Railway Investigation Report
Railway Investigation Report
Railway Investigation Report
R09T0151
MAIN-TRACK DERAILMENT
Summary
At 1415, Eastern Daylight Time, on 05 June 2009, westward Canadian Pacific Railway freight
train 235-04 derailed 4 head-end locomotives and 27 cars at Mile 174.41 of the Belleville
Subdivision in Oshawa, Ontario. The derailed cars included a dangerous goods tank car loaded
with hydrogen peroxide (UN 2015), which was not breached. Locomotive fuel tanks were
punctured and about 3000 gallons of diesel fuel were released and ignited. Emergency
responders extinguished the fire and contained the spilled diesel fuel. As a precaution,
105 nearby homes were evacuated. There were no injuries.
-2-
The Accident
In the early morning of 05 June 2009, Canadian Pacific Railway (CP) freight train 235-04 (the
train) departed from Montréal, Quebec, destined for Toronto, Ontario. The train consisted of
7 head-end locomotives (the first 4 online and the trailing 3 isolated) and 111 cars; it was
8137 feet long and weighed 8649 tons. Prior to departure, the train had passed all required
inspections and no defects had been noted. The initial portion of the trip to Smiths Falls,
Ontario, was without incident.
A new crew took control of the train at Smiths Falls (Mile 0.0) and at 0900 1 departed for
Toronto Agincourt Yard (Mile 197.0) on the Belleville Subdivision (see Figure 1). The operating
crew included a locomotive engineer and a conductor, both of whom met fitness and rest
standards, were qualified for their respective positions and were familiar with the territory.
The trip was uneventful until the approach to Oshawa, Ontario. According to the locomotive
event recorder (LER), between 1358:00 and 1402:00, with the train operating in throttle positions
5 to 8, the third locomotive, CP 4652 experienced continuous wheel slip (Mile 163.50 to 165.70).
However, the wheel slip was not sufficient to trigger a warning alarm in the lead locomotive. At
1408:26, the train was travelling at 47 mph with the throttle in position 2 near Mile 169.76, in the
vicinity of Courtice Road, approaching the west Darlington Siding switch (see Figure 2). The
throttle was gradually advanced from position 2 to 8 as the train entered an ascending grade.
1 All times are Eastern Daylight Time (Coordinated Universal Time minus 4 hours).
-3-
Meanwhile, a motorist driving on the road alongside the main track observed a large column of
smoke originating from one of the locomotives. The motorist made a 911 call, stating that the
leading wheels on the fourth locomotive were on fire, with flames reaching the height of the
locomotive. The call was routed to the local fire department, which contacted CP Police.
CP locomotive 4652 was fourth in the consist with its leading end approximately 200 feet from
the cab of the leading unit.
At 1414:37, while the train was proceeding at about 40 mph, an in-cab audible traction motor
ground fault alarm was activated. The throttle was moved from position 8 to 2 over the next
12 seconds and speed decreased to 35 mph. At 1414:55, a train-initiated emergency brake
application occurred as the train was crossing the Oshawa east siding switch at Mile 174.41. The
engineer bailed off the locomotive independent brake and the train came to rest at 1415:13 in
Oshawa.
The crew immediately initiated an emergency radio broadcast and stated that the train was in
emergency. At the same time, the rail traffic controller (RTC) was contacted by CP Police to
report locomotive CP 4652 as being on fire. Emergency 911 operators also received calls from
local residents reporting spot fires along the railway right-of-way from Bowmanville
(Mile 163.58) to Oshawa (Mile 174.40).
It was determined that the last 4 locomotives and the following 27 cars had derailed. The
derailed cars included 7 box cars (3 loaded), 5 covered hopper cars (3 loaded), 7 tank cars
(6 loaded) and 8 empty multi-level, flat cars (see Photo 1). One of the derailed cars, dangerous
goods (DG) tank car FMLX 17113, was loaded with hydrogen peroxide (UN 2015). The car
remained intact and no product was released; however, about 3000 gallons of diesel fuel spilled
from damaged locomotive fuel tanks into adjacent ditches and ignited. Emergency responders
were dispatched, extinguished the fires and contained the spilled diesel fuel. As a precaution,
105 nearby homes were evacuated. The evacuation order remained in effect until noon the next
day, when FMLX 17113 was secured and removed. There were no injuries.
