Availability analysis of crank-case manufacturing in a two-wheeler

automobile industry
Savita Garg
a,
*
, Jai Singh
b
, D.V. Singh
c
a
Department of Mathematics, Mukand Lal National College, Yamuna Nagar, Haryana 135 001, India
b
Punjab College of Engineering and Technology, Lalru Mandi (Mohali), India
c
Department of Mathematics, National Institute of Technology, Kurukshetra 136 119, India
a r t i c l e i n f o
Article history:
Received 3 December 2008
Received in revised form 5 September 2009
Accepted 9 September 2009
Available online 17 September 2009
Keywords:
Availability
Preventive maintenance
Corrective maintenance
RungeKutta method
a b s t r a c t
The paper describes the availability of crank-case manufacturing system in an automobile
industry. The units discussed here fail either directly from normal working state or indi-
rectly through partial failure state. The machines are subjected to both preventive and cor-
rective maintenance. Failure and repair times of the units are independent. The problem is
formulated using probability consideration and supplementary variable technique. The
system of equations governing the working of system consists of ordinary as well as partial
differential equations. Lagrange method and RungeKutta method is used to solve partial
differential equation and ordinary differential equation respectively. The study reveals that
successful program of preventive and routine maintenance will reduce equipment failures,
extend the life of the equipment, and increase the system availability to considerable
margin.
2009 Elsevier Inc. All rights reserved.
1. Introduction
Reliability, in general, can be dened as the ability of a system to perform its required functions under stated conditions
for a specied period of time. Reliability technology is an important phenomenon of the existing era. This technology is
widely used to increase the efciency of the system. To overcome day-to-day problems, now a day, the system analysts
and engineers are interested in the analysis of reliability models to implement them for practical utility. The paper consists
of a brief discussion of method of calculating availability for idealized and faulty preventive maintenance. The problem is
also reduced to the case when no preventive maintenance (PM) is done. The paper also presents the steady state and tran-
sient study of the system in evaluation of availability. The primary action of the preventive maintenance is to control the
condition of the system and to ensure its availability. Efciency of PM also affects the operational behavior of the system.
Perfect and efcient PM means no damage and no error during maintenance operation. The faulty and inefcient PM means
that either replaced component may not have worked properly or maintenance personal may have damaged the equipment.
Corrective maintenance (CM) is the maintenance action carried out to restore a defective system or component to an oper-
ative condition. Many authors have studied preventive maintenance systems. Barlow and Hunter [1] studied the preventive
maintenance models with minimal repairs. Singh [2] computed the state probabilities of a complex system having four types
of components with preemptive repeat priority repairs. Zhang [3] studied the stochastic behavior of an (N+1)-unit standby
system under preemptive priority repair rule and obtained the expressions for transient as well as steady states of the sys-
tem using supplementary variable technique and Laplace transforms. Gandhi and Wani [4] evaluated maintainability index
of mechanical system using digraph and matrix method. Grall et al. [5] presented a preventive maintenance structure for a
0307-904X/$ - see front matter 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2009.09.016
* Corresponding author.
E-mail address: savitaphd@rediffmail.com (S. Garg).
Applied Mathematical Modelling 34 (2010) 16721683
Contents lists available at ScienceDirect
Applied Mathematical Modelling
j our nal homepage: www. el sevi er . com/ l ocat e/ apm
gradually deteriorating single unit system. Kumar [6], Singh [7], Gupta et al. [8], Garg et al. [9] and some other workers ap-
plied reliability technology to various industrial systems obtaining important results.
In the present paper, the production of a crank-case in a two-wheeler automobile company has been discussed in detail.
The raw material used is raw crank-case. The paper focuses on the system availability, and obtained the valuable expressions
for availability under various conditions. The system availability has been discussed for two states: transient state and steady
state. Numerical analysis of availability of the system for transient state and steady state has been done by using RungeKut-
ta method and software package MATLAB 5.1 respectively. Based on the analysis of system, suggestions have been made to
optimize the production of system.
2. System description and notations
2.1. System description
The crank-case assembly line, discussed in this paper consists of the following nine sub-systems. The units under study
are specialized single purpose machines.
1. Module machine-1 (M
1
).
2. Module SPM widma-1 (W
2
).
3. Module machine-2 (M
3
).
4. Module SPM widma-2 (W
4
).
5. Module machine-3 (M
5
).
6. Fine boring machine (B).
7. Horizontal milling machine (H).
8. Tapping machine-1 (T
8
).
9. Tapping machine-2 (T
9
).
