VERTICAL split rim (VSR) failures are a current source of major wheel failures in North American heavy-haul operations and are of great concern to wheel manufacturers and railways alike, with both forged and cast wheels having suffering VSRs while in service.

AmstedExtensive testing over the past few years using X-ray diffraction techniques has shown that the axial residual stress pattern within the wheel rim is significantly different for new AAR class C wheels than AAR class C wheels that have failed due to a VSR, and non-failed AAR class C wheels that have been operating in service.

VSRs almost always begin at areas of tread damage, resulting from shelling or spalling, and cracking that propagates into the rim section under load. At the rim locations tested, the new wheels have a relatively "flat" axial residual stress profile, compressive but near neutral, caused by the rim quenching operation, while wheels that have been in service have a layer of high axial compressive stress at the tread surface, and a balancing zone of axial tensile stress underneath.

The magnitude and direction of this axial tensile stress is consistent with the crack propagation of a VSR failure. When cracks from tread surface damage propagate into this sub-surface axial tensile zone, a VSR can occur under sufficient additional service loading, such as loads caused by in-service wheel-rail impacts from tread damage. Furthermore, softer class U untreated wheels, which were removed from service and tested, have a balancing axial tensile stress layer that is deeper below the tread surface found in used class C wheels.

Following observations and studies of many VSRs by Amsted Rail, we understand that VSRs almost exclusively start at areas of tread damage, and then propagate below the wheel tread for some period of time before a large circumferential section of the wheel fractures off. Both cast and forged wheels have suffered VSRs in service, and multiple wheel designs have experienced VSRs. Typically, the front rim face of the wheel breaks off, although VSR failures have noted instances where the back rim face and flange have broken off the wheel.

Amsted-fig-1Figure 1 shows a classic VSR wheel failure, with a large circumferential piece of the front rim face broken off. The initial failure area is shown at an area of tread damage, with cracking propagating "down" into the rim towards the wheel plate. The crack then proceeds to the left and right "away" from the area of initial fracture, with several crack arrest marks noted.

No clear cause was found for VSRs. Microcleanliness, macrostructure, chemistry, and manufacturing records all appeared to be normal when failures were investigated. Indeed, a 2011 TTCI study showed that 28 out of 30 VSRs passed the AAR microcleanliness test, with the two wheels that failed manufactured prior to adoption of the test. Further, Amsted Rail's internal studies showed sulphur and phosphorous values for a group of VSR wheels was less than the average value of sulphur and phosphorous for other production wheels, and ultrasonic signal distributions were the same for VSR and non-VSR wheels.

As a result, investigations into axial residual stress began, with axial defined as across the rim section from back rim face to front rim face.

Wheel samples were sent to Lambda Research in Ohio for core drilling and X-ray diffraction testing from the wheel tread surface to determine the residual stress present. A new wheel and a VSR wheel were found to have different residual stress patterns, with the VSR wheel showing a stress reversal from the as-manufactured compression, to axial residual tensile stress slightly below the tread surface. The core drilling and X-ray diffraction measurements for the VSR wheel were taken on the tread surface some distance away from the failed area due to concern about relaxation of stresses at the failure. The maximum depth of measurements was about 8.4mm, thus another method of sampling was needed to find internal rim stress.

After consulting with Lambda Research, we focused on testing of radial wheel slices (cut flange to bore) for measurement of axial residual stress. We determined that the stress relaxation effect was extremely small, so further testing proceeded using cut radial slices without a correction factor. A second X-ray diffraction testing supplier, American Stress Technologies (AST), also measured the axial residual stress in wheel rims, with results from samples previously tested by Lambda found to be similar to AST's results.

X-ray diffraction measurements on the face of radial slices were typically taken at distances of 38.1mm and 76.2mm from the front rim face of the slices. The 38.1mm distance is near the location of many VSR failures, and the 76.2mm distance is near the tape line. Prior to taking X-ray diffraction measurements, measurement areas were electro polished to remove any residual stresses associated with saw cutting. Measurements were taken along a line below the tread surface, typically at 3.18mm intervals, to a depth of 50.8mm. Slices were removed from areas not within the VSR failure area due to concerns about stress relaxation from the previous fracture.

Initial work involved cast wheel slices cut from five VSRs, four new, and eight used (five class C and three class U) wheels. Slices were tested to determine if there were differences in the axial residual stress pattern. The class U wheels are untreated (not rim quenched), while slices from two forged wheels were also analysed to see if the residual stress pattern was different from cast wheels.

amsted-fig23Figure 2 shows average axial residual stress results for lines of X-ray shots taken in the five different wheel types at a distance of 38.1mm from the front rim face of the wheel. Figure 3 shows average axial residual stress results that were taken on some of the same new, used C, used U and VSR wheels at a distance of 76.2mm from the front rim face of the wheel. The depth below tread surface is shown on the x-axis, while stress is plotted on the y-axis. Stress values below zero are compressive and stress values above zero are tensile.

