NPP: Power Grid Mutual Safety Assessment - Nuclear Power Plant Instrumentation and Control Systems for Safety and Security - page 423

The aim of this chapter is to introduce an ap-
proach to series integration of CI safety assess-
ment methods using integration of FL methods
and BBNs under uncertainty.

System criticality as a safety value. A high
criticality Crt S ( ) of a system corresponds to its
marginal (pre-emergency) state, in which its fur-
ther use is prohibited or inexpedient or its recov-
ery to operable condition is not possible or expe-
dient. The main distinction between the margin
state in a reliability theory and a high criticality
in the safety theory is consideration of system
failure consequences in pre-emergency condition.
Criticality assessments may be represented
on qualitative and quantitative scales. This paper
considers linguistic criticality assessments. Thus,
for example, criticality can be represented as lin-
guistic variable with terms {High (H), Medium
(M), Low (L)}.
An illustration of semantic interpretation of
linguistic terms of criticality, condition, e.g., NPP
reactor, is presented in Table 9.
Procedure. In order to apply the suggested
approach one will need: to chose a test object
(system), for which safety rating will be estab-
lished, the result is determination of the child
system for using BBN block; specify which sys-
tems define safe state of the test object; the result
is determination of the parent system for using
BBN block; determine components of the parent
system and parameters of their states; the result
is logic and linguistic model of the parent system
for determining their safety rating in terms of
parameters of components for FLI block.

Joint FL-BBN Assessment
of NPP Safety

General approach. The suggested approach is
based on the following assumptions:

• Any NPP may be represented as a collec-
tion of hierarchical layers of objects, and
namely, systems components and elements;
• Any object in NPP may be represented as
a BBN.

The NPP hierarchy is a basic premise for
representation of its safety assessment integra-
tion methods architecture as hierarchy as well.
This means that parameters of conditions of, e.g.,
elements are used as input data for components
safety assessment. Further these assessments
serve as input data for determining subsystems
safety. In this way the safety assessment runs
from the bottom to the top, from systems of the
lowest hierarchy layer to systems of a higher layer.
The system safety ois a function of safety of its
subsystems, components and elements.
On the other hand, subsystem safety assess-
ments may be unitized in prediction (diagnostics)
of their components condition. In this case safety
assessment runs from the top to the bottom from
systems of the highest hierarchy layer to lower
layer systems.
Consequently, both upward and downward
integration of methods is possible. Such an inte-
gration of different methods results in compensa-
tion of insufficiency of data for models of higher
level due to “excessive” data in another, lower
hierarchy layer.
The hybrid safety assessment method sug-
gested by this approach is given in Figure 16.

1. Fuzzy Logic Inference (FLI) Block:
(Bottom-up analysis) for NPP safety sys-
tems assessment on the basis of parameters
of its components. In order for the block to
solve problems it should have solved the
subtask of selecting the most important
system components, which condition defines
system safety. The task of forming of a set
of informative (essential) parameters , the
values of which allow distinguishing system
conditions, must be solved. The basic data
are deterministic input data - parameters