Maintenance Planning for Rail Asset Management—Current Practices (2020)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

23 4.1 Defect Rate Analysis The previous chapter presented the direct results of the survey. This chapter attempts to look at the survey data in a more focused fashion in an effort to identify the relationship among transit system size, level of traffic, development of rail defects, and need for increased testing. Freight railroads typically define loading as traffic density in MGT. Transits do not typically evaluate MGT because actual loading applied to the track structure is a function of the number of passengers. In the absence of MGT data, then, two other quantities were used herein: passenger miles and passenger car-miles as a proxy for loading on the rail. Passenger miles were obtained from APTA’s (2019a) Public Transportation Factbook. Ridership trend information for each indi- vidual agency was obtained from APTA’s (2018, 2019b) Q4 2018 and Q4 2019 Public Transpor- tation Ridership Reports. Passenger car-mile data were taken from FTA’s NTD. The number of passenger car-miles per mile was also calculated to get a measure of the number of cars (and thus axles) passing over a section of track per year. This measure defines the level of loading of the track (which affects the development of fatigue defects in the rail, which directly relates to the number of cycles of loading on that rail). Such a simplified analysis allows for estimation of the relative level of loading cycles experienced by the different transit agencies. Figure 22 and Figure 23, respectively, present defects per track mile as a function of passenger miles and as a function of passenger car-miles. Both figures show an increasing trend, with more defects per mile occurring with increasing traffic. Figure 24 presents the number of defects per track mile, plotted as a function of passenger car-miles per track mile. This figure provides further visualization of the effects of repeated loading on a system. Note that this figure cor- responds to the rate of loading of the rail, because the number of car-miles per track mile is a measure of traffic density and the corresponding number of loading cycles. As can be seen in this figure, there is a well-defined relationship between defect rate (defects per mile) and level of loading, as defined by car-miles per track mile. (The one apparent outlier at 350,000 car-miles per track mile is Bay Area Rapid Transit, which reported no detected defects.) Simply put, the greater the level of traffic, the greater the rate of defect development. 4.2 Frequency of Inspection Next, the frequency and effectiveness of testing were examined. Frequency of UT as a func- tion of system size, illustrated in Figure 25, shows a positive correlation. The size of the bubbles in the chart corresponds to the number of defects found by the specific transit system in 2019. Larger systems tend to schedule UT more often than smaller systems; Figure 25 shows that larger transit systems indeed experience more defects, thereby warranting their more frequent test- ing. The same information for visual inspection led to the results displayed in Figure 26; again, C H A P T E R 4 Evaluation of Survey Results

24 Maintenance Planning for Rail Asset Management—Current Practices 0 0.2 0.4 0.6 0.8 1 1.2 0 2,000 4,000 6,000 8,000 10,000 12,000 D ef ec ts /M ile Million Passenger Miles Defects per track mile 2019 Defects per track mile 2018 Figure 22. Defects per track mile as function of passenger miles. 0 0.2 0.4 0.6 0.8 1 1.2 0 50 100 150 200 250 300 350 400 D ef ec ts /M ile Million Car-Miles Defects per track mile 2019 Defects per track mile 2018 Figure 23. Defects per track mile as function of passenger car-miles. 0 0.2 0.4 0.6 0.8 1 1.2 0 100,000 200,000 300,000 400,000 500,000 600,000 D ef ec ts /M ile Car Mile/Track Mile Defects per track mile 2019 Defects per track mile 2018 Figure 24. Defects per track mile versus passenger car-miles per track mile.

Evaluation of Survey Results 25 the size of the bubbles in the chart indicates the number of defects found by a specific transit system in 2019. Less of a correlation, however, is found among system size, frequency of visual inspection, and number of defects found. This lesser correlation may be attributable to external requirements for visual inspection, as defined in some of the more commonly used standards. It should also be noted that visual inspection encompasses much more than simply rail defects and rail condition. 4.3 Broken Rail Analysis Next, the quantity of broken rails in each system was examined. The number of broken rails in each system in 2019 was first plotted against the system’s respective number of track miles. Figure 27 shows this plot as a bubble chart, with the size of the bubbles corresponding to the number of passenger miles (representing cumulative loading) in 2019. This measure was further examined by plotting broken rails per track mile as a function of passenger miles, because passenger miles is a metric for the annual loading on a system. Figure 28 represents loading 0 1 2 3 4 5 0 100 200 300 400 500 600 700 800 Te st s pe r Ye ar System Size (Track Miles) Figure 25. Frequency of UT versus system size. Bubble size represents defects found in 2019. 0 1 2 3 4 0 100 200 300 400 500 600 700 800 In sp ec tio ns p er W ee k System Size (Track Miles) Figure 26. Frequency of visual inspection versus system size. Bubble size represents defects found in 2019.