All the sub-systems listed above are single units subjected to revealed as well as unrevealed failure. The rst machine in
the crank-case line is the module machine-1, is used to drill on the sides of the crank-case. Module SPM widma-1 machine is
used to drill on the face of the raw crank-case. The module machine-2 is used for side tapping. The module SPM widma-2 is
used for tapping. Module machine-3 is used for further tapping. Fine bores are drilled with the help of mico ne boring ma-
chine. Milling is done with the help of horizontal milling machine. Then tapping is done on tapping machine-1 and nally on
tapping machine-2. Undergoing all these processes serially, we get a nished crank-case. Fig. 1 gives the schematic ow
chart of the crank-case production process.
2.2. Notations
In addition to the notations used for sub-systems, we have also used the following notations.
- the sub-system/unit is running without any failure
g unit is in good state but not operative
m unit is under preventive maintenance
r unit is under repair or repair continued
A
z
(A = B, H) indicates the working state of ne boring and horizontal milling machine w.r.t. z, (z = -, g, m, r)
M
x;y
i;j
indicates the working states of the sub-systems M
i
and M
j
w.r.t. x, y (x, y = -, g, r, m:: i = 1, 3, 5: j = i + 2, i + 4 if
i = 1; j = i 2, i + 2 if i = 3; j = i 2, i 4 if i = 5). The ordered pairs
x
i
and
y
j
refer the working status of the
units M
i
and M
j
W
x;y
k;l
indicates the working states of the sub-systems W
k
, W
l
w.r.t. x, y (x, y = -, g, r, m:: k = 2, 4: l = 4 if k = 2; l = 2 if k = 4)
T
x;y
u;v
indicates the working states of the sub-systems T
u
, T
v
w.r.t. x, y (x, y = -, g, r, m:: u = 8, 9: v = 9 if u = 8; v = 8 if u = 9)
k
i
respective constant transition rate causing a unit of sub-systems M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
from
normal to degraded state, (i = 1, 2, . . . , 9)
pdf probability density function
l
i
(x),
C(x)
repair rate and pdf of PM time of the sub-systems M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
for its transition from
degraded to normal state, and has an elapsed repair time x; (i = 1, 2, . . . , 9)
k
j
(y) the rst order probability that the sub-systems M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
transits from normal to
failed state in the interval (y, y + D), conditioned that it has not failed up to time y; (j = 10, 11, . . . , 18)
l
j
x; Dx repair rate and pdf of repair time of the M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
to return it from failed to normal
state, and has an elapsed repair time x; (j = 10, 11, . . . , 18)
(continued on next page)
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1673
P
0
(t) probability that the system is working with full capacity at time t
P
z
(t) probability that the system is in state z at time t
P
z
(x, t) probability that the system is in state z at time t and has an elapsed repair time x
P
z
(x, y, t) probability that the system is in z state at time t and failure occurs within the interval (y, y + dy], and repair
is completed within the interval (x, x + dx]
b
i
constant probability that PM of M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
is carried out satisfactorily, and this
makes the system operative; (i = 1, 2, . . . , 9)
1 b
i
constant probability that PM of M
1
, W
2
, M
3
, W
4
, M
5
, B, H, T
8
and T
9
is carried out unsatisfactorily, and this
leads the system to failed state immediately thereafter
3. Assumptions
In the present analysis, it is assumed that
(1) All units are initially operating and are in good state.
(2) Each unit has three states viz., normal, degraded, and failed.
(3) Each unit is good as new after repair.
(4) The operating unit is repaired on failure or at certain age, whichever occurs rst.
(5) All transition rates which brings the sub-system to degraded state from normal state are taken as constant.
(6) Failure and repair time of all the sub-systems are arbitrarily distributed.
(7) Independent maintenance facilities are available to handle scheduled preventive maintenance and corrective mainte-
nance. When the failure of a sub-system or unit is revealed, its PM begins immediately.
(8) At a time, preventive maintenance of only one unit is performed.
(9) If any one of the units is under corrective repair, the scheduled preventive maintenance of other working units of the
sub-systems is not initiated till the repair of the failed unit completes.
(10) There is no simultaneous failure among sub-systems.
Raw Material
Finished Crank Case
Module machine-2
Module SPM widma-1
Module machine-1
Module SPM widma-2
Module machine-3
Fine boring machine
Horizontal milling
machine
Tapping machine-1
Tapping machine-1
Fig. 1. Flow diagram of crankcase production system.
1674 S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683
(11) On pending failure state (revealed failure) of any sub-system, there is a buffer stock, which keeps the system operative
for about eight hours approximately. Thus, partial failure (pending failure state) of any sub-system does not cause the
failure of the system.
Based on the above notations and assumptions, the transition diagram of the system is as in Fig. 2.
4. Mathematical analysis of the system
Probability considerations give the following differential-difference equations associated with the transition diagram.
d
dt
T
0
_ _
P
0
t S
0
t 1
@
@x

