MYcsvtu Notes

Fundamentals of Reliability Engineering and

Applications

www.mycsvtunotes.in
Reliability Engineering
Outline

MYcsvtu Notes
• Reliability definition
• Reliability estimation
• System reliability calculations

www.mycsvtunotes.in
2
Reliability Importance

• One of the most important characteristics of a product, it is a

measure of its performance with time (Transatlantic and

MYcsvtu Notes
Transpacific cables)

• Products’ recalls are common (only after time elapses). In

www.mycsvtunotes.in
October 2006, the Sony Corporation recalled up to 9.6 million
of its personal computer batteries

• Products are discontinued because of fatal accidents (Pinto,

Concord)

• Medical devices and organs (reliability of artificial organs)

3
Reliability Importance
• Business data

MYcsvtu Notes
www.mycsvtunotes.in
Warranty costs measured in million dollars for several large
American manufacturers in 2006 and 2005.
(www.warrantyweek.com) 4
 Some Initial Thoughts
Repairable and Non-Repairable
Another measure of reliability is availability (probability

MYcsvtu Notes
that the system provides its functions when needed).
Maximum Reliability level

www.mycsvtunotes.in
With R
epairs
Reliability

No Rep
airs

Time
Some Initial Thoughts
Warranty
• Will you buy additional warranty?
• Burn in and removal of early failures.
(Lemon Law).

MYcsvtu Notes
Early Failures

www.mycsvtunotes.in
Increasing
Constant Failure
Failure Rate
Failure Rate

Rate

Time
 Reliability Definitions

Reliability is a time dependent characteristic.

MYcsvtu Notes
It can only be determined after an elapsed time but can be
predicted at any time.

www.mycsvtunotes.in
It is the probability that a product or service will operate
properly for a specified period of time (design life) under
the design operating conditions without failure.

7
 Other Measures of Reliability

Availability is used for repairable systems

MYcsvtu Notes
It is the probability that the system is operational at
any random time t.

www.mycsvtunotes.in
It can also be specified as a proportion of time that
the system is available for use in a given interval
(0,T).

8
 Other Measures of Reliability

Mean Time To Failure (MTTF): It is the average

time that elapses until a failure occurs.

MYcsvtu Notes
It does not provide information about the distribution
of the TTF, hence we need to estimate the variance

www.mycsvtunotes.in
of the TTF.

Mean Time Between Failure (MTBF): It is the

average time between successive failures.
It is used for repairable systems.
9
Mean Time to Failure: MTTF

 

MYcsvtu Notes
MTTF   tf (t )dt   R(t )dt
0 0

1 n
MTTF   ti

www.mycsvtunotes.in
n i 1

1
2 is better than 1?
2
R(t)

1
Time t
0
10
Mean Time Between Failure: MTBF

MYcsvtu Notes
www.mycsvtunotes.in
11
Other Measures of Reliability

Mean Residual Life (MRL): It is the expected remaining life, T-t,

given that the product, component, or a system has survived

MYcsvtu Notes
to time t.
1 
L(t )  E[T  t | T  t ]    f ( )d  t

www.mycsvtunotes.in
R(t ) t
Failure Rate (FITs failures in 109 hours): The failure rate in a
time interval [ t1]ist2the probability that a failure per unit
time occurs in the interval given that no failure has occurred
prior to the beginning of the interval.

Hazard Function: It is the limit of the failure rate as the length

of the interval approaches zero.
12
Basic Calculations

Suppose n0 identical units are subjected to a

MYcsvtu Notes
test. During the interval (t, t+∆t), we observed
nf(t) failed components. Let ns(t) be the
surviving components at time t, then the MTTF,

www.mycsvtunotes.in
failure density, hazard rate, and reliability at
n0

t
time t are: MTTF  n , fˆ (t ) 
i 1

n t
0
i
n (t ) f

n f (t ) n (t )
ˆ (t )  , Rˆ (t )  Pr (T  t )  s
ns (t )t n0

13
 Basic Definitions Cont’d
The unreliability F(t) is
F( )t 1  () R t

MYcsvtu Notes
Example: 200 light bulbs were tested and the failures in
1000-hour intervals are

Time Interval (Hours) Failures in the

www.mycsvtunotes.in
interval
0-1000 100
1001-2000 40
2001-3000 20
3001-4000 15
4001-5000 10
5001-6000 8
6001-7000 7
Total 200
14
 Calculations
Time Failure Density Hazard rate
Interval f (t ) x 104 h(t ) x 104

MYcsvtu Notes
100 100
Time Interval Failures 0-1000 200  103
 5.0
200  103
 5.0

(Hours) in the
interval

www.mycsvtunotes.in
0-1000 100
1001-2000
40 40
 2.0  4.0
1001-2000 40 200  103 100  103
2001-3000 20
20
2001-3000
3001-4000 15 20
 1.0  3.33
4001-5000 10 200  103 60  103
5001-6000 8
6001-7000 7 …… …….. ……
Total 200
7 7
6001-7000 200  103
 0.35
7  103
 10

