You are on page 1of 41

Shape Optimization of the Hole in an Orthotropic Plate

Shaojie Wang*
Renewable Energy School, North China Electric Power University, Beijing, China
Aizhong Lu
Renewable Energy School, North China Electric Power University, Beijing, China
Xiaoli Zhang
Renewable Energy School, North China Electric Power University, Beijing, China
Ning Zhang
Renewable Energy School, North China Electric Power University, Beijing, China
Address correspondence to Shaojie Wang, Renewable Energy School, North China Electric
Power University, No. 2 Beinong Road Huilongguan, Changping District, Beijing 102206,
China. E-mail: wangshaojie@ncepu.edu.cn

Abstract
The exploration in this work is how to minimize the stress concentration around the edge of the
hole in an orthotropic plate. The study first presents the analytical solution of the stress
distribution around arbitrary holes using the complex variable method and then carries out the
shape optimization using the mixed penalty function method. In the optimization process,

optimal holes and stress distributions under the different factors are investigated, i.e., the loading,
the Youngs modulus and the fiber direction. Finally, we come to the conclusion that in the
biaxial compressive load state, the shape and the stress are mainly affected by the loading,
followed by the fiber direction and the Youngs modulus. In the pure shear condition, all three
factors determine the optimum results.
Keywords: conformal mapping, mixed penalty function method, orthotropic plates, shape
optimization, stress concentration
1. INTRODUCTION
The orthotropic plate with various holes is being used in an increasing number of fields,
such as aerospace, automobiles, ships, construction, mining, and bio-engineering. Usually, the
stress concentration will occur along the edge of the hole as the plate is subjected to loads
(Figure 1). If the strength of the material is not high enough, failure will undoubtedly occur,
usually from the region near the hole. Therefore, it is necessary to analyze the stress distribution
around the hole for structural design.
Many studies have been carried out for various types of holes, such as circular (Ozbay
and Ozer 2005), elliptical (Hufenbach, Schaffer, and Herrmann 1990; Gao 1996; Engels,
Zakharov, and Becker 2001), triangular (Theocaris and Petrou 1986; Sadegh 1988; Daoust and
Hoa 1991; Ukadgaonker and Rao 1999), rectangular (Jong 1981; Rao et al. 2010; Chauhan and
Sharma 2015) and some irregular shaped holes (Ukadgaonker and Kakhandki 2005). Most of
2

these works focus on the stress distributions under different loads or material properties, but they
do not address the problem of whether the stress concentration degree is acceptable for a certain
material strength, and if not, how can we reduce the stress concentration? To achieve this goal
and ensure the structural safety, we can optimize the shape of the hole. There have been some
studies on shape optimization for isotropic materials (Cherepanov 1974; Bjorkman and Richards
1976; Kristensen and Madsen 1976; Bjorkman and Richards 1979; Richards and Bjorkman 1980;
Dhir 1981; Lu et al. 2014). In the optimization process, many methods are available, for
example, the photo-elastic test method, the finite element method (FEM) (Week and Stelnke
1983; Lee, Kikuchi, and Scott 1989; Lin and Lin 1999) and the complex variable method. The
first two methods are not very efficient because the photo-elastic experiments require a large
number of tests and the FEM needs a relatively large amount of computation. Especially for
FEM, in every iterative step of the hole optimization, most parameters, such as the mesh, the
external loads, and the boundary conditions must first be reset before calculating the stress
distribution. Thus, the amount of computation will be larger if the process of the optimization is
greater. On the contrary, the conformal mapping in the complex variable method does not need
to consider this problem. It is therefore convenient and efficient in solving the problem of the
hole structure and is widely used in many research fields.
In general, the process for obtaining the optimal shape of a hole requires solving a series
of forward problems, although the optimization of the hole shape is an inverse geometry

problem. From the subsequent analysis, we learn that the process for determining the optimal
shape of a hole is equivalent to the process for solving a series of forward problems. In solid
mechanics, a forward problem is mainly about analyzing the stress and displacement fields for an
object under certain conditions like the geometry, material properties, and external loads.
As mentioned earlier, most of the optimum works are on isotropic materials. For
orthotropic materials, some researchers acquired optimal results using the FEM (Pedersen,
Tobiesen, and Jensen 1992). There were also some analytical results for similar problems.
Sharma, Patel, and Trivedi (2014) obtained the best fiber angle in a perforated orthotropic plate
and the stacking sequence in a symmetric composite plate containing an elliptical hole when it is
subjected to in-plane loading. However, under the circumstance where the material properties are
unchanged, the question of what type of hole perforating the anisotropic plate induces the
minimum stress concentration was not addressed.
This research will carry out the shape optimization of a hole in an infinite orthotropic
plate. Based on the method presented by the author to solve the stresses around an arbitrary
shaped hole (Lu et al. 2015), combined with the mixed penalty function optimization method, we
acquire the optimal shapes that have the minimum stress concentration along the edge of the hole
under different conditions.
2. SOLUTIONS FOR THE STRESS DISTRIBUTION
2.1 Conformal Mapping Functions
4

