Interesting Hub Flange Failure
Arthur Shapiro
light pinging noises coming from the rear of his bicycle. They got a little
bit worse as we were on one climb. It sounded as if his rear derailleur might
be just barely touching one spoke, but visually that didn't appear to be the
case.
The noise got louder and more frequent as we started up a fairly healthy
climb. Then suddenly something broke. It appeared to be one of the spokes on
his Bontrager rear wheel.
Then we looked closer:
http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
and the subsequent "Next Picture" link to picture #02. (Others in the set are
of no interest beyond the club. The pictures will be up for one week.)
Is there enough visual evidence to speculate on the cause / nature of this
flange failure?
Art
M-gineering
leading and trailing spokepair fight each other. The aluminium in
between is under compression and the hub can cope with the residual
stress in tension. Here two spokes work together in tearing the hub apart
--
/Marten
info(apestaartje)m-gineering(punt)nl
Tim McNamara
arthur...@cox.net (Arthur Shapiro) wrote:
My guess is just too much material removed from the flange. Note that
the spoke heads sit not in individual spoke holes but in slots.
Werehatrack
I would add to the observations above that the low-spoke-count wheels
have loading delta during rotation, and may be operating at a higher
intial tension. If the Bontrager hubs are machined from billets
instead of forged, that might be a contributing factor as well. It is
often the case that the closer you get to the bleeding edge of
technology, the more fragile the tech gets.
--
My email address is antispammed; pull WEEDS if replying via e-mail.
Typoes are not a bug, they're a feature.
Words processed in a facility that contains nuts.
Werehatrack
<rau...@earthWEEDSlink.net> said:
>
>I would add to the observations above that the low-spoke-count wheels
>have loading delta during rotation,
make that "higher loading delta during rotation relative to that
typical of a high-spoke-count wheel."
Michael Press
arthur...@cox.net (Arthur Shapiro) wrote:
> I was riding with a fellow club member on a Thursday club ride, and heard some
> light pinging noises coming from the rear of his bicycle. They got a little
> bit worse as we were on one climb. It sounded as if his rear derailleur might
> be just barely touching one spoke, but visually that didn't appear to be the
> case.
>
> The noise got louder and more frequent as we started up a fairly healthy
> climb. Then suddenly something broke. It appeared to be one of the spokes on
> his Bontrager rear wheel.
>
> Then we looked closer:
>
> http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
It literally exploded!
> and the subsequent "Next Picture" link to picture #02. (Others in the set are
> of no interest beyond the club. The pictures will be up for one week.)
>
> Is there enough visual evidence to speculate on the cause / nature of this
> flange failure?
--
Michael Press
bicycle_disciple
bicycle_disciple
carl...@comcast.net
Dear B.D.,
The greek delta letter is often used to indicate change in equations.
On a low spoke count wheel, the change in load may be greater.
That is, on a 36-spoke wheel, the bottom 5 spokes lose considerable
tension and then regain it. The total tension loss (or change or
delta) is spread over 5 spokes.
On an 18-spoke wheel, the same tension loss is spread over only 3
spokes, so the total tension loss _may_ be greater.
Unfortunately, we have only FEA's for 36-spoke wheels and don't know
what fewer spokes do. Here's an FEA for 36-spokes:
http://www.astounding.org.uk/ian/wheel/index.html
Ian's FEA suggests that the 5 spokes lose tension at roughly these
relative rates (67~69, 244~250, ~350):
/ / | \ \ 36-spoke wheel
-7 -25 -35 -25 -7 total ~99
But what happens to an 18-spoke wheel with the same load?
It could produce a much greater change in the middle spoke:
/ | \ 18 spokes?
-7 -85 -7 total ~99 much greater change
Or the same change:
/ | \ 18 spokes?
-32 -35 -32 total ~99 same change
Or even slight less change:
/ | \ 18 spokes?
-33 -33 -33 total ~99 slightly less change
The assumption is that the correct model tends toward the greater
change example, meaning that a low spoke-count wheel's spokes see a
greater delta or change in tension, from o to say 85 rather than just
from 0 to 35.
The greater the range of loading delta (spoke tension change), the
sooner the flange will fatigue and fail.
Cheers,
Carl Fogel
Joe Riel
> This will surely make it to my blog.
