Optimization of Spring Weight Using Genetic Algorithm IJERTV1IS9507 PDF

International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 1 Issue 9, November- 2012
Optimization Of Spring Weight Using Genetic Algorithm

A. M. Satyanarayana, B. T. N. Charyulu, C. S. C. V. Ramanamurthy Naidu, D.
G. Appala Naidu
1. Abstract:
Optimization is a technique through
which better results are obtained under
In this paper, a non traditional
certain circumstances. The present work
optimization technique, namely Genetic
deals with the problem of weight
Algorithm is used. Optimal design of
minimization of Spring.
/Spring is solved using Genetic Algorithm.
Spring problem has large number
From the results obtained, it is observed
of multivariable and non-linear equations /
that the optimum dimensions and weight
in equalities. Hence traditional
obtained by GA are closer and they are
optimization techniques cannot be applied
RRTT
much better than the values obtained by
in these cases. Traditional optimization
Consol-Opt cad
techniques have a possibility for the
IIJJEE
In the present study, weight

solutions to get trapped into local minima.
minimization of a spring has been
Also the algorithm developed for one type
investigated. Powerful evolutionary
of problem may not be suitable for another
algorithm (Genetic Algorithm) has been
type of problem.
used to solve this problem.
Keywords: Optimization, Spring, Genetic
A. M.Satyanarayana, Roll
No.10A01D1505, NOVA College of Algorithm, Weight.
Engg.& Tech.Jangareddigudem,
2. Introduction
W.G.Dist,A.P Corresponding author,
Email:satyamallapu@gmail.com, Design optimization can be defined
B. Assistant Professor, Department of
as the process of finding the maximum or
Mechanical Engineering, NOVA College
of E&T ,Jangareddigudem,W.G.Dist minimum of some parameters which may
A.P
call the objective function and it must also
C. Professor, Department of Mechanical
Engineering, KHIT, Guntur, A.P satisfy a certain set of specified
D. G.AppalaNaidu,Roll No.10A01D1501
requirements called constraints. Many
NOVA College of Engg.& Tech.
W.G.Dist,A.P methods have been developed and are in
www.ijert.org 1
ISSN: 2278-0181
use for design optimization. All these successfully applied in the area of
methods use mathematical programming Mechanical Engineering. Popular
techniques for solutions. applications include machine elements
In these cases it is difficult to apply design, heat transfer, scheduling, vehicle
traditional optimization techniques. Non- routing, etc.
conventional techniques are applied to 2.1. Optimal Problem Formulation
such cases. These are potential search and The objectives in a design
optimization techniques for complex problem and the associated design
engineering problems. Genetic algorithms parameters vary from product to product.
are found to have a better global Different techniques are to be used in
perspective than the traditional methods. different problems. The purpose of the
Dr. Shapour Azarm [1, 2] has formulation procedure is to create a
worked on optimization of helical spring mathematical model of the optimal design
using consol-optcad and he extended his problem, which then can be solved using
RRTT
work to fly wheel using the same an optimization algorithm. An
optimization procedure. optimization algorithm accepts an
IIJJEE
S.Vijayarangan, V.Alagappan [3] optimization problem only in a particular

was applied Genetic Algorithm technique format. Figure 1 show the common steps
for machine component i.e. Leaf spring. involved in an optimal design formulation
By using Genetic Algorithm, the optimum process. The first step is to realize the need
dimensions of the leaf spring were found for using optimization for a specific design
to be minimum weight with adequate problem. Then the designer needs to
strength and stiffness. choose the important design variables
Kalyanmoy Deb [5] presented associated with the design problem. The
different types of optimization algorithms formulation of optimal design problems
viz., traditional and non-traditional requires other considerations such as
algorithms. Genetic and simulated constraints, objective function, and
annealing algorithms were explained with variable bounds. Usually a hierarchy is
examples in non traditional algorithms. followed in the optimal design process,
Genetic Algorithms have good although one consideration may get
potential as optimization techniques for influenced by the other.
complex problems and have been
www.ijert.org 2
ISSN: 2278-0181
the constraints are formulated to satisfy

stress and deflection limitations. Often, a
component needs to be designed in such a
way that it can be placed inside a fixed
housing, there by restricting size of the
component. The nature and number of
constraints to be included in the
formulation depend on the user.
There are usually two types of
constraints that emerge from most
considerations. Either the constraints are
of an inequality type or of an equality
type.
2.3. Objective function and Variable
RRTT
Fig.1 Flow chart for general Optimal bounds:
design Procedure. The common engineering

