WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

THE OPTIMAL DESIGN AND SIMULATION OF HELICAL

SPRING BASED ON PARTICLE SWARM ALGORITHM AND
MATLAB
XIAO QIMIN [1], LIU LIWEI[2], XIAO QILI[3]
Funda mental Department, The First Aviation College of Air Force
Xinyang, Henan, 464000 P. R. China
[2]
Phys ical Department, Xinyang Normal University
Xinyang, Henan, 464000 P. R. China
[3]
School of Computer and Information, Zhejiang Wanli University
Ningbo, Zhejiang, 315140 P. R. China

[1]

Abstract: - Optimal problem is often met in engineering practice. The method to solve complex optimal
problem is always studied by people. Springs are important mechanical members which are often used in
machines to exert force, to provide flexibility, and to store or absorb energy. Helical spring is the most popular
type of springs. The method of helical spring optimization is a typical one which can be used to solving other
mechanical optimal design problem. Particle Swarm Optimization algorithm is a good method in solving
optimal problem. MATLAB is a high-performance language for technical computing and is an easy tool for us
to simulate the optimization. In this paper, we mainly introduce the optimization of helical spring based on
particle swarm algorithms and simulation in MATLAB. Directed by the theory of Particle Swarm Optimization
algorithm, with the minimum weight of helical spring as objective function, with d, D2 and n as design
variables, with shear stress, maximum axial deflection, critical frequency, bucking, fatigue strength, coils not
touch, space and dimension as constraint conditions, the complex helical spring optimal design mathematics
model with three design variables and fourteen inequality constraints conditions is established. When the model
is simulated in MATLAB the minimal optimal value of variables and the minimal weight of helical spring can
be obtained. Simulating Result shows that Particle Swarm Optimization is practical in solving complicated
optimal design problems and effectively on avoiding constraint of solution. The fundamental idea, the method
of establishing mathematic model, the simulation process in MATLAB of helical spring can be used for
reference to other similar mechanical optimal design.
Key-Words: - Particle Swarm Optimization (PSO), fitness value, local best value, global best value, helical
spring, optimal design, mathematic model , objective function, design variables, constraints condition, shear
stress, deflection, critical frequency, bucking, fatigue strength.
is a high-performance language for technical
computing. It integrates computation, visualization,
and programming in an easy-to-use environment
where problems and solutions are expressed in
familiar mathematical notation. Typical uses include
Math and computation Algorithm development Data
acquisition Modeling, simulation. Simulated the
mathematic model of mechanical element or
machines in MATLAB, the fitness optimal value of
design variables and objective function can be easily
obtained. In this paper, we will focus on the optimal
design of helical spring based on PSO and its
simulation in MATLAB. Section 2 outlines
fundamental idea of PSO searching for optima and
its realization steps. Section 3 established the
optimal design mathematical model of helical spring
directed by theory of PSO algorithm. In Section 4

1 Introduction
Computer Aided Design (CAD) is widely used in
engineering practice particularly in mechanical
design, analysis, optimize and drawing [1]. The
automation design of component [2] and machine [3]
can be realized by applying CAD. Optimal design
problem is often met in industry design. Many
optimal algorithms are promoted to solve optimal
problem. Particle Swarm Optimization (PSO) is a
population based stochastic optimization technique,
inspired by social behavior of bird flocking or fish
schooling [4]. This algorithm with characteristics of
easy realization, high precision and rapid
convergence arouses the attention of academics and
displays the superiority in solving practical optimal
problem especially a complex one. This algorithm is
widely used in mechanical optimization. MATLAB

ISSN: 1109-2734

84

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

the simulation process and simulation results and

analysis are given. Finally, conclusions are given.

generation to the next generation. k is the number of

iterations [6].
The iteration process should be gradually
narrowed in order to ensure the particles gradually
convergence closely to the optimal solution and not
oscillation back and forth in solution space.
To solve this question, inertia weight should be
obtained from equation (3) or equation (4),

2 PSO and Its Realization Steps

PSO is a population based stochastic optimization
technique developed by Dr. Eberhart and Dr.
Kennedy in 1995, inspired by social behaviour of
bird flocking or fish schooling. This algorithm with
characteristics of easy realization, high precision
and rapid convergence arouses the attention of
academics and displays the superiority in solving
practical optimal problem especially complex
optimal problem.

w(k+1)=aw(k) a (0,1)
w(k)= wmax

The foundation idea [4] of PSO is as following. In

PSO, each single solution is called particle. Each
particle has a velocity and a position that are
evaluated by the fitness function to be optimized.
Each particle searches for optima by updating its
velocity and position. The detail searching process
is as following. PSO is initialized with a group of
random particles and then searches for optima by
updating generations. In each iteration each particle
is updated by following two best values. The first
one is the fitness it has achieved so far. This value is
called p best. Another best value that is tracked by the
particle swarm optimizer is the best value obtained
so far by any particle in the population. This best
value is a global best and called gbest. When finding
the two best values, particles update its velocity and
position. In n dimensional space, if xi = (xi1 , xi2 , ,
xin ) represents the position of the i-th particle, vi =
(vi1 , vi2 , , vin ) represents the velocity of the i-th
particle, p besti = (pi1 , p i2 , , p in) represents the fitness
value of the i-th particle, g besti = (g i1 , g i2 , , gin )
represents the global best value of whole particle
swarm, then the velocity and particle position of the
particle are updated with following equation (1) and
equation (2) [5].

