Fatigue Study on Composite Material Bearings

• 1
  Key takeaways
• 2
  What a fatigue study on composite material bearings measures
• 3
  Load, speed, and temperature limits of composite bearings
• 4
  How to run your own fatigue study on composite material bearings
• 5
  Machining composite bearing housings and mating parts
• 6
  Reading the failure surface after a fatigue run
• 7
  Composite vs metal bearings: when each one fits

A rolling bearing fails in two different ways. Metal races spall: a subsurface crack grows until a flake of steel leaves the raceway. Composite races do not spall the same way. Fiberglass-nylon 66 is a polymer matrix with short glass fibers, so the damage starts as localized creep, fiber pull-out, and surface pitting. The race geometry slowly changes, clearance opens, and the rolling elements start to skid instead of roll.

That difference changes what a fatigue study is looking for. On steel you count cycles to spalling and read a classic S-N curve. On composite you watch for a stiffness drop, a rise in running temperature, and a growing radial clearance. The end of life is usually a functional limit, not a broken part. The bearing still turns, but it no longer holds the shaft where the design needs it.

The published work on fiberglass-nylon 66 bearings ran the same bearing through constant speed, variable speed, and reversing conditions. Constant speed produced the shortest life. Reversing produced the longest. That ranking surprises people at first, until you look at where the stress sits.

Under constant speed the outer race sees the same contact patch cycle after cycle in a narrow band. The polymer there works the hardest and heats up first. Under reversing duty the load direction flips, the loaded zone moves, and the material gets a partial recovery window between passes. That recovery is why reversing looks better in the data.

• 1
  Damage modeCreep, pitting, and fiber pull-out rather than subsurface spalling.
• 2
  End-of-life signalClearance growth and temperature rise, not a fracture.
• 3
  Test variable that matters mostContact stress and running temperature, not elapsed hours.

Composite bearings carry far less load than steel of the same envelope. A glass-filled nylon race deforms under contact pressure, so the allowable load is set by a stress limit, not by a strength limit. Push past it and the race flattens at the load line. The bearing does not break, it just stops being round enough to run smoothly.

Speed is limited by heat, not by centrifugal force. Every rotation shears the polymer and generates heat at the contact. At low speed the heat escapes through the housing. At high speed it builds faster than it leaves, the nylon softens, and deflection climbs. A composite bearing that runs at 60 °C in one application can run at 95 °C in a smaller housing at the same load and speed.

Temperature is the single hardest boundary. Nylon 66 absorbs moisture and loses stiffness as it warms. Below the glass transition the bearing behaves predictably. Above it, creep accelerates and the fatigue data you collected at room temperature no longer predicts anything useful.

Shaft and housing fit matter as much as the bearing itself. Composite races expand more than steel when they warm up. A press fit that is correct cold can become loose hot, and a loose race skids. We normally cut the housing bore to hold a light interference at the expected running temperature, not at 20 °C.

• 1
  Static loadSet by allowable contact stress, not by yield strength.
• 2
  SpeedLimited by heat generation versus heat removal.
• 3
  TemperaturePolymer softening above the glass transition kills the fatigue model.

Run the bearing in the housing it will actually live in. A test rig with a rigid steel block tells you about the material, not about your product. The housing wall thickness and the fit set how much heat leaves the race, and that changes the result more than most people expect.

Measure three things at once: radial clearance, housing temperature near the loaded zone, and running torque. Clearance is the end-of-life signal. Temperature tells you how far you are from the softening point. Torque rise usually shows up before either of the other two when the race starts to skid.

Cycle the load the way the machine does. If the application reverses, test reversing. If it runs one direction at constant speed, test that, and expect the shortest life. Do not run a constant-speed bench test and then apply the number to a reversing axis.

Keep the sample count honest. Composite fatigue scatters more than steel fatigue because fiber orientation and molding flow vary from part to part. Three samples give you a direction. Ten give you a number you can put in a design review.

• 1
  Test in the real housingWall thickness and fit drive heat removal.
• 2
  Log clearance and temperatureThese two track the actual failure mode.
• 3
  Match the duty cycleConstant speed and reversing give very different lives.
• 4
  Run enough samplesComposite scatter needs a larger sample than steel.

The bearing is only half the assembly. The housing bore, shoulder, and shaft finish decide whether the fatigue life you measured survives in production. Bore roundness matters more than bore size. An out-of-round bore loads two spots on the race and leaves the rest of the circumference idle, which concentrates stress exactly where the study says the material is weakest.

We hold housing bores to ±0.005 mm and check roundness on the same setup. For polymer races we usually aim for Ra 0.8–1.6 μm on the bore. A rougher bore lets the race cold-flow into the tool marks under load. A mirror finish can be worse, because it gives the polymer nothing to key into and the race creeps around the housing.

Shaft shoulders and retaining features need a radius that matches the bearing corner. A sharp internal corner cuts into a composite race during press-in. We machine a small relief or radius at the shoulder so the race seats on the face and not on the edge.

For prototypes we cut housings from 6061-T6 or 7075 aluminium and, when weight is critical, from carbon fibre or PEEK. A prototype housing in aluminium lets you change the fit after the first fatigue run without retooling. Production housings in die-cast aluminium or stainless follow once the fit is locked.

• 1
  Roundness firstAn out-of-round bore concentrates race stress.
• 2
  Bore finishRa 0.8–1.6 μm gives the polymer something to key into.
• 3
  Shoulder radiusMatch the bearing corner to avoid cutting the race.

A failed composite race tells you which limit you hit. Polished, flattened load lines mean contact stress was too high for the material. A brown or discolored band means heat built up faster than the housing could remove it. Loose, fretting marks on the outside diameter mean the fit was wrong and the race was walking in the bore.

Fiber pull-out with a rough, hairy surface points at a different problem. The glass fibers have separated from the nylon matrix, which happens when the part sees moisture and heat together. If your washdown cycle or humid environment is doing this, a different polymer grade or a metal bearing solves it faster than a redesign of the load path.

Skid marks on the rolling elements are worth a close look. If the elements slide instead of roll, the race is already deformed and the clearance has opened. That is end of life, even if the bearing still turns freely by hand.

Keep the failed parts. A photo of the race surface plus the clearance and temperature log is enough for us to tell you whether the design needs a bigger bearing, a cooler housing, or a change back to metal.

• 1
  Flat load linesContact stress exceeded the material limit.
• 2
  Discolored bandHeat could not leave the race fast enough.
• 3
  Fretting on the ODFit was too loose and the race was moving.
• 4
  Rough, hairy surfaceMoisture and heat separated fibers from the matrix.