Strad Research https://doi.org/10.37896/sr8.

7/036 ISSN: 0039-2049

Multi-State System Reliability Analysis with Two

Types of Common Cause Shock Failures
Changkuoth Jock Chol #1 and Dr. G. Y. Sagar*2
#1,*2
Department of Statistics, College of Natural and Computational Sciences, Gambella University, Ethiopia.

Abstract— In this article, Reliability analysis of a three unit identical system is discussed. The units of system may affect by two types
of failures namely, Lethal Common Cause Shock (LCCS) and Non-Lethal Common Cause Shock (NCCS) failures. Using stochastic
process, the set of differential equations of the current model are derived to obtain reliability measures such as reliability of the
system and Mean Time to Failure (MTTF) in the case of series and parallel. Also the Maximum Likelihood Estimates (MLE) of the
above said measures are discussed and presented in numerical illustration by using simulation. Tables display the findings and
suggest that LCCS and NCCS are the most dominant causes of failures while studying the performance of the systems in reliability
theory.

Keywords— 3-Unit system, Stochastic process, Reliability, MTTF, LCCS & NCCS Failures, M L Estimation, Simulation

I. INTRODUCTION
Reliability is the most commonly accepted analysis tool for solution of engineering problems. It is the
parameter which is used to assess the effectiveness of the system/item and availability of the system under
proper working conditions for a given period of time. Initially, the reliability evaluation techniques have
been used in aerospace industry and military applications, there after nuclear power plants, electricity
supply and continuous process plants were rapidly applied the developments of reliability techniques.
Situations where the failures of some or even majority of system units could lead to partial ability or
partial system down time to perform required operations are quite common in electrical/mechanical
systems. These types of models are also used to describe multi-channel systems (eg. telecommunications
and transportation).
While evaluating the system reliability, we need to consider the Common Cause Shock (CCS) failures
which can severely degrade the reliability of devices, systems etc. These events are purely external causes
which produces multiple failures. As per the reliability literature, in particular two types of CCS failures
viz. Lethal common cause shock failures, which is the occurrence of simultaneous outage of all units in the
system and the other is non-lethal common cause shock failures, which is the occurrence of random
number of units to simultaneous outage of several units in the system. Some attempts have been made in
this direction by several authors. Billinton and Allan [1] discussed the role of common cause shock
failures in different frame works. Chari et al [2] derived the reliability measures of a two unit system in the
presence of common cause shock failures. Dhillon [3], [4] discussed the role of common cause failures as
well as human errors in system reliability aspects. Reddy [5] has developed reliability measures for two
component non-identical system with common cause failures. Sagar et al [6], [8] and Awgichew et al [7]
examined the reliability measurements with common cause shock failures for two unit identical system.
They derived M L estimates of two unit system reliability measures such as frequency of failures in the
presence of CCS failures. Sreedhar et al [9], [10] analysed two unit non identical system with CCS failures.
They studied M L estimation approach for estimating reliability indices.
II. MODEL, ASSUMPTIONS AND NOTATIONS
A. System Description
In this article, as we discussed in the introduction, none of the authors studies three unit identical
systems with LCCS and NCCS failures and maximum likelihood estimation as well. We examined the
reliability of the redundant system in series as well as parallel configurations. There are four different
possible states for the system operation: perfect state, minor failed state, major failed state, and completely
failed states. The failure rates of each unit are constant in nature, but they follow exponential distribution.
B. Assumptions
In this paper, we consider the following assumptions:
1. The system operates effectively until one or more units are functioning.

