INVESTIGATION OF TIE PLATE ICING/ICE JACKING: PHASE II for Transport Canada Prepared by Yin Gao Proprietary Report P-23-019 June 30, 2023 MxV Rail A subsidiary of the Association of American Railroads 350 Keeler Parkway | Pueblo, Colorado USA 81001 www.mxvrail.com Disclaimer: This report was prepared for Transport Canada by Transportation Technology Center, Inc. (dba MxV Rail), a subsidiary of the Association of American Railroads (AAR), Pueblo, Colorado. It is based on investigations and tests conducted by MxV Rail with the direct participation of Transport Canada to criteria approved by them. The contents of this report imply no endorsements whatsoever by MxV Rail of products, services, or procedures, nor are they intended to suggest the applicability of the test results under circumstances other than those described in this report. The results and findings contained in this report are the sole property of Transport Canada. They may not be released by anyone to any party other than Transport Canada without the written permission of Transport Canada. MxV Rail is not a source of information with respect to these tests, nor is it a source of copies of this report. MxV Rail makes no representations or warranties, either expressed or implied, with respect to this report or its contents. MxV Rail assumes no liability to anyone for special, collateral, exemplary, indirect, incidental, consequential, or any other kind of damages resulting from the use or application of this report or its contents. i Executive Summary In Phase II of the cold climate research for Transport Canada, MxV Rail investigated the tie plate icing/ice jacking issue using a railroad survey, a track panel test, and a numerical simulation. The survey collected the industry’s experience and knowledge in tie plate icing. A track panel test and finite element (FE) modeling were used to investigate how track performance (gage strength) changes for different severities of tie plate icing and identify key factors that affect track gage strength. The research detailed in this report was completed between December 2022 and June 2023 under Transport Canada contract T8009-180251-001-TOR. The railroad survey was distributed among the members of the Railway Research Advisory Board (RRAB), Class I railroads, regional railroads, and short lines. Among the 11 responses received, nine provided railroads’ experience in tie plate icing in their territories. The survey also documented 1) track issues caused by tie plate icing, 2) areas of concern, 3) tie plate icing identification methods, and 4) tie plate icing remediation methods. A track panel was built to investigate tie plate icing within a controlled environment. Steel shims with different thicknesses were fabricated to simulate tie plate icing. Additionally, the results of the track panel test were used to validate a FE model that was used to simulate more complicated cases during the investigation of the tie plate icing issue. The key findings from the track panel testing and FE modeling include the following: Track Panel Testing • Track gage strength was highly related to 1) the number of ties with tie plate icing and 2) the distance the rail was lifted above the tie plate (rail base above or below tie plate shoulder). In the tested cases, the largest widened gage was recorded at 0.38ʺ when both rails had five consecutive ties that simulated both tie plate icing and missing field-side rail spikes. • Steel shims (simulating ice jacking) with 1/2ʺ thickness (rail base above tie plate) caused up to 0.2ʺ in gage widening compared to 1/4ʺ steel shims (rail base below tie plate). • Tie plate icing simulated on one rail or both rails did not show a significant difference in track gage strength. Finite Element Modeling • The model was validated by the track panel test results and then used to simulate more severe cases that were not tested in the track panel test. • The number of consecutive ties simulating tie plate icing varied from 0 to 11. None of the simulated cases exceeded 1ʺ gage widening (the limit for Class 4 and 5 track according to Rules Respecting Track Safety, Transport Canada) under a 4-kip gage widening load. • The number of simulated missing field-side rail spikes varied from 0 to 9 in the model. The model showed that missing spikes could substantially increase the gage widening if more than five spikes were missing. When the case of nine missing spikes was simulated, the resulting gage widening was above 1ʺ. ii Table of Contents 1.0 Introduction ............................................................................................................ 1 2.0 Railroad Survey ..................................................................................................... 2 3.0 Track Panel Test .................................................................................................... 5 3.1 Test Setup .................................................................................................... 5 3.2 Test Matrix .................................................................................................... 