IIJJEE
2.2. Design variables and Constraints: objectives involve minimization of overall

cost of manufacturing or minimization of
The formulation of an
overall weight of a component or
optimization problem begins with
maximization of net profit earned or
identifying the underlying design
maximization of total life of a product or
variables, which are primarily varied
others. Instead the designer chooses the
during the optimization process. Other
most important objective as the objective
design parameters usually remain fixed or
function of the optimization problem and
vary in relation to the design variables.
the other objectives are included as
The constraints represent some-
constraints by restricting their values with
functional relationships among the design
in a certain range. The objective function
variables and other design parameters
can be of two types. Either the objective
satisfying certain physical phenomenon
function is to be maximized or minimized.
and certain resource limitations. Some of
The final task of the formulation
these considerations require that the design
procedure is to set the minimum and the
remain in static or dynamic equilibrium. In
maximum bounds on each design variable.
many mechanical engineering problems,
www.ijert.org 3
ISSN: 2278-0181
Certain Optimization algorithms do not hK(X)=0, k=1,2,………,K;

require this information. In these Xi (L)  Xi  Xi (U) , i =l,2,……….,N;
problems, the constraints completely The constraints must be written in a way
surround the feasible region. In general, all so that the right side of the inequality or
N design variables are restricted to lie with equality sign is zero.
in the minimum and the maximum bounds 3.1. Optimization Algorithms.
as follows: Certain problems involve linear
Xi (L)  Xi  Xi (U) for i = 1, 2… N. terms for constraints and objective
In any given problem the function but certain other problems
determination of the variables bounds involve nonlinear terms for them. Some
(L) (U)
Xi and Xi may be difficult. One way algorithms perform better on one problem,
to remedy this situation is to make a guess but may perform poorly on other
about the optimal solution and set the problems. That is why the optimization
minimum and maximum bounds so that literature contains a large number of
RRTT
the optimal solution lies within these two algorithms, each suitable to solve a
bounds. The chosen bound may not be particular type of problem. The
IIJJEE
correct. The chosen bound may be optimization algorithms involve repetitive

readjusted and the optimization algorithm application of certain procedures they need
may be simulated again. to be used with the help of a computer.
3. Optimization Problem Format: 4. Working Principle.
The optimization problem can be When the simple Genetic
mathematically written in a special format, Algorithms is implemented it is usually
known as nonlinear programming format. done in a manner that involves the
Denoting the design variables as a column following operators such as fitness
vector X= (xl, x2,……….., xN)T, the objective proportionate,reproduction, crossover and
function as a scalar quantity f(X), mutation. The Genetic Algorithm begins
J, inequality constraints as gj(X)  0, and k with an initial population of size is equal
equality constraints and hK(X)=0; to 10 (zeroth generation) of strings which
Mathematical representation of the consists of binary coding selected at
nonlinear programming problem: random. Thereafter convert the binary
Minimize f(X) coding in to decimal value and calculate
Subject to gj(X)  0, j= 1,2,………,J; the actual values of design variables with
www.ijert.org 4
ISSN: 2278-0181
in their range by using mapping rule. Next

calculates the value of objective function
by substituting the design variables and
thereafter violation coefficient (C) and the
corresponding modified objective function
is obtained.
This modified objective function is
used to calculate the fitness values of each
individual string and also calculate the
average fitness value. First, the
reproduction operator makes a match of
two individual strings for mating, and then
crossover site is selected at random with in
the string length and position values are
Fig.2. Flow chart of Genetic algorithm
RRTT
swapped between the same strings.
5. Optimal Design of Spring.
Finally the mutation is applied to
5. 1 Introduction
IIJJEE
the randomly selected strings. It is the