-4-
The temperature at the time was 15°C with clear skies and light winds.
Site Examination
The 4 derailed locomotives and the first derailed car remained upright along the right-of-way.
The fourth and fifth locomotives had sustained fire damage. The fourth locomotive, CP 4652,
was identified as the first to derail. Initial inspection revealed that traction motor (TM) combos
(TM and wheel set assembly) Nos. 1 and 3 on CP 4652 exhibited discoloration consistent with
heat and fire exposure. The leading No. 1 TM combo on CP 4652 had severely overheated and
the axle had broken in the area of the bull gear side suspension bearing (see Photo 2). The
fibreglass bull gear cover was burned and the pinion gear teeth were damaged and worn away.
There were skid flats and burn marks on the wheels consistent with canting and sliding on the
gauge corner of the rail. All 4 TM combos from CP 4652 were shipped to the TSB Laboratory for
further examination.
-5-
Photo 2. Left: CP 4652 No. 1 TM combo Right: Broken axle with bull gear
The derailed cars were scattered on both sides of the Park Road South overpass. Most of the
derailed cars, including DG tank car FMLX 17113, were west of the overpass. Further east, cars
had jackknifed against the overpass support piers. East of the overpass, a block of empty 94-foot
long tri-level auto carriers had jackknifed and came to rest in various positions along the
right-of-way. Several cars came to rest in the backyards of adjacent dwellings. Twelve cars were
extensively damaged and subsequently scrapped. The remaining 15 cars exhibited various
forms of derailment damage; they were eventually repaired and returned to service.
About 1300 feet of track was damaged or destroyed. The most severe damage stretched from
the Oshawa siding switch to where the locomotives came to rest. Extending eastward from the
derailment site at Mile 174.41, impact damage was observed on the track infrastructure up to
the Rundle Road grade crossing at Mile 167.10.
The single main track consists of 136-pound continuous welded rail, manufactured and laid in
1985. The rail was laid on hardwood ties (60 ties/100 feet). The ballast was crushed rock and
slag. The track was in good condition and was inspected in accordance with regulatory
requirements.
-6-
Approaching Oshawa, the track runs roughly parallel and north of Highway 401. It makes
4 sweeping curves over 10 miles before cutting through an urban residential neighbourhood in
the area of the derailment. At any point along these curves, crews can look back from the
head-end and inspect the train as required by CROR Rule 111(c), which states, in part, that
crews should “make frequent inspections of both sides [of their train] to ensure that it is in
order”. The Park Road South overpass crosses over the tracks just west of the east siding switch
at Oshawa (Mile 174.41). The siding at Oshawa is 12 300 feet long. It runs parallel to and south
of the main track between Mile 174.46 and Mile 177.0.
1 2 3 4
Photo 3. Left side view of CP 4652 showing traction motor assembly position numbering
The last scheduled service for CP 4652 was performed at CP’s Toronto Diesel Shop (TDS) on
03 May 2009. No major repairs were performed. The required wheel inspection form was
partially completed with only the wheel tread thickness recorded.
CP 4652 returned for unscheduled service on 16 May 2009 for a thin flange on the No. 4 TM
assembly wheel set, a burnt out headlight, missing weather stripping and a bouncy ride. The
No. 4 TM wheel set had been replaced with a dummy wheel set for transit; the corresponding
TM cut-out switch in the locomotive cab was locked out and tagged.
-7-
Repair work on CP 4652 began on 19 May 2009. During the repairs, several additional defects
were identified and all 4 TM combos were scheduled for change out. Over the following 2 days,
subsequent shifts removed and replaced the 4 combos. TM combos Nos. 1, 2 and 3 were
replaced with combos built up at the TDS on 23 and 24 February 2009 and on 26 March 2009,
respectively. The No. 4 TM combo was taken from another locomotive that was being serviced.
When a TM combo is changed out, a Traction Motor Removal and Application (TMRA) form
must be filled out, signed and dated by the tradespersons and his or her supervisor. TMRAs for
CP 4652 were not properly completed or signed by the appropriate staff.
On 21 May 2009, CP4652 was released from the shop. On 02 June 2009, it received a safety
inspection and returned to service. It was marshalled online as the third locomotive on train CP
234-03 departing for Montréal. Though the No. 4 TM combo had been replaced and the
locomotive inspected, the lock-out tag in the locomotive cab was not removed, so the No. 4 TM
remained disabled. The journey to Montréal was uneventful.
