From: jbrandt@hpl.hp.com (Jobst Brandt)
Newsgroups: rec.bicycles.tech
Subject: Re: What causes Rear Axle to break?
Date: 11 Mar 1999 22:03:20 GMT

Aka Tom N writes:

> I have broken 2 rear axle in the last 5 yrs.  I ride a Campagnolo 6
> sp rear hub, 126mm spacing, Campagnolo axle.  I'm actually a light
> weight, but climb a lot of hills.  The first time I didn't even know
> I broke an axle until I got a flat tire, and released the quick
> release...that's when things fell apart.  So question is, what
> causes a rear axle to break?

That's not bad for that kind of hub... if you rode the bike a bit.  If
you look at the hub with the freewheel removed, you will note that the
axle with spacers and jam nut extend the width of 3-4 sprockets
farther on the right side than the left.  These hubs were designed for
four speed freewheels and as more gears were added, the axle extension
was increased to the point where they broke frequently.  This was part
of the impetus for the cassette hub.

Additionally, rear dropouts until recently were still slotted
longitudinally, a leftover from early days before derailleurs, when
one had to move the wheel to make up for differing chain length.  This
slot leaves the jam nut unsupported front and rear so that the pull of
the chain can more easily bend the axle fore and aft. This effect is
self reinforcing because the hardened jam nut frets its way into the
face of the dropout.  These wear marks are often so deep that the
wheel will not drop out without a sharp forward rap with the hand.

Vertical dropouts help but only a little.  The real solution is a
cassette hub from Shimano, that has almost no overhang (Campagnolo
cassettes have overhang), or you could go to an ultra six freewheel
that would get you back to 120mm spacing.  Even that doesn't get
entirely rid of the problem that was there when 4 and 5-speed
freewheels were used.

> Faulty axles from Campy?

No, but not good enough material to do the job.  Higher strength steel
would do it if you can get someone to make it.  By the way, Campagnolo
axles don't bend.  Bent ones always have a crack already so don't both
straightening them.  If you happen to overhaul your hub and the axle
is still straight, switch ends so that the fatigued end is on the
left.  Put a mark on the middle with a file to indicate that it has
been reversed so you won't waste time next time you see it.

Jobst Brandt      <jbrandt@hpl.hp.com>

From: jbrandt@hpl.hp.com (Jobst Brandt)
Newsgroups: rec.bicycles.tech
Subject: Re: Remove freeweel without a puller?
Date: 26 Apr 2001 21:56:32 GMT

Tony Raven writes:

> Not having ultrasound or dye testing it I don't know it wasn't
> cracked but there was no visible crack and I find it hard to
> envisage a crack that would cause the axle to be curved along its
> length.  Perhaps you could also explain the subtle semantic
> differences in "axles don't bend, they break" and "I did not say
> that your axle was not bent"

It's in this thread and I explained why bending an axle is extremely
improbable and that it is a fatigue failure.  Fatigue failures do not
bend, they crack.  Next time you get a "bent" axle, break it and see
for yourself that there is an old part of the crack and a new part.
It doesn't take expensive analytical methods.  Don't be so obstinate.
I explained it and told how to prove it, and you keep bitching.  I
think there is a team effort to see how long I can take this griping.

Jobst Brandt      <jbrandt@hpl.hp.com>

From: jbrandt@hpl.hp.com (Jobst Brandt)
Newsgroups: rec.bicycles.tech
Subject: Re: Remove freeweel without a puller?
Date: 27 Apr 2001 18:47:23 GMT

Tony Raven writes:

>> It's in this thread and I explained why bending an axle is extremely
>> improbable and that it is a fatigue failure.  Fatigue failures do not
>> bend, they crack.  Next time you get a "bent" axle, break it and see
>> for yourself that there is an old part of the crack and a new part.
>> It doesn't take expensive analytical methods.  Don't be so obstinate.
>> I explained it and told how to prove it, and you keep bitching.  I
>> think there is a team effort to see how long I can take this griping.

