Professional Documents
Culture Documents
USING
mmm
CHAPTER 4
USING RELIABILITY
4.1 INTRODUCTION
In earlier chapters we dealt with quality assurance and control and explored the notion or -
quality during the manufacture of the product or the rendering of a service. Our discussion
focused on a specific instant in time. In this chapter we examine the concept of reliability.
Reliability is a measure of the quality of the product over the long run. In here the concept
of reliability is an extended time period over with the expected operation of the product is
considered; that is we expect the product will function according to certain expectations
over a stipulated period of time.
Reliability is the probability of product performing is intended function for a stated period of
time under certain specified conditions. Four aspects of reliability are apparent from this
definition. First, reliability is a probability-related concept; the numerical value of this
probability is between 0 and 1. Second, the functional performance of the product has to
meet certain stipulations. Product design will usually ensure development of a product that
meets or exceeds the stipulated requirements. For example, .if the breaking strength of a
nylon cord is expected to be 1000 kg, then in its operational phase, the cord must be able to
bear weights of 1000kg. Or more. Third, reliability implies successful operation over a
certain period of time. Although no product is expected to last forever, the time
requirement ensures satisfactory performance over at least a minimal stated period (say, 2
years). Fourth, operating or environmental conditions under which product use takes place
are specified. Thus, if the nylon cord is specified to be used under dry conditions indoors,
then satisfactory performance is expected for use under those conditions. In the context of
these four aspects, the reliability of the nylon cord might be described as having a
probability of successful performance of .92 in bearing loads of 1000 kg for 2 years under
dry conditions.
Most products go through three distinct phase from product inception to wear-out. Figure
4.1 shows a typical life-cycle Curve for which the failure rate A is plotted as a function of
time. This curve is often referred to as the bathtub curve; it consists of the debugging phase,
th e c h a n c e -fa ilu re ph ase a n d th e w e a r -o u t p h a se (B e s te r fie ld 1 9 9 0 ; D e lM a r an d S h e ld o n
1 9 8 8 ; H a lp e rn 1 9 7 8 ). T h e d e b u g g in g ph ase, also k n o w n as th e in fa n t-m o r ta lity p h a se ,
exhibits a d ro p in th e fa ilu re ra te as in itia l p ro b le m s id e n tifie d d u rin g p ro to ty p e te s tin g a re
iro n e d o u t. T h e c h a n c e -fa ilu re ph ase, b e tw e e n tim e s t i and X.2 is th e n e n c o u n te re d ; fa ilu re s
o ccur ra n d o m ly an d in d e p e n d e n tly . This p h a se , in w h ic h th e fa ilu re ra te is c o n s ta n t,
typ ically re p re s e n ts th e usefu l life o f th e p ro d u c t. F o llo w in g th is is th e w e a r -o u t p h ase, in
w h ich an in cre a se in th e fa ilu re ra te is o b s e rv e d . H e re , a t th e e n d o f th e ir usefu l life , p a rts
ag e an d w e a r o u t
F{t) = X .r Mt > 0
W h e r e A d e n o te s th e fa ilu re ra te
M TIF = VX
1 -1 dt At
Jo
Where F{t) represents the cumulative distribution function at time t. The cumulative
distribution function for the exponential distribution is shown in Figure 4.25 Figure 4 2
shows the reliability function. R(t) for the exponential failure distribution. At time 0, the
reliability is 1, as it should be. Reliability decreases exponentially with time. In general, the
failure-rate function r(t) is given by the ratio of the time-to-failure probability density
function to the reliability function. We have,
r(t) m
R{t)
For the exponential failure distribution
Ae
At
r(t):
-At
implying a constant failure rate, as mentioned earlier.
Example 4.1: An amplifier has an exponential time-to-failure distribution with a failure rate
of 8% per 1000 h. What is the reliability of the amplifier at 5000 h? Find the mean time to
failure.