6 3.3 Test Results ................................................................................................. 7 3.4 Takeaways ................................................................................................... 9 4.0 Numerical Simulation of Tie Plate Icing .................................................................. 9 4.1 Model Validation ......................................................................................... 10 4.2 Number of Consecutive Ties ...................................................................... 11 4.3 Missing Spikes............................................................................................ 12 4.4 Tie Degradation .......................................................................................... 14 4.5 Rail Longitudinal Forces ............................................................................. 15 5.0 Test Summary, Analysis, and Future Research Recommendations .................... 16 References .................................................................................................................... 19 List of Figures Figure 1. Tie plate icing1 .............................................................................................. 1 Figure 2. “Cauliflowering” of snow/ice near rail ............................................................ 2 Figure 3. Two conditions for tie plate icing1 ................................................................. 2 Figure 4. Survey responses of typical areas of concern of tie plate icing .................... 3 Figure 5. Survey responses of track conditions found when there is tie plating icing .. 4 Figure 6. Survey responses of current practices of remediations ................................ 4 Figure 7. Track panel test setup .................................................................................. 5 Figure 8. Steel shims used to simulate tie plate icing .................................................. 6 Figure 9. LTLF used in the track panel test ................................................................. 6 Figure 10. Gage widening for different cases when tie plate icing occurs on one side of rail only ....................................................................................... 8 Figure 11. Gage widening for different cases when tie plate icing occurs on both rails .................................................................................................. 8 iii Figure 12. Finite element model for tie plate icing ....................................................... 10 Figure 13. Test and simulation cases for one rail with tie plate icing ........................... 11 Figure 14. Test and simulation cases for two rails with tie plate icing .......................... 12 Figure 15. Schematic of missing field-side rail spikes ................................................. 13 Figure 16. Effect of missing spikes .............................................................................. 13 Figure 17. Effect of wood tie degradation .................................................................... 14 Figure 18. Rail longitudinal forces setup in the model ................................................. 15 Figure 19. Effect of rail longitudinal forces ................................................................... 15 List of Tables Table 1. General questions about tie plate icing ............................................................. 3 Table 2. Test matrix......................................................................................................... 7 Table 3. Model validation .............................................................................................. 10 Table 4. Risk factor analysis ......................................................................................... 17 1 1.0 INTRODUCTION Snow and ice can impose hazards on the safety and integrity of train operations and impact rail infrastructure. As a winter occurrence, tie plate icing (also referred as ice jacking, Figure 1) has resulted in train derailments in the past,1-3 and it needs to be thoroughly investigated and closely monitored. Tie plate icing occurs when ice builds up between the base of the rail and the tie plate. The ice buildup will cause the rail to sit vertically above the shoulder of the tie plate, leaving the fasteners as the primary means of providing resistance to lateral rail motion. Figure 1 shows the ice buildup lifting the rail and raising the spikes. Figure 1. Tie plate icing1 MxV Rail investigated and documented tie plate icing/ice jacking during three field inspection trips to three railroads in cold weather conditions in January 2022 and April 2022 (Phase I).4 Tie plate icing is possible any time when snowfall accumulates around the base of the rail and tie plates, making it difficult to identify. Defined as snow/ice buildup around the rail base and showing cracking due to rail movement, “black marbling” or “cauliflowering” (Figure 2) of snow by rail movement, changes in track