The conformal mapping method can be used to solve the problem in this paper
(Lekhnitskii 1968). For the problem of a hole in an isotropic plate, only one mapping function

z = ( ) is required. However, for the problem of a hole in an anisotropic plate, three mapping
functions (i.e., z = ( ) , z1 =1 (1 ) and z2 =2 ( 2 ) ) are involved (Lu et al. 2015).
The general expression of the mapping function that transforms the outer region of the
hole in the z -plane into the outer region of the unit circle in the -plane can be given as

z = ( )=R Ck k
k 1

(1)

where R and Ck reflect the size and shape of the hole, respectively. As the
cross-section of the hole is known, R and Ck can be determined.
The expressions of 1 ( 1 ) and 2 ( 2 ) are given by Lu et al. (2015),
n
1 n

z1 1 ( 1 ) 1R 1 Ck 1k 1 R C k 1k (2)
k 1

1 k 1

n
1 n

z2 2 ( 2 ) 2 R 2 Ck 2k 2 R C k 2k
k 1

2 k 1

(3)

where

1 (1 i1) / 2, 2 (1 i2 ) / 2, 1 (1 i1) / 2, 2 (1 i2 ) / 2 (4)


1 1 i1 (5)
2 2 i2

(6)

where k and k ( k 1, 2 ) are constants related to the material properties and 1 0 ,

2 0 .
2.2 Basic Equations
For an orthotropic plane of linear elasticity, when the coordinate axis z is perpendicular
to one plane of elastic symmetry, the compatibility equation without considering the body forces
can be given as (Muskhelishvili 1963; Lekhnitskii 1981)

a22

4 F
4 F
4 F
4 F
4 F

2
a

(2
a

a
)

2
a

a
0 (7)
26
12
66
16
11
x4
x3y
x2y 2
xy3
y 4

where aij are constants related to the material properties (see Appendix). The solutions
of Eq. (7) are related to the roots of the following equation:

a11 4 2a16 3 (2a12 a66 ) 2 2a26 a22 0 (8)


The four roots of Eq. (8) must be conjugate complex roots (i.e., 1 , 1 , 2 and 2 ).

1 and 2 here are exactly the same as those expressed by Eqs. (5) and (6), respectively.
Here, 1 and 2 are material parameters in the global rectangular coordinate system.
When the main material directions (fiber direction) are along the x and y -directions in the
local rectangular coordinate system xoy , we can easily obtain the material parameters 1
and 2 . In this paper, the fiber direction (material main direction) is defined as the x
direction, as shown in Figure 2. The values of 1 and 2 corresponding to the global
rectangular coordinate system xoy can be respectively given as (Lekhnitskii 1981)

1 cos sin
(9)
cos 1 sin

2 cos sin
(10)
cos 2 sin

where is the angle between x and the x -axis.

The solution of Eq. (7) can be seen as solving the stress boundary conditions, given by
Eqs. (11) and (12).

2Re 1 ( z1 ) 2 ( z2 ) f1

(11)

2Re 11 ( z1 ) 22 ( z2 ) f2

(12)

where z1 and z2 are points on the boundary.

f1 and f 2 in Eqs. (11) and (12) can be determined according to the loads acting along
the edge. In this study, there exist no loads along the edge of the hole, so f1 f2 0 . 1 ( z1 )
and 2 ( z2 ) can be given in the following forms:

1 ( z1 ) B* z1 10 ( z1 )

(13)

2 ( z2 ) B'* iC '* z2 02 ( z2 )

(14)

where B * , B '* and C '* can be determined according to the stress components acting at
infinity (i.e., x , y and xy ) as

x (22 22 ) y 22 xy
B
2 (2 1 )2 (22 12 )
*

(15)

B'*

C
'*

(12 12 21 2 ) y x 2 2 xy

(16)

2 (2 1 )2 (22 12 )

(1 2 ) x 2 (12 12 ) 1 ( 22 22 ) y (12 12 ) ( 22 22 ) xy
22 ( 2 1 )2 (22 12 )

(17)

10 ( z1 ) and 02 ( z2 ) in Eqs. (13) and (14) should have the forms:


n

10 ( z1 ) 10 1 ( 1 ) 1*0 ( 1 ) ak 1 k

(18)

k 0

02 ( z2 ) 02 2 ( 2 ) 02* ( 2 ) bk 2 k

(19)

k 0

where ak , bk ( k 0,..., n ) can be determined by Eqs. (11) and (12) (Lu et al. 2015) [20].
2.3 Solution for the Stress
From Eqs. (2), (3), (13)(19), we have

1 ( z1 ) B* kak 1 k 1 / 1R 1 kCk 1 k 1 +1 R
n

k 1

k 1

1
2
1

2 ( z2 ) B'* iC '* kbk 2 k 1 / 2 R 1 kCk 2 k 1 + 2 R

k 1

k 1

k 1

kC k 1k 1 (20)

2
2

k 1

kC k 2 k 1 (21)

The stress components in Cartesian coordinates can be determined by substituting Eqs.


