Reassessment of the Response to Rail Safety Recommendation R93-01
Testing for Vertical Split Head Defects on Curved Track
On 18 September 1992, a Canadian National Railway (CN) eastward freight train derailed 43 cars when the north rail broke under the train on the two-degree curve at Mile 62.25 on the Aberdeen Subdivision at East Warman, Saskatchewan. There were three tank cars loaded with anhydrous ammonia and 18 tank cars of molten sulphur among the derailed cars. Sulphur leaked and spilled from four of the derailed tank cars. There were no injuries (TSB Occurrence No. R92W0240).
On 04 February 1993, a Canadian National Railway (CN) eastward freight train derailed 35 freight cars over a rail head fracture initiated by a vertical split head defect, as the train entered the curve at Mile 21.2 of the Caramat Subdivision, 21 miles west of Hornepayne, Ontario (TSB Occurrence No. R93T0033).
On 17 February 1993, a Canadian Pacific Railway (CP) eastward freight derailed two locomotives and ten cars over a broken rail caused by a vertical split head defect, on the curve at Mile 23.9 of the Nelson Subdivision. The rail was last tested on 11 December 1992 by Sperry Test Car SRS 127. No defects were identified at the point of derailment by the operator during this test. However, a review of the tapes by Sperry representatives on 22 February 1993 confirmed that the presence of a defect was recorded, but had been missed by the operator (TSB Occurrence No. R93C0020).
Board Recommendation R93-01
The Board determined that the accident record for main track derailments illustrated a disproportionate number attributable to undetected rail defects on curved track, in particular, vertical split heads. In view of the frequency and potential consequences of these undetected rail defects, the Board recommended that:
The Department of Transport reassess the adequacy of current Canadian railway procedures and equipment for main track rail testing for:
- identifying rail defects on curved track, and
- identifying vertical split head defects.
Response from Transport Canada (August 1993)
TC is pursuing the issue of the efficiency of rail testing, whether in main, secondary main or branch lines and in tangent or curved track. In addition, Railway Safety Inspectors have reviewed the detection of rail defects in crossings and turnouts with the railways and steps are being taken to improve surveillance of these areas.
While a disproportionate number of defects are expected to occur in curved track because of the more critical loading of rail in curves, it is clear that the detection is to some degree dependent on the surface condition of the rail, including the effect of rail lubrication. In addition, as the Board has stated, detection of flaws is also dependent on the proficiency of the detector operator. The railways have taken action to evaluate the performance of their contractors and have shortened rail testing intervals if operating circumstances change or the number of detected defects increase. Testing output tapes are being reviewed to determine if operator errors exist. Software development is underway to alert operators to anomalies which require further inspection.
In addition, more stringent rail inspection procedures are being introduced to detect potential rail problems in grade crossings. The department is working with the railways to determine what improvements can be made and what benefits will follow in terms of higher rates of identification of defects. As noted previously, the department is receiving computerized rail defect summaries from the railways to aid in determining the seriousness of the problem.
The Board's Report refers to 19 main-line derailments between 1982 and 1991 which have occurred within 100 days of testing by rail testing vehicles. Recognizing that rail testing technology and procedures have been improved since many of these accidents occurred, the department will examine the degree to which these changes might have enabled early identification of the rail flaws and, thereby, have prevented the accidents. The need for further remedial measures will be determined from this analysis. The Surface Group will be contacting Board staff in this connection and will advise you of results.
The Board's recommendations are being brought to the attention of Surface Group regional offices across Canada and will be taken into consideration in the on-going work to keep abreast of developments in rail testing.
Assessment of the Response to Recommendation R93-01
The response does not provide any details as to what the inspectors have reviewed with respect to the detection of rail defects in crossings and turnouts; nor does it indicate what the results were. The staff question the contention that rail testing technology and procedures have been improved. Although new technologies have been introduced for monitoring track geometry, this technology does not assist in detecting internal rail defects. In some cases, the railway companies have increased the frequency of the testing and the inspection procedures have become more stringent. However, to the staff's knowledge, no significant advances in technology have been developed and implemented within the 1982-1991 time frame for internal defect detection. Information on the 19 main track derailments cited has already been passed to TC staff. As yet, we have not been informed of the results of the analysis of this data.