@
@t
l
i
x
_ _
P
i
x; t k
i
P
0
t; i 1; 2; . . . ; 9 2
@
@x

@
@y

@
@t
l
j
x
_ _
P
j
x; y; t k
j
yP
0
t 1 b
j9
l
j9
xP
j9
x; t; j 10; 11; . . . ; 18 3
where
T
0

9
i1
k
i

18
j10
k
j
y
and
Fig. 2. Transition diagram.
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1675
S
0
t
_

9
i1
b
i
l
i
xP
i
x; tdx
_

18
j10
l
j
xP
j
x; y; tdxdy 4
4.1. Boundary conditions
P
i
0; t k
i
P
0
t; i 1; 2; . . . ; 9
P
j
0; y; t k
j
yP
0
t; j 10; 11; . . . ; 18 5
4.2. Initial conditions
P
0
0 1
P
z
x; 0 0; z 1; 2; . . . ; 9
P
z
x; y; 0 0; z 10; 11; . . . ; 18 6
where
b
i

1 for idealized maintenance
0 for faulty maintenance
_
; i 1; 2; . . . ; 9 7
In the case, when no preventive maintenance is performed, then
k
i
l
i
0; i 1; 2; . . . ; 9 8
4.3. Solution of equations
Eq. (1) is a linear differential equation of rst order and other Eqs. (2) and (3) are Lagranges type linear partial differential
equations of rst order. Solving the above system of equations under the boundary conditions (5) and initial conditions (6),
we get the state probabilities as given below.
P
0
t e
T
0
t
1
_
S
0
te
T
0
t
dt
P
i
x; t e

_
l
i
xdx
k
i
P
0
t x
_
k
i
P
0
te
_
l
i
xdx
dx; i 1; 2; . . . ; 9
P
j
x; y; t e

_
l
j
xdx
_
fk
j
yP
0
t 1 b
j9
l
j9
xP
j9
x; tge
_
l
j
xdx
dx k
j
y xP
0
t x
_ _
;
j 10; 11; . . . ; 18 9
From the above relations, all the probabilities are obtained in terms of P
0
t , where P
0
t is as given in Eq. (1).
5. Availability function
The time-dependent availability, AV(t) of the system is given by
AVt P
0
t
_

9
i1
P
i
x; tdx 10
5.1. Expression of availability under different eld conditions
(a) Availability function AV
I
t of the system in case idealized preventive maintenance is performed (Here b
i
= 1, i =
1, 2 ,. . . , 9), is obtained as
1676 S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683
AV
I
t e
T
0
t
1
_ _

9
i1
l
i
xP
i
x; tdx
_

18
j10
l
j
xP
j
x; y; tdxdy
_ _
e
T
0
t
dt
_ _

9
i1
e

_
l
i
xdx
k
i
P
0
t x
_
k
i
P
0
te
_
l
i
xdx
dx
_ _
dx 11
(b) Availability function AV
F
(t) of the system in case faulty preventive maintenance is performed (here b
i
= 0, i = 1, 2 ,. . . , 9),
is obtained as
AV
F
t e
T
0
t
1
_ _

18
j10
l
j
xP
j
x; y; tdxdy
_ _
e
T
0
t
dt
_ _

9
i1
e

_
l
i
xdx
k
i
P
0
t x
_
k
i
P
0
te
_
l
i
xdx
dx
_ _
dx 12
From relations (11) and (12)
AV
I
t e
T
0
t
_ _

9
i1
l
i
xP
i
x; tdx
_ _
e
T
0
t
dt
_ _
AV
F
t 13
Relation (13) explore that time-dependent availability of the system is greater if idealized preventive maintenance is
performed.
(c) Availability function AV
C
(t) of the system if only corrective repair is performed, is obtained (as states 1, 2, . . . , 9 van-
ished as no preventive maintenance is performed) as:
AV
C
t e
T
0
t
1
_ _