Solution. The constant failure rate A is obtained as
A= 0.08/1000 h
A = 0.00008/h
The reliability at 5000 h is
-Xt
R(t)
(0 0 0 0 8 X 5 0 0 0 )
=e
_ -04
0.5703
89
The mean time to failure is
Weibull Distribution
The Weibull distribution is used to model the time to failure of products that have a varying
failure rate. It is therefore a candidate to model the debugging phase (failure rate decreases
with time), or the wear-out phase (failure rate increases with time) (Henley and Kumamoto
1981). The Weibull distribution was introduced in Chapter 3. It is a three-parameter
distribution whose density function is given in eq.(4.43) as
/<0= £
'w '
U i - y ^ exp
PS
!
t> y
a l « 2 .1 al
The parameters are a location parameter, y ( - < y < ), a scale parameter, a ( a - 0) and a
shape parameter P (P> 0). The probability density functions for y = 0, a - 1, and several
values o f P (P = 0.5,1,2 and 4) are shown in figure.
Appropriate parameters are used to model a wide variety o f situations. If y = 0 and P=1 the
weibull distribution reduces to the exponential distribution. For reliability modeling the
location parameter y will be zero. If a = 1 and p = 0.5 fo r example, the failure rate decreases
with time and can therefore be used to model components in the debugging phase.
Likewise, if a = 1 and p =3.5. the failure rate increases with time and so can be used to
model products in the wear-out phase In this case, the Weibull distribution approximates
the normal distribution.
The reliability function for the Weibull distribution is given by
ia
1
R(t) = exp -
a
MTTF —a -+ 1
P
The failure-rate function r(t) for the Weibull tim e-to-failure probability distribution is
f(t) fit*-'
K0-
R(t) a/
Figure 4.3 shows the shape o f the failure-rate function for the Weibull failure density
function, for values of the parameter P o f 0.5,1 and 3.5.
90
(3 = 3 5
P=1
Time (t)
Example 4.2 Capacitors in an electrical circuit have a time -to-faiiure distribution that case
be modeled by the Weibull distribution with a scale parameter of 400 h and a shape
parameter of 0.2. What is the reliability of the capacitor after 600 h of operation? Find the
mean time to failure. Is the failure rate increasing of decreasing with time?
Solution: The parameters of the weibull distribution are a - 400 h and (3 = 0.2. The location
parameter y is 0 for such reliability problems. The reliability after 600 h of operation is given
by
R(t) = exp -
a
=. 3381
= 48000 h
91
,. 0.2/°2-1
r ^ " (4 0 0)°2
= 0.0603t“ ° 8
This function decreases with time it would model components in the debugging phase.
Most products are made up of a number of components. The reliability of each component
and. the configuration of the system consisting of these components determines' the
system reliability (that is, the reliability of the product) Although product
design, manufacture, and maintenance influence reliability, improving reliability is largely
the domain of design. One common approach for increasing the reliability of the system is
through redundancy in design, which is usually achieved by placing components in parallel.
As long as one component operates the system operates. In this section we demonstrate
how to compute system reliability for systems that have components in series, in parallel, or
both.
Rs = Rix R2 X ... x Rn
The system reliability decreases as the number of components in series increases. Although
over design in each component improves reliability, its impact would be offset by the
number of components in series. Moreover, manufacturing capabilities and resource
limitations restrict the maximum reliability of any given component. Product redesign that
reduces the number of components in series is a viable alternative.
Suppose the system has n components in series, each with exponentially distributed time-
to-failure with failure rate (Xi, X2 ... Xn ), The system reliability is found as the product of the
component reliabilities
92
e q u iv a le n t fa ilu re ra te o f I = 1 A ,. Thus, if eac h c o m p o n e n t th a t fails is re p la c ed im m e d ia te ly
w ith a n o th e r th a t has th e s a m e fa ilu re ra te , th e m e a n tim e to fa ilu re fo r th e
system is given by
MTTF - — — -
2>
,=i
A = .0 5 /4 0 0 0 h
= 1 2 .5 - x 1 0 ~6/h
R = e x p [-(1 2 .5 x 10"6)2 0 0 0 ]
= .9 7 5
R = e x p [-( 1 0 x 10"6)2 0 0 0 ]
= .7 7 9
M T T F = 1 /( 1 0 x 1 2 5 x 1 0 ^ )
= 8000 h
E x a m p le 4 .4 : R e fe r t o E xa m p le 4 .3 c o n c e rn in g th e a u to m a tic focus o f a te le v is io n c a m e ra ,
w h ic h has 1 0 s im ila r c o m p o n e n ts in series. It is d e sire d fo r th e fo c u s u n it to h a v e a re lia b ility
o f .9 5 a fte r 2 0 0 0 h o f o p e ra tio n . W h a t w o u ld b e th e m e a n t im e to fa ilu re o f th e in d iv id u a l
c o m p o n e n ts ?