gage, and rail cant can be typical indications of the occurrence of tie plate icing, as documented in the Phase I study. Both track conditions and weather conditions are needed to generate tie plate icing. Wood-tie tracks with standard tie plates and cut spikes, combined with significant rail movements, due to poor tie condition, fouled ballast, joints, grade crossings, etc., are often found to be areas of concern. In addition, two conditions, 1) rail jacked by snow and 2) rail jacked by ice, are documented by railroads. The first condition happens when snow gets blown against the bottom of the rail and then becomes packed layer by layer to jack up the rail. The second condition occurs when weather conditions create freeze-thaw cycles where snow/ice melts, and the water gets trapped in the gap between the bottom of the rail and tie plate and freezes again. Figure 3 depicts how the two conditions could cause rail to be jacked up by snow or ice under certain conditions. 2 Figure 2. “Cauliflowering” of snow/ice near rail Figure 3. Two conditions for tie plate icing1 Phase I work led to an improved understanding of the inspection, identification, and formation mechanism of tie plate icing. In the current phase of research (Phase II), MxV Rail continued to investigate tie plate icing/ice jacking to improve the understanding of the issue through 1) a railroad survey, 2) a track panel test, and 3) a comprehensive numerical simulation. The results and findings of the Phase II research are included in this report. 2.0 RAILROAD SURVEY MxV Rail designed and distributed a survey to collect information of tie plate icing in North America. This survey was intended to collect individual working experiences in the railway industry. The survey was distributed among the members of the Railway Research Advisory Board (RRAB), Class I Railroads, regional railroads, and short lines. Eleven questions regarding each railroads’ experience with and the identification and remediations of tie plate icing were included in the survey. Survey responses were kept anonymous, and a total of eleven responses were received. The following table and charts show a summary of the survey responses. As Table 1 indicates, the majority of respondents have experienced tie plate icing and track issues caused by tie plate icing. 3 Table 1. General questions about tie plate icing Survey Question Yes No 1) Experienced tie plate icing 9 2 2) Tie plate icing caused track issues 6 5 According to the survey, tie plate icing can occur anywhere on a railroad track (Figure 4). The responses comprised all the selective typical areas of concern. Three answers in “Other” included “All tracks,” “Tangent tracks,” and “Tunnels.” Figure 4. Survey responses of typical areas of concern of tie plate icing Poor ballast drainage and rail pumping were the most common track conditions chosen when tie plate icing occurred (Figure 5). Two respondents mentioned frost heave as another track condition. In addition, even though the condition “Tracks with high rail neutral temperature” was not selected in this survey, one Class 1 railroad stated that it experienced tie plate icing on those tracks during the Phase I study. 4 Figure 5. Survey responses of track conditions found when there is tie plating icing Based on the survey results, the two most common ways to address the effect of tie plating were speed reduction and the manual removal of ice/snow (Figure 6). Reactive actions were generally taken to remediate tie plate icing, most likely due to the nature of tie plate icing. Tie plate icing is extremely dependent on localized conditions (both track and weather), and the location of the icing occurrence can be difficult to predict. Figure 6. Survey responses of current practices of remediations 5 3.0 TRACK PANEL TEST The main purposes of the track panel test were 1) to simulate tie plate icing in a controlled environment and 2) to validate the computer model. It is important to validate the model before expanding the modeling effort to simulate other cases. The following test scenarios were performed during the track panel test: • Installing shims on multiple ties to simulate tie plate icing on consecutive ties • Adjusting shim thickness to simulate different level of tie plate icing severity • Removing spikes to simulate broken/missing spikes 3.1 Test Setup A 32-tie track panel was built on a flat, open space (Figure 7). The panel was constructed with 14ʺ standard AREMA plates, a four-cut-spike pattern, new wood ties, and AREMA 136 RE rail. The ties were numbered from 1 to 23 and spaced 19.5ʺ apart. Figure 7. Track panel test setup Two smooth-surface steel shims measuring 6ʺ×7.5ʺ×0.5ʺ and 6ʺ×7.5ʺ×0.25ʺ, respectively, were used to mimic the condition of an actual snow surface and simulate different tie plate icing situations (Figure 8). The shoulder of a tie plate was 7/16ʺ tall. Therefore, the 0.25ʺ-thick steel shim simulated mild tie plate icing (rail base below tie plate shoulder), and the 0.5ʺ-thick steel shim simulates severe tie plate icing (rail base is above tie plate shoulder). The