Weibull and Exponential Distributions

RELI ABI LI TY ENGI NEERI NG UNI T
ASST4403
Lec t ur e 8: WEI BULL AND EXPONENTI AL DI STRI BUTI ONS
1
L i Learning outcomes
Familiarity with the properties of Weibull and
exponential distributions
Ability to find characteristics of a random variable
with Weibull distribution
Ability to find characteristics of a random variable
with exponential distribution p
2
Weibull distribution
|: shape parameter
1
( )
t
t
f t
|
|
q
|
| |
|
\ .
| |
|
Weibull distribution
q: characteristic life
(scale parameter)
( ) f t e
q
|
q q
\ .
=
|
\ .
(Hyper-exponential)
(E ti l)
f(t)
|=3.5
|=1.0
|=0.5
(Exponential)
(Normal)
|=2.0
(Rayleigh)
3
t
What is behind
Used first to model fatigue/
t th strength
An extreme value distribution
(the smallest extreme value)
Most flexible
(Waloddi Weibull:
1887-1979, Swedish
Most flexible
engineer)
4
Properties
|: shape parameter
q: characteristiclife
1
( )
t
t
f
|
|
q
|
| |
|
\ .
| |
|
Properties
( ) f t e
q
|
q q
\ .
| |
=
|
\ .
{ }
( ) 1
t
F t P T t e
|
q
| |
|
\ .
s
{ }
( ) 1 F t P T t e
q
\ .
= s =
1
( ) 1 0.632 F t e q

= = =
5
Weibull Distribution
1
f(t)
h t i ti lif ti t hi h
0.8
0.9
1
f(t)
q, characteristic life: time at which
cumulative probability = 63.2%
0 5
0.6
0.7
0.3
0.4
0.5
63.2%
0.1
0.2
t
0
Note: when |=1, q=
q
6
| , q
Properties (contd)
|: shape parameter
q: characteristiclife
1
1 q
| |
= I +
|
The mean value
Properties (cont d)
1 q
|
I +
|
\ .
except when 1 q | = =
Note:
2
2 2
2 1
1 1 o q

(
| | | |

I + I +
`
( | |
except when 1 q | = =
The standard
Note:
1 1 o q
| |
= I + I +
`
( | |
\ . \ .

)
deviation
The median (the random variable has 50% chance of being
lower and 50% greater than this value)
|
q
/ 1
5 . 0
) 5 . 0 ln ( = =
median
t t
7
Gamma function
If z=n is a positive integer then If z=n is a positive integer, then
Otherwise the value I(z) can be found in a table
N t l f ll >0 Note also for all z>0
Example: I(4)=(4-1)!=3!=321=6
I(2.6)=1.42962 ( )
I(3.7)= I(2.7+1) =2.7 I(2.7)= 2.7 1.54469=4.1707
8
Gamma function table Gamma function table
9
0.
S &
0.
Same q &
growing |
0.
0.
Same | &
growing q
0.
0.
10
Effect of | and q on the shape of the Weibull pdf
3-Parameter
ib ll di ib i Weibull distribution
|
q
|
|
|
|
.
|
\
|
|
|
|
t
t
1
: Location Parameter
q
q

q
|
|
.
\
|
|
.
|
\
|
= e
t
t f ) (
1 f(t)
: Location Parameter
f(t)
0
time
63.2%
11
time
q
Weibull distribution : examples