18
j10
l
j
xP
j
x; y; tdxdy
_ _
e
T
0
t
dt
_ _
14
6. Special cases
6.1. All failure and repair rates are constant
When all failure and repair rates are constant, then the Eqs. (1)(3) reduce to the following system of equations.
d
dt

18
i1
k
i
_ _
P
0
t

9
i1
b
i
l
i
P
i
t

18
j10
l
j
P
j
t 15
d
dt
l
i
_ _
P
i
t k
i
P
0
t; i 1; 2; . . . ; 9 16
d
dt
l
j
_ _
P
j
t k
j
P
0
t 1 b
j9
l
j9
P
j9
t; j 10; 11; . . . ; 18 17
The availability function, AV
(t)
of the system in this case is given by
AV
t

9
i0
P
i
t 18
6.1.1. Numerical analysis
A particular crank-case production system is chosen for case studies having sub-systems as described in Section 2. The
data for PM pertaining to this particular system is as follows.
k
1
k
3
k
5
:00135; k
2
k
4
:0007; k
6
k
7
:0002; k
8
k
9
:00046
l
1
l
2
l
3
l
4
l
5
l
6
l
7
l
9
1:5; l
8
1:7
19
Making use of above data, the numerical results of availability for idealized PM (AV
(I)
), faulty PM (AV
(F)
) and when no PM
is done (AV
(C)
), calculated by relation (18) for monthly failure and repair rates, is given in Tables 112.
The study of the Tables 112 shows that the availability of two-state models is less than the three state models having
same failure and repair rates. Thus it becomes necessary to impart preventive maintenance to the system on reaching to
pending-failed state to enhance its availability in cost effective manner. Also Table 112 demonstrates that the availability
of the system is high if preventive maintenance performed is perfect as compared to faulty preventive maintenance.
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1677
6.2. Steady-state availability of the system with constant transition rates
When t ?1,
d
dt
!0 and
@
@t
!0; Eqs. (1)(3) reduce to linear simultaneous algebraic equations.
Table 1
Availability for idealized, faulty and no preventive maintenance for the month of April.
Month April
Failure rates k
10
:005; k
11
:015; k
12
:01; k
13
:019; k
14
:022; k
15
:0001; k
16
:013; k
17
:004; k
18
:006
Repair rates l
10
1:75; l
11
1:27; l
12
1; l
13
1:14; l
14
1:43; l
15
1; l
16
1:45; l
17
1:66; l
18
:4
Time
(In days)
Availability
AV
I
b
i
1; i 1; 2; . . . 9 AV
F
b
i
0; i 1; 2; . . . 9 AV
C
k
i
l
i
0; i 1; 2; . . . 9
5 .9252178388. .9203134466 .9249108947
10 9237223156 .9186761219 .9234071385
15 .9235390465 .9184782537 .923223014
20 .9235149562 .9184523203 .9231988152
25 .9235117834 .9184489135 .9231956285
30 .9235113654 .9184484660 .9231952088
Table 2
Availability for idealized, faulty and no preventive maintenance for the month of May.
Month May
Failure rates k
10
:014; k
11
:01; k
12
:05; k
13
:032; k
14
:02; k
15
:0001; k
16
:011; k
17
:002; k
18
:02
Repair rates l
10
2:85; l
11
1:4; l
12
2:94; l
13
1:08; l
14
1:95; l
15
1; l
16
1:35; l
17
1:21; l
18
1:47
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9158152333 .9126527273 .9154712209
10 .9157157156 .9125423877 .9153702911
15 .9157153901 .91254204 .9153699627
20 .9157153888 .9125420386 .9153699615
25 .9157153888 .9125420386 .9153699615
30 .9157153888 .9125420386 .9153699615
Table 3
Availability for idealized, faulty and no preventive maintenance for the month of June.
Month June
Failure rates k
10
:0072; k
11
:02; k
12
:002; k
13
:002; k
14
:0054; k
15
:0001; k
16
:002; k
17
:006; k
18
:01
Repair rates l
10
1:56; l
11
1:2; l
12
:85; l
13
2:7; l
14
3:7; l
15
1; l
16
1:2; l
17
2:18; l
18
1:41
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9641303277 .9599775701 .9639762837
10 .9640608375 .9598559925 .9639057393
15 .9640603732 .9598549551 .9639052703
20 .9640603646 .9598549419 .9639052647
25 .9640603675 .9598549148 .9639052647
30 .9640603675 .9598549148 .9639052647
Table 4
Availability for idealized, faulty and no preventive maintenance for the month of July.
Month July
Failure rates k
10
:022; k
11
:007; k
12
:011; k
13
:014; k
14
:007; k
15
:0001; k
16
:005; k
17
:001; k
18
:02
Repair rates l
10
2:17; l
11
1:96; l
12
2:1; l
13
1:45; l
14
1:08; l
15
1; l
16
1:56; l
17
2:86; l
18
1
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9448803139 .941092834 .9446482711
10 .9447539745 .9409450755 .9445203506
15 .9447532998 .9409443374 .9445196696
20 .944753296 .9409443334 .9445196657
25 .9447532959 .9409443334 .9445196656
30 .9447532959 .9409443334 .9445196656
1678 S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683

