Phase I was developed in direct consultation with railway engineers from across the industry with the objective to:

  1. Reduce the frequency of interruptions and improve service reliability in all weather conditions.
  2. Develop equipment and track infrastructure health monitoring technologies and systems.
  3. Develop a strategy for maximizing system capacity and minimizing rail maintenance in a cold environment.
  4. Train the next generation of engineers with the background and necessary skills to address the research needs of the railway industry.

Phase II builds upon the accomplishments and knowledge developed from Phase I. The research is focused on improving the efficiency and safety of Canadian railways by:

  1. Assessing and improving track performance and reliability.
  2. Assessing ballast quality and degradation.
  3. Establishing a quantitative risk management process.
  4. Optimizing rail and rolling stock for cold environments.

The summaries of each key research focus are presented below.

Current Research

(1) Assessing and improving track performance and reliability

Phase I completed extensive trials (over 12,000 km of measurements) with a commercial vertical track displacement (VTD) measurement system, and developed a new methodology to remove the influence of track conditions so that the subgrade stiffness can be mapped. The results showed the underlying causes of track geometry issues, and where maintenance is incapable of keeping up with degradation, even if maintenance has been recently completed. Phase II expands upon this past work by completing the development of the analysis of VTD measurements.

(2) Track geometry, modulus, and the resulting dynamic loads

During the testing of the VTD measurement system on CN track, the National Research Council of Canada (NRC) installed an instrumented wheelset (IWS) system on the same rail car. The dataset was collected concurrently with VTD measurements, providing a remarkable opportunity to evaluate the magnitude of dynamic service loads and the impact they have on the reliability of track structures. An analysis of the IWS datasets will determine the relationships between track conditions (roughness, joints, modulus, and change in modulus), train speed, and the loads applied to the infrastructure.

(3) Assessing ballast quality and degradation

A now-graduated PhD student from Phase I of CaRRL (Dr. Kirk Scanlan) was able to demonstrate the relationship between poor ballast conditions and poor track performance with the use of ground penetrating radar (GPR). The results show that the renewal of ballast through undercutting has long-term improvements in track performance. CN will be implementing the use of GPR to assess ballast and guide undercutting over the next two years. CaRRL will work with CN to review previous findings with before-and-after GPR measurements of track that has undergone renewal. CaRRL will also work with industry partners to develop GPR datasets from the use of higher frequency (>1 GHz) antennas to evaluate the size of pore space within the granular material. The results will directly inform the use of GPR for analyzing the extent and degree of ballast fouling and the effectiveness of current maintenance practices.

(4) Establishing a quantitative risk management process

In 2014, Transport Canada mandated that Canadian railways conduct risk assessments for the transport of dangerous goods along corridors considered critical. In response to this requirement, CN and CaRRL developed a hazard-ranking tool and calibrated a semi-quantitative risk ranking method to strengthen CN's approach to transport risk assessments. In order to be fully quantitative, the risk assessment methodology requires the further development of:

  • a conceptual model process of rail transportation;
  • Failure Modes and Effects Analysis (FMEA) to evaluate the probability and consequences of hazardous events;
  • a risk register; and
  • a framework generating the quantitative risk assessment.

This model will reflect temporal variation of risk with time of year and account for the variation of the prevalence of derailment causes within the physiographic regions of Canada.

Concurrent with the development of the hazard-ranking tool, an analysis of the number, distribution, and causes of derailment on the Canadian railway network as documented in the Transportation Safety Board's Railway Occurrence Database System (RODS) was conducted. Phase II continues the investigation of the RODS database to generate the metrics of seasonal and regional accident rates (derailments, collisions, crossing accidents, and others) into a meaningful statistical format that can be used to populate the rail transport risk model. This model of the rail system operations will include yard operations, loading facilities, and transport; show what sections of the network are at greater risk and what mitigation strategies are effective at reducing those risks; and consider the characteristics of the Canadian rail system and the surrounding natural and man-made environment. This will provide a meaningful operational risk metric that considers public safety, the environment, and infrastructure.

