Wheel Profile Maintenance Guidelines (2015)

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.

TCRP WOD 65 Part 1 Wheel wear is another reason for wheel truing. Wheels wear into worn shapes in service, with most wear on the tread and flange, as Figure 2 shows. Figure 2. Worn, Unworn Wheels, and Unworn Rail Wear on wheel treads and flanges is usually characterized by measuring tread hollowness and flange thickness. The worn wheel in Figure 2 is a hollow wheel with hollow depth labeled as “Wheel hollowness”. For rail transit agencies running on track governed by FRA rules or adopting Association of American Railroads (AAR) interchange rules (AAR 2012), the tread hollow limit is 0.158 inch (4 millimeters (mm)), and the flange thickness limit is 0.938 inch (23.8mm). It should be noted that the 0.158 inch (4mm) hollow limit was adopted by AAR in 2004. The survey showed that most rail transit agencies have a flange thickness limit (0.938 inch (23.8mm) or 0.875 inch (22.2mm)), but do not have a wear limit on tread hollowness. One possible reason is that wheels in most rail transit systems do not wear as severely hollow as freight railroad wheels do before they are trued. However, hollow wear on wheels is unavoidable, and hollow wheels do cause problems in transit cars. Observations showed that hollow wheels in rail transit cars lead to car lateral instability (hunting) and degraded ride quality (Smith and Kalousek 1991); however, no comprehensive study of the effects of hollow wheels or wheels with mismatched diameters on transit vehicle stability has been performed recently. Instability caused by hollow wheels is commonly observed in freight car operation. A freight car stability test was performed by TTCI in 2001 (Sawley et al. 2005). The “new” wheelsets used for the test had been recently turned (at less than 5,000 miles) with AAR-1B narrow flange profiles. The worn wheelsets selected were revenue service worn wheels with moderate hollow wear. Table 2 lists the details of the wear of these worn wheelsets as installed for the first series of worn-wheel tests. These tests were termed as Configuration 1. Table 2 shows that the wheelsets in Configuration 1 are diagonally worn within a truck; e.g., the L1 and R2 wheels are hollow, whereas the R1 and L2 wheels are not. To test whether the hollow wheel pattern in a truck influences the stability, tests were conducted using 5

TCRP WOD 65 Part 1 Configuration 2. As Table 3 shows, the left and right side of the first and fourth axles were exchanged so each truck had hollow wheels only on one side. Table 2. Details of Worn Wheels in Configuration 1 (Freight Car Stability Test) Axle No. Wheel ID Hollow Wear Flange Wear* 1 L1 0.075 inch (1.9mm) 1.047 inch (26.6mm) R1 0 1.339 inch (34.0mm) 2 L2 0 1.350 inch (34.3mm) R2 0.102 inch (2.6mm) 1.055 inch (26.8mm) 3 L3 0.051 inch (1.3mm) 1.201 inch (30.5mm) R3 0 1.284 inch (32.6mm) 4 L4 0 1.205 inch (30.6mm) R4 0.039 inch (1.0mm) 1.164 inch (29.6mm) * Measured 0.625 inch (15.9mm) up from a point on the tread 3.0625 inch (77.8mm) from the back face of the wheel. Table 3. Details of Worn Wheels in Configuration 2 (Freight Car Stability Test) Axle No. Wheel ID Hollow wear Flange wear* 1 L1 0 1.339 inch (34.0mm) R1 0.075 inch (1.9mm) 1.047 inch (26.6mm) 2 L2 0 1.350 inch (34.3mm) R2 0.102 inch (2.6mm) 1.055 inch (26.8mm) 3 L3 0.051 inch (1.3mm) 1.201 inch (30.5mm) R3 0 1.284 inch (32.6mm) 4 L4 0.039 inch (1.0mm) 1.164 inch (29.6mm) R4 0 1.205 inch (30.6mm) * Measured 0.625 inch (15.9mm) up from a point on the tread 3.0625 inch (77.8mm) from the back face of the wheel. Figure 3 is a plot of the maximum standard deviation of carbody lateral acceleration over 2,000 feet (609.6meters (m)) of track. It shows a critical speed of 55 miles per hour (mph) (88.5kilometers per hour (kmh)) for the new AAR-1B wheels and a critical speed of 50 mph (80kmh) for the hollow wheels. It also shows that the car with hollow wheels has higher lateral accelerations at speeds below the onset speed than is seen with new AAR1B wheels. It also shows that the hollow wheel distribution pattern (hollow wheel on one side of the truck or diagonally implemented in the truck) has little effect on stability. 6

TCRP WOD 65 Part 1 Figure 3. Freight Car Hollow Wheel Stability Test Wheel wear on treads can lead to the formation of false flanges. There are two types of false flange: one kind is due to wheel hollowing so the false flange is on the field side of tread; the other kind is a raised ridge in the flange root. Field side false flanges with severe hollowness on tread not only generated high contract stresses on low rail in curve, but could also cause stock rail (in a switch) rollover derailment (Kerchof 2004, Wolf 2006). Wheel wear on treads and flanges could also generate high impact on frogs. Hollow wheels (wheels with false flange) were present in the New York City Transit system (Cabrera and Gobbato 2000), which contributed to the fast wearing of the Frog noses and risers on standard frogs, and the associated vibration and noise. It was recommended in the report to investigate the most appropriate profile for re-trued wheels that will counteract the development of false flanges. It was also suggested to limit the false flange (hollowness) to a maximum of 0.125 inch (3.2mm), as part of the wheel maintenance criteria. The Denver RTD light rail transit system uses a 0.05-inch (1.3mm) tread wear (hollowness) limit for wheel maintenance. The wheel profile was checked against the template profile every 40,000 miles, and the wheel is trued if the tread shows 0.05-inch (1.3mm) gap. Setting up a wear limit for wheels is a complicated issue. It depends on vehicle and track design, maintenance standards, and operational environment. Wear limit also needs to be justified through economic analysis. Section 3 discusses the effects of wheel wear on contact geometry and vehicle performance. 2.3 Wheel Diameter Differences in One Axle The two wheels on one axle can wear into asymmetric shapes, which results in a wheel diameter difference on one axle. Wheel diameter differences caused by wear or truing has significant effects on car performance. 7

TCRP WOD 65 Part 1 No comprehensive study of the effects of wheel diameter differences on rail transit vehicle stability has been performed recently. In 2006, a freight car tolerance study (Tunna et al. 2006) was conducted to investigate the effect of the opposite wheels on the same axle with mismatching circumferences. As Figure 4 shows, the lateral accelerations of a car with one tape size mismatch in circumference increased gradually with speed above 40 mph (64.4kmh). In contrast, the car with matching wheel circumferences had a sudden increase in lateral accelerations at the critical speed. In the case of the one tape mismatch, the acceleration was above the AAR Chapter 11 limit of 0.13 g for speeds less than or equal to 70 mph (112.7kmh) — at the time of the study, the AAR lateral acceleration limit was 0.26g so at that time neither case exceeded the limit below 70 mph (112.7kmh). Similar trends can be found in Figure 3, even though these two studies were conducted for different purposes. The same conclusion is that wheel diameter differences, caused by either asymmetric wear or by wheel truing, can lead to car instability. Figure 4. Effect of Wheel Diameter Difference on Freight Car Stability (1 Tape equals 0.125 inch on the circumference) When a wheelset has wheels with mismatched circumferences, longitudinal wheel and rail creep forces are generated that steer the wheelset away from the centerline. The conicity of the wheels reduces the rolling radius difference (RRD) until an equilibrium rolling line is reached. The wheelset may have gained enough momentum to pass the equilibrium rolling line and develop longitudinal creep forces in the opposite direction. In this way, a cyclic pattern of wheelset displacement and forces can develop. This is shown for a wheelset with mismatching wheel circumferences in Figure 5. The cyclic pattern begins at the start of the run, before the lateral input to excite hunting, and continues throughout. In contrast, for a wheelset with equal wheel circumferences, longitudinal forces are produced by the lateral track input, but these soon die away. The wavelength of the oscillations of the normal wheelset is approximately 50 feet (15.2m), which corresponds to the wheelset’s kinematic wavelength. The wavelength of the oscillation of the wheelset with wheel circumferences that mismatch by one tape is approximately 17 feet (5.2m). Three cycles of oscillation appear to be in one kinematic wavelength. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 20 30 40 50 60 70 80 90 Speed (miles/hour) La te ra l A cc el er at io n (g ) Normal 1 Tape Speed (mph) 8

TCRP WOD 65 Part 1 Figure 5. Longitudinal Forces on the Right Wheel of the Leading Wheelset at 60 mph (96.6kmh) Clearly, both test and simulation show that car stability and ride quality are sensitive to wheel diameter differences in one axle. It is one of the key parameters for rail transit agencies to control in service and maintenance quality, as Table 4 shows. Table 4. Transit Agency Wheel Diameter Tolerances System Diameter Tolerance after Wheel Re-profiling Denver RTD 0.05 inch (1.27mm) within an axle 0.05 inch (1.27mm) for DS100 Truck, 0.25 inch for SD160 Truck 2.3 inches (58.4mm) truck-to-truck within the same car PATH 0.125 inch (3.2mm) variation left-to-right within an axle 0.125 inch (3.2mm) axle-to-axle within a truck 0.25 inch (6.4mm) truck-to-truck within a car BART In service, 0.03 inch (0.8mm) within an axle; After cutting, 0.005 inch (0.13mm) 0.3125 inch (7.9mm) axle-to-axle within a truck 0.50 inch (12.7mm) truck-to-truck within a car SEPTA 0.125 inch (3.2mm) within the same axle 0.25 inch (6.4mm) axle-to-axle in the same truck 0.50 inch (12.7mm) truck-to-truck in the same car WMATA 0.0625 inch (1.6mm) within the axle 0.25 inch (6.4mm) axle-to-axle in the same truck 0.50 inch truck-to-truck in the same car Chicago Metra Electric 0.125 inch (3.2mm) variation left-to-right within an axle 0.25 inch (6.4mm) axle-to-axle within a truck 0.25 inch (6.4mm) truck-to-truck within a car CTA 0.047 inch (1.2mm) within an axle 1 inch (25.4mm) axle-to-axle in the same truck 1 inch (25.4mm) truck-to-truck within the same car -3000 -2000 -1000 0 1000 2000 3000 4000 0 100 200 300 400 500 600 700 Distance (feet) Lo ng itu di na l F or ce (l bf ) 1 Tape Normal 9

TCRP WOD 65 Part 1 MiniProf™ (Greenwood Engineering A/S, Denmark) profilometers and software have been widely used for wheel and rail cross section profile measurement. The twin-head MiniProf can help to make accurate measurement of the two wheels on same radial position, but it cannot be used directly to measure wheel diameter. Recent developments in MiniProf extended its function to calculate wheel diameter, but additional input of inner diameter has to be measured with special MiniProf wheel instruments. Wheel tapes can also measure diameter, but the location of measurement is dependent on the flange wear. A portable wheel diameter measurement device is needed for rail transit agencies to accurately measure wheel diameters and variations caused by asymmetric wear or machining. The allowable wheel diameter difference maintenance limit depends on the vehicle and truck design, especially the truck suspension and other component maintenance limit. The effect of wheel diameter difference on vehicle performance will be investigated in Task 2 of this study. 2.4 Wheel Diameter Tolerance in One Truck The wheel diameter difference in one truck does not affect car stability, but it may affect wheel load equalization. The freight car tolerance study (Tunna et al. 2006) showed that the vertical car performance affected by mismatching circumferences between the leading and trailing wheels were the empty cars with pitch and bounce track inputs. Figure 6 shows the minimum vertical load on the left wheel of the leading wheelset. The minimum vertical load is lower with the mismatching wheel circumferences, but it is still above the AAR Chapter 11 limit. Figure 6. Effect of Mismatching Wheel Circumference on the Leading and Trailing Axle with Empty Freight Car and Pitch and Bounce Input 0% 20% 40% 60% 80% 100% 120% 0 10 20 30 40 50 60 70 80 90 Speed (miles/hour) M in . V er tic al L oa d Normal 1 TAPE Ch. XI Limit Max 11 Speed (mph) 10

TCRP WOD 65 Part 1 The effects of wheel diameter difference in one truck on transit vehicle dynamic performance, such as wheel load equalization, will be examined based on American Public Transportation Association (APTA) criteria (APTA 2007) in Task 2 of this project. 2.5 Wheel Diameter Tolerance in One Car The wheel diameter difference from truck-to-truck in one car may cause unbalanced loading, leading to carbody vertical and pitch vibrations and deteriorated vertical ride quality. These effects were investigated in Part 2 Report of this study. 2.6 Wheel Flange Angle The maximum flange angle of the designed wheel profiles applied in transit operation ranges between 63 and 75 degrees. Table 5 lists the wheel flange angles adopted by different rail transit agencies. Table 5. Maximum Flange Angle of Transit Agency Designed Wheels System Light Railcars* (degrees) Heavy Railcars** (degrees) Commuter Railcar*** (degrees) Denver RTD 66 NA NA PATH NA 68 NA BART NA 68 NA MBTA 72 NA 75 NJTC 75 72 SEPTA 60-65 (in specified tolerance) 63 72 WMATA 63 Chicago Metra Electric 75 CTA 68 Houston Metro 70 DART 70 *Light railcars: Two trucks or three trucks with articulation. Examples include MBTA (Boston, green line), Denver RTD, NJ Transit, Baltimore, Pittsburgh, Charlotte, MUNI (San Francisco), San Diego, San Jose (Valley), Portland, St. Louis, and SEPTA. These cars can be high floor, low floor, or a combination of both, and are formerly referred to as street cars or trolley cars. **Heavy railcars: These types of cars have two trucks, examples include: NYC Transit, PATH, SEPTA (Philadelphia, subway), WMATA, MARTA, Baltimore, CTA, Los Angeles, MBTA (Boston) and BART. ***Commuter railcars: These types of cars have two trucks, examples include: Metro North, LIRR, METRA (Chicago), SEPTA (Philadelphia, commuter service), Caltrans (California), MARC (Baltimore), MBTA (Boston). Increasing the design wheel flange angle to reduce the risk of flange climb derailment has been a common practice for rail transit agencies. Due to historic reasons, some older rail transit agencies have adopted the relatively low wheel flange angles in the range of 63 to 65 degrees. The low flange angles are prone to flange climb derailment and have less compatibility with different truck designs. Newer rail transit agencies generally start with a wheel profile having a flange angle of 72 to75 degrees, as recommended by APTA (APTA SS-M-015-06 2007). 11

