See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228453626

Imperialist Competitive Algorithm and its Application in Optimization of

Laminated Composite Structures

Article · June 2011

CITATIONS READS

16 1,129

4 authors:

Behzad Abdi Hamid Mozafari

Ulster University Universiti Teknologi Malaysia
33 PUBLICATIONS   108 CITATIONS    41 PUBLICATIONS   114 CITATIONS   

SEE PROFILE SEE PROFILE

Amran Ayob Roya Kohandel

Universiti Teknologi Malaysia Universiti Teknologi Malaysia
88 PUBLICATIONS   337 CITATIONS    8 PUBLICATIONS   33 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Nano composite View project

EXPERIMENTAL AND SIMULATION ANALYSE OF THE MECHANICAL BEHAVIOUR OF MULTILYERED HYBRID BASALT/GLASS FIBRE PRESSURISED COMPOSITE TUBE View project

All content following this page was uploaded by Amran Ayob on 18 March 2014.

The user has requested enhancement of the downloaded file.

European Journal of Scientific Research
ISSN 1450-216X Vol.55 No.2 (2011), pp.174-187
© EuroJournals Publishing, Inc. 2011
http://www.eurojournals.com/ejsr.htm

Imperialist Competitive Algorithm and its Application in

Optimization of Laminated Composite Structures

Behzad Abdi
Faculty of Mechanical Engineering, Universiti Teknologi Malaysia
81310 UTM Skudai, Johor, Malaysia
E-mail: [email protected]

Hamid Mozafari
Faculty of Mechanical Engineering, Universiti Teknologi Malaysia
81310 UTM Skudai, Johor, Malaysia
E-mail: [email protected]

Amran Ayob
Faculty of Mechanical Engineering, Universiti Teknologi Malaysia
81310 UTM Skudai, Johor, Malaysia
E-mail: [email protected]

Roya Kohandel
Faculty of Civil Engineering, Universiti Teknologi Malaysia
81310 UTM Skudai, Johor, Malaysia
E-mail: [email protected]

Abstract

This paper presents an application of the imperialist competitive algorithm (ICA) in

optimization of composite structures design. The recently introduced algorithm has proven
its excellent capabilities, such as faster convergence and better global optimum
achievement. In this paper, imperialist competitive algorithm (ICA) is used to demonstrate
its application in finding the optimal design of laminated composite structures due to the
various failure criteria. The proposed method can be used in various multi-objective
optimization problems. The effectiveness of the proposed method, in comparison to
Genetic Algorithm (GA), is proven through solving several examples of composite
structure problems.

Keywords: Imperialist Competitive Algorithm (ICA), Genetic Algorithm, Laminated

Composite Structure, Failure Criteria.

1. Introduction
Laminated composite materials find a wide range of applications in structural design, particularly in
the field of automotive, aerospace and marine engineering. This is primarily due to the high specific
strength and stiffness values with minimum weight that these type of materials offer. Although
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 175

composite materials are attractive replacement for metallic materials for many structural applications,
the design and analysis is more complex than those of metallic structures.
Composite materials are usually understood as the combination of two or more materials on a
microscopic scale to form a useful third material [1]. Laminated composites are usually designed due
to the designer’s needs by choosing the thickness, orientation and number of laminae. The thickness
and orientation of the laminae are usually limited to some set values due to manufacturing limitations.
Searching for the optimum solution in laminated composite structures is a discrete optimization
problem. Evolutionary algorithms, such as GA, PSO and SA were suggested in the past decades for
solving optimization problems in different fields. Also, these methods have been recently applied to
laminated composite problems. The genetic algorithm has been widely used to find the optimal design
of laminated composite structures due to the ability of its algorithm to solve this kind of optimization
problems [2-13]. The advantages of the use of GAs include the following: (i) they do not require
gradient information and can be applied to problems where the gradient is hard to obtain or simply
does not exist; (ii) if correctly tuned, they do not get stuck in local minima; (iii) they can be applied to
nonsmooth or discontinuous functions; and (iv) they furnish a set of optimal solutions instead of a
single one, thus giving the designer a set of options. On the other hand, the use of GAs has a number of
known disadvantages, which include the following: (i) they require the tuning of many parameters by
trial and error to maximize efficiency; (ii) the a priori estimation of their performance is an open
mathematical problem; and (iii) an extremely large number of evaluations of the objective function are
required to achieve optimization, which can make the use of GAs economically. The basic parts of a
GA are: the variable coding, the selection method, the genetic operators and the how the constraints are
handled [14].
In computer science, the Imperialist Competitive Algorithm (ICA) [15] is a computational
method that is used to solve many types of optimization problems. Like most of the methods in the
field of evolutionary computation, ICA does not need the gradient of the function in its optimization
process.
Specifically, ICA can be thought of as the social counterpart of genetic algorithms (GAs). ICA
is the mathematical model and computer simulation of human social evolution, while GAs is based on
the biological evolution of species.
Figure 1 shows the flowchart of the Imperialist Competitive Algorithm. This algorithm starts
with an initial population. Each individual of the population is called a country. Countries are the
counterpart of chromosomes in GAs. After evaluating the cost function of each country, some of the
best of them (in optimization terminology, countries with the least cost) are selected to form the initial
empires by controlling the other countries (colonies). All the colonies are divided among the initial
imperialists based on their power. The power of each country, the counterpart of fitness value in the
GA, is inversely proportional to its cost. The initial imperialist form states together with their colonies
to form the initial empires.
176 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel
Figure 1: Flowchart of the Imperialist Competitive Algorithm (ICA). [17]