track conditions simulated by steel shims were similar to what was documented during the field trips in Phase I. 6 Figure 8. Steel shims used to simulate tie plate icing (*Note: The ice and snow in the first two photos was from a snowfall during testing and did not provide any test value.) During Phase II, the lateral track gage strength was quantified using a lateral track loading fixture (LTLF). Track gage widening was measured at both the rail head and the rail web. The track was loaded by an incremental gage widening load of 0, 1, 2, 3, 4 kips. The track gage at 4- kip load was the focus in the following analysis. Figure 9. LTLF used in the track panel test 3.2 Test Matrix The test cases were designed to adjust the conditions of the middle five ties, Ties 14, 15, 16, 17, and 18. The following factors were considered in the test design: • Baseline test – No steel shims installed. • Severity of tie plate icing: 1/2ʺ or 1/4ʺ steel shims. • Number of ties having tie plate icing: One tie, three ties, or five ties. 7 • Missing spikes: only remove field side rail spikes on Tie 16. • Number of rail having tie plate icing: One rail or two rails. The test matrix was generated based on the above factors. Table 2 covers the first four factors. The same test cases (Case 2 to Case 14) were conducted for both one rail and two rails (number of rails having tie plate icing). Table 2. Test matrix Case # Tie 14 Tie 15 Tie 16 Tie 17 Tie 18 Case 1: baseline Case 2 1/4ʺ shim Case 3 1/2ʺ shim Case 4 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim Case 5 1/4ʺ shim 1/2ʺ shim 1/4ʺ shim Case 6 1/2ʺ shim 1/2ʺ shim Case 7 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim Case 8 1/4ʺ shim 1/4ʺ shim 1/2ʺ shim 1/4ʺ shim 1/4ʺ shim Case 9 1/4ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/4ʺ shim Case 10 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim Case 11 (missing spike) 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim 1/4ʺ shim Case 12 (missing spike) 1/4ʺ shim 1/4ʺ shim 1/2ʺ shim 1/4ʺ shim 1/4ʺ shim Case 13 (missing spike) 1/4ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/4ʺ shim Case 14 (missing spike) 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 1/2ʺ shim 3.3 Test Results The gage widening values at the highest gage widening load (4 kips) were plotted for each test case. Figure 10 and Figure 11 present the test results documenting the gage widening when simulated tie plate icing on one rail and on both rails, respectively. Due to the complexity of track components in an actual track panel, such as spike orientation, initial position of rail, etc., the trend of gage widening can be complicated. However, the overall trend was as expected. Five ties with 1/2ʺ steel shims produced 0.310ʺ and 0.333ʺ gage widening for one rail and both rails, respectively. Removing a field side rail spike on Tie 16 would increase the gage widening, but not by much (usually less than 10%). The highest gage widening was measured at 0.380ʺ in Case 14 (both rails), a 10% increase from Case 10. 8 Figure 10. Gage widening for different cases when tie plate icing occurs on one side of rail only Figure 11. Gage widening for different cases when tie plate icing occurs on both rails 9 3.4 Takeaways The major findings from the track panel test are summarized below: • Baseline (without tie plate icing) had the strongest gage strength. • Typically, gage widening increased as the number of ties simulating tie plate icing and the severity of tie plate icing increased. • One-quarter-inch steel shims caused 0.2ʺ gage widening (0.1ʺ for the baseline case), regardless of the number of ties having tie plate icing. Using 1/2″ steel shims further increased gage widening up to 0.19ʺ compared to its corresponding 1/4ʺ steel shim cases. Track strength did not change significantly from one rail to both rails simulating tie plate icing issue, indicating that if tie plate icing occurs, track strength will be reduced similarly no matter if it occurs on one rail or both rails. • Missing one field-side rail spike did not result in an obvious reduction in track strength. Cases with more missing spikes were simulated and presented in Section 4.3. • The gage widening test load was maximized at 4 kips in the track panel test, resulting in a maximum 0.38ʺ gage widening. The actual train operation environment can have a lateral load or gage widening load several times (5 to 10 times) higher than 4 kips. Therefore, tie plate icing can indeed generate safety concerns of train operations. 