Vous êtes sur la page 1sur 12

Composite Structures 88 (2009) 443–454

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Design optimization of composite laminated structures using genetic


algorithms and finite element analysis
F.S. Almeida, A.M. Awruch *
Graduate Program in Civil Engineering, Federal University of Rio Grande do Sul, Av. Osvaldo Aranha, 99, 90035-190 Porto Alegre, RS, Brazil

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

Article history: A technique for the design optimization of composite laminated structures is presented in this work. The
Available online 17 May 2008 optimization process is performed using a genetic algorithm (GA), associated with the finite element
method (FEM) for the structural analysis. The GA is adapted with special operators and variables codifi-
Keywords: cation for the specific case of composite laminated structures optimization. Some numerical examples are
Multiobjective optimization presented to show the flexibility of this tool to solve different kinds of problems. Two cases of multiob-
Genetic algorithms jective optimization of plates under transverse or in-plane load are studied. In these examples the min-
Composites laminated structures
imization of two objectives, such as weight and deflection or weight and cost, are simultaneously
Finite element analysis
performed and a pareto-optimal set is obtained by shifting the optimization emphasis using a weighting
factor. The stiffness maximization of a composite shell under pressure load is presented in the last exam-
ple, where the geometrically nonlinear behavior of the structure is considered. Some aspects of the opti-
mization performance, such as the apparent reliability and the computational cost, are investigated in
each application.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction problems typically involve multimodal search spaces, which may


lead gradient based methods to converge to locally optimal regions
In the recent decades, structural applications with composite in the design space [3]. Many other optimization techniques have
laminated materials have been growing, requiring great effort on been tested as an alternative to the gradient based methods, hav-
the development of analysis and design techniques. An advantage ing the genetic algorithm (GA) stand out the others because it per-
of using fiber-reinforced composites over conventional materials is fectly adjusts to the characteristics of the composite optimization
that the former can be tailored to specific requirements of certain problem. GAs are probabilistic search methods seeking to mimic
applications [1]. However, the large number of design variables the biological reproduction and natural selection process through
and the complex mechanical behavior associated with such mate- random but structured operations. The design variables, usually re-
rials turn the structural design much more difficult and laborious stricted to discrete values, are coded as genes using binary or inte-
than those involving conventional materials. These characteristics ger numbers, and grouped together in chromosomes strings that
have motivated the use of optimization methods in the sense of represent an organism (a possible solution in the design space). In-
turn the composite material structural design a more systematic stead of working with just one search point in the design space, GA
and well defined task, becoming less dependent to the designer uses a population of designs that, by reproduction and selection
sensitivity and achieving the maximum material performance [2]. operations, evolve through successive generations. Many search
The earlier works in the field of composite structures optimiza- points dispersed in the design space prevent the GA to get stuck
tion employed the same methods already used to optimize conven- in locally optimal regions, avoiding a premature convergence of
tional material structures. These methods are based on gradients of the process. New designs are generated by the reproduction pro-
the objective and constraints functions with respect to the design cess that consists in the application of the genetic operators in par-
variables, which are considered to be continuous in the design ents selected from the existing population. These genetic operators
space. Such works resulted in limited success because composite are counterparts of the natural genetic mechanisms, acting over
laminate design falls on a discrete optimization problem, since in the chromosomal strings of the organisms [4]. The selection of par-
practice the variables are restricted to few values imposed by the ents for the reproduction process and the selection of organisms to
manufacturing process. Moreover, the composite optimization fill each new generation are both probabilistic. However, the
chances of selection of each organisms is proportional to its fitness,
* Corresponding author. Tel.: +55 51 3308 3587; fax: +55 51 3308 3999. as happen in the nature where fittest organisms have more
E-mail address: amawruch@ufrgs.br (A.M. Awruch). chances to reproduce and to continue in the next generation.

0263-8223/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2008.05.004
444 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

The organism fitness is obtained directly from an objective func- very common constraint adopted in most of optimization prob-
tion using simple structure information and gradient evaluations lems, and it is used in this work too. The Tsai–Wu failure criterion
are not required. [7] is used for the failure prediction in a ply. A safety factor against
In real designs cases, when the structural geometry is usually failure kf can be obtained in linear problems with the Tsai–Wu fail-
complex and the prediction of the structural behavior must be ure function using the material strength parameter for traction,
accurate, it is necessary to use numerical tools, such as the FEM, compression and shearing at each of the principal material axes.
for the structural analysis. These methods are computationally In geometrically nonlinear analysis the structural failure is verified
expensive and may turn the optimization processes with GA at each load step of the incremental solution method, which is
impracticable when a large amount of analyses is required. Many stopped if material failure is detected.
researchers have proposed modifications to the classical GA struc-
ture to take advantage of composite laminates characteristics and
minimize the computational cost. Some of these new strategies 3. Genetic algorithms for composite laminate structure
are applied in this work, consisting essentially in a GA restructur- optimization
ing of the variable codification and the genetic operators.
When more than one objective is handled in the design process, This work uses a GA provided with many modifications with re-
a multiobjective optimization problem may be solved considering spect to the classical GA structure stated by Goldberg [4]. Although
all the objectives simultaneously, providing a set of optimum de- the main concepts and the sequence of operations remain similar
signs (pareto-optimal set), depending of the emphasis given to to the original formulation, a new scheme for the variable codifica-
each one of the objectives. The pareto-optimal set may be very use- tion and special genetic operators are introduced, increasing the
ful when the critical objective is not known a priori. Various performance of the method in the case of composite structures
researchers have studied the problem of multiobjective optimiza- optimization.
tion of laminated structures, but the use of this approach together An initial population containing P organism is first created in a
with GA and the finite element method (FEM) have not been random process. In order to create successive generations, parents
widely explored. are chosen from the current population based on their fitness.
In the present work, two examples of multiobjective optimiza- Next, the genetic operators are applied to create children in order
tion of composite laminate plates using GA and FEM are studied. A to form a children population. An elitist selection scheme is used to
third application deals with the single optimization of a semi- obtain the new generation taking organisms from the current pop-
cylindrical shell considering the geometrically nonlinear behavior ulation and from the children population just created. This process
of the structure. is repeated until the convergence criterion is met. A description of
the variable codification, the genetic operators and the selection
scheme for the construction of new generations are given in the
2. Composite structure analysis following subsections.