Mechanical springs are used in
random flipping of the bits or gene that is
machines to exert force, to provide
changing a zero to one and vice versa, so
that the first cycle of operation is flexibility, and to store or absorb energy.
completed which is called as iteration.
In general, springs may be classified as
This iteration is termed as generation in
wire springs, flat springs or special-shaped
genetic Algorithms. This new generation
creates a second population (i.e. springs and there are variations within
generation -1). This second population is
these divisions. Wire springs include
further evaluated and tested for optimal
helical springs of round or square wire and
solution which is obtained at zero
violation and at the higher fitness value. are made to resist tensile, compressive or
Genetic algorithm is a population based
torsional loads. Cantilever and elliptical
search and optimization technique. It is an
types belongs to flat springs, the wound
iterative optimization procedure. Genetic
algorithm works as follows motor or clock-type power springs, the flat
www.ijert.org 5
ISSN: 2278-0181
spring washers are often called Belleville storage capacity. In such analysis only one
springs. objective is considered at a time for a
The main applications of the springs are as given application. In this case consider the
follows: optimum design of a valve spring.
(i) To act as a reservoir 5.2 Objective Function

The objective function is to minimize
energy, e.g. springs in
the weight of the spring.
clocks, toys or movie-
The weight of the spring is the
cameras.
sum of the weight of active and
(ii) To absorb shocks and
inactive coils.
vibrations, e.g. vehicle
 2
suspension spring. Minimum of N  Q Cd 3
4
RRTT
(iii) To return the mechanical N and Q are the number
IIJJEE
part to its original position of active and inactive coils
when it has been respectively.
temporarily displaced, e.g. Q is a constant and N
springs in valves, clutches depends on the spring gradient k.
and linkages. Where k is defined as:
(iv) To measure force, e.g. FU  FL Gd 2

k 
 8D 3 N
spring balance
In certain applications So it can be written a

 2   Gdc 3 
springs are designed with a specific Minimum of   Q Cd 3
4  8k 
objective such as minimum weight,
minimum volume or maximum energy
www.ijert.org 6
ISSN: 2278-0181
increase the maximum shear stress in the
spring, possibly resulting in a premature
d
fatigue failure. In a high speed mechanism
the natural frequency of the spring is taken
as 13 times the fundamental frequency of

D
the applied force.
f = 13 fan
This requirement (fn  f) is
expressed as follows:
Fig.5.1 Helical compression

fig 5.1 Helical compression spring
g1: K1C-1d2  1
spring
Where
RRTT
Gf
5.3 CONSTRAINTS:- K1 
112800( FU  FL )
IIJJEE
There are 11 constraints in this problem.
They are 5.3.2 Buckling (g2):-
The critical length of spring is a

5.3.1 Surging (g1):- function of boundary conditions and is
If the spring is used in a high speed normally presented graphically. An
mechanism and a condition of resonance approximate equation to avoid buckling
exists with the forcing function, large conservatively assuming both ends hinged
amplitude longitudinal vibration will is
result. This will appreciably affect the
d2
L=Nd(1+A)  11.5k
force exerted by the spring on the 4 Fu
components of the mechanism also, such Gd 4

Where N=
8D 3 k
internal vibrations will significantly
www.ijert.org 7
ISSN: 2278-0181
g2: K2C-5  1 number, IU and a minimum number, IL.
Experience indicates that extreme values
GFU (1  A) are IL  4
Where K2 
22.3k 2
IU  20
5.3.3 Minimum number of coils (g3):-
The corresponding constraints are:
A lower limit on the number of
g4: K4C  1
coils must be imposed in order to obtain
g5: K5C-1  1
accuracy and control in manufacturing. A
value of three is normally used as an

Where
absolute minimum but a larger value can
1
K4 
be selected to keep the pitch angle small IU
RRTT
(Preferably 120-150). This requirement is K5  IL
Gd 4
IIJJEE
Where N=
8D 3 k
5.3.5 Pocket Length (g6):-
(5.3.3.2)
When space is limited, the length
Substituting (5.3.3.2) into (5.3.3.1)
of a spring in its machine setting must be
and normalizing results to the constraint:
below an upper limit. The maximum
g3: K3C3 d-1  1
length (Lm) at maximum force (Fu) is
L+Qd  Lm
8kNmin
where K3  Substituting (5.3.2.1) into (5.3.2.2)
G
5.3.4 Spring Index (g4):- and normalizing gives
The spring index (defined by g6: K6d2C-3 +L6d  1
C=D/d) must be bounded by a maximum
www.ijert.org 8
ISSN: 2278-0181
5.3.6 Maximum Allowable The strength of the spring material
Outside Diameter (g7):- is a function of the wire diameter. This
Maximum allowable outside relation was found for the given limits of
diameter, OD the wire diameter for different materials.
g9 : K9 d-1  1
D+d  OD
g10 : K-110 d  1
or in normalized form
g7: K7 (Cd +d)  1