4.0 NUMERICAL SIMULATION OF TIE PLATE ICING Numerical modeling was conducted using Ansys Workbench to assess the influence of tie plate icing on track performance, primarily on gage strength. The modeling efforts characterized the effects of various parameters on track strength and integrity. The simulated parameters included 1) the number of ties having tie plate icing, 2) the number of missing spikes, 3) new and degraded wood ties, and 4) rail longitudinal forces. The model used for tie plate icing simulation had eighteen ties with 14ʺ plates and cut spikes, the same as the track panel test. The contact types for rails/spikes, rails/tie plates, plates/spikes, and tie plates/ties were set to be frictional contacts. The coefficient of friction was set at 0.4 for the frictional contacts used in the model. Spikes and ties were set to be bonded. The plate element type was a solid tetrahedron element (sizing: 0.5ʺ), and the rail, spikes and ties were a solid hexahedra element, with sizing of 0.78ʺ, 0.2ʺ, and 4ʺ, respectively. The total nodes and elements were 1,076,067 and 381,512, respectively. The two rails were constrained at both ends to simulate the condition of continuous rails, rails that should not move much longitudinally due to the loading conditions in this study. All the track components were assumed to be elastic. In addition, the gage widening load was 4 kips in the simulation. The track panel tests showed the widened gage was not critical when the rail base is still below the tie plate shoulder. Therefore, only the cases with severe tie plate icing (worst scenario: rail base above the tie plate shoulder) were simulated. 10 Figure 12. Finite element model for tie plate icing 4.1 Model Validation Tie plate icing simulated by 1/2ʺ steel shims (rail base above tie plate shoulder) was the extreme case during the track panel test. Therefore, Cases 3, 6, 10, and 14 (for one rail and both rails) along with the baseline case were used to validate the model. Table 3 shows the comparison of test results and simulation results. Most of the simulation cases matched the actual tests within a 10% difference. Two cases had over a 30% difference, which could be due to the initial setup (initial positions of track components) during that test. Overall, the trend and magnitude of the simulation results were reasonable. Figure 13 and Figure 14 plot the test cases and corresponding simulation cases for a visualized comparison. Table 3. Model validation Test Case Gage Widening (Test) Gage Widening (Model) Difference (%) HH (in.) WB (in.) HH (in.) WB (in.) HH WB Baseline 0.099 0.028 0.102 0.027 3% -4% One 1/2ʺ shim, one rail 0.154 0.043 0.145 0.045 -6% 5% Three 1/2ʺ shims, one rail 0.303 0.128 0.210 0.086 -31% -33% Five 1/2ʺ shims, one rail 0.310 0.138 0.305 0.163 -2% 18% Five 1/2ʺ shims, one rail, one missing spike 0.297 0.205 0.315 0.181 6% -12% One 1/2ʺ shim, two rails 0.223 0.067 0.190 0.065 -15% -3% Three 1/2ʺ shims, two rails 0.266 0.110 0.260 0.108 -2% -2% Five 1/2ʺ shims, two rails 0.333 0.183 0.359 0.195 8% 7% Five 1/2ʺ shims, two rails, two missing spikes 0.380 0.216 0.372 0.213 -2% -1% 11 4.2 Number of Consecutive Ties Due to testing safety concerns, the number of consecutive ties with tie plate icing and the number of missing spikes were limited in the track panel test. The model, however, did not have the same limitations, and researchers increased the tie plate icing severity by adding more ties with tie plate icing and more missing spikes. Cases 15, 16, and 17 were simulated to have seven, nine, and eleven consecutive ties experiencing tie plate icing. No missing spikes were simulated in these three cases. The gage widening values at the rail head for these three cases were: 0.359ʺ, 0.401ʺ, and 0.427ʺ for one rail with tie plate icing (Figure 13) and 0.398ʺ, 0.431ʺ, and 0.452ʺ for both rails (Figure 14). The increase rate of widened gage was slowed after five ties with tie plate icing, indicating that the effect of the number of ties with tie plate icing decreased when the number reached five ties. Figure 13. Test and simulation cases for one rail with tie plate icing 12 Figure 14. Test and simulation cases for two rails with tie plate icing In addition, simulations of three scenarios of ties having tie plate icing were performed: • Every other tie having tie plate icing. • Three consecutive ties with tie plate icing with one tie without tie plate icing in between. • Five consecutive ties with tie plate icing with one tie without tie plate icing in between. The gage widening values had a minimal difference (<1%) compared to Cases 3, 6, and 10 (one tie, three ties, and five ties having tie plate icing, respectively), meaning that the number of consecutive ties having tie plate icing dominated the widened gage. 4.3 Missing Spikes Case 16 (nine ties with tie plate icing) was selected for investigating the effect of missing spikes. Since field-side rail spikes are the key spikes used to restrain the rail lateral movement, those spikes were removed to simulate the missing spike cases (Figure 15). The lateral load was applied at the middle tie of the nine consecutive ties. Missing spike cases involve five scenarios: • One missing spike on the middle tie. • Three spikes missing (removed two rail spikes on the ties adjacent to the middle tie). • Five spikes missing. • Seven spikes missing. • Nine spikes missing. 