>
> B.D
>
> http://cozybeehive.blogspot.com
That [method used on your blog, and described by analyticcycling] is a
poor way to measure rotational inertia. I discussed this some years
ago, I believe on the (now defunct) hardcore bicycling science mailing
list. To get an accurate measurement you want to decrease the length
of the pendulum. The easiest way to do that is to hang the top
section of the rim on a knife edge.
--
Joe Riel
bicycle_disciple
Why is it poor? How off is one potentially going to be? I thought the
100 swings sort of evens things out, and you get a good figure, but I
understand this is not a foolproof method.
B.D
bicycle_disciple
>
> - Show quoted text -
Thanks Carl, this weekend I'll probably go over the details in that
link. So delta just pertains to change in loading, and with lower
spokes, there's a higher change? And in the life of a low spoked
wheel, the greater loading per each cycling reduces its time to
failure at some spot? Correct?
carl...@comcast.net
Dear BD,
The greater the change in tension, the sooner the failure, as long as
the tension doesn't reach the yield point for anything involved.
But whether a wheel with fewer spokes behaves that way is the
question.
Someone with a tension gauge and an 18-spoke wheel could probably
determine what's actually going on more quickly and more convincingly
than anyone can do an FEA for such wheels.
Part of the problem is that low-spoke-count wheels tend to be much
stiffer, so it's not clear what happens.
Another part of the problem is that there's disagreement about whether
stiffer wheels with fewer spokes need higher tension.
I suspect that most of the low-spoke-count wheels have been developed
in just the same way as the original 36-spoke wheels--trial and error
come first, then maybe some theorizing, and maybe not.
The spoke holes in the flange in question actually resemble an old
highwheeler trick, but the idea was to allow cross-7 lacing that would
normally have had the shafts of spokes covering the heads of other
spokes:
http://i3.tinypic.com/5ymzs40.jpg
Marten has pointed out elsewhere in this thread that spokes in this
arrangement look like a bad idea because the two spoke heads close
together are trying to pull the hub apart.
http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
The modern version is even worse than the highwheeler version, since
the two spoke holes are now a single elongated hole, with no
reinforcing material between them.
Cheers,
Carl Fogel
Joe Riel
You are computing the rotation inertia via the formula
Irot = m*R^2*(g*(T/2/pi)^2/R - 1)
where
R = length of the pendulum
m = mass of wheel
T = period of oscillation
Let's assume most of the error is in the measurement of R
(that is almost certainly the case with the setup you show).
The sensitivity of Irot with respect to R is
(dIrot/dR)*(R/Irot) = 1 - R^2/R0^2,
where Irot = m*R0^2. We know that R0 < Rw (wheel radius).
From the the pictures it looks like R ~ 3*Rw, so the
magnitude of the sensitivity is greater than 3^2-1 = 8.
If you use the technique I suggest (use the inside top edge
of the rim as a pivot point on a fixed knife edge), the
sensitivity is reduced more than eight times.
--
Joe Riel
bicycle_disciple
> On Fri, 23 Nov 2007 17:28:35 -0800 (PST), bicycle_disciple
>
>
>
>
>
Yes I noticed how the spoke head are in one elongated hole, trying to
pull that area apart. Actually, I have the wheel in question with me
in my basement, I was surprised I hardly noticed this before!
B.D
jobst....@stanfordalumni.org
> Then we looked closer:
http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
Cross laced spokes in hubs with a larger number of spokes have the
flange section between two spokes in compression. With few spokes, as
shown in this picture, this part of the flange is in tension and fails
similarly to radial spoking where stress in spokes arise primarily
from tension rather than wheel loads that in this wheel cause these
high tension loads to vary with each wheel revolution.
Jobst Brandt
jobst....@stanfordalumni.org
now that we need a former metallurgist's advice and failure analysis?
Jobst Brandt
jobst....@stanfordalumni.org
> Then we looked closer:
http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
Cross laced spokes in hubs with a larger number of spokes have the
flange section between two spokes in compression. With few spokes, as
shown in this picture, this flange section is in tension and fails
similarly as with radial spoking where stress in spokes arise
primarily from static tension rather than wheel loads that in this
wheel cause spoke tension loads to vary with wheel revolutions.
Jobst Brandt
Tom Sherman
> Where's: http://www.jimbeam.com/beam/default.aspx?
> now that we need a former metallurgist's advice and failure analysis?
Darning holes in the sock puppet?
