or Connect
Cycling Forums › Forums › Bikes › Cycling Equipment › Spoke stress relief test
New Posts  All Forums:Forum Nav:

Spoke stress relief test

post #1 of 26
Thread Starter 
When bent at the elbow, spokes are expected to have residual stresses.

It's often claimed that squeezing spoke pairs together will relieve
these residual stresses by raising the spoke tension above normal.

Unfortunately, before and after stresses have never been shown to
confirm or refute this theory, nor has any formal testing ever been
published to show whether squeezed spokes outlast unsqueezed spokes.
Fatigue testing for spokes is incredibly tedious, taking months on
expensive rigs, so there's little hope in that direction.

A further complication is that squeezing may also improve the wheel
build in other ways, such as seating spoke heads and nipples,
releasing spoke wind-up, and simply encouraging care in other steps.
So even if squeezed spokes do last longer, it may have nothing to do
with stress relief.

Neutron diffraction images are supposed to reveal stresses, but the
neutron diffractor at Fogel Labs has been on the blink for a long
time.

However, Fogel Labs has a propane torch that works just fine. Heat is
generally agreed to relieve residual stresses, and a propane torch
will heat a spoke bend to a cheery orange glow in a few seconds,
relieving any residual stresses. The heated spoke visibly bends to a
new angle as the stresses are relieved.

Alas, initial testing showed an odd problem when old-fashioned carbon
spokes were sacrificed on the altar of stress-relief. Bent in the
middle and toasted, the carbon spokes bent impressively to a new
angle. But they turned out to bend the wrong way, opposite of the
direction in which modern stainless steel spokes bend when toasted.

Hmmm . . . the general opinion in materials circles was that the
carbon steel went through a phase change that overwhelmed the change
in residual stresses. Stainless steel, on the other hand, is not
expected to suffer such phase changes, so testing proceeded with
modern spokes.

The next problem was that factory elbow bends are too tiny, too short,
and too awkward for any useful testing. If they bent after toasting,
the angle was too small to be seen, even with camera magnification.

So a spoke-bending rig was cobbled together:

http://server5.theimagehosting.com/i..._spoke_rig.jpg
or http://tinyurl.com/y9kmws

A U-bend is made in the middle of the spoke by spreading the jaws of
the vise. The vise tension will easily draw the elbow bend straight:

http://server5.theimagehosting.com/i...rig_165kgf.jpg
or http://tinyurl.com/ycnr2f

After tensioning, the elbow end must be cut off to remove the spoke:

http://server5.theimagehosting.com/i..._elbow_cut.jpg
or http://tinyurl.com/yenfz9

At first, the tensioned spoke was squeezed with pliers to mimic the
typical hand-squeezing. But pliers are so powerful that they left
bends in the spokes that were already at about 165 kgf:

http://server5.theimagehosting.com/i...eeze_bends.jpg
or http://tinyurl.com/tdkf3

It turned out to be much simpler to skip the pliers and just use the
vise and tension gauge to raise spoke tension to the desired levels.

Here are four straight 2mm stainless steel Sapim Leader spokes, each
bent in the vise rig and faintly marked at their ends and one side on
2mm graph paper:

http://i12.tinypic.com/2zgtfs5.jpg

From left to right:
Hand-bent, no tension 0 kgf
Elbow straightened, ~76 kgf / 167 lbs (Park gauge mark 21)
Tensioned more, ~121 kgf / 266 lbs (Park gauge mark 25)
Tensioned like crazy, ~179 kgf / 394 lbs (Park gauge mark 28)

Previous experiments found no wheels where tension rose more than
55~65 lbs when opposite pairs of spokes were squeezed with 60 lbs
of force. The spokes bend impressively, giving rise to mistaken
calculations of huge tension increases, but the rim is merely
bending into a faint N-shape that provides slack for the spokes.

After the U-bends were all heated to an orange glow with a propane
torch, all residual stresses were presumably relieved. The faint marks
on the graph paper were enhanced with a red Sharpie:

http://i11.tinypic.com/2qsc507.jpg

The hand-bent spoke shows a noticeable change. In fact, it had to be
set a little to one side to avoid hitting the next spoke. Obviously,
the residual stresses were relieved by heating.

The other three spokes show no significant change after heating.

The test suggests that practically all residual stress at the spoke
elbow is mechanically relieved long before the spokes are brought to
ordinary tension.