Real problems of composite structure design depend of reliable 3.1. Composite laminate codification
structural analysis. In the case of composite laminates, the deter-
mination of the mechanical behavior is difficult even for simple A pair of chromosomes is adopted for the representation of each
geometric configurations. It happens because of some complex laminate in this work, similar to the scheme implemented by Sore-
mechanisms inherent to the material, like coupling between mekun [8]. As only symmetric laminates are studied in this work,
stretching, bending and twisting deformations, depending on the just half of the layers need to be coded in the gene strings, and
stacking sequence. These coupling effects and the usually complex so the total number of genes in a chromosome is proportional to
structural geometric configurations inhibit the use of closed-form half of the maximum admissible number of layers in a specific de-
solutions in the structural analysis. In consequence, numerical sign problem. Each layer is represented by a pair of genes (with one
method must be adopted for the accurate prediction of the struc- gene in each chromosome), being the first pair referred to the out-
tural behavior. ermost layer and the succeeding pairs of genes referred to the in-
In this work, a triangular flat plate and shell element with 18 ner layers. In the first chromosome, named ‘‘orientation
degrees of freedom called DKT (discrete Kirchhoff triangle) is used. chromosome”, is stored information about fiber orientation and
This element was developed by Bathe and Ho [5] for the nonlinear about the number of plies contained by each layer of the laminate.
analysis of isotropic plates and shells. The original formulation was The second chromosome, called ‘‘material chromosome”, is used to
modified by the introduction of specific constitutive matrices of store the material properties of the layer, containing information
the membrane and bending parts in order to allow the analysis like ply thickness, elastic and strength constants.
of symmetric composite laminated structures. Any other element Each design variable must assume one of the admissible dis-
formulation could be adopted, since in GA based optimization no crete values defined in the optimization process. These values are
gradient evaluation is needed, and so no additional modification represented by positive integer numbers. They are used to code
to the analysis tool is necessary. In the solution of geometrically de variable value in its correspondent gene. Since there are two
nonlinear problems an incremental iterative scheme referred as kinds of variables, the orientation and the material variable, two
generalized displacement control method (GDCM) is used (see code alphabets are used in the codification. An empty stack code,
[6]). This method is able to solve nonlinear problems with multiple represented by the number zero, is used in problems where the
critical points and snap-back points allowing the complete deter- variation of the number of layers is considered. This code is intro-
mination of the post buckling behavior of the structure. The only duced in a pair of genes, representing the layer to be deleted, by a
parameter to be set in the GDCM is the basic load increment ki, specific genetic operator. Another genetic operator acts adding lay-
which corresponds to the ratio between the first load increment ers to the laminate by changing the value zero of a pair of genes by
and the complete load. This parameter determines the sensibility any other admissible value. The maximum number of layers in the
of the method to the nonlinear characteristic of the problem. laminate is limited by the number of genes in the chromosome.
Additionally to the structure displacements, the analysis tool In Fig. 1 is presented an example of the decodification of a pair
must be able to determinate accurately the stress components at of chromosomes in a laminate stacking sequence, based in a given
the composite layers in order to predict material failure. This is a orientation and material gene codification alphabets. These chro-
F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454 445

more of its characteristics. This may lead to a less random opera-


tion, which is supposed to be more efficient in guiding the evolu-
tion towards optimization objectives. New operators named
orientation alteration, material alteration, ply addition and ply
deletion, are introduced to replace the classical mutation in the
present GA [8,9]. A description of these new operators is presented
below, followed by an example of their applications as it is shown
in Fig. 3.
Orientation and material alterations are implemented similarly
to the classical mutation, but are independently applied to orienta-
tion and material chromosomes, respectively. Different orientation
and material operator probabilities (poa and pma) may be adopted,
which is useful, since each chromosome may converge at different
velocities in most optimization processes. Additionally, these oper-
Fig. 1. Decodification of chromosomes in a laminate stacking sequence. ators are not applied to genes with empty stack code, when it is
present in the chromosome.
Two distinct operators are used to vary the number of layers in
mosomes could represent a laminate with up to 32 plies, since the laminate. The first one, named ‘‘ply addition”, acts in the chro-
each one of the 8 genes in the strings represent 2 plies, and only mosomal strings, at a given probability ppa, introducing a new layer
one half of the symmetric laminate is coded. However, the pres- close to the laminate mid-plane (end of chromosomes), and
ence of the code ‘‘0” in the first pair of genes indicates that this removing an existing pair of genes with empty stack code. The
layer does not exist, indicating that the laminate has only 28 plies. genes that represent the new layer are randomly created, assuming
any value contained in each respective alphabet codification. In an
3.2. Genetic operators opposite way, the ‘‘ply deletion” operator acts taking out the last
pair of genes of the organism chromosomes (innermost layer)
3.2.1. Crossover and adding a pair of genes with empty stack code at the outermost
Crossover is an essential GA operator, having the fundamental position. This operator is applied with a given probability ppd. Both
task of creating new organisms (children) in a reproduction pro- operators manipulate at the innermost laminate layer because it
cess. It acts combining genetic information taken from a pair of has little effect on bend or twist behavior of the plate or shell, caus-
organisms (parents) selected from the current population. The cre- ing no abrupt changes in the design.
ated child will hopefully be better than, or at least equivalent, in
fitness to its parents. The crossover operation is applied by first
generating a random number to define the crossover point. Then,
the gene strings of both material and orientation chromosomes
are split at the same point in both parents [8]. The left part of par-
ent 1 and the right part of parent 2 are combined to form a child, as
is shown in Fig. 2. The crossover operator is usually applied with
some probability, but in this work crossover is always used to cre-
ate children.

3.2.2. Mutation
Mutation is the class of genetic operators responsible to main-
tain the genetic diversity of the population by introducing new
information in the chromosomal strings of each child after it is cre-
ated by the crossover operation. These operations provide a ran-
dom search capability to GA, which may be useful to find
promising areas in the design space, and prevent crossover to lose
its effect due to a standardization of the population. In the classical
implementation of this operator, to each gene is given a small
probability to switch to any other permissible value, excepting
its current value [3].
In spite of the randomness of the mutation process, it is possible
to incorporate to this operator some knowledge about the response
of composite laminates with respect to the alteration of one or Fig. 3. (a) Orientation and material alteration; (b) ply addition; (c) ply deletion.

Fig. 2. Crossover operation.


