First, lets review some facts about bearings.

• All bearings have a finite life.
• In the laboratory, life predictions are extremely accurate and the cause of failure is normal material fatigue.
• In the field, many factors must be considered in order to estimate service life.
• The most common causes of failure in railway service are loose components, mechanical damage, lubrication failure,
  and spalling.

A comparison of failure modes in laboratory testing and in railway service shows the following differences:

Laboratory Fatigue(L10) Spalling:

• Primarily due to sub-surface initiated fatigue from cyclic stress
• Failure defined as a spall size of 0.01 square inches on the roller or raceway surface

Railway Service Life Spalling Due To:

• Cyclic stress
• Brinell damage
• Lubricant breakdown
• Water etched components

Loose bearings resulting from:

• Oversize cone bores
• Undersize axle journals
• Absence of cap screw clamp load

Mechanical damage from:

• Displaced adapters
• Derailment
• Wheel impacts

Failure Defined as:

• Bearing fails in service or is removed for overheating
• Does not meet reconditioning standards in the AAR Roller Bearing Manual

From this you can conclude that laboratory L10 Life is certainly not the same as railway service life.

Bearing Fatigue Life in General Terms

Predicting bearing life under laboratory conditions has become very precise. Items that influence fatigue life include material quality, manufacturing process, internal design, lubrication and the application conditions such as load, speed, and temperature.

From years of testing by many companies, standards for estimating bearing lives have been developed. The Anti-Friction Bearing Manufacturers Association (AFBMA) has standardized on methods of determining bearing capacity and calculating L10 fatigue life. It is important to note that these standards are revised periodically as a result of improvements in bearing steels, manufacturing techniques and technology.

Definition of L10 Fatigue Life

A calculated life (hours, miles, etc.) that a group of identical bearings operating under controlled conditions will obtain before the first evidence of fatigue occurs, typically called spalling or surface flaking.

For a given population of identical bearings, 90% will meet or exceed the predicted life, and 10% will fail before reaching that value.

Failure is defined as the development of a spall of approximately .01 square inches on the cone or cup raceway or roller. Figure 1 shows an area slightly larger than .01 square inches which would represent a spalling failure in the laboratory.

Figure 1

To the laboratory researcher, this is typically the end of useful service life. However, in the railroad industry provisions have been made for reconditioning bearings to extend their use. Under current reconditioning standards bearings are commonly returned to service containing repaired spalls up to 0.375 inch x 0.375 inch or 0.14 square inches. This represents a spalled surface 14 times larger than laboratory failure.

Causes of Fatigue Failure of the Bearing Elements

The quality and cleanliness of bearing steels today are vastly improved compared to the steels of 25 years ago. However, due to imperfections and non-metallic inclusions in the steel, normal fatigue failure will eventually develop in any bearing.

Fatigue of the bearing components occurs due to the cyclic stressing between the rollers and raceways. The rollers support the applied load and in doing so create high stresses underneath the raceway surfaces. Figure 2 shows the sub-surface stress distribution for a bearing that is running with an adequate lubricant film between the races and rollers. This distribution would be typical of bearings run under laboratory conditions and relates well to the life calculations.

Figure 2

Keeping in mind that the maximum stresses are some distance below the raceway surface, Figure 3 shows pictorially how surface flaking results from initial sub-surface cracking. The high sub-surface stresses can cause cracking to initiate at inclusions, and the cracking then propagates until it finally breaks out to the surface.

Figure 3

To reiterate, the sub-surface fatigue failure mode is normal and is the basis for life calculations

Major Influences on Bearing Life - Laboratory and Railway Service
Bearing Loading

Fatigue life is most heavily influenced by the applied load on the bearing and is inversely proportional to the (10/3) power of the load. As an example, if the bearing experiences a 10% overload, the life is reduced to 73% of the original life. Similarly, a 25% overload reduces the bearing life to less than 50% of the original life. Figure 4 shows the load-life relationship of a 6½ X 12 bearing for the load range from unloaded to fully loaded on a 100 ton freight car.

Figure 4

Fatigue life can also be reduced by brinelling or water etching of the races and rollers and by lubricant breakdown due to water contamination. All of these factors have the effect of concentrating the applied bearing load and causing a reduction in fatigue life similar to that caused by increased load.