(20) and (21) into the following equations:
x 2 Re 12 1 ( z1 ) 22 2 ( z2 )

y 2Re 1 ( z1 ) 2 ( z2 )
xy 2Re 11 ( z1 ) 22 ( z2 )

(22)
(23)
(24)

Then, the stress components , and in orthogonal curvilinear coordinates


can be calculated by the two equations (Muskhelishvili 1963):

x y (25)
2i ( y x 2i xy )

2 ( )
2 ( )

(26)

Along the edge of hole where 1 2 ei and 1 , the stresses in these


points can be solved very easily by Eqs. (20)(26). In this paper, the compressive stress is
positive, and the tensile stress is negative.
3. THE PROCESS OF SOLVING AN INVERSE PROBLEM
Determining the optimal shape of the hole is similar to solving an inverse problem. The
material properties and the load acting at infinity for both the inverse and forward problems are
known. However, for the inverse problem, Ck (k 1,,n) in Eq. (1) are unknown coefficients
to be determined.
Given the external load, x , y and xy , and the material parameters 1 and 2
from Section 2, the distributions of the tangential stress for any point on the edge of the hole are
dependent on the values of Ck (k 1,,n) . For the given initial values of Ck (k 1,,n) , i.e.,

Ck(0) (k 1,, n) , the maximum absolute tangential stress max along the edge of the hole
can be calculated using the method for solving the forward problems (Section 2). By repeatedly
revising Ck(0) (k 1,, n) , a group of Ck* (k 1,,n) values to minimize the value of max

can finally be obtained. The cross-sectional shape determined by Ck* (k 1,,n) is the optimal
shape, which has the minimum stress concentration.
In mathematics, this problem is a minimization problem in optimization techniques, for
which the corresponding objective function can be given as

F ( X ) min max
[0,2 ]

(27)

In this paper, the mixed penalty function method is used to optimize the calculations
(Fletcher 1987; Huang and Wang 2012), where the mathematical model can be expressed as

F ( X ), X D Rn

(28)

D : g j ( X ) 0, j 1,2,..., m
hj ( X ) 0, j m 1, m 2,..., p
The penalty function form is written as

P ( X ) F ( X ) r ln
jI1

1
1
1
2
min[0, g j ( X ) ]
g j ( X ) jI2 r
r

h (X )

j m 1

2
j

(29)

where I1 j g j ( X ) 0;1 j m , I 2 j g j ( X ) 0;1 j m and r is the penalty


factor.
In this study, F ( X ) min max and X D [C1, C2 ,..., Cn ] , where C1 , C2 ,, and
[0,2 ]

Cn are the design variables.


The inequality constraint is

10

| C
k 1

| 1 (30)

In Eq. (30), Ck are complex numbers, i.e., Ck Ck1 iCk 2 , where Ck1 and Ck 2 are
the real and imaginary parts of Ck , respectively.
When optimizing the shape of the hole, it does not lose the generality for R 1.0
because it is equivalent to a hole that rotates around an angle in the global coordinate system

xoy when R is the complex number ei .

4. EXAMPLES OF OPTIMIZATION
4.1 The Parameters and the Convergence
In the optimization process, five complex coefficients of the mapping function, C1 , C2 ,
, C5 , are taken as the design variables. The reason we take five coefficients is that the results
are not significantly different if more coefficients are selected. R is equal to the real number

1.0 , and the loads at infinity are x P , y Q and xy 0 .


The optimization examples are based on the following three cases:
1.

Constant fiber orientation and material parameters but different loads. Take the vertical
load P 1MPa and ratios between the horizontal and vertical loads Q / P 0 , 1, 1 ,

1.5 , 2.5 and 5 .


2.

Constant load and fiber orientation but different Youngs modulus in the y -direction.
The constant material parameters are Ex ' 20 GPa , Gx ' y ' 15 GPa and x ' y ' 0.21 , and
11

the ratios of the elastic modulus in the y -direction to that in the x -direction are

Ey ' / Ex ' 0.5 , 2 , 3 and 5 .


3.

