This report summarizes the Canadian Rail Research Laboratory's (CaRRL) findings after performing an in-depth investigation into the potential safety benefits gained from the introduction of enhanced train control (ETC) technology to the Canadian railway environment. CaRRL was contracted by Transport Canada (TC) to perform this study as a follow-up to a previous ETC study performed by the Train Control Working Group (WG), the final report from which was submitted to the Advisory Council on Railway Safety (ACRS) in September 2016.
The mandate provided to CaRRL included four key components:
- Clearly define the functionality of a potential ETC system;
- Perform a detailed assessment of Railway Occurrences Database System (RODS) records to accurately characterize the proportion of ETC-preventable occurrences;
- Develop risk prioritization criteria for ETC implementation in Canada; and
- Apply the risk prioritization criteria and perform a risk analysis for select rail corridors.
This report is subdivided into two complementary parts. Part A addresses the first two mandate items, while Part B addresses the final two. While this report is structured in two halves, the results presented and discussed in Part B are intimately reliant on the assumptions and analyses made in Part A.
ETC System Functionality
The ETC system envisioned by CaRRL consists of a four-tiered hierarchical framework, with subsequent levels building on the previous in terms of complexity and functionality. A hierarchical system framework was adopted because the WG report indicated a "one size fits all" ETC approach would not be appropriate in the Canadian railway environment. ETC Level 1 through 3 systems are designed as overlays on the existing train control system, while the ETC Level 4 system includes a complete replacement of the existing train control infrastructure. All four of the ETC systems proposed in this report are theoretical in nature but the ETC Level 1 through 3 systems should be implementable with existing technologies. The implementation of the Level 4 ETC system would require significant additional technological development.
The ability of an ETC system to prevent an incident is directly dependent on the functionality designed into the system. In recognition of this dependency the first step undertaken by CaRRL was to fully define the functionality of the proposed ETC system(s) under review. For the purpose of this analysis, CaRRL chose to closely parallel the US Positive Train Control (PTC) system's core functional objectives.
- Prevention of over speed derailments;
- Prevention of train to train collisions;
- Prevention of train occupying improperly aligned switches; and
- Prevention of train entering a foreman's work authority.
The most basic ETC system envisioned by CaRRL (Level 1) is a crew assist and monitoring system that is minimally invasive and locomotive centric (i.e., no buildout into the wayside). ETC Level 2 is a crew assist and enforcement system that incorporates an interface with the train braking system (allowing the system to stop a train instead of only issuing warnings) and selective buildouts into the wayside through the monitoring of key switches. The ETC Level 3 system further increases in complexity by including, amongst other features, full buildouts into the wayside and positive enforcement of operating authorities. The Level 3 ETC system is intended to closely parallel the US PTC system. Finally, the Level 4 system involves a complete re-design of existing train control infrastructure into a communication-based moving block system. At Level 4, all requirements for wayside signalling would be eliminated and all operating authorities would be contained within the ETC system.
Detailed RODS Assessment
RODS is maintained by the Transportation Safety Board of Canada (TSB) and contains information on federally reportable railway incidents. CaRRL performed their ETC assessment on an extracted version of the RODS database provided by the TSB in spring 2017. It is important to note that the RODS database is continuously updated while the download provided to CaRRL is a static snapshot that will not include revisions or alterations incorporated after spring 2017. CaRRL focused their assessment on the 14,036 occurrences reported in the ten-year period between January 1, 2007 and December 31, 2016 within the provided RODS dataset.
To assess whether individual occurrences would have been ETC preventable, a number of assumptions are required to 1) describe how and where the ETC system would be installed and operated and 2) address specific operational circumstances encountered during the assessment. Both sets of assumptions are critical to the final results as different assumptions would alter whether specific occurrences would be ETC preventable or not. For occurrences determined to be ETC preventable, the minimum ETC Level required for occurrence prevention and key system functionality are identified. If an occurrence is determined to be non-ETC preventable, the primary impediment to ETC preventability is identified; however, multiple factors may render an occurrence not preventable with ETC.