Raceway Integrity

Another factor that can have significant effect on fatigue life is the integrity of the raceway. One dramatic example is the void (hole) in the raceway that is left when a spall is repaired under current AAR reconditioning standards.

Figure 5 shows the Hertzian contact stress distribution along the roller for an undamaged bearing and one that has a 3/8 diameter repaired raceway spall.

Figure 5

The stress plot shows that the repaired spall not only increases the contact stress significantly in the vicinity of the repair but also introduces a high stress riser at the edge of the repair. As with increased bearing load, the effect of higher contact stress has an exponential effect on life reduction. When the roller is directly over the void, there is less raceway surface to support the load. For this case, the calculated life of the raceway with a 3/8 void is only 10% of the undamaged raceway.

Even after repairing the spalled area by hand grinding, sub-surface cracking may still be present as shown in Figure 6.

Figure 6

Under the influence of the high stress at the edge of the spall, the sub-surface cracking may continue and flake out adjacent to the initial repaired area.

Bearing Components Loose on the Journal

Oversize cone bores, undersize journals, and low axial clamp forces separately and collectively affect bearing life in service. None of these conditions are present during laboratory L10 Life testing. They are, however, the major contributors to in-service bearing failures in the form of overheated and burned off bearings. Separate studies by BRENCO and the AAR found loose cones and turning cones associated with 40% - 50% of the in-service failures.

Mechanical Damage
Mechanical damage can shorten the life of a bearing in a number of ways:

A displaced adapter can come into contact with the end cap or backing ring causing the bearing to loosen on the journal. The displaced adapter can also produce a concentrated load condition on the outer race reducing fatigue life as described in the bearing loading section.

Derailment can produce heavy brinelling of the races and broken components, especially the outer race. The effects of slid-flat wheels and built-up tread on bearing life are the object of some recent research activity. It has been known for some time that heavy wheel pounding in service can cause roller retainers (cages) to break, leading to bearing failure. The effect of wheel pounding on brinelling, spalling, and bearing loosening on the journal is not well understood at this time.

In Summary

The life calculation is a starting point for service life expectancy of railroad bearings and relates to fatigue failure only. In the railroad application, many other conditions such as bearing clamp, brinelling, repaired spalls, lubrication, and external damage can have the effect of reducing service life. However, the user can take action especially in the area of bearing clamp, that will effectively extend the service life of the bearings. These actions essentially address the 50% of bearing field failures caused by loose components.

Assure that all prescribed bearing installation procedures are
being followed. Axle journals should be verified for size, and
tapped holes in the axle ends should be free of any condition (dirt,
debris, burrs, metal chips, etc.) which would cause the cap screws
to bind on the threads.

In accordance with current AAR instructions, previously used cap
screws should be checked for mechanical damage and lubricated
prior to installation.

Follow the prescribed cap screw torquing procedure set forth in the
AAR Wheel and Axle Manual.

In addition to inspecting for loose cap screws and loose backing rings (both indications of loss of clamp on the axle), thorough field inspections of bearings in service can detect displaced adapters, damaged or loose seals, evidence of heat damage from thawing operations, and wheel conditions that could generate high impact forces into the bearing.

Over the past decade, all the leading bearing manufacturers have taken advantage of the cleaner steels and advanced manufacturing technologies to significantly extend the laboratory life of the railroad bearing. In fact, calculated laboratory life remains far in excess of the actual bearing life achieved in the field. If we are to realize any of these gains in actual service, increased attention needs to be given to the proper installation, maintenance, and frequent field inspection to insure that conditions detrimental to bearing life are quickly corrected.

References

Technical Forum 88-1, "Bearing Retention" and 88-3, "Proper Mounting of Bearing Enhances Retention" describe in more detail the necessary steps to improve bearing retention. A videotape, "Proper Bearing Mounting", is also available on this topic. Technical Forum 89-2 discusses in detail various procedures for inspecting bearing for indications of damage or distress.

The Technical Forum is an information resource for the rail industry and is provided as a courtesy of Amsted Rail Group. Suggestions, inquiries or comments are welcomed and should be directed to:

Editor, Technical Forum
BRENCO, Incorporated
P.O. Box 389
Petersburg, Virginia 23804
804-863-1713

BACK TO TECH SHEETS