Page 132 - 360.revista de Alta Velocidad - Nº 5

Grande, Zacarías. Blanco López, Marta. García Tamames, Alberto. Castillo, Enrique.

Note also that the limit between “Tolerable” and “Intolerable”, indicated by the line in red, is
between 9 and 10 and that the limit between “Acceptable” and “Tolerable”, indicated by the
bottom line in red, is between 6 and 7.
Consequently, the above Table 4 significantly improves the European standard 50126 and the
Spanish standard UNE-EN-50126, since:

1. It uses seven frequency levels instead of six.
2. Quantifies the frequencies accurately.

3. It uses frequency levels (proportional to 5 where n is the level) that multiply by five the
n
frequencies of the previous ones, which allows levels to be associated with frequencies,
that is, qualitative with quantitative information, which does not occur neither in European
standard 50126 nor in the Spanish UNE-EN-50126.
4. Quantifies the consequences in terms of deaths and using also a factor 5 to pass level, which
avoids ambiguities.
5. Declassifies as “acceptable without any agreement” (“Negligible risk”) low frequency cases
with serious consequences
Nevertheless, the use of tables, such as Table 5, can also be criticized because they provide
a risk per year and this depends on the number of trips per year and its length (the larger
the number of trips and its length, the larger the risk). In other words, using this yearly risk
discriminates the different events by length and number of trips. In our opinion the risk must be
given in fatalities per kilometer. The fact that a user travels more time per year or uses longer
or shorter trips should not change the required safety level.

The most important conclusions that can be deduced from the above are:

1. The European standard 50126 and the Spanish UNE-EN-50126 are not the most appropriate,
and if used could easily lead to the conclusion that the risks associated with the event,
“very unlikely with serious consequences” are “acceptable without any agreement”, that
is, do not need a safety analysis, if such event had been considered “Incredible.”

2. This will not occur in the case of using the recommendations indicated by the RSSB, or more
modern methodologies, such as those based on fault trees or in Bayesian networks.

3. The imposition of the ADIF methodology, based on European and Spanish regulations, as
the only accepted ones, giving as a reason the difficulty of comparing results if different
methods were used, should be eliminated, or at least thoroughly reviewed, by the risks
which can be overlooked in its application, which is also mandatory.
4. Finally, the risk should be given by km and not per year.

2. Probabilistic Safety Analysis (PSA)

Once the need of a detailed probabilistic safety analysis (PSA) has been detected we have to
identify all possible risks and proceed to evaluate the whole risk of the line. To this end, we can
use the above method, which is very cumbersome or use an alternative.
In this paper we present a method to simplify the process based on Bayesian networks, which
can be used as the main tool for probabilistic representation of multidimensional variables in
order to analyze railway safety.

130 360.revista de alta velocidad