(1)

xid (k-1)= xid (k) +vid (k+1)

(2)

2.2 PSO Parameter Control

There are two key steps when applying PSO to
optimization problems, the representation of the
solution and the fitness function. One of the
advantages of PSO is that PSO take real numbers as
particles. For example, we try to find the solution
for f(x) = x12 + x22 + x32 , the particle can be set as (x1,
x2 , x3 ), and fitness function is f(x). Then we can use
the standard procedure to find the optimum. The
searching is a repeat process, and the stop criteria
are that the maximum iteration number is reached or
the minimum error condition is satisfied.
There are not many parameter need to be tuned
in PSO. Here is a list of the parameters and their
typical values.
The number of particles, the typical range is
about 20 to 40. Actually for most of the problems 10
particles is large enough to get good results. For
some difficult or special problems, one can try 100
or 200 particles as well.
Dimension of particles, it is determined by the
problem to be optimized.
Range of particles, it is also determined by the
problem to be optimized, we can specify different
ranges for different dimension of particles.
Learning factors, c1 and c2 usually equal to 2.
The stop condition, the maximum number of
iterations the PSO execute and the minimum error
requirement.

where i=1,2,,m( m is the number of particle in

particle swarm), d=1,2,,n( n is the dimension of
solution), c1 and c2 are learning factors, r1 and r2 are
random number between (0, 1), w is inertia weight,
used to control the effect on velocity of the older

ISSN: 1109-2734

(4)

where wmax and wmin are the maximum and

minimum of inertia weight, k max is the maximum
number of iteration.
Particles update and iterate its position and
velocity by equation (1) to (4) until get an optimum
or achieve to the specified largest number of
iteration.

2.1 The Theory of Particles Search for

Optima

vid (k+1)=wvid(k)+c1 r1 (p best (k) xid (k)) +

c2 r2(gbest (k) xid (k))

wmax wmin
k
kmax

(3)

85

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

2.3 The Implement Steps of PSO

2.4 The Feature of PSO

As stated in above, the implement steps [7] of global

particle swarm optimization are as following.
Firstly, initialize particle position xi, particle
velocity vi, the number of iterations k, the
dimension of solution n.
Secondly, calculate the current fitness value xi
and pi for each particle.
Thirdly, compare the fitness value pi in step two
with best fitness value pbesti. If pi < pbesti, then set
current value as the new best fitness value and save
the position, that is pbesti= pi and xbesti= xi.
Fourthly, compare particle with best fitness value
pbesti to all particles with best fitness value gbesti.
If pbesti < gbesti, then the particle with the best
fitness value of all particle as gbesti, and save the
current particle position, that is gbesti = pbesti,
xbest= xbesti.
Fifthly calculate particle velocity according
equation (1) and update particle position according
equation (2). When the calculation mentioned above
finished, returns to step two, and then new particle
produced. The calculation is ended while the
specified number of iteration is achieved. The flow
chart [7] of global particle swarm optimization
algorithms can be shown in Fig.1.

In the process of iteration, being only one current

optimal position needed to be returned, PSO
algorithm is simple. The structure of the PSO
algorithms is relatively simple and the calculate
speed is fast. Just being the same reason, once the
current optimal position is returned, the other
particles will move closely to it quickly. If this
optimal position is a local optimal position, PSO
will not be able to re-search in the solution space.
So PSO algorithm can easily fall into local optimum.
This emergence phenomenon is called pre-mature
phenomenon.
In order to solve the pre-mature phenomenon,
two measures are often adopted. First, randomly
initialize the particles in accordance with a certain
laws to increase the diversity of particles. Second,
search for variation to the best particle. In this twopronged approach to increase the global search
capability, to overcome the shortcomings of
convergence the local extreme value, but also to
ensure the convergence speed and accuracy of
search.

