. The peak value of sr vs. temperature difference at different temperature drops Fig. 18. The peak value of sr vs. temperature difference at different heat transfer co-
in per meter. efficients of MR.

difference between NG and MR increases. At the same time, the hMR ¼ 1.0 kW/(m2 K), hNG ¼ 1.5 kW/(m2 K), temperature drop rate
results also show that the peak value of equivalent stress is slightly DTr ¼ 2.5 K/min and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar. The
influenced by the temperature drop rate as the temperature drop of results also show that the peak value of equivalent stress increases
NG is the same as that of MR in the cool-down process. as the temperature difference between NG and MR increases. The
Fig. 21 presents the data on the peak value of equivalent stress as peak value of the equivalent stress is slightly influenced by the MR
a function of the temperature difference between NG and MR at temperature as the temperature difference between NG and MR is
different MR temperatures for heat transfer coefficient less than 10 K in the cool-down process. The peak value of
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1123

relation between the peak value of the equivalent stress and the
temperature drop rate is depicted in Figs. 22e24.
Fig. 22 presents the data on the peak value of the equivalent
stress as a function of the temperature drop rate at different heat
transfer coefficients between MR and the wall of plate-fin struc-
tures for temperature difference DT ¼ 10 K between MR and NG,
heat transfer coefficient hNG ¼ 1.5 kW/(m2 K) between NG and the
wall of plate-fin structures and fluid pressure PNG ¼ 71 bar,
PMR ¼ 4 bar. The results show that the peak value of the equivalent
stress increases slightly with the increase of temperature drop rate
as the temperature drop rate of NG is the same as that of MR in the
cool-down process. Meanwhile, the greater heat transfer coefficient
between MR and the wall of plate-fin structures is, the larger the
peak value of the equivalent stress is.
Fig. 23 presents the peak value of the equivalent stress vs. the
temperature drop rate at different heat transfer coefficient between
NG and the wall of plate-fin structures for temperature difference
Fig. 19. The peak value of sr vs. temperature difference at different heat transfer co- DT ¼ 10 K between NG and MR, heat transfer coefficient
efficients of NG. hMR ¼ 1.3 kW/(m2 K) between MR and the wall of plate-fin struc-
tures and fluid pressure PNG ¼ 71 bar, PMR ¼ 4 bar. Similarly, the
results show that the peak value of the equivalent stress is slightly
affected by the temperature drop rate as the temperature drop rate
of NG is the same as that of MR in the cool-down process.
Fig. 24 depicts the relation between the peak value of the
equivalent stress and the temperature drop rate at different tem-
perature differences between NG and MR for the heat transfer co-
efficient hMR ¼ 1.3 kW/(m2 K), hNG ¼ 1.5 kW/(m2 K) and fluid
pressure PNG ¼ 71 bar, PMR ¼ 4 bar. The results indicate that the
peak value of the equivalent stress also increases slightly with the
increase of temperature drop rate in the cool-down process.
To sum up, the peak value of the equivalent stress in the plate-
fin structures also increases with the increase of temperature drop
rate because the temperature maldistribution is more obvious as
the temperature drop rate is greater in the cool-down process.
However, the peak value of the equivalent stress is slightly
impacted by the temperature drop rate as the temperature drop
rate of NG is the same as that of MR in the cool-down process. This
is because the temperature maldistribution is slightly influenced by
Fig. 20. The peak value of sr vs. temperature difference at different temperature drop
the temperature drop rate due to the excellent heat conductivity
rates. performance of plate-fin structures in heat exchanger as the tem-
perature drop rate of NG is the same as that of MR in the cool-down
equivalent stress is obviously impacted by MR temperature as the process.
temperature difference between NG and MR is more than 10 K in At the same time, a cool-down experiment, in which the liquid
the cool-down process. nitrogen was used as the cryogenic medium of plate-fin heat
For main cryogenic heat exchanger in a large LNG cold-box, the exchanger, was also used to verify the simulation results. In this
boiling heat transfer and the condensation heat transfer are per- experiment, the liquid nitrogen directly flows into the heat and cold
formed in the passage of MR and NG, respectively. Due to the channels of heat exchanger. In the whole cool-down process, the
complexity and instability of the boiling heat transfer and the temperature drop rate of the fluid in the heat channel is the same as
condensation heat transfer, the temperature difference between that of the fluid in the cold channel. Table 3 lists the experiment
NG and MR will vary greatly if the operations are not reasonable for results. In the first experiment, the average temperature drop rate
the main cryogenic heat exchanger in the cool-down process. This and the local maximum temperature drop rate are 1 K/min and
will result in the crack and fatigue failure at the brazed joint of 26 K/min, respectively (the cool-down duration is defined as the
plate-fin structures. At the same time, as the temperature differ- elapsed time from pouring liquid nitrogen into the heat exchanger
ence between NG and MR is more than 10 K in the cool-down to the temperature of platinum resistance reaching that of liquid
process, the peak value of equivalent stress is obviously influ- nitrogen. The average temperature drop rate is the ratio of the
enced by MR temperature and the heat transfer coefficient between temperature difference (between liquid nitrogen and room tem-
MR or NG and the wall of plate-fin structures which is the key perature) and cool-down duration). The average temperature drop
parameter to describe the heat transfer performance of heat rate is less than the limit value, at which the heat exchanger is
exchanger. heavily allowed to suffer in a small-scale cold-box in the cool-down
process. The result shows that plate-fin structures are safe in the
cool-down process. In the second experiment, the average tem-
5.3. The influence of temperature drop rate for stress perature drop rate and the local maximum temperature drop rate
are 9.7 K/min and 265 K/min, respectively. The average tempera-
In order to further investigate the influence of temperature drop ture drop rate is far greater than the limit value. Plate-fin structures
rate for stress of plate-fin structures in the cool-down process, the are no distortion, no cracks in the brazed joints, and the strength is
1124 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Fig. 24. The peak value of sr vs. the temperature drop rate at different temperature
Fig. 21. The peak value of sr vs. temperature difference at different temperatures of
differences.
MR.

