See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/257267405

Reliability optimization with high and low level

redundancies in interval environment via
Genetic Algorithm
ARTICLE in INTERNATIONAL JOURNAL OF SYSTEMS ASSURANCE ENGINEERING AND MANAGEMENT OCTOBER
2013
DOI: 10.1007/s13198-013-0199-9

CITATION

DOWNLOADS

VIEWS

106

234

3 AUTHORS, INCLUDING:
Laxminarayan Sahoo

Asokekumar Bhunia

Burdwan Raj College

University of Burdwan

22 PUBLICATIONS 85 CITATIONS

114 PUBLICATIONS 827 CITATIONS

SEE PROFILE

SEE PROFILE

Available from: Laxminarayan Sahoo

Retrieved on: 09 September 2015

Int J Syst Assur Eng Manag

DOI 10.1007/s13198-013-0199-9

ORIGINAL ARTICLE

Reliability optimization with high and low level redundancies

in interval environment via genetic algorithm
Laxminarayan Sahoo Asoke Kumar Bhunia
Dilip Roy

Received: 15 May 2013 / Revised: 23 September 2013

 The Society for Reliability Engineering, Quality and Operations Management (SREQOM), India and The Division of Operation and
Maintenance, Lulea University of Technology, Sweden 2013

Abstract This paper deals with redundancy allocation

problem in interval environment that maximizes the overall
system reliability subject to the given resource constraints
and also minimizes the overall system cost subject to the
given resources including an additional constraint on system reliability. Here, the reliability of each component is
assumed as interval valued and the cost coefficients as well
as the amount of resources are imprecise and interval
valued. These types of problems have been formulated as
an interval valued nonlinear integer programming problem.
In this paper, we have formulated two types of redundancy,
viz. component level redundancy known as low-level
redundancy and the system level redundancy known as
high-level redundancy. These problems have been transformed as an unconstrained optimization problem using
penalty function technique and solved using genetic algorithm. Finally, two numerical examples (one for low-level
redundancy and another for high-level redundancy) have
been solved and the computational results have been
compared.

L. Sahoo (&)
Department of Mathematics, Raniganj Girls College, Raniganj
713358, India
e-mail: lxsahoo@gmail.com
A. K. Bhunia
Department of Mathematics, The University of Burdwan,
Burdwan 713104, India
e-mail: bhuniaak@rediffmail.com
D. Roy
Centre for Management Studies, The University of Burdwan,
Burdwan 713104, India
e-mail: dr.diliproy@gmail.com

Keywords Genetic algorithm  Interval

environment  Low-level redundancy  High-level
redundancy  Reliability optimization

1 Introduction
Development of modern technological system design
depends on the selection of components and configurations
to meet the functional requirements as well as performance
specifications. For a system with known cost, reliability,
weight, volume and other system parameters, the corresponding design problem becomes a combinatorial optimization problem. The best known reliability design
problem of this type is referred as the redundancy allocation problem. The basic objective of redundancy allocation
problem is to find the number of redundant components
that either maximize the system reliability or minimize the
system cost under several resource constraints. Redundancy allocation problem is basically a nonlinear integer
programming problem. Most of these problems can not be
solved by direct/indirect or mixed search methods due to
discrete search space. According to Chern (1992), redundancy allocation problem with multiple constraints is quite
often hard to find feasible solutions. This redundancy
allocation problem is NP-hard and it has been well discussed in Tillman et al. (1977) and Kuo and Prasad (2000).
Earlier, several deterministic methods like heuristic methods (Nakagawa and Nakashima 1977; Kim and Yum 1993;
Kuo et al. 1978; Aggarwal and Gupta 2005; Ha and Kuo
2006), reduced gradient method (Hwang et al. 1979),
branch and bound method (Kuo et al. 1987; Tillman et al.
1977; Sun and Li 2002; Sung and Cho 1999), integer
programming (Misra and Sharma 1991), dynamic

123

Int J Syst Assur Eng Manag

programming (Nakagawa and Miyazaki 1981; Hikita et al.