VOLUME 8, ISSUE 7, 2021 344 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

2. The system has four states: perfect, minor partially failed, major partially failed, and completely
failed.
3. The system units fail individually and also simultaneously due to lethal common cause shock
failures or non-lethal common cause shock failures in Poisson manner.
4. Individual, lethal common cause shock and non-lethal common cause shock failures are
independent to each other.
5. A repair man is available and ready to restore minor and major faults whether they are failed
individually or simultaneously due to common cause shocks.
6. The repair times of failed units depend on the failure mode and are assumed exponentially
distributed.
C. Notations:

s/t – Laplace transform/Time scale variable

 / /  – Failure rate of individual unit / LCCS / NCCS
0 / 1 – Repair rates
T – Time to failure of a unit
p(q) – The probability of simultaneous failures of units due to NCCS / LCCS
Pi(t) – Probability that the system is in state (i = 0, 1, 2, 3) at time t
RLNS (t ) / RLNP (t ) – Reliability of the system when units are in series / parallel
Rˆ (t ) / Rˆ (t )
LNS LNP – M L estimate of reliability function for series / parallel system
ELNS (T ) / ELNP (T ) – Mean time to failure for series / parallel system

III.STATE TRANSITION DIAGRAM AND DESCRIPTION

In view of the stated assumptions, we formulate state transition diagram of the model in Fig.1. The state
description of the current model highlights that initially all the units are functioning perfectly and it in a
state of s0. After any one of the three units is down and others are functioning, it switches to state s1 which
is regarded as minor partially down state. If two units have failed, it will be passed to s2 that is the major
partially down state. In both cases, to restore the system we use general repair. State s3 indicates
completely down state due to failure of all the three units. The quantities that appear in Fig.1 are defined as:
0  3(   pq 2 ) 

1  (  p 3   ) 
2  (  p 2   ) 
 (1)
c     p 
0   

1  2 

λc

λ0 λ1
λ2
s0 s1 s2 s3
μ0 μ1

Fig.1 State transition diagram

VOLUME 8, ISSUE 7, 2021 345 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

IV. MATHEMATICAL MODEL

The set of differential equations associated with the current mathematical model for the above state
transition diagram are:
P0' (t )  (0  c ) P0 (t )  0 P1 (t ) (2)
P1' (t )  0 P0 (t )  (1  0 ) P1 (t )  1 P2 (t ) (3)
P (t )  1 P1 (t )  (2  1 ) P2 (t )
2
'
(4)
P (t )  c P0 (t )  2 P2 (t )
3
'
(5)
Initial conditions: P0(0) = 1, and other state probabilities are zero at t = 0
Taking Laplace transformation of equations (2) to (5) and using initial conditions, we obtain
2 2 2
r1  r1 K  L r2  r2 K  L r3  r3 K  L
P0 (t )  exp( r1t )  exp( r2 t )  exp( r3 t ) (6)
( r1  r3 )( r1  r2 ) ( r1  r2 )( r2  r3 ) ( r1  r3 )( r2  r3 )
0 ( r1  2  1 ) 0 ( r2  2  1 ) 0 (r3  2  1 )
P1 (t )  exp( r1t )  exp( r2 t )  exp( r3t ) (7)
( r1  r3 )( r1  r2 ) ( r1  r2 )( r2  r3 ) ( r1  r3 )( r2  r3 )
0 1 0 1 0 1
P2 (t )  exp( rt
1
) exp( r2 t )  exp( r3t ) (8)
( r1  r3 )( r1  r2 ) ( r1  r2 )( r2  r3 ) ( r1  r3 )( r2  r3 )

P3 (t )  1  [ P0 (t )  P1 (t )  P2 (t )] (9)

 A1
r1   r sin( )  
3 
  A1 
r2  r sin       (10)
3  3 
   A
r3  r sin       1 
 3  3

Here

A1 A2 A13 
q  A3  2 
3 27 
2 
A 
2 1/ 2
r 1
 3 A2 
3 
 4q  
sin 1  3  

 r  
3 
where
K  (1  2   0  1 ) 
 (11)
L  (12   0 2   0 1 
A1  ( 0  1  2  0  1  c )
A2  ( 0 1   0 c  1c  10   0 2  1c  2 c  0 1  0 2  12 )
A3  ( 0 1c   0 c 2  12 c  0 12 )

