Postmarket Surveillance - Medical Device Design: Innovation from Concept to Market - page 292

life when mean rank is 0.63 is the mean life (but only for the straight-line ideal); if you had to
straighten your line the actual mean life is given by Equation (12.3).

(12.3)

Actual mean eanlife(measured t @0 63)

o

Clearly this is invaluable information to the designer. If you have designed your device to
last for 100 uses then t o had better be >100! Let us reexamine the data in Table 12.1 and plot
these on the Weibull plot ( Figure 12.6 ).

Using Equation (12.2) we obtain t o :

( 180 − −

−−−

100 100 )( 80 )

t 0 100

( 180 100 ) ( 100 80 )

73 uses

We now replot the data using the modified life values ( t - t o ) ( Figure 12.7 ).

If we now use the 0.63 mean rank and take the corresponding life value this gives us the
estimate of average life. In this case it corresponds to about 72 uses. To obtain the estimate of
actual average life we use Equation (12.3):

Actual average life 73 72 145 uses

The beauty of this type of analysis is that you can use your service data to constantly update
these three values. Clearly, while all is steady one is happy. If the values start to drop then
it must indicate something has changed and this could be your device, or the way it is being

Life

1 10

100 1000

180

=

Mean rank

100

=

80

0.1

Figure 12.6
Weibull plot for data in Table 12.1(b) .