93
.95 = exp[ -As (2000)]
or
A s = 0.0000256/h
As =AS/1 0
= 0 00000256/h
= 2.56/106h
MTTF = 1 / A
= 390,625h
Suppose we have n components in parallel, with the reliability of the ith component
denoted by R, , I - 1, 2......., n. Assuming that the components operate randomly and
independently of each other, the probability of failure of each component is given by F, —1.
- R, Now, the system fails only if all the components fail. Thus, the probability of system
failure is
Rs= 1 - Fs
= 1 - n (1- R.)
94
Rs = 1 - n (1- R,)
= 1 - it 1 - e*'
In the special case where all components have the same failure rate A, the system reliability
is given by
Rs= 1 - (1 - e -Kt )n
For this situation, the mean time to failure for the system with n identical components in
parallel, assuming that each failed component is immediately replaced by an identical
component, is given by
Example 4.5: For the system shown in Figure 4.5, determine the system reliability for 2000 h
of operation, and find the mean time to failure. Assume that all three components have an
identical time-to-failure distribution that is exponential with a constant failure rate of
0.0005/h. What is the mean time to failure of each component? If it is desired for the
system to have a mean time to failure of 4000 h. What should the mean time to failure be
for each component?
Solution: The failure rate of each component is A = 0.0005/h. For 2000 h of operation, the
system reliability is
Rs = 1 - {1 - exp[-(0.0005)2000]}3
= 1 - ( 63212)3
=0.7474
MTTF = — ?— (1 + 1 /2 + 1 /3 )
0.0005
= 3666.67 h
MTTF = — = — -—
1 0.0005
= 2000 h
By placing three identical components in parallel, the system MTTF has been increased by
about 83 3%.
For a desired system MTTF of 4000 h. we now calculate the required MTTF of the individual
95
components. We have
4000 = — (1 + 1/ 2 + 1 /3 )
X
Where A. is the failure rate for each component Solving A ,we get
A = 1.8333
4000
= 0.00046/ h
Complex systems often consist of components that are both in series and in parallel.
Reliability calculations are based on the previously discussed concepts, assuming that the
components operate independently.
Example 4.6 : Find the reliability of the eight-component system shown in Figure 4.6; some
components are in series and some are in parallel The reliabilities of the components are as
follows: Rai=.92, Ra2=.90, Ra3 —.88, Ra4=.96, Rbi=.95, Rb2=-90, Rb3=.92 and Rci=.93.
Solution; We first find the reliabilities of each subsystem For the subsystem with
components Ai, A2, A3, and A4 the reliability is
Ri = 1 - ( 1 -R a i M ( 1 - R asRa4)
= 1 - [1 - (92)(,90)][1 - ( 88)(.96)]
= 9733
Similarly, the reliability of the subsystem with components Bi, B2, and B3 is
Rs= Ri X R2 X R3
= (.9733)(.9996)(.93)
= .9048
96
Exam ple 4 .7 : Find th e system failure rate and th e m ean tim e to failu re fo r th e eight
com ponent system shown in Figure 4 6 The fa ilu re rates (n u m b er o f units per hour) fo r th e
com ponents are as follows XAi= 0 0006, XA2 = 0 0 04 5 , Aa3 = 0 .0 0 3 5 , AA4= 0 .0 0 1 6 , XBi - 0 .0 0 60 ,
AB2 = 0 .0 0 6 0 , AB3 = 0 .0 0 6 0 ,.. Aa = 0 .0 0 50 .
S olution: First w e com pute failure rates fo r th e subsystems. W e have a failure rate o f (AAi +
AA2) = 0 .0 0 5 1 fo r th e A 1 /A 2 subsystem; fo r th e A 3 /A 4 subsystem, th e failu re rate is (AA3 +
Am ) = 0 .0 0 51 . T he m ean tim e to failu re fo r th e A 1/ A 2/A 3/A 4 subsystem is
1 \
MTTF1 1+ -
0.0051 .
