Track Maintenance Costs on Rail Transit Properties (2009)

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.

CHAPTER TWO LITERATURE REVIEW GENERAL This literature review presents selected information available from published and public sources on track maintenance costs and the influences on track maintenance cost within the United States rail transit industry. The topics of this section are presented in the following order: • Defining track maintenance • Size and shape of the rail transit industry • Literature review. DEFINING TRACK MAINTENANCE At the most fundamental level, track maintenance costs reflect track maintenance practices. Track maintenance is well understood by its practitioners, as evidenced by continuing success (and resurgent growth) of the wheel/rail technology. This report does not include structures maintenance; for example, bridges and tunnels. Track maintenance definitions and understandings are as follows: 1. Track 2. Maintenance demand 3. “Acceptable” track conditions 4. General maintenance approaches 5. Life-cycle costs 6. Direct and indirect costs 7. Light rail, heavy rail 8

Track Track is the system of materials from the subgrade to top of rail in ballasted track or from the bottom of a rail support device (fastener, block tie) to the top of rail in ballast-less track. Maintenance Demand Maintenance demand is the level of effort, materials, and equipment to provide acceptable track. Acceptable Track Conditions Acceptable track conditions are as defined by APTA and American Railway Engineering and Maintenance-of-Way Association (AREMA) track safety standards. Track is properly maintained or “acceptable” when the track condition is acceptable for the designed operating parameters over that track. Any length of track that meets the applicable standards for the designed operation on that track is considered to be “acceptable.” Any flaw in the track that causes it not to comply with the track standards for the designed operation is an unacceptable track condition. General Maintenance Approaches The execution of track maintenance varies by maintenance philosophies or budget realities. Approaches to track maintenance range from preventive maintenance, where developing conditions are corrected as they occur, and crisis maintenance, where corrections occur at failure (service degradation by slow order for a track condition is, by the foregoing definition, a failure), as well as “spot” or “programmed” maintenance. Most if not all maintenance practitioners adhere to preventive maintenance as a goal, although budget constraints require a balanced approach somewhere between ideal maintenance and crisis maintenance. Life-Cycle Costs Life-cycle costs are the sum of all costs of a specified track throughout its economic life, from first installation through removal or replacement. These costs include the material purchase and initial track construction, routine track inspections, and periodic maintenance to the end of its economic life, as well 9

as disposal or recycle costs; for example, tie disposal, disposal of spikes and anchors, and including costs to collect and sell to scrap dealers. The nature of track requires the definition of life cycle to be stated for an arbitrary period, often assumed to be on the order of 25 years, within which the all the track performance cross-influences are adequately captured. Economic life is defined as a point in time where the trend of annual maintenance costs of an existing component or system of components exceeds a threshold value. Technically, a threshold value for identifying useful economic life is when repair costs have reached some percentage of the replacement and future maintenance costs. A key criterion for economic life is track that meets the definition of acceptable track conditions, described previously. Track with deferred maintenance requires “temporary” slow orders in place until repairs or maintenance is performed are one level, whereas “slow orders” to continue service occur when track has exceeded its economic life. Direct and Indirect Maintenance Costs The definitions of direct maintenance effort and the supporting organizational effort/cost to implement productive maintenance are fundamental for this project. For this report, direct maintenance is an effort to perform a specific maintenance task, such as replacing a frog or a rail. The effort and costs of direct maintenance are defined as those functions directly involved in the maintenance task that should be common to all rail applications. For this report only, the following are direct maintenance efforts/costs for specific maintenance tasks: • Labor and material to perform a task. This does not include the effort to assemble crews, material, and equipment; travel to a site; or management overheads. Labor for direct maintenance includes all craftsmen (track laborers, welders, machine operators, and any helpers) and their direct supervision (the crew foremen for most organizations). Material costs are the direct cost of components, including delivery of the material to the receiving point in the system; • Expendables (fuel, etc.); • Track inspection and reporting; • Employee fringe benefits; and • Premiums for constraints such as working in tunnels or other constricted areas, at night, etc. 10

