Information Technology Reference
In-Depth Information
- Train accidents usually belong to category M while accidents sustained
by persons belong to category P.
This already results in an initial classification of severity categories: M-F,
P-F, M-S, P-S, M-L and P-L. In order to judge this classification, it can be
compared to a common numerical scale, the so-called fatalities-and-weighted-
injuries (FWI) scale, where it is often assumed that one fatality is equivalent
to 10 serious injuries and 100 light injuries, respectively. Table 3 shows that
this approach also seems numerically plausible and quite natural.
Tabl e 3 . Plausibility check for severity categories
Combination
FWI range
Typical
FWI EN 50126
M-F
2FWI
5
Catastrophic
P-F
1 FWI<2
1
Catastrophic/ critical
M-S
0.2 FWI<1
0.5
Critical
P-S
0.1 FWI<0.2
0.1
Critical
M-L
0.02 FWI<0.1
0.05
Marginal
P-L
0.01 FWI<0.02
0.01
Marginal
-
FWI<0.01
n. a.
Insignificant
Often, proposals are made to add an additional severity class on top of
combination M-F or to subdivide M-F into two subclasses (e.g. FWI<10 and
FWI 10). However, this does not seem to be practicable for setting design
targets for technical systems for the following reasons:
- The outcome in terms of FWI for an accident of combination M-F is
more or less random, so it would be very dicult to distinguish between
further subclasses
- EC Regulation 352/2009 already sets the most demanding safety target
for a potentially catastrophic scenario. So, for the process of setting design
target, a further distinction is not needed.
5
Assessment of Accident Scenarios
In the next step, it must be checked whether the classification of accident
scenarios with respect to the proposed severity categories is feasible. In order
to classify accident scenarios, two different approaches are feasible:
1. Expert judgement has to be treated with care as experts often seem to
remember worst cases well and tend to overestimate typical bad outcomes
2. The statistical evaluation of accident data often has problems in that
the data for severe accidents is rare and that sometimes data quality is
questionable