18
i1
k
i
_ _
P
0

9
i1
b
i
l
i
P
i

18
j10
l
j
P
j
Table 5
Availability for idealized, faulty and no preventive maintenance for the month of August.
Month August
Failure rates k
10
:0076; k
11
:014; k
12
:005; k
13
:012; k
14
:0053; k
15
:0001; k
16
:0026; k
17
:0015; k
18
:006
Repair rates l
10
:76; l
11
:85; l
12
1:47; l
13
:93; l
14
:7; l
15
1; l
16
1:09; l
17
1; l
18
1:35
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9453473868 .9389864199 .9451186269
10 .94473096 .9382182443 .9444972614
15 .9447204977 .938205274 .944486737
20 .9447202796 .9382049745 .9444865178
25 .9447202744 .938204968 .9444865126
30 .9447202743 .9382049678 .9444865125
Table 6
Availability for idealized, faulty and no preventive maintenance for the month of September.
Month September
Failure rates k
10
:011; k
11
:017; k
12
:0154; k
13
:0144; k
14
:009; k
15
:0001; k
16
:0015; k
17
0; k
18
:0046
Repair rates l
10
1:23; l
11
1:04; l
12
1:75; l
13
1:5; l
14
:44; l
15
1; l
16
1:7; l
17
0; l
18
2:3
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9392934535 .932100768 .9390415136
10 .9375555235 .9280510305 .9372936771
15 .9373799864 .9258668077 .9371172668
20 .9373613692 .923871764 .9370985619
25 .9373593908 .9219003709 .9370965747
30 .9373591806 .9199352231 .9370963635
Table 7
Availability for idealized, faulty and no preventive maintenance for the month of October.
Month October
Failure rates k
10
:02; k
11
:02; k
12
:004; k
13
:0044; k
14
:0053; k
15
:0001; k
16
:003; k
17
:0006; k
18
:0026
Repair rates l
10
2:13; l
11
1:35; l
12
:45; l
13
1:59; l
14
3:3; l
15
1; l
16
1:3; l
17
3:3; l
18
:9
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9598672551 .9547899559 .9596966498
10 .9590845943 .9536542682 .9589090558
15 .9590080231 .9535441177 .9588320587
20 .9590001337 .9535328712 .9588241264
25 .958999318 .9535317189 .9588233065
30 .9589992337 .9535316008 .9588232217
Table 8
Availability for idealized, faulty and no preventive maintenance for the month of November.
Month November
Failure rates k
10
:011; k
11
:01; k
12
:013; k
13
:021; k
14
:012; k
15
:0001; k
16
:0064; k
17
0; k
18
:012
Repair rates l
10
:85; l
11
:7; l
12
2:9; l
13
1:3; l
14
:84; l
15
1; l
16
1:89; l
17
0; l
18
1:7
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9331347848 .9264621922 .9328578276
10 .9325258605 .9237912216 .932244252
15 .9325140232 .9218178779 .9322323477
20 .9325137485 .9198614526 .9322320716
25 .9325137414 .9179094637 .9322320645
30 .9325137412 .9159616243 .9322320643
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1679
l
i
P
i
k
i
P
0
; i 1; 2; . . . ; 9
l
j
P
j
k
j
P
0
1 b
j9
l
j9
P
