Civil Engineering Reference
In-Depth Information
Since the individual components of the safety system are characterized by their
high availabilities and, consequently, very low failure probabilities (p
1
,p
2
,p
3
,
p
4
p
1
) etc. of these safety components
can each be assumed to be approximately equal to 1. Consequently, the sequence
frequency at the upper end of the branching of the event tree of Fig.
6.2
1), the probability of the availability of (1
F
m
,
1
¼
f
m
1
ð
p
1
Þ
ð
1
p
2
Þ
ð
1
p
3
Þ
ð
1
p
4
Þ
f
m
The radioactive release caused in this case, C
m,1
is negligible. On the other hand,
the radioactive releases, C
m,6
or C
m,7
as a consequence of a failure of the electricity
supply followed by a failure of the ECCS and of the integrity of the outer
containment would be very large, because the core would melt and a large fraction
of the radioactive inventory would be released from the outer containment. The
frequency of occurrence, however, of this maximum accident is extremely low,
amounting to F
m,6
¼
p
1
.
In a detailed event tree analysis, many more details must be considered, such as
the individual functions of the ECCS, etc. Interdependencies of the different events
may lead to systematic consequential failures and to the elimination of branches in
an event tree.
f
m
(1
p
1
)
p
2
p
3
f
m
p
2
p
3
and F
m,7
¼
f
m
6.3 Fault Tree Analysis
The fault tree analysis approach is used for numerical assessment of the failure
probabilities of larger units of the safety system. It breaks these larger systems
down into single components, concluding about the failure probability of a larger
unit from the failure probabilities of such individual components by taking into
account the way in which the logical functions of the single components are
interrelated. If common mode failures are possible they must be accounted for.
Often, fault trees must be developed to such detail that available data on single
equipment components or human error can be applied from experience. Uncer-
tainties in reliability data are taken into account by entering not only single values,
but distribution functions for the failure probabilities of single components. For
other components, such as emergency power diesel systems, statistical data directly
available from experience are applied. When determining the failure rate of pipings
under pressure, methods of probabilistic fracture mechanics must be used in
addition.