From this data, and other Amsted Rail residual stress testing, we proposed a VSR failure mechanism as follows:

• tread damage (shelling or spalling) leads to cracks that propagate to various depths under service rolling and impact loads

• cracks that propagate into the sub-surface zone of tension can rapidly change orientation, and

• service loads such as rolling contact, bending moments, and pounding from impacts can lead to a VSR failure.

Similar axial residual stress patterns for used class C wheels were also found during tests at TTCI using a slit cutting technique. In addition, we found that used class U wheels (untreated and therefore softer) have compressive axial residual stresses at the surface, but the balancing axial tensile stresses were located deeper below the tread surface at a depth of approximately 25.4mm to 31.8mm.

Testing of two wheels revealed axial tensile stress at every X-ray measurement point along a line below and parallel to the tread surface. This confirms the axial tensile stress induced by service loads forms a zone of tension below the tread surface.

Anecdotal evidence suggests VSRs were not common when class U wheels were in service. With the axial tensile zone deeper below the tread surface, at locations more difficult for cracks associated with tread damage to propagate, it is logical VSRs are less common with class U wheels. We note wheel loads were generally lower when class U wheels were in widespread use.

Subsequent work focused on determining whether there is a correlation between total indicated runout (TIR) and axial residual stress. Seven 91.4cm (36 inch) wheels were tested for TIR on a machining centre, and runout at various locations around the circumference of the wheel tread was recorded. Four wheels were forged and three wheels were cast. A line of X-ray measurements was taken from the tread surface down into the rim at a distance of 63.5mm from the back rim face.

The low TIR slices were taken from areas of the wheel with as close to zero TIR as possible. The maximum compression and tension values for X-ray diffraction readings were recorded along the measurement line. The axial stress exhibited the same general pattern as VSR and service worn wheels in past analysis, with high values of axial compressive stress near the tread surface, and balancing axial tensile stress below the surface. A CB38 wheel was also similarly tested.

A group of tread damaged wheelsets (both forged and cast wheels) were machined in wheel truing lathes at two major North American railways. Such machining is known to remove tread damage and cracking, and also restore the wheel flange and tread profile.

Of particular interest was the effect of machining on the axial residual stress profile in the wheel. Wheel truing, long known to restore flange and tread profiles and remove tread damage from wheel tread surfaces, can also reduce the deep axial compression at the tread surface and balancing axial tension below the tread surface. This reduction in axial residual tensile stress below the tread surface should provide for additional safety margin in reducing VSR wheel failure frequency.

Wheel truing also removes tread damage and associated cracking that can lead to VSR formation.

We have not heard reports of VSR failures in other parts of the world and VSRs appear to be most common in North America. We suspect tread condition may be partly responsible for this situation. Operations that run very heavy gross rail loads, such as on Australia's heavy-haul network, have no experience of VSR failures, but their wheels are trued more frequently than many wheels in North America.


These tests have shown that a distinct pattern of axial residual stress is evident for newly manufactured class C wheels, used class C wheels, and VSR class C wheels. New wheels have an axial residual stress near zero - slightly compressive - while used and VSR class C wheels have compression at the tread surface, and balancing axial tension deeper in the rim. VSR wheels have slightly greater axial stress levels than used wheels.

Tread damage, caused by shelling and spalling cracking, is a key contributor to VSR wheel failures. These cracks propagate into the rim section under rolling/pounding service loads. Axial residual tensile stresses generated in service are also a key factor in the formation of VSRs. A zone of axial tensile stress, about 12.7mm to 19mm deep, exists below the wheel tread surface in VSR and used class C wheels. When the sharp cracks from tread damage reach the high axial tensile stress zone below the surface, the cracks change direction, begin to propagate more rapidly, and become a VSR.

For VSR wheels, variation in axial residual stress was noted in measurements from slices removed at different circumferential positions. Values along the tread width at the same circumferential point can also vary while relaxation from the failure may play an important role in localised axial stress effects.

The studies also found that higher out-of-round wheels exhibited a greater average maximum axial compression value and a greater average maximum axial tension value than low out-of-round wheels. Used cast and forged class C wheels exhibit similar axial residual stress patterns when tested using X-ray diffraction. Class U (untreated) wheels have an axial tensile residual stress zone deeper below the tread surface than class C wheels. Our understanding is that this is because class U wheels are softer and deform more than the harder class C wheels.

We also found that corrective machining of wheel tread damage not only removes tread damage and cracks, and restores the flange and tread profile, but it also reduces the magnitude of compressive and tensile axial residual stresses in the wheel rim. If performed more frequently by US and Canadian railways, it could help to reduce the number of VSRs and as a result contribute to an improvement in safety on North America's heavy-haul network.

*This article is based on a paper presented at the 17th International Wheelset Congress held in Kiev, Ukraine, in September 2013.