2 94 .11 8 h
1 ' 1 1
MTTF .L !+- +■
0.0060 v 2 3)
3 0 5 .5 5 5 h
= 0 .0 1 1 6 7
MTTFs = 8 5 .6 9 h
97
system reliability is higher than for comparable systems with components in parallel. In
parallel systems discussed previously, all components are assumed to operate
simultaneously: Figure 4.7 shows at standby system with a basic component and two
standby components in parallel. Typically a failure-sensing mechanism triggers the
operation of a standby component when the currently operating component fails.
x!
For a system that has a basic component in parallel with one standby component, the
system will be operational at time t, as long as there is no more than one failure. Therefore,
the system reliability would be
R = e Xt + e Xt (At )
For a standby system with a basic component and two standby components (shown in
Figure 4.7), the system will be operational if the number of failures is less than or equal to 2.
The system reliability is
Basic
Component
In general, if there are n components on standby along with the basic component for a total
of (n + 1) components in the system] the system reliability is given by
98
f n
At (A, ) 2 (A /
R =e 1+ A H---------- + ■+ ....+
s t 2! 3! nl
V
MT7F^!±1
A
We have discussed OC curves for acceptance sampling plans in previous chapters. In this
section we discuss OC curves for life and reliability testing plans. A common life testing plan
involves choosing a sample of items from the batch and observing their operation for a
certain predetermined time. If the number of failure exceeds a stipulated acceptance
number, the lot is rejected; if number of failures is less than or equal to the acceptance
number, the lot is accepted. Two options are possible. In the first option, an item that fails is
replaced immediately by an identical item. In the second, failed items are not replaced.
In the calculations for OC curves, we assume that the time to failure of each item is
exponentially distributed with a constant failure rate A. The parameters of a life testing plan
are the test time T, the sample size n, and the acceptance number c. The OC curve shows
the probability of lot acceptance, P a, as a function of the lot quality as indicated by the
mean life { 0 ) or the mean time to failure of the item. Under the stated assumptions, the
number of failures within a specified period adheres to the Poisson distribution. The Poisson
distribution is used to calculate the probability of lot acceptance.
APPLICATION OF RELIABILITY
99
This study intends to present in detail the various factors as,mentioned earlier involved in
the Reliability and Maintainability of Telecommunication equipments. It is well-known that
reliability is the probability that a system or equipment or component will perform the
required functions under specified conditions for a specified length of time without
degradation of failure. A lot of thinking has gone into the subject of reliability and
considerable work also done resulting in various specifications and standards such as the
Mentally Specification and Standard (MIL). International! Electrochemical Commission
Standard (IEC Standard) and the Japan Industrial Standard (JIS). These standards are useful
in assessing the reliability of the equipments/systems by performing reliability
demonstration tests.
RELIABILITY
co
R(t)= epl(t)dt
o
Figure 1 shows how the failure rate A(t ) varies with time for an electronic equipment. As
can be seen in the figure the failure rate is divided into three general patterns.
-S'
co
<
uL>
13
CO
u_
TIME (t)
This is called the bathtub curve. The failure rate decrease sharply during the early stages of
service and then enters the so called "constant failure intensity period", during which
failures occur only occasionally.
O P T IM U M RELIABILITY
The reliability requirements are very high in such applications where repair and
maintenance are not all possible e.g. space applications. But in the field of
telecommunications where repair and maintenance activities are possible the reliability
requirements can be made optimum with the costs involved in the design, production and
100
maintenance of any electronic equipment. Fig. 2 shows such an optimum condition.
Maintenance
and other
COSTS
cost
RELIABILITY------------------------------------
As it can see design and production costs increases with increase in reliability of the
equipment while the maintenance deceases. An optimum figure for reliability can be
obtained as in Fig. 2.
The more useful definitions connected with telecommunication applications are the
availability and maintainability, The reliability of a system is, of course, totally dependent on
the reliability of the constituent parts, The extent to which we can depend on system is
enhanced when (i) each component is failure free and (ii) it is promptly repaired when it
encounters any failure. The extent to which a system 'can be used as required is called its
"Availability". The quantification for the system unavailability is made in terms of equipment
reliability and that for the repair, in terms of MTTR = Mean Time To Repair.