Start
Is there an empire
Imperialistic Competition with no colonies
Initialize the empires Yes

Compute the total cost of all empires Eliminate this empire

No
Assimilate colonies

Unite Similar Empires

Revolve some colonies Exchange the positions of that
imperialist and the colony
Stop condition
Is there a colony in an satisfied
empire which has lower cost Yes
than that of the imperialist Yes

No END
No

After forming initial empires, the evolution begins. The colonies in each empire start moving
toward their relevant imperialist country. This movement is a simple model of assimilation policy
which was pursued by some of the imperialist states. Besides assimilation, revolution is another
operator of this algorithm. Revolution occurs in some of the colonies by making random changes to
their position in the socio-political axis. The total power of an empire depends on both the power of the
imperialist country and the power of its colonies.
Imperialistic Competition is another step of the algorithm. All empires try to take the
possession of colonies of other empires and control them. The imperialistic competition gradually
brings about a decrease in the power of weaker empires and an increase in the power of more powerful
ones. The imperialistic competition is modeled by picking some (usually one) of the colonies of the
weakest empire and making a competition among all empires to possess these (this) colonies. The
above steps continue until a stop condition is satisfied by reaching to an acceptable suboptimal solution
[16-20].
One of the important issues in the optimization problems is the constraints handling. In the
optimization of laminated composite structures one usual constraint is the symmetry and balance of the
laminate. In general, it is taken into account using a data structure strategy which contains of writing
code only for half of laminate and considering that each stack of the laminate is made of two laminate
with the same orientation but opposite signs [3, 6]. Data structure, repair strategy and penalty functions
are the most common ways to treat constraints [4, 6 and 9].
The structural failure constraints in optimization of laminated composite structure are often
handled by using penalty function. The different forms of construction of the objective function with
the penalty terms were studied by Le riche and Haftka [6]. Several failure criteria were studied by
Lopes and Luerson [14] and by Narayana and et al [21].
In optimization of laminated composite structures, two types of objective (cost) function have
been proposed. When considering the strength of the structure, the buckling load [2, 4 and 22],
strength, stability and the stiffness in one direction are maximized [23]. When considering the benefit
of cost, material cost, manufacturing cost and weight are minimized [24].
In this paper, the imperialist competitive algorithm is introduced and subsequently its
application in optimizing laminated composite structures. The rest of this paper is organized to
introduce Tsai – Wu (TW) and maximum stress failure criterion (MS) that used as constraints in the
optimization of composite structure. The results of imperialist competitive algorithm are compared
with the previous study which had used the genetic algorithm for optimization.
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 177

2. Formulation of Imperialist Competitive algorithm

2.1. Creation of Initial Empires
The goal of optimization is to find an optimal solution in terms of the variables of the problem. An
array of variable values to be optimized is formed. In the genetic algorithm, this array is called
“chromosome”, but in ICA the term “country” is used. In an N var -dimensional optimization problem, a
country is a 1× N var array. This array is defined as follow,
Country = [ p1 , p2 , p3 ,…, pNvar ] (1)
where pi is the variables to be optimized. The variable values in the country are represented as floating
point numbers. Each variable in the country can be interpreted as a socio-political characteristic of a
country. From this point of view, all algorithms search for the best country with the best combination
of socio-political characteristics such as culture, language, economical policy, and even religion. From
optimization point of view this leads to finding the optimal solution of the problem, Figure 2 shows the
interpretation of country using some of socio-political characteristics.