plate-fin structures as the temperature drop rate of the fluid in the


74 heat channel is the same as that of the fluid in the cold channel in
ΔΤ=10Κ,hMR=1.3kW/(m K)
2

the cool-down process. This agrees well with the simulation results.
ΔΤ=10Κ,hMR=1.0kW/(m2K) However, in the practical engineering, the temperature drop
72
ΔΤ=10Κ,hMR=0.8kW/(m2K) rate of NG may be different from that of MR in the cool-down
process of PFHE. This will result in that the temperature differ-
Stress (MPa)

70
ΔΤ=10Κ,hMR=0.5kW/(m2K)
ence between NG and MR is changed with time in the cool-down
process. Fig. 25 depicts the schematic diagram of temperature
difference change between NG and MR caused by temperature drop
68 rate of MR. TNG and TMRI are the temperature change curve in the
ideal cool-down process in which the temperature drop rate of NG
66 is the same as that of MR and remain constant in the cool-down
process. DTI is the temperature difference between NG and MR in
the ideal cool-down process. However, the NG temperature cannot
64 change synchronously due to the thermal inertia of the heat
0 5 10 15 20 25
exchanger as the temperature drop rate of MR increases in the cool-
Temperature Drop Rate
down process (as TMRP shown in Fig. 25). The temperature differ-
Fig. 22. The peak value of sr vs. the temperature drop rate at different heat transfer ence between NG and MR will increase from the temperature dif-
coefficients of MR. ference (DTI) between NG and MR in the ideal cool-down process to
DTP. This will induce stress cracking in plate-fin structures because
the temperature difference has obviously effect on the peak value of
equivalent stress in plate-fin structures. It means that the stress
cracking induced by the temperature drop rate is essentially
ascribed to the increase of the temperature difference between NG
and MR as the temperature drop rate of NG is different from the
temperature drop rate of MR in the practical cool-down process.
Therefore, the temperature difference between NG and MR is the
main reason to induce stress cracking of plate-fin structures in the
practical cool-down process.