Indirect maintenance efforts are the costs of preparing crews and materials for a task, travel from a staging area to a site, delays for example resulting from limited track access, mid-level supervision, material stores costs (material stock-pile efforts, including purchasing activities, inventory, etc.), equipment procurement and maintenance, clothing allowances, training, and organization overheads that are not captured in the direct costs or other category. Heavy Rail, Light Rail A heavy rail system usually has an exclusive right-of-way (ROW) with no other intervening transportation form, including road crossings. A light rail system shares the ROW in some manner with other transportation forms, largely road crossings and in-street operation. The term light rail, despite its implications, has nothing to do with weight. The vehicles and track for light rail have weights similar to heavy rail configurations. Hearsay suggests that the term “light rail” was devised as a political euphemism for streetcars and trolleys to allow funding consideration over objections that the latter were considered obsolete. Both modes may be operated in transit commuter services that shape ROW with freight railroads. RAIL TRANSIT INDUSTRY Track included in this literature review encompasses heavy rail, light rail, and commuter rail transit service, each producing somewhat different track performance, maintenance, and costs. It is useful to introduce, at least, the size of the industry (Table 1), its costs (Table 2), and cost by size and ridership (Table 3). The following data were reported to and audited by the FTA. The data are from the 2004 National Transit Database, the most recent year available. The cost data in Tables 2 and 3 are from a single year, using aggregates. The aggregates combined by the author, are offered as loosely indicative of industry means. 11

TABLE 1 TRANSIT RAIL INDUSTRY SYSTEMS AND RIDERSHIP 2004 Heavy Rail Light Rail Commuter Rail Totals Number of Rail Transit Systems 14 27 20 61 Annual Ridership (unlinked trips) 2,747,616,634 349,915,503 413,898,363 3,511,430,500 Annual Vehicle Train Miles 94,025,617 41,969,242 49,988,272 185,983,131 Length Route miles (main lines only, double track = 1 route mile) 878.81 681.19 3,793.20 5,353.20 Track miles, including each direction of multiple tracks, yards, and sidings At-grade exclusive ROW (track miles) 736.70 294.80 3,312.10 4,343.60 ROW with crossings (track miles) 32.20 544.90 3,253.70 3,830.80 Number of crossings 27 1,386 2,661 4,074 shared ROW (track miles) 0.00 248.00 85.70 333.70 Number of crossings 0 2,279 66 2,345 subtotal at-grade (track miles) 768.90 1,087.70 6,651.50 8,508.10 Above grade aerial structures (track miles) 485.90 62.90 66.80 615.60 track on elevated fills 100.50 57.80 458.70 617.00 subtotal above grade track 586.40 120.70 525.50 1,232.60 Below grade tunnels (track miles) 794.40 66.10 39.00 899.50 track in open cuts (track miles) 59.80 46.70 68.10 174.60 subtotal below grade (track miles) 854.20 112.80 107.10 1,074.10 Total track miles (including double track, yard, sidings) 2,209.50 1,321.20 7,284.10 10,814.80 (Reference: National Transit Database, the FTA). The rail transit industry is composed of 61 transit systems, predominantly under public management, that receive subsidies from local, state, and federal grants, formula distributions, and agreements. In 2004, these agencies expended a little more than $10 billion on rail transit system maintenance and improvements of the FTA-reported total $44 billion expenditures for public transportation (includes buses, rail, and paratransit). 12

TABLE 2 RAIL TRANSIT INDUSTRY COSTS Cost Heavy Rail Light Rail Commuter Rail Totals Capital Expense Guideway $1,398,244,515 $1,413,882,577 $936,633,072 $3,748,760,164 Systems $495,753,019 $149,530,198 $83,501,424 $728,784,641 Stations $977,821,226 $240,246,591 $389,902,370 $1,607,970,187 Maintenance facilities $349,769,250 $126,473,275 $155,947,263 $632,189,788 Revenue vehicles $329,551,033 $380,843,591 $726,291,642 $1,436,686,266 Other capital $174,862,274 $115,533,225 $259,839,969 $550,235,468 Other vehicle amount $18,472,816 $3,597,565 $4,177,254 $26,247,635 Administration buildings $11,910,417 $660,499 $4,407,005 $16,977,921 Fare collection equipment $39,391,281 $10,498,514 $16,158,436 $66,048,231 Total capital expense $3,795,775,831 $2,441,266,035 $2,576,858,435 $8,813,900,301 Operating Expense (facilities only = “non-vehicle”) $1,224,234,345 $156,016,534 $623,914,117 $2,004,164,996 Total (Capital + Operating) $5,020,010,176 $2,597,282,569 $3,200,772,552 $10,818,065,297 TABLE 3 RAIL TRANSIT INDUSTRY UNIT COSTS Heavy Rail Light Rail Commuter Rail Average of Modes Average guideway cost per mile (guideway capital expense/total track miles) $632,833 $1,070,150 $128,586 $346,632 Average guideway cost per rider (guideway capital expense/annual riders) $0.51 $4.04 $2.26 $1.07 Average total cost per mile (total cost/total track miles) $2,272,012 $1,965,851 $439,419 $1,000,302 Average total cost per rider (total cost/annual riders) $1.83 $7.42 $7.73 $3.08 Guideway (that portion of the transit line included between all outside lines of curbs or shoulders, etc., and all appurtenant structures) costs represent about one-third of the total rail transit expenses; the FTA data include new extensions as well as maintenance funds. The industry maintains more than 5,000 route miles (almost 11,000 track miles, including double track, sidings, and yards), serving about 3.5 billion riders per year. 13

14 For the following data presentations: • Commuter rail is presented in this literature review for perspective. The panel chose not to include it in the scope of this project. Cog railroads, excursion (generally historic) railroads, cable car operations, and trams are other acknowledged rail-based systems that are not covered in this report. • Ridership is shown as the number of individual boardings, meaning that an individual passenger may be counted multiple times in traveling to a destination if the traveler transferred to a separate transit system, and separately for a return trip on the same route. • Transit industry configurations (track sections, stations, etc.) are from industry infrastructure databases, such as the FTA and APTA. (See the Bibliography for sources.) Heavy Rail, Light Rail, and Commuter Rail Tables 1–3 show the three major forms of rail transit: heavy rail, light rail, and commuter rail. These forms have many things in common, but they also have many differences that affect track maintenance. A heavy rail system is completely separate from public vehicles and other rail modes, and operates on its own “exclusive” ROW. It is 100% electrified, with power typically delivered by a “third rail.” Light rail is also separated from other rail modes, but does having street crossings (“at-grade crossings”). In many cases, it shares traffic lanes in streets with automobiles. Light rail is generally electrified, with power delivered by an overhead catenary. Commuter rail shares track with freight rail operators, using passenger cars that are very similar to traditional intercity passenger coaches. (Some operations successfully use refurbished passenger coaches.) Many commuter rail trains use either diesel locomotives or overhead catenary electrified power.

TABLE 4 RAIL MODE COMPARISON Heavy Rail Light Rail Commuter Rail Right-of-Way Exclusive, not shared with any other rail or auto facilities. No grade crossings. Not shared with other rail modes. Has street crossings and shares traffic lanes. Shared with freight Typical Motive Power Third rail Catenary Diesel locomotives Axle Load (maximum or “crush” load) 30,000 lb 30,000 lb 70,000 lb (locomotive) Speeds (route dependent): • Maximum • Typical 80 mph 50 to 75 mph 60 mph 35 mph (or street speed) 79 mph* 60 mph Train Traffic Density (average trains/year, over maximum density route) ~38,000 trains/year ~26,000 trains/year ~6,000 trains/year (commuter trains only) Ridership (annual trips) 2,747,616,634 349,915,503 413,898,363 *100+ mph permissible under special FRA dispensation. 15

