Common practices of
risk analysis
Common practice for predicting
incidences are by applying various
methods of root-cause analysis, such as FTA, ETA, FMEA, and HAZOP. A comparison
of these methods is available in an article by
Silvianita et al. (2011). Theoretically, such methods can reveal many kinds of
failure modes.
Limitations of the common practices
These methods are quite helpful in failure analysis during investigation, but not for avoiding risks. The main problems with these methods is that it they require huge investment, which makes them impractical. Another
problem is that incidences are often associated with unpredictable
situations, typically due
to slips,
unexpected events,
inter-unit coordination problem and
human-machine mismatch.
There are several explanations why the effect of these methods is
limited:
- Perrow (1984)
explained that accidents in operating
risky technology such as nuclear power
stations are due to
complexity. Root-cause analysis conducted manually by
human beings cannot handle all the combinations of concurrent
faults. The
incidences materialize mainly combinations that were skipped in the analysis.
- Hollnagel (1983) pointed out that the same sequence of events may lead to both success
and failure. Therefore, root-cause analysis is useful only when going
backward, by backtracking prior
incidences, but is useless for predicting
future incidences.
- Taleb (2007)
argued that Black Swan
accidents cannot possibly be predicted, because there
is no data about prior events.
-
Norman (1990) highlighted the problem of information as primary source for the
automation problem.
-
Bainbridge (1983) pointed out that training for
normal operation hampers problem solving in
emergency
-
Zonnenshain and Harel (2009) demonstrated that often, operational
errors are the result
of inconsistency in the
state of the extended system.
-
Zonnenshain and Harel (2015) stressed that the methods above are trigger-oriented, while the analyses of many accidents indicate that the contribution of the
trigger is negligible compared to that of the conditions that enable the
accidents.
Updated on 26 Mar 2017.