TCRP WOD 65 Part 1 A wheel profile with a higher flange angle can reduce the risk of flange climb derailment and can have much better compatibility with any new designs of vehicles and trucks that may be introduced in the future compared to wheels with lower flange angles. Also, with a higher lateral to vertical (L/V) ratio limit, high flange angles will tolerate greater levels of unexpected track irregularity. Measurements showed the wheel flanges usually wear into a steeper flange angle in service, which decreases the flange climb derailment risk (Shu and Tunna 2007). However, wheels with high flange angles and nonconformal contact on rails can result in higher wear rates than wheels with shallow flange angles. Rail transit agencies suffering flange climb derailments may need to change to wheels with steeper flange angles to reduce derailments, but they may also have to contend with excessive wear through rail grinding and lubrication (Griffin 2006). A detailed study on wheel flange climb derailment and criteria was published from a previous study for TCRP (Wu et al. 2005). 2.7 Other Wheel Inspections Transit railcar maintenance also includes inspections of any cracks or fatigue shells on wheel flanges and treads. If a crack or shelling defect on the tread or flange surface is larger than 0.375-inch (9.5mm) diameter, the wheel is usually considered nonserviceable. If a crack or shelling defect on the tread or flange surface is smaller than 0.375-inch (9.5mm) diameter, the wheel is to be scheduled for truing. Other wheel inspections may include the following: Wheel separation Loose retaining ring if applicable Rotation of wheel hub with respect to axle and/or wheel tire with respect to wheel center Safety wired axle caps 2.8 Wheel Truing 2.8.1 Truing Cycle and Wheel Life Wheel truing is performed to remove any defects on wheels such as flats, shellings, and spalls, or to restore the worn wheel shape to a designed shape (truing template). Wheels are usually trued several times until the wheel rim thickness reaches its limits. Table 6 lists the fixed truing cycles and wheel life periods in different rail transit agencies. Table 6. Transit Agency Wheel Truing Cycles and Wheel Life System Truing Cycle Wheel Life Denver RTD 40,000 miles 400,000 miles (6~7 years) Houston Metro 30,000 miles Power Truck 350,000 miles Center Truck 250,000 miles PATH 3 years New fleet about 8 years with 40,000 miles/year SEPTA (Streetcar lines) 150,000 miles 10 years WMATA 1 year 400,000 miles or 4.5 years NJT Commuter Rail 60,000 miles 250,000 miles CTA NA 5 years BART 108,000 miles 324,000 miles 12

TCRP WOD 65 Part 1 Most rail transit agencies trued wheels based on fixed cycle (certain number of mileage or service years) regardless of condition. Many rail transit agencies also trued wheels because of diameter mismatch, not just hollow or flange wear. Some rail transit agencies trued wheel either in a fixed cycle or based on the service conditions. For example, NJT wheel truing (light rail) is performed either at fixed intervals, i.e., every 30,000 to 40,000 miles depending on the truck design, or as periodic measurements indicate the need for corrective action (Lovejoy et al. 2012). 2.8.2 Truing Surface Roughness Two types of wheel re-profiling machines are commonly used. Figure 7 shows the milling type that has a cutting head with many small cutters. The arrangement of the cutters forms the wheel profile. Figure 8 shows the lathe type truing machine. The single cutter cuts the wheel by following the shape of a template. Figure 7. Milling Type Wheel Truing Machine 13

TCRP WOD 65 Part 1 Figure 8. Lathe Type Wheel Truing Machine Several rail transit agencies have reported flange climb derailments occurring at curves or switches in yards when the cars were just out of the wheel re-profiling machines. This type of derailment is often caused by the wheel surface roughness after wheel truing. Figure 9 shows the rough wheel surface just after truing with a milling type truing machine. The wheel surface trued by a lathe type truing machine is usually smoother than that trued by a milling type machine. 14

TCRP WOD 65 Part 1 Figure 9. Rough Wheel Surface from Milling Type Truing Machine The rough surface produced by wheel truing increases the effective coefficient of friction between wheel and rail, which significantly reduces the L/V ratio limit for flange climb. A low flange angle further increases the derailment risk. Several remedies may improve the surface condition: Frequently inspecting the cutting tools  especially for the milling type machine. Dulled tools can produce a very rough surface. Sometimes the grooves on the wheels were obvious. Addressing the final surface turning. In this step, there is no significant material removal, but rather a light cut for smoothing the surface. WMATA has included this step in its wheel re-profiling procedures. Further, lubrication after wheel truing can be an effective way to prevent flange climb derailment on newly trued wheels. WMATA now manually lubricates all wheels immediately after truing. CTA has installed wayside lubricators on the curves as well as guardrails in their yards to prevent derailment. 2.8.3 Truing Templates Most rail transit agencies trued wheels to restore their shape to the original design shape. The original wheel design profiles usually came from car manufacturers. They may or may not be the optimal profiles for the vehicle and track in a transit, according to current understanding of rail and wheel contact mechanics. An optimal wheel profile has to be compatible with the rail profiles in various track configurations, including tangent, curves, and special trackwork, such as switches, frogs, and guards and restraining rails. 15

TCRP WOD 65 Part 1 Once a new wheel profile has been accepted, any changes to the wheel profile (especially tread and flange width) must be evaluated by both vehicle and track designers. New wheels generally wear much faster during the wear-in period, and then reach a relatively stable shape compatible with rail shapes. The length of the wear-in period depends on the conformality of wheel and rail shapes. The design wheel profile is usually compatible with new rail profile, but less compatible with the worn rail shape. The worn wheel does not necessarily need to be restored to its design shape; instead, it should be trued to be compatible with the majority of the existing worn rails. NJT has developed intermediate wheel profiles for their wheel truing template. The template is determined by software incorporated in the wheel truing machines. As many as 20 variants of corrective actions are recommended by the machine so as to minimize the removal of metal from the wheels. With this program in place, the NJT light rail system has increased resilient wheel life dramatically, typically achieving 200,000 to 250,000 miles of service before tire replacement is necessary (Lovejoy et al. 2012). The survey of PATH found that the design wheel generates two-point contact on the worn rail, as Figure 10 shows. The two-point wheel/rail contact results in not only high contact stress and wear, but also poor steering performance. However, the worn wheel has a conformal contact on worn rail, as Figure 11 shows, which is in favor of good truck curving performance. Figure 10. New Wheel Contact on Worn High Rail 16

TCRP WOD 65 Part 1 Figure 11. Worn Wheel Contact on Worn High Rail The design wheel profile (truing template) used by PATH is not compatible with the majority of the existing rails, and thus it is not recommended for wheel truing. Detailed wheel/rail dynamic analysis will be conducted in Task 2 of this study to generate guidelines for wheel maintenance, including how to generate an optimized wheel profile for truing. A new wheel truing template for PATH will be proposed to demonstrate the procedures using these guidelines. 17

TCRP WOD 65 Part 1 C H A P T E R 3 State-of-the-Art Wheel Profile Design and Maintenance Principles Wheel profiles have a significant effect on wheel/rail contact and overall vehicle and track dynamic performance. A design of a new wheel profile well suited for a specific vehicle, track, and service environment can improve vehicle and track dynamic performance and reduce wear and damage on wheel and rail. In spite of a large number of publications on this topic, wheel profile design still remains a challenge for truck design and maintenance. 3.1 Wheel Profile Design Methodology Researchers have adopted various methods with different targets and strategies to develop a new theoretical wheel profile, and the following are examples of profile design based on: Target RRD function (Smith and Kalousek 1991, Shevtsov et al. 2005) Target contact angle (Shen et al. 2003) Target conicity and wide contact range (Polach 2009) Typically, a wheel profile was designed using a trial and error approach to reach design targets. Wheels cannot be designed without reference to rail profiles. Theoretical or measured rail and/or wheel profiles were usually selected as “seeds” or references during the design process. Smith and Kalousek 1991 developed a numerical procedure for design of a wheel profile described by a series of arcs. Although the procedure was specifically developed for steered axle rail transit vehicles, some important aspects of it can be applied to conventional rail transit systems as well. Shevtsov et al. 2005 proposed a procedure for design of a wheel profile that improves wheel and rail interaction by reducing wear while taking into account rolling contact fatigue. The procedure uses an optimality criteria based on a RRD function. The criteria accounts for stability of a wheelset, minimum wear and contact stresses of wheels and rails as well as safety requirements. Using the proposed procedure, Shevtsov et al. designed a new wheel profile and conducted simulations using ADAMS/Rail software. Shen et al. 2003 proposed a wheel profile design method using a target wheel/rail contact angle function and rail profile information. A computer program was developed to produce an independent wheelset profile for a rail transit car. Polach 2009 investigated the relationship between the equivalent conicity, contact angle, and location of the contact area in nominal position, the contact stress, and lateral contact spreading on worn rail profiles. New wheel profiles were created with a target conicity and at the same time wide contact spreading. Persson and Iwnicki 2004 and Novales et al. 2006 used optimization procedures based on a genetic algorithm to design a wheel profile for railway vehicles. Two existing wheel profiles were chosen as “parents,” and “genes” were formed to represent these profiles. These genes were mated to produce offspring genes and then reconstructed into profiles that had random combinations of the properties of the parents. Each of the offspring profiles were evaluated by running a computer simulation of the behavior 18

TCRP WOD 65 Part 1 of a vehicle fitted with these wheel profiles and calculating a penalty index. An inverted penalty index was used as the fitness value in the genetic algorithm. The method was used to produce optimized wheel profiles for two variants of a typical vehicle, one with a relatively soft primary suspension and the other with a relatively stiff primary suspension. The development of the standard AAR-1B wheel profile for North American railroads involves methodologies of traditional manually modifying wheel profiles, computer–aided analysis (NUCARS®* simulations), and revenue service tests (Leary et al. 1991). Until 1990, the AAR 1:20 wheel profile was the AAR standard for interchange service. Experience showed that a substantial amount of the tread and flange wear occurred before a steady state of wear was established by the wheel. Additionally, when the 1:20 wheel wore in to a new profile after 20,000 to 50,000 miles of service, problems with lateral stability and truck hunting were frequently encountered. The awareness of the need for a new standard profile was heightened by the introduction of a “Heumann” profile wheel by the Canadian railroads. The Canadian National (CN) Heumann profile was created based on expansions of rail shapes. This profile was designed to provide single-point contact with the then prevailing AREMA 115-pound rail section. The reported reduction in wheel wear brought about by this wheel was quite significant. As a result, the Research and Test Department of the AAR was requested by its Mechanical Division to develop an alternative wheel profile that would provide better performance than the 1:20 wheel profile. The development of the now standard AAR-1B wheel profile advanced through four different research phases. The first phase was concerned with the general definition of the problem, accompanied by a statistical field survey of existing worn wheel and rail shapes. In the second phase, a method was derived that allowed the development of statistically and analytically based profiles. These profiles were established from known wheel and rail shapes and kept within generally accepted safety and performance criteria. In the third research phase, the candidate profiles developed in the first two steps were put through a series of acceptance tests, design adjustments were made to the flange throat, and a partial 1:20 profile was added to the tread. In the final phase, revenue service tests were carried out. The results of the wear tests indicated very favorable economics in terms of reduced wear and fuel consumption. Over the last decade, heavy haul operation in North America railroads has been changing the vehicle and track service environment, and freight railroads are facing new challenges related with maintenance of wheel and rail interface (Tournay 2009). In 2006, failures of primary suspension adapter pads and loaded car hunting were reported on a particular type of heavy axle load 286,000-pound grain railcars. Associated with these failures was some degradation of polymer elements in the constant contact side bearings used on these railcars. Initial observations revealed that the railcars were hunting under load at speeds approaching 50 mph (80 kmh). The root causes have been found to be system related: a combination of low truck warp restraint, high wheelset conicity and loaded body inertial and suspension characteristics. This leads to coupled resonance at speeds as low as 47 mph (75 kmh) between the kinematic motion of the wheelset and body yaw. An interim solution is the use of stiffer and more durable adapter pads; longer term solutions are a truck with higher warp stiffness combined with management of the wheel/rail interface to reduce conicities from the observed high values of 0.7. A new standard freight car wheel profile is being evaluated at TTCI under the AAR Strategic Research Initiatives (SRI) Program to replace the AAR-1B profile for improving railcar stability and reducing wheel and rail rolling contact fatigue in freight rail operations (Wu 2007). *NUCARS® is a registered trademark of Transportation Technology Center, Inc., Pueblo, Colorado 19

