Professional Documents
Culture Documents
1. Plot the repair parameters m(t), M(t) and h(t). Find Mean down time.
0-10 12
11-20 15
21-30 22
31-40 19
41-50 18
51-60 14
Discrete distributions of m(t), M(t), h(t) etc are defined as follows
for Grouped Data:
Consider repairs taking place for N items in a time T (divided into equal
intervals t).
N=100 m(t)=1
MDT = 5*0.12 + 15*0.15 + 25*0.22 + 35*0.19 + 45*0.18 + 55*0.14
(i.e taking mid values of intervals)
=30.8hours
Note: this is Mean Down Time because of waiting
period at workshop.
2. Given that the Mean Time Between Failure (MTBF) of this particular
instrument is 260 hours of use, what is its Availability?
Availability is MTBF/(MTBF+MDT)
=260/(260+30.8)
=89.4%
Continuous form of Repair functions:
They follow similar rules as the function f (t ) and F (t ) for non-repairable components
1. m(t ) dt 1
0
t
2. M (t ) m(t ') dt '
0
3. MTTR t. m(t ) dt;
0
Note : (t MTTR ) . m(t ) dt t 2 . m(t ) dt MTTR 2 .
2 2
0 0
4. MTTR (1 M (t )) dt
0
From these we can work out the MTBF and MDT of that machine.
We can also guess the type of distribution of m(t) using probability chart
paper or Weibull analysis.
Data from the operating
Uptime logbook of a machine.
Downtime
Uptime
Downtime (Time in hours)
Calculate :
1. MTBF
2. MDT
3. Distribution of m(t)
4. Availability of this machine.
Total=8205.3 Total=112.5
MTBF=8205.3/12 MDT=112.5/12
=683.8 hr =9.4 hr
What is the distribution of m(t)?
Let us order the downtimes, and plot it vs Cumulative function M(t).
i.e, y bx c
(b is the slope equal to , and y intercept c is equal to - log e ).
Note : x coordinates are in the form of log (t).
OR
1
MTTR= (1 ) 10 (2) 10 hr (agrees with other estimate of 9.4hr)
tmed 6.9hr
In the same way we can fit a Weibull distribution to the uptimes.
We can also find the failure rate ( ROCOF ) from the "uptime" Weibull fit;
1
t
0
t
(t ) 0.00143 0.00143/ hr (constant failure rate).
700
ht ' ht
M (t ) m ( t ') dt ' h e dt ' 1 e
0 0
1 1
MTTR ; .
h h
In many practical cases, repair times are log-normally distributed:
(i.e. distribution of log e (t) is normally distributed).
1 t
M (t ) log e ( z ).
tmed
2
MTTR tmed e 2
Example :
1.Exponential Distribution
A particular type of component can be repaired at the constant rate of 0.01 per hour.
What is the probability of completing a repair in 50 hr?
What is the probability of exceeding 50 hr for a repair?
Ans :
h 0.01 / hr
MTTR=1/0.01=100hrs.
1.Probability of completing a repair before 50hrs = M(50)
=1-e- ht 1 e-0.0150
=0.3937
2.Probability of exceeding 50 hr for a repair =1-0.3937=0.6065
Example :
2.Lognormal Distribution
Ans :
M (3) 0.9
1 3
log e 0.9
0.45 tmed
( z ) 0.9 (From Gaussian tables, z =1.28)
1 3
1.28= log e
0.45 tmed
tmed 1.686 hr
2 0.452
MTTR tmed e 2 1.686 e 2
1.866 hr
END