446 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

PN i
i¼1 X g P
An ¼ ; ð1Þ
N
where X ig is the total number of generations analyzed in the ith opti-
mization procedure. In GA optimization procedures it is very com-
mon to occur repeated analyses because one specific design may
Fig. 4. Gene swap operation.
appear in many generations during the process. When a memory
containing information about the performed analyses is used, asso-
ciated to the GA, these repeated analyses can be avoided, resulting
3.2.3. Gene swap in an important reduction of the computational cost. In these cases,
The main characteristic of the gene swap operator is the ability another measurement of the computational cost can be stated tack-
to modify laminate stack sequence without changes of the total ing the average number of the effectively performed analyses (Ar),
number of plies with fibers oriented on each permissible direction. obtained dividing the number of analyses that were effectively car-
This allows GA to change the bending behavior of the laminate ried out by the number of GA executions (N).
without modifying its in-plane mechanical response. The imple- The criterion to stop the optimization process, which was used
mentation of this operator follows Refs. [8,9], where two pairs of in all examples presented here, is based in two parameters: the
genes are randomly chosen and have their position shifted in the upper limit of the number of generations (NLG) and the maximum
chromosome, resulting on a new stacking sequence. Such opera- number of generations with no improvement of the best design
tion occurs at a given probability pgs, usually with a larger value (NSD). Once one of these limits is reached, the optimization process
than those corresponding to mutation operators probabilities. A is stopped and the best laminate of the last generation is taken as
description of the gene swap operation is given in Fig. 4. the optimization result. NLG and NSD are defined in each optimiza-
tion procedure, depending on the complexity of a specific problem.
3.3. Selection scheme
4.1. Weight and deflection minimization of a composite laminated
There are many ways to obtain the population of successive plate under transverse load
generations in a GA. In classical algorithms new generations are
formed only by children created by genetic operators applied to This example deals with the design of a composite laminated
the current population. This process has many drawbacks since square plate, subjected to a uniform pressure load on its surface.
there is no warranty of improvement or maintenance of achieved Minimization of the structural weight and deflection are the design
evolution when all the old organisms are replaced. To solve this objectives. The two objectives must be considered in conjunction
problem new selection schemes were created, being one of them with the constraints imposed by material failure and maximum
the elitism scheme, which consists in transfer some good organ- values of contiguous plies thickness with the same fiber orienta-
isms from the old population to a new generation, preserving tion. These are two opposite objectives, since improvements in
desirable genetic information. This papers deals with a multiple one of them leads to depreciation of the other, and they must be
elitist scheme proposed by Soremekun [8]. In the implementation, considered at the same time in the optimization, requiring a mul-
both, parent and child populations of size P are independently tiobjective approach. In its formulation, the objective function
ranked from best to worst fitness. These two populations are then must contain both objectives, which are weighted by a factor
combined and ranked together, resulting in a combined population which controls the emphasis given to each one of the objectives
with 2P organisms. Then, the best Ne individuals of the combined in the optimization. As a result of the variation of the weighting
population are transferred to the new generation. The best individ- factor used in the objective function, this problem has a set of opti-
uals of child population that have not already been used are taken mal solutions (pareto-optimal set) instead of a single solution.
to fill the remainder of the new generation. The number of top ele- The structure geometry, boundary conditions and the mechan-
ments (Ne) to be transferred to the new generation is a GA param- ical properties of the composite material are presented in Fig. 5.
eter to be adjusted at each application. The elastic constants are the Young’s modulus in the fiber direction
(E1) and transverse to the fiber direction (E2), the shear modulus
4. Numerical examples and discussion (G12) and the Poisson’s ratio (m12), respectively. Strength parame-
ters for traction and compression for longitudinal and transversal
Three examples of composite laminated structures design using directions are given by F1t, F1c, F2t, and F2c, respectively. The
GA optimization and FEM analysis are presented in the following remainder parameters are the shear strength (F6) and the specific
sections. To prove the success of the optimization procedure and weight (q). The structure must support a design pressure load of
to characterize the design space of the problems, all the possible 0.1 MPa with no material failure (the Tsai–Wu failure function
laminate configurations are previously analyzed. Additionally, N must be lower than 1.0 for the whole plate) and thickness of con-
optimizations are carried out for each example to obtain the algo- tiguous plies with the same fiber orientation must not be greater
rithm reliability and computational cost. This is possible to accom- than 2 mm or less than 0.75 mm.
plish because results of previous analyses of the design space are In the plate design, the laminate is restricted to be symmetric
stored. This information is used by the GA to evaluate the objective with 8 layers, being represented in the GA by a pair of chromo-
function. somes, with 4 genes each one. The fiber orientation angle and
The apparent reliability (R) is determined by taking the number the thickness of each layer are the optimization variables. They
of optimizations where the GA finds at least one global optimum must assume one of the discrete values given in Table 1, where
(No), divided by the total number of applications of the GA (N). It the codification adopted in the GA is also given. The number of
defines the chances of obtaining the global optimum in a single genes combined with the number of possible values of each vari-
optimization process. As the structural analysis employing the able leads the size of the design space (SDS) to be equal to 65536.
FEM is usually the most time consuming task in the optimization In this example the fitness evaluation FIT must consider both,
procedure, the GA cost (An), which is determined by the average weight and deflection reduction, at the same time. It is done by
number of analyses required in a single optimization process, is Eq. (2a), where the fitness value is taken as the inverse of the objec-
given by the following expression: tive function, and the weighting factor a is introduced to allow the
F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454 447

Fig. 5. Structure geometry, boundary conditions and composite properties.