That is, just 76 kgf is enough to eliminate residual stresses in 2mm
stainless steel spokes bent into a U-shape.

I'll try to accomodate any suggestions for other tests.

Anyone who wants to pursue such tests can do so with some spokes (the
longer the better), a propane torch, a vise, bolts to stick in the
vise-jaw-plate holes, wire-cutters, and graph paper. A tension gauge
is nice, but not necessary.

Here's the vise, which needed 8x1.25mm bolts:

http://www.harborfreight.com/cpi/cta...temnumber=5655

It's on sale for about $40, and Harbor Freight often offers discounts
in newspaper ads and fliers.

Here's a nice site with a bad name for printing graph paper pdf's at
any size:

http://incompetech.com/beta/linedGraphPaper/easy.html

Here are long Sapim spokes without nipples, $6.50 per 20:

http://www.biketoolsetc.com/index.cg...id=SA-LSG14304

BikeToolsEtc also sells nipples.

Cheers,

Carl Fogel
post #2 of 26
Thread Starter 

Re: Spoke stress relief test

On Thu, 02 Nov 2006 14:47:20 -0700, carlfogel@comcast.net wrote:

>When bent at the elbow, spokes are expected to have residual stresses.
>
>It's often claimed that squeezing spoke pairs together will relieve
>these residual stresses by raising the spoke tension above normal.
>
>Unfortunately, before and after stresses have never been shown to
>confirm or refute this theory, nor has any formal testing ever been
>published to show whether squeezed spokes outlast unsqueezed spokes.
>Fatigue testing for spokes is incredibly tedious, taking months on
>expensive rigs, so there's little hope in that direction.
>
>A further complication is that squeezing may also improve the wheel
>build in other ways, such as seating spoke heads and nipples,
>releasing spoke wind-up, and simply encouraging care in other steps.
>So even if squeezed spokes do last longer, it may have nothing to do
>with stress relief.
>
>Neutron diffraction images are supposed to reveal stresses, but the
>neutron diffractor at Fogel Labs has been on the blink for a long
>time.
>
>However, Fogel Labs has a propane torch that works just fine. Heat is
>generally agreed to relieve residual stresses, and a propane torch
>will heat a spoke bend to a cheery orange glow in a few seconds,
>relieving any residual stresses. The heated spoke visibly bends to a
>new angle as the stresses are relieved.
>
>Alas, initial testing showed an odd problem when old-fashioned carbon
>spokes were sacrificed on the altar of stress-relief. Bent in the
>middle and toasted, the carbon spokes bent impressively to a new
>angle. But they turned out to bend the wrong way, opposite of the
>direction in which modern stainless steel spokes bend when toasted.
>
>Hmmm . . . the general opinion in materials circles was that the
>carbon steel went through a phase change that overwhelmed the change
>in residual stresses. Stainless steel, on the other hand, is not
>expected to suffer such phase changes, so testing proceeded with
>modern spokes.
>
>The next problem was that factory elbow bends are too tiny, too short,
>and too awkward for any useful testing. If they bent after toasting,
>the angle was too small to be seen, even with camera magnification.
>
>So a spoke-bending rig was cobbled together:
>
>http://server5.theimagehosting.com/i..._spoke_rig.jpg
>or http://tinyurl.com/y9kmws
>
>A U-bend is made in the middle of the spoke by spreading the jaws of
>the vise. The vise tension will easily draw the elbow bend straight:
>
>http://server5.theimagehosting.com/i...rig_165kgf.jpg
>or http://tinyurl.com/ycnr2f
>
>After tensioning, the elbow end must be cut off to remove the spoke:
>
>http://server5.theimagehosting.com/i..._elbow_cut.jpg
>or http://tinyurl.com/yenfz9
>
>At first, the tensioned spoke was squeezed with pliers to mimic the
>typical hand-squeezing. But pliers are so powerful that they left
>bends in the spokes that were already at about 165 kgf:
>
>http://server5.theimagehosting.com/i...eeze_bends.jpg
>or http://tinyurl.com/tdkf3
>
>It turned out to be much simpler to skip the pliers and just use the
>vise and tension gauge to raise spoke tension to the desired levels.
>
>Here are four straight 2mm stainless steel Sapim Leader spokes, each
>bent in the vise rig and faintly marked at their ends and one side on
>2mm graph paper:
>
>http://i12.tinypic.com/2zgtfs5.jpg
>
>From left to right:
> Hand-bent, no tension 0 kgf
> Elbow straightened, ~76 kgf / 167 lbs (Park gauge mark 21)
> Tensioned more, ~121 kgf / 266 lbs (Park gauge mark 25)
> Tensioned like crazy, ~179 kgf / 394 lbs (Park gauge mark 28)
>
>Previous experiments found no wheels where tension rose more than
>55~65 lbs when opposite pairs of spokes were squeezed with 60 lbs
>of force. The spokes bend impressively, giving rise to mistaken
>calculations of huge tension increases, but the rim is merely
>bending into a faint N-shape that provides slack for the spokes.
>
>After the U-bends were all heated to an orange glow with a propane
>torch, all residual stresses were presumably relieved. The faint marks
>on the graph paper were enhanced with a red Sharpie:
>
>http://i11.tinypic.com/2qsc507.jpg
>
>The hand-bent spoke shows a noticeable change. In fact, it had to be
>set a little to one side to avoid hitting the next spoke. Obviously,
>the residual stresses were relieved by heating.
>
>The other three spokes show no significant change after heating.
>
>The test suggests that practically all residual stress at the spoke
>elbow is mechanically relieved long before the spokes are brought to
>ordinary tension.
>
>That is, just 76 kgf is enough to eliminate residual stresses in 2mm
>stainless steel spokes bent into a U-shape.
>
>I'll try to accomodate any suggestions for other tests.
>
>Anyone who wants to pursue such tests can do so with some spokes (the
>longer the better), a propane torch, a vise, bolts to stick in the
>vise-jaw-plate holes, wire-cutters, and graph paper. A tension gauge
>is nice, but not necessary.
>
>Here's the vise, which needed 8x1.25mm bolts:
>
>http://www.harborfreight.com/cpi/cta...temnumber=5655
>
>It's on sale for about $40, and Harbor Freight often offers discounts
>in newspaper ads and fliers.
>
>Here's a nice site with a bad name for printing graph paper pdf's at
>any size:
>
>http://incompetech.com/beta/linedGraphPaper/easy.html
>
>Here are long Sapim spokes without nipples, $6.50 per 20:
>
>http://www.biketoolsetc.com/index.cg...id=SA-LSG14304
>
>BikeToolsEtc also sells nipples.
>
>Cheers,
>
>Carl Fogel