3 The Optimal Design of Helical

Spring Based on PSO
Springs are mechanical members, which are
designed to give a relatively large amount of elastic
deflection under the action of an externally applied
load. Springs are used for a variety of purposes. The
following is a list of the important purposes and
applications of springs [8].
Controlling of motion in machines. This category
represents the majority of spring applications such
as operating forces in clutches and brakes.
Reduction of transmitted forces as a result of
impact or shock loading. Applications here include
locomotive or automotive suspension system
springs and bumper springs.
Storage of energy. Applications in this category
are found in clocks and movie cameras having recoil
starters.
Measurement of forces. Scales used to weigh
people is a very common application for this
category.
In general, springs may be classified as either
wire springs, flat springs, or special-shaped springs,
and there are variations within these divisions. Wire
springs include helical springs of round or square
wire and are made to resist tensile, compressive, or
torsion loads.

Fig.1 Flow chart of particle swarm algorithm

ISSN: 1109-2734

86

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

3.1 The Aim of Helical Spring Optimization

3.3 Establish Objective Function

The optimal design of a helical spring involves the

following consideration.
Aim of helical spring optimal design. To design
the helical spring with minimum weight is usually
the aim of optimization. So we must establish the
objective function.
Design variables which determine the weight of
helical spring. The selection of these variables will
be stated following.
Conditions which constraint the realization of
helical spring optimal design. There are many
conditions which effect on the objective function,
involving space into which the spring must fit and
operation, values of working forces and deflections,
accuracy and reliability needed, tolerances and
permissible
variations
in
specifications
environmental conditions such as a corrosive
atmosphere, cost and quantities needed.
The designer uses these factors to select a
material and specify suitable values for the wire size,
the number of coil, the diameter and free height, the
type of ends, and the spring rate needed to satisfy
the working force-deflection requirements. Because
of the large number of interdependent variables, the
problem is not as simple as inserting numbers into a
procedure that will produce a complete set of results.
Because of this fact, there are a rather large number
of researchers who have attempted to simplify the
spring-design problem by the use of charts and
monographs.
The weight of a helical spring is an important
measure of quality. When spring is designed for
minimal weight, the weight can be defined as
objective function. The weight of the helical spring
can be obtained from equation (5) [9].
W= (n+2)D2 (d24)=0.252 d 2 D2 (n+2) (5)
where W is the weight of spring, n is the number of
active coils, D2 is mean spring diameter, d is wire
diameter, is the density of spring wire materials.
For steel-made spring =7.8106kg/mm3. So the
weight of spring can be finally expressed with
equation (6).
W=1.92105 d 2 D2 (n+2)
(6)

The objective function can be stated as equation (8).

F(x)=1.92105 d 2 D2 (n+2)
=1.92105 x12 x2 (x 3 +2)
(8)

3.4 Specify Constraints Conditions

3.4.1 Condition of Shear Stress
When a helical spring is loaded by the axial force F,
shear stress is exerted in the spring wire. By using
superposition, the shear stress in the inside fiber of
the spring may be computed using the equation (9).
=8FD2 /d 3 +4F/d 2
(9)
If define spring index C=D2/d as a measure of
coil curvature. With this relation, equation (9) can
be arranged to give as equation (10) shows.
= 8FD2 (1+0.5/C) /d 3
(10)
Or designating Ks =1+0.5/C, then
8 FD2
(11)
= Ks
d 3
where Ks is called a shear multiplication factor. For
most springs, C will range from about 6 to 12.
Equation (11) is quite general and applies for both
static and dynamic loads.
The stress equation can be presented as equation
(12).
8 FD
2
(12)
=K
3
d
where K is called the Wahl correction factor. This
factor includes the direction shear, together with
another effect due to curvature. Curvature of the
wire increases the stress on the inside of the spring
but decreases it only slightly on the outside. The
value of K may be obtained from equation (13).
or

The weight of spring is determined by d, D2 and n.

So set d, D2 and n as design variables. Design
variables can be stated as equation (7).

ISSN: 1109-2734

(13)

K 1.66(d/D2 )0.16 [10]

(14)

The maximum shear stress max should not larger

than allowable shear stress []. That is
max =8KF ma xD2 /d 3 []
(15)
Fma x is the maximum axial load. So the condition of
shear stress can be expressed as equation (16).
g1 (x)= []+4.23Fmax x 20.84 x12.84 0
(16)
Materials for springs should have high elastic
properties, which are stable with time. It is poor
practice to make springs of low-strength materials.
The mass of geometrically similar springs at a given
load and elastic deflection is inversely proportional
to the allowable stress. The is due to the fact that

3.2 Specify design va riables

x= (x1 , x2 , x3 )T = (d, D2 , n)T

K=(4C-1)/(4C-4)+0.615/C

(7)