Table 1 Tv is referred to the violation of the limit of contiguous plies thick-


Genes alphabet and possible discrete variable values ness with the same fiber orientation. Tv is equal to the exceeding
Orientation gene alphabet Material gene alphabet value violating the limit fixed to the thickness of contiguous plies
Code Orientation angle Code Ply thickness (mm)
with the same fiber orientation. As an example, if the thickness of
each one of two contiguous plies with the same fiber orientation
1 1 ply at 0° 1 0.75
is equal to 1.5 mm, the exceeding value violating the limit (i.e.
2 1 ply at 45° 2 1.00
3 1 ply at +45° 3 1.50 Tv) is 1.0, since this work adopts a limit of 2.0 mm. The constant
4 1 ply at 90° 4 2.00 1.0 is added to Tv in order to avoid a division by zero in Eq. (2a)
when Tv = 0.
To allow the analysis of the optimization performance, GA is
executed 50 times for each a, which is taken varying from 0.0 to
variation of the emphasis given to each objective. The constraints 1.0 with increments equal to 0.05. A population size P = 50 and
are considered by a penalty formulation of the objective function, the elitist scheme parameter Ne = 5 are adopted (see Section 3.3).
where unfeasible designs have a reduction on their fitness propor- The genetic operators are used with the probabilities poa = 4%,
tionally to the magnitude of the constraints violation. The objective pma = 2%, and pgs = 80%, while the probability of ply addition (ppa)
function uses the dimensionless variables W* and D*, that represent and ply deletion (ppd) operations are set to zero, since the number
the total weight and the central deflection of the plate normalized of layers must remain fixed during the optimization (see Sections
by their maximum and minimum values. This approach is more 3.2.2 and 3.2.3). The parameters used as a criterion to stop the pro-
efficient than a formulation that uses directly the weight and dis- cess are NLG = 250 and NSD = 125.
placement value or even a simple dimensionless value of these The distribution of weight and central displacement of all the
variables divided by a reference value (see [10]). The lower and feasible designs in the problem is shown in Fig. 6. Points A–P in this
higher weight limits can be easily obtained by taking all the plies
thickness equal to 0.75 mm or equal to 2.00 mm, respectively.
The maximum and minimum displacements to be used are ob-
tained by the results of optimizations performed with a equal to
0.0 and equal to 1.0, respectively. When a is equal to 0.0 only the
displacement is reduced, and the optimization result is a design
that have the smallest displacement. When a is taken equal to
1.0 the GA obtains the lightest structure, which may have a large
displacement, and it is used as the maximum value in the normal-
ization. The normalization of the variables is given by Eq. (2b),
where the coefficient 1.0 is added to avoid nulls values of W* and
D*.
(
FIT ¼ ðTvþ1ÞðaW1 þð1aÞD Þ ; if FF 6 1;
1
ð2aÞ
FIT ¼ FFðTvþ1ÞðaW  þð1aÞD Þ
; if FF > 1

where
W  W min D  Dmin
W ¼ þ 1; D ¼ þ 1: ð2bÞ
W max  W min Dmax  Dmin
The parameters FF and Tv are introduced in Eq. (3a) to penalize
unfeasible designs. The first one represents the maximum value of
the failure function evaluated in the structure, while the parameter Fig. 6. Weight and central displacement of feasible designs.
448 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

figure are the designs that form the pareto-optimal set, which must of repeated structures is avoided using a memory, since Ar is found
be obtained by the GA, according to the emphasis given to each of to be about 40–57% of An, depending on the value of a. Further-
the objectives. Details of the pareto-optimal set are presented in more, the number of analyses effectively performed Ar is about
Table 2. 4.82–7.09% of the SDS, which means that only a small part of the
Good reliability levels were obtained for the optimization with design space is explored, but the total number of analyses is still
most of the a values, as can be observed in Table 3, where the col- huge. A reduction of the average value Ar can be obtained when
umn r represents the standard deviation of the apparent reliability the parameters of GA are modified. However, in all tests accom-
R. The loss of reliability in the optimizations using a equal to 0.55 plished here this reduction was obtained together with a reduction
occurs because the GA finds many times the design identified by L on the reliability levels, mainly for values of a presenting critical
instead of the design identified by J, which is the correct solution in reliability levels.
this case. It happens because both designs J and L have practically As can be seen in the Tables 2 and 3, the GA is successful in find-
the same fitness value for a equal to 0.55. The same occurs for a ing most of the pareto-optimal designs, but the designs identified
equal to 0.70, when the GA finds the design identified by O in many by the points B and D are not obtained. These points are possible
optimization processes instead of the design identified by N, which solutions of the optimization, but they are located out of a convex
is the correct solution. In optimizations where a is taken as being curve defined by the other optimal points. Due to this fact the GA
equal to 0.80, 0.85 and 0.95, the low level of reliability occurs be- does not find the points B and D, since the fitness is evaluated as a
cause the designs obtained in these cases have values of the fitness convex combination of the objectives. Figs. 7a and b show the dif-
function very close to the optimal designs identified by O and P. ference of the fitness values of the points B and D with respect to
The average number of analyses required (An), the average their neighbor points in a range of a where the optimal solution
number of analyses effectively performed (Ar) and the ratio be- changes from point A to C and from point C to E, respectively
tween Ar and the size of the design space (SDS), which is equal (see Table 3). The figures show that the fitness of the points B
to 65,536, are also shown in Table 3 for each value of a. An expres- and D are never greater than those of their neighbor points at
sive reduction of the computational cost is observed if the analysis the same time and so they cannot be obtained by the GA, no matter
the value of the weighting parameter.
All optimal designs obtained in the pareto optimal set present
Table 2 outer layers with fibers oriented at 90° and the maximum admis-
Pareto-optimal designs
sible thickness (2 mm). These layers are the most important for
Optimal Laminate Weight Deflection Weighting the plate bending behavior, and the previous characteristics give
design (N) (mm) factor a the best stiffness properties. The inner layers of the staking se-
A [902,0, +452,0, 902,0, 451,0]S 219.7 7.9 0.0–0.20 quence of the different designs vary to obtain the different results
B [902,0, 452,0, 902,0, +450,75]S 211.9 8.8 – for the structural weight and deflection, according to the emphasis
C [902,0, 451,75, 902,0, +451,0]S 204.0 9.6 0.25
given for each of the optimization procedures. The lightest design
D [902,0, 451,75, 902,0, +450,75]S 196.2 10.8 –
E [902,0, 451,0, 902,0, +451,0]S 188.4 11.8 0.30 obtained here has less than the half of the weight of the heaviest;
F [902,0, 450,75, 902,0, +451,0]S 180.5 13.2 0.35 however, the last one presented a displacement more than seven
G [902,0, 450,75, 902,0, +450,75]S 172.7 15.0 0.40 times lower than the first one. The chose of the design to be used
H [902,0, 450,75, 901,75, +451,0]S 164.8 17.3 0.45 in a specific application depends on how critical is the weight for
I [902,0, 450,75, 901,75, +450,75]S 157.0 19.9 0.50
J [902,0, 450,75, 901,0, +451,0]S 149.1 23.3 0.55
such application and on the magnitude of the allowable
L [902,0, 450,75, 901,0, +450,75]S 141.3 27.2 0.60 displacement.
M [902,0, 450,75, 900,75, 450,75]S 133.4 32.1 0.65
N ½901;0
2 ; 45
1;0
901;0 S 125.6 38.5 0.70 4.2. Cost and weight minimization of an in-plane loaded composite
O ½901;0
2 ; þ45
0;75
 451;0 S 117.7 46.6 0.75–0.85
laminate plate
P ½901;0
2 ; 0
0;75
þ 450;75 S 109.9 59.5 0.90–1.0