15
Failure Density vs. Time

MYcsvtu Notes
×10-4

www.mycsvtunotes.in
1 2 3 4 5 6 7 x 103

Time in hours
16
Hazard Rate vs. Time

MYcsvtu Notes
×10-4

www.mycsvtunotes.in
1 2 3 4 5 6 7 × 103

Time in Hours

17
Calculations

MYcsvtu Notes
Time Interval Failures
(Hours) in the Time Interval Reliability R(t )
interval 0-1000 200/200=1.0
0-1000 100

www.mycsvtunotes.in
1001-2000 40
1001-2000 100/200=0.5
2001-3000 20
3001-4000 15
4001-5000 10 2001-3000 60/200=0.33
5001-6000 8
6001-7000 7
…… ……
Total 200

6001-7000 0.35/10=.035

18
Reliability vs. Time

MYcsvtu Notes
www.mycsvtunotes.in
1 2 3 4 5 6 7 x 10 3
Time in hours

19
Exponential Distribution

Definition (t)

MYcsvtu Notes
 (t )     0, t  0 Time

www.mycsvtunotes.in
f (t )   exp(t )

R(t )  exp(t )  1  F (t )

20
Exponential Model Cont’d

Statistical Properties

MYcsvtu Notes
1
MTTF    5  106 Failures/hr


www.mycsvtunotes.in
MTTF=200,000 hrs or 20 years

Variance 
1
2

1
Median life (ln )2 Median life =138,626 hrs or 14
 years
21
Empirical Estimate of F(t) and R(t)
When the exact failure times of units is known, we
use an empirical approach to estimate the reliability

MYcsvtu Notes
metrics. The most common approach is the Rank
Estimator. Order the failure time observations (failure

www.mycsvtunotes.in
times) in an ascending order:

t1  t 2  ...  t i 1  t i  t i 1  ...  t n 1  t n
 Empirical Estimate of F(t) and R(t)
F (ti ) is obtained by several methods

MYcsvtu Notes
i
1. Uniform “naive” estimator
n
i

www.mycsvtunotes.in
2. Mean rank estimator n 1
i  0.3
3. Median rank estimator (Bernard) n  0.4
i 3/8
4. Median rank estimator (Blom)
n 1/ 4
Empirical Estimate of F(t) and R(t)
Assume that we use the mean rank estimator
ˆ i
F (ti ) 

MYcsvtu Notes
n 1
ˆ n 1 i
R(ti )  ti  t  ti 1 i  0,1, 2,..., n
n 1

www.mycsvtunotes.in
Since f(t) is the derivative of F(t), then

ˆ (t )  Fˆ (t )
F
fˆ (ti )  i 1 i
ti  ti 1  ti
ti .(n  1)
1
fˆ (ti ) 
ti .(n  1)
24
Empirical Estimate of F(t) and R(t)
1
ˆ (t ) 
ti .(n  1  i )
i

MYcsvtu Notes
Hˆ (ti )   ln ( Rˆ (ti )

Example:

www.mycsvtunotes.in
Recorded failure times for a sample of 9 units are
observed at t=70, 150, 250, 360, 485, 650, 855,
1130, 1540. Determine F(t), R(t), f(t), ( )t ,H(t)

25
Calculations

i t (i) t(i+1) F=i/10 R=(10-i)/10 f=0.1/t  =1/(t.(10-i)) H(t)

MYcsvtu Notes
0 0 70 0 1 0.001429 0.001429 0
1 70 150 0.1 0.9 0.001250 0.001389 0.10536052
2 150 250 0.2 0.8 0.001000 0.001250 0.22314355

www.mycsvtunotes.in
3 250 360 0.3 0.7 0.000909 0.001299 0.35667494
4 360 485 0.4 0.6 0.000800 0.001333 0.51082562
5 485 650 0.5 0.5 0.000606 0.001212 0.69314718
6 650 855 0.6 0.4 0.000488 0.001220 0.91629073
7 855 1130 0.7 0.3 0.000364 0.001212 1.2039728
8 1130 1540 0.8 0.2 0.000244 0.001220 1.60943791
9 1540 - 0.9 0.1 2.30258509

26
Reliability Function

1.2

MYcsvtu Notes
1

Reliability 0.8

www.mycsvtunotes.in
0.6

0.4

0.2

0
0 200 400 600 800 1000 1200

Time

27
Probability Density Function
0.001600

MYcsvtu Notes
0.001400

0.001200

0.001000

Density

www.mycsvtunotes.in
0.000800
Function
0.000600

0.000400

0.000200

0.000000
0 200 400 600 800 1000 1200

Time

28
Failure Rate
Constant
0.001900

MYcsvtu Notes
0.001700

0.001500

0.001300

www.mycsvtunotes.in
0.001100
Failure Rate
0.000900

0.000700

0.000500

0.000300

0.000100
0 200 400 600 800 1000 1200

Time

29
Exponential Distribution: Another Example

Given failure data:

MYcsvtu Notes
Plot the hazard rate, if constant then use the
exponential distribution with f(t), R(t) and h(t) as

www.mycsvtunotes.in
defined before.

We use a software to demonstrate these steps.

