Decades of derailment investigations have yielded a body of knowledge that will reduce the number of derailments in the long run

Derailments are never a good thing. But when they occur, as they have on every railroad over the years, they open a window into vehicle / track interaction, albeit poor interaction in these cases. In the a ermath, investigators comb through the wreckage and operating and maintenance records, to identify the cause or causes. And under the right light, the ndings yield lessons that can lead to operating, maintenance, or even regulatory changes. For a look into

some of the lessons learned, Mike Roney, of Iron Mustache Consulting, examined two decades of derailment investigations by the Transportation Safety Board of Canada.

Roney, a former General Manager of Track and Structures, and Chief Engineer at CP Rail, who in 2019 was inducted into the International Heavy Haul Association’s Heavy Haul of Fame for his “lifetime vision and dedication to the advancement of heavy haul railroad technology, engineering and operations,” examined derailment incidents through the lens of the TSB investigators and through the corrective action plans that were provided in response to the TSB’s ndings.

“Twenty years of derailment investigations by the Transportation Safety Board of Canada is a story about how knowledge is advancing and how we are making it work for us,” Roney told delegates at the 2022 Wheel/Rail Interaction conference. “It’s also a story about how there has been collaboration between railways and the regulators to make things better.” e Transportation Safety Board of

Canada (TSB), like the National Transportation Safety Board (NTSB) in the U.S., is an independent federal agency that is there to advance transportation safety by investigating occurrences in air, marine pipeline, and rail. e TSB selects and investigates derailments to determine the cause of incidents that might indicate a trend or indicate a risk to the public. e TSB reports identify the factors that caused or contributed to the occurrence, and the safety de ciencies that need to be addressed. While the TSB does not have a mandate or authority to implement corrective actions, it monitors recommendations and actions of Transport Canada and the railways for mitigating actions that eliminate or reduce the so-called safety de ciencies found in these investigations.

Overall, the TSB investigated 327 derailments between 2000 and 2020; 90 derailments re ecting 28% of the total derailment costs identi ed a track issue as a primary cause; 100 derailments re ecting 31% of the costs were attributed to human factors; 58 derailments re ecting 16% of the costs were attributed to a mechanical-related cause. e majority of the track-related derailments (49%) were attributed to three causes: broken rails, broken joint bars, and broken welds. e remainder were attributed to wide gauge (17%), track buckles (12%) wheel climb and rail rollover (10%), geotechnicalrelated (7%) and geometry defects, other than wide gauge (5%). “ at’s actually a very good result for the track geometry departments,” Roney said. irty ve derailments in this sample identi ed broken rail as a primary cause; 21 of the derailments, the majority of which were prior to 2015, resulted from undetected internal aws. e four major derailments that were investigated in 2002, for example, had defects that TSB determined should have been detected by a rail aw detector car if it had the right instrumentation, if it had been run frequently enough, and if rail surface conditions such as rolling contact fatigue had not blocked or interfered with the ultrasonic signal. e TSB identi ed wide gauge as a primary cause in 13 derailments over those

“A lot of the early broken-rail derailments were characterized as detectable,” Roney said. In 2015, two out of the three should have been detected. In 2019, one of four should have been detectable. e positive trend, particularly since 2015, indicates that railroads have done a better job of ferreting out defects that are detectable with ultrasonic inspection.

A TSB study showed that mainline derailments as a whole decreased by a factor of three between 1980 and 1988 (although the downward trend attened out between 1988 and 1993). e TSB attributed the reduction in derailments between 1980 and 1988 to improved installation and repair of CWR and increased use of automated rail defect detection and track geometry measurement technology. “I would add that replacement of older rail steels with cleaner steels rolled in 1985 or later also had a positive impact,” Roney said.

Rail grinding, particularly the conversion from corrective to rail pro le grinding, was another important part. Joint elimination and the movement away from standard carbon rails to intermediate grade rail steels in tangents and light curves, and premium head-hardened rail in sharp curves were also important, he said. “Improved steelmaking has reduced the indications of tensile residual stresses in the rail web, and there is less clustering of transverse defects than in the past. And when they occur, rail breaks have tended to be cleaner breaks, which are picked up by the signal system before a train nds them.

“Advancements in internal rail defect detection and more frequent ultrasonic testing –moving from 25 million gross tons (MGT) to 5 MGT testing intervals , in particular, has been a big factor in reducing the number of broken rails,” he said.

20 years. ey also indicated that in ve of those derailments, poor inspection practices missed things that should have been seen. “Granted, some of them were wintertime inspections during which inspection was di cult and there was no visible indication in the snow that the rail was spreading,” Roney said. “Nonetheless, the circumstances led us into corrective actions, such as use of gaugerestraint-measurement systems, which are very e ective at nding hidden and dynamic instances of wide gauge.” e data shows that 85% of the wide-gauge derailments that were investigated occurred in 2012 or earlier. ere were no major derailments between 2013 and 2016; only one in 2018 and one in 2020 — a dramatic improvement in preventing widegauge derailments, he said.