87

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

Table 2 Ultimate tensile strength of carbon steel

spring wires B (Mpa)
Carbon steel spring wire
Wire diameter d
Class of wire
(mm)
B
C
D
1.00
1660~2010 1960~2360 2300~2690
1.20
1620~1960 1910~2250 2250~2550
1.40
1620~1910 1860~2210 2150~2450
1.60
1570~1860 1810~2160 2110~2400
1.80
1520~1810 1760~2110 2010~2300
2.00
1470~1760 1710~2010 1910~2200
2.20
1420~1710 1660~1960 1810~2110
2.50
1420~1710 1660~1960 1760~2060
2.80
1370~1670 1620~1910 1710~2010
3.00
1370~1670 1620~1910 1710~2010
3.20
1370~1670 1570~1860 1710~1960
3.50
1320~1620 1570~1810 1660~1910
4.00
1320~1620 1520~1760 1620~1860
4.50
1320~1570 1520~1760 1620~1860

springs of weaker materials must be made with

larger diameters to retain the specified rigidity and,
consequently, the coils are subject to higher twisting
moments than those of springs of stronger materials.
The effectiveness of using high-strength materials is
also associated with the lower stress concentration
in springs than in other machine components and the
smaller size of the cross section of the coils.
A great variety of spring materials are available
to the designer, including plain carbon steels, alloy
steels, and corrosion-resisting steels, as well as
nonferrous materials such as phosphor bronze,
spring brass, beryllium copper, and various nickel
alloys [11]. Table 1 provides spring materials and
allowable stress.
Table 1 Spring materials and allowable stress
Common
Allowable shear stress [] (Mpa)
name,

Specification
Carbon steel
wires, B,C,D
0.3B
0.4B
0.5B
grade
65Mn
60Si2Mn
480
640
800
60Si2MnA
50CrVA
450
600
750

Table 3 Ultimate tensile strength of 65Mn spring

wires B (Mpa)
Wire diameter d
B
(mm)
1~1.2
1800
1.4~1.6
1750
1.8~2
1700
2.2~2.5
1650
2.8~3.4
1600

The common materials for springs are as follows.

High-carbon steels. Being less expensive, highcarbon steels are widely used for springs from 0.8 to

14mm in diameter. Not for use above 130 C or at

subzero temperature.
Manganese steel. Manganese steel has higher
mechanical properties and better harden-ability,
which enable them to be applied for springs from 2
to 12 mm in diameter.
Chrome vanadium steel. Chrome vanadium steel
has higher mechanical properties and an especially
high endurance limit, high heat resistance and good
processing properties. Also used for shock and
impact loads. Widely used for aircraft engine valve
springs and for devices enduring temperatures to

220 C. Available in annealed size 0.8 to 12mm in

diameter.
Chrome silicon steel. This alloy is an excellent
material for highly stressed springs requiring a long
life and subjected to shock loading. Rockwell
hardness of C50 to C53 are quite common, and the

material may be used up at 250 C, be available

from 0.8 to 12mm in diameter.
Tensile strength of carbon steel spring wires is
given in table 2.
Tensile strength of 65Mn spring wires [12] is
given in table 3.

ISSN: 1109-2734

3.4.2 Condition of maximum axial deflection

When a helical spring is loaded by the axial force F,
deflection is also exerted in spring. To obtain the
equation for the deflection of a helical spring, we
shall consider an element of wire formed by two
adjacent cross sections. An element, the length of it
is dx, cut from wire of diameter d. This element is
on the surface of wire which is parallel to the spring
axis. When loaded by force F and after deformation
it will rotate through the angle and occupy the new
position. From the expression of Hookes law for
torsion, we have equation (17).
=8FD2 /d 3 G
(17)
The load F has a moment arm of D2 /2, and so the
deflection is as equation (18) shows.
=8F D23 n/ d 4G
(18)
where n is the number of active coils of spring,
F= Gd 4 8nD23 , where G is shear modulus of
elasticity[8], for steel-made spring G=8104 Mpa,
for bronze-made spring G=4104 Mpa.
To ensure the axial deflection is smaller than the
maximum deflection, the maximum working load

88

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

For
a
spring
with
fixed
ends,
fb=3.56105d/n D22 . For a spring with one fixed

which lead to maximum axial deflection should

larger than the working force, that is FmaxF. So
the condition of maximum axial deflection can be
expressed as equation (19).
g2 (x)=G x14 /8 x 23 x3 Fma x0
(19)

end, one pinned end fb=1.78105d/n D22 [13].

The fundamental critical frequency fb should be
from 15 to 20 times the frequency of the force or
motion of the spring fw to avoid resonance with the
harmonics. For fb=3.56105d/n D22 , so condition of
critical frequency can be expressed as equation (23).

3.4.3 Condition of critical frequency

If a wave is created by a disturbance at one end of a
swimming pool, this wave will travel down the
length of the pool, be reflected back at the far end,
and continue this back-and-forth motion until it is
finally damped out. The same effect happens to
helical spring, and it is called spring surge. If one
end of a compression spring is held against a flat
surface and the other end is disturbed, a
compression wave is created that travels back and
forth from one end to the other exactly like the
swimming-pool wave. So spring manufactures have
taken slow-motion movies of automotive valvespring surge.
When helical spring are used in applications
requiring a rapid reciprocating motion, the designer
must be certain that the physical dimensions of the
spring are not such as to create a natural vibratory
frequency close to the frequency of the applied force,
otherwise resonance may occur resulting in
damaging stresses, since the internal damping of
spring material is quite low [8].
The governing equation for a spring placed
between two flat and parallel plates is the classical
wave equation and is

g3 (x)=15 f w 3.56105 x1 / x 22 x3 0

3.4.4 Condition of buckling

A compression spring whose free length H0 is more
than four times its mean diameter D2 should be
checked for buckling. If the spring is properly
guided, such as inside a tube over a bar, the amount
of buckling can be greatly reduced.
Define ratio of the height H0 to diameter D2, b=
H0/ D2. The critical ratio bc can be obtained like
this. For the springs which has the squared and
ground ends are on rigid parallel surfaces
perpendicular to the springs axis, bc=5.3. For the
springs which has one end on a rigid surface, one
end hinged, bc=3.7. For the springs which has both
surface on which the ends rest are hinged on a pin,
bc =2.6 [14].
When b is greater than the above values,
buckling should be checked and satisfied with
b= H0 / D2 < b c
(24)
For helical spring, b can be obtained from
equation (25).
b= H0 / D2 =0.5n+1.5(d/D2 )
(25)
So condition of buckling can be expressed as
equation (26).
g4 (x)=1.5x1 /x2 +0.5x3 b c0
(26)

2u m 2u
(20)
=
y 2 kl 2 t 2
where k is spring constant, l is the length of spring
between plates, m is the mass of spring, y is the
coordinate along length of spring, u is motion of any
particle at distance y.
The solution to this equation is well known. The
natural frequency in radians per second turns out to
be

fb =

1 k
d
=
2 m 2nD22

G
2

3.4.5 Condition of fatigue strength

Springs are made to be used, and consequently they
are always subject to fatigue loading. In many
instances the number of cycles of required life may
be small, say, several thousand for a padlock spring
or a toggle-switch spring. But the valve spring of an
automotive engine must sustain millions of cycles of
operation without failure. So it must be designed for
infinite life.
In the case of shafts and many other machine
members, fatigue loading in the form of completely
reversed stress is quite ordinary.
Springs subject to variable stresses with a large
number of cycles (for instance, number of cycles of
the stress N103) should be checked for fatigue

(21)

Where k is spring constant, k = Gd 4 8 D23 n , is

the density of spring wire materials.
If the spring has one end against a flat plate and
the other end free, the frequency is
f b =0.25 k m
(22)
This equation also applies when one end is against a
flat plate and the other end is driven with a sinewave motion.

ISSN: 1109-2734

(23)

89

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

strength and static strength. When the springs

subject to the load that alternates from Fmin to
Fmax, the stresses for springs can be obtained from
equation (27) and equation (28) [13].
8F D
(27)
min = K min3 2
d
8F D
(28)
max = K max3 2
d
The fatigue strength condition is
S= ( 0 +0.75 min ) / max[S]
(29)
where 0 is endurance limit of the spring for a zeroplus cycle, which can be found in table 4. [S] is
allowable factor of safety, if accuracy of design
calculation and mechanical properties is higher, then
[S]=1.3~1.7, if lower, [S]=1.8~2 [14].

3.4.7 Condition of index

Springs are manufactured either by hot-working or
cold-working processes, depending upon the size of
the material, the spring index C=D2 /d where D2 is
mean spring diameter, d is wire diameter, and the
properties desired.
Small-size springs of wire under about 8 to 10
mm in diameter are manufactured by cold coiling,
larger size, by hot coiling.
Most cold-coiled springs are made of wire that
has been heat-treated before coiling so that the
springs are only tempered after being coiled. All
hot-coiled springs and the most critical cold-coiled
ones, in particular, those of most alloy steels, are
hardened in the coiled condition. Wire for critical
springs, made of high carbon steel with 1% carbon,
is subjected to lead patenting, i.e. immersed after
being heated to a high temperature in a molten lead
bath. This increases grain size and the wire,
following final drawing, is work-hardened to a high
degree and strengthened. This also improves the
elastic properties of the material.
Helical springs are produced in quantity on
special automatic spring winding machines. In other
cases the spring is wound around mandrel on a lathe.
For proper the spring index is of extreme
importance. The smaller the index the more difficult
it is to wind the spring. Table 5 provides the values
of the common spring index C [16].
Compression springs are wound so as to allow a
certain clearance between the coils. Extension
springs have closely wound coils. To obtain closed
coils the wire is stretched during the process of
winding to subject it to tensile elastic strain. When
the wound spring is removed from the mandrel there
occurs an elastic spring-back of the material, the
spring expands in diameter and the coils are so
tightly pressed against each other that an initial
tension is given to the spring increasing its carrying
capacity.

Table 4 Endurance limit of the spring

for a zero-plus cycle
Number
of
cycles of the 104
105
106
107
stress N
0
0.45B 0.35B 0.33B 0.3B
When the spring works under the condition with
number of cycles of the stress N=106 , then 0 =0.33b .
So the condition of fatigue strength can be
expressed as equation (30).
0.33 B x12.84
F
g 5 (x ) = [S ]
0.75 min 0 (30)
0.84
Fmax
4.23Fmax x2

3.4.6 Condition of coils not touch

When there is no load applied on a compression
spring, the spring is free, and the height of it is H0 .
As the load is applied, the coils move closer
together, but do not touch. The maximum deflection
max exerts when the maximum load applied. In this
case, to satisfy the requirement of coils not touch,
following equation should be satisfied.
H0 max Hb
(31)
where H0 free height, for closed and ground ends,
H0 =nt +(1.5~2)d, for closed ends, not ground, H0 =nt
+(3~3.5)d. t is pitch, t(0.28~0.5)D2 . When
calculating, we usually set t=0.4 D2 . max is the
maximum deflection under the action of the load
Fma x. Hb is solid height(the minimum operating
height)when the number of inactive turn n 2 =2 and
the springs have plain endsHb (n+1.5)d [15]. So
the condition of coils not touch can be expressed by
equation (32).
(32)
g6 (x)=8Fma x x23 x3 /G x14 +x1 x3 0.4x2x3

ISSN: 1109-2734

Table 5 Common spring index C

d (mm)
C
0.2~0.4
7~14
0.5~1
5~12
1.5~2.2
5~10
2.5~6
4~9
7~16
4~8
18~50
4~6
Springs for static or limit short-term action, as
well as springs subject to variable stresses with a
cycle factor close to unity, are subjected to plastic
deformation, which is called pressing. This is done

90

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

When we design a spring, the maximum number

of coil nmax and minimum number of coil nmin
must be specify, that is nminnnmax. So the
condition of active coil can be expressed as equation
(35) and (36).
g9 (x)= n min x3 0
(35)
g10 (x)= x3 n max 0
(36)

by holding them for 6~48 hours under the load of

the same sign as the design working load, but one
producing stress above the elastic limit. As a result,
the external fibers have a permanent set. In the free
state of the spring, the external fibers, interacting
with the inner fibers, have residual stresses of a sign
opposite to the working stresses. When such a
spring is loaded, these stresses are subtracted from
the working stresses, thereby increasing the load
capacity of the spring by 20% to 25%. When we
design a spring, the maximum spring index Cmax
and minimum spring index Cmin must be specify,
that is CminCCmax. So the condition of index
can be express as equation (33) and equation (34).
g7 (x)= Cmin x2 / x1 0
(33)
g8 (x)= x2 / x1 Cmax0
(34)

3.4.9 Condition of space and dimension

To satisfy the requirement of fit, operation and
dimension specification, the wire diameter d, the
outside spring diameter D must be severely
limited[18], that is d min dd max , D2min D2 D2max. So
the conditions of space and dimension can be
expressed as the following equations.
g11 (x)= d min x1 0
(37)
g12 (x)= x1 d max0
(38)
g13 (x)= D2min x2 0
(39)
g14 (x)= x2 D 2max 0
(40)

3.4.8 Condition of active coil

The helical spring is the most popular type of spring.
There are three types of ends commonly used for
compression springs. In each case, when there is on
load, the coils are separated. As the load is applied,
the coils move closer together, but do not touch.
Helical compression springs are used to exert force
on mating parts.
It is desirable that compression springs have as
such possible with that mating parts at the ends of
the springs. The three types of ends used for
compression springs are described as follows.
Squared and ground ends, Y type. This type of
end enjoys the advantage of not becoming readily
tangled during manufacturing. In the case, the load
is transmitted in a perfectly axial direction.
Squared or closed ends, Y type. This type does
not tend to become tangled during the
manufacturing process as readily as the plain ends
ground.
Plain ends ground, Y type. This type is better
than the plain ends. However, during storage,
handling and assembly of spring, the ends tend to
become readily tangled.
Regardless of the type of ends used, a partially
dead or inactive coil nd exists at each end of the
spring. This interactive coil must be subtracted from
the actual total number nt of coil to find the number
of active coils. The following is an approximate rule
for finding the active number of coils [17].
Squared and ground ends, subtract a total of 2
turns, that is nd =2, n= nt2.
Squared or closed ends, subtract a total of 1.5
turns, that is nd =1.5, n= nt1.5.
Plain ends ground, subtract a total of 1 turns, that
is nd =1, n= nt1.