TCRP WOD 65 Part 1 3.2 Wheel Profile Design and Maintenance Criteria The objectives of optimum wheel and rail profiles are to provide: Stable performance over the range of normal train speeds Safety from derailment under adverse but realistic operating conditions Maximized wheel and rail life Wheel and rail profile design is a matter of optimizing several criteria. Some criteria must be satisfied, but some can be compromised to achieve an overall optimum solution. The following criteria are usually used to design a wheel profile: Lateral Stability — should be achieved for normal operating speeds in the empty and loaded conditions; hunting performance depends on the nonlinear conicity function Maximum Contact Angle — should be greater than 72 degrees to avoid flange climbing derailments; APTA recommends at least 72 degree (suggested tolerance +3° and -2°) angle for commuter cars (APTA SS-M-015-06 2007) L/V Ratio — should be less than 0.8 to avoid flange climbing derailments Wear Index — should be as low as possible to avoid wear on wheels and rail in curves Contact Stress — should be as low as possible; high contact stress contributes to rolling contact fatigue and metal flow Contact Position — should be widely spread to avoid concentrated wear; should not be too far toward the field side of the rail to avoid rail rollover moments Rolling Resistance — should be as low as possible to reduce power consumption and draft forces In theory, all these criteria apply to not only new wheel design, but also to wheel maintenance. However, wheels and rails gradually wear and change their profiles in service; therefore, their contact geometry properties can never keep constant. Three of these performance indices, car lateral stability, wheel/rail contact position, and stress, are significantly affected by wheel/rail wear. The effects of wear on these criteria are mostly negative, leading to car instability, deteriorated ride quality, and damage to wheels and rails. 3.3 Wheel/Rail Contact Conicity Conical wheels have a conical taper on tread, as Figure12 shows. The taper is to promote self-centering in tangent track and generates some degree of steering in shallow curves. Figure 12. Wheel and Rail Contact Geometry 20

TCRP WOD 65 Part 1 The kinematical properties of wheel and rail contact, such as rolling radius, contact angles, and wheelset roll angle vary as the wheelset moves laterally relative to the rails. The nature of the functional dependence between these geometrically constrained variables and the wheelset lateral position is defined by the wheel and rail profiles. An important characteristic of the contact between wheels and rails is the rolling radius of a wheel at the contact point. This radius can be different for the right and the left wheel as a wheelset moves laterally. The RRD results in relative creep movement between wheel and rail on the contact points. The forces generated from the contact geometry constraints and wheel/rail friction can not only steer the wheelset to move along the track center, but also lead to hunting. The wheel/rail contact geometry has significant effects on vehicle dynamic performance including curving and hunting. An important parameter to characterize the wheel/rail contact geometry is equivalent conicity (or effective conicity). In general, the effective conicity is defined by Equation 1 (IHHA 2001): 𝜆𝜆 = 𝑅𝑅𝑅𝑅𝑅𝑅 2𝑦𝑦0 (1) where y0 is the wheelset lateral shift. In 2007, APTA drafted a standard to define a stability taper for the measurement of wheel tread taper on wheels used in commuter rail service in relation to vehicle and truck stability (APTA SS-M-017-06 2007). The stability taper is similar to the traditional tread taper, but calculated with a contact location weighting function that uses a normal distribution centered on the mean value of wheelset lateral shift with a standard deviation. Critical hunting speed is inversely proportional to square root of conicity; the higher the conicity the lower the critical speed. A high equivalent conicity can lead to wheelset and truck hunting, whereas a very low conicity can lead to combined oscillation of vehicle body and truck due to a resonance between the truck weaving movement and an eigenmode (natural frequency) of the vehicle body (Smith and Kalousek 1991, Polach 2009). The equivalent conicity defined in Equation 1 is half of the slope of the linearized wheel RRD function in the range of wheel lateral shift before reaching flange contact. For a new wheel with constant taper on the tread, the RRD function is almost linear in the range of wheel/rail clearance, as Figure 13 shows. The equivalent conicity for a new tapered wheel and rail is proportional to the tread taper slope. For a worn wheel, the RRD function is nonlinear, and the equivalent conicity defined in Equation 1 is negative, as Figure 14 shows. The negative equivalent conicity reveals the limit of the definition in Equation 1. 21

TCRP WOD 65 Part 1 Figure 13. Contact of New AAR-1B Narrow Flange Wheel on New AREMA 136-RE Rail, 10-inch Crown Radius, 1:40 Cant, at a Gage of 56.5 inches Figure 14. Contact of Hollow Worn Wheel on Tangent Worn Track at a Gage of 56.5 inches 22

TCRP WOD 65 Part 1 There are several other definitions and methods for equivalent conicity calculation. Among them the following methods are frequently used to calculate the equivalent conicity: Equivalent linearization by the application of Klingel formula defined in International Union of Railways UIC 519 standard 2004 Linear regression of the function of RRD defined in UIC 519 standard 2004 Harmonic quasi-linearization (Polach 2009) These three methods produce a nonlinear equivalent conicity function over the range of wheelset lateral displacement that is different from the single value conicity defined in Equation 1. Polach 2009 investigated the effects of nonlinearity of the equivalent conicity function on car instability performance and proposed two parameters to characterize the car hunting behavior: (1) the equivalent conicity at the specified wheelset lateral shift, and (2) its slope on the equivalent conicity function. His study concluded that: A wheel/rail contact with low equivalent conicity and positive slope of the equivalent conicity function usually leads to a sudden occurrence of a limit cycle wheelset movement with large lateral shift amplitude at a critical speed. The critical speed is usually referred to as hunting speed. A wheel/rail contact with high equivalent conicity and negative slope of the equivalent conicity function usually leads to a limit cycle wheelset movement with an amplitude slowly growing with increasing speed. The limit cycle usually occurs at speeds far below the traditional hunting speed. UIC 518 standard 2005 recommends that wheel/rail contact equivalent conicity values for all axles in a car should be distributed so that the equivalent conicity value 0.2±0.05 occurs in a range of wheelset lateral displacement between ±2mm and ±4mm for the majority of assessed conditions. Because the wheel/rail clearance in Europe is usually less than that in North America, the application of UIC 518 and UIC 519 standards on rail transit vehicle performance assessment needs to be further investigated. 3.4 Correlation between Wheel Wear and Equivalent Conicity Wheel profiles, regardless of tapered or cylindrical wheel, change from new shapes to worn shapes during service. Even a new wheel profile that is based on worn shapes often changes its tread shape due to tread wear. Wheel/rail wear results in wheel/rail contact geometry change and consequently changes car dynamic performances including curving, hunting, and ride quality. Guidelines or standards for wheel wear limits or wheel diameter mismatching tolerance could be generated from the nonlinear equivalent conicity function to improve car dynamic performances and maintenance efficiency. A nonlinear equivalent conicity function is a promising index to characterize variation of wheel/rail contact geometry caused by wheel wear or mismatching after truing. However, the correlation between the wheel wear or wheel diameter mismatching and the nonlinear equivalent conicity function is not fully understood. Further, wheel/rail contact equivalent conicity functions in one truck or in a railcar could be different from axle-to-axle, as Figure 15 shows. The effects of multiple-axle equivalent conicity functions and truck suspensions on railcar performances are unclear. This is an area that requires additional research. 23

TCRP WOD 65 Part 1 Figure 15. RRD Functions Measured in a Freight Railcar 24

TCRP WOD 65 Part 1 C H A P T E R 4 Conclusions and Recommendations The following conclusions and recommendations are made from the survey of wheel profile maintenance practices in rail transit agencies and the literature review: • Wheel slide and wheel flats are mainly caused by braking and low adhesion conditions. New anti- slip technologies and devices are needed to reduce wheel flats. • Wheel diameter difference on one axle has a significant effect on car lateral stability performance. Allowable wheel diameter difference maintenance limit depends on vehicle and truck design, especially the truck suspension and the maintenance limits of other components. • Wheel diameter difference in one truck affects car vertical performance such as the wheel load equalization capability. • Most rail transit agencies surveyed do not have wheel tread wear limits. Wheel wear has significant effects on vehicle and track performance. Setting up wear limits on wheels is a complicated issue. It depends on vehicle and track design, the maintenance standards of truck components, and operational environment. • A new wheel design or truing template should be optimized on the basis of existing rail wear conditions, vehicle design and maintenance standards, and special trackwork maintenance requirements. • Wheel truing template profiles need to be evaluated periodically to take into account existing rail wear conditions. • Rough surfaces on wheels from wheel truing can increase the risk of flange climb derailment. Smooth surfaces and lubrication could reduce the flange climb derailment risk. • The effect of the following maintenance limits on railcar performance will be further investigated in Task 2 of this project: – Wheel diameter differences on one axle, one truck and one car – Wheel wear and patterns – Multiple-axle wheel wear and patterns – Car type and suspension parameters – The nonlinear equivalent conicity function is a promising index to characterize variation of wheel/rail contact geometry caused by wheel wear or mismatching after truing. However, the correlation between the wheel wear or mismatched wheel diameter and the nonlinear equivalent conicity function has not been fully established. – The application of equivalent conicity defined in UIC 518 and UIC 519 standards to North American rail transit car performance assessment needs to be further investigated. Guidelines for wheel profile maintenance will be established through Task 2 of this project. 25

TCRP WOD 65 Part 1 References American Public Transit Association. APTA SS-M-014-06, Standard for Wheel Load Equalization of Passenger Railroad Rolling Stock, 2007. American Public Transit Association. APTA SS-M-015-06, Standard for Wheel Flange Angle for Passenger Equipment, 2007. American Public Transit Association. APTA SS-M-017-06, Standard for Definition and Measurement of Wheel Tread Taper, 2007 Association of American Railroads. Field Manual of the AAR Interchange Rules, Washington, D.C., 2012 Cabrera, A. and G. Gobbato. Redesign of Frog Geometry at New York City Transit to Reduce Vibration, Noise and Accelerated Wear, Proceedings of AREMA Conference, September 2000. Griffin, T. TCRP Report 114: Center Truck Performance on Low-Floor Light Rail Vehicles. Transportation Research Board of the National Academies, Washington, D.C., 2006. International Heavy Haul Association. Guidelines to Best Practices for Heavy Haul Railway Operations: Wheel and Rail Interface Issues. Virginia Beach, Virginia, May 2001. Kerchof, B. Interaction of Tread-hollow Wheel and Worn Switch Point / Stock Rail, Proceedings of AREMA Conference, 2004. Kumar, S. (Tranergy Corp.) TCRP Research Results Digest 17: Improved Methods for Increasing Wheel/Rail Adhesion in the Presence of Natural Contaminants. TRB, National Research Council, Washington, D.C., 1997. Leary, J.F., S.N. Handal, and B. Rajkumar. Development of Freight Car Wheel Profiles – A Case Study. Vol. 144. Wear, 1991. Magel, E. and J. Kalousek. Martensite and Contact Fatigue Initiated Wheel Defects, Proceedings 12th International Wheelset Congress, Qingdao, China, September 1998, pp. 110-111. Nelson, J. and T. Wilson (Ihrig, & Associates, Inc.). TCRP Report 23: Wheel/Rail Noise Control Manual. TRB, National Research Council, Washington, D.C., 1997. New Jersey Transit press release. NJ Transit Unveils Aqua Track to Prevent Wheel-Slip Conditions, http://www.njtransit.com/tm/tm_servlet.srv?hdnPageAction=PressReleaseTo&PRESS_RELEASE_I D=721, Accessed December 2003. Novales, M., A. Orro, M.R. Bugarín. A New Approach for the Design of Wheel Profile Geometries. Proceedings of 7th World Congress on Railway Research, Montreal, Canada, 2006. Parsons Brinckerhoff, Inc. TCRP Report 155: Track Design Handbook for Light Rail Transit, Second Edition, Transportation Research Board of the National Academies, Washington, D.C., 2012. Persson, I. and S.D. Iwnicki, Optimisation of Railway Profiles using a Genetic Algorithm. Vehicle System Dynamics, Supplement to Vol. 41, 2004. Polach, O. Wheel Profile Design for the Target Conicity and Wide Contact Spreading. Proceedings of the 8th International Conference on Contact Mechanics and Wear of Wheel/Rail System, Italy, 2009. Sawley, K., C. Urban, and R. Walker. The Effect of Hollow-Worn Wheels on Vehicle Stability in Straight Track, Wear, Vol. 258, Issues 7–8, 2005. Shen, G., J.B. Ayasse, H. Chollet, and I. Pratt. A Unique Design Method for Wheel Profiles by Considering the Contact Angle Function. Proc. Instn. Mechanical Engineers, Part F: J. Rail and Rapid Transit, Vol. 217, 2003. Shevtsov., I.Y., V.L. Markine, C. Esveld. Design of Railway Wheel Profile Taking into Account Rolling Contact Fatigue and Wear, Wear, 2005. 26

TCRP WOD 65 Part 1 Shu, X. and J. Tunna. Investigation into Flange Climbing Derailment and Distance Criterion with Wheel and Worn Rail Profiles. Research Report R-982, Association of American Railroads, Transportation Technology Center, Inc., Pueblo, CO, 2007. Shu, X. and N. Wilson. TCRP Report 71 : Track-Related Research, Volume 7 : Guidelines for Guard/Restraining Rail Installation. Transportation Research Board of the National Academies, Washington, D.C., 2010. Smith, R.E. and J. Kalousek. A Design Methodology for Wheel and Rail Profiles for Use on Steered Railway Vehicles, Wear, Vol. 144, 1991. Tournay, H. Investigation of Vehicle/Track and Bogie Parameters Leading to Loaded Wagon Lateral Instability. Proceeding of International Heavy Haul Association Conference, Shanghai, China, 2009]. Tunna, J., X. Shu, and J. Dasher. Investigation into the Effects of Geometric Tolerances on Freight Car Dynamics, Proceedings of ASME International Mechanical Engineering Congress, November 2006. UIC Leaflet 519, Method for Determining the Equivalent Conicity, 2004. UIC Leaflet 518, Testing and Approval of Railway Vehicles from the Point of View of Their Dynamic Behavior – Safety – Track Fatigue – Ride Quality, 2005. Wolf, G. Switch Point Derailments: Is It the Point or the Wheel, Journal of Wheel/Rail Interaction, 2006. Wu, H., X. Shu, and N. Wilson. TCRP Report 71 : Track-Related Research, Volume 5 : Flange Climb Derailment Criteria and Wheel/Rail Profile Management and Maintenance Guidelines for Transit Operations. Transportation Research Board of the National Academies, Washington, D.C., 2005. Wu, H. Control of Wheel/Rail Wear and Rolling Contact Fatigue. Proceedings of International Heavy Haul Association Conference, Kiruna, Sweden, 2007. 27