26 Maintenance Planning for Rail Asset Management—Current Practices effect on the quantity of broken rails a system experiences. Note the two small systems with a disproportionate number of broken rails for their size: these will be addressed further in the case examples. Figure 29 and Figure 30 show broken rails as a function of rail defects. Although a linear trend line fits the data well (see Figure 29), a polynomial, or curved, trend line fits the data better (see Figure 30). These figures show an upward trend, which indicates that the more defects that exist, the higher the likelihood of a broken rail. This trend is consistent with the literature, which shows a definite relationship between an increasing number of rail defects and an increasing number of broken rails (service defects) (Zarembski and Palese, 2005a). It should be noted that three transit systems reported no detected defects but still had broken rails: (a) Minneapolis’s Metro Transit (nine broken rails), (b) Phoenix’s Valley Metro (seven broken rails), and (c) Denver’s RTD (two broken rails). Of these systems, Minneapolis, which reported no rail defects in 2018 or 2019, stated that the broken rails it experienced were attrib- utable to cold weather (see case example discussion). Phoenix found one defect in 2019, and RTD found none. Note that Minneapolis, Phoenix, and Denver test once annually. 0 10 20 30 40 50 60 70 0 100 200 300 400 500 600 700 800 Br ok en R ai ls Track Miles Figure 27. Broken rails per track mile, with passenger miles represented. Bubble size corresponds to number of passenger miles in 2019. 0 0.05 0.1 0.15 0.2 0.25 0 2,000 4,000 6,000 8,000 10,000 12,000 Br ok en R ai ls p er T ra ck M ile Million Passenger Miles Figure 28. Broken rails per track mile as function of passenger miles.

Evaluation of Survey Results 27 Chicago’s Transit Authority also reported broken rails from cold weather. Two of these transit systems with significant numbers of broken rails—Minneapolis and Chicago—were selected for follow-up case examples, which will be discussed in greater detail later in this report. Interestingly, Figure 30’s polynomial trend line is nearly linear at the low defect rate levels but then increases in nonlinearity as it approaches New York City Transit Authority, with its relatively large number of defects and track miles. If the NYCT data point were not included, the resulting trend line would be linear but with a slope greater than that in Figure 29, which would correspond to a higher rate of broken rails. It is possible that NYCT’s extensive testing (four times per year) finds more defects than are repaired before they grow to failure and break, thereby driving the trend down. This could in turn point to the effectiveness of increased use of such practices as risk-based UT scheduling. It should be noted that this approximate linear relationship between the number of defects found and the number of broken rails that occur each year is consistent with the results of a broader study of rail defect behavior in Class I railroads, as reported in Zarembski and Palese (2005a). 0 10 20 30 40 50 60 0 100 200 300 400 500 600 700 Br ok en R ai ls Rail Defects Figure 29. Broken rails as function of defects, plotted with linear trend line. 0 10 20 30 40 50 60 0 100 200 300 400 500 600 700 Br ok en R ai ls Rail Defects Figure 30. Broken rails as function of defects, plotted with polynomial trend line.

28 Maintenance Planning for Rail Asset Management—Current Practices 4.4 Broken Rail Derailments The number of broken rail derailments over the past 5 years was examined next. Only four systems had derailments attributable to broken rails in the past 5 years. The total number of broken rail derailments divided by five times the number of broken rails that a given agency experienced in 2019 (representing the number of broken rails over the 5-year time period) resulted in an approximation of how many broken rails result in a derailment. Figure 31 plots this measure as a function of system size. As might be expected, the derailments occurred on larger systems. Note that the derailment rate was again consistent with that found in the railway literature and reported for other rail systems (Zarembski and Palese, 2005a; 2007). 0 0.002 0.004 0.006 0.008 0.01 0.012 0 100 200 300 400 500 600 700 Br ok en R ai l D er ai lm en ts p er B ro ke n Ra il System Size (Track Miles) Figure 31. Broken rail derailments per broken rail as function of system size.