5.3.9 Clash allowance (g11):-
Stresses are limited by the spring

Where
solid condition. Therefore if the maximum
1
K7  5.3.7 Minimum
OD working load approaches the spring
RRTT
Allowable Inside Diameter (g8) :- condition, it is only necessary to provide
IIJJEE
Minimum allowable outside sufficient clash allowance (i.e., the
diameter, ID difference in spring length between
D-d  ID maximum load and spring solid positions)
or in normalized form: to allow for any possible combinations of
g8 : C-1+K8 C-1d-1  1 tolerance stack-up, differential thermal
expansion and wear of parts. The usual
recommendation is to provide a clash
K 8  ID allowance of approximately 10% of the

Where
total spring deflection at the maximum
(5.3.7.3)
work load.
5.3.8 Upper and Lower on wire
Diameter (g9, g10):-
www.ijert.org 9
ISSN: 2278-0181
L2-Ls  0.1  Gf

Where K1 =
112800( FU  FL )
 = LF-L2
G=79982 N/mm3
f=500 Hz, =6.35mm

F  FL
Where = U
k Fu=133.4466N, Fl=80.06796 N
g2: K2 -C5 0
L2=[N(1+A)+Q]d Where
Ls=(N+Q)d
GFU (1  A)
K2= ,Cl
L2-Ls = NAd 22.3K 2
Constraint g11 in normalized form is earance constant =
g11 : K11 C3d-2  1 0.4
g3: K3C3 -d ≤0
RRTT
0.8(FU - FL )
Where k11 
AG
8KN min
Where K3= ; Nmin=3
IIJJEE
The objective is to minimize the G
weight of spring is given by g4: K4C  1
Minimize 1 1
Where K4= = =0.05
IU 20
2  G
W= { dC-3+Q} Cd3 g5: K5C-1  1
4 8K
Subjected to eleven Where K5=IL=4
constraints, Gi (X) 0,i=1,2,3…j g6:K6d2C-3+L6d  1
Where j=number of G (1  A)
Where K6=
8KLm
constraints =1 to 11
Q
Gi(X) can be L6= Lm= 31.75 mm .Q = 2.
Lm
written as
g7 : C-K7d-1 + 1  0
g1: K1d2-C  0
www.ijert.org 10
ISSN: 2278-0181
1 Optimal design of helical

Where K7 =
OD compression spring has been discussed by
OD = 38.1 mm. using Consol-Optcad and optimization
capability of Genetic Algorithms (GA).
g8: K8d-1-C+10
Two design variables and 11
Where K8=ID=19.05 mm. constraints are considered in the helical
g9: K9d-1  1 compression spring problem. The results
obtained for helical compression spring
g10: (K10)-1d  1
are shown in table A. It was observed that
g11: K11C -d  0
3 2
the optimum dimensions and minimum
0.8( FU  FL ) weight obtained by GA are better than the

Where K11=
AG values obtained by the Consol-Optcad
method.
5.3.8. Optimal Design Model
Mathematical model for optimal design of

RRTT
spring is summarized as
IIJJEE
2  G
Minimize W= { dC-3+Q} Cd3
4 8K
6. Results and Discussions
Subject to g1: K1d2-C  0 g2: K2 -C5 0
g3: K3C3 -d ≤0 g4: K4C  1
Method C d (mm) W(N)
Consol-Optcad 9.56246 2.2248 0.09875
GA 10.032258 2.112903 0.089194
g5: K5C-1  1 g6:K6d2C-3+L6d  1 g9: K9d-1  1 g10: (K10)-1d  1
g7 : C-K7d-1 + 1  0 g8: K8d-1-C+10 g11: K11C3 -d2  0
www.ijert.org 11
ISSN: 2278-0181
8. Reference
1. Dr.Shapour Azarm, 1996, "Helical
Table – A:-Helical Compression Spring.
compression spring Optimization via
The minimum weight of Helical Consol-optcad", University of Maryland,
Compression Spring obtained by GA is USA.
reduced by 9.67 %, when compared to the 2. S.Vijayarangan, V.Alagappan,
weight obtained by Consol - Optcad I.Rajendran, 1999, "Design optimization
solution7. of Leaf Springs using Genetic
Algorithms", Journal of IE (I), vol 70 pp:
135-139.
3 David E.Goldberg., 1989, "Genetic
RRTT
Algorithms in search optimization a
Conclusions: Machine learning", Addison-Wesley
IIJJEE
Operation in Engineering Design is publishing company, Singapore.