Wheel Profile Maintenance Guidelines PART 2 Wheel Profiles Design and Maintenance Guidelines for Rail Transit Operation

Table of Contents TABLE OF CONTENTS ............................................................................................................................... i LIST OF FIGURES ...................................................................................................................................... ii LIST OF TABLES ....................................................................................................................................... iv SUMMARY .................................................................................................................................................. 1 CHAPTER 1 Introduction ............................................................................................................................ 3 CHAPTER 2 Development of Wheel Profile Design and Maintenance Guidelines for Rail Transit Operations ....................................................................................................................... 4 2.1 Wheel Profile Design Considerations .................................................................................... 4 2.2 Guidelines for Improving Curving Performance ................................................................... 5 2.3 Guidelines for Improving Hunting Performance ................................................................. 10 2.3.1 Hunting and Safety Criteria .................................................................................... 10 2.3.2 Development of a W/R Contact Geometry Based Hunting Criterion ..................... 14 2.3.3 New Design Wheel Hunting Performance Evaluation ............................................ 20 2.4 Guidelines for Contact Stress and Wear Performance ......................................................... 22 2.5 Guidelines for Compatibility with Special Trackwork ........................................................ 26 2.5.1 Turnouts .................................................................................................................. 26 2.5.2 Switch Point Protectors ........................................................................................... 28 2.5.3 Spring Switches ....................................................................................................... 30 2.6 Wheel Profile Maintenance Guidelines................................................................................ 31 2.6.1 Wheel Diameter Difference .................................................................................... 31 2.6.2 Wheel Wear............................................................................................................. 36 CHAPTER 3 Conclusions and Recommendations .................................................................................... 44 REFERENCES ........................................................................................................................................... 46 APPENDIX Light Railcar Hunting Speed Contour Chart ....................................................................... 47 i

List of Figures Figure 1. Measured Wheel Profiles .................................................................................................. 6 Figure 2. Existing and New Design Wheel Profiles ........................................................................ 6 Figure 3. L/V Ratio, Existing 63-degree Wheel on New Rail, No Track Perturbations ................. 8 Figure 4. L/V Ratio, New Design 70-degree Wheel on New Rail, No Track Perturbations .......... 8 Figure 5. L/V Ratio, Existing 63-degree Wheel on New No. 8 Turnout, Down-and-out Track Perturbations ............................................................................................................. 9 Figure 6. L/V Ratio, New Design 70-degree Wheel on New No. 8 Turnout, Down-and-out Track Perturbations ............................................................................................................. 9 Figure 7. A Constrained Single Axle with Conical Wheel ............................................................ 10 Figure 8. Effect of Conical Wheel Conicity on Hunting Speed ..................................................... 11 Figure 9. Relationship between Hunting Speed and Root Square of Conicity .............................. 11 Figure 10. Time History Comparisons of Conical Wheel Hunting (65 mph) and No Hunting (64 mph) ....................................................................................................... 12 Figure 11. Limit Cycle Movements of Conical Wheel Hunting (65 mph) and No Hunting (64 mph) ........................................................................................................ 13 Figure 12. Two-line Wheels Contact on AREMA 115RE Rails Used for Hunting Simulations ......................................................................................................... 14 Figure 13. Subcritical and Supercritical Hunting Cases ................................................................ 15 Figure 14. Ride Quality of Subcritical and Supercritical Hunting ................................................. 16 Figure 15. Truck Frame Accelerations at Different Speeds (45-degree flange angle wheel) ........ 17 Figure 16. Truck Frame Accelerations at Different Speeds (63-degree flange angle wheel) ........ 17 Figure 17. Truck Frame Accelerations at Different Speeds (70-degree flange angle wheel) ........ 18 Figure 18. W/R Equivalent Conicity and Clearance Effects on Hunting Speed ............................ 19 Figure 19. Hunting Speed Contour Chart for Representative Heavy Railcar ................................ 19 Figure 20. Existing and New Design Wheel Conicity (56.25-inch gage) ...................................... 20 Figure 21. Existing and New Design Wheel Conicity (56.25-inch gage) ...................................... 21 Figure 22. Truck Frame Accelerations of the Existing and New Design Wheel (56.25-inch gage) .............................................................................................................. 21 Figure 23. Measured Rail Profiles in Curve .................................................................................. 23 Figure 24. Contact Stress of the Existing and New Design Wheels on New Rails ....................... 23 Figure 25. Contact Stress of the Existing and New Design Wheels on Worn Rails ...................... 24 Figure 26. Track Curvature Distribution ........................................................................................ 24 ii

Figure 27. Comparison of Wear Index for the Existing and New Design Wheel .......................... 25 Figure 28. Cylindrical Wheel Contact on a Worn Switch ............................................................. 26 Figure 29. Tapered Wheel Contact on a Worn Switch .................................................................. 26 Figure 30. A Worn Wheel Contact on a Worn Switch Point and Stock Rail ................................. 27 Figure 31. A Taper Wheel and Cylindrical Wheel Contact on a Frog ........................................... 28 Figure 32. Switch Point Protector .................................................................................................. 28 Figure 33. A Larger Chamfer Wheel Contacts on a New Switch Point Guard .............................. 29 Figure 34. A Smaller Chamfer Wheel Contacts on a New Switch Point Guard ............................ 29 Figure 35. Spring Switch in a Light Rail Transit System .............................................................. 30 Figure 36. W/R Contact on Spring Switch Points.......................................................................... 31 Figure 37. Wheel Diameter Difference (in the Same Axle) Effect on Hunting (56.25-inch gage) .............................................................................................................. 32 Figure 38. Wheel Diameter Difference Effect on Wear Index ...................................................... 33 Figure 39. Wheel Diameter Difference Effect on Rolling Resistance ........................................... 33 Figure 40. Wheel Diameter Difference Effect on Wheel L/V Ratio .............................................. 34 Figure 41. Wheel Diameter Difference Effect on Wheel Lateral Force ........................................ 34 Figure 42. Wheel Diameter Difference Effect on Wheel Unload .................................................. 35 Figure 43. Measured New and Worn Wheel Conicity, Track Gage 56.25 inches ........................ 36 Figure 44. Hunting Speed Estimation for a Car Equipped with Measured New and Worn Wheels .............................................................................................................. 37 Figure 45. Truck Frame Accelerations of a Car Equipped with Measured New and Worn Wheels (56.25-Inch Gage) ...................................................................................... 38 Figure 46. Measured New and Worn Wheel Conicity, Track Gage 56.5 inches ........................... 38 Figure 47. Measured New and Worn Wheel Conicity, Track Gage 57 inches .............................. 39 Figure 48. Wheel Wear Effect on Hunting .................................................................................... 39 Figure 49. Effect of Wheel Wear on Ride Quality......................................................................... 40 Figure 50. No. 20 Turnout Frog Profiles ....................................................................................... 41 Figure 51. W/R Impact Loads on a New Frog ............................................................................... 41 Figure 52. New and Worn Wheel Wear Index in Curves with New Rails .................................... 42 Figure 53. New and Worn Wheel Lateral Forces in Curves with New Rails ................................ 43 iii

List of Tables Table 1. Case Study Parameters ..................................................................................................... 22 iv

TCRP WOD 65 Part 2 Summary Transportation Technology Center, Inc., (TTCI), a wholly owned subsidiary of the Association of American Railroads (AAR), was contracted by the Transit Cooperative Research Program (TCRP D-07, Task Order 20) to develop wheel profile maintenance guidelines for rail transit operations. The effects of wheel profile and wheel/rail (W/R) interaction on car dynamic performances, including lateral stability, curving, W/R contact stress, wear and performances in special trackwork, have been investigated using NUCARS®1 simulations. The following guidelines were developed from this study: • Both W/R contact conicity and W/R gage clearance have significant and complex effects on car lateral stability (hunting), especially as the wheels and rails wear. A new method for evaluating the combined effects of these two parameters was developed using Hunting Speed Contour (HSC) charts. To demonstrate the new method, two HSC charts were developed for a representative heavy railcar and a representative light railcar. The charts were used to generate the following guidelines for wheel profile design and maintenance: – Hunting speed generally decreases (car becomes more unstable) with the increase of both conicity and W/R clearance. – Wheel and rail wear increases W/R clearance, and its effect on hunting depends on wear pattern:  Worn wheels with high conicity and wide W/R clearances cause the hunting speed to decrease quickly  Worn wheels with low conicity and wide W/R clearances cause the hunting speed to decrease slowly, but may result in sudden onset of hunting instability – Hunting speeds for high conicity wheels are more sensitive to W/R clearance variations than for low conicity wheels. – HSC charts may be used to evaluate new wheel profile designs and also to evaluate worn wheels (and rails), and develop wear and gage clearance tolerances. To provide specific conicity and gage clearance guidelines for a particular vehicle in a rail transit system, a new HSC chart would need to be developed using simulations for the particular case. • Increasing the maximum flange angle can effectively reduce flange climb derailment risk. American Public Transportation Association (APTA) recommends a 72-degree (with tolerance +3 degrees and -2 degrees) flange angle wheel for use in commuter railcars. However, a wheel profile with a flange angle less than 72 degrees (but high enough to prevent flange climb) can also be adopted to provide a smooth transition from an existing low flange angle wheel profile to a new design with high flange angle wheel profile. • Wheel profiles (new and worn) should be compatible with special trackworks: – Impact forces on the switch frog nose generally increase with wheel wear. – Wheels with profiles that are incompatible with the frog generate significant impact on the switch frog. – High flange angle wheels can reduce flange climb derailment risk and reduce excessive switch point tip wear in spring switches. • Wheel diameter differences on an axle can improve hunting performance because of the decrease of W/R gage clearance when the axle shifts (moves laterally) from the track center position toward the smaller radius wheel. However, wheel diameter differences may result in poor curving performances, such as more wear and larger lateral forces on high rails, which may cause gage spreading. 1 NUCARS® is a registered trademark of Transportation Technology Center, Inc. 1

TCRP WOD 65 Part 2 – Systems with many curves may need tighter tolerances on wheel diameter differences than systems with few curves and mostly straight track. • Wheel wear has significant effects on both hunting speed and switch frog impact. Wear limits on wheel treads and flanges can be determined by the HSC chart and impacts with switch frogs. On-track tests are recommended to further validate these guidelines. 2

TCRP WOD 65 Part 2 C H A P T E R 1 Introduction The Transportation Technology Center, Inc. (TTCI) is developing “Wheel Profile Maintenance Guidelines” under Project D-7, Task 20, for the Transit Cooperative Research Program (TCRP) to help transit systems minimize maintenance costs and maximize system performance. The objectives of this study are to: • Investigate the effects of wheel profiles (include both new and worn profiles) on rail transit vehicle performance (safety and ride quality) • Develop wheel profile maintenance guidelines for rail transit operations (light rail and heavy rail systems) • Develop guideline implementation procedures demonstrated with examples The tasks of this project include: • Task 1 – Survey current wheel profiles and maintenance practices • Task 2 – Develop wheel profiles maintenance guidelines • Task 3 – Demonstrate wheel maintenance guideline implementation procedures This report presents the development of wheel profile maintenance guidelines for rail transit operations. The optimum wheel and rail profiles are to provide: • Stable performance over the range of normal train speeds • Safety from derailment under adverse but realistic operating conditions • Maximized wheel and rail life Wheel and car performances are generally evaluated in the following two aspects for safety operations: • Hunting (lateral stability) performance and ride quality • Curving performance In addition, the following aspects should be addressed for wheel profile design and optimization: • W/R contact stress and wear • Compatibility with special trackwork, including frog, switch, and switch point protector Requirements from these aspects often conflict with each other. Wheel profile design and maintenance guidelines have been developed in this report to compromise these different requirements for an overall optimum solution. 3

TCRP WOD 65 Part 2 C H A P T E R 2 Development of Wheel Profile Design and Maintenance Guidelines for Rail Transit Operation Wheel profile shape has significant effects on rail vehicle and track performances. Wheel and rail profile optimizations have been investigated intensively to improve car performances with the increase of car running speed and axle load, as described in the Task 1 literature review report for this study (Shu 2014). Different optimization approaches and concepts have been developed by researchers for specific or general objectives. Various guidelines for wheel profile design and maintenance have been used or proposed for railroad and rail transit systems. Some guidelines have been commonly accepted by the industry, such as use of high flange angle wheels to reduce flange climb derailment risk and low tread conicity wheels to prevent hunting. However, some guidelines remain controversial, especially for hunting performance. A high equivalent conicity can lead to wheelset and truck hunting, whereas a very low conicity can lead to combined oscillation of the vehicle body and trucks due to a resonance between the truck waving movement and an eigenmode (natural frequency) of the vehicle body (Polach 1009, Smith and Kalousek 1991). Dynamic simulation programs provide useful tools to improve vehicle and track design. Car curving performance predictions are more consistent with tests than that of hunting. Curving simulation results usually follow certain trends, while uncertainties exist among hunting simulation results because of the hunting sensitivity to nonlinear parameters in a vehicle and track system. This uncertainty is so strong that from an academic point of view, a researcher has asked: “Does a critical speed for railroad vehicles exist?” (True 1994) However, car hunting is such a persistent dynamic behavior that railroads worldwide have adopted safety standards to prevent hunting. This study intends to re-examine the hunting related criteria and to provide a W/R contact geometry based hunting criterion for wheel profile design and maintenance. 2.1 Wheel Profile Design Considerations Wheel profile shape is one of the fundamental aspects for new transit system design and existing system maintenance. Changing wheel profile shape is a common practice for a rail transit system to resolve vehicle and track dynamic performance issues. Derailment is the leading reason for changing wheel profiles especially for a system with low flange angle wheels. Other reasons may include poor ride quality, corrugation, and excessive wheel and rail wear. The objective of changing an existing wheel profile is to not only fix specific problems but also to improve overall vehicle and track dynamic performances. Wheel maximum flange angle has a significant effect on flange climb derailment. The derailment risk decreases with the increase of flange angle. The experiences of Massachusetts Bay Transportation 4