--
Tom Sherman - Holstein-Friesland Bovinia
"Localized intense suction such as tornadoes is created when temperature
differences are high enough between meeting air masses, and can impart
excessive energy onto a cyclist." - Randy Schlitter
Tom Ace
> Cross laced spokes in hubs with a larger number of spokes have the
> flange section between two spokes in compression. With few spokes, as
> shown in this picture, this part of the flange is in tension and fails
> similarly to radial spoking where stress in spokes arise primarily
> from tension rather than wheel loads that in this wheel cause these
> high tension loads to vary with each wheel revolution.
I'm not following you here. Don't conventional 36-spoke wheels
have flange sections that alternate being in compression and
in tension?
Tom Ace
jobst....@stanfordalumni.org
>> Cross laced spokes in hubs with a larger number of spokes have the
>> flange section between two spokes in compression. With few spokes,
>> as shown in this picture, this part of the flange is in tension and
>> fails similarly to radial spoking where stress in spokes arise
>> primarily from tension rather than wheel loads that in this wheel
>> cause these high tension loads to vary with each wheel revolution.
> I'm not following you here. Don't conventional 36-spoke wheels have
> flange sections that alternate being in compression and in tension?
Not only do these spoke pairs pulling in opposite directions not
overlap and compress the material between, there is no material
between. These spokes appear to be laced into slots that contain two
spokes. The bridge between has got to fail. Most hubs are deigned to
have opposing spokes pull against each other when properly laced.
This wheel was not and the hub was designed to not do that, judging
from the slots containing two spokes. Being black that isn't easy to
see but the failed section shows no material between the ends of the
slot.
Look at response #1 to see the effect phrased slightly differently.
Jobst Brandt
Sir Ridesalot
Hi there Jobst.
I just lightened and cropped the image of the hub and posted it on
Flickr @:
http://www.flickr.com/photos/73832500@N00/2058503237/
You can clearly see that the 2 spokes are in a slot and not 2
individual spoke holes.
Great design huh?
Cheers from Peter
Ben C
> bicycle_disciple <1.crazy...@gmail.com> writes:
>
[...]
>> Why is it poor? How off is one potentially going to be? I thought the
>> 100 swings sort of evens things out, and you get a good figure, but I
>> understand this is not a foolproof method.
>
> You are computing the rotation inertia via the formula
>
> Irot = m*R^2*(g*(T/2/pi)^2/R - 1)
>
> where
> R = length of the pendulum
> m = mass of wheel
> T = period of oscillation
>
> Let's assume most of the error is in the measurement of R
> (that is almost certainly the case with the setup you show).
> The sensitivity of Irot with respect to R is
>
> (dIrot/dR)*(R/Irot) = 1 - R^2/R0^2,
>
> where Irot = m*R0^2. We know that R0 < Rw (wheel radius).
> From the the pictures it looks like R ~ 3*Rw, so the
> magnitude of the sensitivity is greater than 3^2-1 = 8.
>
> If you use the technique I suggest (use the inside top edge
> of the rim as a pivot point on a fixed knife edge), the
> sensitivity is reduced more than eight times.
It might be harder to keep the oscillation in the plane of the wheel
with that method though. At least if you use string, you can make a sort
of triangular swing that's relatively stable.
Ben C
> Arthur Shapiro writes:
>
>> I was riding with a fellow club member on a Thursday club ride, and
>> heard some light pinging noises coming from the rear of his bicycle.
>> They got a little bit worse as we were on one climb. It sounded as
>> if his rear derailleur might be just barely touching one spoke, but
>> visually that didn't appear to be the case.
>
>> The noise got louder and more frequent as we started up a fairly
>> healthy climb. Then suddenly something broke. It appeared to be
>> one of the spokes on his Bontrager rear wheel.
>
>> Then we looked closer:
>
> http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
>
>> and the subsequent "Next Picture" link to picture #02. (Others in
>> the set are of no interest beyond the club. The pictures will be up
>> for one week.)
>
>> Is there enough visual evidence to speculate on the cause / nature
>> of this flange failure?
>
> Cross laced spokes in hubs with a larger number of spokes have the
> flange section between two spokes in compression. With few spokes, as
> shown in this picture, this flange section is in tension
Isn't the situation just the same with 32H or 36H? The basic pattern is
the same as with this failed hub-- if you number the spokes A, B, C, D,
etc. then if A and B are crossing over and compressing the flange, then
B and C will be pulling the section of flange between them apart. C and
D compressing again, D and E pulling.