This example uses GA to obtain the optimum design of an in-


plane loaded plate of composite materials. The objective is to find
the lightest and cheapest laminate satisfying the design constraints
Table 3
with respect to structural stability and material failure. Two unidi-
Optimization results with GA for the square plate
rectional composite materials, Kevlar-epoxy and Graphite-epoxy,
a Optimal designs R (%) r (%) An Ar Ar/SDS (%) may be used to construct the stacking sequence, being the first
0.00 A 100 0.0 7401 3641 5.56 cheaper but heavier and less resistant than the last one. The vari-
0.05 A 100 0.0 7780 3781 5.77 ables to be manipulated by the GA are the number of layers of
0.10 A 100 0.0 7759 3852 5.88
the laminate, the material of each one of the layers and the fiber
0.15 A 100 0.0 7640 3842 5.86
0.20 A 100 0.0 7587 3923 5.99
orientation angles of the plies. Plate geometry, load and boundary
0.25 C 98 2.0 8659 4645 7.09 conditions are presented in Fig. 8. This figure also shows the elastic
0.30 E 100 0.0 8299 4463 6.81 constants, strength parameters, specific weight, ply thickness and
0.35 F 100 0.0 7681 4014 6.12 the parameter of cost per unit weight (C) for the Kevlar-epoxy
0.40 G 100 0.0 7949 4242 6.47
and Graphite-epoxy.
0.45 H 100 0.0 7694 4299 6.56
0.50 I 100 0.0 7750 4392 6.70 The safety factor for material failure kf is given by the Tsai–Wu
0.55 J 84 5.2 8386 4472 6.82 failure criterion [7], while the safety factor for structural elastic
0.60 L 98 2.0 8035 4095 6.25 instability kb is obtained solving the eigenvalue problem involving
0.65 M 100 0.0 8146 4146 6.33 the linear and the geometrically nonlinear stiffness matrices. A de-
0.70 N 80 5.7 7830 3747 5.72
0.75 O 94 3.4 8205 3568 5.44
sign is considered to be feasible when both kf and kb are grater or
0.80 O 78 5.9 7805 3175 4.84 equal to 1.0. For simplicity, the design cost is considered to be pro-
0.85 O 68 6.6 7798 3161 4.82 portional to the amount of each material contained in the different
0.90 P 92 3.8 8559 3404 5.19 plies of the laminate. This value is obtained by multiplying the por-
0.95 P 86 4.9 8067 3379 5.16
tion of the plate weight corresponding to one given material by its
1.00 P 92 3.8 8001 3347 5.11
cost per unit weight (denoted by C).
F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454 449

Fig. 7. Differences of the fitness values of points B and D with respect to their neighbor points. (a) When FIT(B) > FIT(A), then FIT(C) > FIT(B); (b) when FIT(D) > FIT(C), then
FIT(E) > FIT(D).

Fig. 8. Composite laminate plate under in-plane load.

The laminated is supposed to be symmetric and formed by 6–12 the unfeasible designs, driving the GA to feasible areas of the de-
layers. Each layer has 2 plies that may assume the following orien- sign space. A small ‘‘bonus”, proportional to the safety factor, is
tations: 0°2, ±45° and 90°2. They are represented in the orientation incorporated to the objective function in the case of feasible
gene alphabet by the codes 1, 2 and 3, respectively. The material
gene alphabet is formed by the codes 1 and 2, corresponding to
Kevlar-epoxy and Graphite-epoxy, respectively. Empty stacks code
can be used to reduce the number of layers. Due to the symmetry,
only 6 genes are necessary in each of the two chromosomes used to
represent the composite material in the GA. Considering the num-
ber of variables and the number of possible values of these vari-
ables, the size of the designs space (SDS) is 55,944. A distribution
of weight and cost of all the feasible designs in the problem is
shown in Fig. 9. Points A to F in this figure are the designs that form
the pareto-optimal set, which must be obtained by the GA. Details
of the points of the pareto-optimal set are presented in Table 4.
Due to the simultaneous minimization of cost and weight, the
multiobjective approach is used in the formulation of the objective
function, given in Eq. (3), which contains both objectives. The
weighting factor a is introduced to allow a shifting on the empha-
sis given to each of the objectives, driving the GA to converge to
one of the points in the pareto-optimal set, according to the value
attributed to this factor. The constraints of the problem are also
considered in the objective function. In this example, a feasible de-
sign is determined by the safety factor k*, which is given by the
minimum value between kb and kf. A penalization is applied to Fig. 9. Weight and cost of feasible designs.
450 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

Table 4 Table 5
Designs of the pareto-optimal set Optimization results with GA for the plate with in-plane load