TCRP WOD 65 Part 2 Authority’s Type 8 car demonstrated that derailment accidents were significantly reduced by increasing wheel maximum flange angle from 63 to 75 degrees (Griffin 2006). In addition to the maximum wheel flange angle, wheel tread taper is another parameter for wheel profile design and maintenance. The hunting speed (speed at which a railcar bogie or bogies begin to hunt) generally increases with the decrease of equivalent conicity. The equivalent conicity generated from the modified wheel profile and existing rail profile has to be designed to work with the car suspension system to ensure the hunting speed is higher than the current operation speed. Once the maximum flange angle and tread taper are determined, the flange root that connects the wheel tread and flange could vary with different shapes. The flange root shape has significant effects on both curving (especially shallow curves) and hunting performances. Compromise on these two aspects (curving and hunting) is needed for optimal performance. A rule of thumb for flange root design is to start from the existing worn flange root shape and further modify it to optimize the curving and hunting performances. A worn flange root shape is usually conformal to the majority of existing worn rails and generates low contact stresses and less wear. Other parts of a wheel that have potential contact with rails, such as the wheel flange back, tread end near the field side, and the wheel face, also have to be compatible with special trackwork, such as switches, frogs, guardrails, and switch point protectors. Incompatible wheel shapes and contacts with special trackworks not only generate high W/R impact forces and excessive wear on wheels and rails, but also could result in derailment. The following subsections demonstrate with examples how all these aspects can be addressed comprehensively, and how new guidelines were developed for new wheel profile design and worn wheel maintenance. 2.2 Guidelines for Improving Curving Performance Wheel wear on a representative heavy rail transit system was investigated, and an improved wheel profile was designed. The existing new wheels on this system are designed with a flange angle of 63 degrees. Rail transit systems using low flange angle (60 to 63 degrees) wheels often experience flange climb derailments on yard track and small number turnouts. The desirable wheel flange angle is above 72 degrees because a higher flange angle gives a higher wheel lateral force to vertical force (L/V) ratio limit required for wheel climb based on the Nadal criterion (Shu 2014). Measurements showed the wheels and rails in the investigated rail transit system wore into a relatively consistent flange/gage face angle about 70 degrees, as Figure 1 shows. Adopting the existing 70-degree worn flange angle will provide a higher wheel L/V ratio limit to reduce the risk of wheel flange climb and achieve a smooth transition from existing low flange angle wheels. 5

TCRP WOD 65 Part 2 Figure 1. Measured Wheel Profiles If wheels with a flange angle greater than 70 degrees are directly introduced, such as the 75-degree flange angle wheel that has been commonly adopted for many newly built transit systems, the resulting incompatible contact pattern with the existing worn rails will require aggressive wheel truing and rail grinding. The new improved design wheel profile thus uses a 70-degree flange angle and similar flange root shape of the worn wheels, as Figure 2 shows. Figure 2. Existing and New Design Wheel Profiles Computer simulations were conducted to evaluate the performance of the improved wheel profile in comparison to the existing wheel profiles. A typical heavy rail vehicle model with the following specifications was used in this study: • Cylindrical bushing primary suspension • Four-point air spring leveling secondary suspension system • Lateral and vertical damper in secondary suspension • Articulated truck frame • Axle spacing: 7.5 feet 6

TCRP WOD 65 Part 2 • Truck Center Spacing: 52 feet • Wheel load: 9.45 kips • Wheel diameter: 27 inches Vehicle model parameters were measured through characterization tests. The measured primary suspension longitudinal and lateral stiffness and damping were reduced by half to simulate a worn truck condition. Measured track geometries, the standard American Railway Engineering and Maintenance of Way Association (AREMA) 115RE rail profile, and 0.5 W/R friction coefficient representing dry W/R contact conditions were used in the simulations. A number of yard track geometries, all without restraining rails, were included in the simulations to compare the curving performances of the current 63-degree flange angle new wheel and the improved new design 70-degree flange angle wheel. These were: • A 755-foot radius curve with 1-inch superelevation, 56.5-inch gage and 100-foot spirals. • A 500-foot radius curve with 1-inch superelevation, 56.5-inch gage and 100-foot spirals. • A 320-foot radius curve with 1-inch superelevation, 57.0-inch gage and 100-foot spirals. • A 250-foot radius curve with 1-inch superelevation, 57.0-inch gage and 100-foot spirals. • A No. 8 turnout with no superelevation and 56.5-inch gage. The track geometries were represented as design case smooth track geometry with no track irregularities for yard curves, and a “Down-and-Out” perturbation for the No.8 turnout. The Down-and- Out perturbation consisted of a combination of track geometry irregularities that were of a magnitude at the limit stated in the transit system’s track standard. This consisted of a downward vertical cusp of 1.25- inch amplitude on the high rail combined with a 2-inch outward lateral alignment cusp on the high rail and an inward cusp on the low rail of a magnitude sufficient to ensure that the maximum permitted gage was not exceeded. These irregularities were of 31-foot wavelength with a haversine (1-cosine) shape. In all cases, the vehicle speed was 15 mph, which is the speed limit for yard track. The W/R friction coefficient is also a factor that has a large effect on the potential for derailment. Therefore, all simulation cases were carried out for W/R friction coefficients of 0.3, 0.4, 0.5, and 0.6. All of the simulation cases were modeled using TTCI’s NUCARS vehicle dynamics simulation program. The results are presented as graphs of the maximum wheel L/V ratio as a function of W/R friction coefficient. The maximum W/R L/V ratio was evaluated by using FRA’s Code of Federal Regulations, Track Safety Standards, Part 213, Subpart G, Section 213.333” (FRA March 2013). Also plotted on the graphs is a line representing the Nadal limit for the particular W/R profile combination. Where the simulation results closely approach or cross this line indicates the potential for a flange climb derailment. Figure 3 show the results for the cars equipped with the existing 63-degree wheel profiles running on new rails. The maximum L/V ratios on curves with 250- and 320-feet radii were close to the limit at 0.5- friction coefficient; the car derailed at 0.6-friction coefficient conditions even without any track perturbation. Figure 4 shows the significant improvement on curving for the car equipped with the 70- degree flange angle wheel; the maximum L/V ratios on curves with 250- and 320-feet radii were close to the limit at 0.6, but no derailments occurred. 7

TCRP WOD 65 Part 2 Figure 3. L/V Ratio, Existing 63-degree Wheel on New Rail, No Track Perturbations Figure 4. L/V Ratio, New Design 70-degree Wheel on New Rail, No Track Perturbations 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.3 0.4 0.5 0.6 Le ad A xl e H ig h R ai l W he el L /V Friction Coefficient New Design Wheel (70 Deg) NADAL R=755 ft R=500 ft R=250 ft R=320 ft 8

TCRP WOD 65 Part 2 Figure 5 shows the car with the 63-degree flange wheel exceeding the L/V ratio limit at 0.5-friction coefficient, derailing at 0.5-friction coefficient on a No. 8 turnout. Figure 6 shows that even though the maximum L/V ratio of the 70-degree flange angle wheel exceeded the limit at 0.6-friction coefficient, no derailment occurred on the No.8 turnout. Figure 5. L/V Ratio, Existing 63-degree Wheel on New No. 8 Turnout, Down-and-out Track Perturbations Figure 6. L/V Ratio, New Design 70-degree Wheel on New No. 8 Turnout, Down-and-out Track Perturbations 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.3 0.4 0.5 0.6 Le ad A xl e H ig h R ai l W he el L /V Friction Coefficient NADAL 9

TCRP WOD 65 Part 2 Increasing the maximum flange angle can effectively reduce flange climb derailment risk. APTA recommends a 72 degree (with tolerance +3 degrees and -2 degrees) flange angle for wheels used in passenger railcars. However, wheel profiles with flange angles less than 72 degrees (but high enough to prevent flange climb) can also be adopted for a smooth transition from low flange angle wheels. 2.3 Guidelines for Improving Hunting Performance 2.3.1 Hunting and Safety Criteria Lateral instability or hunting is an inherent characteristic of rail vehicles with solid axles and noncylindrical wheel profiles. The tapered wheel profile results in coupled axle lateral and yaw movements. With increasing speed, axle hunting, characterized by sustained flange-to-flange lateral motion, will occur as the energy from forward motion is transferred to the axle oscillation. Hunting has been a core topic of W/R interaction research. The critical speed is sensitive to not only the car suspension parameters but also the track perturbations in terms of both the shape and amplitude. A railcar’s suspension system deteriorates over time in service, and track geometries also degrade over time and vary along the track. These parameters change so dramatically in vehicle and track systems that the critical speed also changes from case-to-case. The maximum wheel flange angle is a simple geometry index for evaluating wheel profile effects on flange climb derailment. Similarly, researchers also hope to find a simple geometry index to evaluate wheel profile effects on hunting. So far, the most promising index appears to be the conicity. For a constrained single axle with a conical wheel without a flange, as Figure 7 shows, an analytical formula can be derived as Equation 1. NUCARS simulations using the transit car model with conical wheels without a flange validated that the hunting speed is inversely proportional to square root of conicity, as Figures 8 and 9 show. The parameter varied solely in the model is the conical wheel profile with different tread slopes (conicity). However, wheels must have flanges for safe operation. Once the wheel flange is introduced, the relationship between the hunting speed and conicity becomes much more complicated and cannot be expressed in a simple analytical equation. Figure 7. A Constrained Single Axle with Conical Wheel 10

TCRP WOD 65 Part 2 )mI( )k(kr V Speed Hunting 2 2 0zz y000 C   + + = λ y (1) Figure 8. Effect of Conical Wheel Conicity on Hunting Speed Figure 9. Relationship between Hunting Speed and Root Square of Conicity y = 31.213x - 0.1036 R² = 0.985 0 50 100 150 200 250 0 1 2 3 4 5 6 7 8 H un tin g Sp ee d (m ph ) Inverse of Root Square of Conicity 11

TCRP WOD 65 Part 2 The hunting criteria adopted by the railroad industry are different from that in academic research. The hunting speed in Figures 8 and 9 was defined as the speed at which the axle lateral motion starts to increase without damping after the excitation of track perturbations, as Figure 10 shows. Figure 11 shows the axle movement was stable at 64 mph (blue line, converged to track center position) but unstable (hunting) at 65 mph and reached a stable limit cycle over certain distances. Figure 10. Time History Comparisons of Conical Wheel Hunting (65 mph) and No Hunting (64 mph) 12

TCRP WOD 65 Part 2 Figure 11. Limit Cycle Movements of Conical Wheel Hunting (65 mph) and No Hunting (64 mph) From an academic point of view, the hunting speed was reached when the real part of the eigenvalue of a linear system changed to positive from negative, with the calculated hunting speed referred to as “linearized critical speed.” However, the vehicle and track system is strongly nonlinear. Therefore, the linearized critical speed calculated by eigenvalue analysis usually does not match test results very well. The “nonlinear critical speed” is usually calculated through parametric study by using dynamic simulations. The critical speed is defined as the speed at which the amplitude of axle lateral oscillation (coupled with axle yaw motion) starts to increase and may reach a stable limit cycle motion. The axle limit cycle motion is not necessarily involved with flange contact. It may be stable with or without flange contact, and it may also burst into an instable state and result in derailment. The critical speed must be judged case-by-case. The nonlinear critical speed simulation requires a large amount of work because there are many vehicle and track parameters influencing hunting and each parameter can vary in certain ranges. The lowest of critical speed from all these parametric variation cases was defined as the hunting speed (True 1994). The hunting speed in the railroad industry is defined differently by using truck frame accelerations instead of axle movements, such as defined in the FRA 49 CFR Parts 213 Vehicle/Track Interaction Safety Standard (FRA March 2013) and UIC Leaflet 518, Testing and Approval of Railway Vehicles from the Point of View of Their Dynamic Behavior – Safety – Track Fatigue – Running Behavior (UIC Leaflet 518 2009). The FRA 213 standard defines truck hunting as: “Truck hunting is defined as a sustained cyclic oscillation of the truck evidenced by lateral accelerations in excess of 0.3 g root mean square (mean-removed) for more than 2 seconds (FRA March 2013).” 13

