Rail transportation safety investigation R19T0147
The TSB has completed this investigation. The report was published on 05 April 2022.
Table of contents
Canadian National Railway Company
Remote control locomotive system
Dual hump yard assignment YDHF60
Mile 0.0, Halton Subdivision
View final report
On 15 August 2019, at about 0110Footnote 1 Eastern Daylight Time, the Canadian National Railway Company (CN) remote control locomotive system (RCLS) YDHF60 dual hump yard assignment (the assignment) was pulling 82 cars northward along track W100 (W100) in CN’s MacMillan Yard, which is located within the Concord industrial district of Vaughan, Ontario. The assignment was controlled remotely by a single CN yard operating employee (YOE) using a Beltpack. As the assignment negotiated a 15-degree left-hand curve (in the direction of travel), the trailing end of the 26th car behind the locomotives string-lined, causing that car to derail along with the next 8 cars. All 9 cars were empty multi-level vehicular flat cars (i.e., autorack cars). During the derailment, the 27th to 29th cars overturned on their sides to the inside of the curve and pinned the YOE under the 27th car. Consequently, the YOE was fatally injured.
On 15 August 2019 at about 0045, CN 2100 west industrial yard assignment coupled a cut of 34 loaded cars to the south end of 24 autorack cars that had previously been left standing on W100. The cut of 34 cars weighed 3391 tons and was 2074 feet long, and the 24 autorack cars weighed 1245 tons and were 2252 feet long. Five minutes later, the foreman of the CN 2100 assignment informed the trainmaster by radio that they had completed the switching and the trainmaster then notified CN Car Control.
At about 0100, the occurrence assignment arrived at W100 with a cut of 24 cars from track E008 that weighed 1938 tons and was 1652 feet long, and coupled the cut of cars onto the 58 cars standing on W100. With the locomotive hump set and the total of 82 cars joined together, the assignment weighed a total of 7086 tons and was 6166 feet long.
At the time, unbeknownst to the yardmaster, the yardmaster trainee, and the YOE, the cut of 34 loaded cars that had been recently added to the south end of the autorack cars placed 44% of the assignment’s weight in the rear 25% of its length, making it “tail-end heavy.”
After making the joint between the 24th and 25th cars, the YOE walked southward and released the hand brakes on the 3 autorack cars (25th car to the 27th car), then entrained the leading end of the 27th car while the assignment was stopped.
At about 0108, the assignment began to pull the 82 cars northward through the 15-degree left-hand curve en route to the west pullback track. About 2 minutes later, there was a communication failure between the controlling locomotive and the Beltpack. Following the loss of communication and as designed, the RCLS system automatically placed the locomotives in idle, made a full service air brake application, applied the locomotive independent brakes and brought the assignment to a controlled stop.
The site observations were all consistent with classic string-line derailment characteristics.
The investigation determined that the trailing end of the 26th car string-lined; the left side wheels of the trailing truck climbed the rail, derailing the car to the inside of a 15-degree curve. As the 26th car derailed, it pulled laterally on the A-end coupler of the 27th car. This caused the 27th, 28th, and 29th empty autorack cars, which remained coupled together, to roll off their trucks, separate from the head-end cars, and overturn on their side about 2 feet from the track. The YOE, who was riding on the left side of the 27th car, was pinned beneath the leading A-end, and fatally injured.
The YOE’s actions while switching before the accident, and during RCLS operation using the Beltpack, were in accordance with company requirements and his training.
Operation of remote control locomotive systems
An RCLS operating system is programmed in such a way that the selected operating speed is attained as quickly as possible while operating within the parameters that are programmed into the RCLS to determine throttle output. Once the selected speed is attained, the RCLS automatically controls the locomotive throttle and brakes to maintain the speed.
While the RCLS’s aggressive throttle modulation is dependent on the differential between the current speed of the consist and the selected speed, it can also occur if the operator makes a series of small progressive speed increases or immediately selects Max speed (15 mph). As a result, at the time of the occurrence, RCLS operators, like the occurrence YOE, did not have the ability to directly control incremental locomotive throttle changes to facilitate the slow, smooth acceleration of an assignment.
Ratio of lateral-to-vertical forces and potential for derailment
A combination of lateral (L) and vertical (V) forces exists at the wheel-rail interface. The ratio of lateral-to-vertical (L/V) forces indicates the potential for a derailment to occur. When a high lateral force and low vertical force are present (e.g., as with an empty car), the high lateral force will tend to push the wheel flange up and over the gauge face of the rail, resulting in a wheel-climb derailment.
A single-wheel L/V ratio in excess of about 0.82 is indicative of the potential for a freight car wheel to climb (or lift) onto the head of a rail and cause a derailment. Empty long cars equipped with hydraulic long-travel end-of-car cushioning devices (EOCCDs), such as the autorack cars in this occurrence, are particularly vulnerable to these forces. A truckside L/V ratio in excess of 0.65 is indicative of the potential for a freight car truck to roll a rail and cause a derailment.