The RODS database includes many categories of occurrences that are not preventable by typical ETC systems and technologies. For this reason, it was expected that the proportion of all RODS occurrences that would have been ETC preventable would be small. ETC functionality was expected to achieve much more substantial occurrence preventability in the targeted areas where the system was intended to be of benefit. To provide as full of an analysis as possible, CaRRL has performed three separate preventability assessments:
- ETC preventability for all occurrences in the snapshot of the RODS dataset,
- ETC preventability for Movement Exceeds Limits of Authority-type occurrences, and
- ETC preventability for Main-Track Train Collisions and Derailments.
The breakdown in ETC preventability considering all 2007-2016 RODS occurrences is as follows:
- Level 1 ETC system 3.55% of all 2007-2016 RODS occurrences (498 of 14,036)
- Level 3 ETC system 4.57% of all 2007-2016 RODS occurrences (642 of 14,036)
- Level 4 ETC system 5.96% of all 2007-2016 RODS occurrences (837 of 14,036)
- Not ETC preventable 94.04% of all 2007-2016 RODS occurrences (13,199 of 14,036)
No preventability is presented for the Level 2 ETC system as it is an intermediary between Levels 1 and 3 and preventability will depend on the individual switches monitored. The total numbers of ETC-preventable occurrences at each Level are dominated by Movement Exceeds Limits of Authority (MELA) type occurrences. The specific breakdown in preventability by ETC Level considering only MELA-type occurrences is;
- Level 1 ETC system 36.64% of all 2007-2016 MELA occurrences (463 of 1,168)
- Level 3 ETC system 45.12% of all 2007-2016 MELA occurrences (527 of 1,168)
- Level 4 ETC system 58.39% of all 2007-2016 MELA occurrences (682 of 1,168)
- Not ETC preventable 41.61% of all 2007-2016 MELA occurrences (486 of 1,138)
ETC preventable MELA-type occurrences include those where the proposed ETC functionality was intended to provide the most benefit including prevention of trains passing signals at stop, exceeding the limits of their authority as well as unauthorized entry into foreman's authorities.
A review of only the combination of RODS-reported main-track train collisions and main-track train derailments (rail accidents) between 2007 and 2016 identified the following ETC preventability:
- Level 1 ETC system 2.16% of all 2007-2016 RODS occurrences (22 of 1,018)
- Level 3 ETC system 3.24% of all 2007-2016 RODS occurrences (33 of 1,018)
- Level 4 ETC system 3.93% of all 2007-2016 RODS occurrences (40 of 1,018)
- Not ETC preventable 96.07% of all 2007-2016 RODS occurrences (978 of 1,018)
The full RODS ETC assessment results highlight that only a very small proportion of all 2007-2016 occurrences would be preventable with implementation of any of the four Levels of the ETC system; although the ETC system was not expected to prevent all RODS occurrences. Within key occurrence type categories (ex. MELA or Main-Track Train Collisions and Derailments) where ETC is expected to provide key incident preventability, there are significant numbers of ETC preventable occurrences. In addition, the least complex ETC system (incorporating the Level 1 functionality) provides a large component of overall preventability.
One occurrence category where the majority of overall preventability is not predominantly associated with the ETC Level 1 functionality is main-track switch in abnormal position. For these types of occurrences, only 4 of the 82 total occurrences (4.88%) were preventable at ETC Level 1, while 66 occurrences (80.49%) would have been preventable at ETC Level 3. For the main-track switch in abnormal position category, ETC Level 4 provided no additional occurrence preventability. ETC preventability being heavily weighted towards Level 3 for main-track switch in abnormal position type occurrences is a consequence of the proposed functionality not including universal switch monitoring until Level 3.
Corridor Risk Assessment, Risk Prioritization Criteria, and Cost-Benefit Analysis
To facilitate the corridor-based assessment, the full RODS database was broken down into 19 mainline (> 10 MGT) corridors (comprised of 78 associated subdivisions) defined by Canadian National (CN) and Canadian Pacific (CP). These 19 corridors contain 62% of all 2007-2016 RODS occurrences.