Here are four more spokes bent into a U, photographed, and then heated
to an orange glow with a propane torch.

The yellow tape strips are just a confused marking system.

From left to right:

Hand-bent, no tension 0 kgf
Low tension, ~47 kgf / ~103 lbs (Park mark ~16)
More tension ~62 kgf / ~136 lbs (Park mark ~19)
Even more tension ~79 kgf / ~174 lbs (Park mark ~21)

Low tension is tricky and awkward, so take them with a large grain of
salt.

Notice that the 3rd spoke bent oddly, which may be part of the
v-shaped vise-rig's behavior, which is still straightening the tiny
spoke elbow at that point.

The Park gauge is not intended for such low tensions, much less spokes
that are still faintly curved in the rig, but the tensions were
obviously relatively lower, whatever their absolute values were.

Again, the bend of the first spoke, hand-bent with no tension applied
to relieve residual stresses, expanded visibly when it was heated,
suggesting that residual stresses were being relieved.

The second spoke, with only a hundred pounds of tension applied, also
clearly expanded its bend, though not as much as the first spoke,
suggesting that a hundred pounds of tension had relieved some of its
residual stress.

The oddly bent third spoke's bend, tensioned to around 130~140 pounds,
arguably changed a little. (It's hard to fiddle the spokes into
exactly the same place for measurement, but I think that it shows a
little change).

If so, the third spoke contracted a tiny bit, instead of expanding,
but that could be due to where it ended up in the bending process,
which straightens the initial curve into a fairly sharp vee.

The fourth spoke, with around 175 lbs of tension, also arguably
changed, expanding its bend a little bit. (Again, it's hard to match
the original alignment, and the change is much smaller than the
obvious change in the untensioned, hand-bent spoke. The graph paper is
2mm squares.)

Original unrelieved spokes:
http://i12.tinypic.com/48z7vwx.jpg

Same spokes after heating:
http://i12.tinypic.com/2j42hjt.jpg

These results seem to support the theory that no residual stresses
are relieved on the spoke with no tension, that about half the
residual stresses are relieved at very low tension (~100 pounds), and
that by the time that low bicycle wheel tension is reached (~175 lbs)
the residual stresses have pretty much been eliminated.