The unusual thing about this Bontrager hub is that the holes aren't
evenly spaced. The sections of flange that are being pulled apart by
spokes are shorter-- just the length of those slots. The sections in
compression are longer. I don't know why or whether that made any
difference to why it failed.
Mike
"Ben C" <spam...@spam.eggs> wrote in message
news:slrnfkg9oo....@bowser.marioworld......
>> Cross laced spokes in hubs with a larger number of spokes have the
>> flange section between two spokes in compression. With few spokes, as
>> shown in this picture, this flange section is in tension
The section under tension didn't fail - the failures are both in sections
that were under compression.
> Isn't the situation just the same with 32H or 36H? The basic pattern is
> the same as with this failed hub-- if you number the spokes A, B, C, D,
> etc. then if A and B are crossing over and compressing the flange, then
> B and C will be pulling the section of flange between them apart. C and
> D compressing again, D and E pulling.
Yes, but see below.
> The unusual thing about this Bontrager hub is that the holes aren't
> evenly spaced. The sections of flange that are being pulled apart by
> spokes are shorter-- just the length of those slots. The sections in
> compression are longer. I don't know why or whether that made any
> difference to why it failed.
If you look closely at the photos, the crack on the left is in line with the
spoke that was attached in that slot. I think the crack on the right is
unrelated to the primary failure, and happened after the crack on the left.
The left crack looks very much like the result of the spoke effectively
cleaving the flange. The spoke tension is partially translated into tension
in the flange that is orthogonal to the direction of the spoke, effectively
trying to split the flange with a pressure that increases with the spoke
tension. Imagine the spoke end as the tip of a chisel, applying pressure
against the flange.
Mike
Joe Riel
While it's been a while since I used the technique, I don't
recall that being an issue. To confirm that, I took five
minutes to try. Not a problem. The wheel stays in plane.
--
Joe Riel
Ben C
weight:http://www.yellowjersey.org/vel9.jpg
[http://www.ocrebels.com/images/GelsonsNBNovember2007/pic02.htm]
[...]
> If you look closely at the photos, the crack on the left is in line with the
> spoke that was attached in that slot. I think the crack on the right is
> unrelated to the primary failure, and happened after the crack on the left.
> The left crack looks very much like the result of the spoke effectively
> cleaving the flange. The spoke tension is partially translated into tension
> in the flange that is orthogonal to the direction of the spoke, effectively
> trying to split the flange with a pressure that increases with the spoke
> tension. Imagine the spoke end as the tip of a chisel, applying pressure
> against the flange.
That sounds plausible, and if you look at an old hub (that didn't fail)
you often see little notches where the spokes were pulling on it.
The interesting thing is that the left spoke (as it appears in the
picture) is not one that pulls on the flange when you pedal hard-- it's
going the other way.
So if that is the one that failed first, the initial damage may have
been done when the wheel was first put together. That initial notch
gradually fatigued into a big enough crack for it to break.
jobst....@stanfordalumni.org
>> If you look closely at the photos, the crack on the left is in line
>> with the spoke that was attached in that slot. I think the crack
>> on the right is unrelated to the primary failure, and happened
>> after the crack on the left. The left crack looks very much like
>> the result of the spoke effectively cleaving the flange. The spoke
>> tension is partially translated into tension in the flange that is
>> orthogonal to the direction of the spoke, effectively trying to
>> split the flange with a pressure that increases with the spoke
>> tension. Imagine the spoke end as the tip of a chisel, applying
>> pressure against the flange.
> That sounds plausible, and if you look at an old hub (that didn't
> fail) you often see little notches where the spokes were pulling on
> it.
> The interesting thing is that the left spoke (as it appears in the
> picture) is not one that pulls on the flange when you pedal hard--
> it's going the other way.
Pedaling torque has close to no effect on spoke tension as you can see
from FEA diagrams in "the Bicycle Wheel". The cause of flange failure
is high stress from tension cycled by lead bearing forces when these
spokes pass through the load affected zone.
> So if that is the one that failed first, the initial damage may have
> been done when the wheel was first put together. That initial notch
> gradually fatigued into a big enough crack for it to break.
Stop guessing. This is a bad design as others have noted and there
isn't much more to it.