As raw occurrence counts within each corridor will be subject to changes in the operational environment (track usage), an attempt was made to normalize the ETC preventability results by considering the track class, existing control method, 2016 train miles, and 2016 train counts. These data were also provided to CaRRL for each subdivision. While the normalization results suggest that implementing the ETC system will result in a greater rate of RODS occurrence prevention in corridors with existing occupancy control system (OCS) or OCS/ABS (automatic block system) + centralized traffic control (CTC) (mixed) train control systems, the results can be misleading as these corridors also exhibit the least amount of usage (train volumes). The potentially misleading normalization results clearly demonstrate the need for a risk-based assessment of ETC implementation.
The impact of ETC implementation on rail transport risk for specific corridors is evaluated following two methodologies:
- Normalized ratios developed from the full RODS database based on train miles, existing control method, and track class; and
- 2007-2016 corridor-specific observations.
In addition to evaluating the number of occurrences that would have been prevented through ETC implementation, severity indicators are used evaluate the reduction in the consequences of these incidents. The severity indicators CaRRL adopted are derived from information in RODS and include:
- Number of rolling stock involved in the occurrences;
- Number of rolling stock derailed in the occurrences;
- Number of cars involved transporting dangerous goods (DG);
- Number of occurrences with serious injuries; and
- Number of fatalities.
The following table summarizes the proportion of each severity indicator that would be preventable depending on the specific ETC Level implemented. These data are derived considering every occurrence category within the RODS dataset, whether they contain ETC-preventable occurrences or not (see Tables 1-9 and 1-10 in Part A).
|Severity Indicator||Proportion That Would Have Been Prevented|
|Non-ETC Equipped||ETC Level 1||ETC Level 3||ETC Level 4|
|Number of Occurrences||0.00%||3.55%||4.57%||5.96%|
|Rolling Stock Involved||0.00%||2.39%||3.02%||3.86%|
|Rolling Stock Derailed||0.00%||0.85%||1.07%||1.21%|
|DG Cars Involved||0.00%||0.03%||0.15%||0.23%|
|Occurrences with Serious Injuries||0.00%||0.47%||0.47%||0.53%|
The above table highlights that while an ETC system may prevent between 3.5% (Level 1) and 6% (Level 4) of all RODS occurrences, the proportion of each severity indicator is not reduced to the same degree. This suggests the occurrences being prevented are of an overall lower consequence and severity compared to those that are not. This point is further demonstrated by collecting occurrences of major consequence and observing that the primary occurrences contributing to each severity indicator are not ETC preventable. For example, 90% of railway-related fatalities are associated with either trespasser (452 of 751) or crossing (223 of 751) incidents; types of incidents that the proposed ETC system is not designed to address.
The finding that severity indicators related to the transport of dangerous goods and risk to life are not significantly affected by ETC further contributes to the discussion of the risk factors used during corridor prioritization of ETC implementation. It is the opinion of the authors that while comprehensive risk ranking should consider at minimum the risk factors recommended by the United States Federal Railroad Administration (FRA), the ranking procedure for prioritizing ETC implementation may be simplified to consider only a subset of risk factors:
- Train volume (i.e., train miles) [W1/2];
- Method of control [W4];
- Number of tracks [W5];
- Class of track [W6]; and
- Track grade and curvature [W7].
However, this ETC prioritization should be complemented by a qualitative evaluation of the amounts of dangerous goods transported with respect to exposed population and sensitive environmental areas. Weights assigned to each risk factor (W1, W4, etc.) provide a systematic methodology to elicit the experience of rail operators during the ETC prioritization process.
Finally, with the marginal overall safety benefits observed from ETC implementation (relative to all RODS-reportable occurrences), widespread implementation of ETC may clearly not be the best investment to improve overall rail safety in Canada. The proposed ETC functionality (developed to be similar to the US PTC system) attempts to address targeted safety concerns related to a small fraction of all railway incidents. The authors suggest that an optimal safety investment strategy include investigations into the key factors for the most significant severity occurrences. Results from these investigations may then lead to the incorporation of other risk mitigation technologies (ex. Intelligent Traffic Management Systems) into a different prospective ETC framework that may be more effective at reducing overall rail transport risk.
Learn More: To obtain a copy of the report, please contact the Rail Safety Directorate at RailSafety@tc.gc.ca