SAFRAN comes with a default probability scale with five probability categories: Very Low, Low, Medium, High and Very High. The assumed correspondence between these categories and annual probabilities (expressed in 1/y) and probabilities during the life time of the facility are sown in Table 3.7. This table also shows the expected number of occurrences of the accident during the life time of the facility. The values in Table 3.7 have been derived by assuming that the occurrence of the accidents follows a Poisson distribution. This means that the annual probability of the accidents is constant during the life time of the facility and does not depend on previous occurrences.
To derive the scale we first defined values of the probability (P) that the accident will occur at least once during the life time (TLifeTime) the facility for each probability categories. From these values we calculated values of the annual probabilities (p) using the following equation, obtained from the exponential distribution:
Table 3.7. Probability scale included in SAFRAN (default values are shown).
Definition of columns in Table 3.7:
Columns in Table 3.7 that can be modified by the user: All values in this table can be directly modified by the user.
SAFRAN users can add one or more probability scales, with different probability values and assuming different life times of the facility and specific areas within the facility. The SAFRAN default values of annual probabilities (column 3) shown in Table 3.7 have been derived from assumed values of the probability during lifetime of the facility (column 2) using equation 3.10 and assuming a lifetime of the facility of 100 years. SAFRAN does not calculate annual probabilities from specified values of probabilities during the lifetime of the facility. All values in the probability scale have to be entered by the user. In order for SAFRAN to be able to establish a relationship between probability values and qualitative probability categories the probability values should be entered in the same format as in the default probability scale. The name of the qualitative categories can also be changed by the user.