1992) and other well-developed mathematical programming techniques were used to solve such redundancy
allocation problem. However, these methods have both
advantages and disadvantages. Dynamic programming is
not useful for reliability optimization of a general system as
it can be used only for few particular structures of the
objective function and constraints that are decomposable.
In branch and bound method, the effectiveness depends on
sharpness of the bound and required memory increases
exponentially with the problem size. As a result, with the
development of genetic algorithm (Goldberg 1989) and
other evolutionary algorithms, most of the researchers have
given more attention on redundancy allocation problem as
these methods provide more flexibility and require fewer
assumptions on the objective as well as the constraints.
Also these methods are also effective irrespective of
whether the search space is discrete or not.
In almost all the studies referred above, the design
parameters of redundancy allocation problem have usually
been taken to be precise values. This means that complete
probabilistic information about the system is known. In this
case, it is usually assumed that for every event probability
involved is perfectly determinable. However, in real life
situations, there are not sufficient statistical data available
in most of the cases. In this case the system is either new
or it exists only as a project in which the data/information
can not be collected precisely due to human errors,
improper storage facilities and other unexpected factors
relating to environment. Therefore, one has to consider
situations where parameters are imprecise. To tackle the
problem with such imprecise numbers, generally stochastic, fuzzy and fuzzy-stochastic approaches are applied and
the corresponding problems are converted to deterministic
problems for solving them. In stochastic approach, the
parameters are assumed as random variables with known
probability distributions. In fuzzy approach, the parameters, constraints and goals are considered as fuzzy sets with
known membership functions or fuzzy numbers. On the
other hand, in fuzzy-stochastic approach, some parameters
are viewed as fuzzy sets and other as random variables.
However, for a decision maker to specify the appropriate
membership function for fuzzy approach and probability
distribution for stochastic approach and both for fuzzystochastic approach is a formidable task.
To overcome these difficulties for handling the imprecise numbers by different approaches, one may represent
the same by interval valued number as this representation is
the best representation among others. Due to this new
representation, the objective function as well as constraint
functions of the reduced redundancy allocation problem is
interval valued, which is to be maximized under given
constraints. As all the parameters viz., reliability, cost,

123

weight and amount of resources are interval valued so we

call this problem as interval valued programming problem
(IVPP).
In this paper, we have formulated two types of redundancy, viz. component level redundancy known as lowlevel redundancy and the system level redundancy known
as high-level redundancy, for a five-link bridge system
where the objective function as well as constraint functions
are in interval environment. Studies of the system reliability, with component reliability as interval valued, have
already been initiated by some researchers (Gupta et al.
2009; Bhunia et al. 2010; Sahoo et al. 2010, 2012b; Bhunia
and Sahoo 2011; Mahato et al. (2012)). On the other hand,
considering non-interval valued (precise/fuzzy/stochastic)
component reliability, a number of researchers has presented different situations and solutions methodologies on
redundancy allocation problem. Over the last few years,
several techniques were proposed for solving constrained
optimization problem with precise coefficients with the
help of genetic algorithm. Among these, penalty function
techniques are very popular in solving the same by genetic
algorithms (Miettinen et al. 2003). In this work, we have
used GA based approaches for solving interval valued
nonlinear integer programming problem (IVNLP) type
redundancy allocation problem. To find the optimal solution of this type of problem by GA, order relations of
interval numbers take an important role for GA operators.
In these works, we have used the definitions of Sahoo et al.
(2012a) for order relations between interval numbers. For
solving such types of problems, we have developed a real
coded elitist GA with tournament selection, uniform
crossover and 1-neighborhood mutation (Bhunia et al.
2010). Finally, to illustrate the proposed method, for highlevel as well as low-level redundancies, two numerical
examples have been considered and solved.

2 Assumptions and notations

The following assumptions and notations have been used in
the entire paper.
2.1 Assumptions
1.
2.
3.
4.
5.

Reliability of each component is imprecise and interval valued.

Failures of components are statistically independent.
All redundancy is active and there is no provision for
repair.
The components as well as the system have two
different states, viz. operating state and failure state.
The cost coefficients as well as the amount of
resources are imprecise and interval valued.

Int J Syst Assur Eng Manag

Fig. 1 Low-level redundancy

x1

x2

xn

2.2 Notations
n
xj

x
h

r~i rjL ; rjR 

q~i qiL ; qiR 
R~j xj RjL xj ; RjR xj 

~ j QjL ; QjR 
Q
~
RS x RSL x; RSR x
C~S CSL ; CSR 
c~j cjL ; cjR 

~j wjL ; wjR 
w

R~T RTL ; RTR 

g~i x giL x; giR x
b~i biL ; biR 
lj, uj
pc
pm

The number of subsystems

The number of components in
j-th subsystem, arranged in
parallel in case of low-level
redundancy
(x1, x2,,xn)
The number of redundant
subsystems,
arranged
in
parallel in case of high-level
redundancy
Interval valued reliability of jth component
1 - [riL, riR]
1  1  rjL ; rjR xj , the reliability
of
j-th
parallel
subsystem
1 - [RjL, RjR]
Interval
valued
system
reliability
Interval valued system cost
Interval
valued
cost
coefficients for the j-th
component
Interval
valued
weight
coefficients for the j-th
component
Interval valued target system
reliability
i-th
constraint
function,
i = 1, 2, , m
Availability of i-th resource
(i 1; 2; . . .; m
Lower and upper bounds of xj
Probability
of
crossover/
crossover rate
Probability
of
mutation/
mutation rate

Fig. 2 High-level redundancy

max_gen
b yc

Maximum
number
generation
Integral value of y

of

3 Low-level and high-level redundancy

Let us consider a n component system. Now, we can either
provide redundant components, which give a system design
diagram as shown in Fig. 1, or provide a total redundant
system as shown in Fig. 2. The component level redundancy is known as low-level redundancy whereas the
system level redundancy known as high-level redundancy.