But there could be all sorts of other interesting explanations.

I tried to get rid of the pesky elbow bend that the vise-rig
straightens out at around 175 lbs of tension, but it's too small to
grip with pliers strong enough to twist it. Now I'm thinking about
sawing up a hub section and somehow clamping it into place so that the
vise-rig doesn't have to straighten the elbow at all, eliminating it
as a possible point of confusion.

Cheers,

Carl Fogel
post #3 of 26

Re: Spoke stress relief test

Carl Fogel writes:

> Original unrelieved spokes:
> http://i12.tinypic.com/48z7vwx.jpg


> Same spokes after heating:
> http://i12.tinypic.com/2j42hjt.jpg


> These results seem to support the theory that no residual stresses
> are relieved on the spoke with no tension, that about half the
> residual stresses are relieved at very low tension (~100 pounds),
> and that by the time that low bicycle wheel tension is reached (~175
> lbs) the residual stresses have pretty much been eliminated.


> But there could be all sorts of other interesting explanations.


> I tried to get rid of the pesky elbow bend that the vise-rig
> straightens out at around 175 lbs of tension, but it's too small to
> grip with pliers strong enough to twist it. Now I'm thinking about
> sawing up a hub section and somehow clamping it into place so that
> the vise-rig doesn't have to straighten the elbow at all,
> eliminating it as a possible point of confusion.


I think some unexplained changes may be caused by the method of
heating. If this were done in an oven that slowly and uniformly heats
samples to the temperatures of interest, in contrast to one-sided
from outside to inside, I think the experiment might show different
results... and then it might not.

Jobst Brandt
post #4 of 26
Thread Starter 

Re: Spoke stress relief test

On 03 Nov 2006 21:40:27 GMT, jobst.brandt@stanfordalumni.org wrote:

>Carl Fogel writes:
>
>> Original unrelieved spokes:
>> http://i12.tinypic.com/48z7vwx.jpg

>
>> Same spokes after heating:
>> http://i12.tinypic.com/2j42hjt.jpg

>
>> These results seem to support the theory that no residual stresses
>> are relieved on the spoke with no tension, that about half the
>> residual stresses are relieved at very low tension (~100 pounds),
>> and that by the time that low bicycle wheel tension is reached (~175
>> lbs) the residual stresses have pretty much been eliminated.

>
>> But there could be all sorts of other interesting explanations.

>
>> I tried to get rid of the pesky elbow bend that the vise-rig
>> straightens out at around 175 lbs of tension, but it's too small to
>> grip with pliers strong enough to twist it. Now I'm thinking about
>> sawing up a hub section and somehow clamping it into place so that
>> the vise-rig doesn't have to straighten the elbow at all,
>> eliminating it as a possible point of confusion.

>
>I think some unexplained changes may be caused by the method of
>heating. If this were done in an oven that slowly and uniformly heats
>samples to the temperatures of interest, in contrast to one-sided
>from outside to inside, I think the experiment might show different
>results... and then it might not.
>
>Jobst Brandt


Dear Jobst,

Could be, but . . .

Why would the bent but untensioned spoke repeatedly show large,
visible changes in its bend angle, but the bent and increasingly
heavily tensioned spokes refuse to bend, as if their residual stresses
had been effectively relieved?

One possibility has been raised by Dave Van Tol, who wondered if some
phase change could affect things. But if so, the change behaves
differently for the never-tensioned spoke bend as opposed to all the
others.

Another possibility is that all the tensioned spokes are actually
being over-tensioned for a moment or two as the elbow bend
straightens.

That is, it might somehow take 200 kgf of tension to straighten that
pesky little elbow and let the spoke slip into the hole and snug the
spoke head up against the bolt. Then the spoke relaxes back down to
much lower tension.

If so, then all the spokes would be effectively getting wildly high
tension to relieve stress before I get a tension gauge on them. That
would fool me--the spokes that showed 76, 121, and 179 kgf would have
actually all been raised for a second or two to 200 kgf.

But the second set of spokes deliberately included two lightly
tensioned spokes whose elbows were visibly not straightened out. (The
part where the spoke slithers into the hole in the bolt is absurdly
satisfying.)

