probabilistic nets provide an enormously strong tool for decision analysis, including prior
analysis, posterior analysis and pre-posterior analysis.
In the following the use of Bayesian nets for different purposes in risk assessment will be
illustrated by examples. All examples can be calculated using free demonstration software on
BPN such as e.g. Hugin Lite downloadable from: http://www.hugin.com.
Example 10.2– Classical fault tree and event tree risk analysis by Bayesian Probabilistic
Nets
When Bayesian probabilistic nets are applied for the analysis of the reliability of systems as a
substitute for fault trees and or event trees their use follow straight-forwardly from the
descriptions in the foregoing.
Consider a power supply system composed of an engine, a main fuel supply for the engine
and electrical cables distributing the power to the consumers. Furthermore, as a backup fuel
support a reserve fuel support with limited capacity is installed. The power supply system fails
if the consumer is cut off from the power supply. This in turn will happen if either the power
supply cables fail or the engine stops, which is assumed to occur only if the fuel supply to the
engine fails.
A Bayesian network based system model for the power supply is illustrated in Figure 10.5. In
Figure 10.5 also the unconditional probabilities for the parent events and the conditional
probability tables for the children events are illustrated.
Figure 10.5: Illustration of Bayesian nets for the power supply risk analysis.
All probabilities given in the network are for simplicity assumed to be annual probabilities.
Executing the Bayesian network now provides the probability structure for the different states
of the system as illustrated in Figure 10.6.