Figure 3. Traction motor (the arrows point to the location of the suspension bearings)
-8-
A thin film of oil provides lubrication between the finely machined surface of the locomotive
axle journal and the inner surface of the suspension bearing. The inner surface of the brass
suspension bearing is lined with babbitt metal. Babbitt metal is an anti-friction alloy lining for
brass bearings; it is composed primarily of lead, with various amounts of tin, antimony and
copper. The babbitt provides a load bearing surface and facilitates lubrication. While the train is
in motion, oil is constantly applied to the axle journal surface through the capillary action of a
wick that contacts the journal surface and extends down into the oil reservoir located in the axle
cap.
There is a large quantity of accumulated research that identifies a failure mechanism, known as
liquid metal embrittlement (LME), which can occur in axle steel when used with plain-type
suspension bearings. LME occurs when the ductility of a solid metal is reduced by exposure of
the surface to a liquid metal. The loss of ductility results in a reduction in the axle toughness.
Toughness is the property that gives a material its resistance to crack initiation and propagation.
The progression of LME usually occurs in the following sequence:
• The brass backing is heated to its melting point and penetrates the axle, initiating
multiple fractures; and
• The fractures form a front which advances from the surface towards the axle core
compromising the integrity of the axle.
When properly assembled, the plain type suspension bearing is expected to last for the life of a
wheel set. By policy, CP replaces bearings at wheel set change-outs and the average wheel set
life for this service is 250 000 miles. However, an increase in friction due to inadequate surface
finish or to a breakdown in the lubrication system will cause premature bearing wear and
failure. At track speeds, increased friction can generate enough heat to melt the babbitt and
bearing metal, initiating LME.
On arrival, the locomotive wheel sets are placed in storage and are to be monitored to ensure
the coating is not compromised. When needed, a wheel set is removed from storage, brought
into the work area and allowed to adjust to room temperature. A mild solvent is used to remove
the protective coating and the wheel set is allowed to dry. Critical wheel set dimensions are
-9-
taken to ensure it meets specifications and the axle suspension bearing surface is inspected for
corrosion or impact damage. The locomotive wheel set is then installed into the upper halves of
the TM suspension bearings. The lower halves of the suspension bearings, axle caps and
lubricating wicks are then applied, after which the TM is set upright. Once the axle cap wells
are filled with lubricating oil, the TM combo is ready for use.
The surface finish protection of an axle is very critical, especially in the support
bearing area. These high precision micro-finish surfaces allow a hydrodynamic
layer of oil, between the axle and the bearings, for maximum load carrying
capacity. Therefore, special care must be taken in order to ensure that the fine
surface finish of the suspension bearing journals are maintained, during handling
and storage of axles and wheelsets.
g)Very minor rust or scratch marks (not exceeding one square inch) may be
repaired by hand with Emery cloth (medium roughness, 100 grit or finer),
provided that the repaired area does not exceed 1/4 inch around marks. No
buffing or polishing tools may be used. Repair is not permitted if scratch area
extends more than one inch across the five inch band at the centre of the
suspension bearing journals. No such marks or repairs are allowed in the one
inch band at the exact centre of the suspension bearing journals.
GM Maintenance Instruction 3902 provides detailed TM combo build-up instructions that must
be followed to prevent premature bearing failure. These instructions include requirements to
apply sealant between key motor and bearing surfaces, including behind the suspension
bearing flanges. The sealant reduces the possibility of water and dirt contamination in the
bearing lubrication system. Axle surface finish in the suspension bearing area is limited to a
maximum surface roughness of Ra 7 micro-inches 2 and a wave index of 100 micro-inches per
inch. Excessive roughness or waviness produces peaks (asperities) in the surface that can
2 Surface finish is the allowable deviation from a perfectly flat surface that is made by some
manufacturing process. The most widely used parameter in North America for general surface
roughness is Ra. Ra measures average roughness by comparing all the peaks and valleys along a line
and then averaging them all over the length of the line. A low Ra number indicates a fine (smooth)
surface finish.
- 10 -
penetrate the lubricating film and generate friction and wear.
The TDS did not receive any locomotive wheel sets in 2007. Between January and October 2008,
eight shipments totalling 71 wheel sets were received. There were no further shipments from
October 2008 to June 2009.
Canadian Pacific Railway’s Toronto Diesel Shop Facility and Shop Practices
Up until the late 1990s, locomotive fleet maintenance facilities reported directly to the CP
mechanical department. CP had a national quality assurance program for locomotive TM
combos, with locally deployed quality assurance groups. In the late 1990s, the CP quality
assurance groups were disbanded and locomotive maintenance facilities were placed under the
management of the local service areas. Between 2000 and 2005, CP contracted out TDS
operations to GM. GM managed the shop and work was performed by CP unionized employees
in accordance with CP maintenance regulations and manufacturer specifications. The contract
was not renewed in 2005 and since then, TDS has been under local CP management.
At the time of the derailment, TDS was the home shop for a fleet of 124 four-axle and six-axle
locomotives equipped with plain-type suspension bearings. The facility conducts scheduled
and unscheduled maintenance necessary to keep this fleet of locomotives in service and
periodically handles unscheduled maintenance for other locomotives in CP’s fleet. This includes
the build-up of TM combos.
TM combos are built up on all shifts by experienced tradespeople with the assistance of
apprentices or other experienced staff. The occurrence tradesman received on-the-job-training
(OJT) from his predecessor 4 years earlier. There had not been any additional training since that
time. Tradespeople are directly supervised by a process coordinator, who in turn reports
directly to a process manager who has responsibility for the overview of daily shop operations.
• The protective coating on the axles of stored wheel sets was not monitored or
maintained. The coatings had deteriorated, and corrosion pitting had developed
on the axle surfaces.
• Sealant was not applied between the bearing shells and the traction motor.
- 11 -
• Paperwork (inspection forms, TMRAs) was not systematically filled out and
supervisors often did not sign off on completed work.
• There was no record of inspection and quality control of completed work by peers
or supervisors; the supervisors were only accountable for ensuring that work was
completed on schedule.
• CP neither monitored TDS shop maintenance records nor audited staff to ensure
work was correctly completed.
Laboratory Examination
TDS built up and installed TM combo No. 2 in locomotive CP 3024 near the same time as it built
the TM combos from CP 4652. CP 3024 TM combo No. 2 subsequently failed on
30 October 2009. All 4 TM combos from CP 4652, the failed No. 2 TM combo from CP 3024 and
2 randomly-chosen locomotive wheel sets from the storage area at the TDS were shipped to the
TSB Laboratory for further examination (TSB laboratory report LP161/09).
The TM combos Nos. 1 and 3 from CP 4652 were heavily damaged by the overheating and fire
while the other motors did not show exterior damage. The drain plugs on both TM combos
Nos. 1 and 3 were tight and all axle cap oil reservoirs contained some oil. Sealant, which was
required to be applied behind the bearing flanges in order to prevent contaminant and moisture
migration into the bearings, was not observed in either TM combo.
The axle from TM combo No. 1 fractured near the midpoint of the gear side suspension bearing
area. The pinion gear had been stripped of its teeth and the mate bull gear was void of lubricant
and exhibited minor teeth damage. The fractured stub end of the failed axle (bull gear side)
showed extensive smearing damage consistent with continued operation after failure. Due to
overheating, oil levels were lower on the gear side and the wick had completely disintegrated.
However, the operation and alignment of the wick relative to the suspension bearing window
appeared normal. On the side opposite the bull gear (commutator side), the wick was properly
aligned and in good operating condition.
- 12 -
The axle from TM combo No. 3 was not fractured through. The bull gear cover was damaged
and the bull gear teeth were shiny and dry. The gear-side bearing components were heat
discoloured. The bull gear was devoid of lubricant and the gear case was damaged. The wicks
appeared to be properly aligned.
Ultrasonic inspection revealed 3 secondary cracks inboard of and parallel to the primary
fracture on the axle of combo No. 1. The axle from combo No. 3 had 1.5 inch deep radial cracks
around its entire circumference at the approximate center of the gear side suspension bearing
area (Photo 4).
The cracks in both axles showed dark colour and beach marks consistent with progressive
cracking. There were no metallurgical anomalies in samples taken from the core of the axle.
Brass and lead material was observed on the crack face. The lead and brass material was present
to a finite depth, suggesting that subsequent crack growth was caused by the fatigue extension
of the crack.
Surface roughness in the axle suspension bearing journal area on axles 1, 2 and 3 from CP 4652
ranged from 14 to 22 micro-inches. Each axle journal exhibited corrosion pits and a significant
number of multi-directional scratches consistent with mechanical damage from the use of
abrasive paper. The scratches varied in orientation and depth (see Photo 5). The axle from
CP 3024 had similar surface features and roughness.
- 13 -
Detailed scanning electron microscope (SEM) examination and energy dispersive spectroscopy
(EDS) analysis were performed on samples from the bearing journals of axles 2 and 4. Flattening
of the edges of the scratches was observed on axle 2. Steel corrosion products as well as bearing
material were observed in the corrosion pits and in many of the deeper scratches.
TM combo No. 4 was in good condition. The axle suspension bearing journal surface was not
corroded and did not exhibit any signs of manual reworking; surface roughness measured
8 micro-inches. The scratches were much finer than on the other axles and were aligned in the
circumferential direction. No flattening of scratches or bearing material transfer was observed.
The examination of the 2 randomly chosen wheel sets revealed that corrosion pitting had
developed on 3 of the 4 axle suspension bearing journal surfaces beneath the protective coating
during storage. The pitting significantly increased the surface roughness. After removing the
protective coating, the average recorded surface roughness of the pitted axle surfaces ranged
from 37 micro-inches to 144 micro-inches.
Inspecting suspension bearings to ensure they are safe and reliable requires access to the
underside of the locomotive.
From 2006 to 2009, freight railways in Canada have reported 9 broken axles, 18 failed
suspension bearings and one overheated bearing to TC (Appendix C) for an average of 7 events
per year. In all but one case, the service failure was identified after critical component failure.
VIA Rail has reported nearly twice as many events in the same time period. However, 44 out of
54 events (81%) were identified before bearing failure. There was 1 broken axle and no
derailments.
There are no regulatory requirements that deal directly with locomotive maintenance shop
practices, although safety management regulations address the need for all aspects of railway
operations to implement safety management practices that can monitor performance and
address the need for corrective actions. Since 2001, TC has not conducted inspections or audits
of TDS locomotive traction motor storage, assembly or inspection protocols.
CP’s HBDs work as standalone stations: they are not networked so data from consecutive
detectors cannot be compared to identify developing anomalies. In contrast, CN’s HBD system
is networked. The data is collected at a central point and used to identify failing journal roller
bearings at an early stage of distress.
- 15 -
Oil Sampling
Other Canadian railways have put in place safeguards to identify problematic bearings. For
instance, VIA Rail implemented a protocol for the weekly sampling and testing of suspension
bearing lubricating oil. If sample lead (Pb) content is between 100 and 149 parts per million
(ppm), the oil is changed and the wicks, replaced. If sample Pb content is greater than 150 ppm,
the TM combo is changed out.
On 18 December 1992, CN eastward freight train No. 218 derailed near Oakville, Manitoba,
resulting in a DG release and the evacuation of the town. The Board determined that the
derailment was caused by a locomotive axle failure resulting from inadequate lubrication of a
TM suspension bearing. In consideration of the consequences associated with any kind of
locomotive axle failure, the Board recommended that:
The Department of Transport urge all Canadian railways to implement heat detection
systems on locomotive suspension bearings to warn crew members of failing bearings.
(R94-08)
In response, TC studied locomotive plain-type suspension bearing failures from Canadian and
American railways and determined that the risk of a suspension bearing failure was very low.
TC concluded that modifying the existing locomotive fleet with a suspension bearing heat
detection system was not justifiable. Moreover, the current locomotive fleet equipped with
friction suspension bearings will be eventually replaced with locomotives equipped with the
newer style of roller suspension bearings. To facilitate risk monitoring, TC required that all
federally regulated railways report each occurrence of a suspension bearing found to be
overheating or failing while in service.
The TSB reassessed the recommendation in April 2009. By this time, about 66% of the Canadian
freight locomotive fleet was equipped with roller type suspension bearings. In addition, the
entire VIA Rail passenger locomotive fleet is equipped with either suspension bearing heat
detection systems or roller type suspension bearings. To date, there has been no failure of any
locomotive axle due to a roller suspension bearing overheat. In light of this information, the risk
of locomotive axle failure was considered to be significantly reduced. The Board reassessed the
response to Recommendation R94-08 as Fully Satisfactory.
VIA Rail Canada Derailment near Biggar, Saskatchewan (TSB Report R97H0009)
On 03 September 1997, VIA Rail Canada Inc. (VIA) Train No. 2, travelling eastward at 67 mph,
derailed at Mile 7.5 of the CN Wainwright Subdivision, near Biggar, Saskatchewan. Thirteen of
19 cars and the 2 locomotives derailed. Seventy-nine of the 198 passengers and crew on board
were injured, 1 fatally and 13 seriously. Approximately 600 feet of main track were destroyed.
- 16 -
The investigation determined that the derailment immediately followed the fracture of the lead
axle on the trailing locomotive. The axle fractured as a result of an overheated TM suspension
bearing that failed due to a lack of lubrication. An onboard hot bearing monitoring system
detected the overheated bearing 29 hours before the derailment and sounded an alarm. Various
operating and maintenance employees attempted to diagnose the warning, but inadequate
knowledge and training, coupled with miscommunication, led to the erroneous conclusion that
the failure was in the warning system, and the crew disconnected it.
On 01 December 2001, eastward CP train 346-046 derailed at the Hawk Lake siding, Mile 122.71
of the Ignace Subdivision. CP determined that the No. 1 locomotive axle, from CP 5802, the
second locomotive, had broken on the bull gear side (L-1 position) suspension bearing area. The
No. 2 TM bull gear side suspension bearing had severely overheated and the suspension
bearing lube oil in the No. 4 TM appeared contaminated.
Locomotive CP 5802 was maintained for CP by GM at the TDS facility under a work service
agreement. All 6 of CP 5802’s TM combos were applied at TDS on 23 and 24 November 2001;
the locomotive had been released into service on 26 November 2001 and the No. 1 wheel set
axle failed 4 days later. Subsequent examination determined that corrosion in the axle
suspension bearing journal area was likely a factor in No. 1 failure and No. 2 overheat. In both
instances, randomly dispersed zones of corrosion measuring up to 2 inches in width and score
marks of varying sizes were observed. Tear down inspections of the 4 other CP 5802 TM
combos revealed varying degrees of corrosion and scoring on the axle suspension bearing
journal areas.
To address this issue, the TSB issued Rail Safety Advisory Letter 01/02 on 18 January 2002 and
stated that “TC may wish to review the manner in which locomotive axles and wheel set
assemblies are handled, stored, reworked and inspected prior to their installation into
locomotive traction motor assemblies.” In response, TC noted that railway companies are
required to annually document and file information concerning locomotives set off from a train
enroute because of suspension bearing failure. CP advised TC that its maintenance regulation
DL-12-01 was reviewed with GM at TDS and locomotive wheel set handling practices were
reviewed to ensure compliance. TC officials were satisfied with CP’s action and planned to
follow up with a comprehensive safety audit at TDS in spring 2002. TM combo build-up and
DL-12-01 procedures were not part of any subsequent TC safety audit.
- 17 -
Analysis
Neither train handling nor track condition was considered to be contributory in this accident.
The report will focus on TM maintenance procedures and practices and the effectiveness of
suspension bearing inspection and monitoring.
The Accident
Recorded wheel slip indications suggest that the No. 1 axle likely broke near Mile 163.0. Upon
approaching Oshawa, CP 4652 was trailing smoke and flames and had been in distress for at
least 11 miles. Impact marks observed on the track infrastructure show that the wheel set first
derailed at the Rundle Road grade crossing (Mile 167.10). The train continued on until CP 4652’s
No. 1 wheel set displaced and came into contact with the east Oshawa siding switch at
Mile 174.41, causing the derailment of 4 locomotives and 27 cars.
The locomotive CP 4652 was on its first return trip since the TM combos were replaced at CP’s
TDS. TM combos are expected to last 250 000 miles in service, but the No. 1 axle broke after
travelling roughly 500 miles and failure of the No. 3 axle was imminent. While plain type
journal suspension bearing failures can be caused by a number of factors, the low service
mileage since installation strongly indicates a deficiency in shop procedures. 4
The components of CP 4652 TMs Nos. 1 and 3 were exposed to high temperature. The presence
of brass and lead material observed on the axle crack faces suggests that friction caused
overheating and melting of the bearing and babbitt material permitting contact with the
suspension bearing journal surfaces of axles No. 1 and 3. The quantity of oil and oil residue
observed in the support bearing wick boxes suggests that initial oil quantity was not a factor.
