Reliability
SMJ 4812 Project Mgmt and
Maintenance Eng.
Definition of Reliability
“Probability that a system or
product will perform in a
satisfactory manner for a given
period of time when used under
specified operating condition”
Reliability - 4 main elements
1. probability – numerical
representation - number of times
that an event occurs (success)
divided by total number trials
2. Satisfactory performance –
criteria established which
describe what is considered to be
satisfactory system operation
3. Specifed time – measure against which
degree of system performance can be
related - used to predict probability of
an item surviving without failure for a
designated period of time
4. Specified operating conditions expect a
system to function - environmental
factors, humidity, vibration, shock,
temperature cycle, operational profile,
etc.
LIFE CYCLE CURVE
typical life history curve for infinite
no of items – ‘bathtub curve
comparison of failure rate with time
3 distinct phase – debugging ,
chance failure and wear-out phase
Life Cycle Curve
Debugging (Infant mortality)
Phase
rapid decrease in failure rate
Weibull distribution with shape
parameter < 1 is used to describe
the occurrences of failure
Usually covered by warranty period
Chance failure phase
Constant failure rate – failure occur in
random manner
Exponential and also Weibull with =1
can be used to describe this phase
Wear-out phase
Sharp rise in failure rate – fatigue,
corrosion (old age)
Normal distribution is one that best
describes this phase
Also can use Weibull with shape
parameter > 1
Maintainability
Pertains to the ease, accuracy,
safety and economy in the
performance of maintenance
actions
Ability of an item to be maintained
Maintainability is a design
parameter, maintenance is a result
of design
Measures of
Maintainability
MTBM – mean time between maintenance,
include preventive and corrective
maintenance
MTBR – mean time between replacement,
Mgenerate spare part requirement
- mean active maintenance time
M
ct – mean corrective maintenance time or
Mmean time to repair
pt – mean preventive maintenance time
Frequency of maintenance for a
given time is highly dependent on
the reliability of that item
Reliability frequency of
maintenance
Unreliable system require
extensive maintenance
Reliability function [R(t)]
R(t) = 1 – F(t)
F(t) = probability of a system will
fail by time (t) = failure
distribution function
Eg. If probability of failure F(t) is
20%, then reliability at time t is
R(t) = 1 – 0.20 = 0.80 or 80%
Reliability at time (t)
R(t) = e-t/
e = 2.7183
= MTBF
1 = failure rate
So,
R(t) = e-t
Failure Rate ()
Rate at which failure occur in a
specified time interval
= number of failures
total operating hours
Can be expected in terms of failures per
hour, % of failure per 1,000 hours or
failures per million hours
Example 1
10 components were tested. The
components (not repairable) failed
as follows:
Component 1 failed after 75 ours
Component 2 failed after 125 hours
Component 3 failed after 130 hours
Component 4 failed after 325 hours
Component 5 failed after 525 hours
Determine the MTBF
Solution:
Five failures, operating time = 3805 hours
525
325
130
125 5 x 525
75
Solution
= 5 / 3805 = 0.001314
Example 2
The chart below shows operating time and breakdown time
of a machine.
20.2 2.5 6.1 7.1 24.4 4.2 35.3 1.8 46.7
Operating time Down time
a) Determine the MTBF.
Solution:
Total operating time = 20.2 + 6.1 + 24.4 + 4.2 + 35.3 + 46.7
= 136.9 hours
Solution
= 4 / 136.9 = 0.02922
Therefore;
= MTBF = 1/ = 34.22 hours
b) What is the system reliability for a mission time of 20
hours?
R = e-t t = 20 hours
R= e-(0.02922)(20)
R = 55.74%
Reliability Component
Relationship
Application in series network,
parallel and combination of both
Series Network
Most commonly used and the
simplest to analyze
Input A B C Output
All components must operate if the system is to
function properly.
R = RA x RB x RC
If the series is expected to operate
for a specified time period, then
Rs (t) = e ( 1 2 3 ...n ) t
Example
Systems expected to operate for 1000 hours.
It consists of 4 subsystems in series, MTBFA =
6000 hours, MTBFB = 4500 hours, MTBFC =
10,500 hours, MTBFD = 3200 hours.
Determine overall reliability.
A = 1 /MTBFA = 1/6000 = 0.000167
B = 1/MTBFB = 1/4500 = 0.000222
C = 1/MTBFC = 1/10500 = 0.000095
D = 1/MTBFD = 1/3200 = 0.000313
Therefore; R = e-(0.000797)(1000) = 0.4507
Parallel Network
A number of the same components
must fail order to cause total
system failure
A
C
Example
Consider two units A and B in
parallel. The systems fails only
when A and FBs(t)failed.
= Fa(t) Fb(t)
A = [1-Ra(t)][1-Rb(t)]
= 1-Ra(t) Rb(t) + Ra(t) Rb(t)
B
Rs(t) = 1- Fs(t)
= Ra(t) + Rb(t) – Ra(t) Rb(t)
If A and B are constant failure rate
units, then:
a t
Ra(t) = e Rb(t) = e bt
1 1 1
And Rs(t) = Rs (t ) dt
0
a b a b
s = MTBF
Consider 3 components in
parallel
Rs = 1 – F s
Fa = 1- Ra Fb = 1- Rb Fc = 1- Rc
A
Rs = 1 – (1-Ra)(1-Rb)(1-Rc)
If components A, B and C are identical, thenB
the reliability,
Rs = 1 – (1 – R)3 C
For a system with n identical components,
Rs=1- (1-R)n
Combined series parallel
network
B
A
Rs = RA [RB+RC-RBRC]
Combined series parallel
network
A C
B D
Rs = [1-(1-RA)(1-RB)][1-(1-RC)(1-RD)]
Combined series parallel
network
A
E
B D
C F
Rs=[1-(1-RA)(1-RB)(1-RC)][RD] x [RE+RF-(RE)(RF)]
Combined series parallel
network
For combined series-parallel
network, first evaluate the parallel
elements to obtain unit reliability
Overall system reliability is
determined by finding the product
of all series reliability