You are on page 1of 7

Physics Procedia 33 ( 2012 ) 498 504

1875-3892 2012 Published by Elsevier B.V. Selection and/or peer review under responsibility of ICMPBE International Committee.
doi: 10.1016/j.phpro.2012.05.095

2012 International Conference on Medical Physics and Biomedical Engineering

Comparison of Two Methods of Finite Element Modeling for
Elbows with Unequal Wall Thickness
Junhua Dong
a
, Xiangfu Bao
b
, Xize Zheng
c

a
School of Chemical Engineering and Technology, Hebei University of Technology,Tianjin,China,300130
b
Troitec Automotive Electronics Co.,Ltd
c
Tianjin Jinneg Solar Cell Co.,Ltd


Abstract
In Finite Element Stress Analysis of elbow, its unequal wall thickness can be obtained by two methods: eccentric
circle method and multi-spot spline curve drawed according to wall thickness of elbow. In this work, an elbow with
constant inner diameter was taken as an illustration and its simulation results based on these two modeling methods
were compared under different ratios of central line bend radius to mean diameter of pipe R/D. It is found that
modeling method has no effect on the stress analysis results of elbows. But eccentric circle method has the virtue of
being easier to implement and can be used without restriction of R/D, so it is an ideal method of finite element
modeling for unequal wall thickness elbows. Because the FEA results of equal thickness elbows are recognizably
higher than those of elbows with unequal wall thickness, considering non-uniform thickness of elbows is necessary to
set up a reasonable safety evaluation for elbows.

2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [name organizer]

Keywords: modeling method; elbows; unequal wall thickness ;Finite Element Analysis; ANSYS
1.Introduction
Elbows can not only redirect the fluid flow direction, and can improve flexibility of pipeline to
reduce its vibration and force of constraint, so elbows are widely used in the pipeline of modern
industry such as petrochemical industry,chemical industry and electric power. When straight pipe is
bended, outside wall becomes thinner and inside wall becomes thicker, so cross section of elbows made
by bending pipe have unequal wall thickness shape and certain ovality. Beause of the influence of
curvature and defects in manufacturing process, elbows are often the weak link of pipe system, so many
scholars used numerical simulation method to study the stress of elbows under different loads
[1-4]
. But
most of stress analysis of elbows are obtained by regarding elbows as internal pressure vessels with
Available online at www.sciencedirect.com
2012 Published by Elsevier B.V. Selection and/or peer review under responsibility of ICMPBE International Committee.
Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504 499

constant wall thickness, stress analysis for elbows with unequal wall thickness or elliptical cross section
are rare.
Because unequal thickness of the wall of elbow will influence its stress distribution, it should be
considered in stress analysis of elbow. In Finite Element Stress Analysis of elbow, its unequal thickness
of the wall can be obtained by two methods: eccentric circle, multi-spot spline curve drawed according to
wall thickness of elbow. In this paper, the effect of this two methods on stress results will be studied .
2.Two methods of finite element modeling of elbows with unequal wall thickness
2.1.Modeling method of multi-spot spline curve
The wall thickness variations can be calculated by relative distortion of longitudinal fiber of elbows
when elbows subjected to pure bendin
[5]
, namely
l
l
t
t
0
(1)
In Eq.(1), t
0
is the thichness of straight pipe under the same pressure;
0
t
t
is relative variation
of thichness;
l
l
is relative distortion of longitudinal fiber of elbows.
A lot of plastic deformation appear in the elbows during bending, so =0.5. The relative distortion of
longitudinal fiber in the circumferential direction of elbows(seen in Fig. 1) is shown in Eq.(2):

R
r
Rd
Rd d r R
l
l
m m
sin ) sin (
(2)

Rr
m
K R
r
t
t
4
sin
2
sin
0

Where K
Rr
is the ratio of central line bend radius to mean diameter, K
Rr
= R/D=R/2r
m.

Fig. 1 Elbows
The thickness of elbows at different locations in the circumferential direction is as follow:

)
4
sin
1 (
0
Rr
K
t t
3
Thus, at the location of - thickening ratio of wall thickness is as follow:
% 100
4
1
% 100
0
0
1
Rr
K t
t t

At the location of - thinning ratio of wall thickness is:
500 Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504
% 100
4
1
% 100
0
0
2
Rr
K t
t t

The diameter of inner wall circle or outer wall circle is fixed, and outer wall or inner wall can be
obtained by drawing a multi-spot spline curve through many points determined by wall thickness
according to Eq.(3). As an example, cross-sectional images of elbows with unchange inner diameter is
illustrated in Fig. 2.