A weibull distributed random variable has shape
t 2 d l t 1 000 h Fi d parameter 2 and scale parameter 1,000 hr. Find
The mean
St d d d i ti Standard deviation
The median
The probability the random variable takes a value p y
between 800 and 1,200 hours
The time at which the random variable has 90%
chance to be greater than chance to be greater than
12
Weibull distribution : examples Weibull distribution : examples
The mean
)
2
1
1 ( 1000 )
1
1 ( + I = + I =
|
q
Standard deviation
hr 23 . 886 88623 . 0 1000 ) 5 . 1 ( 1000
2
= = I =
|
Standard deviation
)
2
1
1 ( ) 1 1 ( 1000 )
1
1 ( )
2
1 (
2
2
(
+ I + I =
(
+ I + I =
| |
q o
The median
| | hr 25 . 463 88623 . 0 1 1000
2
= =
Th b bilit th d i bl t k l
hr t t
median
55 . 832 ) 5 . 0 ln ( 1000 ) 5 . 0 ln (
2 / 1 / 1
5 . 0
= = = =
|
q
The probability the random variable takes a value
between 8,000 and 24,000 hours
Th ti t hi h th d i bl h 90%
13
chance to be greater than
Weibull distribution : examples Weibull distribution : examples
between 800 and 1 200 hours between 800 and 1,200 hours
) 800 ( ) 1200 ( ) 1200 800 Pr(
2 2
800 1200
( (
= = s s
|
|
|
|
|
|
F F T
2904 . 0 1 1
1000 1000
=
(
(

(
(
|
.
\
|
.
e e
hr t e t F
t T t T
t
59 324 1 0 1 1 0 ) (
1 . 0 ) Pr( 9 . 0 ) Pr(
1000
9 . 0 9 . 0
2
9 . 0
= = =
= < = >
|
.
|
\
|
14
hr t e t F 59 . 324 1 . 0 1 1 . 0 ) (
9 . 0 9 . 0
= = =
Exponential Distribution
A special case when |=1 in Weibull distribution
Very important in reliability engineering
More to come More to come
15
Exponential Distribution
1
f(t)
0 7
0.8
0.9
( )
0.5
0.6
0.7
f(t)
0.3
0.4
0.5
0.1
0.2
0
t, time
16
t
e t f

= ) (
Exponential Distribution Exponential Distribution
f(t)
0, 0 ( ) ,
t
t f t e

> > = ( ) { } F t P X t = s
f( )
0
1
t
x t
e dx e

= =
}
t
63.2%
= 1/
t
17
Exponential Distribution Exponential Distribution
f(t)
Q: Why 63.2%?
f( )
1
( ) 1 1 0.632
t
t
F t e e

=
= = = =
1
=
mean
Standard
deviation
o
1
=
t
63.2%
deviation
o =
= 1/
t
18
Exponential distribution : examples
Redo the example for Weibull distributed random
i bl ith h t 1 d l variable with shape parameter 1 and scale
parameter 1,000 hr.
19
Exponential distribution :
l examples
The mean
)
1
1
1 ( 1000 )
1
1 ( + I = + I =
|
q
Standard deviation
hr 1000 1 1000 ) 2 ( 1000
1
= = I =
|
Standard deviation
)
1
1
1 ( ) 2 1 ( 1000 )
1
1 ( )
2
1 (
2
2
(
+ I + I =
(
+ I + I =
| |
q o
The median
| | hr 1000 ! 1 ! 2 1000
2
= =
Th b bilit th d i bl t k l
hr t t
median
15 . 693 ) 5 . 0 ln ( 1000 ) 5 . 0 ln (
1 / 1 / 1
5 . 0
= = = =
|
q
between 8,000 and 24,000 hours
20
Exponential distribution :
l examples
between 800 and 1 200 hours between 800 and 1,200 hours
) 800 ( ) 1200 ( ) 1200 800 Pr(
800 1200
( (
= = s s
|
|
|
|
|
|
F F T
1481 . 0 1 1
1000 1000
=
(
(

(
(
|
.
\
|
.
e e
hr t e t F
t T t T
t
6 2302 1 0 1 1 0 ) (
1 . 0 ) Pr( 9 . 0 ) Pr(
9 0
1000
9 0
9 . 0 9 . 0
9 . 0
= = =
= < = >
|
.
|
\
|
hr t e t F 6 . 2302 1 . 0 1 1 . 0 ) (
9 . 0 9 . 0

21

Weibull and Exponential Distributions

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Weibull and Exponential Distributions

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Weibull and Exponential Distributions

Uploaded by

Copyright:

Available Formats

RELI ABI LI TY ENGI NEERI NG UNI T

Weibull distribution : examples

You might also like