Constant load and Youngs modulus but different fiber directions. Take the clockwise
angle between the fiber ( x -direction) and x -axis in global rectangular coordinates, i.e.,
the fiber angle = 0, 15, 30, 45, 60, 75 and 90.
In the process of optimization, the initial values of the five complex coefficients C1 , C2 ,

, C5 are all set to zero. In other words, the initial hole shape is circular. The absolute value of
the convergence precision for the design variables is set to 1.0107 . After a number of
iterations, the convergence precision can finally be satisfied. The corresponding values of C1 ,

C2 , , C5 are optimal, and the shape determined by the five coefficients is optimal, with the
minimum tangential stress concentration along the hole edge.
For instance, under the condition of Q / P 1 , 0

and Ey ' / Ex ' 5 , the

convergence can achieve the prescribed precision after 1230 iterations. The process of iteration is
shown in Figure 3. The value of Max along the edge of the hole shows a generally
decreasing trend. In the process, the penalty function effectively corrects the deviations of the
design variables.
4.2 Results and Discussion

12

Case 1: Figure 4 shows the optimizations in different loading conditions when = 0 and

Ey ' / Ex ' 2 . The optimal shapes are shown in Figure 4a, and the corresponding distributions of
the tangential stress on the hole edge are drawn in Figure 4b.
The optimal results indicate that the loading has a significant effect on the hole shape.
When Q / P 0 , the optimal hole is a lanky olive with a ratio of the semi-major to the
semi-minor axis Rx / Ry 1.8193/ 0.1192=15.26 , which is more like a crack in the plate. The
stress distribution at the hole edge is nearly a horizontal line with a value nearly equal to

P 1 MPa , and it sharply decreases to 1 MPa near = 180 and 360, which are the cuspidal
points of the shape. Here, is the angle in the -plane. Under the pure shear state ( Q / P 1
), the optimal shape is a symmetric rounded rectangle. Not only compressive stress but also
tensile stress occurs on the edge, and their absolute values are equal. Four maximum
compressive stress points are symmetrically located near the corners, and the maximum tensile
stress occurs at the middle parts of the top and bottom sides (see Figure 4a).
In the compressive load state, the optimal shape is close to a circle when Q / P 1 and
to an ellipse when Q / P 1.5 , 2.5 and 5. Comparisons between standard ellipses (circle) with a
ratio of the major to the minor axes being Q / P and the optimal shapes are shown in Figure
4a. From the calculation data, the major axis and the maximum external load are in the same
direction. Meanwhile, the ratio of the optimal shapes semi-major axis to the semi-minor axis is
close to Q / P only when the biaxial external loads are close. When Q / P is relatively large,

13

the ratio of the semi-axes is smaller than Q / P (Table 1). Compared with the optimum hole in
an isotropic plate (Kristensen and Madsen 1976), the ratio between the major axis and minor axis
is theoretically equal to Q / P . The orthotropic material shows little difference. For the stress
aspect, from the existing theories of the isotropic plate, we know that the maximum tangential
stress along the optimal hole edge equals Q P , which is illustrated by the dotted lines in
Figure 4b. However, for an orthotropic material, the maximum tangential stress is slightly larger
than Q P , except for in two high-curvature areas where the tangential stress decreases.

Theoretically, in the global rectangular coordinate system of this paper, the imaginary
parts of the mapping function coefficients are equal to zero, as the shape is symmetrical about
the x -axis. When the hole is symmetrical about the y -axis, even the real terms of the mapping
function coefficients should be equal to zero. Although our calculation results are not zero, they
have a precision of 104 .
Case 2: In this case, we mainly discuss the effect of the Youngs modulus on the optimal
results. Figure 5 illustrates the results when we take = 30 and Q / P 2.5 . It can be noted
from Figure 5 that the Youngs modulus has little effect on the shape but a relatively significant
effect on the tangential stress. The maximum tangential stress decreases but the minimum
tangential stress increases as Ey ' / Ex ' increases from 0.5 to 5, meaning that the tangential stress
distribution becomes more uniform (see Figure 5b and Table 2).

14

Figure 6 provides the optimal results under the pure shear condition, Q / P 1 . When

= 90, the hole shapes are symmetrical rounded rectangles about the x and y -axes with the
height/width ratio increasing with Ey ' / Ex ' from 0.5 to 5 (see Figure 6a). Instead, when = 0,
the height/width ratio decreases with Ey ' / Ex ' . For other fibers, the optimal holes are not
symmetrical, and the shapes are not affected so significantly as when = 0 and 90. According
to the calculations, the Youngs modulus has the minimal effect on the shape when = 45. The
tangential stress distributions are illustrated in Figure 6b, and the maximum tangential stresses
are listed in Table 3. It can be seen that compressive and tensile stresses of the same magnitudes
can be found on the edge of the hole. The maximum compressive and tensile stresses, occurring
near the corners, increase as Ey ' / Ex ' increases from 2 to 5 . There also exist four zero-stress
points on the hole edge for each Ey ' / Ex ' , which are located near = 45, 135, 225 and 315,
respectively.
Case 3: In this case, we will discuss the effect of the fiber angle on the optimal results.
Figure 7 illustrates the optimal results under different when Q / P 5 and Ey ' / Ex ' 0.5 .
We can see that under biaxial compressive stresses, the fiber angle has a certain effect on the
shape. All the optimal shapes are ellipse-like, with nearly the same major and minor axes (see
Figure 7a). The shapes are symmetrical about the x and y -axes only for = 0, 90 and