Optimal Laminate Weight Cost kb kf a Pareto-optimal Optimal Near An Ar Ar/SDS


design (N) (uc) set designs optimal (%)
designs
A ½45ge ; 90ge ge ge
2 ; 02 ; 45 ; 02 S
ge
24.49 73.46 1.04 55.12
B ½90ge ge ge
2 ; 02 ; 452 ; 02 S
ke
25.44 64.62 1.30 30.93 R (%) r (%) R (%) r (%)
C ½90ge ge ge
2 ; 02 ; 45 ; 452 S
ke
26.39 55.77 1.50 31.84
0 F 100 0.0 100 0 3829 1238 2.21
D ½45ge ; 90ge
2 ; 0ke
2 ; 45 ke
; 90ke
2 S 27.34 46.93 1.64 18.00
0.1 F 100 0.0 100 0 3839 1714 3.06
E ½45ge ; 45ke ; 90ke 4 ; 0ke
2 S 28.30 38.09 1.56 17.16
0.2 F 100 0.0 100 0 3862 1709 3.05
F ½45ke ke
3 ; 904 S 29.25 29.25 1.30 16.04
0.3 E 100 0.0 100 0 3750 1683 3.01
0.4 E 100 0.0 100 0 3732 1723 3.08
0.5 D 40 9.8 100 0 4805 2209 3.95
0.6 C 20 8.0 100 0 4498 2336 4.18
designs. This ‘‘bonus” promotes an additional goal in the optimiza- 0.7 B 8 5.4 100 0 5999 2868 5.13
tion process, seeking to maximize the value of the safety factor. As 0.8 A 84 7.3 100 0 4172 2037 3.64
a consequence of this formulation, the result of the optimization 0.9 A 92 5.4 100 0 4740 2410 4.31
process is supposed to be the safest design having the best 1 A 88 6.5 100 0 4622 2219 3.97
weighted combination of weight and cost.
Parameters W* and C*, used in Eq. (3a), are the dimensionless to-
tal weight and cost of the plate, respectively. They are given by Eq.
but having smaller values of safety factor, are considered as valid
(3b) using the maximum and minimum possible values of weight
solutions. These are called ‘‘near optimal designs” and are found in
and cost, which are easily obtained by the extreme combination
every optimization and for all values of a, showing that GA is effi-
of materials and number of layers. Instead of working directly with
cient in finding near optimal solutions. Table 5 also shows the aver-
a linear combination of the dimensionless weight and cost, the
age number of analyses required (An), the average number of
implemented objective function uses the square of these variables,
analyses effectively performed (Ar) and the ratio between Ar and
already weighted by the factor a. This alternative formulation is
the size of the design space (SDS), for each value of a. The algorithm
necessary because the results of optimizations that uses a linear
has explored only a little part of the design space, but for optimiza-
combination of W* and C* as objective function is always one of
tions presenting low reliability levels the search was extended to
the extreme points A or F (see Fig. 9). It occurs because the pare-
more generations to seek for the best solution, requiring more anal-
to-optimal set is arranged in a straight line in the weight-cost
yses. Using computer memory to avoid repeated analyses during the
plane. The new formulation provides a curved distribution of opti-
optimization gives a significant reduction on the computational cost,
mal points, allowing the GA to find them. The optimal solution is
as can be seen from the difference between An and Ar in Table 5.
defined as the point which is located at the closest ‘‘distance” from
Although the variation of the number of layers is allowed in this
the origin of the weight-cost reference system, being this ‘‘dis-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi example, all the designs that form the pareto optimal set are com-
tance” given by ‘‘ ½aðW  Þ2 2 þ ½ð1  aÞðC  Þ2 2 ”. posed by 20 plies. As layers with two plies and the symmetry con-
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 dition have been used, laminates with total number of plies that
>
> are multiple of 4 have been obtained. In this case, the optimization
>
< OBJ ¼ ½aðW  Þ2 2 þ ½ð1  aÞðC  Þ2 2 þ 106 k ; if k P 1;
process showed that laminates with 12–16 plies are unfeasible,
> qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 while laminates with 24 plies are less efficient than those with
>
>
: OBJ ¼ ðk Þ2 ½aðW  Þ2 2 þ ½ð1  aÞðC  Þ2 2 ; if k < 1; 20 plies. The transition of the heaviest and less expensive design
(F) to the lightest and most expensive design (A) is accomplished
ð3aÞ
modifying gradually the material of the different layers. In each
where subsequent design contained in the pareto set, the outer layers
composed by Kevlar-epoxy were replaced by a graphite–epoxy
W  W min C  C min
W ¼ þ 1; C ¼ þ 1: ð3bÞ layer, taking the maximum advantage of the stiffest material. The
W max  W min C max  C min
variation of the weight observed from design A to F is much lower
To study the GA performance, 25 optimization processes were exe- than the variation of the cost. This implies that, unless for weight
cuted for each a, which is taken varying from 0.0 to 1.0 with incre- critical applications, the cheapest design is the best one. However,
ments equal to 0.1. The GA is used with a population size P = 30 this conclusion may be strongly affected by the way the cost is
and a elitist parameter Ne = 4, together with the following probabil- evaluated or even if the cost parameter (denoted by C) is modified.
ities for the genetic operators: poa = 4%, pma = 2%, ppa = 4%, ppd = 8%
and pgs = 80%. The parameters of the criteria used to stop the process 4.3. Stiffness maximization of a composite laminated shell with
are NLG = 300 and NSD = 100. The results of the optimizations per- geometrically nonlinear behavior
formed here show that the GA can obtain one optimal design for
every tested a, but not all of the 25 GA executions performed with This example deals with the application of the GA to obtain
each value of a are successful. This happens because GA has diffi- the design of a composite shallow shell with the maximum stiff-
culty to search for the optimal design in regions where the objective ness with respect to a pressure load. The nonlinear behavior of
function has low gradients, as a consequence of the introduction of the structure is considered in the fitness evaluation by taking into
the ‘‘bonus” proportional to the safety factor. There are many points account the critical load level and the maximum displacement of
(designs) with fitness values very similar to the optimal design fit- the structure. These values are obtained by the geometrically
ness value, having the same weight and cost of the optimal design, nonlinear analysis of the problem. The structural optimization is
but with a smaller safety factor. Table 5 shows the apparent reliabil- performed by the GA through the manipulation of the laminate
ity (R) and its standard deviation (r), obtained from GA the 25 opti- stacking sequence, considering a fixed number of layers. Con-
mizations for each value of a, considering two situations. In the first straints referred to material failure and number of contiguous
column of Table 5 optimizations founding only one optimal design plies with the same fiber orientation are imposed to the problem.
are computed as successful, while in the second column all designs Two cases of optimization are studied in this example, consid-
with the same weight and cost of those included in the first column, ering the same geometry of the structure, but different magnitudes
F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454 451

