Language selection

Railway Investigation Report R02M0050

Main-track derailment
Canadian National
Freight train No. Q13711-13
Mile 38.85, Bedford Subdivision
Milford, Nova Scotia
13 August 2002

The Transportation Safety Board of Canada (TSB) investigated this occurrence for the purpose of advancing transportation safety. It is not the function of the Board to assign fault or determine civil or criminal liability. This report is not created for use in the context of legal, disciplinary or other proceedings. See Ownership and use of content.

Summary

On 13 August 2002, at 1636 Atlantic daylight time, Canadian National freight train No. Q13711-13, travelling westward from Halifax, Nova Scotia, to Montréal, Quebec, with 112 loaded container platforms, derailed near Milford, Nova Scotia, at Mile 38.85 of the Bedford Subdivision. Seven container platforms from the last two five-pak container cars, and approximately 2.85 miles of track were damaged. There were no injuries and no dangerous goods involved.

Ce rapport est également disponible en français.

Factual information

Canadian National (CN) freight train No. Q13711-13 (the train) departed Halifax, Nova Scotia, at approximately 1410 Atlantic daylight time,Footnote 1 travelling westward destined for Montréal, Quebec (see Figure 1). The operating crew consisted of one locomotive engineer and one conductor both located in the lead locomotive. They were qualified for their respective positions and met fitness and rest standards. As the locomotives passed Mile 42.60, a train-initiated emergency brake application occurred, bringing the train to a stop at Mile 42.90. After conducting the necessary emergency procedures, the train crew determined that seven container platforms on the last two five-pak container cars were derailed in an upright position on the right-of-way. There was approximately 2.85 miles of damaged track. A track buckle was noted at Mile 38.85 (see Photo 1).

Figure 1. Schematic of geographic area (Source: Railway Association of Canada, Canadian Railway Atlas)
Schematic of geographic area
Photo 1. Post-derailment view of track buckle, looking westward, at Mile 38.85
Image of the post-derailment view of track buckle, looking westward, at Mile 38.85

The train was 7540 feet long, weighed approximately 6230 tons, and was powered by two locomotives. The train consist comprised 33 loaded container cars, each of which consisted of one, three, four or five platforms, amounting to 112 platforms in total. A train safety inspection and No. 1 air brake test were performed before departure from Halifax and no exceptions were noted.

The Bedford Subdivision extends from Fairview Junction, Nova Scotia, Mile 5.0, to Truro, Nova Scotia, Mile 64.0. Train operations from Mile 15.6 to Mile 61.5 are controlled by the Centralized Traffic Control System authorized by the Canadian Rail Operating Rules and supervised by a rail traffic controller located in Montréal. The subdivision is predominantly single main track and handles both passenger and freight traffic. Approximately 15 million gross tons of traffic are carried over the track annually. In the derailment area, the track is located on a slight descending grade (0.5 per cent) in the direction of train travel. There is a reverse curve of approximately three degrees in each direction, with the initial point of derailment located within the exit spiral for the first curve (a right-hand curve).

The authorized timetable speed was 65 mph for passenger trains, and 50 mph for freight trains. Trains consisting of entirely intermodal equipment, whether loaded or empty, were authorized to operate at express speed, which was defined as 5 mph above the freight zone speed.

Recorded information

The event recorder transcript indicated that, in the five-minute period before the emergency brake application, train speed varied between 57 mph and 46 mph, with the locomotive throttle in the No. 8 position and no dynamic brakes or air brakes applied. At a time of 1636:48, with the throttle in the No. 7 position and train speed at 45 mph, a train-initiated emergency brake application occurred. Train speed was recorded as 0 mph at 1637:29.

A review of recorded data from a wayside inspection system located at Mile 22.2 did not show any abnormal wheel bearing temperature, or dragging equipment, as the train passed this location. An inspection of the derailed cars did not reveal any indication of pre-derailment equipment defects.

Track and site information

The track structure consisted of 115-pound RE continuous welded rail (CWR) rolled and laid in 1989. The rail was secured with seven spikes per hardwood tie on 14-inch double-shouldered tie plates, and anchored every second tie. The crushed rock ballast was in good condition, with approximately 18-inch shoulders at the end of the ties and full cribs.