= 45, Q / P 1 , but not for any other angles. From the calculations and Figure 7b, it can be
seen that only the compressive stress occurs around the hole, and the maximum values are

15

negligibly influenced, and they are all approximately 6.3MPa as the fiber angle ranges from 0
to 90. However, the minimum values near the points = 9 or 270 are greatly different for
different fiber angles. In other words, the range of the stress fluctuation Max Min is
different with . From all the samples we calculated, the smallest range occurs when = 45.
However, the largest range varies according to the elastic modulus. Specifically, the range is
widest at = 90 when Ey ' / Ex ' 1 and = 0 when Ey ' / Ex ' 1 as shown in Table 4.
In the pure shear state, the fiber angle has a powerful influence on the optimal shape (see
Figure 8). In Figure 8, we take a constant Ey ' / Ex ' 5 . As the fiber rotates from 0 to 90, the
optimal shape turns from a horizontal symmetric rounded rectangle to a tilted ellipse-like shape
and finally to a vertical symmetric rounded rectangle (see Figure 8a). Similar to what we
discussed earlier, under a pure shear stress, both compressive and tensile tangential stresses will
occur along the hole edge, and they have the same magnitudes (see Figure 8b and Table 5). The
results also show that the magnitude initially increases, reaching its maximum value at = 45,
and then decreases (see Table 5). In this case ( Ey ' / Ex ' 5 ), the maximum compressive
tangential stress reaches its minimum value 3.9727 MPa at = 90 and maximum value

4.3342 MPa

at = 45. Similar conclusions can also be verified for

Ey ' / Ex ' 0.5 , Table 5).


4.4. Comparisons

16

Ey ' / Ex ' 1 (

In an orthotropic plate, the stress concentration can effectively be reduced by using the
optimal hole. To prove this point, two examples are illustrated in this section. Under the
conditions of = 0, Ey ' / Ex ' 3 and Q / P 2.5 , the results of an optimal hole are compared
with those of a standard ellipse of which the semi-axis ratio is 2.5 (see Figure 9). Although the
optimal hole is very close to that of a standard ellipse, the tangential stress distribution of the
optimal hole is more uniform, and the stress concentration is reduced by 7.62% compared to that
of a standard ellipse.
Figure 10 illustrates the results of the optimal hole and a circle under the pure shear
condition when = 45 and Ey ' / Ex ' 3 . In this case, the tangential stress distributions of the
two holes are similar, and the tangential stress for the optimal hole is 12.43% smaller than that
for the circle. From the two comparisons in Figures 9 and 10, there are only tiny differences
between the shapes of the optimal holes and the corresponding selected standard holes, but the
tangential stress concentration is still effectively alleviated.
5. CONCLUSIONS
To reduce the stress concentration around a hole in an orthotropic plate, shape
optimization is carried out by using an analytical method and a mathematical optimization
method. The optimal shapes and the corresponding tangential stress distributions at the hole edge
are obtained. In the calculation process, the governing equations for the shape and stress are

17

written as FORTRAN codes, and the results can therefore be acquired in a quick and accurate
manner. The following conclusions are obtained:
The loading has a great influence on the optimal shape. Under the pure shear condition,
the optimum holes, mainly rounded rectangles or irregular ellipse-like shapes, are significantly
affected by the fiber angle and Youngs modulus. Specifically, when the fiber and the loads are
in the same direction, the holes are symmetric rounded rectangles. Not only compressive stress
but also tensile stress occurs under pure shear stress, and their magnitudes are equal. However,
for a certain fiber angle or Youngs modulus, the maximum tangential stresses occurring on the
hole edge are affected. A suitable elastic modulus and fiber direction can reduce the degree of
the stress concentration of the optimal shape, i.e., decreasing the anisotropy of the material and
setting the fiber in the horizontal or vertical direction.
Under biaxial compressive stresses, the optimal shape is mainly affected by the fiber
direction, followed by the Youngs modulus. The optimal holes are approximately circular or
elliptical, with the ratio of the semi-major to the semi-minor axis nearly equal to that of the
loading in the corresponding direction when the biaxial external loads are close. If there is a
great difference between the biaxial external loads, the semi axis ratio is smaller than the ratio of
the loads. The optimal shape can effectively reduce the stress concentration. Considering the
stresses, the maximum tangential stress along the edge approximates the sum of the external
loads. Where the curvature is greatest, the stress decreases rapidly. Different from the pure shear

18

state, increasing the degree of the materials anisotropy can reduce the stress concentration, but
altering the fiber direction does not significantly improve the stress concentration.
Acknowledgements
The study is supported by the Natural Science Foundation of China (Grant Nos.
11172101 and 11572126) and the Fundamental Research Funds for the Central Universities
(Grant No. NCEPU2016XS59).