A = 1 - UA
When a system fails completely, it is not available totally for use. Flowever, a definition is
required to determine the degree of quality degradation that constitutes unavailability. For
example, in digital systems (as per CCITT recommendation) a period of unavailability begins
when the bit error ratio in each second is more than 1 x 10 Exp(-3) for a period of 1-
consecutive second, These ten seconds constitute the unavailable time, The period of
unavailability terminates once this condition is improved,
For system planning, we do not require such a strict definition, At this stage value for
unavailability is obtained from the system failure,
UA = MTTR
MTBF + MTTR
This is the ratio of the mean repair time (or mean unavailable time) to the duty time, When
the value of MTBF is large and that of MTTR is small, the availability will increase as
Availability = 1 minus Unavailability,
101
MTTR is given in hours. This is a function of maintenance efficiency or the complicity of the
fault To reduce MTTR an effective and efficient maintenance support must be available, A
mathematical relationship can be given relating duty time (t), time allowed for maintenance
(T), Reliability R(t), Maintainability M(t), and Availability A(T, t) as follows
Since for telecom applications availability of the equipment is the equipment is the mam
criteria, even with the equipment having reliability of the order of 0.9, availability of the
equipment can be achieved to the higher order by improving the maintenance techniques
In industrial sector it is always desired that each equipment must be operative for long time
to give maximum output and hence profit to the industry. Under given operative conditions
the behavior analysis of each equipment may help in improving the system design and
maintenance planning.
In paper industry, we have various systems, viz., chipping, feeding, pulp formation, washing,
bleaching, screening, paper formation and collection. The important functionary part, which
produces brightness in paper, is the bleaching system. In paper industry, chips from storage
are fed into digester [1,2,3] to form the pulp, which is processed through washing system
[4,5] The washed pulp is kept in a chamber where chlorine, at controlled rate, is passed
through the pulp for few hours. The pulp is passed over filter over filter and opener in two
stages to get chlorine free white-pulp. The pulp so obtained is sent to the screening section
for final section for final cleaning of the pulp. This paper discusses the behavior of the
various equipment's used in bleaching system.
SYSTEM DESCRIPTION
The bleaching system has four subsystems xi , yi (i = 1,2, j = 1,2 of the same nature and
capacity)
The subsystem (Xi) called filter consists of two units and is said to be in failed state (due to
fall in quality) when one unit fails;
The subsystem (Yi) called opener, consists of two units and is said to be in failed state (due
to fall in quality) when one unit fails.
102
Xi, Yj , 1= 1,2 j = 1,2 : Capital letters indicates good state of units while small letters indicate
the failed state.
Dash on P represents the first derivative of the probabilities. Limit of integration is taken
from 0 to ° ° .
Transition diagram for the system is given as:
_0
Pi(Zj,t) =0
d. + aT+^
P0 (0) = 1 otherwise = 0 ;
Pi (0,t) = a :,P0 (t) (i = 1,2; i = 2, ,j = 3,4) ........... 2
Integrating the above system of equations under given conditions the solutions, hence
reliability is given by
R(t) = Po (t)
103
2(al + a l ) t N{t)e 2 ( « i +£Z 2 > +1 3
W h ere
x.
(j = 1/2 ; i = 2, ,j = 3 ,4 , i =2)
PA R TIC U LA R CASE
The o p erative conditions fo r a particular industry are planned and repair facilities are
produced depending upon w o rkers availability and skillness ie., constant repair fo r each
eq u ip m en t, th e n
R(t) o f th e system is given by
In process industry each e q u ip m en t m ust rem ain o p erative fo r a long tim e , hence steady
<2A J l 2A j
and m ean tim e to system failure is given by
MTSF - +2«2
Example
Taking P i = p2 = 0.2
104
Table 1
Effect o f Failure Rate
Table 2
Effect o f Repair Rate
(Taking O i=0.03, a 2=,02)
Pi p2= o .i p 2 = 0.3 P z= 0 .5
Table 1 and 2 show th a t fa ilu re and re p a ir rate o f X, has m ore e ffe ct upon system a va ila b ility
th a n Y, fa ilu re and re p a ir rates, w h ile MTSF is affe cted equally, i e. X, requires m ore
a tte n tio n th a n Y, keeping in v ie w cost constraints, re p a ir is arranged fo r m in im u m fa ilu re in
th e system .