The oil quality, wick performance and alignment were all within manufacturer specifications.
Therefore, the lubrication system was likely assembled and functioned as designed and the
source of the friction was surface roughness in the axle journals due to corrosion pitting and
abrasion sanding.
The pitting alone would have been sufficient to cause friction and rapid bearing failure at main
track speeds. Contrary to company regulations, efforts were made to remove the corrosion by
sanding the axle suspension bearing journal surface. However, the sanding may not have
removed the root of the corrosion pits, leaving them exposed to service loads. Even after
sanding, the finished surface roughness still remained 2 to 3 times the allowable limit. As soon
as CP 4652 was placed into road service, the asperities of the rough axle suspension bearing
journal surface penetrated the lubricating layer of oil and friction began to wear and melt the
suspension bearing babbitt metal. Once the babbitt and bearing material melted, liquid metal
penetrated the axle suspension bearing journal surface, which initiated liquid metal
embrittlement cracking in the axles Nos. 1 and 3, and resulted in the catastrophic failure of axle
No. 1. While axle No. 3 was only cracked, its failure was imminent.
4 M. Hanson, “Assuring Support Bearing Performance”, Railway Locomotives and Cars, December 1972,
p. 10.
- 18 -
During the investigation, a number of shortcomings with respect to shop procedures and
practices for locomotive axles were identified. CP 4652 TM combos Nos. 1, 2 and 3 were built up
using locomotive wheel sets which were stored outdoors on a dead end track where first-in,
first-out practices were not followed. Therefore, as the axle shipments were received between
January 2008 and October 2008, some axles from the earlier shipments were exposed to a
corrosive environment for an extended period of time before being installed in a TM combo.
Contrary to GM Maintenance Instructions and CP Locomotive Maintenance Regulations, the
protective coating on stored locomotive wheel set axles was not monitored or maintained,
which accelerated corrosion pitting on the axle surfaces. Although excessive corrosion pitting
had developed on the axle suspension bearing journal surface, the axles were not rejected and
were further damaged when abrasive paper was used to refinish the journal surface.
The failure to properly assess the critical role of axle suspension bearing journal surface in the
service life of a TM combo is indicative of shortcomings in the quality control process. The
adequacy of, and compliance with, existing standards and procedures, sign-off of completed
work, and documentation of activities are key components of a robust quality control program.
When quality control is robust, it ensures that procedures have been undertaken and applied in
a consistent manner. Where company quality control practices are not robust, regulatory
auditing and inspection activities present a last line of defence. However, the Railway Locomotive
Inspection and Safety Rules do not address shop storage, handling and assembly practices. Since
CP’s TDS operations had not been audited by TC, the regulator was also unaware of
shortcomings in shop practices. In the absence of adequate shop supervision, company quality
control processes and regulatory audit, company management and regulators were unaware of
the shortcomings at the CP TDS facility. As a result, non-compliance with safety-critical
procedures was not identified and rectified.
This was not the first time there were TM combo problems at TDS. A similar situation occurred
in 2001 when locomotive CP 5802 had all 6 TM combos replaced at TDS at the end of November
and the No. 1 wheel set axle failed 4 days later. Examination determined that corrosion in the
axle suspension bearing journal area was likely a factor in the failure of the No.1 axle and the
No. 2 overheat. Tear down inspections of the 4 other CP 5802 TM combos also revealed varying
degrees of corrosion and scoring on the axle suspension bearing journal areas. Since about 66%
- 19 -
of the Canadian freight locomotive fleet is equipped with roller type suspension bearings and
the entire VIA Rail passenger locomotive fleet is equipped with either suspension bearing heat
detection systems or roller type suspension bearings, the overall industry risk of locomotive
axle failure as a result of suspension bearing overheat is considered to have been reduced since
the early 1990s. However, when localized company shop practices and quality control activities
are deficient, the risk of locomotive axle failure and derailment is increased.
When a train enters a densely populated urban area, train crews tend to be more cautious upon
approaching level crossings in urban areas and become more attentive to people entering the
right-of-way. In addition, crews are required to identify and communicate each signal.