References
Bjorkman, G. S., and R. Richards. 1976. Harmonic hole-An inverse problem in elasticity. ASME
Journal of Applied Mechanics 43:41418. doi:10.1115/1.3423882
Bjorkman, G. S., and R. Richards. 1979. Harmonic hole for nonconstant field. ASME Journal of
Applied Mechanics 46:57376. doi:10.1115/1.3424608
Chauhan, M. M., and D. S. Sharma. 2015. Stresses in finite anisotropic plate weakened by
rectangular hole. International Journal of Mechanical Sciences 101102:27279.
doi:10.1016/j.ijmecsci.2015.08.007
Cherepanov, G. P. 1974. Inverse problems of a plane theory of elasticity. Journal of Applied
Mathematics and Mechanics 38:91531. doi:10.1016/0021-8928(75)90085-4
Daoust, J., and S. V. Hoa. 1991. An analytical solution for anisotropic plates containing
triangular holes. Composite Structures 19:10730. doi:10.1016/0263-8223(91)90018-t
Dhir, S. K. 1981. Optimization in a class of hole shapes in plate structures. ASME Journal of
Applied Mechanics 48:90508. doi:10.1115/1.3157754
Engels, H., D. Zakharov, and W. Becker. 2001. The plane problem of an elliptically reinforced
circular hole in an anisotropic plate or laminate. Archive of Applied Mechanics 71:601
12. doi:10.1007/s004190100167
Fletcher, R. 1987. Practical methods of optimization, 2nd ed. New York: Wiley-Interscience.
Gao, X. L. 1996. A general solution of an infinite elastic plate with an elliptical hole under
biaxial loading. International Journal of Pressure Vessels Piping 67:95104.
doi:10.1016/0308-0161(94)00173-1
Huang, C. C., and S. Wang. 2012. A penalty method for a mixed nonlinear complementarity
problem. Nonlinear Analysis 75:58897. doi:10.1016/j.na.2011.08.061

19

Hufenbach, W., M. Schaffer, and S. Herrmann. 1990. Calculation of the stress and displacement
field of anisotropic plates with elliptical hole. Ingenieur Archiv 60:50717.
Jong, T. D. 1981. Stresses around rectangular holes in orthotropic plates. Journal of Composite
Materials 15:31128.
Kristensen, E. A., and N. F. Madsen. 1976. Madsen N. F. On the optimum shape of fillets in
plates subjected to multiple in-plane loading cases. International Journal for Numerical
Methods in Engineering 10:100719. doi:10.1002/nme.1620100504
Lee, M. S., N. Kikuchi, and R. A. Scott. 1989. Shape optimization in laminated composite plates.
Computer
Methods
in
Applied Mechanics
& Engineering
72:2955.
doi:10.1016/0045-7825(89)90120-5
Lekhnitskii, S. G. 1968. Anisotropic plates. New York: Gordon and Breach.
Lekhnitskii, S. G. 1981. Theory of elasticity of an anisotropic body. Moscow: Mir Publishers.
Lin, C. C., and C. H. Lin. 1999. Stresses around pin-loaded hole in composite laminates using
direct boundary element method. International Journal of Solids and Structures 36:763
83. doi:10.1016/s0020-7683(98)00045-6
Lu, A. Z., H. Y. Chen, Y. Qin, and N. Zhang. 2014. Shape optimization of the support section of
a
tunnel
at
great
depths.
Computers and Geotechnics
61:19097.
doi:10.1016/j.compgeo.2014.05.011
Lu, A. Z., N. Zhang, X. L. Zhang, and W. S. Li. 2015. Analytic method of stress analysis for an
orthotropic rock mass with an arbitrary-shaped tunnel. International Journal of
Geomechanics 15:04014068. doi:10.1061/(asce)gm.1943-5622.0000408
Muskhelishvili, N. I. 1963. Some basic problems of the mathematical theory of elasticity.
Groningen: Noordhoff.
Ozbay, M., and D. Ozer. 2005. The analysis of elasto-plastic stresses in the composite laminate
with a circular hole subjected to in-plane loads by means of finite element method.
Journal
of
Reinforced
Plastics
and
Composites
24:62131.
doi:10.1177/073168405045022
Pedersen, P., L. Tobiesen, and S. H. Jensen. 1992. Shapes of orthotropic plates for minimum
energy concentration. Mechanics of Structures and Machines 20:499514.
doi:10.1080/08905459208905178
Rao, D. K. N., M. R. Babu, K. R. N. Reddy, and D. Sunil. 2010. Stress around square and
rectangular cutouts in symmetric laminates. Composite Structures 92:284559.
doi:10.1016/j.compstruct.2010.04.010
Richards, R., and G. S. Bjorkman. 1980. Harmonic shapes and optimum design. ASCE Journal of
Engineering Mechanics 106:112534.
Sadegh, A. M. 1988. On the Greens Functions and boundary integral formulation of elastic
planes with cutouts. Mechanics of Structures and Machines 16:293311.
doi:10.1080/08905458808960265

20

Sharma, D. S., N. P. Patel, and R. R. Trivedi. 2014. Optimum design of laminates containing an
elliptical hole. International Journal of Mechanical Sciences 85:7687.
doi:10.1016/j.ijmecsci.2014.04.019
Theocaris, P. S., and L. Petrou. 1986. Stress distributions and intensities at corners of equilateral
triangular holes. International Journal of Fracture 31:27189. doi:10.1007/bf00044050
Ukadgaonker, V. G., and V. Kakhandki. 2005. Stress analysis for an orthotropic plate with an
irregular shaped hole for different in-plane loading conditionsPart 1. Composite
Structures 70:25574. doi:10.1016/j.compstruct.2004.08.032
Ukadgaonker, V. G., and D. K. N. Rao. 1999. Stress distribution around triangular holes in
anisotropic plates. Composite Structures 45:17183. doi:10.1016/s0263-8223(99)00024-0
Week, M., and P. Stelnke. 1983. An efficient technique in shape optimization. Journal of
Structural Mechanics 11:43349. doi:10.1080/03601218308907451

21

Table 1. Optimization data: mapping function coefficients ( C1 , C2 ,, C5 ) and maximum


tangential stress ( Max ( MPa )) along the hole edge and the ratios of the shapes semi axes in
the y and x directions
Q/P

C1

C2

0.000663 0.000176

C5

+0.01816

0.00007

+0.00011

2.025

1.0184/1.0174 = 1

.0010

13

946

985

185

0.000302

+0.00018

+0.00002

0.00009

+0.00006

21i

654i

492i

132i

693i

+0.00058

+0.01593

+0.00018

17

841

974

743

13

+0.00089

+0.00031

0.000027

0.00011

0.000204

305i

385i

40i

199i

21i

+0.01066

0.00003

2.5 0.393144 0.000169

Ry / Rx

C4

78

1.5 0.188818

Max

C3

0.001776 2.542

0.001626 3.589

97

60

220

650

71

+0.00085

+0.00068

0.000186

0.00029

+0.00002

790i

849i

62i

055i

675i

0.619959

+0.00004

+0.00400

+0.00005

22

0.000879 6.195

1.2070/0.8261 = 1
.4611

1.4064/0.6157 =
2.2842

1.6248/0.3833 = 4

65

499

692

909

59

0.000346

+0.00007

+0.00009

0.00000

0.000011

60i

343i

008i

566i

39i

23

.2390

Table 2. Influence of Ey ' / Ex ' increasing when = 30 and Q / P 2.5


Ey ' / Ex ' 0.5

Ey ' / Ex ' 2

Ey ' / Ex ' 3

Ey ' / Ex ' 5

Max

3.7383

3.6086

3.5873

3.5608

Min

3.0807

3.2917

3.3002

3.3410

Max Min

0.6576

0.3169

0.2871

0.2198

24

Table 3. Optimization data in different Ey ' / Ex ' when Q / P 1 : mapping function


coefficients and maximum and minimum tangential stresses ( MPa ) along the hole edge

Ey ' / Ex '

C1

C2

C3

C4

C5

45

0.5

0.008197

+0.000481

0.052426

+0.002483

+0.004251

3.886

3.88

95

37

03

87

05

63

+0.002098

+0.001126

+0.008520

+0.001351

0.015277

45i

64i

78i

23i

52i

+0.007339

+0.001508

0.051993

+0.003970

+0.002148

3.743

3.74

19

02

37

18

78

30

+0.003357

0.007021

0.002389

0.001464

+0.017089

74i

22i

23i

60i

35i

0.014547

0.000629

0.026417

+0.003092

+0.006190

3.977

3.97

75

28

15

87

40

83

0.000064

0.003476

0.014888

0.000380

+0.020395

19i

25i

71i

34i

52i

0.006804

0.004396

+0.006897

+0.003424

+0.004847

4.334

4.33

37

68

57

23

41

45

25

Max Min

90

0.5

0.002296

0.005852

0.009373

0.003618

+0.029479

90i

35i

98i

32i

91i

0.190623

0.000915

0.073298

0.000684

+0.003166

3.304

3.30

40

15

76

67

78

47

+0.000100

+0.000155

0.000103

0.000822

0.000167

48i

92i

95i

16i

42i

+0.198624

0.000325

0.076677

0.000127

0.004819

3.361

3.36

91

34

85

24

83

18

+0.000887

0.000440

0.000288

+0.000216

0.000388

43i

40i

65i

94i

40i

+0.313867

+0.006373

0.072918

+0.003178

0.006182

3.581

3.58

54

09

33

49

35

15

+0.000861

+0.000779

0.000617

0.000691

0.000849

96i

47i

81i

85i

48i

+0.426961

0.000932

0.077165

0.000742

0.010352

3.972

3.97

12

25

46

61

31

25

0.000129

+0.000043

0.000008

+0.000050

+0.000041

06i

09i

74i

76i

99i

26

Table 4. Range of stress distributions for different under the loading condition of Q / P 5