Jobst Brandt
Ben C
[...]
>> So if that is the one that failed first, the initial damage may have
>> been done when the wheel was first put together. That initial notch
>> gradually fatigued into a big enough crack for it to break.
>
> Stop guessing. This is a bad design as others have noted and there
> isn't much more to it.
But what's wrong with the design? As far as I can see on any normal 3x
hub similar sections of the flange are in compression and in tension
just like with this one.
The only apparent differences are the slots and the irregular spacing of
the holes. That's a different design certainly but it's not obvious to
me why it's bad or necessarily the cause of this failure.
Tim McNamara
Ben C <spam...@spam.eggs> wrote:
What's wrong with the design? The slots, for one thing, and the spokes
pulling away from each other with the slot in between. Isn't that
obvious from the photos?
carl...@comcast.net
Dear Tim,
The more I look at it, the less obvious it seems and the more I wonder
if Mike and Ben are right.
It could be just a typical hub failure, with neither the tiny slot nor
the pull-apart spacing having much to do with the failure.
***
Here's a similar pull-apart hub failure, but there's no slot:
http://felixwong.com/gallery/show.php?photo=broken_bike_parts9c.jpg
Another similar pull-apart hub failure, again with no slot:
http://www.bikexprt.com/witness/product/wheels.htm
***
Solid flanges fail without slots or pull-apart design:
http://materials.open.ac.uk/mem/mem_ccf1.htm
Bigger picture:
http://materials.open.ac.uk/mem/images/images_cc/ccf4.jpg
***
Radial non-drive side, solid flange failure over 3 holes:
http://pardo.net/bike/pic/fail-001/FAIL-032.html
***
Solid 2-cross failure:
http://pardo.net/bike/pic/fail-001/tune-broken.jpg
***
Radial front failure:
http://i7.photobucket.com/albums/y270/andrewmeade/MVC-789S.jpg
***
Cross-3 drive-side failure:
http://photos1.blogger.com/blogger/1531/2525/320/IMG_0536.jpg
It's from http://midweekwayfarers.blogspot.com/2006_03_01_archive.html
***
Cheers,
Carl Fogel
carl...@comcast.net
A few more:
Cross-3 drive side:
http://www.lemlem.de/smolka_live/wochenbuch/aktuell/00063/images/1036548307.jpg
***
Here's a nice picture that shows how the break occurs in a pull-apart
flange with a slot, this time on the non-drive side:
http://gallery.roadbikereview.com/showphoto.php?photo=16000&si=hub
The long crack has gone completely through the flange along the spoke.
As the crack propagates, the bridge over the slot is levered outward
and cracks at the other end of the slot at about 90 degrees, just like
the original picture that started the thread.
Here's the original again for easy comparison:
http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
Cheers,
Carl Fogel
Werehatrack
<sunset...@REMOVETHISyahoo.com> may have said:
>jobst....@stanfordalumni.org wrote:
>> Where's: http://www.jimbeam.com/beam/default.aspx?
>> now that we need a former metallurgist's advice and failure analysis?
>
>Darning holes in the sock puppet?
Restuffing a straw man?
bicycle_disciple
> On Sat, 24 Nov 2007 18:55:15 -0700, carlfo...@comcast.net wrote:
> >On Sat, 24 Nov 2007 17:32:56 -0600, Tim McNamara
> ><tim...@bitstream.net> wrote:
>
> >> Ben C <spams...@spam.eggs> wrote:
>
> >>> On 2007-11-24, jobst.bra...@stanfordalumni.org
>
> >***
>
> >Cheers,
>
> >Carl Fogel
>
> A few more:
>
> Cross-3 drive side:
>
>
> ***
>
> Here's a nice picture that shows how the break occurs in a pull-apart
> flange with a slot, this time on the non-drive side:
>
> http://gallery.roadbikereview.com/showphoto.php?photo=16000&si=hub
>
> The long crack has gone completely through the flange along the spoke.
>
> As the crack propagates, the bridge over the slot is levered outward
> and cracks at the other end of the slot at about 90 degrees, just like
> the original picture that started the thread.
>
> Here's the original again for easy comparison:
>
> http://www.ocrebels.com/images/GelsonsNBNovember2007/pic01.htm
>
> Cheers,
>
> Carl Fogel
I wonder if Jobst is right here more... bad design.