Figure 2: The candidate solutions of the problem, called country. [15]

Culture Language Economical Policy …..… Religion

c o u n t r y = [ p 1 , p 2 , p 3 , ..., p N ]
var

The cost of a country is found by evaluating the cost function f at variables

( p1 , p2 , p3 ,..., pN var ) . So,
cost = f (country) = f ( p1, p2 , p3 ,..., pNvar ) (2)
To start the optimization algorithm, initial countries of size N Country are produced. N imp of the
most powerful countries to form the empires are selected. The remaining N col of the initial countries
will be the colonies each of which belongs to an empire.
To form the initial empires, the colonies are divided among imperialists based on their power.
That is, the initial number of colonies of an empire should be directly proportionate to its power. To
proportionally divide the colonies among imperialists, the normalized cost of an imperialist is defined
by
Cn = cn − max{
i
ci } (3)
where cn is the cost of the nth imperialist and Cn is its normalized cost. Having the normalized cost of
all imperialists, the normalized power of each imperialist is defined by

Cn
pn = N imp
(4)
∑
i =1
Ci

The initial colonies are divided among empires based on their power. Then the initial number of
colonies of the nth empire will be
N .C.n = round{ pn . N col } (5)
178 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel

where N .C.n is the initial number of colonies of the nth empire and N col is the total number of initial
colonies. To divide the colonies, N .C.n of the colonies are randomly chosen and given to the
nth imperialist. These colonies along with the nth imperialist form the nth empire. Figure 3 shows the
initial empires. As shown in this figure, bigger empires have greater number of colonies while weaker
ones have less. In this figure, Imperialist 1 has formed the most powerful empire and consequently has
the greatest number of colonies.

Figure 3: Generation of initial empires. [15]

2.2. Assimilation: Movement of Colonies toward the Imperialist

Pursuing assimilation policy, the imperialist states tried to absorb their colonies and make them a part
of themselves. More precisely, the imperialist states made their colonies to move toward themselves
along different socio-political axis such as culture, language and religion. This process is modeled by
moving all of the colonies toward the imperialist along different optimization axis, as shown in Figure
4. Considering a two-dimensional optimization problem, in this figure the colony is absorbed by the
imperialist in the culture and language axes. Then the colony will get closer to the imperialist in these
axes. Continuation of assimilation will cause all the colonies to be fully assimilated into the imperialist.
Figure 4 shows a colony moving toward the imperialist by units. The new position of the
colony is shown in a darker color. The direction of the movement is the vector from the colony to the
imperialist. In this figure x is a random variable with uniform (or any proper) distribution. Then

Figure 4: Movement of colonies toward their relevant imperialist. [15]

x ~ U ( 0,β × d ) (6)
where β is a number greater than 1 and d is the distance between the colony and the imperialist state.
The condition β > 1 causes the colonies to get closer to the imperialist state from both sides.
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 179

Assimilating the colonies by the imperialist states does not result in direct movement of the
colonies toward the imperialist. That is, the direction of movement is not necessarily the vector from
colony to the imperialist. To model this fact and to increase the ability of searching more area around
the imperialist, a random amount of deviation is added to the direction of movement. Figure 5 shows
the new direction. In this figure θ is a parameter with uniform (or any proper) distribution. Then
θ ~ U ( -γ, γ ) (7)
where γ is a parameter that adjusts the deviation from the original direction. Nevertheless the values of
β and γ are arbitrary, and in many implementations a value of about 2 for β and about π / 4 for γ
results in good convergence of countries to the global minimum.

Figure 5: Movement of colonies toward their relevant imperialist in a randomly deviated direction. [15]

2.3. Revolution; A Sudden Change in Socio-Political Characteristics of a Country

Revolution is a fundamental change in power or organizational structures that takes place in a
relatively short period of time. That is, instead of being assimilated by an imperialist, the colony
randomly changes its position in the socio-political axis. Figure 6 shows the revolution in Culture-
Language axis. The revolution increases the exploration of the algorithm and prevents the early
convergence of countries to local minimums. The revolution rate in the algorithm indicates the
percentage of colonies in each colony which will randomly change their position. A very high value of
revolution decreases the exploitation power of algorithm and can reduce its convergence rate. In the
present simulations the revolution rate is 0.3 that is 30 percent of colonies in the empires change their
positions randomly.

Figure 6: Revolution; a sudden change in socio-political characteristics of a country. [15]

180 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel

2.4. Exchanging Positions of the Imperialist and a Colony

While moving toward the imperialist, a colony might reach a position with lower cost than the
imperialist. In this case, the imperialist and the colony change their positions. Then the algorithm will
continue with the imperialist in the new position and the colonies will be assimilated by the imperialist
in its new position. Figure 7.a depicts the position exchange between a colony and the imperialist. In
this figure the best colony of the empire is shown in a darker color. This colony has a lower cost than
the imperialist. Figure 7.b shows the empire after exchanging position between the imperialist and the
colony.