4 Formulation of reliability-redundancy optimization

problems
Let us consider a network with n subsystems. The goal of
the redundancy allocation problem is to determine the
number of redundant components in each of n parallel
subsystems so as to maximize the overall system reliability
subject to the given resource constraints and also to minimize the overall system cost subject to the given constraint
on system reliability.

123

Int J Syst Assur Eng Manag

The general form of the redundancy allocation problem

is as follows:
Maximize R~S x
subject to g~i x  b~i ; i 1; 2;   ; m

1  lj  xj  uj ; xj integer, j 1; 2; . . .; n

This formulation is designed to achieve a minimum total

system cost, subject to R~T , a target limit on the system
reliability.
For low-level redundant system (see. Fig. 1), the corresponding optimization problems are as follows:
Maximize R~S x f R~1 x1 ; R~2 x2 ; . . .; R~q xq ; . . .; R~n xn
subject to g~i x  b~i ; i 1; 2; . . .; m
1  lj  xj  uj ; xj integer, j 1; 2; . . .; n

where R~S h 1  1  f ~
r1 ; r~2 ; r~3 ; . . .; r~q ; . . .; r~n

5.1 Interval
An interval number A~ is a closed interval denoted by A~
aL ; aR  and is defined by A~ aL ; aR  fx : aL 
x  aR ; x 2 <g, where aL,aR are the left and right limits
respectively and < is the set of all real numbers. Actually,
every real number can be treated as an interval, such as for
all x 2 <, x can be written as an interval [x, x] which has
zero width. The basic arithmetical operations like, addition,
subtraction, multiplication, division and integral power of
interval numbers are available in the book of interval
analysis (Hansen and Walster 2004). Here we shall discuss
two important operations of interval numbers as follows:
5.2 n-th root of an interval
According to Karmakar et al. (2009), the nth root of an
interval A~ aL ; aR  is defined as
1

Minimize C~S x
subject to R~S x f R~1 x1 ; R~2 x2 ; . . .; R~q xq ; . . .; R~n xn  R~T

4
where R~j xj 1  1  r~j xj ; j 1; 2;   ; n
For high-level redundant system (see Fig. 2), the corresponding optimization problems are as follows:
Maximize R~S h 1  1  f ~
r1 ; r~2 ; r~3 ; . . .; r~q ; . . .; r~n h
subject to g~i h  b~i ; i 1; 2; . . .; m
5

~ n aL ; aR n
A

8 p p
n
n
>
< apL ; aR 
p

n
n
aL ; aR  0; aR 
>
:
u

where R~j xj 1  1  r~j xj ; j 1; 2; . . .; n and r~j 2 0; 1

l  h  u; h integer and r~i 2 0; 1, i 1; 2; . . .n

if aL  0 or if n is odd
if aL  0; aR  0 and n is even
if aR \0 and n is even

where / is the empty interval.

Again applying the definitions of power and different
roots of an interval, the rational power of an interval A~
aL ; aR  is defined as follows:
p

~ q A~p q
A
5.3 Interval power of an interval
Sahoo et al. (2012a, b) defined the interval power of an
interval as follows: Let A~ aL ; aR  and B~ bL ; bR be two
interval numbers, then we define

~ B aL ; aR bL ;bR 
A
(

eminbL log aL ;bL log aR ;bR log aL ;bR log aR ; emaxbL log aL ;bL log aR ;bR log aL ;bR log aR if aL  0

a complex interval
if aL \0

123

6
h

5 Prerequisites mathematics

The goal of the formulation (1) is to determine the

number of redundant components so as to maximize the
overall system reliability. This problem belongs to the
category of constrained nonlinear integer programming
problems (NIPP).
The general form of the cost minimization problem is as
follows:
Minimize C~S x
subject to R~S x  R~T

Minimize C~S h
subject to R~S h  R~T

Int J Syst Assur Eng Manag

5.4 Order relation of interval numbers

According to the assumption (1), the objective function of
redundancy allocation problem would be interval valued.
So, to find the optimal solution of the said problem, the
order relations of interval numbers take important role in
decision-making.
Let A~ aL ; aR  and B~ bL ; bR  be two closed intervals. Then these two intervals may be of the following
three types:
Type I:
Type II:
Type III:

Two intervals are disjoint

Two intervals are partially overlapping
One of the intervals contained in the other one

It is to be noted that both the intervals A = [aL, aR] and

B = [bL, bR] will be equal in case of fully overlapping
intervals. That is A = B if aL = bL and aR = bR.
Here we shall consider the definitions of order relations
developed by Sahoo et al. (2012a).
Definition 5.4.1: The order relation [ max between
the intervals A~ aL ; aR  hac ; aw i and B~ bL ; bR 
hbc ; bw i, then for maximization problems
1.
2.

A~ [ max B~ , ac [ bc for Type I and Type II intervals,

either
ac  bc ^ aw \bw
or
A~ [ max B~ ,
ac  bc ^ aR [ bR for Type III intervals

According to this definition, the interval A~ is accepted

for maximization case. Clearly the order relation A~ [ max B~
is reflexive and transitive but not symmetric.
Definition 5.4.2: The order relation \min between the
intervals A~ aL ; aR  hac ; aw i and B~ bL ; bR 
hbc ; bw i, then for minimization problems
1.
2.