of the pressure load and thickness of the laminate, what results in


different levels of nonlinearity. In the first case the structure is sub-
jected to a pressure load q = 0.25 MPa, and the laminate have a to-
tal thickness h = 12.6 mm. The whole design space is previously
analyzed, allowing the study of the effect of some GA parameters
over the reliability and computational cost of the optimization pro-
cess by numerous applications of the GA. The second case consid-
ers a pressure load q = 0.125 MPa and the laminate thickness
h = 6.3 mm. The nonlinear structural behavior observed in this case
is stronger than in the first case, arising some difficulties to the
analysis requiring a larger number of steps to produce an accurate
response for most laminate configurations. A strategy for the auto-
matic refinement of the basic load increment ki, used in the nonlin-
ear analysis by the GDCM [6], is introduced to reduce the
computational cost. The effect of this strategy is studied in four
optimization processes performed for the second case. In both
cases the laminate is symmetric and formed by 14 layers that are
represented in the GA by a chromosome containing 7 genes. The
material chromosome is not necessary in this example, since the
unique layer characteristic that is allowed to be changed in the Fig. 11. Critical load level and maximum displacement of feasible designs.
optimization process is the fibers orientation.
The boundary conditions and structural geometry are shown in
Fig. 10, together with the elastic and strength parameters of the ering the number of genes in the chromosomal string and the num-
glass–epoxy composite used in this example. FEM is used to solve ber of possible values for each gene, the size of the design space in
the geometrically nonlinear problem. this example is 2187. The whole design space has been analyzed to
The fitness value of each design obtained by the GA during the allow the evaluation of the GA, and the feasible designs are shown
optimization process is defined by Eq. (4), where two parameters in Fig. 11, where the optimal design is indicated.
are used to characterize the structural stiffness and other two The optimal design, indicated in Fig. 11, is defined by the stack-
parameters are used to consider the constraints violation. The ing sequence [(904, ± 45)2, 902]S and the following value of the
parameters used to evaluate the structural stiffness are the critical parameters were obtained: NCcrit = 0.563, Umax = 0.0272 m and
load level (NCcrit), determined when the curve pressure  displace- FIT = 20.698. As expected, the optimal design has most of its plies
ment at the central point A reaches the first peak, and the value of with fibers oriented at 90°, providing the best bending properties.
the maximum displacement at the same point (Umax), which is ta- The 45° plies are introduced in order to satisfy the constraint re-
ken at the end of the load increment or when material failure is ob- lated to the number of contiguous plies with the same fiber orien-
served. In the case of unfeasible designs, the maximum load level tation. The curve load  displacement at point A for the optimal
acting on the structure without material failure (NCmax), and the design is presented in Fig. 12, together with the same curve for a
number of violations of the limit of 4 contiguous plies with the design configuration with a smaller fitness value. In this case the
same fiber orientation (Vnlc) are used as penalizations. stacking sequence is [±453, 04, ±452]S, NCcrit = 0.228, Umax =
0.0307 m and FIT = 7.427. In the analyses executed for all the
!
NC crit  NC 2max points of the designs space the basic load increment ki in the GDCM
FIT ¼ : ð4Þ was taken equal to 0.05.
U max  ðV nlc þ 1Þ
Nine combinations of the GA parameters P and NLG, presented in
The constant value 1.0 is added to Vnlc in order to avoid a division by Table 6, are used to study the influence of these parameters in the
zero in Eq. (4) when Vnlc = 0. In the first case studied in this example, optimization performance. Considering that the maximum number
considering q = 0.25 MPa and h = 12.6 mm, each of the 14 layers of analyses (Amax) to be executed in a optimization process is deter-
that form the laminate contains two plies, oriented at 0°2, ±45° or mined by the product (NLG + 1)  P, it is possible to impose an
90°2. These orientations are represented in the chromosomal string upper limit to An by adjusting these two parameters. The values
by the codes 1, 2 and 3, respectively. The resulting laminate is of P and NLG are obtained by working with ranges of Amax/SDS equal
formed by 28 plies with an individual thickness of 0.45 mm. Consid- to 12.5%, 25% and 50%. Using the expression for Amax stated before

Fig. 10. Shallow shell geometry, boundary conditions and material properties.
452 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

Fig. 12. Central displacement  load level for the optimal design and for a low fitness value.

Table 6 tions 1, 2 and 3, referred to Amax/SDS = 12.5%, low computational


Combination of the parameters P and NLG
cost is obtained together with unacceptable levels of reliability.
Amax/SDS = 12.5% Amax/SDS = 25.0% Amax/SDS = 50.0% The other combinations have generated good reliability levels
Comb P (Ne) NLG Comb P (Ne) NLG Comb P (Ne) NLG (over 95%, excepting combination 4 which has R = 90%) and the in-
(NSD) (NSD) (NSD) crease observed in the computational cost is not proportional to
1 7 (1) 38 (30) 4 7 (1) 77 (44) 7 7 (1) 155 (55) the increase in Amax/SDS to 25% and 50%. Combinations with smal-
2 14 (2) 19 (15) 5 14 (2) 38 (22) 8 14 (2) 77 (27) ler values of P have shown to be not so expensive in terms of com-
3 21 (3) 12 (10) 6 21 (3) 25 (15) 9 21 (3) 51 (18) putational cost than the others belonging to the same range of
Amax/SDS. A reduction of 30% in the number of analyses to be exe-
cuted in each optimization is observed as a consequence of the
and adopting the population size P equal to 7, 14 and 21 (propor- adoption of a strategy to avoid the repetition of analyses that have
tional to 1, 2 and 3 times the number of genes in the chromosome), been performed previously. This aspect is more expressive in opti-
it is possible to determine the value of NLG for each range of Amax/ mizations using smaller values of P, as for example in combination
SDS. The values of the parameters Ne and NSD are conveniently ad- 7, where the saving in computational cost is of about 44%.
justed in each of the nine combinations, but these parameters are As it is shown in Fig. 13, the optimizations using small popula-
also decisive in the efficiency of the process. The genetic operators tions and many generations have the best performance in terms of
probabilities adopted here are poa = 4% and pgs = 80%, being the computational cost, maintaining good reliability levels. Although
remainder set of probabilities equal to zero because they are not the GA performance is highly dependent on the problem to be
used in this problem. The GA is executed 50 times for each of the solved, the tendencies observed in this example may be extended
nine combinations of parameters in order to study their influence to many other applications, helping to determine suitable values
in the apparent reliability (R) and in the computational cost of of the parameters P and NLG. Furthermore, if optimizations result-
the optimization process. ing in designs with fitness values grater or equal to 99% of the opti-
Fig. 13 shows the performance of the GA considering each one mum design (quasi optimal designs) are considered as being
of the combinations of P and NLG tested in this work. In combina- successful, the reliability level is increased to 100% in all the nine
combinations, as it is shown in Fig. 13. In fact, GA can find designs
very close to the optimum, exploring only few points in the design
space.
In the second case studied in this example, considering
q = 0.125 MPa and h = 6.3 mm, each of the 14 layers that form
the laminate contains only one ply, which must be oriented at
45°, +45° and 90°. These orientations are represented in the chro-
mosomal string by the codes 1, 2 and 3, respectively. The individ-
ual ply thickness (0.45 mm) and the size of the design space (2187)
are the same of the previous case. The GA parameter are adjusted
with the values P = 7, Ne = 1, NLG = 155, NSD = 55, poa = 4% and
pgs = 80%, according to the best combination found for the previous
case.
The reduction of the laminate thickness leads to a stronger non-
linear behavior of the structure, even if the load is reduced by the
same factor. As a consequence of this fact, many laminate configu-
rations present in the design space of this example result in struc-
tures that cannot be accurately analyzed if the basic increment
load ki is taken equal to 0.05, as it was adopted in the previous case.
Fig. 13. GA performance with different combinations of P and NLG. There is not a method to determine a priori the best value for the
F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454 453

parameter ki. If the adopted value is too big, the nonlinear solution
is inaccurate or even may not converge. In the other hand, if the va-
lue is smaller than the necessary, the analysis process may became
very slow, which is very undesirable when an optimization by GA
is used. The challenge in this case is to find a value for ki that pro-
vide accurate analyses for all the designs with the minimum com-
putational cost.
In this example a strategy for the automatic refinement of the
parameter ki is implemented to reduce the cost of the analyses
and to guarantee the convergence and the accuracy of the nonlin-
ear solution. The first step in the implementation is carried out
monitoring the structural response during the nonlinear solution
and identifying anomalous behaviors that can be interpreted as
solution instability or error. When this situation occurs, the pro-
cess is stopped and restarted using a smaller value for ki. The
way the structural response is monitored and the factor of reduc-
tion of the parameter ki are problem dependent and can be imple-
mented in different forms. In the present example the error is
identified when the increment in the displacement of the central Fig. 14. Load level  central displacement for the best design obtained in optimi-
point of the shell in a specific load step is greater than 10 times zations 1, 2, 3 and 4.

the maximum displacement verified in the previous load step.