Figure 7: Exchanging the positions of a colony and the imperialist. [15]

(a) Before (b) After

2.5. Uniting Similar Empires

In the movement of colonies and imperialists toward the global minimum of the problem some
imperialists might move to similar positions. If the distance between two imperialists becomes less
than a threshold distance, they both will form a new empire which is a combination of these empires.
All the colonies of two empires become the colonies of the new empire and the new imperialist will be
in the position of one of the two imperialists. Figure 8 shows the uniting process of two empires before
uniting and resulting from uniting two empires, respectively.

Figure 8: The uniting process of two empires. [15]

(a) Before (b) After

2.6. Total Power of an Empire

The total power of an empire is mainly contributed by the power of imperialist country. However the
power of the colonies of an empire has a negligible effect, on the total power of that empire. This fact
is modeled by defining the total cost of an empire by
T .C . = Cost ( imprialist n ) + ξ mean{Cost(colonies of empire n )} (8)
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 181

where T .C.n is the total cost of the nth empire and ξ is a positive small number. A small value for ξ
causes the total power of the empire to be determined by just the imperialist and increasing it will
increase the role of the colonies in determining the total power of an empire. The value of 0.1 for ξ has
shown good results in most of the implementations.

2.7. Imperialistic Competition

All empires try to take possession of colonies of other empires and control them. The imperialistic
competition gradually brings about a decrease in the power of weaker empires and an increase in the
power of more powerful ones. The imperialistic competition is modeled by just picking some (usually
one) of the weakest colonies of the weakest empire and making a competition among all empires to
possess these (this) colonies. Figure 9 shows a bigger picture of each empires taking possession of the
mentioned colonies. In other words, these colonies will not definitely be possessed by the most
powerful empires, but these empires will be more likely to possess them.

Figure 9: Imperialistic competition. [15]

To start the competition, first a colony of the weakest empire is chosen and then the possession
probability of each empire is found. The possession probability p p is proportionate to the total power
of the empire. The normalized total cost of an empire is simply obtained by
N .T .C.n = T .C.n − max{T .C.i } (9)
i

where, T .C.n and N .T .Cn are the total cost and the normalized total cost of nth empire, respectively.
Having the normalized total cost, the possession probability of each empire is given by

N .T .C.n
p pn = Nimp
(10)
∑ N .T .C.
i =1
i

To divide the mentioned colonies among empires vector P is formed as following

P = ⎡ p p1 , p p2 , p p3 ,..., p pN ⎤ (11)
⎣ imp ⎦

Then the vector R with the same size as P whose elements are uniformly distributed random
numbers is created,
R = ⎡ r1 , r2 , r3 , ..., rN ⎤
⎣ imp ⎦
(12)
r1 , r2 , r3 , ..., rN imp U (0,1)
182 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel

Then vector D is formed by subtracting R from P

D = P - R = ⎡ D1 , D 2 , D 3 , ..., D N im p ⎤
⎣ ⎦ (13)
= ⎡ p p1 − r1 , p p 2 − r2 , p p 3 − r3 , ... , p p N − rN im p ⎤
⎣ im p ⎦
Referring to vector D the mentioned colony (colonies) is handed to an empire whose relevant
index in D is maximized.
The process of selecting an empire is similar to the ones used in selecting parents in genetic
algorithm. But this method of selection is much faster because it is not required to calculate the
cumulative distribution function and the selection is based only on the values of probabilities. Hence,
the fast process of selecting the empires will increase its execution speed.
The main steps of ICA is summarized in the pseudo-code given in Figure 10. The continuation
of the mentioned steps will cause the countries to converge to the global minimum of the cost function.
Different criteria can be used to stop the algorithm. One method is to use a number of maximum
iteration of the algorithm, called maximum decades, to stop the algorithm. Another stop criterion to
end the imperialistic competition is when there is only one empire. On the other hand, the algorithm
can be stopped when its best solution in different decades cannot be improved for some consecutive
decades.

Figure 10: Pseudo code of the Imperialistic Competitive Algorithm. [15]

1) Select some random points on the function and initialize the empires.
2) Move the colonies toward their relevant imperialist (Assimilation).
3) Randomly change the position of some colonies (Revolution).
4) If there is a colony in an empire which has lower cost than the imperialist, exchange the positions of that colony
and the imperialist.
5) Unite the similar empires.
6) Compute the total cost of all empires.
7) Pick the weakest colony (colonies) from the weakest empires and give it (them) to one of the empires
(Imperialistic competition).
8) Eliminate the powerless empires.
9) If stop conditions satisfied, stop, if not go to 2.