ISSN: 1109-2734

3.5 Mathematic model of helical spring

optimal design
According to objective function and constraints
conditions the optimal design mathematical model
[19] of helical spring can be states as equation (41).
min F ( x ) = 1.92 10 5 x12 x2 ( x3 + 2 )

(41)
s.t.g u ( x ) 0 u = 1,2,3,,14

T
T
x = (d , D2 , n ) = ( x1 , x2 , x3 )
The helical spring optimal design model is a
complex one which has three design variables and
fourteen inequality constraints conditions.

4 Simulation of Helical
Optimization in MATLAB

Spring

MATLAB is a high-performance language for

technical computing. It integrates computation,
visualization, and programming in an easy-to-use
environment where problems and solutions are
expressed in familiar mathematical notation. Typical
uses include Math and computation Algorithm
development Data acquisition Modeling, simulation,
and prototyping Data analysis, exploration, and
visualization Scientific and engineering graphics
Application development, including graphical user
interface building MATLAB is an interactive
system whose basic data element is an array that
does not require dimensioning. This allows us to
solve many technical computing problems,
especially those with matrix and vector formulations,
in a fraction of the time it would take to write a

91

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

program in a scalar non-interactive language such as

C or FORTRAN. So it is easy for us to use
MATLAB to simulate the optimization of helical
spring. In this paper, we simulate the optimization
of helical spring with language C.
When simulating helical spring optimal design
there are almost as many ways to create a springdesign program as there are programmers, and there
is nothing unusual about the program presented here.
It works. The program, which is for the design of
compression springs, can be used as a starting point
for creation of other programs. It contains of seven
separate subroutines, all of which utilize the same
memory location. The subroutines, which should be
used in the order in which they are present, are as
following. Enter and display the outside diameter.
Enter and display the total number of coils. Enter
and display the number of dead coils. Compute and
display the number of active coils. Select a material
and enter and display the exponent and coefficient.
Enter and display the wire diameter. Compute and
display the torsion yield strength. Enter and display
the maximum torsion stress desired when the spring
is closed solid. Compute and display the solid height,
the free height, and the force required to compress
the spring solid. Compute and display the spring
constant. Enter and display any desired operating
force F. Compute and display the corresponding
values of the torsion stress and the spring deflection.
Separate subroutines are used in this program in
order to avoid re-entering all the data when only a
single parameter is to be changed. In this way it is
easy to see the effect of the single change. For
example, having run through the program once it
may be desirable to try a different wire size. This
can be done by entering the new wire size in
subroutine 4 and proceeding from that point.

From the example above, values of parameters are

known. F max =40N, Fmin =0, f w =25Hz, d min =1mm,
d max =4mm, n min =4.5, n ma x=50, D2min =10mm,
D2max =30mm, C min =4, C ma x=12, min =10mm.
As stated above, by consulting table 1 and table 2,
we can get []=350Mpa.To this spring, we can
specify G=8 104Mpa, [S]=1.9, bc=3.7. So the
constraints conditions can be obtained.
g1 (x)= 350+169.2 x20.84 x12.84 0
g2 (x)=103 x14 /8 x23 x3 400
g3 (x)=375 3.56105 x1 / x 22 x3 0
g4 (x)=1.5x1 /x2 +0.5x3 3.70
g5 (x)=1.9 2.07 x12.84 x20.84 0
g6 (x)=4103 x23 x3 + x1 x3 0.4 x2 x3 0
g7 (x)=4 x3 / x1 0
g8 (x)= x3 / x1 120
g9 (x)=4.5 x3 0
g10 (x)= x3 50 0
g11 (x)=1x1 0
g12 (x)= x1 40
g13 (x)= 10 x2 0
g14 (x)= x2 30 0

4.3 Write M file

Enter MATLAB, then write the M file [7] of
objective function and constraints conditions and
named as Spring.m.
function [f,g]=Spring_Opti(x)
f=1.92*(10e5)* x(1) ^2*x(2) *(x(3)+2) ;
g(1)= 350+169.2*x(2)^0.84/x(1)^2.84;
g(2)= 1*(10e+3)*x(1)^4/( x (2)^3* x(3)) 40;
g(3)= 3753.56*10e5*x(1)/( x(2)^2 * x(3)) ;
g(4)=1.5* x(1)/ x(2) +0.5* x(3) 3.7;
g(5)=1.9 2.07* x(1) ^2.84/ x(2) ^0.84;
g(6)=4*(10e3)* x(2) ^3* x(3) / x(1) ^4 + x(1) *
x(3)0.4* x(2) * x(3) ;
g(7)= 4x(2)/x(1) ;
g(8)= x(2)/x(1) 12;
g(9)= 4.5 x(3) ;
g(10)= x(3) 50;
g(11)= 1x(1) ;
g(12)= x(1) 4;
g(13)=10 x(3) ;
g(14)= x(3) 30;
x0=[ x1 x2 x3 ] ;
x=constr(Spring_Opti,x0)

4.1 Exa mple

Example, A compression spring with one end fixed
and another end free, the spring material is 65Mn,
maximum load F ma x=40N, the minimum load F min =0,
the frequency of the force f w =25Hz, the life of
spring is 104h, the wire diameter d ranges about 1 to
4mm, the mean spring diameter D2 ranges about 10
to 30mm, the number of active coil n ranges about
4.5 to 50, the spring index C is not smaller than 4,
the working temperature is 50C, the deflection is
not smaller than 10mm. Design this helical spring
for minimal weight.