VOLUME 8, ISSUE 7, 2021 346 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

V. SOME RELIABILITY CHARACTERISTICS

In this section, we derived some performance measures when three units of the system are in series and
in parallel modes.
A. Series System
In this case, all units of the system are in good working condition. The states s 1 to s2 and s2 to s3 are
absorbing states and hence no transition is allowed. Therefore, the reliability function is given by:

RLNS (t )  P0 (t )
 exp( (4   p (1  3q 2 ))t ) (12)

And the mean time to failure is:


ELNS (T )   RLNS (t ).dt
0

1
 (13)
4   p (1  3q 2 )
B. Parallel System
The reliability function for parallel system is:

RLNP (t )  P0 (t )  P1 (t )  P2 (t )
 M1 exp(r1t )  M 2 exp( r2t )  M 3 exp( r3t ) (14)

Where
M1 
 ( r12  r1 K  L )  0 ( r1  2  1 )  0 1  ( r1  r3 )(r1  r2 )
M2   (r
2
2

 r2 K  L)  0 (r2  2  1 )  01 ( r2  r3 )(r1  r2 )
M3   (r
3
2
 r3 K  L )  0 ( r3  2   )     ( r  r )( r  r )
1 0 1 1 3 2 3

also K , L, r1 , r2 , r3 are defined in equations (11) and (10)

and MTTF of parallel system is:

ELNP (T )   RLNP (t ).dt
0

(0 1  0 2  10  12  0 2  0 1 )

 (15)
r1r2 r3

Where 0 , 1 , 2 , 0 , 1 and r1 , r2 , r3 are defined in (1) and (10)

C. Numerical Illustration
For illustration purpose by fixing   0.01, 0  1, 1  1.5, p  0.3 and for different values of time-variable
t = 0, 2, 4, 6, 8, 10, 12, 14, 16, 18 and 20 units of time, we get different values of reliability for series and
parallel cases as shown in table 1.

VOLUME 8, ISSUE 7, 2021 347 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

TABLE I

RELIABILITY FOR SERIES AND PARALLEL SYSTEMS

Time Series System Parallel System

(t)
β=0.2, ω=0.2 β=0.3, ω=0.3 β=0.4, ω=0.5 β=0.2, ω=0.2 β=0.3, ω=0.3 β=0.4, ω=0.5

0 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

2 0.6863 0.5918 0.5103 0.8761 0.8293 0.7804

4 0.4710 0.3502 0.2604 0.7703 0.6908 0.6074

6 0.3233 0.2073 0.1329 0.6776 0.5756 0.4722

8 0.2219 0.1227 0.0678 0.5960 0.4796 0.3670

10 0.1523 0.0726 0.0346 0.5243 0.3997 0.2852

12 0.1045 0.0430 0.0277 0.4612 0.3330 0.2217

14 0.0717 0.0254 0.0090 0.4057 0.2775 0.1723

16 0.0492 0.0150 0.0046 0.3569 0.2313 0.1339

18 0.0338 0.0089 0.0023 0.3139 0.1927 0.1041

20 0.0232 0.0053 0.0012 0.2761 0.1605 0.0809

TABLE II

MTTF FOR SERIES AND PARALLEL SYSTEMS

µ = 1, p = 0.2

Series System Parallel System

(β, ω) (β, ω)
λ

(0.1, 0.1) (0.2, 0.2) (0.3, 0.4) (0.1, 0.1) (0.2, 0.2) (0.3, 0.4)

0.01 10.163 6.378 4.647 35.702 21.605 14.702

VOLUME 8, ISSUE 7, 2021 348 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