TC states that it is working with the railways to determine what improvements can be made and what benefits will follow in terms of higher rates of identification of defects. Again, there is no substance as to what specifically is being done. Notwithstanding the lengthy response, it does not address the safety deficiency of the effective and timely identification of defects on curved track. Furthermore, no specific comments were provided with respect to the identification of vertical split head defects. Since the response for recommendation R93-01 does not address the concerns, the Board assessed the response as Unsatisfactory.
Additional Response from Transport Canada (August 1994)
TC has reviewed the findings of the Board and concurs that appropriate action has been taken by the railway to reduce recurrence of this type of accident. Statistics provided by the Board indicate a decrease of rail-caused mainline derailments for previous years. It appears that changes to the testing system have increased the chances of detecting the rail defects and TC is satisfied with the measures taken by the railways.
Canadian railways have examined their current procedures and equipment for main track rail testing for identifying rail defects on curved track and identifying vertical split head (VSH) defects. In addition, they also evaluated the training and suitability of the working conditions of the operators of rail testing vehicles.
Consequently, initiatives taken by the railways have improved reliability of detection of sizeable VSH defects in track. Furthermore, the railway industry now carries out more frequent ultrasonic tests and has improved operator qualifications so that they can better interpret indications. Railways are continuing to work with the supply industry to improve detection technology. If successful, a pattern recognition system will all but eliminate the need for operator interpretation of probe indications. The industry has also increased awareness of rail defect types and now has a better ability to visually detect them. Furthermore, they have developed systems that provide to their local supervisors continuous evaluation of rail wear for immediate action or to plan timely replacement.
Reassessment of the Response to R93-01 (September 1994)
This recommendation sought improvement for identifying rail defects on curved track and identifying vertical split head defects. Although the latest response attempts to provide assurances in this respect, the response contains little information as to how current operating procedures or equipment modifications have led to the improvements being sought through this recommendation. Again, since the response does not indicate the extent to which the underlying safety deficiency is being addressed in substantive terms, the Board reassesses the response to remain Unsatisfactory.
Reassessment of the Response to R93-01 (March 2006)
Information received from the railway industry indicates a higher level of testing is being scheduled and the frequency of vertical split head rail failures is reducing. However, TC has not responded with details of its assessment of the adequacy of current Canadian railway procedures and equipment for main track rail testing for identifying rail defects on curved track and identifying vertical split head defects. Therefore, the Board reassesses the response to Recommendation R93-01 as Satisfactory in Part.
Recommendation Status (March 2006)
As the residual risk of having undetected rail defects and vertical split heads in curved track has been significantly reduced, and it is unlikely that continued assessment will yield further results the status of this file is changed to be Inactive.
Additional Response from Transport Canada (March 2011)
Transport Canada advises that, in addition to adding probes to scan the rail at different angles to improve detection, a new procedure for rail testing is being introduced that will increase the frequency and accuracy of testing. The new procedure encompasses a continuous testing over a subdivision instead of the present method of stopping at each suspect location for hand detection and verification. After the run is complete, the information gathered by the rail flaw detection car is transmitted to a specially trained group for analysis and comparison with previous rail flaw detection records and subsequent field verification. The continuous test run method facilitates more accurate detection of defects and more frequent testing with the fleet of testing cars. The data from each test becomes part of the historical data for that subdivision allowing the analysts to quickly identify anomalies for field testing and verification. This method also assists in detecting defects in the early stage of growth so appropriate remedial action can be taken and longer term planning put in place. Improvements in other rail testing areas are continuing. In addition to conventional ultrasonic testing and induction, Sperry Rail has introduced a new X Fire ultrasonic gage wheel allowing more accurate detection of defects. Improvements in detection using guided ultrasonic waves and new technologies using laser-based systems show promise. In general, the rail industry has advanced detection of rail defects greatly over the last several years in its ability to detect defects and its ability to conduct more frequent inspections, thus reducing the risk of in-service failures and accidents. Advancements in technology will continue to help improve this process.
Reassessment of the Response to R93-01 (May 2011)
With rail flaw detection vehicles using induction testing, which is most effective in detecting defects in the rail head, and ultrasonic testing, which has improved markedly in recent years with more probes added at different angles to detect detail fractures, the increased success in identifying rail defects on curved track and identifying vertical split head defects has significantly reduced the risk. A review of TSB's Railway Occurrence Database (RODS) reveals that there were 27 vertical split head main track derailments between 1992 and 1996. However, there were only 5 vertical split head main track derailments from 2006 to 2010. Given that there has been a significant decrease in the number of vertical split head derailments, the Board reassesses the response to Recommendation R93-01 as Fully Satisfactory.
- Date modified: