You are on page 1of 12

Chapter 4

DESIGN OPTIMIZATION OF COMPOSITE PLATE WITH GA & PSO

In laminated composite structures, each ply has its greatest stiffness and strength properties,
along the direction, through which the fibers are oriented in. By orienting the layers at
different angles, the structure can be designed for a specific loading environment. Along with
structural performance and weight, cost is an area of great interest when considering
optimization studies in structural design. Obviously, reducing the amount of material required
for the structure minimizes the cost of a laminate composite.

4.1 Introduction
This chapter deals with design problem of a composite laminated plate. The main objective is
to obtain the optimum solution in form of minimum composite weight for various type of
mechanical loading. The laminate angles and their stacking sequence are taken as design
variables while the constraints are supplied in the form of their failure strengths. Previously it
was discussed that by taking the initiating failure mode into consideration it is possible to
predict failure type quite accurately. So that here strength of composite is calculated based on
its mode of failure.
Another objective of this chapter is to compare the various popular methods of composite
design optimization. Minimum Constrained Optimization is used to design the laminated
plate for different cases of mechanical loading. Solution for two variable problems is also
obtained by optimizing stacking angles and number of plies.
PSO has proven to have superior computational efficiency when compared to GA and other
such evolutionary algorithms [64,65]. PSO has a flexible and well balanced mechanism to
enhance and adapt to the global and local exploration and exploitation abilities within a short
computation time [66]. Here, PSO seems like a more efficient alternative for the problem on
hand-the multi-objective design optimization of composites [67,68].
Two more popular optimization strategies are used one is Genetic algorithm and other one is
Particle swarm optimization. By taking different loading conditions it is shown that a

29

stacking sequence which is optimum design solution for one loading environment may not be
suitable for other loading cases or even it may give worst results.

Figure 4.1: Composite laminated plate

4.2 Problem Formulation


Design optimization of simply supported composite plate is performed here. The composite is
designed for weight optimization under given load conditions. All the stress and strain
calculation is done by Classical Laminate Plate Theory (CLPT). The mathematical
formulation of problem is done in MATLAB. All the mathematical explanations and
equations are given in appendix.
4.2.1 Design objective
Minimization of weight of laminated plate
min W = a b h
Where

(4.1)

is density of composite, a, b and h are length, width and thickness of plate

respectively. In above expression only thickness of laminate is variable so our objective


function is reduced to minimization of laminate thickness. So objective function is given by
min
W = a b min
h

30

The values of a & b are equal and taken as 10 meter for the sack of comparison with that of
Ref [65].
4.2.2 Design Variables
Design variables taken are stacking sequence, number of plies and ply angle. Here number of
plies is considered to be a continuous variable that can take non-integer values. The initial
assumption of stacking sequence considered is [ ] NS and then for [ ( 1 )N 1 /( 2 )N 2 ] .
a) Number of piles
0
0
b) Ply angle 0 90
c) Stacking sequence
Stacking sequence is not considered as design variable in problem formulation but few
discreet stacking sequences are explored. These stacking sequences are
a)

[( 90 )N 1 / ( 0 )2 N 2 / ( 45 )2 N 3 ]T

b)

[ 02 N 1 90 2 N 2 ]S

c)

[( 1 )N 1 /( 2 )N 2 ]

d)

[0]NS

e)

[ ] NS

4.2.3 Design Constraints


For composite optimization problem constraints are supplied in form of strength parameters.
These parameters depend upon failure theory considered. Different parameters associated
with lamina strength are given bellow
a) Longitudinal tensile strength

XT

b) Longitudinal compressive strength


c) Transverse tensile strength

XC

YT

d) Transverse compressive strength

YC

e) Shear strength S

31

If

and

are longitudinal, transverse and shear stresses respectively. Then

the constraints are given by


X T > L ; X C < L ;
Y T > T ; Y C < T ; S >| |

4.2.4 Loading Conditions


In plane loading is imposed on the composite plate. Different loading cases for which
optimization is performed are broadly classified as
a) Axial loading in x and y directions
b) Biaxial loading
c) General 2-D loading
Optimization procedure is shown in Figure 4.1. Three widely accepted optimization
techniques in composite designing are used.

Figure 4.2: Flow chart for composite laminated plate optimization

32

4.3 Results and Discussion


Results obtained by three optimization methods are presented.

4.3.1 Minimum Constraint Optimization


Table 4.1 summarizes the optimization results for applied loading of 1800 N/mm in X
direction. The ply thickness is kept constant for all the cases and failure mode for this type of
loading is fiber breakage. By obtained results it is clear that 0 0 angle ply give better results
as compared to other stacking sequence. This is obvious because we are applying load in X
direction. The optimized weight of laminate is 256 kg.
For compressive loading in X direction the possible failure mode is fiber compressive failure.
Results for this case are represented in Table 4.2. 00 angle ply again is found to be better than
other stacking the optimized weight is 288 kg. Table 4.3 shows optimized results for tensile
load in Y direction, mode of failure in this type of loading is matrix cracking. Here quasi
isotropic ply laminate performed better than other angles plies. Weight of laminate in this
case is 448 kg. Table 4.4 gives results for compressive loading in Y direction; failure mode
considered is matrix crushing. Optimized value is 448 kg for quasi isotropic laminate.
In Table 4.5, weight of plate is optimized for both tensile and compressive load in X and Y
directions. It is seen that optimized weight is given by quasi isotropic laminate.
4.3.2 GA Design Optimization
Weigh optimization of composite plate under biaxial loading are shown in Table 4.6. Biaxial
loading of value 500 N/mm is applied in both tensile and compressive directions. Design
optimization is performed to obtain optimum weight. GA parameters involved are given
below
a)
b)
c)
d)
e)
f)

Population type: Double Vector


Population: 20
Initial scores: [1,450]
Selection: Rank
Lower bound for design variable: [0, 0]
Upper bound for design variable: [30, pi/2]

33

In Table 4.6 optimized weight of laminated plate is obtained. Different stacking sequences are
considered for comparison purpose. It is found that optimum cross ply laminate perform
better than other stacking sequences.
Table 4.9 express results of laminate weight optimization through GA. Design variable
considered are value of ply angle and number of plies. Single and double angle ply laminates
are observed under biaxial loading. Results are same for both cases.

4.3.3 Particle Swarm Optimization


Particle Swarm Optimization (PSO) is applied for weight optimization of composite plate
under biaxial loading. Optimum results for different stacking sequences are given in Table
4.7. Minimum weight of laminate is obtained for cross ply laminate. Table 4.10 express
results for angle ply laminate. Value of optimum angle given by PSO is 450.
4.3.4 Comparison of Minimum Constraint Optimization, GA and PSO
Table 4.11 concludes results obtained by all three schemes for biaxial loading can be
analyzed to compare GA and PSO results. So by evaluating these results it may be found that
all three Optimization techniques give same results.

4.4 Summary
Composite laminated plate is optimized for minimum thickness under various loading
conditions. Three optimization tools used are Minimum Constraint Optimization, GA and
PSO. It is seen that all three optimization methods give exactly same results.

Table 4.1: Minimum Constraint Optimization of composite plate in Region of fiber breaks
for tensile loading in X direction

S.

Loading Case

Stacking sequence

34

Optimum laminate

No

weight (Kg)

.
1
2
3

Nx=1800
N/mm
Nx=1800
N/mm
Nx=1800

[0]NS

N=8

256

[90N 1 , 0 N 2 ]S

N1=15;N2=5

640

N1=2;
N2=8; N3=3

416

[ ( 90 )N 1 / ( 0 )N 2 / ( 45 )N 3 ] S

N/mm

Table 4.2: Minimum Constraint Optimization of composite plate in Region of fiber compressive failure
for compressive loading in X direction

S.
No.

Loading Case
Nx=-1800

N/mm
Nx=-1800

N/mm
Nx=-1800

N/mm

Stacking sequence

Optimum laminate

weight (Kg)

[0]NS

N=9

288

[90N 1 , 0 N 2 ]S

N1=8;N2=3

352

[( 0 )N 1 ( 90 ) N 2 / ( 45 )N1=2;N2=9;N3=3
N 3 ]S

448

Table 4.3: Minimum Constraint Optimization of composite plate in Region of Matrix Cracking Failure
for tensile loading in Y direction

S.
No

Loading Case

Stacking sequence

Optimum laminate

weight (Kg)

.
1.

Ny=1800 N/mm

[0]NS

N=52

1664

2.

Ny=1800 N/mm

[90N 1 , 0 N 2 ]S

N1=14;N2=5

608

3.

Ny=1800 N/mm

[ ( 0 )N 1 ( 90 ) N 2 / ( 45 )N1=2;N2=9;
N 3 ]S
N3=3

448

Table 4.4: Minimum Constraint Optimization of composite plate in region of matrix crushing failure
for compressive loading in Y direction

S.
No

Loading Case

Stacking sequence

Optimum laminate
weight (Kg)

35

1.

Ny=-1800 N/mm

[0]NS

N=45

1440

3.

Ny=-1800 N/mm

[90N 1 , 0 N 2 ]S

N1=4;N2=11

480

3.

Ny=-1800 N/mm

[( 90 )N 1 / ( 0 )N 2 / ( 45 )N 3 ] S N1=9;N2=2;N3

448

=3

Table 4.5: Minimum Constraint Optimization of composite plate under biaxial loading

S.
No
.
1.
2.
3.

Loading Case

Stacking

sequence
Nx=500 N/mm
Ny=500 N/mm
Nx=500 N/mm
Ny=500 N/mm
Nx=500 N/mm
Ny=500 N/mm

Optimum laminate
weight (Kg)

[0]NS

N=45

1440

[90N 1 , 0 N 2 ]S

N1=4;N2=3

224

[( 90 )N 1 / ( 0 )N 2 / ( 45N1=2;N2=2;N3=2
)N 3 ]S

192

Table 4.6: GA Optimization of composite plate under biaxial loading

S.
No
.
1.
2.
3.

Loading Case

Stacking

sequence
Nx=500 N/mm
Ny=500 N/mm
Nx=500 N/mm
Ny=500 N/mm
Nx=500 N/mm
Ny=500 N/mm

Optimum laminate
weight (Kg)

[0]NS

N=42

1344

[90N 1 , 0 N 2 ]S

N1=3.00;N2=3

192

[( 90 )N 1 / ( 0 )N 2 / ( 45N1=2;N2=2;N3=3
)N 3 ]S

224

Table 4.7: PSO Optimization of composite plate under biaxial loading

S.
No
.
1.
2.

Loading Case

Stacking

sequence
Nx=500 N/mm
Ny=500 N/mm
Nx=500 N/mm
Ny=500 N/mm

Optimum laminate
weight (Kg)

[0]NS

N=42

1344

[90N 1 , 0 N 2 ]S

N1=3.00;N2=3

192

36

3.

Nx=500 N/mm
Ny=500 N/mm

[( 90 )N 1 / ( 0 )N 2 / ( 45N1=2;N2=2;N3=3
)N 3 ] S

224

Table 4.8: Minimum Constraint Optimization results under biaxial loading for angle ply

S.
No
.
1

Loading Case
Nx=500 N/mm
Ny=500 N/mm

2
Nx=500 N/mm
Ny=500 N/mm

Stacking

Optimized

sequence

angle

=.7854

[ ]NS

[( 1)N 1 /(2 )N 2 ] S

1=.7853
2=.7853

Optimum laminate
N

weight (Kg)

N=5

160
192

N1=3
N2=3

Table 4.9: GA Optimization results under biaxial loading for angle ply

S.
No
.
1

Loading Case

Optimized

sequence

angle

Optimum laminate
N

weight (Kg)
160

Nx=500
N/mm
Ny=500

Stacking

[ ]NS

N/mm
Nx=500

[( 1)N 1 /(2 )N 2 ] S

N/mm
Ny=500

=.7854

1=.7853
2=.7853

N=5

192
N1=4
N2=2

N/mm

Table 4.10: PSO Optimization results under biaxial loading for angle ply

S.
No
.
1

Loading Case
Nx=500

Stacking

Optimized

sequence

angle

[ ]NS

=.79

37

Optimum laminate
N

weight (Kg)

N=5

160

N/mm
Ny=500
2

N/mm
Nx=500
N/mm
Ny=500

[( 1)N 1 /(2 )N 2 ] S

1=.785
2=.785

192
N1=4
N2=2

N/mm

Table 4.11: Optimum weight of the laminate by Optimization techniques for biaxial loading
(Nx= 500 N/mm, Ny= 500 N/mm).

S.
N

Laminate weight by
Stacking sequence

o.

Min Constraint

Laminate weight by
GA Optimization

Laminate weight by
PSO Optimization

Optimization

[0]NS

1440

1344

1344

[90N 1 , 0 N 2 ]S

224

192

192

192

224

224

[( 0 )N 1 ( 90 ) N 2 / ( 45 ) N 3 ] S

38

Current Function Value: 0.0082005

0.005

Constraint violation

4
5
6
7
Iteration
Maximum Constraint Violation: 1.8839e-008

0.4

0.3
0.2
0.1
0

4
5
Iteration

Figure 4.3: Minimum Constraint Optimization results for biaxial loading

[( 0 )N 1 ( 90 ) N 2 / ( 45 ) N 3 ]S
-3

Best Function Value: 0.0082203

x 10

8
7

Func tion value

Function value

0.01

6
5
4
3
2
1
0

0.5

1.5

2.5
Iteration

39

3.5

4.5

Figure 4.4: Solution Convergence of GA Optimization results for biaxial loading and unidirectional ply laminate

40

You might also like