0.02 7.225 5.081 3.918 27.459 18.452 13.131

0.03 5.605 4.223 3.387 22.542 16.215 11.925

0.04 4.579 3.613 2.983 19.275 14.544 10.970

0.05 3.869 3.157 2.665 16.947 13.250 10.195

0.06 3.351 2.803 2.408 15.205 12.218 9.553

0.07 2.955 2.520 2.197 13.851 11.375 9.014

0.08 2.643 2.289 2.019 12.769 10.674 8.553

0.09 2.390 2.097 1.868 11.885 10.082 8.155

0.10 2.182 1.935 1.739 11.149 9.575 7.809

VI. ESTIMATION AND SIMULATION

A. Estimation
In this, we have attempted Maximum likelihood estimation to estimate the system reliability and MTTF of
the present model. However, the system is under the influence of NCCS and LCCS failures in addition to
individual failures.
Let the samples x1 , x2 ,........., xn ; y1 , y2 ,........., yn and w1 , w2 ,........., wn with size ‘n’ representing times between
individual, NCCS and LCCS failures which will obey exponential law.
Let the samples z11 , z12 ,........., z1n ; z21 , z22 ,........., z2 n with size ‘n’ number of times between repairs of the
units with exponential population law.

xˆ , yˆ , wˆ , zˆ1 , zˆ2 are the maximum likelihood estimates of  ,  ,  , 0 , 1 respectively.

1 1 1 1
Where, xˆ  ; yˆ  ; wˆ  ; zˆ1  ; zˆ2  ; x 
1  xi ; y   yi w   wi ; z   z1i ; z  z 2i
1 2
x y w z1 z2 n n n n n

B. Simulation
We compute M L estimates such as Rˆ LNS (t ), Rˆ LNP (t ) of the present model by using Monte-Carlo
simulation. For a range of specified values of the rates of  ,  ,  , 0 , 1 and for the sample size n=5(5)15
were simulated in each case with N=20000(30000)100000 in order to evolve mean square error (MSE) in
each case by using C++ (software).

VOLUME 8, ISSUE 7, 2021 349 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

TABLE III

RELIABILITY ESTIMATION FOR SERIES SYSTEM

 = 0.1, β = 0.2, ω = 0.3, p = 0.3, t = 1

SAMPLE SIZE (n = 5)

N RLNS (t ) Rˆ LNS (t ) MSE

20000 0.577989 0.492346 0.025525

50000 0.577989 0.491844 0.025762

80000 0.577989 0.492411 0.025832

SAMPLE SIZE (n = 10)

N RLNS (t ) Rˆ LNS (t ) MSE

20000 0.577989 0.522508 0.011552

50000 0.577989 0.523343 0.011446

80000 0.577989 0.523294 0.011507

SAMPLE SIZE (n = 15)

N RLNS (t ) Rˆ LNS (t ) MSE

20000 0.577989 0.532862 0.007407

50000 0.577989 0.533398 0.007394

80000 0.577989 0.533402 0.007395

VOLUME 8, ISSUE 7, 2021 350 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

TABLE IV

RELIABILITY ESTIMATION FOR PARALLEL SYSTEM

 = 0.1, β = 0.2, ω = 0.3, µ0 = 1, µ1 = 1.5, p = 0.3, t = 1

SAMPLE SIZE (n = 5)

N RLNP (t ) Rˆ LNP (t ) MSE

20000 0.868255 0.830932 0.003984

50000 0.868255 0.831099 0.004021

80000 0.868255 0.831404 0.003957

SAMPLE SIZE (n = 10)

N RLNP (t ) Rˆ LNP (t ) MSE

20000 0.868255 0.846471 0.001455

50000 0.868255 0.846431 0.001455

80000 0.868255 0.846591 0.001446

SAMPLE SIZE (n = 15)