3. Review of Composite Material Failure Criteria

Two failure theories that are commonly used in the design of laminated composite structures are
described briefly below:

3.1. Maximum Stress Failure Criterion

Due to this theory, a lamina fails when any of the stress components σ LL , σ TT and τ LT reaches its
corresponding strength value [22, 27].
σ LL ≥ X T or σ TT ≥ X C
σ TT ≥ YT or σ TT ≥ YC (14)
τ LT ≥ + S or τ LT ≥ − S
where σ LL , σ TT and τ LT are longitudinal, transverse, shear stress along material axes L, T ,
respectively and X T , X C , YT and YC are tensile and compressive strengths in X − Y direction.
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 183

3.2. Tsai – Wu Failure Criterion

This theory is a quadratic failure criterion, which is most extensively used in the design of laminated
composite structures [21, 27]. This failure criterion states that the lamina fails when the following
condition is satisfied,
Fijσ iσ j + Fiσ i ≥ 1 i. j = 1, 2,…, 6 (15)
where Fi , Fij are strength parameters and σ i , σ j are stress components. In principal material
coordinates ( L, T ) , for the case of a lamina, this failure criterion becomes
FLLσ LL
2
+ 2 FLT σ LLσ TT + FTT σ TT
2
+ FSSσ LT
2
+ FLσ LL + FT σ TT ≥ 1 (16)

4. Numerical Results
Two different optimization problems are solved by imperialist competitive algorithm and the results
are compared with genetic algorithm results. These problems are the same as in the examples in Ref.
[14] that are solved by using genetic algorithm. The first example presents a validation of the proposed
imperialist competitive algorithm and second example presents an optimization problem under the
constraint of two different failure criteria.

4.1. Example 1 – Optimization of Weight Under Failure Mode Criteria of Maximum Strain and
Buckling
The main purpose of this example is to validate the imperialist competitive algorithm (ICA) by
comparing it to a well known laminated composite optimization procedure [6]. The purpose of this
example is to find the minimal weight of a laminated composite plate under the constraints of laminate
symmetry and balance, maximum number of contiguous plies with the same orientation as well as the
maximum strain and buckling. The allowable orientation angle values are 02 , ± 45 and ±90 degrees.
Therefore, the optimization problem can be described as [14]
find : {θ k , n} , θ k ∈ {02 , ± 45, ± 90} , k = 1 − n
minimize : weigth of composite (17)
⎪⎧strain failure(λstrain )
subject to: ⎨
⎪⎩ buckling failure(λbuckling )
where θ k , n are the orientation of each stack of the laminate and the total number of stacks,
respectively. As mentioned, each stack is composed of two layers to guarantee balance.
Fig. 11 shows a rectangular plate, with simply supported at edges and subjected to compressive
in- plane loads per unit length N x and N y . The thickness of each layer is 0.127 mm and the length and
width of plate are 0.508 m and 0.127 m respectively. The classical lamination theory and the linear
elastic buckling analysis are used in this study [1]. The buckling load factor λbuckling represents the
failure buckling load divided by the applied load, and is calculated as [29],
⎛ ⎡ ⎛ r ⎞
4 2
⎛ r ⎞ ⎛ s ⎞
2
⎛ s ⎞ ⎤ ⎞
4

⎜π ⎢ D11 ⎜ ⎟ + 2 ( D12 + 2 D 66 ) ⎜ ⎟ ⎜ ⎟ + D 22 ⎜ ⎟ ⎥ ⎟
2

⎝ a ⎠ ⎝ a ⎠ ⎝b ⎠ ⎝ b ⎠ ⎥⎦ ⎟
(18)
⎜ ⎣⎢
λ b u c k lin g = m in ⎜ 2 2 ⎟
r ,s
⎜ ⎛ r ⎞ ⎛ s ⎞ ⎟
⎜ ⎜ ⎟ Nx +⎜ ⎟ Ny ⎟
⎝ ⎝ a ⎠ ⎝b ⎠ ⎠
where Dij s are coefficients of the laminate bending stiffness matrix, r and s determine the amount of
half waves in the x − and y − direction, respectively. a , b are length and width of plate, respectively.
184 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel
Figure 11: Laminated composite plate subjected to compressive loads [14].