45

75

90

Ey ' / Ex ' 5

0.7520

0.6392

1.4366

1.7206

Ey ' / Ex ' 0.5

3.8276

2.1312

2.4540

2.8062

27

Table 5. Maximum compressive and tensile stresses ( MPa ) for different when Q / P 1

Ey ' / Ex ' 5

Ey ' / Ex ' 0.5

15

30

45

60

75

90

3.9789

4.2468

4.3303

4.3345

4.3331

4.2639

3.9725

3.9789

4.2467

4.3303

4.3342

4.3331

4.2640

3.9727

3.3190

3.5388

3.8305

3.8863

3.8255

3.4789

3.3047

3.3193

3.5388

3.8302

3.8863

3.8255

3.4789

3.3047

28

Figure 1. An anisotropic plate with an arbitrary shaped hole under in-plane loading.

29

Figure 2. Sample of material parameters.

30

Figure 3. Design history under condition of Q / P 1 , = 0 and Ey ' / Ex ' 5 .

31

Figure 4. Optimal results for different Q / P when = 0 and Ey ' / Ex ' 2 .

32

Figure 5. Optimal results for different Ey ' / Ex ' when = 30 and Q / P 2.5 .

33

Figure 6. Optimal results for different Ey ' / Ex ' when Q / P 1 .

34

Figure 7. Optimal results for different when Q / P 5 and Ey ' / Ex ' 0.5 .

35

Figure 8. Optimal results for different when Q / P 1 and Ey ' / Ex ' 5 .

36

Figure 9. Comparison between the optimal shape and an ellipse when = 0, Ey ' / Ex ' 3 and

Q / P 2.5 .

37

Figure 10. Comparison between the optimal shape and a circular hole when = 45, Ey ' / Ex ' 3
and Q / P 1 .

38

Appendix
For an orthotropic material, as the local coordinate axes coincide with the materials main
direction, the generalized Hookes law in the local coordinate system can be written as

a12
a13

1
1 a11


a a a
0
2
2 12 22 23
3
3 a13
a23
a33

a44
4
4
5
5

0
a55


6
a66
6
where: 1, 2 , 3 are x , y , z ; 4 ,5 , 6 represent yz , xz , xy ; 1, 2 , 3 mean

x , y , z and 4 , 5 , 6 are

a33

1
1
1
1
1


, a22
,
yz , xz , xy . And aij aji , a11
E y
Ex
2
2
2

1
1

1
1
yx , a13
zy , a44


zx , a23

, a12
, a55
and a66
,
E y
Ez
Gyz
Gxy
Ez
Gxz
Ez

where Ex , E y and Ez are the Youngs moduli in the directions of the x , y and z
axes, respectively. Gyz , Gxz and Gxy are the shear moduli in the yoz , xoz and xoy
planes, respectively. yx , zx and zy are the corresponding Poissons ratios, where

yx

E y
E
E
xy , zx z xz and zy z yz .
Ex
E y
Ex

In the global coordinate system, when the z -axis is perpendicular to one plane of elastic
symmetry, the flexibility matrix can be given as

39

a11
a
21
a31

0
0

a61

a12 a13 0
0 a16
a22 a23 0
0 a26
a32 a33 0
0 a36

0
0 a44 a45 0
0
0 a54 a55 0

a62 a63 0
0 a66

Based on Eqs. (7) and (8), a44 , a55 and a45 ( a54 ) need not be determined. The other aij in
the global system ( x , y , z ) can be calculated according to the material parameters aij in the
local system ( x , y , z ). The calculation formulae are

a11 a11 cos4 (2a12 a66 )sin 2 cos2 a22 sin 4


a22 a11 sin 4 (2a12 a66 )sin 2 cos2 a22 cos4

a33 a33
a12 a12 (a11 a22 2a12 a66 )sin 2 cos2
a13 a13 cos2 a23 sin 2
a23 a13 sin 2 a23 cos2

a16 a22 sin 2 a11 cos2 (a12 0.5a66 )cos(2 ) sin 2

a26 a22 cos2 a11 sin 2 (a12 0.5a66 )cos(2 ) sin 2

a36 (a23 a13 )sin 2


a66 a66 4(a11 a22 2a12 a66 )sin 2 cos2

40

41

You might also like