The lowest tension spoke (around 100 lbs tension) showed clear
movement when heated--not nearly as much movement as the
never-tensioned spoke, but more movement than the faint/possible
movements of the more heavily tensioned spokes. It's the second from
the left in this picture:

http://i12.tinypic.com/2j42hjt.jpg

Lacking an oven that can heat even thin spokes to a glowing orange, I
can only try to hold the torch at a distance and let the heat rise
slowly.

It's an annoying experiment. It seems to support the theory that the
residual stresses are rapidly relieved as tension increases to the low
end of what we use for bicycle wheels, but there are lots of potential
holes in the theory.

I'll see what I can do to clean up the elbow problem, but I haven't
any idea of where to borrow the right kind of oven.

Cheers,

Carl Fogel
post #5 of 26

Re: Spoke stress relief test

Carl Fogel writes:

> The lowest tension spoke (around 100 lbs tension) showed clear
> movement when heated--not nearly as much movement as the
> never-tensioned spoke, but more movement than the faint/possible
> movements of the more heavily tensioned spokes. It's the second
> from the left in this picture:


> http://i12.tinypic.com/2j42hjt.jpg


> Lacking an oven that can heat even thin spokes to a glowing orange,
> I can only try to hold the torch at a distance and let the heat rise
> slowly.


> It's an annoying experiment. It seems to support the theory that
> the residual stresses are rapidly relieved as tension increases to
> the low end of what we use for bicycle wheels, but there are lots of
> potential holes in the theory.


I realize we don't have a test laboratory at hand so lets try a
different approach. Take a spoke in your front wheel and give it,
let's say a 30° over-the-thumb bend at midspan. Re-tension it to the
same tension as the other (tight) spokes.

If that spoke loses its "kink" and becomes a spoke that is straight
when you relax its tension, then you have your proof. If it doesn't,
then tighten it up again and squeeze it in the manner proposed for
stress relieving and see if that changes anything.

You may recall, that when tensile testing spokes, I noticed that a
fairly severely bent spoke became straight as an arrow, so to speak,
after taking it to yield but not significantly beyond.

I think we are getting close. I don't like the heat method because
too many things get involved and it is destructive.

Jobst Brandt
post #6 of 26
Thread Starter 

Re: Spoke stress relief test

On Thu, 02 Nov 2006 14:47:20 -0700, carlfogel@comcast.net wrote:

>When bent at the elbow, spokes are expected to have residual stresses.
>
>It's often claimed that squeezing spoke pairs together will relieve
>these residual stresses by raising the spoke tension above normal.
>
>Unfortunately, before and after stresses have never been shown to
>confirm or refute this theory, nor has any formal testing ever been
>published to show whether squeezed spokes outlast unsqueezed spokes.
>Fatigue testing for spokes is incredibly tedious, taking months on
>expensive rigs, so there's little hope in that direction.
>
>A further complication is that squeezing may also improve the wheel
>build in other ways, such as seating spoke heads and nipples,
>releasing spoke wind-up, and simply encouraging care in other steps.
>So even if squeezed spokes do last longer, it may have nothing to do
>with stress relief.
>
>Neutron diffraction images are supposed to reveal stresses, but the
>neutron diffractor at Fogel Labs has been on the blink for a long
>time.
>
>However, Fogel Labs has a propane torch that works just fine. Heat is
>generally agreed to relieve residual stresses, and a propane torch
>will heat a spoke bend to a cheery orange glow in a few seconds,
>relieving any residual stresses. The heated spoke visibly bends to a
>new angle as the stresses are relieved.
>
>Alas, initial testing showed an odd problem when old-fashioned carbon
>spokes were sacrificed on the altar of stress-relief. Bent in the
>middle and toasted, the carbon spokes bent impressively to a new
>angle. But they turned out to bend the wrong way, opposite of the
>direction in which modern stainless steel spokes bend when toasted.
>
>Hmmm . . . the general opinion in materials circles was that the
>carbon steel went through a phase change that overwhelmed the change
>in residual stresses. Stainless steel, on the other hand, is not
>expected to suffer such phase changes, so testing proceeded with
>modern spokes.
>
>The next problem was that factory elbow bends are too tiny, too short,
>and too awkward for any useful testing. If they bent after toasting,
>the angle was too small to be seen, even with camera magnification.
>
>So a spoke-bending rig was cobbled together:
>
>http://server5.theimagehosting.com/i..._spoke_rig.jpg
>or http://tinyurl.com/y9kmws
>
>A U-bend is made in the middle of the spoke by spreading the jaws of
>the vise. The vise tension will easily draw the elbow bend straight:
>
>http://server5.theimagehosting.com/i...rig_165kgf.jpg
>or http://tinyurl.com/ycnr2f
>
>After tensioning, the elbow end must be cut off to remove the spoke:
>
>http://server5.theimagehosting.com/i..._elbow_cut.jpg
>or http://tinyurl.com/yenfz9
>
>At first, the tensioned spoke was squeezed with pliers to mimic the
>typical hand-squeezing. But pliers are so powerful that they left
>bends in the spokes that were already at about 165 kgf:
>
>http://server5.theimagehosting.com/i...eeze_bends.jpg
>or http://tinyurl.com/tdkf3
>
>It turned out to be much simpler to skip the pliers and just use the
>vise and tension gauge to raise spoke tension to the desired levels.
>
>Here are four straight 2mm stainless steel Sapim Leader spokes, each
>bent in the vise rig and faintly marked at their ends and one side on
>2mm graph paper:
>
>http://i12.tinypic.com/2zgtfs5.jpg
>
>From left to right:
> Hand-bent, no tension 0 kgf
> Elbow straightened, ~76 kgf / 167 lbs (Park gauge mark 21)
> Tensioned more, ~121 kgf / 266 lbs (Park gauge mark 25)
> Tensioned like crazy, ~179 kgf / 394 lbs (Park gauge mark 28)
>
>Previous experiments found no wheels where tension rose more than
>55~65 lbs when opposite pairs of spokes were squeezed with 60 lbs
>of force. The spokes bend impressively, giving rise to mistaken
>calculations of huge tension increases, but the rim is merely
>bending into a faint N-shape that provides slack for the spokes.
>
>After the U-bends were all heated to an orange glow with a propane
>torch, all residual stresses were presumably relieved. The faint marks
>on the graph paper were enhanced with a red Sharpie:
>
>http://i11.tinypic.com/2qsc507.jpg
>
>The hand-bent spoke shows a noticeable change. In fact, it had to be
>set a little to one side to avoid hitting the next spoke. Obviously,
>the residual stresses were relieved by heating.
>
>The other three spokes show no significant change after heating.
>
>The test suggests that practically all residual stress at the spoke
>elbow is mechanically relieved long before the spokes are brought to
>ordinary tension.
>
>That is, just 76 kgf is enough to eliminate residual stresses in 2mm
>stainless steel spokes bent into a U-shape.
>
>I'll try to accomodate any suggestions for other tests.
>
>Anyone who wants to pursue such tests can do so with some spokes (the
>longer the better), a propane torch, a vise, bolts to stick in the
>vise-jaw-plate holes, wire-cutters, and graph paper. A tension gauge
>is nice, but not necessary.
>
>Here's the vise, which needed 8x1.25mm bolts:
>
>http://www.harborfreight.com/cpi/cta...temnumber=5655
>
>It's on sale for about $40, and Harbor Freight often offers discounts
>in newspaper ads and fliers.
>
>Here's a nice site with a bad name for printing graph paper pdf's at
>any size:
>
>http://incompetech.com/beta/linedGraphPaper/easy.html
>
>Here are long Sapim spokes without nipples, $6.50 per 20:
>
>http://www.biketoolsetc.com/index.cg...id=SA-LSG14304
>
>BikeToolsEtc also sells nipples.
>
>Cheers,
>
>Carl Fogel


A small piece of right-angle metal, some crude drilling, and a washer
or two got rid of the elbow-straightening of the original vise-rig.

It didn't seem to make any difference, but it eliminates the
possibility that straightening the elbow involved significant extra
unmeasured tension.

Here are 7 spokes before heating:

http://i11.tinypic.com/33424uc.jpg

Here they are after heating:

http://i12.tinypic.com/29c04sy.jpg

Left to right, they increase in tension and decrease in bend-change
after heating to orange with a propane torch.

Hand bent 0 kgf / no tension large change
Park ~15 ~41 kgf / ~90 lbs small change
Park ~17 ~51 kgf / ~112 lbs a little larger change, odd bend
Park ~19 ~62 kgf / ~136 lbs no visible change
Park ~21 ~76 kgf / ~167 lbs a little change (poorly placed)
Park ~25 ~121 kgf / ~266 lbs a little change
Park ~28 ~179 kgf / ~394 lbs maybe a little change other way?