TCRP WOD 65 Part 2 The industry uses an acceleration-based hunting speed criterion instead of an axle motion based criterion for the following reasons: • It is difficult to measure the dynamic axle displacements in the field. • The definition of hunting using axle motion is not clear; for some cases, the axle lateral motion bursts into hunting with hard flange contact, while for other cases, the axle lateral oscillation slowly grows into flange contact, while the truck and carbody accelerations may exceed safety limits before flange contact. • The truck frame acceleration hunting criterion is more conservative because it provides a safety warning before the axle oscillates in full scale with hard flange impact. • Truck frame acceleration is easy to measure and more sensitive to hunting movement. Earlier signs of hunting can be more easily captured by truck frame accelerations than axle lateral displacements, so an acceleration-based hunting criterion is more conservative than that based on axle lateral displacement. 2.3.2 Development of a W/R Contact Geometry Based Hunting Criterion New cars fleets are required to meet vehicle and track safety standards through on-track tests. Hunting performance is usually evaluated through truck frame and/or carbody acceleration measurement. However, for daily vehicle and track maintenance, rail transit agencies prefer a W/R contact geometry based hunting criterion and guidelines because wheel and rail geometry measurement is easier than that of acceleration and, for a given rail transit system, the vehicle parameters are generally fixed, while wheel and rail wear, and track geometry can vary. A W/R contact geometry based hunting criterion was developed through NUCARS simulations. Simple two-line wheel profiles were evaluated in the model to investigate the effects of flange angle, tread conicity, and track gage on hunting. The two-line wheel profile consists of two segments: tread and flange. The wheel tread and flange-segment profiles are two lines connected by a small arc at the flange root, as Figure 12 shows. Figure 12. Two-line Wheels Contact on AREMA 115RE Rails Used for Hunting Simulations The wheel tread conicity, flange angle, and track gage were varied in the parametric study: • Wheel flange angle: 45, 63 and 70 degrees • Tread conicity: 0.1, 0.2, 0.3 and 0.4 • Track gage: 56.1, 56.25, 56.5, and 57 inches The Task 1 report described several definitions and methods for equivalent conicity calculation (Shu 2014). The equivalent conicity defined in UIC 519 (UIC Leaflet 519 2004) was used in this study. Figure 13 shows two types of hunting responses: the subcritical and supercritical Hopf Bifurcations. The Hopf bifurcation theory and methodology have been used to investigate nonlinear dynamic system stability for decades (Strogatz 1994). At a critical state, a nonlinear dynamical system could lose stability 14

TCRP WOD 65 Part 2 in two different trends, namely supercritical and subcritical Hopf bifurcation. Hopf bifurcations occur when a railway vehicle runs above a critical speed (True 1994). An unstable vehicle with supercritical Hopf bifurcation generally has the following characteristics: • High conicity, usually with high flange angle or negative slope of equivalent conicity • Limit cycle of wheelset movements (sustained wheelset lateral displacement and yaw) with amplitudes growing with running speed • Limit cycle starts at low speed and may not exceed safety limit An unstable vehicle with subcritical Hopf bifurcation generally has the following characteristics: • Low conicity, usually with low flange angle and positive slope of equivalent conicity • Sudden occurrence of a limit cycle with large amplitudes that exceed safety limits • No limit cycle movements until a critical speed was reached Figure 13 shows the truck frame acceleration root mean square (rms) value for the car equipped with a 45-degree flange angle running on 56.25-inch gage track. The wheelset accelerations jumped to 1.4 g at 75 mph, demonstrating a typical subcritical Hopf bifurcation instability. However, the truck frame acceleration RMS value for the car equipped with 63-degree flange angle running on 57-inch gage track gradually increased with running speed, demonstrating a typical supercritical Hopf bifurcation instability. Based on a hunting defined as flange-to-flange contact axle oscillation, it is clear the hunting speed for the car with the 45-degree wheels is about 75 mph since the truck frame acceleration suddenly increased, indicating the wheel flange contact. However, the hunting speed is unclear for the car equipped with 63- degree wheels, because it is unclear at which speed the flange-to-flange contact axle oscillation occurred. Application of an acceleration-based hunting criterion, such as the 0.3 g rms FRA 213 track safety standard, is straight forward. For the simulated cases, the hunting speed for the car equipped with 63- degree wheel running on 57-inch gage track is about 66 mph, and the hunting speed for the car equipped with 45-degree wheel running on 56.25-inch gage track is about 71 mph. Figure 13. Subcritical and Supercritical Hunting Cases 15

TCRP WOD 65 Part 2 Figure 14 shows the ride quality of the simulated car equipped with 45-degree wheels is better than that with 63-degree wheels at speeds lower than 79 mph, which is consistent with their hunting performances (the hunting speed of the car equipped with 45-degree wheels is higher than that with 63-degree wheels). The examples in Figure 14 demonstrate that an acceleration-based hunting criterion is adequate for hunting performance evaluation. Figure 14. Ride Quality of Subcritical and Supercritical Hunting Figures 15 through 17 show the truck frame acceleration rms values at different speeds for the car equipped with 45-, 63- and 70-degree flange angles and different conicity wheels running on measured track geometry from Transportation Technology Center’s hunting test track (Railroad Test Track) with different track gages. Simulation speeds were from 50 mph to 120 mph to determine the hunting speed at which truck frame lateral acceleration values reach 0.3 g rms. Each graph shows results for several combinations of gage and conicity. For example, the line marked g561, 0.1 is for 56.1-inch track gage with 0.1-tread conicity. 0 0.2 0.4 0.6 0.8 1 1.2 55 60 65 70 75 80 85 90 95 100 Ca rb od y La te ra l A cc el er at io n W ei gh te d RM S Va lu e (m /s ^2 ) Speed (mph) Flange angle 45, gage 56.25, conicity 0.2 Flange angle 63, gage 57, conicity 0.2 ISO 2631, a ittle uncomfortable limit 16

TCRP WOD 65 Part 2 Figure 15. Truck Frame Accelerations at Different Speeds (45-degree flange angle wheel) Figure 16. Truck Frame Accelerations at Different Speeds (63-degree flange angle wheel) 17

TCRP WOD 65 Part 2 Figure 17. Truck Frame Accelerations at Different Speeds (70-degree flange angle wheel) Figures 15 through 17 appear to show that the highest hunting speed increases with flange angle. But the flange angle effects on hunting are different for subcritical (such as in Figure 15) and supercritical (such as in Figures 16 and 17) hunting. In comparison to the flange angle, the conicity and track gage effects on hunting are more consistent. The hunting speed decreased with the increase of conicity and track gage, regardless of subcritical or supercritical hunting responses. The question is how to quantify the relationship between the hunting speed and these varied parameters. The first step used in this study is to prioritize the parameters based on their effects, then combine multiple parameters into one parameter if possible using multiple parameter regression analysis. The fewer the parameters are, the easier the analyses is. Wheel flange angle and track gage variations result in different W/R clearances. The clearance was defined as the axle lateral shift from track center to the position where the wheel contacts the rail at maximum flange angle. Because the hunting speed was defined based on truck frame acceleration rms values, the W/R contact clearance was more relevant to the hunting speed than wheel flange angle. The wheel flange angle and track gage effects on hunting can be replaced by one parameter, the W/R clearance. In Figure 18, the simulation results were regrouped based on conicity. For each group with the same conicity, the hunting speed and the W/R clearance was fitted with an exponential function. The r-square quality of fit for each group is above 0.97. The fitting results were interpolated to obtain the same hunting speed for each conicity group. A group of hunting speed contour (HSC) functions, with two W/R contact geometry parameters, the conicity and W/R clearance, was developed by using the interpolation results, as Figure 19 shows. 18

TCRP WOD 65 Part 2 Figure 18. W/R Equivalent Conicity and Clearance Effects on Hunting Speed Figure 19. Hunting Speed Contour Chart for Representative Heavy Railcar 40 60 80 100 120 140 160 0 2 4 6 8 10 12 14 16 18 H un tin g Sp ee d (m ph ) W/R Clearance (mm) Simulation, Tread Conicity 0.1 Fitted, Tread Conicity 0.1 Simulation, Tread Conicity 0.2 Fitted, Tread Conicity 0.2 Simulation, Tread Conicity 0.3 Fitted, Tread Conicity 0.3 Simulation, Tread Conicity 0.4 Fitted, Tread Conicity 0.4 19

TCRP WOD 65 Part 2 The HSC chart in Figure 19 was developed based on a specific type of transit car model. It can be used to determine the hunting speed of this type of transit car with new and worn wheels. This chart provides the following guidelines: • The same hunting speed can be achieved by using either a high conicity with tight W/R clearance or a low conicity with wide W/R clearance. • The hunting speed generally decreases with the increase of conicity and W/R clearance. • Wheel and rail wear increase W/R clearance, so the worn wheel hunting speed decreases quickly as a wheel wear into high conicity and wide W/R clearance. • The worn wheel hunting speed decreases slowly when wheel wear results in low conicity and wide W/R clearance. • High conicity wheel hunting speeds are more sensitive to W/R clearance variation than that of low conicity wheels. 2.3.3 New Design Wheel Hunting Performance Evaluation The hunting performances of the existing and new design wheels were evaluated by using the HSC chart. Figure 20 shows the conicity of the existing 63-degree wheel and the new design 70-degree wheel with 56.25-inch gage, which was standard on tangent track in the transit system studied by TTCI. Figure 20. Existing and New Design Wheel Conicity (56.25-inch gage) The existing 63-degree wheel and AREMA 115RE rail generate 0.37 conicity (before contact at maximum flange angle) and 4.94-millimeter (0.19 inch) W/R clearance; the new design 70-degree wheel and AREMA 115 RE rail generate 0.22 conicity and 6.73-millimeter (0.26 inch) W/R clearance. The new design 70-degree wheel generates lower conicity but wider W/R clearance than the existing 63-degree wheel. By plotting these parameters in the HSC chart in Figure 21, the hunting speeds of these two 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0 1 2 3 4 5 6 7 8 9 10 Eq ui va le nt C on ic ity Axle Lateral Displacement (mm) Existing 63-Deg Flange Angle New Design 70-Deg Flange Angle 20

TCRP WOD 65 Part 2 wheels were estimated to be 70 mph and 65 mph for the new design 70-degree wheel and existing 63- degree wheel, respectively. Figure 21. Existing and New Design Wheel Conicity (56.25-inch gage) Based on the HSC chart, the new design 70-degree wheel will improve car lateral stability. This was validated through simulations using these two types of wheels, as Figure 22 shows. Figure 22. Truck Frame Accelerations of the Existing and New Design Wheel (56.25-inch gage) 21

TCRP WOD 65 Part 2 The application of the HSC chart includes the following steps: • Generate a HSC chart for a representative type of car: – Build the car model by using parameters measured through characterization test – Generate a series of simplified wheel profiles consisting of tread (with different tread slopes) and flange (with different flange angles) lines – Conduct W/R contact geometry analysis by using a simplified wheel profile and a representative rail profile to obtain W/R conicity and clearance – Conduct simulations using these simplified wheel profiles at different speeds with track gage variations – Conduct regression analysis between the predicted hunting speed and W/R contact geometry parameters in terms of conicity and W/R clearance – Interpolate the fitting results to obtain a group of HSC functions with W/R conicity and clearance parameters • Validate the HSC chart through simulation or test – Measure wheel and rail profiles – Conduct W/R contact geometry analysis on measured wheel and rail profiles to obtain conicity and clearance – Estimate the hunting speed from the HSC chart by using the measured W/R conicity and clearance – Conduct simulations by using wheel and rail profiles and track geometry, or conduct on-track test to measure hunting performances – Validate the HSC chart by comparing the estimated hunting speed to the hunting speed measured from tests or predicted from simulations. • Evaluate the car stability by using the HSC chart and W/R contact geometry parameters based on the guidelines described in subsection 2.3.2. 2.4 Guidelines for Contact Stress and Wear Performance Table 1 lists the curvatures, track superelevation, and running speed used in the wear analysis simulations. The speeds correspond to 1-inch cant deficiency for the curve. Table 1. Case Study Parameters Case Curve Radius (feet) Curvature (degree) Superelevation (inch) Speed (mph) Track Gage (inch) 1 250 22.92 0 7.87 57 2 500 11.46 0 11.12 56.5 3 750 7.64 4 30.66 56.5 4 817 7.01 4 32.00 56.5 5 955 6.00 4 34.60 56.5 6 1,145 5.00 4 37.88 56.5 7 1,430 4.01 4 42.34 56.5 8 1,910 3.00 4 48.93 56.25 9 2,864 2.00 4 59.91 56.25 10 5,729 1.00 1.5 59.83 56.25 New and worn rail profiles were used in the simulations. The worn rails were measured on curves, as Figure 23 shows. 22

TCRP WOD 65 Part 2 Figure 23. Measured Rail Profiles in Curve Figure 24 shows the new design 70-degree flange angle wheel running on a new 115RE rail generates lower contact stress than that of the existing 63-degree flange angle wheel for almost all simulated curves. The maximum contact stress on tight curve was lowered by about 31 percent. Figure 24. Contact Stress of the Existing and New Design Wheels on New Rails Figure 25 shows that the new design 70-degree flange angle wheel running on a worn 115RE rail generates lower contact stress than that of the existing 63-degree flange angle wheel on tight curves (<955-feet curve radius, 6-degree curve) and curves with radii larger than 2,600 feet (2.2-degree curve). The maximum contact stress on tight curve was lowered by about 42 percent. The design wheel generates higher contact stress than the existing wheel on 2- to 6-degree curves. However, these curves only account for 12 percent of the total track length, as Figure 26 shows. 23

TCRP WOD 65 Part 2 Figure 25. Contact Stress of the Existing and New Design Wheels on Worn Rails Figure 26. Track Curvature Distribution 0 10 20 30 40 50 60 70 <955 955 to 1433 1433 to 2860 >2860 Tangent Pe rc en t o f Tr ac k Le ng th (% ) Curve Radius (feet) 24

TCRP WOD 65 Part 2 The W/R wear was evaluated by using the wear index, defined in Equation 2. It is calculated as the sum of the tangential forces (Tx, Ty and Mz) multiplied by the creepages (γx, γy and ωz) at the contact patch. Rolling resistance is defined as the sum of wear indices on all wheels in a car. High rolling resistance can induce either RCF or high rates of wear. It is also an indicator of the energy consumption at the W/R interface. Figure 27 shows the wheel wear index of the wheel with 70-degree flange angle is slightly higher than that of the 63-degree wheel on curves with radii less than 750 feet, but a little lower on curves with radii from 750 to 1,430 feet. Sharp curves with radii smaller than 750 feet are generally in the yards in the system that was investigated. The yard tracks were designed to 4.5-inch underbalance speed operation. Because all simulations were conducted with 1-inch overbalance, the actual wheel and rail wear on sharp curves (radii less than 750 feet) is expected to be lower than the simulation results. Since the curve radii on the mainline are mostly above 750 feet, the rate of W/R wear of the proposed 70-degree wheel profile will be similar to the existing 63-degree wheel profile. Figure 27. Comparison of Wear Index for the Existing and New Design Wheel Using high flange angle wheels may increase wheel and rail wear, especially when the new design wheel flange angle is higher than the maximum flange angle on worn wheels and rails. Adopting new design wheel with a flange angle close to that of the worn wheel can smooth the transition by reducing the cost from wheel truing and rail grinding. Wear Index T T Mx x n y y z z= + +∑ γ γ ω (2) 25