The first derailed wheel marks were approximately 45 feet west of the initial indications of the lateral shift in the track structure (i.e. Mile 38.85). At that location, the track structure had shifted approximately 18 inches to the outside of the curve. There was no sign of any appreciable spike lift for the first 35 feet. Track destruction in the derailment zone precluded the identification of any signs of pre-derailment track conditions (e.g. rail creep). The train crew did not see or otherwise detect any problem with the track structure when the locomotives passed over that location.

Track buckle characteristics

A track buckle is a lateral shift of the track, and, as a phenomenon, has been extensively studied by the railway industry. It occurs when longitudinal compressive stresses building up in the railFootnote 2 overcome the lateral resistance of the track structure. Most track buckles occur on curves. They are more likely to happen in the presence of one or more of the following factors:

A weakened track structure will prevail if ballast is missing from the cribs or ends of ties. If ballast is disturbed, such as after track work, the lateral stability of the track can be reduced by more than 50 per cent.

The action of a train on downgrades can increase rail creep and, consequently, the amount of compressive rail stress. Train vehicle forces can contribute to track buckling by exerting additional longitudinal forces during acceleration and braking. In curved sections of track, rolling stock equipment can contribute to buckling by increasing lateral wheel forces, especially on unstable subgrade. Track with no apparent visual indicators of track misalignment or buckling has been known to buckle ahead of, or beneath, a passing train.

Track inspection and maintenance

Employees performing track patrols relied on visual inspections to identify potential track buckle conditions. The portion of the track where the derailment occurred was inspected by a relief track supervisor on 11 August 2002, two days before the accident. The last previous inspection was 07 August 2002. During these two inspections, no exceptions in the general area of Mile 38 were identified.

A CN Track Evaluation and Service Test (TEST) car examined the Bedford Subdivision around 1100 on the day of the accident, several hours before the arrival of the train. No exceptions were recorded. Line and surface profile measurements through the reverse curve were within allowable limits.

Due to the alignment of the track in the derailment area, and the effects of train forces on the reverse curve, surfacing work was conducted every two years. Historically, the low side in these curves would settle. Frequent track surfacing was required to keep the proper superelevation. In 1999, surfacing work was conducted in the area in association with the installation of approximately 500 railway ties. The most recent surfacing work was a "surface lift,"Footnote 3 completed on 05 and 06 July 2002 from Mile 38 to Mile 40 using a track tamper and a ballast regulator.

Daily maximum temperatures at that time were 21°C (70°F). After surfacing was completed, train movements were temporarily protected with a slow order of 30 mph. Between the time of the last surfacing work and the accident, approximately 1 200 000 gross tons of traffic had passed over this location. This was well above the requirements of CN's standard practice circulars (SPCs) that require protection in most cases until the ballast is well compacted, thus providing adequate lateral resistance.

There had been no rail repaired due to breaks, or rail replaced, nor was there any rail destressing performed in these curves during the previous five years.

CN's SPC 3706 states that, during "rehabilitation ballasting" and "ballast lift" programs, if there are indications that rail stress is out of adjustment, destressing must be carried out in accordance with CN's Recommended Method 3205-0. Some of the warning signs of a "tight rail condition"Footnote 4 and the potential for a track buckling problem are:

The Railway Track Safety Rules (TSR), Part II, Subpart F (V), state:

In the event of fire, flood, severe storm, or other occurrence which might have damaged track structure, a special inspection must be made of the track involved as soon as possible after the occurrence.

At the derailment location, there were no warning signs or previous reports of the rail stress being out of adjustment. The surfacing work that had been performed was a surface lift, not a ballast lift,Footnote 5 the latter being more destabilizing to the track structure.

Effect of high ambient temperatures on rail

Whenever the temperature of the CWR exceeds the neutral rail temperatureFootnote 6 at which the rail was laid or last adjusted, longitudinal compressive forces are created, resulting in a tight rail condition. The tighter the rail becomes, the smaller the lateral force required to cause a track buckle. In 115-pound CWR, each increase of 8°F in the temperature of the rail increases the compressive stress in the rail by about 2200 pounds. A rail temperature increase of 15.5°C (28°F), which can occur on hot summer days, can create a compressive force of 61 425 pounds in a 115-pound RE rail. Neutral rail temperature can be modified by the amount of rail traffic, track maintenance activities (e.g. ballast cleaning or tie renewal programs), and extreme weather conditions, such as extraordinarily high or low temperatures.