COST FUNCTION
R (t)^ G Q + G l e
-<V1 + G 2 e - V2 - ............5
s. ± s 2 —4 s 2
0,2 = - L -
~2
P1P2
G0 =
2
E i - P i + P2 + 2 ( a i + 0 2 ) ;
e2 - P1P2+ 2 oiP2 + 2 02 p i ;
105
q _ + (P\ + 02W\ + P\ A
0 W , - ^ 2)
£ __ 6 \ + ( P \ + P i )® 2 + A A?
0 ACA^,)
If service facility is always available, it rem ains busy fo r t i m e 't ' during th e interval (o ,t). Let
Cx and C 2 are revenue per unit tim e and service cost per unit tim e respectively, th e n th e
profit function H(t) fo r interval (0 ,t) is given by
dH(t) n
Putting ----------- = 0 w e get
dt
{n e l + d2)
log P l G2<Cl G0 “ C2> i
a
d 2H{t)
Putting gives positive f o r 't' i.e. H (t) is m in im u m f o r 't ' given by (7)
Dt2
Table 3
Profit Function
(Taking a j= 03, a 2= 0 2 3 x= p2 = -2)
t R(t) G (t)
0 .0 1.0 0.1
2 0 .0 0 .6 6 7 13.23
4 0 .0 0 .6 6 6 7 26.57
6 0 .0 666 6 7 3 9 .9 0
8 0 .0 .66 6 66 5 3.2 3
1 00 .0 .66 6 66 6 6.5 7
For given values o f failure and re p air rates, Table 3 gives th e reliability and p ro fit function,
which indicate th a t th e reliability o f th e system reaches to a constant value a fte r a certain
period w h ile th e p ro fit function increases. If it is run fo r a long tim e , it m ay give considerable
p ro fit under favorable conditions.
106
4.8 BAYESIAN RELIABILITY ANALYSIS FOR AN OUTLIER MODEL
This study considers the Bayesian reliability analysis for an exponential failure model on the
basis of some ordered observations in the presence of a very "early failure", which is
suspected to be a possible "outlier". Bayes estimators of the mean life and reliability are
obtained under the assumption of a suitable parametric outlier model, squared error loss
function, and suitable prior densities of the parameters involved.
INTRODUCTION
In a study of Bhattacharya and Singh (1979) as well as Bhattacharya and Tyagi (1989), the
problem of reliability estimation based on a first few ordered observations from the
exponential distribution for situations when the first failure occurs at a very early stage of a
life test and is suspected to be an "early failure" or ; an "outlier", was studied under the
assumption of a suitable parametric model. This parametric outlier model was referred to as
the "SB model" in a subsequent paper of Bhauacharya and Singh (1986), wherein the case
when multiple outliers may be present, was also discussed. In this context the paper of
Gather and Kale (1981) may also be mentioned. The Bayesian reliability analysis for the SB
model was later studied by Bhaitacharya and Singh (1988). The results of this last mentioned
paper are somewhat simplified here by assuming different prior densities.
The SB model for a single outlier is described clearly in the paper of Bhattacharya and Singh
(1988) and we retain all the assumptions of that paper except that the nuisance parameter
v there in is now assumed to have a generalized uniform prior (cf. Bhattacharya, 1967)
given by the pdf
It is assumed that v ad the scale 9 of the exponential failure model are a priori independent.
From (1), it is that g(v) decreases when v moves away from zero towards unity, so that the
probability that the observation with the mean 9 vis the smallest observation will be large.