Past Research

(1) Analysis of Canadian train derailments from 2001 to 2014

Disruptions to rail service can have costly implications, not only in terms of monetary loss, but also to the environment, the public and railroad employees. Derailments account for a large number of these disruptions and are caused by a number of factors. This study investigated long-term trends in the number of derailments on Canadian railways from 2001 to 2014, with a focus on main track rail. The total number of derailments were considered, as well as just those that involved dangerous goods cars. To reflect changes in rail traffic volumes over the study period, these trends were normalized against gross tonne-km of goods transported. Another area of focus of this research was to determine the leading causes of derailments and to assess both frequency and severity for these causes. It was expected that a number of causes would show some degree of seasonality, with subgrade issues more common in the summer and mechanical issues more common in the winter. Spatial trends were developed based on the physiographic regions of Canada to assess the effects of physical geography on the safe operation of railways. Four of the leading derailment causes were selected for this analysis. This analysis was accomplished by analyzing data from two primary sources. Derailment data was obtained from the Railway Occurrence Database System, a database of Canadian rail incidents maintained by the Transportation Safety Board of Canada (TSB). An abbreviated version is publicly available on-line, but a more extensive database was provided for this study by the TSB. This database contains information on all types of rail incidents that are self-reported by the railway operators. Rail traffic data was obtained from publicly available tables on the Statistics Canada website. A decreasing trend in main track derailments, as well as the subset of derailments with dangerous goods cars involved, was observed from 2001 to 2014. During this time period, it was found that the cause associated with the greatest number of derailments was the "rail, joint bar and rail anchoring" incident cause, followed by "track geometry," "environmental conditions" and "wheels." These four causes were included in the seasonal and spatial analyses, and it was observed that derailments due to rail and wheel breaks were more common in the winter, while derailments attributed to subgrade and track geometry issues were more common in the summer. Spatially, a higher number of derailments occurred in the Cordillera, Interior Plains and Canadian Shield regions, while comparatively few occurred in the St. Lawrence Lowlands and Appalachian regions. Decreasing or relatively consistent trends were observed in each region.

(2) An investigation of the effects of axle spacing on the rail bending stress behavior

Under heavy and frequent trainloads, large stresses can develop in the rail, of which the bending stress is an important portion. Bending stress may cause fatigue defects to grow and also result in rail breaks, which is the dominant failure mode according to the records of derailments caused by rail issues reported by the Transportation Safety Board of Canada. In this study, the rail bending stresses under different track and loading conditions when the axle spacing between adjacent railcars varies were investigated. Finite element models of different complexities were established using ABAQUS. The Winkler model was also used in the investigation for comparison and reference. Three levels of track modulus, which are 13.79 MPa, 27.58 MPa, and 41.37 MPa were studied, representing soft, medium and stiff track conditions, respectively. Two rail sections, the 115 RE rail and the 136 RE rail, were used, which are common rail sections in North American freight railways. Location effects of wheel loads on the rail bending stress behavior when the axle spacing varies were also examined. It was demonstrated that when wheel loads were applied at the middle of the rail head surface, under each track modulus and for each rail section, the maximum bending stress at the rail head generally follows a pattern of first increasing and then decreasing when the axle spacing increases, while the maximum bending stress at the rail base fluctuates in a small variation range and does not show a clear pattern. This study provides useful guidance in the aspect of studying the effects of axle spacing on the rail bending stress behavior.

(3) Quantifying the distribution of rail bending stresses along the track using train mounted deflection measurements

Heavy and frequent train loads generate large bending stresses in rail. These stresses contribute to the propagation of transverse fatigue defects, which are among the leading causes of broken rail derailments in North America. A thorough assessment of the rail structural condition requires reliable methods for estimating rail bending stresses. This is often challenging due to the many uncontrollable environmental, operational, and structural factors that affect the magnitude of rail bending stresses along the thousands of miles of track. A new methodology was developed for estimating rail bending stresses over long distances using train-mounted vertical track deflection (VTD) measurements. Mathematical correlations between track modulus, rail deflection, rail stress, and applied load form the basis of the method. To develop the correlations, a new finite element modelling method was developed which allowed the simulation of a stochastically varying track modulus along the track. Track models with different track modulus distributions were developed and the resulting VTD and rail bending stresses under moving wheel loads were calculated. The mathematical correlations between the inputted track modulus, modelled VTD and rail bending stresses were quantified using statistical approaches. Based on the results, equations were proposed to estimate the statistical properties of track modulus and rail bending stresses over track windows using the VTD measurements. A framework was also developed to estimate the probability distributions of maximum tensile and compressive bending stresses in the rail head and base, which are necessary for calculating the rail reliability under applied loading. The accuracy of the proposed equations was first verified using a numerical case study for which a random track modulus distribution was considered and artificial noise was added to the modelled VTD. Subsequently, datasets collected from a study site were used to validate the methodology for estimating rail bending stresses. The rail-mounted strain gauges and the wheel impact load detector system at the study site provided information about the rail bending strains under known applied loads. This allowed validation of the maximum bending stresses estimated using train-mounted VTD measurements.