j9
; j 10; 11; . . . ; 18
Table 9
Availability for idealized, faulty and no preventive maintenance for the month of December.
Month December
Failure rates k
10
:013; k
11
:013; k
12
:007; k
13
:019; k
14
:0033; k
15
:0001; k
16
:0046; k
17
:002; k
18
:006
Repair rates l
10
1:5; l
11
:84; l
12
:65; l
13
1:39; l
14
2:9; l
15
1; l
16
1:05; l
17
1:79; l
18
2:44
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9461497579 .9413019395 .945923732
10 .9456241529 .9406552618 .9453940231
15 .9456101056 .9406377679 .9453798899
20 .9456096441 .9406371904 .9453794258
25 .9456096277 .9406371699 .9453794093
30 .945609627 .9406371691 .9453794087
Table 10
Availability for idealized, faulty and no preventive maintenance for the month of January.
Month January
Failure rates k
10
:0031; k
11
:003; k
12
:0023; k
13
:006; k
14
:008; k
15
:0001; k
16
:004; k
17
:003; k
18
:0094
Repair rates l
10
:95; l
11
2:27; l
12
2:5; l
13
3:45; l
14
:8; l
15
1; l
16
2:1; l
17
1:24; l
18
1:2
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9716723028 .9668865948 .9715502258
10 .9714711406 .9666022482 .9713470817
15 .9714681974 .9665983243 .971344114
20 .9714681484 .9665982617 .9713440646
25 .9714681475 .9665982607 .9713440638
30 .9714681475 .9665982606 .9713440638
Table 11
Availability for idealized, faulty and no preventive maintenance for the month of February.
Month February
Failure rates k
10
:01; k
11
:0067; k
12
:007; k
13
:0054; k
14
:006; k
15
:0001; k
16
:003; k
17
:0011; k
18
:0053
Repair rates l
10
2:2; l
11
1:5; l
12
1:05; l
13
1:5; l
14
:4; l
15
1; l
16
:98; l
17
2:1; l
18
1:64
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9624541906 .956488515 .9622953869
10 .9608401131 .9544140331 .960671954
15 .960634842 .9541540715 .9604655925
20 .9606078859 .9541202320 .9604384981
25 .9606043414 .9541158205 .960434936
30 .9606038753 .9541152453 .9604344677
Table 12
Availability for idealized, faulty and no preventive maintenance for the month of March.
Month March
Failure rates k
10
:0055; k
11
:013; k
12
:0051; k
13
:02; k
14
:01; k
15
:0001; k
16
0; k
17
:001; k
18
:01
Repair rates l
10
1:09; l
11
:88; l
12
1:47; l
13
1:45; l
14
1:6; l
15
1; l
16
0; l
17
2:13; l
18
2:94
Time
(In days)
Availability
AV
I
AV
F
AV
C
5 .9551979616 .9500524019 .9550075664
10 .955028451 .9489496636 .9548362029
15 .9550267636 .9480500369 .9548345014
20 .9550267448 .9471530971 .9548344824
25 .9550267446 .9462570256 .9548344822
30 .9550267445 .9453618021 .9713440638
1680 S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683
Solving recursively the above system of equations and using the normalizing condition

