In the early hours of July 6, 2013, an unattended Montreal, Maine and Atlantic Railway train carrying 72 loaded tank cars of crude oil rolled down a grade into Lac-Megantic, Quebec. It derailed in the downtown and burned, and 47 people died. The Transportation Safety Board of Canada released its final report in August 2014. This guide relies on that report rather than the news coverage, which got the cargo and the release sequence wrong in ways that still circulate. It looks at hazard identification, cross-border mutual aid, a town with no capacity for a fire of that size, and what commodity flow studies can tell a planner.

The night of July 5 and 6, 2013: the sequence as the TSB reconstructed it

The train was operated by a single engineer. It consisted of five locomotives, two non-tank cars and 72 tank cars loaded with crude oil from North Dakota, bound for a refinery in Saint John, New Brunswick, on a route that took it from Montreal across Quebec and through northern Maine. On the evening of July 5 the engineer parked the train on the main track at Nantes, about seven miles west of Lac-Megantic, where the line descends toward the town. He applied hand brakes and left the lead locomotive running so that its compressor would keep air in the locomotive brake system, then went off duty to a hotel. The TSB found that the number of hand brakes applied was not enough to hold the train on that grade without help from the air brakes. It also found that the test meant to confirm the hand brakes would hold had been done with the locomotive air brakes still applied, which meant the test could not show whether the hand brakes alone were sufficient.

Late that night a fire broke out on the lead locomotive, which the TSB found had been running with an unresolved mechanical problem. The Nantes fire department responded, and a railway employee came to the scene. The firefighters shut the engine down, consistent with the railway’s own instructions, put out the fire, and left once it was out. With the engine off, the compressor stopped. Air leaked from the locomotive brake system over the next hour, the locomotive brakes lost their holding force, and the hand brakes could not hold the train by themselves. Shortly before 1:00 a.m. the train began to roll toward Lac-Megantic.

According to the TSB, the train reached about 65 mph and derailed at about 1:15 a.m. on a curve near the center of town, where the track speed was far lower. Sixty-three tank cars derailed, most of them breached, and the TSB puts the release at about 6 million litres of crude oil, which ignited. The fire and explosions destroyed much of the downtown core. The death toll of 47 is consistent across the TSB report, the Quebec coroner’s work and government accounts, and unlike the casualty figures for many of the historical disasters on this site, it is not in dispute. I had no role in the response or the investigation, and what follows comes from the published record.

The TSB report and the early coverage: which record to use and which stories to drop

The TSB is Canada’s independent safety investigator for transportation accidents. Its mandate is to find out what happened and why, and to recommend changes, and it does not assign civil or criminal liability. That separation matters here because the disaster also produced criminal charges, a railway bankruptcy and years of litigation. Three railway employees were tried on criminal negligence charges arising from the deaths, and a jury acquitted all three in January 2018. Anyone who reads courtroom coverage and the TSB report side by side will find them answering different questions, and the TSB report is the one built to explain the mechanism.

The coverage in the first days got several things wrong, and some of those errors still appear in training material. Early stories often described the brakes as having failed, when what the TSB documented was a train secured in a way that depended on a running locomotive, which then stopped running. Early stories also tended to treat the cargo as ordinary crude and to repeat the classification on the shipping documents as though it described the oil, which the TSB later found it did not. The order of events at Nantes was also told in several conflicting versions before the investigation established it.

The TSB report is a safety investigation of why the train ran away and why the cars released their contents. It gives the firefighting and evacuation much less space than it gives the mechanical, operational and regulatory story, and in my reading it does not attribute any of the deaths to the conduct of the response. The response lessons in this guide are partly the report’s own findings, such as its finding on emergency response assistance plans, and partly planning questions I draw from the circumstances. I identify which is which as I go. For the complete record, read the TSB’s report, which the board publishes on its website, in preference to any summary, including this one.

The crude on the documents and the crude in the cars: hazard identification

The shipping documents identified the cargo as petroleum crude oil, UN 1267, a Class 3 flammable liquid, assigned to Packing Group III. Packing groups rank the degree of danger within a hazard class, and Packing Group III is the least hazardous of the three for flammable liquids. The TSB had samples of the oil tested after the accident, and the results met the criteria for Packing Group II, which means the oil was more volatile than its paperwork said. The TSB found that the crude had been misclassified, and that the sampling and testing behind the classification were not adequate for a product whose properties could vary from one source to the next. Its broader concern was that the hazard had not been accurately communicated to the railways, regulators and communities along the route.

The trap for a responder is that the packing group does not appear on a placard. A rail tank car of UN 1267 carries the same Class 3 placard and the same identification number whichever packing group the shipper assigned, and the Emergency Response Guidebook points to the same guide, Guide 128, for all of it. That guide already calls for isolating 800 meters, or half a mile, in all directions when a rail car is involved in a fire. So the placard information, read correctly, would have pointed a responder toward the right broad tactics on the night, and the misclassification did its damage earlier, in what everyone upstream of the fire department assumed about how this oil would behave.

In 2013, crude from the Bakken formation in North Dakota was moving by rail in quantities that had grown quickly over a few years, and the industry and regulators were still catching up with its properties. The general point for planners outlasts that particular commodity. The identification number tells you what a material is called and which guide to open, but it does not tell you how a specific batch will behave in a fire, and the packing group printed on a shipping paper is only as reliable as the testing behind it. A plan that sizes its isolation zones and water needs from the paperwork alone inherits whatever error is in the paperwork.

There was one more identification problem specific to a runaway. The engineer and the train documents were not with the cars when they arrived in town, so nobody at the derailment could hand the first-arriving officer a consist. In Lac-Megantic the crude trains were a familiar sight and the cargo was not a mystery, but a jurisdiction that relies on train crew contact for its first information has no fallback when there is no crew.

Six million litres in a downtown: why no small fire service could fight it

Lac-Megantic is a town of roughly 6,000 people, and its fire service was sized for the fires such a town normally has, which are structure fires, vehicle fires and the occasional commercial building. Nothing in that service, and nothing in the fire service of any town of that size I am familiar with, was built to extinguish millions of litres of burning crude spread through a downtown, with breached tank cars still feeding the fire and the flowing oil reaching drains, the lake and the Chaudiere River. Burning crude of that volume calls for foam in quantities that municipal departments do not keep, delivered at application rates that require specialized equipment, and it calls for water supply beyond what a downtown grid can provide while the same grid is being damaged by the fire.

The tank cars made the scale worse. The 72 cars were of the type known in Canada as Class 111 and in the United States as DOT-111, and the TSB found that the design’s weaknesses in a derailment had been known for years. At Lac-Megantic most of the derailed cars breached, which turned a derailment into a release of nearly the whole cargo within a short time, in the middle of town, close to occupied buildings. The TSB’s January 2014 recommendations began with stricter standards for tank cars carrying flammable liquids, so that a derailment releases less product.

For a small department, the realistic doctrine for a fire like this is defensive. That means isolating and evacuating at the distance the guide calls for or greater, protecting exposures where water allows, and keeping firefighters away from tank cars exposed to flame, while the fire is allowed to burn until enough foam, water and specialized help have arrived to do more. That doctrine is sound, but it means the town loses whatever is inside the zone, and it means the most important work happens before the derailment, in decisions about what passes through the downtown and where the help will come from. Planners should take the duration of the fire and the timeline of the response from the TSB report and the Quebec authorities’ accounts, because figures repeated in secondary sources vary.

Help from across the region and across the border: the planning questions it raises

Lac-Megantic sits less than an hour’s drive from the Maine border, and fire departments from across the surrounding region responded, including companies that came across from Maine. The TSB report does not analyze cross-border mutual aid in any depth, and I am not aware of a published finding that it failed at Lac-Megantic. So this section deals with the questions the response raises for any agency near an international border, and should not be read as a list of things that went wrong in Quebec.

The legal questions come first, because they are slow to answer during an incident. A firefighter crossing a border needs authority to act in the other jurisdiction, and the department sending them needs to know who carries liability for their actions and who pays if they are injured. At the state and provincial level in the northeast, the International Emergency Management Assistance Memorandum of Understanding among New England states and eastern Canadian provinces, Quebec included, provides a framework. The Emergency Management Assistance Compact that governs interstate aid in the United States does not reach into Canada. A volunteer company in a border town usually depends on whatever local agreement its own municipality has signed, if it has signed one at all, and on how its own workers’ compensation and insurance treat out-of-country work. The border crossing itself, with apparatus, personnel and in some cases hazardous materials such as foam concentrate, is easier to clear in advance than at 2:00 a.m.

