Chap 2: Reliability and Availability Models

Reliability
R(t) = prob{S is fully functioning in [0,t]}
Suppose from [0,t] time period, we measure out of N components,
of which N0 (t): # of components operating correctly at time t
Nf (t): # of components which have failed at time t
R (t )  reliability of the component
N 0(t ) N  Nf (t ) Nf (t )
   1
N N N
dR (t )  1 dNf (t )
 Physical meaning:
dt N dt
dNf (t ) dR(t ) instantaneous rate at which
or  N components are failing
dt dt
24
How many unfailing components are there at time t ?  N 0(t )

dNf (t ) N
 Z (t ) * N 0(t )  N 0 (t)
Nf(t+dt) - Nf(t)
dt
Failing rate of dt

a single component N f (t)

1 dN f ( t )
Z (t )  t t+dt time
N 0 ( t ) dt
dR ( t )

1  dR ( t )  dt f (t )
(2.3)
   N   
N 0(t )  dt  R (t ) R (t ) text p.28

Z(t) is also called the hazard rate.

25
For electronic components, Z(t)’s relationship with respect to time is
a bathtub curve.
Burn-in period
Z(t)
High failure rate
due to faulty design,
manufacturing or Useful time phase
Wear-out
assembly.
Infant phase
mortality (due to aging)
Weak components phase
are removed during
Failure rate can be
the “burn-in” period. time
assumed to be a
constant during the
useful time period,
say 

26
  1 dR ( t )
 
R ( t ) dt
dR ( t )
   R (t )
dt
 t
Exponential Failure Law R(t )  e

e.g.  = 0.01 hr-1, what is R(t) at t = 100 hrs? Ans: e-0.01*100

 For hardware components, exponential failure law is frequently

assumed.

 For software components, the reliability may grow as the software’s

design faults are removed during the testing/debugging phase.
27
In general, we can assume

Z(t) =  ( t )-1 Weibull dist.

 - (t )
R(t) = e

This yields Z(t) Use this to model software

failure rate

<1 >1
e.g.  = -1
-1 =1
R(t) = et t

 t  , R(t)  1  Reliability improves as a function of t

t  0, R(t)  0
28
Formal Definition of R(t):

Let x be a .. representing the life of a system and let F be the

cumulative distribution function (CDF) of x.
Then, 
R (t )  prx  t  1  F (t )   f ( x)dx
t

 For a component obeying the exponential failure law


R (t )   e  x dx  e  t
t

29
Mean Time to Failure (MTTF):
i first failing time
1 t1
The expected time that a system will Identical
operate before the first failure occurs systems
N tN
Discrete case Continuous case
dF (t ) d 1  R (t ) 
MTTF  E T   0 t f (t ) dt  0 t
  
dt  0 t 
t
dt
i dt dt
failure time
MTTF  i

  0 t 
dR (t )
dt   tR (t )   1 R (t ) dt 0

N

dt
 uv' dt
 0 R (t ) dt  R (t )  0 as t  
 uv   u ' vdt
 1
e.g . MTTF  0 e t dt 

Q: what is the reliability of a system obeying the exponential failure
law at t = MTTF? 1
 1 
Ans: R ( t  MTTF )  R  t    e     e 1  1 
  1
 0 . 3678 ...
  e 2 . 718 ...
30
MTTR (Mean Time to Repair)
If also assume a failed system obeys “Exponential Repair Law”,
then 1
MTTR  , where  is the repair rate

Relationship between MTBF (Mean time between failure) , MTTR
& MTTF:
time
MTTF MTTR MTTF MTTR

Time of Time of
first failure 2nd failure

MTBF = MTTR + MTTF

 If MTTF >> MTTR Then MTBF  MTTF

31
Availability
Instantaneous (or point) Availability
A(t) = prob {the system is functioning at time t}

A(t) Regardless of the # of times it may have

failed & been repaired during [0, t]

t time
0

MTTF
Steady-State Availability = MTTF+MTTR

 For a system without repair, A(t) = R(t)

32
Assume exponential “failure” & “repair” law

See also page 67, text chapter 4
O F

Time domain:
Po’(t) = - Po(t) + PF(t)
PF’(t) = Po(t) - PF(t)
with initial state Po(0) = 1 & PF(0) = 0
Laplace domain: Po(0)
SPo(S) - 1 = - Po(S) + PF(S)
SPF(S) - 0 = Po(S) - PF(S)
PF(0)
33
 
 Po(S) 
1
 
A

B
S  (   ) S (S     ) S S  (   )
 
     
  LT
S S  (   ) F(t) L(F(t)) = f(S)
Inverse
Similarly LT
  1 1/S
PF(S) 
 

  t 1/S2
Inverse LT
S S  (   ) to return to tn n!/Sn+1
time domain eat 1/(S-a)
Po(t) =  
+ e-(+)t
+ + Physical meaning:
  Po(t) = prob {the system is
PF(t) = - e-(+)t
+ + functioning at time t }
 
 A(t) = + + + e-(+)t
34
Q1: unavailability? Ans: PF(t)
e.g.  = 0.01 &  = 0.1
Q2: R(t)? Still e-t
1.0
Q3: Steady-state availability?
0.9
Ans: A( t) = 
0.90909...
+ A(t)
t

MTTF 

  
MTTF  MTTR   

 

35
Modeling:
Series-Parallel Reliability Block Diagrams

 A series-parallel block diagram represents the logical structure of

a system with regard to how the reliabilities of its components
affect the system reliability.