(4) A quantitative evaluation of the impact of soft subgrades on railway track tructure

The majority of Canada's railways were constructed approximately 100 years ago. Railway loads have since increased significantly, resulting in the imposing of higher loads on older infrastructure, particularly those constructed on soft soil foundations. This combination has resulted in the need to upgrade these lines to handle the increased loads and the expected increase of volume of traffic. The main challenge in upgrading these lines is the limited knowledge about the location of poor subgrades, as the extent and relative stiffness of the foundation have not been mapped and documented over long distances. The lack of this information has also limited the study of the influence of subgrade on track performance as well as quantifying the value obtained from the investment in improving track and substructure. An extensive trial with a newly developed rolling deflection technology, over 12,000 km of track, was conducted to assess its potential to map the variability in subgrade conditions over long distances. It was evident from the collected data that unprocessed deflection measurements are heavily affected by the track surface condition such as joints and geometry irregularities so as to obscure the deflections because of poor subgrade support. A methodology was developed to minimize the influence of the surface condition that occur at short wavelengths and show the variations in track deflections because of changes in subgrade conditions that occur at longer wavelengths. The comparison of the processed data at different subgrade and geology condition confirmed that they are consistent with field conditions and are representative of the subgrade conditions. Mapping the subgrade condition over extensive lengths of track presented the opportunity to investigate the impact of subgrade stiffness on the prevalence of track geometry defects and degradation of track quality indices (TQI). This investigation was consisted of the analysis of 800 km of subgrade data and track geometry measurement from two subdivisions from different physiographic region of Canada. The analysis showed that the geometry defects have a strong correlation with both subgrade condition and its variability whereas the TQI are only related to the variability of subgrade condition. These results showed that the locations that have a large deflection and a high variability in deflection are those that are difficult to maintain, and at which maintenance is not always able to keep up with the degradation of the track geometry. It also suggested the processed data from rolling deflection measurement systems provides an evaluation of the underlying causes that result in the degradation of track conditions and allow for the identification of sections where it most likely that maintenance will not always be able to keep up with degradation; even if maintenance has done so recently. A methodology was also developed for quantifying the effectiveness of different methods used to improve the railway track performance on soft subgrades. This methodology is comprised of quantifying the changes in track stiffness from before and after vertical track deflection (VTD) measurements, and the evaluation of the roughness of the track that has developed since the track upgrades. A project was discussed as a case study to explain the steps of this methodology. The result showed that replacement of joints with heavier continuously welded rail (CWR) can reduce the track deflection up to 60%. The results of replacing the suggested 600 mm of subballast with 300 mm of subballast and a geogrid showed no change in the performance of the track under the CWR. Field Measurement of the Effect of ballast degradation on track performance This research focused on developing the technology and analytical capabilities to map the extent and degree of ballast degradation. The purpose of this was to facilitate the monitoring of the quality of the ballast and to develop a methodology for informed planning of ballast renewal programs. This research will aid in clarifying the understanding of the processes at work behind sections of track exhibiting performance issues and how track performance issues are affected by ballast renewal programs in subsequent years. Findings of this research will then allow for the more efficient maintenance of track foundations. More than 500 km of ground penetrating radar (GPR) data were collected along CP's Brooks Subdivision and CN's Edson Subdivision in 2012. The majority of GPR data are of high quality, clearly showing the complex structure within the railway embankment along the entire lengths of the surveyed subdivisions. From these GPR data, estimates of ballast degradation were derived for six sections of track (three along CP's Brooks Subdivision and three along CN's Edson Subdivision). Those ballast quality estimates have then been compared to various measures of track performance derived from the extensive track geometry records provided by both CP and CN. The GPR-derived ballast quality estimates do not appear to be spatially correlated with the track performance measures. Sections exhibiting degraded ballast do not appear to be those exhibiting track performance issues. An analysis of the efficiency of CN's ballast renewal program was conducted. The unprecedented access to long histories of both track geometry and ballast maintenance records along the Edson Subdivision, provided by CN, has allowed for the analysis of the efficiency of ballast renewal operations in the long-term correction of track geometry issues. Investigations have demonstrated that patterns in track geometry after ballast renewal programs are highly variable. Situations exist where ballast renewal has resulted in the quiescence of track geometry variability in following years. While, in contrast, situations exist where ballast renewal has had a negligible effect on the variability of track geometry measures in the following years.