The critical normal strain failure factor λstrain is described as [14]

⎡ ⎛ ε 1u a ε 2u a γ 1u2a ⎞⎤
λ stra in = m in ⎢ m in ⎜ , , ⎟⎥ (1)
k
⎢ ⎜ S f ε 1k S f ε 2k S f γ 1k2 ⎟⎥
⎣ ⎝ ⎠⎦
Where ε iua , i = 1, 2 and γ 12ua are the allowable strains. ε ik , i = 1, 2 and γ 12ua are the strains in the
principal direction of the k th lamina and S f is a safety factor. Therefore, the critical failure factor λcr is
the smallest between λbuckling and λstrain .
The elastic material properties of the layers are E1 = 127,550 MPa , E2 = 13, 030 MPa ,
G12 = 6410 MPa and Poisson’s ratio v12 = 0.3 . Also, the ultimate strains are ε1ua = 0.008 , ε 2ua = 0.029 ,
γ 12ua = 0.015 and the safety factor is S f = 1.5 .
The plate is studied under three different loading conditions. The first load case has
N x = 2277 N / mm and N y = 284.6 N / mm , the second load case has N x = 2189 N / mm and
N y = 547.3 N / mm and the third load case has N x = 1716 N / mm and N y = 858.1 N / mm .
The main purpose of this study is to compare the performance of the imperialist competitive
algorithm with genetic algorithm and then comparing the results of this algorithm with those of [6].
“Price of the search” and “reliability” are two terms that are used in this study to evaluate the
performance of the imperialist competitive algorithm. The reliability of an algorithm is the probability
it has of finding a practical optimum which in turn is defined, in this study, as an optimal weight design
with the critical failure factor λcr with in 0.1% of λcr of the global optimum. The price of the search is
the number of analysis necessary to reach 80% reliability, i.e., to have 80% probability of finding a
practical optimum [14].
Table 1 shows the optimal design found in Ref. [6] where all the practical optimum information
can be found in detail.

Table 1: Optimal designs for each load case of Example 1([6, 14])

Load case Optimal design Failure mode Number of practical optima

1 [± 4 5 5 / 0 4 / ± 4 5 / 0 4 / 9 0 2 / 0 2 ]S Strain >13
2 [± 4 5 2 / 9 0 2 / ± 4 5 3 / 0 2 / ± 4 5 / 0 4 / ± 4 5 / 0 2 ]S Strain 3
3 ⎡⎣ 9 0 2 / ± 4 5 2 / ( 9 0 2 / ± 4 5 ) 2 / ± 4 5 5 ⎤⎦ Buckling 13
S

The results obtained by the imperialist competitive algorithm (ICA) formulated in this study
and their comparison with the results of Refs. [6, 14] are listed in Table 2. The imperialist competitive
algorithm shows good performance and achieves results that are as good as the results of Refs. [6,14].
In load case one, the algorithm of Ref. [6] is still better regarding the search price, while the imperialist
competitive algorithm gives the same results as the algorithm of Ref. [14]. The imperialist competitive
algorithm has obtains better results regarding search price and reliability in second and third load cases.
Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 185

Table 2: Price and reliability for Example 1. comparison with Refs [6,14].

Load case 1 2 3
Search Price 470 1110 1350
Imperialist competitive algorithm
Reliability 1.00 1.00 1.00
Search Price 480 1120 1360
Reference [14]
Reliability 1.00 1.00 1.00
Search Price 440 1180 1490
Reference [6]
Reliability 1.00 1.00 0.94

4.3. Example 2 – Optimization of Material Cost of a Laminate Under Ply Failure and Weight
Constraints
In this example, the material cost optimization of a hybrid laminated composite plate under ply failure
and weight constraints is described. The layers considered in this example are carbon – epoxy (CE) and
glass – epoxy (GE). Carbon – epoxy is lighter and stronger, while glass – epoxy has a cost advantages
in terms of lower price per square meter. For example, the price per square meter of glass – epoxy is
about 8 times less than carbon – epoxy. As in Example 1, the laminate is subjected to symmetry and
balance constraints as well as the first ply failure criteria. Also, a maximum weight constraint is used in
this study. Therefore, the optimization problem can be described as [14],
fin d : {θ k , m a t k , n } , θ k ∈ {0 2 , ± 4 5, 9 0 2 }
m a t k ∈ {G E , C E } , k = 1 − n
(20)
m in im ize : m aterial co st
⎧ first p ly failu re co n strain t
su b ject to : ⎨
⎩ m ax im u m w eig h t: 7 0 N
Using the above description about the material cost, each CE and GE layers is assumed to cost
8 and 1 monetary unit (m.u.), respectively. The elastic properties of the CE layers are
E1 = 116, 600 MPa , E2 = 7673 MPa , G12 = 4173 MPa and Poisson’s ratio v12 = 0.27 and mass density
ρ = 1605 kg / m3 and the elastic properties of GE layers are E1 = 37, 600 MPa , E2 = 9584 MPa ,
G12 = 4081 MPa and Poisson’s ratio v12 = 0.26 and mass density ρ = 1903 kg / m3 . the failure
properties of the both laminas are showed in Table 3. The thickness of each layer is 0.1 mm and the
length and width of the plate are 1.0 m . The plate is under in-plane fixed applied loads
( N x = 2000 N / mm and N y = −2000 N / mm )[14].