6. Conclusions

In this paper, the stress distributions in four paths marked in


plate-fin structures are obtained from simulation results in the
cool-down process. The influence of the temperature drop rate of
NG or MR and the temperature difference between NG and MR for
stress of plate-fin structures in the cool-down process were
investigated by means of a direct thermal-stress coupling standard
Fig. 23. The peak value of sr vs. the temperature drop rate at different heat transfer FEM codes in ANSYS software. The simulation results are also
coefficients of NG. verified by the experiment data. The following conclusions can be
drawn from this study.

qualified. The above experiment results further indicate that the (1) The results obtained by numerical simulations indicate that
temperature drop rate has slight effect on the structural safety of the direct stresses are main factor to impact the strength of
H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126 1125

Table 3
The experiment results for cool-down.

No. Cool-down duration Local maximum temperature drop rate Average temperature drop rate Experiment results

min K/min K/min

1 167 26 1 (1) Strength is qualified.


2 21 265 9.7 (2) No distortion.
(3) No cracks in brazed joints.

the brazed joint. The shear stress on x-y plane is slight to different fluids and flow regimes also impact the stress of plate-fin
influence that. At the same time, the equivalent stress rea- structures in the cool-down process. But it can be ascribed to the
ches the peak value at the brazed joint near the fin side. The influence of heat transfer coefficient for the stress of plate-fin
brazing seam between the fin and the plate also suffers larger structures. In this paper, the peak value of equivalent stress is
stress than that in the adjacent region. The main reasons analyzed at different heat transfer coefficients. Therefore, these
inducing the above results are stress concentration at the conclusions can be also applied to other fluids.
brazed joint, the mismatching thermal expansion co-
efficients between filler metal and base metal and thermal References
distortion restricted by external constraints.
(2) The peak value of equivalent stress is obviously impacted by Aiyangar, A.K., Neuberger, B.W., Oberson, P.G., Ankem, S., 2005. The effects of stress
the temperature difference between NG and MR in the cool- level and grain size on the ambient temperature creep deformation behavior of
an alpha Ti-1.6 wt pct V alloy. Metall. Mater. Trans. A-Phys. Metall. Mater. Sci.
down process of heat exchanger. In order to ensure the safe 36A (3), 637e644.
and reliable operation of heat exchanger, the temperature Brown, T.B., Dauda, T.A., Truman, C.E., Smith, D.J., Memhard, D., Pfeiffer, W., 2006.
difference between NG and MR needs to be reasonably Predictions and measurements of residual stress in repair welds in plates. Int. J.
Press. Vessels Pip. 83 (11e12), 809e818.
controlled in the cool-down process of main cryogenic heat Carey, V.P., Mandrusiak, G.D., 1986. Annular film flow boiling of liquids in a partially
exchanger which is used in a large-scale LNG cold-box. heated, vertical channel with offset strip fins. Int. J. Heat Mass Transf. 29 (6),
(3) The temperature drop rate has slight effect on the stress of 927e939.
Celik, E., Sayman, O., Karakuzu, R., Ozman, Y., 2007. Numerical analysis of the in-
plate-fin structures as the temperature drop rate of the NG is fluence of buffer layer thickness on the residual stresses in YBCO/La2Zr2O7/Ni
the same as that of MR. The conclusion is also verified by a superconducting materials. Mater. Des. 28 (7), 2184e2189.
cool-down experiment. But if the temperature drop rate of Chang, K.H., Lee, C.H., 2007. Residual stresses and fracture mechanics analysis of a
crack in welds of high strength steels. Eng. Fract. Mech. 74 (6), 980e994.
NG is different from that of MR in the practical cool-down Chen, H., Gong, J., Geng, L., Tu, S.-T., 2006. Numerical analysis of thermal defor-