Fig. 2 Modeling method of multi-spot spline curve
2.2.Modeling method of eccentric circle
The annular section with variable thickness of elbows can be obtained through two circles with
different centre , as shown in Fig.3. In Fig.3, e is the eccentric distance of two circles,
1 0
t e
or
2 0
t e
. As shown here, section thickness is gradually changing with the maximum
-

Fig. 3 Modeling method of eccentric circle
3.Computational Example
Take an elbow with unchanged inner diameter for example, cross-section of elbows will be builded
according to the modeling method of eccentric circle and multi-spot spline curve, then stress intensity
obtained in two method will be compared.
3.1.Design parameter and material property
Design pressure of pipe is 3.67 MPa, design temperature is 608 , inner paremeter is 864mm, the
ratio of central line radius to mean diameter R/D is respectively 1,1.5,2.0,2.5and 3. Pipes are made of
P92(normalizing and air cooling) and its mofulus of elasticity E is 1.06105MPa, Yield strength
s

is 300MPa and t is 66.7MPa.The fluid density in pipe is
1.05x103 kg/m3. The wall thickness of straight pipe is 50mm. minimum thickness and maximum
thickness of elbow can be calculated according to the Eq.(3).
Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504 501

3.2.Finite-element analysis of pipes
3.2.1.Finite element model
Considering the symmetry of structure and loads, the finite element model of 1/4 structer are created.
The inequal wall thickness of elbows can be obtained by the method of eccentric circle and multi-spot
spline curve, respectively. The length of straight pipe is 6 times outer paremeter of straight pipe which is
much larger than the attenuation length of edge stress. According to the welding groove size, transition
section is installed between elbow and straight pipe. Three-dimensional 20 nodes solid elements
SOLID95 is selected for struture, and there are 4410 elements and 22926 nodes in the finite element
model of structure.
3.2.2.Loads and boundary conditions
To simulate internal pressure, uniform surface force pi is applied on inner wall surface of structure,
whose value is 6.67 MPa.The equivalent axial tension Pc is applied at the end of stright pipes, whose
value can be calculated as follow: .
MPa
R R
PR
P
i
i
c
44 . 26
420 470
420 67 . 6
2 2
2
2 2
0
2

Symmetry constraints are applied on the symmetrical plane of structure and the hoop constraint is
applied at the end of stright pipes to prevent model from moving.
3.2.3.Calculation result and the stress evaluation
Contour maps of stress intensity of elbows with 1.5 R/D are illustrated in Fig.4. As it show, the stress
distribution are almost identical between two modeling methods and maximum stresses all appear in
the inner wall at 45 ,whose value is 76.621MPa, 75.697MPa, respectively.


(a) Modeling method of eccentric circle
502 Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504


(b) Modeling method of multi-spot spline curve
Fig. 4 Contours of stress intensity and position of stress assessment line of the unequal thickness elbow R/D=1.5
As Fig.4 shown, stress assessment line is set through the point with maximum TRESCA stress in
contours of stress intensity. The primary general membrane stress intensity SI and the primary stress plus
secondary stress intensity S are listed in Table 1. It can be seen from Table 1, either general membrane
stress intensity or general membrane stress plus bending stress intensity obtained respectively with two
modeling method of unequal wall thickness elbows is roughly the same and all are lesser than those of
equal wall thickness elbows
Table 1. Results of stress classification and results of assessment of stress intensity
stress intensity
calculation valuesof stress intensity MPa
the allowable limit of
design stress intensity
MPa
results of assessment
eccentric
circle
multi-spot
spline curve
constant wall
thickness
eccentric
circle
multi-spot
spline curve
constant wall
thickness
the primary general
membrane stress intensity
S
I

66.65 66.27 77.07 Sm=66.7 safety safety unsafety
the primary stress plus
secondary stress intensity
S
75.64 75.25 86.77 3Sm=200 safety safety safety
The parametric finite element model (FEM) of this structure was established using ANSYS
parametric design language (APDL) and the stress of elbows with different R/D are studied. Comparing
the results of two method, the largest stress and its site are almost always the same. The stress intensity
will be obtained by making assessment line of stress linearization through the max TRESCA stress and
they are illustrated in Fig.5.
As Fig.5 shown, in the range of parameters for calculation, either general membrane stress or general
membrane stress plus bending stress is almost the same.
Compared with variable thickness elbows, the FEA results of uniform thickness elbows are larger, and
general membrane stress is 12.7% larger and general membrane stress plus bending stress is 11.8% than
those of variable thickness elbows. That is because stresses of elbows under internal pressure are not
evenly distributed, but larger inside and smaller outside
[6]
. The shape of cross section of unequal thickness
elbows keeps coherent with the distribution regularity of stress, so unequal thickness is a favourable cross
section for elbows. According to stress criteria in JB 4732-1995, general membrane of unequal thickness
elbow with R/D=1 don t satisfies the strength requirement, but that of equal thickness elbow with every
R/D can t meet the needs of strength. So it is essential to consider the wall thickness variations in stress
Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504 503