the past aimed at a Design problem with 4 Kalylanmoy Deb., 1995, "Optimization
single objective function with single for Engineering Design Algorithms and
variable and with or without constraints. Examples", Prentice-Hall of India Private
But here the code is used for the Limited, New Delhi.
optimization of helical compression spring 5 Singiresu.S.Rao., 1984, "Optimization
to minimize the weight. It is found that Theory and applications", Wiley Eastern,
the results obtained by genetic algorithms New Delhi.
are better, search space is wide and it aims
at global optimum than that the local
optimum as in a traditional method for the
same input parameters.
It is found that the results obtained
from genetic algorithms are less in weight
as compared to consol opticad method.
www.ijert.org 12

Optimization of Spring Weight Using Genetic Algorithm IJERTV1IS9507 PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Optimization of Spring Weight Using Genetic Algorithm IJERTV1IS9507 PDF

Uploaded by

Copyright:

Available Formats

International Journal of Engineering Research & Technology (IJERT)

Optimization Of Spring Weight Using Genetic Algorithm

In the present study, weight

S.Vijayarangan, V.Alagappan [3] optimization problem only in a particular

the constraints are formulated to satisfy

design Procedure. The common engineering

2.2. Design variables and Constraints: objectives involve minimization of overall

Certain Optimization algorithms do not hK(X)=0, k=1,2,………,K;

correct. The chosen bound may be optimization algorithms involve repetitive

in their range by using mapping rule. Next

the randomly selected strings. It is the

springs. objective is considered at a time for a

follows: optimum design of a valve spring.

(i) To act as a reservoir 5.2 Objective Function

part to its original position of active and inactive coils

when it has been respectively.

temporarily displaced, e.g. Q is a constant and N

springs in valves, clutches depends on the spring gradient k.

and linkages. Where k is defined as:

(iv) To measure force, e.g. FU  FL Gd 2

In certain applications So it can be written a

minimum volume or maximum energy

increase the maximum shear stress in the

spring, possibly resulting in a premature

the natural frequency of the spring is taken

as 13 times the fundamental frequency of

This requirement (fn  f) is

Fig.5.1 Helical compression

There are 11 constraints in this problem.

They are 5.3.2 Buckling (g2):-

The critical length of spring is a

components of the mechanism also, such Gd 4

g2: K2C-5  1 number, IU and a minimum number, IL.

Experience indicates that extreme values

value of three is normally used as an

5.3.4 Spring Index (g4):- and normalizing gives

The spring index (defined by g6: K6d2C-3 +L6d  1

C=D/d) must be bounded by a maximum

5.3.6 Maximum Allowable The strength of the spring material

Outside Diameter (g7):- is a function of the wire diameter. This

diameter, OD the wire diameter for different materials.

g7: K7 (Cd +d)  1

Stresses are limited by the spring

Minimum allowable outside sufficient clash allowance (i.e., the

diameter, ID difference in spring length between

D-d  ID maximum load and spring solid positions)

or in normalized form: to allow for any possible combinations of

g8 : C-1+K8 C-1d-1  1 tolerance stack-up, differential thermal

expansion and wear of parts. The usual

recommendation is to provide a clash

K 8  ID allowance of approximately 10% of the

5.3.8 Upper and Lower on wire

Diameter (g9, g10):-

L2-Ls  0.1  Gf

f=500 Hz, =6.35mm

Constraint g11 in normalized form is earance constant =

g11 : K11 C3d-2  1 0.4

The objective is to minimize the G

weight of spring is given by g4: K4C  1

Subjected to eleven Where K5=IL=4

constraints, Gi (X) 0,i=1,2,3…j g6:K6d2C-3+L6d  1

1 Optimal design of helical