The refinement of the parameter ki is accomplished by taking one
half of its current value. A new structural analysis in the optimiza- which store previous results. In the optimization 4 the number of
tion process using GA to look for a new design is always started analyses is reduced in 50%. Some advantages were observed in
using the value of ki defined by the user, even if the refinement the use of the refinement strategy. First, it is possible to reduce
process was used in the previous analysis. the average time of the analyses (Tan) from 1.000 in the optimiza-
The results of four optimizations executed for this case are pre- tion 1 (when the refinement strategy was not used) to 0.616 in the
sented in Table 7. The optimizations 1 and 2 were executed using optimization 4, resulting in 32% of time saving in the total optimi-
the fixed value ki = 0.025, while the optimizations 3 and 4 were zation time (Ttotal). The second advantage lies in the guarantee of
executed using the value ki = 0.05, together with the refinement the correct solution for the nonlinear analysis of the structure,
strategy. In the optimizations 1, 3 and 4 the optimal solution for avoiding the use of a wrong structural response by the GA, as oc-
the problem is given by the laminate [904, 45, 902]S, with the fol- curred in the optimization 2.
lowing values for the critical load and maximum displacement:
NCcrit = 0.451 and Umax = 0.02895 m. This stacking sequence is al- 5. Final remarks
most similar to that obtained for the best design of the previous
example. It contains many plies with fiber oriented at 90°, in order The GA was successfully applied to obtain the optimal design of
to maximize the shell stiffness, with a 45° ply introduced to attend composite laminate structures such as plates and shells subjected
the problem constraint. The best design found by the optimization to different load conditions. Two examples of multiobjective opti-
2 is the laminate [902, 452, 90, 45, 90]S; however, the nonlinear mization were presented, and some aspects related to the formula-
analysis of this structure, using ki = 0.025, results in a wrong solu- tion of the objective function and its influence on the optimization
tion, invalidating this optimization. The load level  central dis- process were discussed. The performance of the GA in terms of
placement curves for the designs found in optimizations 1, 3 and computational cost and reliability was studied, showing that the
4, together with the curve referred to the right solution (using method is very efficient in finding near optimal solutions, and an
ki = 0.0125) of the design found in optimization 2, are shown in important saving in computer time can be obtained by de use of
Fig. 14. In this figure an additional region with snap-back and suitable values for the GA parameters and when results of different
snap-through points is observed in the curve corresponding to analyses are stored.
the best design of optimization 2. This behavior occurs in many de- In the third example the GA was used together with a nonlinear
signs explored during the optimization process, making the nonlin- FEM analysis to maximize the stiffness of a composite shell. Two
ear analysis of such structures much more difficult since these cases of laminate thickness and load level were considered to ex-
behaviors are correctly reproduced only when small values of the plore different levels of nonlinearity of the structural response.
parameter ki are used. The influence of the population size P and the limit of the number
The values adopted for the GA parameters resulted in a good of generations NLG over the reliability and the computational cost
performance in terms of computational cost, considering the small of the method were investigated in this example. The results dem-
size of the design space in this problem. Furthermore, a great onstrated that when relatively small populations associated with a
reduction in the number of analyses to be executed in each optimi- large limit of the number of generations are used, better perfor-
zation is obtained avoiding repeated analyses using a memory mances of the GA are obtained. A strategy for the automatic refine-
ment of the basic load level (ki), used in the nonlinear analysis, was
Table 7 proposed. It showed to be very important for the improvement of
Optimization performance with or without the refinement strategy
the optimization process, since a sensible reduction of the average
Optimization ki An Ar Ar/SDS (%) Ttotala Tana time of the structural analysis was obtained.
1 483 293 13.4 1.000 1.000
2 0.025 441 297 13.6 0.996 0.982
3 0.05 504 259 11.8 0569 0.643 Acknowledgement
4 With refin. 651 325 14.8 0.682 0.616
a
Ttotal and Tan dimensionless values obtained with reference to the The authors wish to thank the Brazilian agencies CNPq and
optimization 1. CAPES for its financial support.
454 F.S. Almeida, A.M. Awruch / Composite Structures 88 (2009) 443–454

References [6] Yang Y, Shieh M. Solution method for nonlinear problems with multiple
critical points. AIAA J 1990;28(12):2110–6.
[7] Daniel IM, Ishai O. Engineering mechanics of composite materials. Oxford
[1] Walker M, Smith RE. A technique for the multiobjective optimization of
University Press; 1994.
laminated composite structures using genetic algorithms and finite element
[8] Soremekun GAE. Genetic algorithms for composite laminate design and
analysis. Compos Struct 2003;62:123–8.
optimization. M.Sc. thesis (Master of Science in Engineering Mechanics).
[2] Gürdal Z, Haftka RT, Hajela P. Design and optimization of laminated composite
Polytechnic Institute and State University, Blacksbourg, Virginia, USA; 1997.
materials. Wiley & Sons; 1999.
[9] Nagendra S, Jestin D, Gürdal Z, Haftka RT, Watson LT. Improved genetic
[3] Soremekun GAE, Gürdal Z, Haftka RT, Watson LT. Composite laminate design
algorithm for the design of stiffened composite panels. Comput Struct
optimization by genetic algorithm with generalized elitist selection. Comput
1996;58(3):543–55.
Struct 2001;79:131–43.
[10] Almeida FS. Laminated composite material structures optimization with
[4] Goldberg DE. Genetic algorithms in search, optimization, and machine
genetic algorithms. M.Sc. thesis – PPGEC/UFRGS, Porto Alegre, Rio Grande do
learning. Addison-Wesley Publishing Company; 1989.
Sul, Brazil; 2006 [in Portuguese].
[5] Bathe KJ, Ho L. A simple and effective element for analysis of general shell
structures. Comput Struct 1981;13:673–81.

Vous aimerez peut-être aussi