The operational questions follow. Radio is the one I know best. A Quebec municipal fire service and a Maine fire department operate on systems licensed by different national regulators, on frequency plans built separately, and they share no talkgroup or channel unless someone has arranged it beforehand. Use of frequencies near the border is subject to coordination between the U.S. and Canadian regulators, so the American national interoperability channels cannot simply be assumed to be available on the Canadian side. Language is a separate issue in Quebec, where the host agency works in French and many arriving American crews work only in English. Hose coupling and equipment compatibility deserve a check before an incident rather than during one. A departmental agreement that addresses these points in a few pages is worth more than a regional framework that nobody at the company level has read.

The industry backstop that did not exist: response assistance plans and the 2014 recommendations

Canada’s Transportation of Dangerous Goods Regulations require shippers of certain dangerous goods to have an Emergency Response Assistance Plan approved by Transport Canada. An ERAP commits the shipper to provide specialized response help, meaning trained people, equipment and technical advice, to an incident involving its product anywhere along the route. The TSB found that crude oil shipped in rail tank cars did not, at the time, require an ERAP. A town like Lac-Megantic therefore had no shipper legally obliged to bring foam, firefighting capability and product specialists to its fire.

In January 2014, before its final report was released, the TSB issued three recommendations. The first called for stricter tank car standards, discussed above. The second called for route planning and analysis for trains carrying large volumes of dangerous goods, so that railways would weigh the risks of the routes they use. The third called for ERAPs to be required where large volumes of liquid hydrocarbons, such as crude oil, are shipped. The final report in August 2014 added two more recommendations, one on physical defences to keep unattended trains from running away and one on Transport Canada’s auditing of railway safety management systems.

Transport Canada later moved to require response assistance plans for large shipments of flammable liquids, including crude. The details have changed over time, so verify the current requirement in the Transportation of Dangerous Goods Regulations or with Transport Canada before writing it into a Canadian plan. The United States has no direct equivalent to the ERAP. American planners rely instead on the railway’s own hazardous materials response resources and contractors, the 24-hour emergency response telephone number required on shipping papers, and services such as CHEMTREC, so the planning question for a U.S. jurisdiction is how long the railway’s resources would take to reach a given point on the line, and that figure should come from the railway in writing.

Commodity flow studies: what they tell you and what they leave out

A hazardous materials commodity flow study is an attempt to describe what dangerous goods move through a jurisdiction, by which mode, along which routes and in what rough volumes. In the United States, local emergency planning committees often carry them out, frequently with Hazardous Materials Emergency Preparedness grant funding from PHMSA. The methods combine data the carriers provide, roadside placard counts on highways, information on fixed facilities and pipelines, and data from state agencies. The result feeds the hazard analysis, the training plan and the decisions about where to place resources.

For rail, the main source in the United States is the railways. Under the Association of American Railroads’ Circular OT-55, a railway will, on request, give a bona fide emergency response agency a list of the hazardous commodities it moves through the community, typically the top 25 by volume over a year, and the AskRail application gives qualified responders the contents of a specific train during an incident. The information is security sensitive and is provided on condition that it be held confidentially, so an agency needs to decide who holds it and how it reaches the people writing the plan. In Canada, Transport Canada directed railways in late 2013 to share annual dangerous goods information with the municipalities they pass through, and Canadian planners should check the current form of that requirement. In May 2014 the U.S. Department of Transportation issued an emergency order requiring railways to notify state emergency response commissions of trains carrying large quantities of Bakken crude through their states. Verify how that notification requirement stands today with PHMSA and your state emergency response commission.

Commodity flow data has limits that Lac-Megantic illustrates. It reports commodities by name and identification number, so it would have described the Lac-Megantic cargo as UN 1267 crude oil and would have said nothing about a batch more volatile than its classification. It is a historical aggregate, which means it can lag behind a rapid change in traffic, and rail crude in the early 2010s was a change of exactly that kind. Annual totals also say little about how cargo moves, and the operational facts that shaped this disaster, such as long unit trains of a single commodity, single-person crews and unattended trains parked on a grade above town, appear in no commodity table. A study tells a planner which hazards to prepare for, and it needs to be paired with a conversation with the railway about operations on the specific line.