analysis for elbows.
!.0 !. ?.0 ?. 3.0
b
0

80
8
90
9
_

u
'

m

m
|

u
u


:
t

:
:
lD
ut` `'
mu't`:ot :'`u uv
uu`om tl`ku::

(a) General membrane stress
1.0 1.5 2.0 2.5 3.0
65
70
75
80
85
90
95
100
105
110
_

u
'

m

m
|

u
u


:
t

:
:

'
u
:

|

u
d
`
u
_

:
t

:
:
lD
ut` `'
mu't`:ot :'`u uv
uu`om tl`ku::

(b) General membrane stress plus bending stress
Figure 5. Stresses of elbows with different R/D
Just as a slight clarification, the thinning ratio of wall thickness 2 of elbows with small R/D(R/D
2) calculated by Eq.(1) exceeds the requirment of no more than 12.5% in GB12459-2005. So the
modeling method of multi-spot spline curve applies only to those elbows with great R/D. In this paper,
only for comparison, the maximum thickness and minmum thickness are taken for the same values in
both modeling methods. But in fact, the thinning ratio of wall thickness 2 of elbows is not confined to
Eq.(1) and can be defined by measureing actual value or required value. Consequently, the modeling
method of eccentric circle is not limited by R/D and is simple and easy to carrying out.
4.Conclusion
According to the manufacturing process characteristics, the thickness of elbows made by bending
straight pipe gradually reduce from inside to outside. The non uniform thick of elbow is bound to
change the stress distribution in elbow and will have an impact on its the result of stress analysis. In this
paper, an effect of two methods of finite element modeling of elbows with unequal wall thickness on
504 Junhua Dong et al. / Physics Procedia 33 ( 2012 ) 498 504
calculated results is analyzed and these calculated results are campared with those of uniform thickness
elbows. The conclusions ss followed:
The stress of uniform thickness elbows are greater than those of elbows with unequal wall
thickness. For the structure of the discussion, general membrane of equal thickness elbow with
different R/D all do not satisfy the intensity evaluation conditions, but that only occur in the
unequal thickness elbow with small R/D. So it is essential to consider the wall thickness
variations of elbow in stress analysis for correctly evaluating its safety.
In the range of parameters for calculation, either general membrane stress or general membrane
stress plus bending stress for both two methods of finite element modeling of elbows is almost
the same, so the results of stress intensity evaluation are no difference btween them. But
eccentric circle method lu: tl v`tu o |`u_ u:` to `m'mut and can be used
without restriction of ratios of curvature and diameter, so it is an ideal method of finite element
modeling for elbows with unequal thickness of the wall.
References
[1] Su, Baoqing; Tian, Qi; Yang, Xiaohe; Wang, Zhimin. Test and study of elbow stress of directly buried heat-supply
pipeline[C], Energy and the Environment - Proceedings of the International Conference on Energy and the Environmental, 2003, v1,
p 377-379.
[2] Wang Zhangqi, Zhang Zuguo, Feng Yanting , et al.Bend Stress Analysis Under Different Loads Based on Finite Element
Method [J], HEBEI ELECTRIC POWER,2009,28(5): 29-33(Ch).
[3] SUN Lan ping, ZHAO Jian ping. Influence of the non-uniform thickness elbow on the stress distribution by FEA method[J],
JOURNAL OF NANJING UNIVERSITY OF TECHNOLOGY, 2006 28 3 49-52(Ch).
[4] Basavaraju, C. Lee, R.L. Stress distribution in elbow due to external moments using finite-element methodology[J], Seismic
Modal Analysis and System Interaction, 1993, v 249, p 103-112.
[5] DING Bomin, HUANG Zhenglin. Design of Chemical Equipment- chemical pressure vessel[M]. Beijing: Chemical Industry
Press, 2003.118-119 (Ch).
[6] YANG Peiliang. About Technologic Allowance for Bend Wall Thickness Calculation[J]. BOILER TECHNOLOGY, 2000,
31(11):16-20 (Ch).

You might also like