Definitions
Distribution function:
If FX(x) is a continuous function of x, then X is a
continuous random variable.
FX(x): discrete in x Discrete rvs
FX(x): piecewise continuous Mixed rvs
Definitions
(Continued)
Equivalence:
CDF (cumulative distribution function)
PDF (probability distribution function)
Distribution function
FX(x) or FX(t) or F(t)
pdf properties:
1.
2.
Definitions
(Continued)
Equivalence: pdf
probability density function
density function
density
f(t) = dF
dt
F (t )
f ( x)dx
f ( x)dx ,
0
For a non-negative
random variable
Exponential Distribution
F(t)
12500
25000
37500
50000
f(t)
Memoryless property
Assume X > t. We have observed that the component has
Memoryless property
Assume X > t. We have observed that the
component has not failed until time t.
Let Y = X - t , the remaining (residual)
lifetime
Gt ( y ) P (Y y | X t )
P( X y t | X t )
P (t X y t )
y
1 e
P( X t )
Definitions
(Continued)
Equivalence:
Reliability
Complementary distribution function
Survivor function
R(t) = 1 -F(t)
R (t ) 1 F (t )
h(t) t = Conditional Prob. system will fail in
(t, t + t) given that it has survived until time t
f(t) t = Unconditional Prob. System will fail in
(t, t + t)
Difference between:
probability that someone will die between 90 and 91,
probability that someone will die between 90 and 91
Weibull Distribution
Frequently used to model fatigue failure, ball bearing failure etc.
(very long tails)
t
Reliability:
R t e
t0
Weibull distribution is capable of modeling DFR ( < 1), CFR ( =
1) and IFR ( >1) behavior.
5.0
1.0
2.0
3.0
4.0
Bathtub Curve
Infant Mortality
(Early Life Failures)
Steady State
Operating Time
Wear out
Early-life Period
Also called infant mortality phase or reliability
growth phase
Caused by undetected hardware/software defects
that are being fixed resulting in reliability growth
Can cause significant prediction errors if steadystate failure rates are used
Availability models can be constructed and solved
to include this effect
Weibull Model can be used
Steady-state Period
Failure rate much lower than in early-life period
Either constant (age independent) or slowly
varying failure rate
Failures caused by environmental shocks
Arrival process of environmental shocks can be
assumed to be a Poisson process
Hence time between two shocks has the
exponential distribution
Bathtub curve
DFR phase: Initial design, constant bug fixes
CFR phase: Normal operational phase
IFR phase: Aging behavior
h(t)
(burn-in-period)
(wear-out-phase)
CFR
(useful life)
DFR
IFR
t
Decreasing failure rate
Failure-Rate Multiplier
7
6
5
4
3
2
1
0
2,190
4,380
failure
rate
1 W 1 , SS
is the Weibull shape parameter
Failure rate multiplier = W ( t) SS
7
6
5
4
3
2
1
0
2,190
4,380
SS
Uniform density
U(0,1) pdf
1.2
1
cdf
0.8
0.6
0.4
0.2
0
0
2.1 2.2
Uniform distribution
The distribution function is given by:
0 ,
F(x)=
xa
b 1a ,
x < a,
a <x<b
x > b.
HypoExponential
HypoExp: multiple Exp stages in series.
2-stage HypoExp denoted as HYPO(1, 2). The density,
distribution and hazard rate function are:
probable
state
Failed state
aging
Erlang Distribution
Special case of HypoExp: All stages have same rate.
Erlang Distribution
Is used to approximate the deterministic one
since if you keep the same mean but increase
the number of stages, the pdf approaches the
delta function in the limit
Can also be used to approximate the uniform
distribution
1.8
Comparison of
probability density functions (pdf)
1.6
1.4
1.2
1
pdf
0.8
0.6
0.4
0.2
0
T=1
time
1.2
cdf
0.8
0.6
U(0,1) cdf
0.4
0.2
T=1
time
HyperExponential Distribution
Hypo or Erlang Sequential Exp( ) stages.
Alternate Exp( ) stages HyperExponential.
Log-logistic Distribution
Log-logistic can model DFR, CFR and IFR failure models
simultaneously, unlike previous ones.
For, > 1, the failure rate first increases with t (IFR); after
momentarily leveling off (CFR), it decreases (DFR) with
time. This is known as the inverse bath tub shape curve
Use in modeling software reliability growth
Defective Distribution
If
Example:
If X is N(0,1), then,
Above method suggests a way to get the random variates with desired distribution.
Choose to be F.
Since, Y=F(X), FY(y) = y and Y is U(0,1).
To generate a random variate with X having desired distribution, generate U(0,1)
random variable Y, then transform y to x= F-1(y) .
This inversion can be done in closed-form, graphically or using a table.
n j
F ( y ) [1 F ( y )]
j
hence
j
j k
FYk ( y )
n j
F ( y ) [1 F ( y )]
j
Rkofn (t ) ( nj )[ R (t )] j [1 R (t )]n j
j k
Series system:
Rseries (t ) [ R (t )]
or Ri (t )
i 1
Parallel system:
n
R
(
t
)
[
1
R
(
t
)]
Minimum of n EXP random
variables is special case of Y1 = min{X1,
parallel
,Xn} where Xi~EXP(i)
Y1~EXP( i)
This is not true (that is EXP dist.) for the parallel case
Voter
R(t)
RTMR (t ) 3R (t ) 2 R (t )
2
TMR (Continued)
Assuming that the reliability of a single
component is given by,
we get:
RTMR (t ) 3e
2 t
2e
3t
TMR
(Continued)
TMR
(Continued)
Lifetime of Lifetime of
Spare
Active
EXP()
EXP()
Total lifetime 2-Stage Erlang
R(t ) (1 t )e
EXP()
EXP()
f Z (t ) f X ( x) fY (t x) dx
0
f t (t ) e e
x
( t x )
dx
2 t
dx
0
te , t 0
2
FZ (t ) f Z ( z )dz 1 (1 t )e
0
R(t ) 1 F (t )
t
(1 t )e , t 0
Convolution: Erlang
Distribution
The general case of r-stage Erlang Distribution
When r sequential phases have independent identical exponential
distributions, then the resulting density is known as r-stage (or rphase) Erlang and is given by:
Convolution: Erlang
EXP()
EXP()
(Continued)
EXP()
r r 1 t
t e
f (t )
(r 1)!
( t ) k t
F (t ) 1
e
k!
k 0
r 1
Warm standby
With Warm spare, we have:
Active unit time-to-failure: EXP()
Spare unit time-to-failure: EXP()
EXP(+ )
EXP()
First event to occur is that either the active or the spare will fail. Time
to this event is min{EXP(),EXP()} which is EXP( + ).
Then due to the memoryless property of the exponential, remaining
time is still EXP().
Hence system lifetime has a two-stage hypoexponential distribution
with parameters
1 = + and 2 = .
f Z (t ) 1e
1 x
2 e
2 ( t x )
dx
12 2t 12 1t
e
e
1 2
2 1
Hot standby
With hot spare, we have:
Active unit time-to-failure: EXP()
Spare unit time-to-failure: EXP()
EXP(2)
EXP()
2-stage hypoexponential
EXP()
TMR
EXP(3)
EXP(2)
i 1
, where X1 ,X2 , , Xr
i
(i = j for i = j).
EXP(1)
EXP(2)
EXP(r)
Y1
EXP((n-1))
Y2
...
EXP(k)
Yn-k+1
EXP((k-1))
Yn-k+2
...
EXP()
Yn
EXP(n
+(s-1) )
...
EXP(n
+ )
EXP(n)
...
EXP(k)