Sunteți pe pagina 1din 23

Institutional Support of Learning from Accidents:

Some Obstacles to Getting a Useful Community-wide


Database in the EU
SRA 2006 - Annual Conference of the Society for Risk Analysis,
Sep 2006, Ljubljana.
Authors:
Frank Huess Hedlund, COWI A/S, Parallelvej 2, DK-2800 Lyngby, Denmark. fhhe@cowi.dk, and,
Technical University of Denmark, Building 321, DK-2800 Lyngby, Denmark. fhh@imm.dtu.dk
Henning B. Andersen,
Ris National Laboratory DK-4000 Roskilde, Denmark; hba@risoe.dk

1
12.

Institutional Support of Learning from Accidents

Some Obstacles to Ge

Key points

In this paper we examine man's perhaps oldest accident prevention


strategy: learning from past mistakes.

The benefit of such learning is obvious. What is not so obvious,


however, is how to provide institutional support and to set up
systems that facilitate such learning.

We are going to suggest that the well-known database operated by


the European Union, the Major Accident Reporting System (MARS)
may not fully live up to the noble intentions behind its creation. We
provide examples of insufficient information contained in the MARS
database and we offer some suggestions for improvement.

2
12.

Institutional Support of Learning from Accidents

Some Obstacles to Ge

Case example bulk storage of methanol


Case: Existing gasoline tank converted to methanol service

Vapours

Liquid

3
12.

Institutional Support of Learning from Accidents

Some Obstacles to Ge

4
12.

Institutional Support of Learning from Accidents

Some Obstacles to Ge

Two records are


relevant for
bulk storage of
methanol

#113 despite rudimentary description,


this turned out to be an eye-opener

#233 despite rich text description,


the record is not useful

#5 no causes, no lessons learned,


indicates possible scope problems

#6 similar, if not identical, accident recur at same site one year later,
suggesting that no lessons have been learned. Causes still not identified.
Record does not meet the most basic criteria for accident investigation.

Insight - Differences between properties of


motor gasoline and methanol

11
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Inherent explosion hazards of


methanol storage

For all practical matters, the vapour space in a methanol tank is


rarely lean or rich. Most of the time it is ignitable, only waiting for
an ignition source for it to explode violently. This was the major
finding in 1994 case story, and it was a direct consequence of
insights gained from the MARS data.

Experience has shown that it is extremely difficult to eliminate all


sources of ignition.

This hazard is not present in gasoline storage tanks

Local experience and knowledge therefore supported an


uncomplicated and straightforward fire prevention case it was
simply a change of one flammable liquid for another. The only
perceived significant change was that methanol, unlike gasoline, is
classified as toxic. Seveso legislation supported this interpretation.
The methanol storage tanks were subject to Seveso legislation, not
because of methanols flammability classification, but because of its
toxicity classification.

12
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Actual methanol tank explosions

13
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Tank lift-off due to internal explosion

14
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Another tank explosion

15
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Methanol tank explosions are


beyond doubt a major accident

Yet, in 1994 the methanol storage tank was subject to Seveso


legislation (upper-tier), not because of methanols flammability
classification, but because of its toxicity classification.

In 2005, according to Seveso II legislation, a 2,000 m3 methanol


tank is not subject to upper-tier requirements (Column 3 limit is
5,000 tons) safety report is not required

An identical hazard analysis can be performed for ethanol bulk


storage (biofuels). Many existing gasoline tanks are converted into
ethanol service. Yet, column 3 limit for ethanol is 50,000 tons
effectively excluding most ethanol tanks from Seveso II legislation.

Our case is simple. These types of installations exist in all member


states. Would have been of great practical value if the EU database
had contained information on the inherent explosion risks of
methanol (and ethanol).

Q: To which degree does MARS live up to its stated objective of


preventing accidents through information sharing ? (scope/model)

16
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Institution of Chemical Engineers IChemE


(UK) - The Accident Database (TAD)
Short comparative study with IChemEs TAD

17
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Comparative analysis MARS-TAD

Number of records

Hits: records with "methanol"

.. relevant to bulk storage

.. hints at internal explosion hazard

18
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

MARS

TAD

600

13,000

19

129

16

1 (2?)

11

Findings 1
value of info on past accidents

Our case example with bulk methanol storage has shown that some
of the data in some of the Short Reports in MARS have a number of
serious shortcomings. Causes are not identified and lessons
learned are left blank. The value of this information in the efforts
to prevent future methanol accidents is limited - but not nil.

Our case also demonstrates that even the most rudimentary data
on past accidents can be of value in accident prevention. Ironically,
it was the Short Report in MARS with the most meagre data that
gave inspiration to an accident causation hypothesis which
eventually proved useful. Also, the mere knowledge of past
accidents undermined the Companys assertion that it had no
knowledge of any prior accidents involving the bulk storage of
methanol.

19
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Findings 2 less than 30 records per


year enter MARS for the entire EU

Over the course of 22 years the EU database has only accumulated


about 600 Short Reports for all its member states. We consider it
highly unlikely that this number represents all major accidents
according to a common sense definition of this concept for such a
large territory. From the data in our glyoxylic acid case we
hypothesise that scope limitations imposed by strict legal
interpretation of the Seveso legislation may be counter productive
and self-defeating for the Seveso legislations stated objective of
accident prevention through information sharing.

We have identified 114 academic citations that mention the MARS


major accident reporting system. The ratio of citations to accident
records is 1:6, a very large ratio.

20
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

Findings 3 public funding appears


to be crucial to sheer survivability

We have undertaken a comparative analysis of accident records in


MARS and in TAD. TAD suffers from many of the fundamental
shortcomings as do MARS. On all parameters, however, the TAD
seems to be superior to MARS. It is noteworthy, and unsettling,
that despite much targeted advertising there has been very limited
consumer interest in TAD. TAD has failed to be commercially viable
and is now in a dormant state.

It is beyond the scope of this paper to examine or speculate about


the reasons for this. Echoing the title of Kletz 2003 publication,
Still going wrong, we acknowledge that there are many barriers to
evolutionary learning. We conclude by emphasising, however, that
the value of accident reporting schemes is not only determined by
the quality of the data they contain but that public funding appears
to be crucial to sheer survivability. In this respect, the MARS
database may come out on top of the TAD

21
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

22
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

23
12. Some Obstacles to Ge

Institutional Support of Learning from Accidents

S-ar putea să vă placă și