18
i0
P
i
1; we get
P
0

_
1

27
s1
N
s
_
1
where
N
k

k
k
l
k
; k 1; 2; . . . ; 18
N
s

1 b
s18
k
s18
l
s9
; s 19; 20; . . . ; 27 20
The expression for the availability of the system, in this case, is obtained as
AV
1

9
i1
N
i
1

27
s1
N
s
21
The expressions for steady-state availability under different eld conditions are obtained as
(i) The long run availability under idealized preventive maintenance (AV
(I)
) obtained from Eq. (21) is
AV
I

1

9
i1
N
i
1

18
K1
N
k
22
(ii) The long run availability under faulty preventive maintenance (AV
(F)
) obtained from Eq. (21) is
AV
F

1

9
i1
N
i
1

18
K1
N
K

9
i1
k
i
l
i9
23
(iii) The long run availability (AV
(C)
) of the system in case only corrective repair is performed obtained from Eq. (21) is
AV
C

1
1

9
i1
D
i
k
i
l
i9
24
where D
i
(i = 1, 2, . . . , 9) is a deterioration factor which is introduced to study the effect of deterioration of the system.
It is the ratio of transition rate which transits the system to failed state from normal state to the transition rate which
transits the system to pending-failed state from normal state.
6.2.1. Numerical analysis
Numerical results of long run availability AV
(C)
corresponding to deterioration factor D
i
are obtained by making use of
software MATLAB 5.1and have been given in Tables 1315 with the case of deteriorating system using the expression (24).
Data taken for CM pertaining to the system is as follows:
Table 13
Effect of deterioration factor of module-1 m/c (D
1
) and module-3 m/c (D
5
) on availability (AV
(C)
) of the system.
D
5
/D
1
7.26 8 10 12 14 15
6.3 .9567 .9566 .9546 .9530 .9514 .9506
8 .9558 .9557 .9536 .9521 .9505 .9497
11 .9534 .9528 .9507 .9491 .9476 .9468
14 .9504 .9503 .9483 .9467 .9452 .9444
16 .9488 .9487 .9467 .9451 .9436 .9428
Table 14
Effect of deterioration factor of widma-1 m/c (D
2
) and module-2 m/c (D
3
) on availability (AV
(C)
) of the system.
D
3
/D
2
15.8 17 18.5 20 21.5 23
7.26 .9567 .9561 .9554 .9546 .9539 .9532
8.5 .9557 .9551 .9544 .9536 .9529 .9521
10 .9545 .9539 .9531 .9524 .9517 .9509
11.5 .9533 .9526 .9519 .9512 .9504 .9497
13 .9520 .9514 .9507 .9499 .9492 .9485
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1681
k
10
k
12
:0098; k
11
:011; k
13
:013; k
14
:0085; k
15
:0001; k
16
:0048; k
17
:002; k
18
:008
l
10
l
13
1:55; l
11
1:3; l
12
l
14
l
16
l
18
1:5; l
15
1; l
17
1:65
Tables 1315 exhibiting system behavior associated with a wear out mode. The closer look at the numerical results shows
that the availability of the system decreases with increasing deterioration factor for a given repair rate. So, to avoid the sys-
tem to reach to a pending failed state, the preventive maintenance should be started without delay. This will result in en-
hanced availability of the system.
The preventive maintenance carried out on degradation of the system is not always ideal, but can be faulty. So, the impact
of idealized and faulty preventive maintenance activities carried out on the availability of the system is studied.
(i) In case of idealized preventive maintenance, i.e., when b
i
= 1, (i = 1, 2 , . . . , 9), then the availability of the system comes
out to be
AV
I
:9569
(ii) In case of faulty preventive maintenance, i.e., when b
i
= 0, (i = 1, 2 , . . . , 9), then the availability of the systemcomes out
to be
AV
F
:9528
(iii) In case of no preventive maintenance, i.e., when k
i
l
i
0, (i = 1, 2 , . . . , 9), then the availability of the system comes
out to be
Table 16
Effect of corrective repair rates of module-1 m/c (l
10
) and widma-1 m/c (l
11
) on availability (AV
F
) of the system.
l
11
/l
10
.5 1 1.55 2 2.5 3
.5 .9268 .9365 .9399 .9414 .9423 .9430
1 .9369 .9468 .9503 .9518 .9528 .9535
1.5 .9404 .9503 .9539 .9553 .9563 .9570
2 .9421 .9520 .9556 .9571 .9581 .9588
2.5 .9421 .9531 .9567 .9582 .9592 .9599
3 .9438 .9538 .9574 .9589 .9599 .9606
Table 15
Effect of deterioration factor of ne boring m/c (D
6
) and tapping-2 m/c (D
9
) on availability (AV
(C)
)of the system.
D
6
/D
9
10 15 20 25 30 40
.5 .9588 .9574 .9560 .9546 .9532 .9504
1 .9587 .9573 .9559 .9545 .9531 .9503
2 .9585 .9571 .9557 .9543 .9529 .9502
4 .9582 .9568 .9554 .9540 .9526 .9498
6 .9578 .9564 .9550 .9536 .9522 .9494
Table 17
Effect of corrective repair rates of module-2 m/c (l
12
) and module-3 m/c (l
14
) on availability (AV
(F)
) of the system.
l
14
/l
12
.5 1 1.5 2 2.5 3
.5 .9281 .9378 .9410 .9427 .9437 .9443
1 .9366 .9465 .9498 .9515 .9525 .9532
1.5 .9395 .9494 .9528 .9544 .9555 .9561
2 .9410 .9509 .9543 .9559 .9570 .9576
2.5 .9418 .9518 .9551 .9568 .9579 .9585
3 .9424 .9524 .9557 .9574 .9585 .9591