Third rails and insulators are normally mounted on cross ties, often maintained by track departments. Overhead catenary equipment is separated from the track structure and typically is not a track maintenance responsibility. Table 4 offers the attributes of the three major transit modes, with emphasis on the differences that affect track maintenance costs. In this table, heavy rail operates at higher speeds with higher train density than the other rail modes. Light rail can approach the operating speeds of heavy rail where the system route is used exclusively by the trains (i.e., without grade crossings). However, these systems largely operate within city centers. When the alignment is along or within roadways, some states require the train speed to be the same as the posted automobile speed. Commuter rail service is a lighter density form of transit, with the track usually shared with mainline freight operations. The allocation of track maintenance costs between the freight and commuter trains is somewhat difficult to define objectively. The principal point of this review is to illustrate that different rail modes produce differences in track maintenance demand resulting from different track loads, track configurations, and traffic. Track configurations are the subject of the following section. Track Standards and Configurations Affecting Maintenance Track maintenance costs can vary with system configurations. Light rail systems’ general deployment within city infrastructure inherently limits speeds. Light rail, by definition, has a far greater percentage of track within city streets. Light rail systems typically allow shorter permissible radii in curves and more severe vertical curves than the other modes, to accommodate street constraints. Heavy rail’s dedicated ROWs allow higher speeds but negotiate inner cities by means of elevated guideways and tunnels in greater percentages of route length than light rail. These configurations introduce more specialty track, such as Direct Fixation track, than other rail modes. Light rail and heavy rail systems are nearly universally electrified, adding catenary or third rail facilities as a significant asset for maintenance compared with non-electrified railroads. Added maintenance costs from electrification are additional track components (primarily in third rail systems), 16

17 corrosion of track components as well as structures and utilities surrounding the track, and added safety work rules. Track design standards and track maintenance standards significantly affect the cost of track maintenance. Track maintenance standards for rail transit have a basis in the broader industry standards of AREMA, with some exceptions for track gauge and curvature (noted earlier). However, light rail and heavy rail transit agencies’ track materials installations appear to lack uniformity. The lack of standardization adds a significant premium to track material delivery times, component unit costs, agency spare parts stocking costs, and maintenance processes (including training to recognize differences among similar components, costs of errors related to component incompatibility, and train delays associated with errors). TECHNOLOGY INFLUENCE ON TRACK MAINTENANCE COSTS Track maintenance costs vary dramatically over time, usually beneficially. In the early days of railroads, it was usual to replace cross ties each year (1); today, the expected life of a timber cross tie is on average 40 years (2). Figure 1 shows unit costs for major track cost components in the U.S. freight industry. They are based on the Association of American Railroad’s (AAR) Total Right-Of-Way Analysis and Costing System (TRACS), an empirically calibrated model of freight railroad maintenance-of-way (MOW) costs from 1970 to 2000.

FIGURE 1 MOW cost breakdown by maintenance component for eastern 30 MGT mainline. The improvements in maintenance costs are attributed to improvements in technology and practices. Rail cost improvements contributed the most to the overall cost savings. Rail costs fell 58%. From $0.62/1000 GTM under 1970 steady-state conditions to $0.26/1000 GTM in 2000. This dramatic cost improvement was due to the introduction and refinement of rail grinding and lubrication techniques and due to the replacement of old rail by much more durable, higher quality rail, especially on curves. The total MOW costs fell 37% under the assumptions of this scenario, from $1.13/1000 GTM to $0.71/1000 GTM. Rail costs were the largest component of MOW costs, comprising 55% of the total in 1970 and 37% in 2000. Ties were the second largest contributor, consisting of 22% of total costs in 1970 and 34% in 2000. Routine maintenance, ballast and surfacing, and turnout costs all contributed approximately equally to the overall costs. Turnout costs did improve slightly more than the ballast and surfacing and routine maintenance costs due to more widespread use of better quality turnouts that have lower angles of incidence and also due to the elimination of underutilized turnouts from the network. Tie costs rose from 1985 to 2000 due to the introduction of premium fastenings on track with high degrees of curvature, despite a reduction in plate cutting and spike kill by using larger tie plates in 2000 (3). 18

19 These trends would be expected to be available in transit. The unit costs from freight are useful in suggesting probable upper limits for what may occur in transit if it is assumed that the maximum axle load density or frequency of maximum load occurrence is greater in freight than in rail transit. MODELING ATTEMPTS TO ESTIMATE MAINTENANCE COSTS Past track maintenance cost-estimating methods generally use statistical techniques (which we will refer to as models) that analyze various factors to associate past costs with influence parameters such as rail traffic characteristics, component age, and resources, to predict maintenance demand and costs. The models in this review have different purposes that range from establishing rail rates (for freight railroad regulatory and legal issues) to attempting maintenance planning (for budgeting, rehabilitation assessment, public funding evaluation, and railroad projections). The different purposes have led to different models that at best should be used with caution and at worst can be misleading for other purposes. General Track Maintenance Modeling Concepts It is useful to define a basis for a model’s assumptions, constructions, and underlying data. Track maintenance “models” presume that railroad traffic produces track maintenance. The presumption fails to capture yard track, as an example, even though yards certainly are a significant system track maintenance cost. Track maintenance modeling also presumes that track maintenance is predictable; that is, there are quantifiable, mathematical relationships between influence factors (e.g., size of loads, number of wheels, curvature, and weather) and maintenance effort. This premise requires encapsulating two types of processes. The first process is a rate of degradation for track and its components. To predict maintenance demand, the degradation mechanics (wear, settlement, etc.) must be known sufficiently to develop accurate predictions. The second process is track management, or the management of the track degradation process through maintenance. Philosophies (i.e., preventive maintenance or crisis maintenance), cyclic budget resources, changes in public officials effect funding, and execution of track maintenance often vary dramatically. Maintenance management processes are absent in any discrete form from all track maintenance prediction models.

20 In addition, most transit cost modeling efforts depend on past documentation (maintenance records, expenditure records, etc.) that may be incomplete, and may use varying accounting systems. Models based on maintenance data should have data over multiple economic cycles and parallel information on funding streams to understand the data. A competent track maintenance model would have a database of the full system, including alignment, components, traffic by specific location (route, track, and engineering station), and the dates of installation of each component. Ideally, maintenance costs are then assigned to each track segment as the maintenance occurs. Predictions can then be based on site-specific traffic and maintenance conditions. Track Maintenance Models The following track maintenance models included in this literature review are indicative of past and current approaches: 1. AAR HAL Phase II Economic Analysis (1997) 2. Total Right-of-Way Analysis and Costing System (TRACS—1994) 3. Degradation Cost of Track (2004) (4) 4. Cost Sharing Allocation Models: a. Speed Factored Gross Tonnage (SFGT—AAR) b. Weighted System Average Cost (WSAC—AAR) c. Commercial Feasibility Study (FRA) d. TrackShare (Zeta-Tech) e. Swedish Railway Marginal Costs (2006) The routine maintenance component (i.e., daily routine of inspections, spot repairs, adjustments, minor repairs) is described as a differential between a base maintenance demand and the demand from increased axle loads (5): MGTBRMNALNRM DFE * 33 ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛= Where: NRM = new axle load routine maintenance (hours),

33 = 263,000-lb freight car axle load (tons), NAL = new axle load (tons), DFE = estimated damage factor exponent, BRM = base (33-ton) routine maintenance (hours/MGT/mile), and MGT = traffic density (MGT/mile/year). Compared with freight rail operations, train operations increase transit maintenance costs. Train operations in transit either interfere with maintenance operations or restrict track access to narrow time periods. The cost amplification from traffic interference may be 50% of the basic cost of maintenance (Figure 2) (5) based on European data. FIGURE 2 Effects of train interference on maintenance costs. (Source: AREMA 2001 Annual Conference.) Consideration of track renewals as an investment has merit, predicated on savings from technology gains (see preceding section, Technology Influence on Track Maintenance Cost). The Massachusetts Bay Transportation Authority took this approach, suggesting that track rehabilitation, in part, is an investment (6) with a measurable return in system efficiencies, including track maintenance cost. The system was evaluated to consider the current age of components, life expectancy, and rate of return on reductions in maintenance for a system in an “ideal” state of repair. The results suggest that timely replacement of aging components has a positive influence from technology implementation on long-term cost trends. Among others, Keeler (7) and Caves et al. (8) used aggregate data on U.S. Class I railroads with a focus on scale economies (i.e., economies with increasing railroad size) and productivity to generate cost- function estimates for the railway industry. The results suggest that system size reduces unit maintenance 21