Therefore, a crew’s visual and mental attention may be focused on the track ahead of them,
rather than behind. When people perform more than one task at the same time, the performance
on at least one of the tasks often declines. If a number of the tasks make use of the same sense,
then performance is further reduced. Therefore, when train crews perform multiple concurrent
tasks while traversing a densely populated urban area, CROR Rule 111(c) may not always be
performed consistently and therefore cannot be consistently relied upon to detect anomalies in
the rear of a train consist.
Other railways have put in place safeguards to identify problematic bearings before
catastrophic failure. For instance, VIA Rail has implemented onboard suspension bearing
monitoring as well as a protocol for weekly sampling and testing of suspension bearing
lubricating oil. CN has an integrated HBD system and collects data at a central point which can
be used to perform real time trending and identify failing wheel journal bearings at an early
stage. While wayside inspection systems are not currently designed to identify overheating
suspension bearings, additional sensors could be incorporated into existing systems to perform
this function. Other already installed locomotive systems such as wheel slip or pinion slip
- 20 -
detection systems may also be calibrated to alert a crew to an advanced stage suspension
bearing failure. Without wayside technology or onboard systems capable of detecting an
overheated suspension bearing at an early stage of distress, locomotives that use plain-type
suspension bearings have an increased risk of a catastrophic axle failure and derailment.
3. Once the babbitt and bearing material melted, liquid metal penetrated the axle
suspension bearing journal surface, which initiated liquid metal embrittlement
cracking in the axles Nos. 1 and 3, and resulted in the catastrophic failure of axle
No. 1.
- 21 -
4. Canadian Pacific Railway 4652 TM combos Nos. 1, 2 and 3 were built up using
locomotive wheel set assemblies that were stored outdoors on a dead end track
where first-in, first-out practices were not followed. As a result, some wheelsets
were exposed to a corrosive environment for an extended period of time.
5. The protective coating on stored locomotive wheel set axles was not monitored or
maintained, which accelerated corrosion pitting on the axle surfaces.
8. Even though CP 4652 traction motors No. 1 and No. 3 were on fire for more than
11 miles, the crew did not become aware that their train was in distress and did
not take preventative action.
Findings as to Risk
1. When local company shop practices and quality control activities are deficient, the
risk of locomotive axle failure and derailment is increased.
2. When train crews perform multiple concurrent tasks while traversing a densely
populated urban area, CROR Rule 111(c) may not always be performed
consistently and, therefore, cannot be consistently relied upon to detect anomalies
in the rear of a train consist.
Other Findings
1. While axle No. 3 was only cracked, its failure was imminent.
Safety Action
Action Taken
In November 2010, the TSB issued a Rail Safety Advisory Letter (RSA) 05/10 to
Transport Canada (TC) and Canadian Pacific Railway (CP) concerning locomotive maintenance
and quality assurance practices at the Toronto Diesel Shop. The RSA documented a total of
10 suspension bearing overheat/broken axle events involving traction motors (TM) built up at
the Toronto Diesel Shop.
• Locomotive wheel sets are now stored inside at all CP locomotive facilities.
• A comprehensive job aid and training document outlining all the activities
associated with the build-up of a traction motor and wheel set assembly is
expected to be developed and rolled out in 2011.
TC rail safety inspectors will monitor CP traction motor and wheel set build-up processes over
the short term when on site to conduct normal inspections.
This report concludes the Transportation Safety Board’s investigation into this occurrence. Consequently,
the Board authorized the release of this report on 09 June 2011.
Visit the Transportation Safety Board’s website (www.bst-tsb.gc.ca) for information about the
Transportation Safety Board and its products and services. There you will also find links to other safety
organizations and related sites.
- 24 -
Appendix A - Glossary
Locomotive Date Applied Date Failed Days Since Build Failed Derailed?
Applied Location Motor #
Year Railway Broken Axle Failed Susp. Bearing 5 Overheated Bearing 6 Failed SB Oil Inspection 7
2006 CN 2 1 0 0
2007 CN 1 1 0 0
2008 CN 1 1 0 0
2009 CN 0 1 0 0
2006-09 CN Total 4 4 0 0
2006 CP 2 2 0 0
2007 CP 1 4 0 0
2008 CP 1 1 0 0
2009 CP 1 6 1 0
2005-09 CP Total 5 13 1 0
2006 VIA 0 3 1 5
2007 VIA 0 4 4 15
2008 VIA 0 1 1 10
2009 VIA 1 1 2 6