~ min B~ , ac \bc for Type I and Type II intervals,

A\
~ min B~ ,
either
ac  bc ^ aw \bw
or
A\
ac  bc ^ aL \bL for Type III intervals,

According to this definition, the interval A is accepted

~ min B~ is
for minimization case. Clearly the order relation A\
reflexive and transitive but not symmetric.

"
#
n
n
1X
1X

xiL ;
xiR ;
x~ 
xL ; xR 
n i1
n i1
Var(~
x r2L ; r2R 
!2
n
n
n
1X
1X
1X

xiL 
xiR ; xiR 
xiL 
n i1
n i1
n i1
 n 
n
p
P
P
and rx~ rL ; rR  Var ~
x 1n
xiL  1n xiR ;
i1
i1
1=2
n
P
2
1
xiR  n xiL 
i1

6 Constraint handling technique

In the application of genetic algorithm for solving the said
reliability optimization problems with interval objectives,
there arises a major huddle question for handling the
constraints. Over the last few years, several techniques
have been proposed to handle the constraints in genetic
algorithms for solving problems with non-interval/fixed
objectives. In our work, we have used penalty function
technique to solve the constrained optimization problem
with interval objective. In this method, the constrained
optimization problem is converted into an unconstrained
optimization problem in which the reduced objective
function involves objective function and a penalty for
violating the constraints. Here, we have used Big-M penalty technique (Gupta et al. 2009).
For the constrained optimization problem
Maximize R~S x
subject to g~i x  b~i ; i 1; 2;    ; m
1  lj  xj  uj ; xj integer, j 1; 2; . . .; n
the general form of reduced problem by Big-M penalty
technique is as follows:

RSL x; RSR x if x 2 S
Maximize R^SL x; R^SR x
M; M;
ifx 62 S
7
and S fx : g~i x  b~i ; i 1; 2;   ; m and 1  lj  xj  uj ;
xj integer, j 1; 2; . . .; ng
For the constrained optimization problem

5.5 Mean, variance and standard deviation of interval

numbers

Minimize C~S
subject to R~S x  R~T

According to Bhunia et al. (2010), mean, variance and

standard deviation of n interval numbers are defined as
follows:
Let x~i xiL ; xiR , i 1; 2; . . .; n, be the ith observation
which is an interval number. Then mean, variance and
standard deviation of x~1 ; x~2 ;   ; x~n are given by

the general form of reduced problem by Big-M penalty

technique is as follows:

CSL x; CSR x if x 2 S
^
^
Maximize CsL x; CsR x
M; M
if x 62 S
8

123

Int J Syst Assur Eng Manag


and S x : R~S x  R~T ; and 1  lj  xj  uj ; xj integer,
j 1; 2; . . .; n g
For the constrained optimization problem
Maximize R~S h
subject to g~i h  b~i ; i 1; 2; . . .; m
l  h  u; h integer
the general form is as follows:

RSL h; RSR h
^
^
Maximize RSL h; RSR h
M; M

if h 2 S
if h 62 S
9

and S fh : g~i h  b~i ; i 1; 2; . . .; m and 1  l  h  u; h

integerg
For the constrained optimization problem
Minimize C~S h
subject to R~S h  R~T
the general form is as follows:

C~SL h; C~SR h
^
^
Maximize CsL h; CsR h
M; M


if h 2 S
if h 62 S
10

and S h : R~S h  R~T ; and 1  l  h  u; h integer

Here R^SL x; R^SR x, C^sL x; C^sR x, R^SL h; R^SR h
and C^sL h; C^sR h are the interval valued auxiliary
objective function. Problem (7) and (8) are integer nonlinear unconstrained optimization problem with interval
objective of n integer variables x1 ; x2 ; . . .; xn whereas
problem (9) and (10) are integer non-linear unconstrained
optimization problem with interval objective of integer
variable h. For solving these problems, we have developed
a real coded genetic algorithm (GA) with advanced operators for integer variable(s).

7 Genetic algorithm
Genetic algorithm is a well-known stochastic method of
global optimization based on the evolutionary theory of
Darwin: The survival of the fittest and natural genetics
(Goldberg 1989). It has successfully been applied in solving optimization problems of different real world application problems. This algorithm is based on the evaluation of
a set of solutions, called population. Basically, the population is initialized by randomly generated individuals.
These populations will be improved from generation to
generation by an artificial evolution process. During each
generation, each chromosome in the entire population is
evaluated using the measure of fitness and the population
of the next generation is created through different genetic

123

operators. This algorithm can be implemented easily with

the help of computer programming. In particular, it is very
useful for solving complicated optimization problems
which cannot be solved easily by analytical/direct/gradient
based mathematical techniques.
For implementing the GA in solving the optimization
problems, the following basic components are to be
considered.