> I am not being obstinate, simply reporting the observation,
> improbable or not, that I have had an axle that is bent in a nice
> curve from one end to the other.

That defies an engineering explanation of how such a bend might occur
without a bending fixture and careful control.  Shafts bend at the
point of greatest bending stress, a place often not apparent to the
average observer.  Typically, bicycle forks flex and bend near the
fork crown, although many people believe they do that at the slim
curved ends, a place where there is no bending moment.  For an axle to
bend in a large curve would require the ends to be held and bent as
one would bend a coat hanger wire by hand, by applying torque at both
ends.  Although one could imagine such a combination of loads in a
bicycle axle, with bearing loads from a high vertical landing acting
at the 1/4 points of the axle span, this is not the common bending
mode of axles.

Chain tension loading and road shock place peak loading at the inboard
end of the right bearing cone, the place where axles usually "bend" as
they crack.  As I said, this doesn't happen all at once or we would
have broken axles all over the place, were they operating that close
to the yield (bending) stress.

Jobst Brandt      <jbrandt@hpl.hp.com>

From: jbrandt@hpl.hp.com (Jobst Brandt)
Newsgroups: rec.bicycles.tech
Subject: Re: Remove freeweel without a puller?
Date: 27 Apr 2001 18:26:06 GMT

Adrew Muzi writes:

> I agreed with you about Q/R axles but I pointed out that it was
> probably a solid axle if it was bent and in fact in his later post
> he confirmed it.  I have straightened solid axles-it's not
> impossible- with judicious use of a hammer.

So how did it get bent?  If axles are soft enough to bend, why don't
the bend on the first steep climb or jump?  The point is that they
"bend" after considerable use and not after any sudden overload.  No
doubt, there have been axles that were bent in incidents where a car
ran over the rear hub.  I am not contesting that one cannot bend and
axle, only that axles that seem to have become bent with time are in
fact cracked.  Solid axles are as subject to this as hollow ones, the
material in the core having no measurable effect on bending strength.

> The problem is one of time and money, often it's noticed and
> corrected in a moment during a flat tire change for someone who
> doesn't know or care and couldn't pay for an overhaul anyway.

That's when we usually notice it unless one is observant of such
phenomena.  For QR folks this is seen easily by spinning the skewer
wobbly jam nut is easily detected.(that spins the axle) when taking
the wheel out.

> Q/R axles don't have enough material to bend without breaking but
> that was not the discussion here.

Tilt!

Jobst Brandt      <jbrandt@hpl.hp.com>

From: jobst.brandt@stanfordalumni.org
Subject: Re: History: horizontal dropout vs vertical?
Newsgroups: rec.bicycles.tech
Message-ID: <dtCK8.6875$3w2.32674@typhoon.sonic.net>
Date: Mon, 03 Jun 2002 04:42:49 GMT

Ted Bennett writes:

>> The wheel drops into place without adjustment or special care but
>> far more important, the jam-nut is supported fore and aft, the
>> direction of bending of the rear axle.  Rear axle failure on
>> Campagnolo freewheel hubs was a common occurrence before vertical
>> dropouts that practically eliminated the problem for those who
>> still ride these antique hubs.

> Jobst, that first sentence contains an assertion that I have never
> before heard, that axles bend in the fore-aft direction.  I find
> that surprising, as I had always assumed that an overload that
> results in a bend would tend to be in a vertical direction.

> Would you expand on that idea?  (Or anyone.)

Well, take your weight plus what you can pull with your other leg when
you stand on the pedal and you get more than twice your weight and
applied at he worst place on the axle.  To verify this, just look at
the inside of a right horizontal dropout and you'll see where the jam
nut augered into the steel from bending fore and aft.  This leaves the
axle supported on a hinge, so to speak.  The left side does not show
such damage.

Normal wheel loads are not concentrated on the right ball bearing in a
bending mode similar to pedaling loads that produce an enormous
bending force applied at the ball bearing by an axle that is greatly
overhung by the use of an originally 4-speed freewheel expanded to 6,
or 7-speeds.