22 costs (i.e., cost per mile); however, whether those benefits are from efficiency of scale or spreading overhead costs is not clear. Other studies have attempted to discern whether costs of track (a “marginal” or influencing cost) have measurable influence on a system’s overall cost performance. Recently, Bitzan (9) and Bereskin (10) conducted studies using aggregate U.S. post-deregulation data, with the latter study estimating marginal costs for MOW among different railroad organizations. European studies that focus on marginal costs and use micro-level data include those by Daljord (11), Tervonen and Idstrom (12), and Gaudry and Quinet (13). Johansson and Nilsson (14) estimated marginal costs using Sweden’s railway network, but this study does not backtrack data from 1994 to 1996, and no analysis has been undertaken on data network changes over time. The study was inconclusive on track maintenance costs. Lack of data resulting from such factors as mergers and acquisitions (eliminating sufficiently long time series) and a focus on day-to- day operations has often restricted micro-level analyses in the railway sector (15). In these modeling efforts, the relationship of track costs to exposure to traffic appeared to be weak. The studies that had variables for specific track costs found that railroad traffic had little or no effect (i.e., was not statistically significant) on track costs. In this vein, Anderson (16) used statistical modeling to estimate the track cost per train and per gross tonne (tonne = metric ton) for track and operating influence parameters using Swedish Rail Agency data from 1999 to 2002. The track parameters in the model include track length, rail lubrication, rail weight, tunnels, bridges, track alignment (curves, grade, superelevation), rail joining (mechanical joints, CWR), and the ages of track components (rail, ties, ballast, switches). Line segment tonnage is used as the independent variable. The model provides a base or constant maintenance cost with added premiums (or marginal) costs for each of the track parameters in the model. Anderson found a significant interaction among track parameters. As might be expected, the results show that rail weight and age affect switch costs. Similarly, track length and alignment affect the other parameters. The track maintenance costs were modeled to a reasonable statistical reliability by including these interactive influences between parameters in the model. However, the statistical relationship between rail traffic and track maintenance was statistically weak, bordering on insignificant. Anderson also found that the results support the concept of economy of scale, implying that efficiencies in track maintenance cost are available as the rail system size (length) increases.

23 LITERATURE SUMMARY The literature on track maintenance costs in this review touches on four points: • Track maintenance costs can be viewed as investments to leverage the benefits of advancing technology, with benefits to system performance and costs as well as to track maintenance unit costs. • Rail transit maintenance costs are higher as a result of train interference (i.e., lack of track access due to train operations, as well as points of entry to the track) compared with rail operations with permissive track maintenance access. • Unit costs may scale inversely with system size, with costs declining as the system size increases. However, the literature is not consistent on whether the benefit exists and does not quantify the amount of the benefit. • Statistical models of rail system costs reviewed appear to have difficulty discerning the influence of rail traffic on modeling results. The last point reflects flaws in past modeling efforts for track issues, including costs. Estimating or predictive models might include the following details: • Route-specific configurations, such as distance, alignment, and turnouts; • Track component details, such as type of switches, frogs, ties, and rails; • Mechanics of vehicle-track interaction (dynamic and kinematic motion) and wheel-rail interaction (curving mechanics); • Track geometry (alignment and support deviations); • Traffic characteristics by location; and • Cost/maintenance documentation by the above parameters to validate the model.

24 However, that the economic modeling efforts identified in the literature review conducted during this study did not show any strong links between maintenance costs and the amount of rail traffic is likely because the modeling efforts identified used statistical methods relying on general characteristics of the traffic and vehicles rather than specific rail vehicle characteristics, track characteristics, and local rail traffic characteristics. Models do exist, however, that relate track maintenance costs to the specific characteristics of track, vehicles, and traffic. These models and the results of their specific applications are usually proprietary and not available to the public.