This is compatible with assumptions of Bhauacharya and Singh (1988) and the labeled
slippage model of Barnett and Lewis (1978,p.37). The contribution of the present paper is
summarized below:
From Bhauacharya and Singh (1988), the join pdf of the observations for the SB model
(single outlier) is given by
where
107
0<x, <x. < xr < oo;0 < 9 < oo,0 < v < 1) . .3
T- ]T x , +{n-r)x, ...4
\ i =i /
It may be noted th a t the statistics T is not the usual Epstein-Sobel (1953) statistics
representing the total tim e on test in the homogeneous case. Given the observed data,
the expression (2) can be considered as a function o f (9,v), and when it is the likelihood
l((9,v), so that we have:
l((9,v)a9 —r (n —l) + v~ . expj - 9" T + v *x, (0 < 9 < oo;0 < v < 1) 5
under the assumption o f the squared error loss function, the Bayes estimation 9 of 6 is
simply the posterior expectation o f (10) and is given by
108
T(a,p)= ^ e ' f - ' d t (Y > 0 ) ............ 13
( r ^ 3) ..............16
W h e re
In a s im ila r m a n n e r, w e can e v a lu a te th e in te g ra l
109
|° 0 2l{6)d6 = ^ ( r + p - 2 ) / ( jc , +T + p Y +p~2}.(« - l) ( r + p - b - 1)~‘ 2F ,[l,r + p ~ 2 ; r + p ~ b; Z )]
Hence, the Bayes estimator (0) can be obtained from (11) by substituting from (17) and (18)
The variance of the posterior distribution g * (0 ) is then obtained by using the formula
V = 0 2l ( 0 ) d 0 / ^ I ( 0 ) d 0 - ( 0 ) 2 ........... 21
The reliability function for the exponential failure model considered here, fo r a prescribed
"mission tim e" t>0, is given by
Under the assumption o f the squared error loss function, the Bayes estimator (R) of the
reliability function R is simply the posterior expectation o f e~lld with respect to . Hence, we
obtained:
The integral in the numerator is evaluated by using (12) and (14). Thus, we obtain:
110
where
Substitution of (18) and (24) in (23) leads to the Bayes estimator (i?)of reliability. The
variance of the posterior distribution of reliability is then obtained by using the formula
In conjunction with (23),(18) and (26). It may be noted that the results of this paper for b = 0
coincide with those of the paper of Bhattacharya and Singh (1988) for m = l and
Bhattacharya and Tyagi (1989)
The demand of users for perfection is probably more than it has been ever before. This
is due to increased awareness of product Quality all round. Quality awareness and control
of Quality must begin when the product is conceived and should not end until the
product-end performance is evaluated by Users. The control of Quality; should be
extended to become a part and parcel in all phases, the product undergoes. The problem
in Defense Industry is that although we have certain level of product Quality, we have not
kept pace with the expectations and demands of Users, which can only be achieved through
total involvement and commitment of everyone in achieving the ultimate Quality goal.
Perfection in Quality leads to Reliability.
Defense Industry is now faced with problems of indigenization of various imported equipment
systems, to achieve self-reliance and self-sufficiency. Defense equipments have certain
peculiar and special features, distinct from commercial requirement. These systems encompass
sophisticated, complex and multi-disciplinary components, depending upon the role; the
system has to play in war. To m eet the demands of User, need has arisen to divert
indigenization of certain components to civil-sector, which has now sufficient capacity
and infrastructure.
INTRODUCTION
To match the capacity of the potential enemy, as well as to keep pace with fast changes
in Defense Technology, ''Modernization of Defense Services" has become the need of
the hour. Based on the tactical roles. General Staff (GS) projects his requirements,
known as GS Quality Requirements (GSQR) coupled with GS Policy Statements, which
specify the conditions and constraints of Defense Weapons, Equipments, and Systems.
The GSQR is translated into technical specification for indigenous development and
111
subsequent production. The GS demands are met either by indigenous development or
by import or by indigenous development of imported Defense Weapons, Equipments
and Systems or a combination of these Three organizations namely Research and
Development (R&D), Quality Assurance (QA) and Ordinance Factory Board (QFB) are
involved in indigenization. Need has arisen to divert indigenization of certain
components/sub-systems to civil-sector viz, Public and Private Sectors, so as to avoid
foreign exchange and fresh investment involved in import and in creation of additional
facilities in Ordinance Factories (OFys) respectively Indigenization of various imported
equipment systems in Defense Industry is to achieve self-reliance and self-sufficiency.
Defense equipment has certain peculiar sad special features, distinct from commercial
requirements. These systems encompass sophisticated, complex and multi-disciplinary
components, depending on tactical role; the system has to play in War. Indigenization in
civil-sector, which has how sufficient capacity and infrastructure, rest with QA, whereas
in OFys, with OFB, who forms Task-Forces (TF), chaired by General Managers and
composed of representatives form QA and R&D, the effective contribution of latter
being little to the Quality-goal. These TF are entrusted with either indigenization of a
component or a system of a product. While indigenization of system components may
theoretically be tenable, there arises a question whether the product, assembled with
indigenized individual components can result in indigenization o f the product, m the
absence of system-integration.
