Backgrounder

Investigation findings for TSB investigation (R15H0021) into the March 2015 derailment and fire of a Canadian National crude oil train near Gogama, Ontario

Investigations conducted by the Transportation Safety Board of Canada (TSB) are complex – an accident is never caused by just one factor. The March 2015 derailment of a Canadian National Railway (CN) crude oil unit train near Gogama, Ontario was no exception. There were many factors that caused this accident, the details of which are contained in the 13 findings as to causes and contributing factors. Furthermore, there were 9 findings as to risk as well as 11 other findings.

Findings as to causes and contributing factors

1. Prior to the arrival of the train, a 16-inch-long portion of the parent south rail head had broken off due to a vertical split head rail failure within the east joint of a plug rail repair, leaving a gap in the south rail.
2. The derailment occurred when the south rail failed catastrophically beneath the train as it traversed the track, resulting in the derailment of the 6th to 44th tank cars, which were loaded with petroleum crude oil.
3. During the plug rail repair 3 days before the derailment, the parent rail was cut and the exposed rail ends were visually inspected for cracks with no defects noted. However, the snow patrol foreman did not perform a dye penetrant test on the cut rail ends as required by CN Engineering Track Standards section 1.7.
4. No specific guidance was provided to CN engineering employees relating to the length of grinding required when a rail end mismatch occurs during plug rail repairs.
5. The 2.5 inch (60 mm) long transition zone created by grinding the plug rail was ineffective and created an abrupt change in rail head height that increased the dynamic loads applied to the east end parent rail head, which also contained a vertical split head defect, and caused the rail to fail.
6. Given the state of the plug rail repair and short transition between rail end mismatch at the east end joint, a slow order should have been placed on the track.
7. The snow patrol foreman was aware of the dye penetrant test but had never performed one before or seen it performed.
8. The online training at CN relating to dye penetrant testing did not specifically highlight the importance of the test as part of plug rail repairs or provide opportunities for practical hands-on training.
9. The large quantities of spilled product, the rapid release of the product, as well as the product's high volatility and low viscosity, contributed to the ignition of large post-crash fires and the pool fire.
10. The absence of tank car thermal protection likely increased the severity of the product release, further fuelling the fire as 15 tank cars sustained thermal tears after exposure to the pool fire.
11. The tank car bottom outlet valve handle arrangement was inadequate to protect against product release during the derailment and contributed to the severity of the release.
12. The speed of the petroleum crude oil unit train increased the severity of the outcome.
13. The severity of the outcome at 43 mph suggests that speed restrictions of 50 mph, which were in place at the time of the accident, would not reduce the severity of a derailment and are not sufficient for unit trains transporting Class 3 flammable liquids.

Findings as to risk

1. If online training for safety-critical tasks is not reinforced by practical training, trainees may not fully comprehend the importance of critical steps within the task, increasing the risk that the task will not be adequately performed.
2. If infrequently performed safety-critical tasks are conducted from memory without the aid of a checklist or independent verification, steps that are important to properly complete the task can be inadvertently omitted, increasing the risk that the task will not be adequately performed.
3. If bottom outlet valve handles continue to be exposed without adequate protection, there is an increased risk of product release during a derailment and site remediation.
4. If flammable liquids continue to be transported in tank cars that are not sufficiently robust to prevent catastrophic failure when involved in an accident, the risk of dangerous goods release during a derailment will remain high.
5. If the new tank car standards are not fully implemented in a timely manner, there is a continued risk of product loss and associated consequences when tank cars carrying flammable liquids are involved in a derailment.
6. If train speed is not adequately restricted for unit trains transporting Class 3 flammable liquids, there is an increased risk of product release and adverse consequences when the train is involved in a derailment.
7. If emerging localized surface collapse, rail end batter, and crushed head rail surface conditions are not fully considered as part of the risk assessment used to plan for regulatory inspections and for railways to develop rail replacement programs, there is an increased risk that problematic sections of rail may not be identified and remediated.
8. If risk assessments do not adequately consider increases in traffic tonnage, the use of heavier rail cars, and the potential for more rapidly degrading track structure, regular track maintenance activities may no longer be sufficient to maintain track to the required standards, increasing the risk of track infrastructure failures.
9. If the risk-based approach to planning regulatory track inspections does not consider all relevant operational factors, including increases in rail traffic tonnage, increases in dangerous goods volumes and emerging leading indicators of track degradation such as localized surface collapse, rail end batter, and crushed head rail surface conditions, the targeted inspections may not be well focused, increasing the risk that degrading track conditions will go undetected.

Other findings

1. Had a dye penetrant inspection been performed, a vertical split head indication would likely have been observed in the cut east rail end, and more of the east parent rail may have had to have been removed.
2. Although clearly visible in the broken thermite weld, the vertical split head defect that caused the rail to fail was either not present or too small to be detected during the ultrasonic test conducted on 02 March 2015 (i.e., 2 days before the broken rail occurred and 5 days before the derailment).
3. The extent to which the jacket and insulation were effective in delaying the internal build-up of pressure could not be determined.
4. For the 15 tank cars that exhibited thermal tears, there was no evidence to support the hypothesis that pressure relief devices with higher start-to-discharge pressure result in more energetic (larger) thermal tears.
5. The 4 CPC-1232 compliant tank cars which were equipped with jackets, insulation and full head shields were more effective in protecting the tank heads against impact punctures during the derailment compared to the half-head shields.
6. The small number of tank cars with breached manways, top fittings, and pressure relief devices suggests that the features incorporated for the protection of top mounted appurtenances were generally effective in reducing the release of product.
7. While 6 of the tank cars separated at the stub sill attachment, none of the stub sill attachment separations caused a breach in a tank.
8. The majority of towns located along a rail line do not meet the criteria of a census metropolitan area and therefore the 40 mph speed restriction of the Rules Respecting Key Trains and Key Routes does not apply.
9. CN's safety management system relied on reactive indicators and did not anticipate the need for increased track maintenance in light of significant increases in dangerous goods volumes and traffic tonnage. 
10. Despite the challenges of responding to a major rail accident and subsequent fire, the emergency response was effective and inclusive.
11. Although the environmental plan was comprehensive and significant mitigating strategies were put in place, site monitoring is ongoing and concerns remain about the possible contamination of the watershed.