TCRP WOD 65 Part 2 2.5 Guidelines for Compatibility with Special Trackwork 2.5.1 Turnouts 2.5.1.1 Switch Points Wheel profile changes have significant effects on existing special trackwork because the special trackwork has been either worn or adjusted into shapes compatible with existing wheels. To avoid derailments on worn switches, the new design wheel has to be checked against worn switches to make sure the new wheel profile will not increase flange climb derailment risk. Figure 28 shows an existing cylindrical wheel contact on a worn switch point; the contact angle on the switch point is about 52 degrees. Figure 29 shows an example candidate new design tapered wheel profile2 contacting on the same worn switch point; the contact angle is about 23 degrees and the lower contact angle cannot effectively resist wheel climbing in a small number switch. The existing cylindrical wheel maximum flange angle is 68 degrees, while the new design tapered wheel maximum flange is 66 degrees, so the candidate new design wheel flange climb derailment risk could be higher than the existing cylindrical wheel in the worst-case scenario. Figure 28. Cylindrical Wheel Contact on a Worn Switch Figure 29. Tapered Wheel Contact on a Worn Switch 2 Not the same new design profile as used for the curving (Section 2.2) and hunting (Section 2.3)analyses 26

TCRP WOD 65 Part 2 It is a common practice in European railroad and rail transit systems to add an increased taper roll-off segment to the wheel tread near the field side, as Figure 29 shows. Compared to the tapered wheels without the roll-off, the wheel tread roll-off provides additional rolling radius difference on curves where high rail contacts on the wheel flange and low rail contacts on the roll-off segment. The larger rolling radius difference promotes steering and improves curving performances. It also delays the onset of wheel hollowing as the wheels wear. AREMA standard (AREMA 2009) requires that switch points rise up ¼ inch above stock rail top. The rise of the switch point is to prevent “false flange” contact on the field side of the wheel tread in the trailing point move. Figure 30 shows that the wheel flange face will contact the stock rail and push it outward if the switch point doesn’t rise up. This is especially important for hollow worn wheels. However, the standard European switch design does not have such requirements; therefore, the switch point and stock rail tops are in the same plane. Instead of raising the switch point to prevent false flange contact, the European car manufacturers often design wheels with a roll-off segment on the tread near the field side, which has more taper than the main part of the tread. Figure 30. A Worn Wheel Contact on a Worn Switch Point and Stock Rail It is not clear why these two switch standards are different. Historically, the European and North American (N.A.) rail networks evolved in two different directions. The N.A. system was based primarily on the reliable movement of heavy freight tonnage at the lowest cost, and the European system focused more on speed. A switch point rise will generate vertical track perturbations that have less of an effect on car performances at low speeds than at high speeds. Both AREMA and European recommended types of switches can meet the dynamic performance requirements for rail transit service because the transit car running speeds are relative low. However, wheel profile design requirements for running on these two different types of switches may need to be different. European car manufacturers usually design wheels with tread roll-off, which is necessary for European types of switches, but the benefits of the roll-off need to be justified for N.A. rail transit systems using AREMA recommended switches because: • High contact stress and rolling contact fatigue could occur near the area where the roll-off segment connects the main part of wheel tread with a sudden change of slope. • The roll-off segment further increases wheel tread slope, which may not be compatible with frog design as discussed in the following subsection. 2.5.1.2 Switch and Crossing Frogs W/R impact on frogs is very sensitive to wheel profile shapes. Most rail transit systems in N. A. have adopted the standard AREMA frog, which was designed for tapered wheels. For transit systems using cylindrical wheels, the standard frog nose is often welded back to be level with the wing rail to be 27

TCRP WOD 65 Part 2 compatible with cylindrical wheels. A new design wheel with a tapered tread will bluntly strike the existing frog nose and result in a depressed nose, as Figure 31 shows. The added taper on the field side exacerbates the problem. To avoid damage, the frog noses would need to be ground accordingly before the new candidate tapered wheels are introduced into service. Figure 31. A Taper Wheel and Cylindrical Wheel Contact on a Frog 2.5.2 Switch Point Protectors Figure 32 shows a typical switch point protector implemented in a yard switch. A recent study showed that wheel tread chamfers on locomotive wheels likely contributed to switch point derailments due to contact between wheel chamfer and the guard (Wilson et al. 2010). Figure 32. Switch Point Protector 28

TCRP WOD 65 Part 2 Figure 33 shows the wheel with a large chamfer (45 degrees) contacting on a new switch point guard at about a 78-degree contact angle. Figure 34 shows the wheel with smaller chamfer contacting on the vertical surface of the guard with a 90-degree contact angle. Both wheels are standard AAR S-622 cylindrical tread narrow flange wheels. The large chamfer (0.8839 inch length, 5/8 inch depth) wheels were permitted to be used prior to 2007. The small chamfer (0.4375 inch long) is currently the largest chamfer allowed in AAR M-107. Figure 33. A Larger Chamfer Wheel Contacts on a New Switch Point Guard Figure 34. A Smaller Chamfer Wheel Contacts on a New Switch Point Guard The contact angle of the larger chamfer wheel on the guard could be lower than 78 degrees as the switch point protector guard wears out. The low contact angle between the larger chamfer and the worn guard facilitates wheel climb. Wheel profiles that include a wheel chamfer need to be carefully designed with consideration of special trackwork compatibility. 29

TCRP WOD 65 Part 2 2.5.3 Spring Switches Spring switches are often used in light rail transit systems, especially in urban city areas. They lower costs by using a mechanical device (spring or retard) instead of a throw motor. It is automatically operated by mechanical devices located under road surface, which also eliminates the safety hazard caused by switch stand. Figure 35 shows a typical spring switch used in a light rail transit system. Figure 35. Spring Switch in a Light Rail Transit System In Figure 35, the mainline switch point is closed. In the trailing move direction of the branch line, the switch point is opened by the wheels; after the wheels pass the switch point, it is closed by the spring force. The spring force must meet requirements from two aspects: • The force has to be big enough to close the point for main line movement. • The force cannot be too high. The high spring force resists the wheel opening the switch point in the trailing move direction for branch-line movement, which may cause flange climb derailment. Figure 36 shows the W/R contact and spring force in a spring switch. To avoid flange climb derailment in a spring switch, the following criterion must be met: 𝜇𝜇2 + 𝜇𝜇3 + FV2 < 𝜇𝜇1+tan𝛼𝛼1−𝜇𝜇1 tan𝛼𝛼 (3) Where, α is the maximum wheel flange angle, 𝜇𝜇1 is the left side W/R friction coefficient, 𝜇𝜇2 is the right side W/R friction coefficient, 𝜇𝜇3 is the friction coefficient between the rail base and the sliding plate underneath the switch point, V2 is the right side vertical wheel load, and F is the spring force. 30

TCRP WOD 65 Part 2 Figure 36. W/R Contact on Spring Switch Points While the friction coefficient 𝜇𝜇3 can be decreased as low as possible by using grease to ease the sliding movement of the switch point on the plate, the W/R friction coefficient 𝜇𝜇2 cannot be controlled because of environmental changes. Lubrication on the switch point is limited by operation. The weather, wheel, and rail surface conditions can dramatically change the W/R friction coefficient. For a rusty switch point top and a new trued wheel, the friction coefficient could easily reach 0.6. Flange climb derailment could occur if the criterion in Equation 3 is not met because of high friction coefficients and low wheel flange angle. Another issue of spring switches is excessive wear occurring on the open point tip (right side switch point tip in Figure 36, which is open for facing movement). The wear was generated during trailing point movement when the wheel opened the switch point on the left side and pushed the right switch point towards the stock rail. A gap between the right side switch point tip and the stock rail exists because of the resistant spring force F and the friction resistance forces in a spring switch. The wheel not only wears out the switch point tip but also bends it towards the stock rail due to the gap. This type of unusual wear does not occur in a regular switch operated with a manual or motor switch machine, where the switch point was hidden under the stock rail without any gap. A wheel with a shallow flange angle (<70 degree), which usually has a larger radius flange root, could generate more wear and metal flow on a spring switch point tip than a wheel with high flange angle (>72 degree). Adopting high flange angle (>70 degree) wheels and lubricating new trued wheels and switch point top can reduce derailment risk and excessive wear in a spring switch. However, because use of lubrication is not failsafe, design and maintenance guidelines should assume dry, high friction conditions. 2.6 Wheel Profile Maintenance Guidelines Wheel maintenance is critical for rail vehicle safety and ride quality. Effects of wheel diameter differences and wheel wear on car performances were investigated in the following subsections to develop wheel profile maintenance guidelines. 2.6.1 Wheel Diameter Difference 2.6.1.1 Effect on Hunting Wheel diameter difference in an axle is usually caused by asymmetric wear or malfunction of a wheel truing machine. Figure 37 shows the effect of wheel diameter difference in the same axle on hunting performance. The wheel profile used in the simulations was the existing new 63-degree wheel profile. 31

TCRP WOD 65 Part 2 Figure 37. Wheel Diameter Difference (in the Same Axle) Effect on Hunting (56.25-inch gage) Figure 37 shows that the hunting speeds increased with wheel diameter differences. Wheel diameter differences in an axle cause the axle to shift away from the track center towards the wheel with smaller diameter, which decreases the W/R clearance between the smaller diameter wheel and rail. The HSC chart in Figure 19 shows that the hunting speed increases with the decrease of W/R clearance when the conicity stays the same. This conclusion only applies for new wheel profiles with diameter differences, because wheel diameter difference does not change conicity. For worn wheels with different diameters, both the conicity and clearance will change because of wear and diameter difference. Therefore, for worn wheels the hunting performance has to be evaluated based on the HSC chart. 2.6.1.2 Effect on Wear and Curving Performance For curving simulation, the smaller diameter wheel was implemented on the high rail to simulate the worst curving scenario for a car with wheel diameter differences. Figures 38 and 39 show that the wheel wear index and rolling resistance generally increase with wheel diameter differences on curves. Diameter differences from 0 inch to 0.2 inch (5.08 mm) were simulated. Figures 40 and 41 show that the wheel L/V ratios and lateral forces significantly increase with wheel diameter difference on curves. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 55 60 65 70 75 80 85 90 Tr uc k Fr am e Ac ce le ra tio n RM S (g ) Speed (mph) Det D=0.00 inch Det D=0.04 inch Det D=0.10 inch Det D=0.16 inch Det D=0.20 inch FRA 213 Limit 32

TCRP WOD 65 Part 2 Figure 38. Wheel Diameter Difference Effect on Wear Index Figure 39. Wheel Diameter Difference Effect on Rolling Resistance 0 100 200 300 400 500 600 0 1000 2000 3000 4000 5000 6000 Le ad in g W he el W ea r In de x (lb -in /in ) Curve Radius (feet) Det D=0.00 inch Det D=0.10 inch Det D=0.16 inch Det D=0.20 inch 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 1000 2000 3000 4000 5000 6000 Su m o f A ll W he el W ea r I nd ex (l b- in /in ) Curve Radius (feet) Det D=0.00 inch Det D=0.10 inch Det D=0.16 inch Det D=0.20 inch 33

TCRP WOD 65 Part 2 Figure 40. Wheel Diameter Difference Effect on Wheel L/V Ratio Figure 41. Wheel Diameter Difference Effect on Wheel Lateral Force 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 1000 2000 3000 4000 5000 6000 Le ad in g W he el L /V R at io Curve Radius (feet) Det D=0.00 inch Det D=0.10 inch Det D=0.16 inch Det D=0.20 inch 0 2 4 6 8 10 12 14 0 1000 2000 3000 4000 5000 6000 Le ad in g A xl e H ig h R ai l L at er al F or ce (k ip s) Curve Radius (feet) Det D=0.00 inch Det D=0.10 inch Det D=0.16 inch Det D=0.20 inch 34

TCRP WOD 65 Part 2 2.6.1.3 Effect on Wheel Unloading Wheel diameter difference effects on vertical wheel unloading were investigated by running the car through measured pitch and bounce track perturbations (AAR Chapter 11, Pitch and Bounce track perturbations, 39-feet wave length and maximum ¾-inch amplitude) at different speeds. Figure 42 shows that the wheel diameter difference effect on the wheel unloading ratio is negligible. Figure 42. Wheel Diameter Difference Effect on Wheel Unload 2.6.1.4 Effect on Vertical Wheel Load Equalization The maximum wheel unloading for a car equipped with axles with 0.25-inch wheel diameter difference between left and right wheels is 8 percent at 2.5-inch wheel drop. The maximum wheel unloading for car equipped with axles with 1.0-inch wheel diameter difference between leading and trailing axle in a truck is 7 percent at 2.5-inch wheel drop, both are well below the 65-percent limit for APTA Class G and R passenger equipment (APTA SS-M-014-06 2007). Wheel diameter difference effects on load equalization should be small as long as the static primary suspension deflection is larger than the wheel diameter difference in an axle. 2.6.1.5 Wheel Diameter Difference Summary In summary, wheel diameter differences on new or freshly turned wheels improve hunting performance due to the decrease of W/R clearance when the axle shifts from the track center position towards the smaller radius wheel. However, wheel diameter differences resulted in poor curving performances, such as more wear and larger lateral forces on high rails, which may cause gage spreading. Wheel diameter differences between axles and trucks may lead to components interfering or fatigue, which will have to be addressed case-by-case. 0 10 20 30 40 50 60 70 80 90 100 40 45 50 55 60 M in m um W he el U nl oa d Ra tio (% ) Speed (mph) Det D=0 mm Det D=1 mm Det D=2.54 mm Det D=4 mm Det D=5.08 mm FRA 213 Limit 35