13 The widened gages generated by missing spikes at the rail head by the head-applied load (HH) for each case were plotted in Figure 16. As can be seen, the widened track changed relatively slowly when changed from no missing spike to three missing spikes. The rate picked up with five or more missing spikes. Understandably, a significant amount of lateral load could still be carried by the field-side rail spikes on adjacent ties in one or three missing spikes cases. However, when there were more consecutive missing spikes, a longer piece of rail lost lateral support from spikes, causing more lateral rail bending and gage widening. Figure 15. Schematic of missing field-side rail spikes Figure 16. Effect of missing spikes It is worth noting that the gage widening load in the simulation cases was 4 kips. This load could easily be doubled (can be as much as 5 to 10 times) in a curve or under other vehicle/track conditions. Since it was an elastic model, the gage widening values would be two to three times greater for the simulated cases if the load level is doubled or tripled, meaning the widened gage could be higher than 1ʺ (the maximum allowable gage widening for Class 4 and 5 track by Transport Canada). In reality, some track components (e.g., spikes) might have failed due to high lateral displacement, which would worsen the track conditions. 14 4.4 Tie Degradation In the model, the tie material considered was new solid-sawn hardwood ties with a tie modulus of 1,700 ksi. However, wood ties will degrade over time due to weathering, decay, and dynamic load. A tie modulus of 400 ksi was used to simulate a degraded tie condition. Gage widening at both the rail head and the rail base was simulated with the reduced wood tie modulus. A widened gage increase that varied from 11 to 15% was recorded for various simulated cases (Figure 17). (a) (b) Figure 17. Effect of wood tie degradation 15 4.5 Rail Longitudinal Forces Either a braking/tractive effort or the rail temperature can generate longitudinal forces in rails. The rail longitudinal forces combined with the lateral load may exacerbate the gage widening caused only by the lateral load. In the simulation, rail longitudinal forces were applied on both rails just outside the ties that have tie plate icing (Figure 18). A rail longitudinal force of 10 kips is considered a typical value generated by locomotives when accelerating or braking, and this force was used for the simulation. About 0.1ʺ to 0.2ʺ gage widening was added to the rail head displacement as shown in Figure 19. Figure 18. Rail longitudinal forces setup in the model Figure 19. Effect of rail longitudinal forces 16 5.0 TEST SUMMARY, ANALYSIS, AND FUTURE RESEARCH RECOMMENDATIONS In this phase of research, MxV Rail designed a railroad survey to collect the industry’s experience on tie plate icing and investigated the track gage strength with the occurrence of tie plate icing through a track panel test and a numerical simulation. The railroad survey was distributed among the members of the Railway Research Advisory Board, Class I railroads, regional railroads, and short lines. Eleven responses were received, and the majority of the responses (9 out of 11) indicated experience with tie plate icing in their territories. The survey documented 1) track issues caused by tie plate icing, 2) areas of concern, 3) tie plate icing identification methods, and 4) tie plate icing remediation methods. The survey shows that tie plate icing can occur anywhere on a track, and visual inspection and manual removal are the usual ways to deal with the tie plate icing issue. A 32-tie track panel was built to both simulate tie plate icing in a controlled environment and validate the computer model. Steel shims with two different thicknesses were used to simulate the severity of tie plate icing. Gage strength was measured under a 4-kip gage widening load using LTLF for all the test cases. The following are the test findings: • Track gage strength was highly correlated with the number of ties simulating tie plate icing and the severity of tie plate icing (rail base above or below tie plate shoulder). The largest widened gage was recorded at 0.38ʺ when both rails had five consecutive ties simulating tie plate icing and missing field-side rail spikes. • The baseline case had 0.1ʺ gage widening. Adding 1/4″ steel shims caused 0.2ʺ gage widening regardless of the number of ties having tie plate icing (up to five ties in this test). When increasing the severity of tie plate icing by replacing 1/4ʺ shims with 1/2ʺ shims, the gage widening could increase up to 0.19ʺ compared to its corresponding 1/4ʺ steel shim cases. Simulated tie plate icing that occurred on only one rail or both rails did not have a significant difference in track gage strength. • One missing field-side rail spike did not result in an obvious reduction in track strength. The results from the track panel test were then used to validate a computer model that had an eighteen-tie track panel. The modeled track gage widening matched the trend and magnitude measured in the track panel tests. The validated model was used to investigate the effect of: • The number of consecutive ties with tie plate icing. • Tie and fastening conditions (degraded ties and missing spikes). • Rail longitudinal forces on track gage strength. 