GA Parameters.
Chromosome representation.
Initialization of population.
Evaluation of fitness function.
Selection process.
Genetic operators (crossover, mutation and elitism).
Termination criteria.

There are basically four parameters used in the genetic

algorithm, viz. p_size (population size i.e., the number of
individuals in the population), m_gen (maximum number
of generations/iterations), pc (probability of crossover) and
pm (probability of mutation/mutation rate). There is no hard
and fast rule for the choice of the values of first two
parameters i.e., p_size and m_gen. These vary from problem to problem according to their dimension. Again, from
the natural genetics, it is obvious that the crossover rate
should be greater than the mutation rate. Generally, the
crossover rate varies from 0.8 to 0.95 whereas the mutation
rate varies from 0.05 to 0.30.
For successful applications of GA, the appropriate
chromosome/individual representation of solutions for the
given problem is an important task. There are different
types of representations available in the existing literature,
viz. binary, octal, hexadecimal and real. Among these
representations, real coding representation is very popular.
In this representation, for a given problem with n decision
variables, a n-component vector x x1 ; x2 ; . . .; xn is used
as a chromosome to represent a solution to the problem. A
chromosome, denoted by vk k 1; 2; . . .; p size is an
ordered list of n genes say vk xk1 ; xk2 ; . . .; xkn . An initial
population of size p_size is randomly generated within the
bounds of the corresponding decision variables according
to the uniform distribution.
In GA, fitness function plays an important role. It represents the fitness value of a chromosome. In this work, the
transformed unconstrained objective function due to penalty technique is considered as the fitness function.
In artificial genetics, the selection operator is the first
operator which is applied to the population to form the
population for the next generation by selecting the above
average chromosome and eliminating the below average
chromosome. Here, we have used the well-known tournament selection scheme of size two with replacement as
the selection strategy. This process selects the better

Int J Syst Assur Eng Manag

8
x 1
>
< ik
xik  1
0
xik
>
: xik 1
xik  1

chromosome/individual from randomly selected two chromosomes/individuals. This selection procedure is based on
the following assumptions:
1.
2.
3.

4.

For feasible chromosomes/individuals, the better one

with respect to fitness value is selected.
For one feasible and another infeasible chromosomes/
individuals, the feasible one is selected.
For both infeasible chromosomes/individuals with
unequal constraint violations, the chromosome with
less constraint violation is selected.
For both infeasible chromosomes/individuals with
equal constraint violations, then any one chromosome/individual is selected.

Crossover is a key operator of genetic algorithm. The

purpose of this operator is to generate the rearrangement of
co-adapted groups of information from high performance
structures. Here, we have used the uniform crossover as the
crossover operator. The different steps of this operator are
as follows:
Step-1
Step-2
Step-3

Step-4

Find the integral value of bpc  p sizec and store

it in Nc
Select two chromosomes vk and vi randomly from
the population
Compute the components xkj and xij j
1; 2; . . .; n of two offspring by either xkj xkj 
g and xij xij g if xkj [ xij or, xkj xkj g and
xij xij  g, where g is a random integer number

between 0 and
xkj  xij
, j 1; 2; . . .; n
Repeat Step-2 and Step-3 for N2c times

Mutation is a background operator that produces random

changes in various chromosomes. Sometimes, it helps to
regain the information lost in earlier generations. Mainly,
this operator is responsible for fine tuning capabilities of the
system. This operator is applied to a single chromosome
only. Mutation attempts to bump the population gently into a
slightly better way, i.e., the mutation changes single or all
the genes of a randomly selected chromosome slightly. The
mutation operator used herein is the one-neighborhood
mutation. The different steps of this operator are as follows:
Step-1
Step-2
Step-3

Step-4

Find the integral value of bpm  p sizec and store

it in Nm .
Select a chromosome vi randomly from the
population.
Select a particular gene xik k 1; 2; . . .; n on
chromosome vi for mutation and domain of xik is
lik ; uik .
Create new gene x0ik corresponding to the selected
gene xik by mutation process as follows:
For k 1; 2; . . .; n

Step-5

if xik lik
if xik uik
if r [ 0:5
if r  0:5

where r is a random number between 0 and 1.

Repeat Step-2 to Step-4 for Nm times

Sometimes, in any generation, there is a chance that the

best chromosome may be lost when a new population is
created by crossover and mutation operations. To remove
this situation the worst individual/chromosome may be
replaced by that best individual/chromosome in the current
generation. This process is called elitism. In this operation
instead of single chromosome one or more chromosomes
may take part. Elitism guarantees that the objective function values do not improve from one generation to another.
7.1 Termination criteria
The termination condition is a condition for which the
algorithm is going to stop. For this purpose any one of the
following three conditions is considered as the termination
condition.
1.
2.

3.

the best individual does not improve over specified

generations.
the total improvement of the last certain number of
best solutions is less than a pre-assigned small positive
number or
The number of generations reaches a maximum
number of generation i.e., max_gen.