TCRP WOD 65 Part 2 2.6.2 Wheel Wear 2.6.2.1 Effect on Hunting and Ride Quality Transit agencies usually adopt wear limits on wheel and rail wear to maintain acceptable vehicle and track performances. A new wheel quickly wears into a shape conformal with existing rails, which decreases contact stress, but may deteriorate car ride quality as it becomes heavily worn. Wheel wear may result in hollow treads and thin flanges. Wheel wear was usually well controlled by wear limits on tread, flange, and flange height. Transit wheel tread wear depth (hollowness) is usually less than that in freight railroads. In addition to flat wheels, worn wheels with thin and tall flanges exceeding limits are usually corrected through wheel truing. The flange thicknesses for the measured new, slightly worn, moderately worn, and heavily worn wheels from the representative heavy rail transit system in Figure 1 were 34, 32, 30 and 28 millimeters (1.7,1.3,1.2,1.1 inches), respectively. Figure 43 shows the conicities of these measured wheels on a new 115RE rail with 56.25 inch track gage. The conicity increased with wear for slightly and moderately worn wheels, but decreased when the wheel became heavily worn. W/R clearance increased consistently with wear as the flange wore out. Figure 43. Measured New and Worn Wheel Conicity, Track Gage 56.25 inches The calculated W/R conicity and clearance were plotted on the HSC chart for hunting performance evaluation, as Figure 44 shows. The estimated hunting speed for the new measured wheel is above 80 mph, while the hunting speeds for the worn wheels are about 70 mph. 0 0.2 0.4 0.6 0.8 1 1.2 0 2 4 6 8 10 12 14 16 18 Eq ui va le nt C on ic ity Axle Lateral Displacement (mm) Measured New Wheel Slightly Worn Wheel Moderately Worn Wheel Heavily Worn Wheel 36

TCRP WOD 65 Part 2 Figure 44. Hunting Speed Estimation for a Car Equipped with Measured New and Worn Wheels Figure 45 shows the predicted car hunting speeds. The car equipped with new wheels has the highest hunting speed (about 81 mph), the hunting speeds with worn wheels decrease to about 67 mph, although the heavily worn wheel’s hunting speed is a little higher than the slightly and moderately worn wheel because of its lower conicity. The estimated hunting speeds from the HSC chart are generally consistent with the predicted hunting speeds from the simulations. 37

TCRP WOD 65 Part 2 Figure 45. Truck Frame Accelerations of a Car Equipped with Measured New and Worn Wheels (56.25-Inch Gage) Figures 46 and 47 show the conicities of these measured wheels with 56.5- and 57-inch track gages, which were used in shallow curves and tight curves, respectively. The conicity decreased and clearance increased with the increase of track gage. Figure 48 shows that the hunting speeds of measured wheel with track gage variations can be estimated from the HSC chart based on their conicity and clearance. Figure 46. Measured New and Worn Wheel Conicity, Track Gage 56.5 inches 0 0.5 1 1.5 2 55 60 65 70 75 80 85 90 95 100 Tr uc k Fr am e Ac ce le ra tio n RM S (g ) Speed (mph) Measured New Wheel Slightly Worn Wheel Moderately Worn Wheel Heavily Worn Wheel FRA 213 Limit 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 2 4 6 8 10 12 14 16 18 20 Eq ui va le nt C on ic ity Axle Lateral Displacement (mm) Measured New Wheel Slightly Worn Wheel Moderately Worn Wheel Heavily Worn Wheel 38

TCRP WOD 65 Part 2 Figure 47. Measured New and Worn Wheel Conicity, Track Gage 57 inches Figure 48. Wheel Wear Effect on Hunting 39

TCRP WOD 65 Part 2 The car hunting speeds of new measured wheels with track gage variations are higher than that of worn wheels. However, the heavily worn wheel wore the tapered tread into a flat shape, which lowered the conicity. So, the car stability with heavily worn wheel is even better than slightly and moderately worn wheel due to its lower conicity. Wheel wear may increase or decrease conicity depending on the wheel and rail geometry parameters, wear patterns, and operation conditions. Traditional wheel wear limits on tread and flange are not directly related with car hunting performances. The proposed W/R contact geometry based HSC chart provides a useful tool to control the car hunting speed with worn wheels above the operational speed. Even though the hunting speed of the heavily worn wheel was similar to that of slightly worn and moderately worn wheels, Figure 49 shows that the low conicity heavily worn wheel with large W/R clearance has a negative effect on ride quality, probably due to the resonance response from carbody. Similar phenomena were observed in tests as described by Polach 2009 and Smith and Kalousek 1991. Further research on secondary suspension and carbody vibration modes is needed to investigate low conicity effect on ride quality. Figure 49. Effect of Wheel Wear on Ride Quality Even though the measured new wheel profile generates low conicity, its maximum flange angle is still 63 degrees; therefore, it still has higher flange climb derailment risk than that of the new design wheel (with 70-degree flange angle). The new design wheel was optimized for both curving and hunting performances. 40

TCRP WOD 65 Part 2 2.6.2.2 Effects on Frog Impact Figure 50 shows four frog cross section profiles at locations A, B, C, and D in a standard AREMA No. 20 turnout. The distances on the track at location B, C, and D were measured as 0.2, 0.8, and 4.58 feet from location A (frog nose), respectively. Figure 50. No. 20 Turnout Frog Profiles Figure 51 compares the impact ratios predicted for the new and worn wheel profiles operating over a new No. 20 turnout frog. The impact ratio was defined as: Impact ratio = maximum impact force static wheel load (4) Figure 51. W/R Impact Loads on a New Frog 1 1.5 2 2.5 3 3.5 30 35 40 45 50 55 60 Im pa ct R at io Speed (mph) Measured New Wheel Slightly Worn Wheel Moderately Worn Wheel Heavily Worn Wheel 41

TCRP WOD 65 Part 2 Figure 51 shows the impact ratio on frog nose increases with wheel wear and running speed. The heavily worn wheel impact ratio was increasing by about 25 percent compared to that of the new wheel. Heavily worn wheels have to be removed to prevent damage to the vehicle and frog. The wheel wear limit for removal can be set up based on the impact ratio. 2.6.2.3 Effect on Wear and Curving Performance Figure 52 shows the wear index of the worn wheels increases on curves with radii from 800 to 2,000 feet, similar to new wheels on tight and shallow curves. Figure 53 shows a similar trend can also be found for lateral forces applied on high rail. Wheel wear effects on curving performances are relatively small compared to hunting and frog impact. Figure 52. New and Worn Wheel Wear Index in Curves with New Rails 42

TCRP WOD 65 Part 2 Figure 53. New and Worn Wheel Lateral Forces in Curves with New Rails 2.6.2.4 Wheel Wear Summary Wheel wear effects on hunting performance depend on W/R wear patterns: • If W/R wear results in high W/R conicity and larger gage clearance, a car may start hunting at speeds lower than that with new wheels and rails. • If W/R wear results in low W/R conicity and larger gage clearance, car hunting stability needs to be evaluated by using the HSC chart. Impacts on frogs increase with wheel wear. Wear limits on the wheel tread and flange can be set up based on W/R impacts, which also depend on frog wear conditions and running speeds. 43

TCRP WOD 65 Part 2 C H A P T E R 3 Conclusions and Recommendations The following conclusions and recommendations are made from this study: • Both W/R contact conicity and W/R gage clearance have significant and complex effects on car lateral stability (hunting), especially as the wheels and rails wear. A new method for evaluating the combined effects of these two parameters was developed using HSC charts. To demonstrate the new method, two HSC charts were developed for a representative heavy railcar and a representative light railcar. The charts were used to generate the following guidelines for wheel profile design and maintenance: – Hunting speed generally decreases (car becomes more unstable) with the increase of both conicity and W/R clearance. – Wheel and rail wear increases W/R clearance, and its effect on hunting depends on wear pattern:  Worn wheels with high conicity and wide W/R clearances cause the hunting speed to decrease quickly  Worn wheels with low conicity and wide W/R clearances cause the hunting speed to decrease slowly, but may result in sudden onset of hunting instability – Hunting speeds for high conicity wheels are more sensitive to W/R clearance variations than for low conicity wheels. – HSC charts can be used to evaluate new wheel profile designs and also to evaluate worn wheels (and rails), and develop wear and gage clearance tolerances. To provide specific conicity and gage clearance guidelines for a particular vehicle in a transit system, a new HSC chart would need to be developed using simulations for the particular case. • Increasing the maximum flange angle can effectively reduce flange climb derailment risk. APTA recommends a 72-degree (with tolerance +3 degrees and -2 degrees) flange angle wheel for use in passenger railcars. However, a wheel profile with a flange angle less than 72 degrees (but high enough to prevent flange climb) can also be adopted to provide a smooth transition from an existing low flange angle wheel profile to a new design with high flange angle wheel profile. • Wheel profiles (new and worn) should be compatible with special trackwork: – Impact forces on the frog nose generally increase with wheel wear. – Wheels with profiles that are incompatible with the frog generate significant impact on the frog. – High flange angle wheels can reduce flange climb derailment risk and reduce excessive switch point tip wear in spring switches. • Wheel diameter differences on an axle can improve hunting performance because of the decrease of W/R gage clearance when the axle shifts from the track center position toward the smaller radius wheel. However, wheel diameter differences may result in poor curving performances, such as more wear and larger lateral forces on high rails, which may cause gage spreading. – Systems with many curves may need tighter tolerances on wheel diameter differences than systems with few curves and mostly straight track • Wheel wear has significant effects on both hunting speed and frog impact. Wear limits on wheel treads and flanges can be determined by the HSC chart and impacts with frogs, which also depend on frog wear conditions and running speed. • On-track tests are recommended to further validate these guidelines. 44

TCRP WOD 65 Part 2 • Effects of wheel profiles with zero and negative W/R conicity (hollow worn wheels) on rail transit car hunting performance are recommended for further investigation. 45

TCRP WOD 65 Part 2 References American Public Transit Association. APTA SS-M-014-06, Standard for Wheel Load Equalization of Passenger Railroad Rolling Stock. Washington, D.C., 2007. American Railway Engineering and Maintenance-of-Way Association. AREMA Manual of Railway Engineering, Vol. 1, Chapter 5, Track. Lanham, MD, 2006. Federal Railroad Administration. 49 CFR Part 213, Track Safety Standards, Subpart G, Train Operations at Track Classes 6 and higher, 213.333 Automated Vehicle Inspection Systems, Washington, D.C., Amended March 13, 2013. Griffin, T. TCRP Report 114: Center Truck Performance on Low-Floor Light Rail Vehicles. Transportation Research Board of the National Academies, Washington, D.C., 2006. International Union of Railways. UIC Leaflet 518. Testing and Approval of Railway Vehicles from the Point of View of Their Dynamic Behavior – Safety – Track Fatigue – Running Behavior, 2009. International Union of Railways. UIC Leaflet 519. Method for Determining the Equivalent Conicity, 2004. Polach, O. “Wheel Profile Design for the Target Conicity and Wide Contact Spreading.” Proceedings of the 8th International Conference on Contact Mechanics and Wear of Wheel/Rail System, Italy, 2009. Shu, X. “Survey of Current Wheel Profiles and Maintenance Practices.” Final Report, TCRP D-7 TASK 20 Task 1, Transportation Technology Center, Pueblo, CO, 2014. Smith, R. E. and J. Kalousek. “A Design Methodology for Wheel and Rail Profiles for Use on Steered Railway Vehicles,” Wear, Vol. 144, 1991. Strogatz, S. H. Nonlinear Dynamics and Chaos. Addison Wesley Publishing Company, 1994. True, H. “Does a critical speed for railroad vehicles exist?” RTD-Vol. 7, Proc. Of the 1994 ASME/IEEE Joint Railroad Conference, Chicago IL, March 22–24, 1994. Wilson, N., D. D. Davis, and S. Anankitpaiboon. “Analysis of Contact Issues Between Locomotive Wheels and Switch Point Guards.” Technology Digest TD-10-18, Association of American Railroads, Transportation Technology Center, Inc., Pueblo, CO, 2010. 46

TCRP WOD 65 Part 2 A P P E N D I X Light Railcar Hunting Speed Contour Chart A.1 Vehicle Model A typical light rail vehicle model consisting of two carbodies and three trucks with the following specifications was used in this study: • Two carbodies articulate on the middle truck • Primary Chevron suspension • Secondary airbag suspension • Lateral and vertical damper in secondary suspension • Axle spacing: 6.3 feet • Truck Center Spacing: 23 feet • Wheel load: Mid truck: 5.2 kips, End truck: 8.2 kips • Wheel diameter: 27 inches Vehicle model parameters were measured through characterization tests. The measured primary suspension longitudinal, lateral stiffness, and damping were reduced by half to simulate a worn truck condition. The standard AREMA 115RE rail profile and 0.5 W/R friction coefficients representing dry wheel and rail contact condition were used in the simulations. A.2 Hunting Speed Contour Chart Hunting speeds of a light railcar with different wheel profiles were obtained through simulations by using the methodologies described in subsection 2.3 of this document. Figure A1 shows the hunting speed contour (HSC) chart for the simulated light railcar. Section 3 of this document discusses the guidelines for using the light rail HSC chart for new wheel design and worn wheel maintenance. 47

TCRP WOD 65 Part 2 Figure A1. Hunting Speed Contour Chart for a Light Railcar 48