If FAR* is less than 1, we conclude that risk is negligible. If FAR* is larger
than 100, we conclude that risk is intolerable, and risk-reducing measures are
required. Now suppose we have indicated an uncertainty factor 10 for the estimate
FAR*. Then if FAR* is larger than 100, we have strong evidence that the
true value FAR is larger than 100/10 = 10. Similarly, if the estimate FAR*
is less than 1, we have strong evidence that the true value FAR is less than
1 × 10 = 10. Thus 10 represents the real criterion for intolerance and negligibility,
respectively. The interval [1,100] is an uncertainty region where the
ALARP principle applies. Decision-makers can draw conclusions about intolerability
(above 100) or acceptance/negligibility (below 1), with the intermediate
24 FOUNDATIONS OF RISK ANALYSIS
region interpreted as tolerable only if risk reduction is impracticable (which
means cost-benefit considerations).
Although such an interpretation seems natural, we have not seen it often
expressed in precise terms in applications.
2.1.3 Reliability Analysis
A reliability analysis can be viewed as a special type of risk analysis or as an
analysis which provides input to the risk analysis. In this section we briefly
review the standard approach for conducting reliability analysis. As this approach
is similar to the one described in the previous section, we will just introduce
the main features of reliability analysis and refer to Section 2.1.2 where
appropriate. We distinguish between a traditional reliability analysis and methods
of structural reliability analysis, as they represent different traditions, the
former dominated by statisticians and the latter by civil engineers.
Traditional reliability analysi ...