This article is from the Bicycles FAQ, by Mike Iglesias with numerous contributions by others.
From: Jobst Brandt <email@example.com>
Date: Fri, 11 May 2001 16:35:42 PDT
Fretting or to fret: to eat or gnaw away, to erode.
In machinery, fretting is the micro-motion of tightly fitting parts
that superficially appear immobile with respect to each other.
Classically, transmission shafts and gears or axles with a press fit
show evidence of motion on disassembly by the presence of rouge, rouge
being iron oxide particles that are generated in such interfaces by
micro-motions far smaller than conventional measuring equipment can
On bicycles such an interface occurs between the square taper on the
pedal crank and its spindle, where rouge is evident on the face of the
steel spindle regardless of whether it was assembled with grease or
not. That fretting occurs is also evident by the need for a retaining
bolt to prevent crank disengagement from its spindle and of pedals
from their crank. Removing a crank requires substantial force with an
extractor, yet continual fretting will disengage the crank in the
absence of a retaining bolt. Likewise pedals are not easily removed,
but without a left hand thread on left pedals, they will unscrew.
In addition to disengaging the press fit of a crank, fretting moves
cranks up the taper until the preload of the retaining and
installation bolt matches the press times the slope of the taper.
That is to say, fretting relaxes surface friction loads in the
interface. Additionally, load distortion of a crank causes it to move
away from the face of the retaining bolt, up the taper of the spindle.
Pedals have similar relative motions in the attachment thread and
pressure face on the shoulder of the spindle. This is also a dynamic
joint that appears to be static. In the case of the pedal, fretting
motion is directional and can cause precession by the "wandering" load
whose center of pressure rotates in the crank thread opposite to the
rotation of the crank. Even without clearance, elastic deformation of
the crank and pedal spindle cause micro motions that, if not countered
by an appropriate thread direction, will unscrew the pedal. The
presence of a left hand thread on the left pedal and on many bottom
bracket right side bearing cups is proof that fretting occurs.
If these motions did not occur, then bolt locking devices, such as
cotter pins, lock nuts and lock washers would not be necessary. Most
nuts and bolts so secured do not come loose in service and therefore
should not rotate. Presence of locking means gives evidence that
fretting is more ubiquitous than most people (mechanics and engineers
Fretting in bearings is a different but similar effect, that is the
bane of steering gears and other mechanical devices that are intended
to rotate but are primarily used in a fixed position (straight ahead).
Automotive patents for anti-fretting steering gears abound. Saginaw,
Gemmer, and Ross steering gears come to mind. In bicycles this effect
is seen in the bearings of the fork, or head bearings, that are meant
to rotate but often experience straight ahead, non rotating use.
Because fretting involves invisibly small motion, it remains difficult
to understand and hard to convey to the user who suffers fretting
symptoms on a piece of machinery. It was long believed that impact
cause Brinelling of bicycle head bearings even though mechanics who
installed cottered cranks should have noticed an inconsistency in that
pounding in cotters with a large hammer with all the shock taken up by
one 1/4" ball under the crank spindle never caused a dent, yet 20
balls loaded by a much smaller force through a rubber tire was
believed to cause dimpled head bearings. Beyond that, the top bearing
that carries practically no load and receives no impact, also became
dimpled and, like the bottom load bearing one, did so in the fore and
aft quadrant. These dimples were not shiny as Brinell indentations
are, but are milky finish typical of tear-outs from asperity welding.
Ball bearings operate in two modes that became apparent in the
computer disk business because their data actuators often move step by
step from track to track, with a radial arm about 1" long, there being
more than 20,000 tracks per inch. Servo control engineers must
analyze bearing drag to be overcome for this purpose. In such small
motions, ball bearings are essentially locked solid with their
lubricant film, the bearing appearing as welded balls acting as
springs. This "pre-roll" stage of motion is the one that causes the
dimples in the bicycle head bearings because they, unlike the disk
bearings, have been lubricant depleted from fretting, not having made
a larger motion for a longer time, motion that would replenish
lubrication between ball and race.
Ball bearings roll on a film of oil that is so thin that it does not
present liquid properties, being several mono-molecular layers thick
as it adheres to ball and race. If it weren't for this behavior, oil
would not remain in the interface. However, with fretting, oil is
displaced and pin point welding takes place. Bicycle head bearing
fretting is caused by fore and aft rocking of the fork crown, a motion
that lies below visible resolution, and is small enough to not
replenish lubricant. Bearing damage appears as dimples from myriad
asperity contacts that welded and broke loose as the ball fretted in
place, leaving a milky finish.
Road bicycles are more subject to this damage than off road bicycles
because they spend more time traveling straight ahead, especially when
coasting downhill. Fretting damage occurs during these times, because
lubrication is not replenished by steering motions. The compound
bearings offered by Shimano seem to have greatly reduced the problem
by taking up fork crown rocking motion in a plain steel on aluminum
spherical cup that is not prone to metal to metal contact, while
steering rotations are borne in a pre-loaded full complement angular
contact ball bearing supported by this plain bearing.