(5) Reliability Centered Maintenance (RCM) and Ultrasonic Leakage Detection (ULD) as a maintenance and condition monitoring technique for freight rail airbrakes in cold weather conditions

Freight rail airbrakes need to improve in reliability to reduce failures, which can lead to harm to the environment, loss of human life and a negative impact on the economy. This research focused on identifying the failures of airbrakes using a Reliability Centered Maintenance (RCM) framework, which uses Failure Modes and Effects Analysis (FMEA) to identify and rank failures. For the purpose of our research, the FMEA result of railroad operating companies was used to research on condition-monitoring techniques for airbrake leakages. Ultrasound Leakage Detection (ULD) was presented as an alternative to the soap and bubble test, as it is a more effective, proactive method to locate and quantify leakages. Experiments were conducted both in the field and in laboratory with simulated and original components to qualitatively and quantitatively evaluate and verify the implications of ULD in cold weather conditions. Correlations between operating pressure, temperature, leakage orifice size, flow-rate and ultrasound intensity were performed to analyze interdependencies. The Principal Component Analysis was applied on the spectral features of dynamic sound signatures to reduce the number of variables and find the correlation between them. The contribution that the frequency ranges made to the factors was estimated to find those having significant impact on spectral feature value for different levels of a particular operating variable. For the sum of contributions from individual spectral features, the frequency range of 2400 - 2500 Hz has the maximum contribution. Frequency range, 1100 - 1200 Hz can be used as a feature for discriminating orifice size, as it has 46% contribution for the Root Mean Square (RMS) value of Power Spectral Density (PSD). The results of this research compared the contribution values of frequency ranges, but do not state the value at which the readings become significant. With extensive quantitative research on the contribution of frequency ranges and an operational inspection strategy, ULD might result in an effective detection method for airbrake leakages in rail service that will be useful for assessing leakage location and severity with more accuracy. This will facilitate reliable airbrakes operations at severe weather and topographical conditions.

(6) Reliability study and maintenance decision making of wheel temperature detectors

In 2011, Canadian Pacific (CP) Railway decided to replace the visual No.1 Air Brake test with a new Automated Train Brake Effectiveness (ATBE) for condition monitoring of rail cars through both physical inspection and measurements by fixed track-side Wheel Temperature Detectors (WTD). To make the most effective use of technology for operational and maintenance decision-making, the new technology should be shown to be reliable, with outputs that are understandable and interpreted accurately. This study used the WTD temperature readings along with records of sensor system failures to develop a method for detecting wheels prone to failure. A set of detector data was checked against neighbouring detectors to improve the classification of a fault with a wheel through multiple measurements and to determine whether there may be a fault with the detector. Studying one train passing consecutive detectors yields useful information about the health of the brakes at each axle of the set of rail cars. Thus, three neighbouring detectors were selected for comparative assessment. Five neighbouring detectors were also selected, but no significant databases were employed and the reliability of detectors was modeled. The best fit to the failure distributions was the normal. Mean-time-between failure (MTBF) for all detectors was calculated to be 2.7 years. For an individual detector the MTBF was about three months. But, for winter operations, the MTBF was found to be only 1.8 months.

(7) Reducing frequency of slow orders resulting from ground hazards

A study on the freezing of railway tunnels in Canada provided a basis on which to access the hazards posed by tunnel icing and identified solutions to limit these hazards, enabling a safer environment for railroaders. The objective of this study was to develop the current understanding of frost penetration along tunnels, the effect of ice pressure and frost shattering on the load-bearing construction, and best practices for mitigating of this hazard to facilitate the reduction of maintenance costs and improvement of safety. A literature-based investigation into the mechanisms of tunnel icing including empirical relations that have been developed to quantify and predict the effects, remedial methods applied to the tunnels to prevent icing, and case studies detailing both successful and unsuccessful tunnel remediation was conducted. A GIS database which includes the location and length of all of the railway tunnels in British Columbia and historical temperature data from the nearest meteorological stations for which data is publicly available was also compiled. Another study on frost heave of degraded ballast identified the susceptibility of the ballast within the rail bed to frost action, allowing the reduction of operational disruptions on the Canadian railway network due to frost heaves through planned maintenance programs. The objective of this project was to develop a frost susceptibility index for use on the Canadian railway network to allow for network-scale monitoring of their development and to allow for diagnosis of the root causes of frost heaves on specific sites. A laboratory-based program was devised to determine the frost heave susceptibility of fouled ballast, using both standard and modified testing methods. The first stage of the study was to assess the physical properties of the fines that are known to contribute to the frost heave susceptibility of soils. Fines were obtained from field samples and were tested to determine the gradation, mineralogy, liquid limit (Swedish fall cone method), and the specific surface area. The second stage of testing oversaw one-dimensional frost heave tests on the fines obtained from the mechanical degradation of ballast. The results of the testing showed that the material generated from the mechanical degradation of ballast is a mix of silt sized particles and rock flour. This material appears to be highly susceptible to frost heave and much more so than the silty subgrade soils that have previously been identified as the root causes of frost heave on railway embankments.