process, the influence of temperature drop rate for the mation and residual stress for the brazed plate-fin structure. In: ASME 2006
stress of plate-fin structures is essentially caused by the Pressure Vessels and Piping/ICPVT-11 Conference. American Society of Me-
chanical Engineers, pp. 373e381.
temperature difference between NG and MR. Therefore, in
Chen Yongdong, C.X., 2010. A technical analysis of heat exchangers in LNG plants
the practical cool-down process, the purpose to control and terminals. Nat. Gas. Ind. 01, 147e148.
temperature drop rate of MR is to control the temperature Chena, H., Gonga, J., Genga, L., Tua, S.-T., 2005. Finite element prediction of residual
difference between NG and MR. stresses and thermal distor-tion in a brazed plate-fin structure. J. Press. Equip.
Syst. 3, 118e124.
Civalek, O., 2004. Application of differential quadrature (DQ) and harmonic differ-
Although the above conclusions are obtained when the fin and ential quadrature (HDQ) for buckling analysis of thin isotropic plates and elastic
plate are AL3003 and the brazing material is AL4004, these con- columns. Eng. Struct. 26 (2), 171e186.
Galli, M., Botsis, J., Janczak-Rusch, J., 2006. Relief of the residual stresses in ceramic-
clusions can be also applied to other materials. Because the excel- metal joints by a layered braze structure. Adv. Eng. Mater. 8 (3), 197e201.
lent thermal conductivity is necessary for any heat exchanger and Jiang, W.C., Chen, H., Gong, J.M., Tu, S.T., 2011a. Numerical modelling and nano-
the vacuum brazing among different materials is a main indentation experiment to study the brazed residual stresses in an X-type
lattice truss sandwich structure. Mater. Sci. Eng. A-Struct. Mater. Prop. Micro-
manufacturing process of plate-fin structures. At the same time, struct. Process. 528 (13e14), 4715e4722.
Jiang, W.C., Gong, J.M., Chen, H., Tu, S.D., 2007. Finite element analysis of creep of
stainless steel plate-fin structure. Acta Metall. Sin. 43 (5), 539e545.
Jiang, W.C., Gong, J.M., Chen, H., Tu, S.T., 2008a. The effect of filler metal thickness
on residual stress and creep for stainless-steel plate-fin structure. Int. J. Press.
Vessels Pip. 85 (8), 569e574.
Jiang, W.C., Gong, J.M., Tu, S.D., Chen, H., 2009a. Modelling of temperature field and
residual stress of vacuum brazing for stainless steel plate-fin structure. J. Mater.
Process. Technol. 209 (2), 1105e1110.
Jiang, W.C., Gong, J.M., Tu, S.D., Fan, Q.S., 2009b. A comparison of brazed residual
stress in plate-fin structure made of different stainless steel. Mater. Des. 30 (1),
23e27.
Jiang, W.C., Gong, J.M., Tu, S.T., 2010. A new cooling method for vacuum brazing of a
stainless steel plate-fin structure. Mater. Des. 31 (1), 648e653.
Jiang, W.C., Gong, J.M., Tu, S.T., 2011b. Effect of brazing temperature on tensile
strength and microstructure for a stainless steel plate-fin structure. Mater. Des.
32 (2), 736e742.
Jiang, W.C., Gong, J.M., Tu, S.T., Chen, H., 2008b. Effect of geometric conditions on
residual stress of brazed stainless steel plate-fin structure. Nucl. Eng. Des. 238
(7), 1497e1502.
Jiang, W.C., Wang, B.Y., Gong, J.M., Tu, S.T., 2011c. Finite element analysis of the
effect of welding heat input and layer number on residual stress in repair welds
for a stainless steel clad plate. Mater. Des. 32 (5), 2851e2857.
Jiang, W.C., Yang, B., Gong, J.M., Tu, S.T., 2011d. Effects of clad and base metal
thickness on residual stress in the repair weld of a stainless steel clad plate.
J. Press. Vessel Technol.-Trans. ASME 133 (6).
Jiang, W.C., Zhang, Y.C., Woo, W., 2012. Using heat sink technology to decrease
Fig. 25. Schematic diagram of temperature difference change caused temperature residual stress in 316L stainless steel welding joint: finite element simulation.
drop rate. Int. J. Press. Vessels Pip. 92, 56e62.
1126 H. Ma et al. / Journal of Natural Gas Science and Engineering 21 (2014) 1113e1126

Khan, K.A., Barello, R., Muliana, A.H., Levesque, M., 2011. Coupled heat conduction Manglik, R.M., Bergles, A.E., 1995. Heat transfer and pressure drop correlations for
and thermal stress analyses in particulate composites. Mech. Mater. 43 (10), the rectangular offset strip fin compact heat exchanger. Exp. Therm. Fluid Sci. 10
608e625. (2), 171e180.
Kim, T.S., Kuwamura, H., 2007. Finite element modeling of bolted connections in Peng, H., Ling, X., 2008. Optimal design approach for the plate-fin heat exchangers
thin-walled stainless steel plates under static shear. Thin-Walled Struct. 45 (4), using neural networks cooperated with genetic algorithms. Appl. Therm. Eng.
407e421. 28 (5e6), 642e650.
Kong, X., Yang, Q., Li, B., Rothwell, G., English, R., Ren, X.J., 2008. Numerical study of Picard, F., Averous, D., Joulia, X., Barreteau, D., 2006. Modelling and dynamic
strengths of spot-welded joints of steel. Mater. Des. 29 (8), 1554e1561. simulation of Thermal stresses in brazed plate-fin heat exchanger. In: 16th
Kuznetsov, V.V., Shamirzaev, A.S., 2007. Boiling heat transfer for freon R21 in European Symposium on Computer Aided Process Engineering and 9th Inter-
rectangular minichannel. Heat. Transf. Eng. 28 (8e9), 738e745. national Symposium on Process Systems Engineering, vol. 21, pp. 659e664.
Lee, J.H., 1988. Some Exact and Approximate Solutions for the Modified von Mises Sen, F., Celik, E., Toparli, M., 2007. Transient thermal stress analysis of CeO2 thin
Yield Criterion. In: Preprint e American Society of Mechanical Engineers, films on Ni substrates using finite element methods for YBCO coated conductor.
November 27, 1988eDecember 2, 1988. Publ by American Soc of Mechanical Mater. Des. 28 (2), 708e712.
Engineers (ASME), Chicago, IL, USA p WA/APM22 27-WA/APM22 27. Xie, Q.Y., Ling, X.A., 2010. Numerical analysis of residual stress for copper base
Ligterink, N.E., Hageraats-Ponomareva, S.V., Velthuis, J.F.M., 2012. Mechanical brazed stainless steel plate-fin structure. J. Mater. Eng. Perform. 19 (5), 611e615.
integrity of PFHE in LNG liquefaction process. In: Barrio, M., Venvik, H.J. (Eds.), Zhao, N., Yang, Y.Q., Han, M., Luo, X., Feng, G.H., Zhang, R.J., 2012. Finite element
2nd Trondheim Gas Technology Conference, pp. 49e55. analysis of pressure on 2024 aluminum alloy created during restricting
Liu, Q., 1997. Modelling the effect of welding residual stresses on fracture toughness expansion-deformation heat-treatment. Trans. Nonferrous Metals Soc. China 22
of a welded joint. Int. J. Press. Vessels Pip. 72 (2), 103e109. (9), 2226e2232.
Liu, Z., Winterton, R., 1991. A general correlation for saturated and subcooled flow
boiling in tubes and annuli, based on a nucleate pool boiling equation. Int. J.
Heat Mass Transf. 34 (11), 2759e2766.