The rail must be laid or adjusted at a temperature that will accommodate stresses caused by fluctuations in the temperature. The preferred rail laying temperature range in the area is 27 to 35°C (80 to 95°F). Railway Recommended Method 3205-0 required that additional safety measures be implemented when the ambient temperature exceeded the preferred rail laying temperature by more than 11°C (20°F), i.e. a temporary slow order of 40 mph should be placed between the hours of 1100 and 2000, and a track patrol would be required between 1100 and 1700. Railway records indicating the temperature at which the rail was last adjusted were unavailable.

Environment Canada records from the nearest reporting station (Halifax Airport) show that, at the time of the derailment, ambient air temperature was 30°C (86°F), the winds were 25 km/h and the skies were clear. Between 10 August and 13 August 2002, the temperature had reached daily maximums of 27 to 30°C (80.6 to 86°F), the warmest temperatures of the summer up to that date. Several property owners who lived near the derailment location stated that local temperatures that day had reached 35°C (95°F). Local radio stations had reported similar temperatures elsewhere in the area. The rail temperature after the derailment was measured at 1700 to be 40.5°C (105°F). It is possible that rail temperatures prior to 1700 were higher, as the maximum ambient temperatures were recorded between 1500 and 1600.

Other information

An examination of the TSB data for the years 1997-2002 indicated that there were 18 other occurrences involving a track buckle.

As well as this occurrence, the TSB has been investigating two other occurrences (report Nos. R02D0069 and R02C0054). A brief explanation of these is provided in Appendix A.

New technologies

Much research has been done, and is ongoing, to develop a non-destructive stress measuring system for CWR. Some examples are as follows - with further information in Appendix B:

Some of the new technologies allow the internal stresses in CWR to be remotely monitored, with alarm systems interfaced with evolving communications-based train control systems, or other wayside information systems in rail traffic control centres. These systems, tested and in use on several railways in the United States and in Europe, allow real-time monitoring of rail stress for the locations and the adjacent zones in the track where they are installed. When predetermined thresholds are reached, the resulting alarms allow immediate remedial action to be taken.

Analysis

Introduction

Although the train was travelling approximately 2 mph above the permissible express speed for a brief period of time, this was a relatively minor speed deviation and it is unlikely that it contributed appreciably to the accident. As the operation of the train otherwise met all company and regulatory requirements, and no defective equipment was identified, it is considered that neither the manner of train operation nor equipment condition played a significant role in this accident.

The accident

The accident occurred at the hottest time of the day, during a period that saw the highest temperatures experienced in the area since the resurfacing work was conducted in July 2002. Given the extreme ambient conditions and the frequent surfacing work in this area, it is likely that the neutral rail temperature had been reduced; therefore, the higher-than-normal ambient temperature created higher-than-normal compressive stresses in the CWR. The frequent surfacing work also resulted in an excessive ballast depth that affected the overall stability of the track structure. When these factors were combined with the lateral force generated by the consecutive passes of rail car wheel sets in curved track, the factors commonly associated with track buckles were present. The forces exerted on the track structure by the moving train and the rail compression could not be contained by the ballast, allowing the track to shift out of alignment and buckle. The last two five-pak cars of the train travelling at express speed could not negotiate the misaligned track (18 inches) and derailed. The location of the track buckle in a curve on a downgrade is consistent with previous data on similar type track failures. Likewise, the track buckle, occurring under a moving train, derailing the end cars is typical of track buckle accidents.

Regular track inspection and maintenance

There is no information to suggest that the frequent surfacing work was not performed in accordance with standard practice requirements and company procedures. Track inspections required by the TSR were performed as required, albeit two days before the accident and at lower ambient temperatures.