30
 Input Data

31

www.mycsvtunotes.in MYcsvtu Notes

 Plot of the Data

32

www.mycsvtunotes.in MYcsvtu Notes

 Exponential Fit

33

www.mycsvtunotes.in MYcsvtu Notes

 Exponential Analysis

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Go Beyond Constant Failure Rate

MYcsvtu Notes
- Weibull Distribution (Model) and
Others

www.mycsvtunotes.in
35
The General Failure Curve
Failure Rate

MYcsvtu Notes
1 3
Constant Failure Rate
Region

www.mycsvtunotes.in
2

Early Life Wear-Out

Region ABC Region
Module

0 Time t
36
Related Topics (1)

Burn-in:

MYcsvtu Notes
Failure Rate

According to MIL-STD-883C,
1
burn-in is a test performed to
screen or eliminate marginal

www.mycsvtunotes.in
components with inherent
defects or defects resulting
from manufacturing process.
Early Life
Region

0 Time t 37
 Motivation – Simple Example

• Suppose the life times (in hours) of several units

MYcsvtu Notes
are: 1 2 3 5 10 15 22 28

1  2  3  5  10  15  22  28
MTTF   10.75 hours

www.mycsvtunotes.in
8

After 2 hours of burn-in

1 2 3-2=1 5-2=3 10-2=8 15-2=13 22-2=20 28-2=26

1  3  8  13  20  26
MRL(after 2 hours)   11.83 hours > MTTF
6
Motivation - Use of Burn-in
• Improve reliability using “cull eliminator”

MYcsvtu Notes
After burn-in
Before burn-in

www.mycsvtunotes.in
Company
1
MTTF=5000 hours
Company
2
39
Related Topics (2)

Maintenance:
Hazard Rate

MYcsvtu Notes
An important assumption for
effective maintenance is that 3
component has an

www.mycsvtunotes.in
increasing failure rate.

Why?
Wear-Out
Region

0 Time t
40
Weibull Model

• Definition

MYcsvtu Notes
 1
t   t  
f (t )    exp        0,   0, t  0
       

www.mycsvtunotes.in
 1
t
 (t )  f (t ) / R(t )   
  

  t  
R(t )  exp       1  F (t )
    

41
Weibull Model Cont.

• Statistical properties

MYcsvtu Notes
 1
MTTF   0 t1/  et dt  (1 

)

www.mycsvtunotes.in
 
2
2 2 1  
Var   (1  )   (1  ) 
     
 

Median life   ((ln 2)1/  )

42
 Weibull Model

43

www.mycsvtunotes.in MYcsvtu Notes

Weibull Analysis: Shape Parameter

MYcsvtu Notes
www.mycsvtunotes.in
44
Weibull Analysis: Shape Parameter

MYcsvtu Notes
www.mycsvtunotes.in
45
Weibull Analysis: Shape Parameter

MYcsvtu Notes
www.mycsvtunotes.in
46
 Normal Distribution

47

www.mycsvtunotes.in MYcsvtu Notes

 Weibull Model
 t  1
h(t )  ( ) .
 

MYcsvtu Notes
 t  1  ( t ) 
f (t )  ( ) e
 

www.mycsvtunotes.in
t   1 (  ) 
F (t )   ( ) e d
0 

t
( ) 
F (t ) 1  e 

t
( ) 
R(t )  e 
Drs. Elsayed and Liao
Drs. Elsayed and Liao
Input Data

www.mycsvtunotes.in MYcsvtu Notes

 Plots of the Data

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Weibull Fit

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Test for Weibull Fit

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Parameters for Weibull

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Weibull Analysis

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Example 2: Input Data

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Example 2: Plots of the Data

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Example 2: Weibull Fit

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Example 2:Test for Weibull Fit

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Example 2: Parameters for Weibull

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
 Weibull Analysis

MYcsvtu Notes
www.mycsvtunotes.in
Drs. Elsayed and Liao
Versatility of Weibull Model
 1
t
Hazard rate:  (t )  f (t ) / R(t ) 
   

MYcsvtu Notes
Hazard Rate

Constant Failure Rate

Region
 1

www.mycsvtunotes.in
0   1

Early Life Wear-Out

Region Region
 1

0 Time t
61
Graphical Model Validation
• Weibull Plot

MYcsvtu Notes
  t  
F (t )  1  R(t )  1  exp     
    

www.mycsvtunotes.in
1
 ln ln   ln t   ln  is linear function of ln(time).
1  F (t )

• Estimate Fˆ (ti ) at ti using Bernard’s Formula

For n observed failure time data (t1 , t2 ,..., ti ,...tn )
ˆ i  0.3
F (ti ) 
n  0.4 62
 Example - Weibull Plot

• T~Weibull(1, 4000) Generate 50 data

MYcsvtu Notes
Weibull Probability Plot
0.99
0.96
0.90
0.75

www.mycsvtunotes.in
0.50 0.632
Probability

0.25
If the straight line fits
0.10
 the data, Weibull
0.05
distribution is a good
model for the data
0.02
0.01

10
-5
10
0
 10
5
63
Data