Planning for a transported hazard you cannot extinguish

The physical facts at Lac-Megantic were visible on a map well before 2013. The line descended toward the town from Nantes and curved through the downtown, which put the place where a runaway or a high-speed derailment would end directly among occupied buildings and close to the lake. A transported hazards annex should identify those locations along every rail line and major highway route in the jurisdiction, meaning grades, curves, sidings and parking locations, and crossings of dense occupancy, schools, hospitals and water supplies, and for each one state what an 800-meter isolation zone would contain and who would evacuate it.

The annex should be honest about capacity. For a small jurisdiction, the plan for a large crude or ethanol train fire should say plainly that local forces will isolate, evacuate and protect exposures, and that extinguishment depends on outside resources, with each of those resources named, its source identified and its realistic travel time stated. Foam is the obvious example, because a fire on the scale of Lac-Megantic needs far more than a municipal department carries, so the plan should state where the nearest large supplies are held, whether at airports, refineries, terminals or regional caches, and how they would be requested. Water supply, drainage and the route by which spilled product would reach a stream or lake belong in the same annex, written with the public works and utility staff who know the system.

Regulation changed after 2013 on both sides of the border. Transport Canada issued an emergency directive on unattended trains within weeks. The United States adopted rules in 2015 for what it calls high-hazard flammable trains, covering tank car standards, braking and routing analysis, and both countries set schedules to retire the older tank cars from flammable liquid service. Those changes reduce the likelihood and size of a release, but they do not move the tracks out of downtowns or add capacity to small fire departments. Emergency managers should take the current requirements from PHMSA, the Federal Railroad Administration and Transport Canada rather than from articles written when the rules were new.

What to do at your agency

• Have your emergency manager or LEPC chair submit a written request to each railway operating in the jurisdiction for its current hazardous commodity list under Circular OT-55, and record in the hazard analysis the date received and who holds the confidential copy.
• Have the planner responsible for the hazardous materials annex mark on a map every rail grade, curve, siding and train parking location within the jurisdiction, and write one paragraph for each naming the occupancies inside an 800-meter radius.
• Have your fire chief or hazmat coordinator ask the railway in writing for the expected arrival time of its hazmat contractor at a named location on the line, and enter the answer and the date in the annex.
• Have your training officer add a 15-minute segment to the next scheduled hazmat refresher showing that the packing group appears on the shipping paper but not on the placard, using the Lac-Megantic misclassification as the example.
• Have your radio system administrator confirm, for each neighboring agency across a state or international border, which channel or talkgroup the two agencies would share at a fire, and test it with one live radio check this month.
• Have the emergency manager list the nearest sources of large foam quantities, including airports, refineries, fuel terminals and regional caches, and confirm each one with a named contact and a request procedure.
• Agencies near an international border should have their fire chief ask the municipal attorney for a copy of any cross-border mutual aid agreement, and add a review of its liability and workers’ compensation terms to the agenda of the next regional chiefs’ meeting.

Takeaways

• On July 6, 2013, an unattended train of 72 crude oil tank cars ran away from Nantes, derailed in downtown Lac-Megantic at about 65 mph according to the TSB, and released about 6 million litres of crude that burned and killed 47 people.
• The TSB’s final report, released in August 2014, is the record to use, because early news coverage got the cargo and the release sequence wrong in ways that still appear in training material.
• The story that the Nantes firefighters caused the disaster is a myth, because the TSB found the train had been left with too few hand brakes to hold it without air from a running locomotive.
• The crude was documented as Packing Group III, while TSB testing found properties meeting Packing Group II, and because the packing group does not appear on a placard, responders cannot see that kind of error from the outside.
• No fire service in a town of roughly 6,000 had the foam, water or equipment to extinguish a fire of that size, so the realistic local role in such an incident is isolation, evacuation and exposure protection while outside resources arrive.
• Crude in rail tank cars did not require an Emergency Response Assistance Plan in Canada in 2013, and the TSB’s January 2014 recommendations addressed that gap along with tank car standards and route planning.
• Commodity flow studies and railway commodity lists tell a planner which materials pass through, but they say nothing about batch-specific volatility or operating practices, so they need to be paired with written answers from the railway about its operations on the specific line.
• Agencies near an international border should settle authority, liability, radio interoperability and language for cross-border aid in advance, because none of those questions can be answered quickly during an incident.