When a train is pulled through a curve, draft (tension) forces tend to stretch or “string-line” the train as wheel flanges are pulled taut against the inside rail of the curve. If the draft force generated by the locomotives is excessive or if there is a significant run-out of train slack, the L/V ratio can reach a critical level that results in a car wheel climbing, or lifting, onto the head of a rail and subsequently derailing.
Train dynamic simulations
Dynamic simulations are theoretical and are often performed in support of derailment investigations. One of the goals for any dynamic simulation is to identify the combination of the factors and forces that produce results which most closely match the physical evidence observed on an accident site. In this case, 7 different simulations were conducted.
Simulations 1 to 3 confirmed that:
- the tonnage added to the tail-end of the autorack cars on W100 left the assignment “tail-end heavy” with the lighter empty autorack cars located in a vulnerable position, and caused both single-wheel and truckside L/V ratios to exceed critical thresholds for the 26th to 28th cars.
- an emergency brake application on the 63rd car recorded the highest single-wheel L/V ratios at the left side wheels of the trailing truck of the 26th car, and created the specific conditions for the trailing end of the 26th car to string-line.
Simulations 4 to 7 evaluated other mitigating strategies that could reduce the risk of a string-line derailment occurring.
Train dynamic simulations confirmed that the combined effects of aggressive throttle response due to RCLS programming, the vulnerable placement of empty autorack cars equipped with EOCCDs between 2 heavier cuts of cars, the weight of the trailing cut of 34 cars added to the assignment behind the autorack cars, and the air brakes that likely remained on the 63rd car, created the circumstances for this accident to occur.
Ensuring safety-critical procedures are carried forward
After 2 similar string-line accidents that occurred on W100 in 2012 and 2013, CN investigated the accidents and implemented corrective actions. Although procedures were implemented to prevent string-line accidents from occurring on track W100, these procedures were not effectively documented or carried forward, and their use was discontinued, which contributed to this accident.
Safety action taken
Transport Canada (TC) conducted an investigation under the Canada Labour Code, Part II (CLC II), and issued 2 directions to CN. TC reviewed CN’s corrective measures and deemed them to be satisfactory. The investigation results were shared with CN and its workplace health and safety committee as per CLC II requirements.
Canadian National Railway Company
CN issued Notices No. 1908-15 and 1908-21 that contained revised instructions for S-Yard industrial released cars, pulling cars on W100 track and train handling while pulling cars from W100 track.
CN training material was updated to highlight hazard areas for tracks with high curvature and instruct employees to ride either the locomotive or the trailing car on tracks with curves of over 12 degrees.
The 15-degree left-hand curve in track W100 was reconfigured to reduce the track’s curvature from 15 degrees to 12 degrees.
A process was developed to verify that safety-critical information communicated by a notice is also included in the next Summary Bulletin and, if required, the respective yard operating manual.
Working with General Electric and Beltpack manufacturer Cattron Intellectual Property Corporation, changes were made to RCLS programming to allow for a more gradual application of the locomotive throttle during RCLS operations.
August 2019 employee fatality at CN MacMillan Yard, Ontario, was caused by a string-line derailment
Read the news release
TSB deploys a team of investigators to a railway accident at MacMillan Yard in Toronto, Ontario
Richmond Hill, Ontario, 15 August 2019 – The Transportation Safety Board of Canada (TSB) is deploying a team of investigators to the site of a Canadian National railway accident at MacMillan Yard in Toronto, Ontario. The TSB will gather information and assess the occurrence.
Map showing the location of the occurrence
Robert Bruder joined the Transportation Safety Board (TSB) in September 2013 as a Regional Senior Investigator for Rail and Pipeline at the Toronto office. Mr. Bruder has an extensive background in railway operations and risk management garnered over a 36-year career with the Canadian National (CN). He managed CN’s Risk Management Department for Eastern Canada from 2004 to 2013, and was extensively involved in the development and implementation of CN’s safety management system, accident/injury investigation, analysis and cause-finding processes, as well as derailment emergency response and mitigation.
Class of investigation
This is a class 2 investigation. These investigations are complex and involve several safety issues requiring in-depth analysis. Class 2 investigations, which frequently result in recommendations, are generally completed within 600 days. For more information, see the Policy on Occurrence Classification.
TSB investigation process
There are 3 phases to a TSB investigation
- Field phase: a team of investigators examines the occurrence site and wreckage, interviews witnesses and collects pertinent information.
- Examination and analysis phase: the TSB reviews pertinent records, tests components of the wreckage in the lab, determines the sequence of events and identifies safety deficiencies. When safety deficiencies are suspected or confirmed, the TSB advises the appropriate authority without waiting until publication of the final report.
- Report phase: a confidential draft report is approved by the Board and sent to persons and corporations who are directly concerned by the report. They then have the opportunity to dispute or correct information they believe to be incorrect. The Board considers all representations before approving the final report, which is subsequently released to the public.
For more information, see our Investigation process page.
The TSB is an independent agency that investigates air, marine, pipeline, and rail transportation occurrences. Its sole aim is the advancement of transportation safety. It is not the function of the Board to assign fault or determine civil or criminal liability.