A revision to the track safety rules by Transport Canada in 2012 increased the requirements for track geometry testing, and spurred railways to increase light geometry testing on lighter used lines and some yard tracks. e problematic gauge areas identi ed as a result has also likely had an impact on the reduction in wide-gauge derailments.

During the 20-year period in question, there were 11 track buckle derailments. “Not surprisingly, track work had been performed at the time of or just prior to the derailment in seven of these cases. It’s likely that the rail had not been properly de-stressed before a tie gang moved in and disturbed the track,” Roney said. Four of the derailments noted insu cient or inadequate anchors and poor tie condition as primary factors. “ e good news is that there was only one investigation (in 2019) of a major track buckle derailment.” e TSB issued a safety recommendation in February 2003 concerning the need to de-stress continuous welded rail. Work done by the TTCI (now MxV Rail) in Pueblo showed how track reacts before and a er tamping, and how neutral temperature tends to drop over time as track is disturbed. ey also provided very good information on how long a rail needs to be unclipped to do a proper destressing, he said.

With a push from the Federal Railroad Administration, railroads developed common standards for CWR management, which included measuring rail temperature, measuring the gap or bypass every time rail was cut, guarding areas that were particularly at risk of track buckling, and de-stressing the rail before doing major programs or before the weather got hot. With signi cant portions of their networks in the U.S., Canadian National and Canadian Paci c adopted the collaborative CWR standard.

Six of the derailments that the TSB investigated cited rail rollover as a primary cause. Also identi ed were: poor tie condition; train handling was noted in three derailments; high l/v ratios in two derailments; and a binding truck in one derailment. Also noted was the e ect of the b/h ratio in rail rollover; that the location of the primary contact band of the rail is an important factor; and that wheel/rail conditions such as at rail heads with hollow-worn wheel treads coupled with wide gauge reduce the b/h ratio and move the contact beyond the base of the rail and start the rail rollover.

CN and CP have installed truck performance detectors to detect higher l/vs from bad actor cars as part of their corrective action plans. Both railways established programs to identify locations where the design superelevation in curves didn’t match the current operating speeds of longer, heavy trains.

“We all know that the rails speak to you,” Roney said. “When the gauge face of the high rail is wearing rapidly and the plates are cutting on the eld side, the curve is under elevated. And if the low rail shows heavy, mushrooming RCF, it’s likely over-elevated. A good understanding has been developed on how to balance those.” e TSB expressly states that it’s not its function to assign fault, so they don’t address the faults of an individual inspector’s ability. “But in looking at some of these incidents in which the issues are pretty obvious, you have to scratch your head and say, ‘what was he thinking?’ at was pretty obvious.” Roney said. (See Understanding the Role of Human Factors in Railway Inspection, published in the May issue of RT&S.) e TSB indicated that poor training was a root cause of 12 of the 90 track-caused derailments it identi ed.

“A er going through these derailments and looking at the training intervals versus a given derailment, we found little relationship between risk and manual inspection frequency: Roney said “But we did think there was something relating to training aptitude versus competency assurance. If we rely on a sole inspector in a critical territory to make all the right decisions, is that the right thing? Or do we need management oversight? Do we need competency assurance by having prociency tests on that inspector?” he asked. He noted that CP had instituted a ‘ride with the inspector program,’ several years ago in which managers were required to go out with the inspectors to “see if their eyes saw the same things.” Another initiative was to establish electronic record keeping so managers could look in on inspectors to see what they were and what they were not nding.

Opportunities for improvement include the ability to:

• better identify disturbed track risks in advance of track work

• target rail neutral temperatures to strike the right balance between the frequency and consequences of track buckles vs. broken rails

• improve understanding of geotechnical and water management hazards to better anticipate extreme weather hazards

• employ mentoring and quality control of manual visual inspections, which is more e ective than training

• increase inspection frequency when tonnage rapidly increases on jointed track

• better de ne combination track geometry defects

• better identify the risk of rolling contact fatigue

Some opportunities for improvement have been incorporated into Transport Canada’s latest rules respecting track safety. ese include a new maximum time intervals between rail for detection tests that are de ned by tonnage and track class. Track geometry test cycles are now de ned for light and heavy geometry inspection vehicles. Quality assurance of track inspection and maintenance activities is now required, and a professional engineer must sign o on CWR management and rail wear management plans, Roney said.

“In the future, we’re going to have better integration of automated inspections and manual validation follow up. ere’s an opportunity to get away from biweekly manual / visual inspections and replace them with more e ective frequent automated inspections. e inspection regime could be better tailored to the risk pro le, the combination of track condition inspection capabilities and tra c mix,” Roney said. “We’ll need to continue working on lowering the vehicle/track interaction stress state. And we’ll need an escalation of what we do for maintenance and renewal action, based on predictive algorithms.”

Bob Tuzik is Publisher and Editor-inChief of Interface Journal (www.interfacejournal.com)

This article is based on a presentation made at Wheel/Rail Seminars’ 2022 Wheel/Rail Interaction conference.