 Components are combined into blocks in

 series
 parallel
or  k-of-n configurations

36
A. Serial system: each element of the system is required to function
correctly for the system to function
1 2 3 n
correctly.
n
R series (t )  
i1
R i(t ) 1
n

e1t e2t ...ent e...nt et  series   i

i 1

B. Parallel system: only one of several elements must be operational

for the system to be operational.
1 R parallel (t )  F parallel (t )  1
n

Assumptions:
2  R (t )  1  F (t )  1  
i1
F i(t )

independent n

random variables
n
 1   1
i1
 R i(t )  2
perfect coverage
 1 
n
  it
so up to n-1 ex : R ( t )  1   e
failures can be i1

tolerated. 37
C. Combination of series & parallel systems
e.g.
Computer 1 Interface 1 Display 1 Bus 1

Computer 2 Interface 2 Display 2 Bus 2

Parallel 1 2 3 4

4
R series  R
i 1
parallel , i (t )

4 2 
   1   1  R j , i ( t )  Where R is for jth
i 1  j 1  component
j,i
of ith unit
Numerical ex: R=0.9
then Rsystem = [1-(1-0.9)2]4 = 0.96
v.s. Rnon-redundant = (0.9)4 = 0.6561
38
e.g.
R1 R1 1

R2 R2 R2 2

R3 3

Rparallel = 1 - ( 1 - Rseries, 1 )( 1 - Rseries, 2 )( 1 - Rseries, 3 )

= 1 - ( 1 - R12 )( 1 - R23 )( 1 - R3 )
Q: Prove the following theorem:
Replication at the component level is more effective than replication at
the system level in improving system reliability using the same # of
components.

Ans: Show that is better than

Assume R=1/2 for each component
2
  1  1 
2
9   1  1   7
Rsystem  1  1  1    Rsystem  1  1      
  2  2  16   2  2   16

39
D. k-out-of-n
e.g. TMR ( Triple Module Redundancy ) is a 2-out-of-3 system.
RTMR(t) = R1(t) * R2(t) * R3(t) all are functioning
+ R1(t) R2(t) ( 1 - R3(t) )
+ R1(t) R3(t) ( 1 - R2(t) ) 1 failed & 2 are functioning
+ R2(t) R3(t) ( 1 - R1(t) )

when R1(t) = R2(t) = R3(t) = R(t)

R2-out-of-3(t) = 3R2(t) - 2R3(t)

In general, n
n
 n (t )    R (t ) 1  R (t ) 
n i
Rk  out  of i 3
i
i k  
e.g.
 3 2  3  3 3! 2
R 2  out  of  3(t )    R (1  R )    R 
 2  3 2!1!

R  R 3
 
R 3
 3 R 2
 2 R 3

   
40
 Rsystem RTMR(t)
Q: Is RTMR(t) > Rsystem with a single component(t) ?
1.0
Let 3R2 - 2R3 = R
3
 R - 2 R + 0.5 = 0  R = 1.0 or 0.5
2 R(t)
single
RTMR(t) = R(t) when the reliability of a
More
single module is 1.0 or 0.5 realistic
region
In fact, when R(t) < 0.5, R(t) > RTMR(t)
0.5 1.0 R(t)
Q: what is the MTTF of a k-out-of-n system when each single module follows the
exponential failure law with a failure rate of  ?
e.g.
n
n!
Rsystem(t )   (n  i )!i !t i
(e ) (1  e )
t ni
2-out-of-3:
i k  1 1 5 
 MTTF=   
 MTTF   Rsystem(t )dt  1  1  ...  1 
3 2 6 

0  n k 2-out-of-5:
  1 1 1 1  77 
MTTF=   5  4  3  2   60 
 

41
Q: what is the reliability & MTTF of the following structure?
Fig. 2.5, p36 m1
P1
m2 A parallel system too
Also a parallel system P2
m3
m = 0.00764/day
p = 0.00139/day
( 1-out-of-3 )
( 1-out-of-2 )

Rsystem(t) = [ 1 - ( 1 - e-0.00139t )2 ][ 1 - ( 1 - e-0.00764t )3 ]

= 6e-0.00903t-3e-0.01042t-6e-0.01667t+3e-0.01806t+2e-0.02431t-e-0.0257t
 


MTTF = R system (t ) dt
0
6 3 6 3 2 1
= - - + + - = 226.09
0.00903 0.01042 0.01667 0.01806 0.02431 0.0257
Recall that Repair rate Failure rate of component i
i i
Ai(t) =  e  (  i i ) t
 i  i  i  i
Equations 1 2 & 3 obtained above can also be used to compute
the system availability  by replacing Ri(t) with Ai(t) 42
Specifically,
n
A series (t )   A i (t ) 4
i 1
n
A parallel (t )  1   1  A i (t )  5
i 1
n
n i
 n ( t )     A ( t ) 1  A ( t ) 
n i
A k  out  of i 6
ik  

Assuming A1(t )  A2(t )  ....  An(t )  A(t )

For steady state availability

i
Use Ai( t   ) = into equations 4 5 & 6 above
i + i
43