7.2 Algorithm
Step-1

Step-2
Step-3

Step-4

Step-5
Step-6
Step-7

Set population size (p_size), crossover

probability (pc ), mutation probability (pm ),
maximum generation (m-gen) and bounds of
the variables li ; ui i 1; . . .; n
t = 0 [t represents the number of current
generation]
Initialize the chromosome of the population
Pt[Ptrepresents
the
population
at
t  thgeneration]
Evaluate the fitness function of each
chromosome of Ptconsidering the objective
function as the fitness function
Find the best chromosome from the population
Pt
t is increased by unity
If the termination criterion is satisfied go to step14, otherwise, go to next step

123

Int J Syst Assur Eng Manag

Step-8

Step-9
Step-10
Step-11
Step-12
Step-13
Step-14
Step-15

Select the population Pt from the population

Pt  1 of earlier generation by tournament
selection process
Alter the population Pt by crossover, mutation
and elitism operators
Evaluate the fitness function value of each
chromosome of Pt
Find the best chromosome from Pt
Compare the best chromosome of Pt and
Pt  1 and store better one
Go to step-6
Print the best chromosome (which is the solution
of the optimization problem)
End

8 Numerical illustrations
In this section, we have considered the redundancy allocation
problem for low-level redundancy (see Fig. 3) and for highlevel redundancy of five-link bridge system (see Fig. 4) for
numerical experiments. In bridge system network, subsystem-5 represents a hub whereas other subsystems represent
servers/client with processors arranged in parallel.
This five-link bridge network system (Kuo et al. 2001)
works successfully as long as one of the paths, (subsystem1-subsystem-2) or (subsystem-3- subsystem-4), is active
independently of subsystem-5. However, if the pair of
subsystems (1, 4) or (2, 3) fails, then subsystem-5 plays an
important role in the system operation. In each subsystem-i
i 1; 2; 3; 4; 5, there is a parallel configuration consisting
of xi identical components having reliability r~i . If R~i be the
system reliability of subsystem-i, i 1; 2; 3; 4; 5 then
R~i 1  1  r~i xi ; i 1; 2; 3; 4; 5:

Fig. 4 High-level redundancy of five-link bridge system

The system reliability of the low-level five-link bridge

network system is given by the expression as follows:
~2 Q
~3 R~4 R~5
~2 R~3 R~4 Q
~1 R~2 R~3 R~4 R~1 Q
R~S x R~1 R~2 Q
~4 R~5 ; where R~i
~1 R~2 R~3 Q
Q
~
1  Qi ; i 1; 2; 3; 4; 5
The system reliability of the high-level five-link bridge
network system is given by the expression as follows:
R~S h 1  1  ~
r1 r~2 q~2 r~3 r~4 q~1 r~2 r~3 r~4 r~1 q~2 q~3 r~4 r~5
h
q~1 r~2 r~3 q~4 r~5 , where r~i 1  q~i , i 1; 2; 3; 4; 5 and h be the
number of redundant subsystems, arranged in parallel.
For low-level redundancy the corresponding system
reliability maximization and cost minimization problems
are of the form as follows:
Example-8.1

Fig. 3 Low-level redundancy of five-link bridge system

123

~2 R~3 R~4 Q
~1 R~2 R~3 R~4
Maximize R~S x R~1 R~2 Q
~
~
~4 R~5
~
~
~
~
R1 Q2 Q3 R4 R5 Q1 R~2 R~3 Q

Int J Syst Assur Eng Manag

subject to,
g~1 x

5
X

Table 1 Values of the parameters related to Examples 8.18.4

c~j xj exp

x 
j

j1

g~2 x

5
X

~j xj exp
w

j1

 b~1  0;

 b~2  0;

rj

cj

b1

wj

b2

RT

[0.64,
0.66]

[3, 5]

[105,
115]

[1.5, 1.6]

[30,
35]

[0.99,
0.999]

[0.73,
0.76]

[4.5, 5]

[2, 2.5]

[0.75,
0.77]

[5.5,
7.5]

[2, 2.25]

[0.83,
0.86]

[5, 7]

[1.5,
1.75]

[0.88,
0.90]

[2, 2.5]

[1.75, 2]

and
Example-8.2
Minimize C~s x

5
h
x i
X
j
c~j xj exp
4
j1

subject to; R~S x  R~T

Table 2 Computational results for examples 8.118.4

~2 Q
~3
~2 R~3 R~4 Q
~1 R~2 R~3 R~4 R~1 Q
Where R~S x R~1 R~2 Q
~4 R~5
~ 1 R~2 R~3 Q
R~4 R~5 Q
For high-level redundancy the corresponding system
reliability maximization and cost minimization problems
are of the form as follows:

Example

xs/h

(2, 2, 1, 2,
2)

(1)

(1, 1, 2,
1, 1)

(3)

Best found
system
reliability

[0.939545,
0.999027]

[0.819842,
0.928286]

Example-8.3

Mean value of
system
reliability

[0.939545,
0.999027]

[0.819842,
0.928286]

Best found
system cost

[53.186,
71.904]

[102.34,
138.159]

Mean value of
system cost

[53.186,
71.904]

[102.34,
138.159]

Standard
deviation of
system
reliability

[0,
0.059482]

[0,
0.108444]

Standard
deviation of
system cost
CPU time in
seconds

[0,
18.718]

[0,
35.819]

0.04000

0.07000

0.03000

0.18000

Maximize R~S h 1  1  ~
r1 r~2 q~2 r~3 r~4 q~1 r~2 r~3 r~4
r~1 q~2 q~3 r~4 r~5 q~1 r~2 r~3 q~4 r~5 h
subject to,

  X
5
h
g~1 x h exp
c~j  b~1  0;
4
j1


  X
5
h
~j  b~2  0;
w
g~2 x h exp
4
j1
and
Example-8.4

  X
5
h
c~j
Minimize C~s h h exp
4
j1
subject to; R~S h  R~T
where R~S h 1  1  ~
r1 r~2 q~2 r~3 r~4 q~1 r~2 r~3 r~4 r~1 q~2
q~3 r~4 r~5 q~1 r~2 r~3 q~4 r~5 h
All the values of the parameters related to problems
8.18.4 are given in Table 1:
The proposed method has been coded in C programming
language. The computational work has been done on a PC
with Intel core-i3 processor in Linux environment. For
each example 20 independent runs have been performed to
calculate the best found system reliability and best found
system cost which are nothing but the optimal values of
system reliability and system cost. Also we have been
computed the statistical measure like mean and variance of

system reliability as well as system cost. In this computation, the values of genetic parameters like p_size, max_gen,
pm and pc have been taken as 100, 100, 0.15 and 0.85
respectively. The computational results have been shown in
Table 2.
It has been observed from the computational results that
the mean system reliability/mean system cost coincides
with the best found system reliability/system cost. This
strict coincidence is due to the fact that each trial run
provides us optimum solution. Also, the lower ends of the
standard deviations, measured in interval form, assume
zero value. It may also be noted that the average CPU time
required for implementing the genetic algorithm, is also
very less.

123

Int J Syst Assur Eng Manag

Lower bound of system

reliability

Lower bound of interval valued

system reliability

Upper bound of system

reliability

Upper bound of interval valued

system reliability

interval valued system

reliability

0.995

0.98

reliability

Interval valued system

0.96
0.94
0.92
0.9
10

20

30

40

50

0.98
0.965
0.95
0.935
0.92
0.45

60

Population size

0.55

0.65

0.75

0.85

0.95

Probability of crossover

Fig. 5 P_size versus interval valued system reliability for Example 1

Fig. 7 P_cross versus interval valued system reliability for Example 1

Lower bound of interval

Lower bound of interval valued

system reliability

valued system reliability

Upper bound of interval

Upper bound of interval valued

system reliability

valued system reliability

1

0.96
0.94
0.92
0.9
10

20

30

40

50

60

Max_gen

0.98

reliability

Interval valued system

reliability

Interval valued system

0.995
0.98

0.965
0.95
0.935
0.92
0

0.05

0.1

0.15

0.2

0.25

0.3

Probability of mutation

Fig. 6 Max_gen versus interval valued system reliability for Example 1

Fig. 8 P_mute versus interval valued system reliability for Example 1

9 Sensitivity analysis

greater than or equal to 30. This implies that our proposed

GA is stable when population size exceeds 30. From Fig. 6
it is clear that our proposed GA is stable when maximum
number of generation is greater than or equal to 10. In
Figs. 7, 8, the values of interval valued system reliability
have been computed with respect to the probability of
crossover within the range 0.450.95 and the probability of
mutation within the range 0.050.30 respectively. From
these figures, it is clear that the proposed GA is stable with
respect to probability of crossover as well as the probability
of mutation.

To investigate the overall performance of the proposed GA

based penalty technique for solving low-level redundancy
as well as high level redundancy, sensitivity analyses have
been carried out graphically on the interval valued system
reliability with respect to different GA parameters separately taking other parameters at their original values.
These have been shown in Figs. 5, 6, 7, 8. From Fig. 5 it is
observed that both the bounds of the interval valued system
reliability be the same for all the values of population size

123

Int J Syst Assur Eng Manag

10 Concluding remarks
In this paper, for the first time, we have formulated two
different redundancies known as low-level redundancy and
high-level redundancy and proposed four problems where
each problem belongs to the category of interval valued
nonlinear integer programming problems. Then we have
solved these problems corresponding to constrained single
objective interval valued reliability optimization problem.
The reduced problem has been converted to unconstrained
interval valued integer programming problem using Big-M
penalty technique and solved by genetic algorithm. To
solve the problem, we have developed a real coded GA for
integer variables with interval valued fitness function,
tournament selection, uniform crossover and one neighborhood mutation and elitism of size one. It is well known
that the penalty coefficient plays a crucial role in solving
constrained optimization problem by penalty function
technique. However, the selection of the value of this
parameter is a formidable task. To avoid this difficulty, we
have used Big-M penalty technique which does not require
any penalty coefficient. This entire approach opens up the
scope for reliability optimization when reliability values
and other design parameters are interval/imprecise valued.
Thus, it can be claimed that the generalization attempted in
this paper can be handled the real life problem with
imprecise parameters. For further research, one may use
the proposed GA and interval approach in solving interval
valued-integer programming problems as well as interval
valued mixed-integer programming problems relating to
several real life application problems.
Acknowledgments The first author would like to acknowledge the
support of the University Grants Commission (UGC), India, for
conducting this research work.

References
Aggarwal KK, Gupta JS (2005) Penalty function approach in heuristic
algorithms for constrained. IEEE Trans Reliab 54:549558
Bhunia AK, Sahoo L (2011) Genetic algorithm based reliability
optimization in interval environment, Innovative Computing
Methods and Their Applications to Engineering Problems, N.
Nedjah (Eds.). SCI 357:1336
Bhunia AK, Sahoo L, Roy D (2010) Reliability stochastic optimization for a series system with interval component reliability via
genetic algorithm. Appl Math Comput 216:929939
Chern MS (1992) On the computational complexity of reliability
redundancy allocation in a series system. Oper Res Lett
11:309315
Goldberg DE (1989) Genetic algorithms: search, optimization and
machine learning. Addison Wesley, Reading, MA
Gupta RK, Bhunia AK, Roy D (2009) A GA Based penalty function
technique for solving constrained redundancy allocation problem

of series system with interval valued reliabilities of components.

J Comput Appl Math 232:275284
Ha C, Kuo W (2006) Multi-path heuristic for redundancy allocation:
the tree heuristic. IEEE Trans Reliab 55:3743
Hansen E, Walster GW (2004) Global optimization using interval
analysis. Marcel Dekker Inc, New York
Hikita M, Nakagawa K, Nakashima K, Narihisa H (1992) Reliability
optimization of systems by a surrogate-constraints algorithm.
IEEE Trans Reliab 41:473480
Hwang CL, Tillman FA, Kuo W (1979) Reliability optimization by
generalized Lagrangian-function and reduced-gradient methods.
IEEE Trans Reliab 28:316319
Karmakar S, Mahato S, Bhunia AK (2009) Interval oriented multisection techniques for global optimization. J Comput Appl Math
224:476491
Kim JH, Yum BJ (1993) A heuristic method for solving redundancy
optimization problems in complex systems. IEEE Trans Reliab
42(4):572578
Kuo W, Prasad VR (2000) An annotated overview of system
reliability optimization. IEEE Trans Reliab 49(2):487493
Kuo W, Hwang CL, Tillman FA (1978) A note on heuristic methods
in optimal system reliability. IEEE Trans Reliab 27:320324
Kuo W, Lin H, Xu Z, Zang W (1987) Reliability optimization with
the Lagrange multiplier and branch-and-bound technique. IEEE
Trans Reliab 36:624630
Kuo W, Prasad VR, Tillman FA, Hwuang CL (2001) Optimal
reliability design fundamentals and application. Cambridge
University Press, Cambridge
Mahato SK, Sahoo L, Bhunia AK (2012) Reliability-redundancy
optimization problem with interval valued reliabilities of components via genetic algorithm. J Inf Comput Sci 7(4):284295
Miettinen K, Makela MM, Toivanen J (2003) Numerical comparison
of some penalty-based constraint handling techniques in genetic
algorithms. J Global Optim 27:427446
Misra KB, Sharma U (1991) An efficient algorithm to solve integer
programming problems arising in system reliability design. IEEE
Trans Reliab 40:8191
Nakagawa Y, Miyazaki S (1981) Surrogate constraints algorithm for
reliability optimization problems with two constraints. IEEE
Trans Reliab 30:175180
Nakagawa Y, Nakashima K (1977) A heuristic method for determining optimal reliability allocation. IEEE Trans Reliab R-26(3):
156161
Sahoo L, Bhunia AK, Roy D (2010) A genetic algorithm based
reliability redundancy optimization for interval valued reliabilities of components. J Appl Quant Methods 5(2):270287
Sahoo L, Bhunia AK, Kapur PK (2012a) Genetic algorithm based
multi-objective reliability optimization in interval environment.
Comput Ind Eng 62:152160
Sahoo L, Bhunia AK, Roy D (2012b) An application of genetic
algorithm in solving reliability optimization problem under
interval component Weibull parameters. Mex J Operat Res
1(1):219
Sun X, Li D (2002) Optimal condition and branch and bound
algorithm for constrained redundancy optimization in series
system. Optim Eng 3:5365
Sung CS, Cho YK (1999) Branch and bound redundancy optimization
for a series system with multiple-choice constraints. IEEE Trans
Reliab 48:108117
Tillman FA, Hwuang CL, Kuo W (1977) Optimization technique for
system reliability with redundancy: a review. IEEE Trans Reliab
26:148155

123