[SystemSafety] Lac-Megantic disaster

[Peter Bernard Ladkin]

On 7/11/13 7:55 AM, Nancy Leveson wrote:
> I am confused. There have been hundreds of hazardous materials accidents involving train derailments
> in the U.S. alone, many very serious and going back for decades. It does not seem hard to predict
> something that has happened so often in the past.

You may be missing the point I was trying to make. Let me be more specific.

Let's take the 30 years 1980-2010.

July 15, Dunsmuir, CA. Derailment caused HazMat to leak into the Sacramento River and killed 
everything downstream (the Sacramento feeds into SFO Bay but the leak was dammed well before this). 
Complex line running through mountains; derailment caused by the catenary effect.

June 30, 1992, Superior, WI. Benzene leak.

March 17, 1993. Fort Lauderdale, FL. Level-crossing collision with tanker, which exploded. 6 killed, 
12 injured.

Dec 30, 1999 Mont St.-Hilaire, Quebec. Hazmat release after derailment. 2 killed, many evacuated; 
large fire burnt for days

MAy 15, 2001, Toledo, OH. Runaway train with HazMat. Was stopped.

July 18, 2001. Howard Street Tunnel fire in Baltimore.

Jan 21, 2002, Minot, ND. Hazmat derailment near town. 1 killed.

June 28, 2004, Macdona, TX. HazMat leak after collision. 4 killed, many injured.

Jan 6, 2005, Graniteville, SC. HazMat leak after derailment (chlorine). 9 killed, 250 injured.

Oct 15, 2005, Texarkana, AK. 1 killed

Oct 10, 2007, Painesville, OH.

That's 11 incidents with HazMat involvement, and includes collisions, not just derailments. Your 
"hundreds" seems an order of magnitude too large.

Of these incidents, six were within or near to towns. And only three involved significant deaths (9, 
6, 4, 2, 1, 1) or injuries (250, "many").

There is a role for people such as Clarke and Perrow who say "look, here is what has happened; here 
is the state of the engineering/technology/operations - the stuff that we do now. Here is what could 
happen, given this state of the practice".

That is what engineers should be doing, to my mind. It is a formal part of Hazard Analysis.

The obvious engineering response is that there should be procedures in place and enforced to 
mitigate this/these hazard(s).

Whereas one suspects that people tolerate 11 incidents in 20 years with an average of 2 deaths per 
incident. I would imagine someone has likely argued that passenger-train collisions are worse, since 
(a) there are more of them, and (b) you tend to get many more killed.

The point about the Worst-Case analysis, the HazAn point, is that, well, yeah, the history might 
show that passenger train collisions have been worse, but current US rail HazMat operations have the 
potential for far worse. As indeed has now happened in Lac-Megantic, with at least 30 dead. And the 
constellation of engineering practices that led to that appear to be, by European standards, 
astonishingly lax.

Hence my statement:

>     It should give engineers pause that sociologists are better at identifying hazards than they are.

> Again, I don't see how this is true. There are U.S. government agencies that existed before their
> book that explicitly deal with this hazard.

Which agencies have apparently been unable to introduce regulations to ensure that an engineer 
cannot leave running equipment overnight; that introduce effective train-company engineering 
involvment when an emergency occurs in the vicinity (the small fire around the parked train in 
Nantes); and that ensure that trains with Hazmats are dependent only on passive measures to ensure 
they stay in place when parked (for example, you don't park them unattended uphill of another town, 
through the center of which the tracks run).

> And I know of at least one university research center
> which studies these types of accidents to try to prevent them (at Texas A&M I think).

In the Waterfall and Glenbrook accidents in Australia, as well as Ladbroke Grove accident in the UK, 
sociologists and organisational theorists were explicitly involved as witnesses to the Royal 
Commissions investigating these accidents. The question is: why, when all the issues were "known", 
does the situation which allowed these accidents to happen persist? I propose that that is a 
question currently best answered by sociologists and organisational theorists. Unfortunately not be 
engineers. I would welcome a world in which the contributions of people such as Clarke and Perrow 
were understood and accepted in engineering. And I mean practical engineering, by people who run, or 
regulate, say rail companies. Or atomic power stations.

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de