Further the undue weightage given to OFB, in formation of TF, generally results in
primary obsession with the development and establishment o f production and the
successful clearance of developmental batches, relegating the Quality and Reliability
aspects over the long run to the background. The present thinking of TF that Quality and
Reliability can be injected into the product at a later stage is misplaced and is likely to
land the indigenization project back to square one, when User encounters absence of
Quality and Reliability and insist on these. This policy also leads to perpetual
concessions/deviations in the process at the conceptual stageitseif.
Methods to achieve product Quality and Reliability in Defense Industry are presented in
this paper, which reviews the functional responsibilities of organizations involved in
indigenisation.
112
RELIABILITY ASSURANCE (RA)
Reliability Assurance is another basic managerial technique for achieving the desired
degree of Reliability of a Product Statistical Methods provide a valuable tool for
reliability engineers, but the achievement of Reliability depends primarily on sound
engineering principles and techniques; in conjunction with managerial methods. The
main elements of RA are concerned with technical specification, planning, prediction,
development testing, verification and the feedback of the field-data.
113
achieved due to conflicting individual disciplines/department's goal. In case any area
cannot live up to its commitment, the other areas are to be notified to facilitate
development of alternative. The overall Quality-goal is made up of each
discipline's/department's individual Quality-goal
The salient factors that determine and ensure product Quality and Reliability are Design,
Materials, Manufacturing process and people.
DESIGN
Most of the potential product problems of Quality and reliability can be identified and
avoided in the design phase, by establishing a close-working relationship and
coordination among specialists and engineers of Design, QA, RA, Manufacturer,
Production control and User. This relationship ensures that the design is compatible with
indigenous resources, as well as with available plant facilities and manufacturing
process; the indigenous product-design can be built at desired Quality and Reliability
level as well as any potential problems of mal-assemblies are designed out: In short, the
design activity should co-ordinate all other activities and review before the design has
taken the final shape for pilot-production.
MATERIALS
To minimise "Differential Variation" due to change of either indigenous or imported
materials or sources or processes all materials needed for indigenisation o f product
should be tested at a central laboratory, created discipline-wise and approved before
their actual use, both in developing and bulk production phases, In case of materials for
explosive product indigenisation, the Ingredients, their sources processes by which
these ingredients are manufactured and their effect on functional aspects, coupled with
shelf-fife, should be estimated and established during development phase The
production control engineer/specialist should be guided by design, QA and RA
engineers/speciahsts.
M ANUFACTURING PROCESS
Manufacturer must develop the plant-process compatible to the product-design, to
ensure consistent Quality-level and degree of Reliability, taking into account the
capability and capacity of existing plant-process Marginal manufacturing process can
produce the desired Quality through the use of "80 percent concept" which allows to
use only 80 percent of specification, for which the process-gauges and check-devices are
provided, to facilitate immediate corrective action. The anticipated process-problems,
both in pilot and bulk production should be spotted and avoided with the assistance of
specialists/engineers of other activities.
PEOPLE
People are the key element in Design, Materials section, Manufacturing processes,
testing and evaluation phases o f product development cycle. Adequate training to
perform operations ensures expected Quality level. Quality-awareness communication
encourages people's involvement in the control of Quality, leading to "Participative
Management" and "Self-motivation".
114
A product to be indigenised may be multi-disciplinary. In such cases, separate sub-
projects are to be allotted discipline-wise, whereas main product-project should be
coordinated by system-integration. In all these projects and sub-projects, the design
engineer/specialist should be the chairman, comprising specialists/engineers from QA,
RA, Manufacturing process, Production control and test-evaluation.
CONCLUSION
Reliability Assurance is part and parcel of Quality Assurance Both Quality and Reliability
are in-built-in product design. Improvement of inherent Quality and Reliability of a
product rests with designer, whereas ensuring these parameters rests with production
and QA organizations. The amount of effort and expenditure devoted to improve both
Quality and Reliability depends on many factors including overall cost, safety and
operational conditions. As such both Quality and Reliability should be planned in detail
from the specification stage, throughout development and production, as well as during
shelf-life of product. For this purpose a product-project team, chaired and co-ordinated
by Designer with the active involvement and cooperation of other organizations should
be formed, as well as central testing and evaluation facilities should be created. In nut
shell, restructuring of the project teams and the review of the responsibility and
functional activity of each organization, along with creation of centralised test
evaluation facilities is the need of the hour for successful indigenisation.
115