Rolling Isn’t That Simple

When you take an assembled ball bearing (inner and outer races, balls, and ball separator) in your hands it has a satisfying “rightness” about it. Made of hard steel, it’s precise and smooth-running. If you spin a clean, dry bearing, it spins a long time. The ideal contact between a ball and a surface having a larger radius of curvature is a point. It just seems right that rolling point contact should have low friction.

Yet we know that Enzo Ferrari in the early 1950s ordered two V-12 engines to be built – one with an all rolling bearing crank, the other with plain journal bearings. In this experiment and in similar comparison tests made by Junior Johnson with NASCAR V8s, plain journal bearings were found to generate no more friction than rolling bearings, and to be far superior in load capacity, durability, and low weight.

Why? One way to look at this is to understand how motorcycle tires roll. If you measure the circumference (distance around) of a motorcycle tire in the center of its tread, you get a bigger number than if you measure circumference around the right or left shoulders of the tire. One result of this is that leaned over in a turn (running on the tire shoulder) you will see a higher number on the tach than when you are upright at the same actual speed.

Another result is that because the tire flattens against the pavement from the weight of bike, fuel, and rider, you will get a different circumference if you measure around the center of the tire or around the right or left edges of the tire’s footprint. This is because the tire’s profile is not flat, like that of a car tire, but is curved. Thus, the center of the footprint rolls forward slightly farther than do its right and left edges; how can this be? The tread rubber, being quite elastic, stretches enough to allow this to happen. The center of the footprint does try to roll farther than the right and left footprint edges, and there is probably some friction as these slightly deformed parts of the footprint lift up off the pavement at the rear of the footprint.

Now extend this reasoning to the balls in a ball bearing. Like a tire, under load, steel balls flatten slightly against their inner and outer races. At the same time, the races are slightly, elastically indented by the pressure of the ball rolling against them. This produces a “footprint” just as a loaded tire does. And just as in the case of a tire, the center of the ball’s “footprint” tries to roll forward slightly farther than do the right and left edges of the footprint. But because steel balls are much more rigid than tires are, the only way this difference in the distance rolled for each turn of the ball can be achieved is by very slight slippage in the footprint.

The pressure between ball and race can be high enough to put it into the realm of ‘elastohydrodynamics’ – the zone of high pressure in which the viscosity of lubricants increases sharply. That produces some friction, but most friction in rolling bearings surely results from over-lubrication, forcing balls and rollers to squish through thick layers of lubricant that take some power to squeegee ahead of the rolling elements. Think of this as going running in the surf. That is indicated by the very low friction achieved in the 13,000-hp reduction gearing in Pratt & Whitney’s new GTF or Geared Turbo Fan engines. By being very careful to supply just enough lubricant to carry away heat, and pumping it away effectively they achieved a claimed 99.3% efficiency for the unit.

A further source of friction in rolling bearings is the shearing of oil films between rolling elements and their separator or cage. Eliminate the separator, then? No, because when one ball or roller rubs on the one ahead of it, the rubbing takes place at double speed.

Bottom line? Short of magnetically-levitated crankshafts rotating in high-vacuum crankcases, the support of rotating parts produces friction.