Track maintenance employees were aware of the favourable results of the inspection performed by the highly instrumented CN TEST car on the date of the accident. However, the TEST car had passed over the track around 1100, which was four to five hours prior to the highest ambient temperature conditions. The TEST car is not designed to detect harmful levels of compressive stress building up in the rails, and track buckling is an instability phenomenon that, in its initial stages, cannot always be identified by track evaluation measurements.

In the absence of any reports of the track stress being out of adjustment, and with few obvious physical signs of excessive rail stress on the date it was inspected, such as spike lift, anchor creep, tie movement or disturbed ballast, it would appear that track maintenance personnel had little indication that the rail needed destressing, and that future high ambient temperatures would negatively affect the integrity of the track structure in the area. This indicates that inspecting for physical signs of track degradation does not identify harmful levels of stress in undisturbed rail or track structure.

Inspection of track during high ambient conditions

The need to fully control rail stress is evident by the amount of compressive stress that can occur in high heat conditions. The surfacing work was conducted at ambient temperatures that were 6°C (10°F) below the preferred rail laying temperature, and there was no indication that the line of the curve had been reduced; therefore, rail stress should not have been augmented by that procedure. However, ambient temperatures on the day of the derailment (which were in the mid-30s) and the rail temperature at 1700 (which was measured to be 40.5°C, or 105°F) were higher than normal. The rail temperature at the time of the accident exceeded the preferred rail laying temperature, increasing the compressive stress in the rail, thereby reducing the lateral force required to create a track buckle.

Railway safety measures, such as extra track patrols and speed restrictions, were not invoked as the ambient air temperature (approximately 35°C, 95°F) was not reported to have exceeded the preferred rail laying temperature by the required 11°C (27°C + 11°C = 38°C, or 100.4°F). It is noted that some railways require track patrols to be conducted at different ambient temperatures. For example:

Signs of the rail being under considerable compressive force would have been more apparent during inspections conducted when rail temperatures were at their highest (i.e. typically in the mid-afternoon to late-afternoon time period). Had a track patrol been conducted in the derailment area during these time periods using lower threshold temperatures (i.e. when temperatures exceeded 32.2°C (90°F) as opposed to 38°C (100°F) specified in CN's SPCs), it would have increased the probability of identifying the early signs (disturbed track structure) of a potential problem, allowing risk reduction measures to be taken to prevent a track buckle.

The need for additional special inspections contemplated by Subpart F (V) of the TSR is somewhat unclear. The TSR mention several cases where a special inspection should be carried out, but do not specifically mention "high heat" as one of the cases. One must infer high heat to be "an occurrence which might have damaged track structure." Somewhat similarly, CN's SPCs required special inspections to be made in relation to a threshold above the preferred rail laying temperature, as opposed to a direct reference to a maximum ambient temperature like some other railways. There are no specific criteria in CN's SPCs to guide employees as to what constitutes "high heat."

Given the absence of specific criteria for the inspection of track exposed to high ambient temperature conditions in the TSR, coupled with the maintenance crew's knowledge of the volume of traffic that had passed successfully since the last resurfacing work was conducted, and the favourable results of the CN TEST car examination, the crew did not inspect the track in the mid- to late-afternoon, the time at which signs of compressive forces would have been more apparent.

Identification of rail stress

The existing inspection methods largely rely on employees to inspect the track structure for any physical signs of a degradation in track structure integrity. To a knowledgeable employee, signs such as spike lift, or rail creep, can indicate the possible existence of a rail that has excessive compressive stress. These inspection methods have been used quite successfully for years but the track must be inspected at the right time of the day, and the employee must identify the physical signs of a tight rail condition. Employees normally have no tools readily available in the field to quantify their assessment. If employees cannot see any physical signs, they have no way of identifying the presence of harmful levels of compressive rail stress. Therefore, they are not likely to suspect that there is a potential problem, particularly if they have no information on the neutral rail temperatures for the sections of track that they are inspecting. Relying on visual inspections alone to identify the physical signs of track degradation does not provide the maximum safety margin as it does not allow the advance identification of harmful levels of residual compressive stress in undisturbed rail or track structure.