Table 3: Strength properties of Glass – Epoxy (GE) and Carbon – Epoxy (CE)

Glass – Epoxy (GE)

XT 1134 MPa YC 150 MPa ε1T 0.0302 mσ f 1.3
XC 1031 MPa S21 75 MPa ε1C 0.0295 p⊥( +||) 0.3
YT 54 MPa Ef1 72, 000 MPa v f 12 0.22 p⊥( −||) 0.25
Carbon – Epoxy (CE)
XT 2062 MPa YC 240 MPa ε1T 0.0175 mσ f 1.1
XC 1701 MPa S21 105 MPa ε1C 0.014 p⊥( +||) 0.3
YT 70 MPa Ef1 230, 000 MPa v f 12 0.23 p⊥( −||) 0.25

The optimization results of the imperialist competitive algorithm and the optimal results of Ref.
[14] are listed in Table 4.
186 Behzad Abdi, Hamid Mozafari, Amran Ayob and Roya Kohandel
Table 4: Optimal material cost and stacking sequence of the laminate for different criteria
Imperialist competitive Algorithm (ICA)
Material cost and weight
Stacking Reliability Search
FC Cost (m.u.) Weight (N)
TW 208 68.23 ⎡ (0 )6
CE
( 9 0 2 )4
GE
⎤ 0.09
⎣ 2 ⎦ s

MS 148 63.11 ⎡ (0 )4
C E
( 9 0 2 )5
G E
⎤ 0.08
⎣ 2 ⎦ s

Reference [14]
Material cost and weight
Stacking Reliability Search
FC Cost (m.u.) Weight (N)
⎡ ( 0 2 )6 (9 )4 ⎤
C E G E
TW 208 68.23 ⎣
0 2 ⎦ s
0.09
MS 148 63.11 ⎡ (0 2 )4
CE
( 9 0 2 )5
GE
⎤ 0.07
⎣ ⎦s

It is interesting to mention that the optimum results obtained by using the imperialist
competitive algorithm are same as in Ref. [14]. From Table 4, all layers with an orientation of 90o are
made of GE, while those with an orientation of 0o are made of CE. Also, as in weight minimization,
each failure criterion yielded a different optimum. From the results of this study and from comparisons
with Ref. [14], when optimizing laminated composite structures, the choice of a failure criterion
corresponding to the real behaviour of the structure is crucial for both economy and safety.

5. Conclusion
In this study, two different optimization algorithms (Genetic algorithm and imperialist competitive
algorithm) are used in optimization of laminated composite plates. The main objective of design
optimization in aerospace composite structures is to minimize the weight of the laminate for a given
loading. Weight minimization and the material cost minimization are two different objectives that were
considered in this study. The Imperialist competitive algorithm Showed good performance in this kind
of optimization problems and compared satisfactorily genetic algorithm. Also, two failure criterions
(maximum stress failure and Tsai – Wu failure) are used in this study. This study shows that the
optimal weight of laminated composite structures depend on the choice of the failure criterion as well
as the load conditions and it can be mentioned that there is not direct relation between optimal weight
and failure criterion.

References
[1] Jones RM., 1999, “Mechanics of composite materials. Philadelphia (PA)”: Taylor & Francis.
[2] Le Riche R, Haftka R., 1993, “Optimization of laminate stacking sequence for buckling load
maximization by genetic algorithm”. AIAA J, 31:951–6.
[3] Nagendra S, Jestin D, Gurdal Z, Haftka R. Watson L. 1994, “Improved genetic algorithm for
the design of stiffened composite panels”. Comput Struct, 58:543–55.
[4] Todoroki A, Haftka R., 1998, “Stacking sequence optimization by a genetic algorithm with a
new recessive gene like repair strategy”. Composites B, 29:277–85.
[5] Seresta O, Gürdal Z, Adams DB, Watson LT., 2007, “Optimal design of composite wing
structures with blended laminates”. Composites B, 38:469–80.
[6] Le Riche R, Haftka R., 1995, “Improved genetic algorithm for minimum thickness composite
laminate design”. Compos Eng, 5:143–61.
[7] Rahul, Sandeep G, Chakraborty D, Dutta A., 2005, “Multi-objective optimization of hybrid
laminates subjected to transverse impact”. Compos Struct, 73:360–9
[8] Murugan MS, Suresh S, Ganguli R, Mani V., 2007, “Target vector optimization of composite
box beam using real-coded genetic algorithm: a decomposition approach”. Struct Multidiscip
Optim, 33:131–46.
[9] Kogiso N, Watson LT, Gurdal Z, Haftka RT, Nagendra S., 1994, “Design of composite
laminates by a genetic algorithm with memory”. Mech Compos Mater Struct, 1(1):95–117.
 Imperialist Competitive Algorithm and its Application in
Optimization of Laminated Composite Structures 187