(8) New destructive and non-destructive methods to quantify fracture toughness of high strength rail steels

Rail breaks, resulting from increasing heavy axle-load operations and harsh environmental conditions, remain a major cause of catastrophic derailment of vehicles in North America. The objective of this research project is to develop convenient testing methods to identify the optimum high-strength rail steels for Canadian weather, which can contribute to reducing derailment risks for the Canadian railway industry. Both destructive and non-destructive testing methods were developed to quantify the fracture toughness of high-strength rail steels. Following the establishment of an extended strain energy density (SED) model for estimating the fracture toughness of high-strength rail steels at 23, -10, and -40 degrees Celsius, the influence of stress triaxiality (defined as the ratio of hydrostatic stress to von Mises stress) on the plastic deformation and fracture behaviour of rail steels was investigated. The critical SED factor, determined by calculating the product of the critical SED and a characteristic distance ahead of the crack tip, was found to correlate well with the mode I critical stress intensity factor (KIc) values among the three types of rail steels. The results also confirmed that the dilatational energy dissipation (also known as damage energy dissipation) at the crack tip is the primary component to correlate the critical SED factor with the KIc of rail steels. A more convenient non-destructive indentation technique was also developed for estimating the fracture toughness of high-strength rail steels. A new constitutive model with coupled stress-triaxiality-dependent plasticity and damage was investigated, and a stress-triaxiality-dependent ductile damage model was developed to estimate the critical damage parameter (Dcr) at the crack tip. Compared to the destructive testing method, the non-destructive indentation technique was found to be more convenient and has the potential to serve as a tool for the in-field health monitoring of rail steels and for the material evaluation, at an early stage, of a new rail steel under development at the University of Alberta.

(9) Landslide risk and resilience in the Ashcroft Thompson River Valley, British Columbia

This research project investigates the hazard frequency and associated risk posed by 12 large landslides in the Thompson River Valley south of Ashcroft, British Columbia, which are collectively referred to as the Ashcroft Thompson River landslides. The Canadian National and Canadian Pacific railways operate busy main lines within the subject 10-kilometer corridor south of the village of Ashcroft, traversing the lower portions of many of the landslides. These landslides have resulted in significant, recurring disruptions to Canada's Class 1 railways, and affect numerous other groups located within the valley, including First Nations communities, residents of the villages of Ashcroft and Spences Bridge, salmonid populations within the Thompson River, and owners/operators of upland agricultural areas. While the Ashcroft Thompson River landslides are typically inactive or very slow moving, numerous periodic reactivations have been documented in the past 150 years, some of sufficient magnitude to temporarily dam the Thompson River. Recent landslide activity in the corridor was correlated to snow pack and stream flow in the Thompson River basin. Trends in the discharge of the Thompson River over the past century were, in turn, related to the phase of the high-level climate phenomenon termed the Pacific Decadal Oscillation (PDO). An awareness of the interannual and interdecadal fluctuations in climate regimes and their hydrologic implications were found to be an essential component for contextualizing historic landslide frequency and future landslide risk, and the anticipated hydrologic impacts of climate change and their potential effects on landslide activity in the corridor were also of interest. A qualitative risk assessment was developed to distinguish the different modes of failure which have been demonstrated by the landslides in the corridor, to elucidate the unique causal factors, consequences and associated risk scenarios. A risk management strategy was also developed, which integrates an understanding of climate factors which may portend landslide activity in the corridor, a kinematic understanding of the landslide failure modes, and an observational approach to geotechnical monitoring and inspection. It was recommended that contemplated risk reduction measures favor flexible options which would be suited to a range of future climate conditions, improve adaptive capacity through more effective anticipation and forward planning, and integrate active monitoring and regular site inspections of slope movements with effective strategies for the documentation, communication and dissemination of hazard and risk information. Further, the economic consequences of a landside which impacts the railway infrastructure in the Thompson River Valley can grow exponentially with the duration of the outage; given the diverse cross-section of stakeholders involved, the consequences of a rapid landslide in this corridor may also include significant safety, environmental, and cultural impacts. Thus, the application of a resilience paradigm for managing critical infrastructure and enhancing a community's ability to cope with disruptions was investigated in the context of the Ashcroft Thompson River landslides.