N RLNP (t ) Rˆ LNP (t ) MSE

20000 0.868255 0.851125 0.000879

50000 0.868255 0.850872 0.000895

80000 0.868255 0.851039 0.000893

VII. RESULT DISCUSSION AND CONCLUSIONS

This paper analyzes the reliability measures of a three unit system in series and parallel under the lethal
and non-lethal common cause shock failures. A study of the model with the support of maximum
likelihood estimation were presented and established empirically. The importance of LCCS and NCCS
failures in these types of models were discussed through numerical illustration and simulation validity in
this article. The following decision can be made based on the analysis carried out in this paper.
Table I show evidence for the reliability of the system at various time values. The reliability is
decreasing in both series and parallel cases when LCCS and NCCS failure rates are increasing. Table II
include the variation in the MTTF corresponding to different failure rates in series and parallel system. It

VOLUME 8, ISSUE 7, 2021 351 http://stradresearch.org/

Strad Research https://doi.org/10.37896/sr8.7/036 ISSN: 0039-2049

is observed that MTTF decreases as the failure rate increases and also there is a great improvement from
series to parallel system. Table III and Table IV show the simulation study in order to establish the validity
of the proposed maximum likelihood estimates. It is observed that the point estimates become more
accurate when the sample size is large and mean square error decreases with increasing the sample size.
The model discussed in this article was found to be great importance in proper maintenance analysis,
and performance evaluation of the system.

ACKNOWLEDGMENT
I, (Changkuoth Jock Chol) am thankful to Prof. (Dr). G Y Sagar for his keen interest and continuous
encouragement to do this paper. He is co-author of this article. Also we would like to express our gratitude
to editors and anonymous reviewers for their useful and rigorous comments which have improved the
quality of the manuscript.

REFERENCES
[1] R. Billinton and R. N. Allan, Reliability evaluation of engineering systems: concepts and techniques, Plenum Press, New York, 1983.
[2] A. A. Chari, M. P. Sastry, and S. Madhusudhana Verma, “Reliability analysis in the presence of common cause shock failures,” Micro-Electronics and
reliability, vol. 31, pp. 15-19, 1991
[3] B. S. Dhillon, “On common cause failures-bibliography,” Micro Electronics and Reliability, vol. 18, pp. 533-534, 1978.
[4] B. S. Dhillon, “Modeling human errors in repairable systems,” in Proc. Annual Reliability and Maintainability Symposium, 1989, pp. 418-424.
[5] Y. R. Reddy, “Reliability analysis for two unit non-identical system with CCS failures,” Ph. D thesis, S. K. University, Anantapur, India,
2003.
[6] G. Y. Sagar, K. Awgichew, M. F. Melkamu and S. M. Abdulfeta, “Simulation Study on Reliability Estimates of a Repairable System with Lethal and
Non-Lethal Common Cause Shock Failures,” Elixir Statistics, vol.126, pp. 52481-52484, 2019.
[7] K. Awgichew and G. Y. Sagar, “Estimation of Availability Measures and Confidence Interval for two unit system with Common Cause Shock failures
and Human Errors,” IOSR Journal of Mathematics (IOSR-JM), Vol. 14(3), pp. 52-59, 2018.
[8] G. Y. Sagar, “Frequency of Failures of a System and Confidence-interval,” Elixir Statistics, vol. 70, pp. 23818-23821, 2014.
[9] B. R. Sreedhar, G. Y. Sagar, K. Pushpanjali, and Y. R. Reddy, “M L Estimation of the reliability measures of a two unit system in the presence of two
kinds of CCS failures,” ARPN Journal of Engineering and Applied Sciences, vol. 7(8), pp. 980 – 986, 2012.
[10] B. R. Sreedhar, K. Pushpanjali, G. Y. Sagar, and Y. R. Reddy, “Evaluation of System Availability and Frequency of Failures with Lethal and NonLethal
CCS Failures By Maximum Likelihood Estimation,” International Journal of Engineering Science and Technology (IJEST), vol. 4(6), pp. 2677-2691,
June 2012.

VOLUME 8, ISSUE 7, 2021 352 http://stradresearch.org/