4.4 Results and analysis

4.2 Specify the parameters data

ISSN: 1109-2734

92

Issue 1, Volume 8, January 2009

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS

Xiao Qimin, Liu Liwei, Xiao Qili

Programming VII, Proc. EP 98 pp, 591-600.

Springer-Verlag, New York, 1998.
[7] Yuan sicong, MATLAB Language and Its
Application in Mechanical Engineering, China
Machine Press, Beijing, 2008.
[8] Joseph
Edward
Shigley,
Mechanical
Engineering Design, First Metric Edit,
McGraw-Hill, 1986.
[9] Liang shangming, Yin guofu. Modern
Mechanical Design Optimization, Chemical
Industry Press, Beijing, 2005.
[10] Smith E.H., Mechanical Engineers Reference
Book, 12th edit., Butterworth-Heinemann,
Oxford, 1998.
[11] Andrew D.Dimarogoneas, Machine Design,
John Wiley & Sons,Inc,2001.
[12] Liu weixin, Modern Mechanical Optimal
Design, Tsinghua university Press, Beijing,
2001.
[13] Kenneth S. Edwards Jr, Robert B. McKee,
Foundamentals of Mechanical Component
Design, McGraw-hill, 1991.
[14] G.M.Maitra,
L.V.Prasad,
Machine
Design,TATA McGraw-Hill,1985.
[15] Robert L.Mott, Machine Elements in
Mechanical Design, 3rd edit. , Prentice-Hall,
Inc, Upper Saddle River, New Jersey, 1999.
[16] M. F. Spotts and T.E.Shoup. Design of
Machine Elements. Prentice-Hall, 1992.
[17] G.M.Maitra, L.V. Prasad. Mechanical Design.
TATA McGraw-Hill, 1985.
[18] Robert H.Creamer, Machine Design, AddisonWesley Publishing Company, 1984.
[19] Mortenson M.E., Geometric Modeling, John
Wiley & Sons, Inc, 1985.

Write program by using MATLAB language for

computing simulation of PSO, the result of
simulation is obtained. They are shown as following.
x =( x1 , x2 , x3 )T=( 1.645, 16.113, 4.501)T
F(x)=5.5g
That is to say, when taken d=1.645, D2=16.113,
n=4.501 respectively, the weight of helical spring is
minimum and equals 5.5g. According to practice,
these parameters can be round to the whole value or
in accordance with the standard series.
As stated above, good optimization results can be
obtained by using PSO.

5 Conclusion
Directed by theory of PSO algorithm the optimal
design mathematical model of helical spring is
established. This mathematical model with three
design variables and fourteen inequality constraints
conditions is a complex optimal design problem.
When simulated in MATLAB, minimal weight of
the helical spring can be obtained, and optimal
values of variables can be obtained too. The result
of simulation shows that by using PSO algorithm
the weight of the helical spring can be greatly
reduced, the design quality and efficiency can be
improved greatly, PSO algorithms can redound the
particles capability to dap out of the Maximum trap
effectively. PSO algorithm is a new method in
solving the complex optimal design problem. The
fundamental idea, simulation process and dataprocessing method of helical spring can be used for
reference to other similar mechanical optimal design.
References:
[1] Ibrahim Zeid, Theory and Practice, McGrawHill, Inc., 1991.
[2] Mikell P.Groover, Emory W.Zimmers Jr.,
Computer-Aided Design and Manufacturing,
Prentice-Hall Inc., 1984.
[3] C.B.Besant, Computer-Aided Design and
Manufacture, Ellis Horwood Limitad, 1998.
[4] Eberhart, R. C. and Shi, Y, Particle swarm
optimization: developments, applications and
resources, Proc. congress on evolutionary
computation 2001 IEEE service center,
Piscataway, NJ., Seoul, Korea., 2001.
[5] Kennedy, J. and Eberhart R. C., Particle
Swarm Optimization, Proceedings of the 1995
IEEE International Conference on Neural
Networks, Piscataway, 1995.
[6] Shi Y., and Eberhart R. C., Parameter selection
in particle swarm optimization, Evolutionary

ISSN: 1109-2734

93

Issue 1, Volume 8, January 2009