Also, it is a challenge for track maintenance employees to perform these visual inspections as, typically, broad geographic areas are affected by high heat conditions. Because employees must physically inspect all of this territory during the heat of the day, which is often limited in duration to a few hours, this makes the task all the more difficult. Visual signs that might indicate a risk of buckling are also indicators of other types of track defects, and they may not always be visible from a hi-rail vehicle moving at the speed normally used during track inspections. Because there is no easy method readily available to track maintenance employees to identify and assess the amount of stress in a rail, and due to the limited time periods available to cover the required territories when rail is at its maximum temperature, there is a risk that stressed rail (i.e. rail that is experiencing an abnormal amount of compressive stress) may go undetected and later cause an unsafe track condition.

Application of new technologies

Because the new technologies described previously are still largely in a developmental stage, they are not in widespread use yet. They are intended to be used in a site-specific manner and are limited in their application as they require the pre-identification of high-risk locations. This occurrence highlights the vulnerability of track inspection methods alone to identify rail that is out of phase with its preferred rail laying temperature. Until such time as stress-measuring devices are regularly used to help track maintenance employees identify the amount of stress in a rail, the detection of unsafe levels of stress will rely largely on existing visual inspection methods that provide a lower margin of safety.

Findings

Findings as to causes and contributing factors

  1. Increased compressive stress in the continuous welded rail due to higher-than-normal ambient temperatures, in addition to an excessive ballast condition due to frequent surfacing, created the circumstances for a track buckle to occur.
  2. The forces exerted on the track structure by the moving train and the rail compression could not be contained by the ballast, allowing the track to shift out of alignment and buckle.
  3. Seven platforms from the last two five-pak cars of a train travelling at express speed could not negotiate the laterally misaligned track and derailed.

Findings as to risk

  1. Inspecting the track for physical signs of track degradation does not allow the advance identification of harmful levels of stress in undisturbed rail or track structure.
  2. Additional railway safety measures, such as extra track patrols and speed restrictions, were not invoked as the ambient air temperature was not reported to have exceeded the preferred rail laying temperature by the railway-specified amount.
  3. Because there is no easy method readily available to track maintenance employees to identify and assess the amount of stress in a rail, and due to the limited time periods available to cover the required territories when rail is at its maximum temperature, there is a risk that rail that is experiencing an abnormal amount of stress may go undetected and later cause an unsafe track condition.
  4. In the absence of stress-measuring devices to help track maintenance employees identify the amount of stress in a rail, the detection of unsafe track conditions will rely largely on existing visual inspection methods that provide a lower margin of safety.

Safety action

After the accident, Canadian National (CN) took the following corrective action in an area of 1.8 miles of the Bedford Subdivision, encompassing the reverse curve where the track buckle occurred:

On 29 August 2003, CN advised that it had also undertaken the following actions:

This report concludes the Transportation Safety Board's investigation into this occurrence. Consequently, the Board authorized the release of this report on .

Appendices

Appendix A - Other TSB investigations involving track buckles - Report Nos. R02C0054 and R02D0069

TSB report No. R02C0054

On 23 July 2002, at approximately 1722 mountain daylight time, Canadian Pacific Railway southward freight train number 771-23 derailed 15 loaded tank cars at Mile 36.6 of the Red Deer Subdivision, near the town of Carstairs, Alberta. Three of the tank cars leaked about 200 litres of ethylene glycol. Highway 2A and adjacent roads were closed for a one-half mile radius around the derailment site. There were no injuries. The investigation identified deficiencies relating to track inspection and maintenance, use of dynamic brakes, employee training, as well as track buckle.

TSB report No. R02D0069

On 03 July 2002, at approximately 1210 eastern daylight time, Canadian National southbound freight train number 353-21-02 derailed 14 cars at Mile 117.68 of the Joliette Subdivision, near L'Assomption, Quebec. There were no injuries and no dangerous goods involved. Approximately 1830 feet of main track, 660 feet of siding track, and a private crossing were destroyed. Other damage included the loss of 150 trees as well as water service to an adjacent tree nursery.

Appendix B - New technologies for measuring stress in CWR

The following list contains some examples of the research that has been done, or is ongoing, to develop a non-destructive stress-measuring system for continuous welded rail (CWR):