17 The findings from the model include: • The number of consecutive ties with tie plate icing varied from 0 to 11. None of the simulated cases exceed 1ʺ gage widening (Class 4 and 5 track according to Rules Respecting Track Safety, Transport Canada) under a 4-kip gage widening load (used in track panel test). However, a higher gage widening load (double or triple) could cause gage widening above the rules of track gage. • The number of missing field side rail spikes varied from 0 to 9 in the model. The model showed that missing spikes could substantially increase the gage widening if there were more than five missing spikes. When the number of missing spikes was nine, the simulated gage widening was above 1ʺ. • Degraded ties (low tie modulus) showed an 11–15% increase in gage widening for various simulated cases. • Rail longitudinal forces can cause an extra 0.1ʺ to 0.2ʺ” gage widening when a 10-kip longitudinal force was applied on the rails. Risk Factor Analysis The factors that drive track strength behavior were investigated in the track panel test as well as the simulation. These factors were categorized based on the severity of each factor’s consequences by the test and simulation results (Table 4). Table 4. Risk factor analysis Risk Factor Risk Category Discussion Ties with tie plate icing High When the number of consecutive ties with tie plate icing is five or more, the track may have a gage widening issue. Ties with missing spike Medium to High When tie plate icing occurs, missing field side rail spikes could exacerbate the loss of track gage strength, especially the number of missing spikes is five or more. Tie condition Low to Medium Degraded ties could cause 10-15% decrease in track gage strength but are not as significant as the first two factors. Rail longitudinal forces Low to Medium Train braking/accelerating or rail temperature could cause rail being loaded in the longitudinal direction, therefore widening track gage. 18 Guidelines for Identification and Mitigation Based on the risk analysis, the following guidelines may need to be recommended/prioritized for the identification and mitigation of tie plate icing. The most important factor in Table 4 is “Ties with tie plate icing.” This factor indicates the presence of a gap between the rail base and the tie plate of that tie. “Gap” management would be key in the identification and mitigation of the tie plate icing issue. To identify potential areas that could cause a gap: • Locate high spikes and missing spikes, especially when consecutive high spikes and missing spikes are found. • Observe black marbling/cauliflowering when snow covers track. • Take vertical track deflection or track cant measurements to identify rail pumping. For mitigation: • Use a Gagelok screw spike at a rail spike position on a standard plate or using curve block plates/elastic fasteners every fourth tie (or depending on railroad’s requirement) in areas that are prone to have tie plate icing, to minimize the gap between the rail base and tie plate. • Fix missing spikes and high spikes before winter. • Improve ballast drainage in areas prone to tie plate icing Future work could include testing the effectiveness of potential mitigation methods. If track geometry and weather data at the area of concerns can be obtained, analysis on track geometry change, especially rail cant, track gage, base gage (if available) would be beneficial to identify tie plate icing in early stage. Also, asymmetric/uneven ice buildup may cause rail to be rolled inward or outward. This could change the wheel/rail (W/R) contact geometry and change the vehicle steering and W/R forces which could exacerbate any gage widening problems. Asymmetric/uneven ice buildup can be further studied if there is interest by the steering committee and railroads. Additionally, researchers could examine the likelihood of tie plate icing in frost heave regions to understand the correlation between frost heave and tie plate icing. 19 References 1. “Railway Investigation Report R11V0057: Main-track train derailment.” Transportation Safety Board of Canada, March 08, 2011. 2. “Derailment Prevention.” Engineering Safety Flash, Canadian National Railway, May 2020. 3. “Railway Transportation Safety Investigation Report R20W0031: Main-track train derailment,” Transportation Safety Board of Canada, February 18, 2020. 4. Gao, Yin. (2022).“Investigation of Tie Plate Icing/Ice Jacking Phase I Report,” P-22-001. AAR/MxV Rail Pueblo, CO. For questions or comments on this document, contact yin_gao@aar.com . MxV Rail 350 Keeler Parkway Pueblo, Colorado USA 81001 A subsidiary of the Association of American Railroads (AAR) www.mxvrail.com mailto:yin_gao@aar.com Cover Executive Summary Table of Contents List of Figures List of Tables 1.0 Introduction 2.0 Railroad Survey 3.0 Track Panel Test 3.1 Test Setup 3.2 Test Matrix 3.3 Test Results 3.4 Takeaways 4.0 Numerical Simulation of Tie Plate Icing 4.1 Model Validation 4.2 Number of Consecutive Ties 4.3 Missing Spikes 4.4 Tie Degradation 4.5 Rail Longitudinal Forces 5.0 Test Summary, Analysis, and Future Research Recommendations References