[10] Gantovnik VB, Anderson-Cook CM, Gurdal Z, Watson LT., 2003, “A genetic algorithm with
memory for mixed discrete–continuous design optimization”. Comput Struct, 81:2003–9.
[11] Gantovnik Vladimir B, Gurdal Zafer, Watson Layne T., 2002, “A genetic algorithm with
memory for optimal design of laminated sandwich composite panels”. Compos Struct, 58:513–
20.
[12] Walker M, Smith RE., 2003, “A technique for the multiobjective optimisation of laminated
composite structures using genetic algorithms and finite element analysis”. Compos Struct,
62:123–8.
[13] Lin CC, Lee YJ., 2004, “Stacking sequence optimization of laminated composite structures
using genetic algorithm with local improvement”. Compos Struct, 63:339–45.
[14] Lopez R.H, Luersen. M.A. Cursi. E.S., 2009, “Optimization of laminated composites
considering different failure criteria”. Composites: Part B 40, 731–740
[15] Atashpaz-Gargari E, Lucas C., 2007, "Imperialist Competitive Algorithm: An algorithm for
optimization inspired by imperialistic competition". IEEE Congress on Evolutionary
Computation. 7. pp. 4661–4666.
[16] Yang Z, Yong W, Cheng P,. 2009, "Improved Imperialist Competitive Algorithm for
Constrained Optimization". Computer Science-Technology and Applications, IFCSTA.
[17] Atashpaz-Gargari E., Hashemzadeh F., Rajabioun R., and Lucas. C., 2008, “Colonial
Competitive Algorithm, a novel approach for PID controller design in MIMO distillation
column process,” International Journal of Intelligent Computing and Cybernetics, 1 (3), 337–
355.
[18] Atashpaz-Gargari E., Rajabioun R., and Lucas C., 2008, “Colonial Competitive Algorithm as a
Tool for Nash Equilibrium Point Achievement,” Lecture notes in computer science, 5073, 680-
695.
[19] Khabbazi A, Atashpaz-Gargari E, Lucas C., 2009, “Imperialist competitive algorithm for
minimum bit error rate beamforming”, International Journal of Bio-Inspired Computation, Vol
1, No 1-2, pp 125 – 133.
[20] Duan H, Xu C, Liu S, and Shao S., 2010, “Template matching using chaotic imperialist
competitive algorithm” Pattern Recognition Letters 31 1868–1875.
[21] Narayana G, Gopalakrishnan. S, Ganguli R., 2008, “Design optimization of composites using
genetic algorithms and failure mechanism based failure criterion”, Composite Structures 83,
354–367
[22] Liu B, Haftka R, Akgun M, Todoroki A., 2000, “Permutation genetic algorithm for stacking
sequence design of composite laminates”. Comput Meth Appl Mech Eng, 186:357–72.
[23] Groenwold A, Haftka R., 2006, “Optimization with non-homogeneous failure criteria like Tsai–
Wu for composite laminates”. Struct Multidiscip Optim, 32:183–90.
[24] Adali S, Richter A, Verijenko VE, Summers EB., 1995, “Optimal design of hybrid laminates
with discrete ply angles for maximum buckling load and minimum cost”. Compos Struct,
32:409–15.
[25] Biabangard-Oskouyi, A., Atashpaz-Gargari, E., Soltani, N., Lucas, C., 2008. “Application of
Imperialist Competitive Algorithm for materials property characterization from sharp
indentation test”. To be appeared in the International Journal of Engineering Simulation.
[26] Krishna Murty AV, Narayana Naik G, Gopalakrishnan S., 2005, “Towards a rational failure
criteria for composite laminae”. Mech Adv Mater Struct – An Int J, 12(2):147–57.
[27] Tsai WS., 1988, “Composites Design”, Fourth edition. Think composite, Dayton, 1988.
[28] Gurdal Z, Haftka R, Hajela P., 1999, “Design and optimization of laminated composite
materials”. New York: Wiley.

View publication stats