The Shimano design addressed this problem the best of any on the
market.

Jobst Brandt  <jobst.brandt@stanfordalumni.org>  Palo Alto CA

From: jobst.brandt@stanfordalumni.org
Subject: Re: Hollow vs solid axles-was: Butted vs. straight-gauge spokes
Newsgroups: rec.bicycles.tech
Message-ID: <Bu%2a.66176$Ik.2790019@typhoon.sonic.net>
Date: Fri, 14 Feb 2003 05:52:33 GMT

Sheldon Brown writes:

http://www.efunda.com/DesignStandards/springs/calc_comp_fatigue_eqn.cfm
http://www.grc.nasa.gov/WWW/RT1996/5000/5220l.htm

>>> In the second of these references, I found:

>>> "In monolithic materials, it has been observed that tensile mean
>>> stresses are detrimental and compressive mean stresses are
>>> beneficial to fatigue life in comparison to a base of zero mean
>>> stress."

>>> There has been some controversy on this list about the observed
>>> greater durability of quick release axles vis-a-vis solid axles, and
>>> this would seem to support those of us who maintain that QR axles
>>> are less prone to breakage than solid ones, despite having less
>>> material.

>> That doesn't apply to axle steel and the difference between QR
>> axles and solid ones is structurally minimal, the material (bore)
>> lying in the neutral axis.  I am fairly sure that the QR axles are
>> better material.  My rear axle failures resulted from horizontal
>> slot dropouts which give no fore and aft bending support to the
>> axle.  "Vertical" dropouts support the forward circumference of the
>> jam nut and therefore, support most of the bending load from chain
>> tension, the one that generally causes fatigue failures.  That is
>> why these dropouts occasionally fail on the right* side.

> There has been a lot of verbiage spilt since then and lots of arcane
> references, but none of them has explained what property makes "axle
> steel" not subject to the same physical laws that govern other
> "monolithic materials."

> I remain unconvinced that the NASA quote doesn't apply, and indeed
> explain the observed greater reliability of quick-release axles.

The QR rear axle operates like the rotating wheel axle in the example
given earlier.  That is, although it has a relatively light
compression it has large tension caused mainly by chain pull.
Considering that a crank is about 170mm and a chainwheel is about
100mm radius, then a 160lb rider outs a 270lb chain force on the axle
to which at least half the rider weight bears on the rear axle
vertically.  When (not) supported by a horizontal dropout slot, this
is a large bending load to which the QR tension adds practically
nothing.  The bending load causes about 500lbs tension and compression
in the forward and trailing part of the axle, the ball bearing being
about an inch away from the face of the jam nut and having a diameter
of somewhat less than 0.2".

>>> Unloaded, the QR axle is in compression, with chain load and some
>>> road shock, depending on how tightly the QR is stressed, the axle
>>> deflects into tension and then back into compression.  This is a
>>> stress reversal.

> Certainly there is occasional tension on parts of a quick-release axle,
> but the greatly predominant mode is compression, as is evidenced by the
> many riders who've ridden long distances on broke QR axles held together
> by their skewers, and only noticed the break after removing the wheel
> for some reason.

The compression from the QR is piffles compared to the dynamic bending
loads the axle supports.  I think the stresses caused by bending are
grossly underestimated by most observers.

> The compression of the skewer creates a bias that prevents the axle
> from experiencing any tensile load except at times of unusual
> stress, as when hitting a pothole.  Even when this happens, the
> magnitude of the tensile stress is greatly decreased by the
> countervailing compressive force from the skewer.  Since it is
> clearly the tensile forces that cause the axle to break, I find the
> NASA hypothesis quite convincing.

I think that a few simple calculations will show that the QR is not
producing much compression in comparison to the tension and
compression caused by chain pull and rider load.

Jobst Brandt
jobst.brandt@stanfordalumni.org
Palo Alto CA