Table 18
Effect of corrective repair rates of ne boring m/c (l
15
) and tapping-2 m/c (l
18
) on availability (AV
(F)
) of the system.
l
15
/l
18
.5 1 1.5 2 2.5 3
.05 .9377 .9451 .9476 .9489 .9497 .9502
.08 .9396 .9471 .9497 .9509 .9517 .9522
.1 .9403 .9478 .9503 .9516 .9524 .9529
.5 .9424 .9500 .9525 .9538 .9545 .9550
1 .9427 .9502 .9528 .9540 .9548 .9553
1.5 .9428 .9503 .9529 .9541 .9549 .9554
1682 S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683
AV
C
:9567
From the above results, it is observed that availability AV
(I)
is higher than AV
(F)
and AV
(C)
, having same failure and re-
pair rates in steady state also. Furthermore, the effect of corrective repair rates of the constituent components of the
system is studied using the relation (23). It is found out from the numerical results that corrective repair rates of M
1
,
W
2
and M
3
signicantly affect the availability of the system. The effect of corrective repair rate on availability, when
PM is faulty, has been depicted in Table 1618.
Tables 1618 also explore that an increase in corrective maintenance rates l
10
and l
11
also increases the availability of
the system to 96%, in case PM is faulty.
7. Finding and suggestions
The availability of three-state system model is higher than the two-state system model in both time dependent and stea-
dy state system, having same failure and repair rates. This necessitates applying PM to the system on reaching to pending-
failed state to enhance its availability.
The availability of the system decreases with increasing deteriorating factor for a given repair rate, exhibiting systems
behavior associated with a wear out mode. So the preventive maintenance should be initiated as soon as possible the system
reaches the pending-failed state.
The preventive maintenance carried out on the degradation of the system is not always ideal, but can be faulty too. Com-
parative studies of the idealized and faulty preventive maintenance show that the system availability is higher in the former
case in time dependent as well as steady state system. So, for performing idealized preventive maintenance, the maintenance
department should consider the following:
(a) A proper maintenance scheduling must be designed and there is free accessibility of the data and information at every
point.
(b) Skilled workmen must be hired and they must have timely access to the spares and other maintenance materials.
(c) Sufcient repair facilities must be available.
Furthermore, an in-depth study and comparison of effect of corrective repair rates of the constituent components of the
system can help in achieving the higher production targets.
8. Conclusion
It is concluded that the availability of the system can be improved considerably by controlling deterioration of the system
using proper maintenance planning and scheduling. Following the above ndings and the results of numerical analysis car-
ried out in the study in different eld conditions, the management can derive cost cutting plans and increased productivity.
Acknowledgements
The author thanks the referees for valuable comments for improvement of the paper.
References
[1] R. Barlow, L. Hunter, Optimum preventive maintenance policies, Operational Research 8 (1960) 90100.
[2] I.P. Singh, A complex system having four types of components with pre-emptive repeat priority repairs, Microelectronics Reliability 29 (6) (1989) 959
962.
[3] Y. Zhang, Reliability analysis of an (N + 1)-unit standby system with preemptive priority rule, Microelectronics Reliability 36 (1) (1996) 1926.
[4] O.P. Gandhi, M.F. Wani, Development of maintainability index for mechanical systems, Reliability Engineering and System Safety 65 (1999) 259270.
[5] A. Grall, L. Dieulle, C. Berenguer, M. Roussignal, Continuous-time predictive-maintenance scheduling for a deteriorating system, IEEE Transactions on
Reliability 51 (2) (2002) 141150.
[6] D. Kumar, Availability of the washing system in paper industries, Microelectronics Reliability 29 (5) (1988).
[7] J. Singh, Reliability analysis of a biogas plant having two dissimilar units, Microelectronics Reliability 29 (5) (1989) 779781.
[8] P. Gupta, J. Singh, I.P. Singh, Behavioral analysis of duplex system under preemptive resume priority repair discipline a case study, Journal of Industrial
Engineering 33 (12) (2004) 413.
[9] S. Garg, J. Singh, Availability analysis of core veneer manufacturing system in plywood industry, Proceedings of International Conference on Reliability
and Safety Engineering (2005) 497508.
S. Garg et al. / Applied Mathematical Modelling 34 (2010) 16721683 1683