Page 239 - Design of Simple and Robust Process Plants

6.2 Basic Theory of Reliability 225
. Mean time to repair (MTTR) is the time needed to repair a component; it
includes not only the repair time but also includes the arrival time of the
repair team and the spare parts. The component that failed also needs to be
made available for maintenance and, after repair, the time monitored to put
the component/unit/plant back in operation.
Now the failure rate is defined as:

k i = f i (2)
R i 1
Failure rate k(t) is the probability that a component or system, that survived up to
time t, will fail in the next time interval.
When the failure rate of components is plotted against time, the graph often has
the form of a bathtub (Figure 6.4). The initial zone A, representing the early fail-
ures, commences with rather high values, but these fall rapidly with time. At a cer-
tain moment the curve becomes relatively low and flat (zone B), and this represents
a relatively constant failure rate. After a certain time, the failure rate increases again
(zone C), which indicates wearing out (also known as ªagingº) failures.

Failure rate ()

A C
Early Constant Wear out
failure B failures failures
Time
Fig. 6.4. The failure rate against time (the bathtub curve).

The early failures in zone A, which are relatively high, can be subdivided into the
following categories:

. Design of the component
. Selection of the component
. Fabrication error of the component
. Installation error of the component
. Instrumentation errors of surrounding equipment, nor necessarily being a
part of the component
. Operation mistakes, manual or automatic.
These potential errors are the main reasons why process plant start-up(s) deal with
frequent starts and stops before being operated at a constant production level. Plant
restarts suffer much less (even after an overhaul) when the number of new compo-
nents is much less, although repaired devices might also fail due to maintenance
errors.