Again, the largest bend changes upon heating are obviously in spokes
with no tension and very low tension. I tried to heat them slowly, but
waving a propane torch may be no substitute for sitting in an oven
that slowly rises to the final temperature over several hours.

Next time, I plan to try hand-bending all spokes to the same angle
instead of letting the vise do the work of straightening them.

I keep thinking that I see odd bends, which may be the result of the
initial bend being off-center or just the way that the wire ends up at
various tensions. The very high-tension spoke, for example, seems to
be bending into a faint safety-pin shape, with the free end curving
back.

Cheers,

Carl Fogel
post #7 of 26
Thread Starter 

Re: Spoke stress relief test

On 04 Nov 2006 03:47:01 GMT, jobst.brandt@stanfordalumni.org wrote:

>Carl Fogel writes:
>
>> The lowest tension spoke (around 100 lbs tension) showed clear
>> movement when heated--not nearly as much movement as the
>> never-tensioned spoke, but more movement than the faint/possible
>> movements of the more heavily tensioned spokes. It's the second
>> from the left in this picture:

>
>> http://i12.tinypic.com/2j42hjt.jpg

>
>> Lacking an oven that can heat even thin spokes to a glowing orange,
>> I can only try to hold the torch at a distance and let the heat rise
>> slowly.

>
>> It's an annoying experiment. It seems to support the theory that
>> the residual stresses are rapidly relieved as tension increases to
>> the low end of what we use for bicycle wheels, but there are lots of
>> potential holes in the theory.

>
>I realize we don't have a test laboratory at hand so lets try a
>different approach. Take a spoke in your front wheel and give it,
>let's say a 30° over-the-thumb bend at midspan. Re-tension it to the
>same tension as the other (tight) spokes.
>
>If that spoke loses its "kink" and becomes a spoke that is straight
>when you relax its tension, then you have your proof. If it doesn't,
>then tighten it up again and squeeze it in the manner proposed for
>stress relieving and see if that changes anything.
>
>You may recall, that when tensile testing spokes, I noticed that a
>fairly severely bent spoke became straight as an arrow, so to speak,
>after taking it to yield but not significantly beyond.
>
>I think we are getting close. I don't like the heat method because
>too many things get involved and it is destructive.
>
>Jobst Brandt


Dear Jobst,

To be honest, I don't follow your idea.

Heating a loose spoke that's been bent should relieve any residual
tensions.

Pulling on a bent spoke could relieve residual stresses, too, but
it also tries to make the bend yield back the other way. This approach
seems to be guaranteed to confuse stress relief with re-bending, not
simplify things.

In any case, the bend won't go arrow-straight at any reasonable
tension because the closer it gets to straight, the more the leverage
is reduced.

At least, that was my experience with lots of bends in spoke mid-spans
caused by weights hanging from ropes.

Maybe at the yield point for the whole spoke, far above the tensions
that we're talking about, things would go straight as the spoke
failed.

But that's still yielding in a bend and bending back the other way,
not relieving residual stresses in a loose spoke.

Anyway, here's what you seemed to be asking for. Maybe I've
misunderstood things. The spoke is loose in the vise-rig with an
angle-piece that prevents the elbow from straightening. I bent a
roughly 30-degree bend on one side:

http://i11.tinypic.com/2zfiycg.jpg

Here's the same spoke, which went roughly straight when tension
reached Park mark 24, roughly 107 kgf, about 235 lbs:

http://i11.tinypic.com/4dbwfmc.jpg

Here's the bend reappearing when tension is released:

http://i11.tinypic.com/317b4pe.jpg

Here's the bend still reappearing after Park mark 28 tension
is released:

http://i12.tinypic.com/2hn5p1e.jpg

At Park mark 28, the bend is surviving roughly 179 kgf of tension,
about 394 lbs. It has yielded considerably back the other way, but
I don't understand what the yielding shows about residual stress
relief.

Maybe you can describe what you're looking for in some other way,
and I can give it a try.

Cheers,

Carl Fogel
post #8 of 26

Re: Spoke stress relief test

In article <eih8me$k3g$1@cascade.cs.ubc.ca>,
gillies@cs.ubc.ca (Donald Gillies) wrote:

> I think the name "stress relieving" is a misnomer - who thought up
> this bogus name ?? I found it important to perform this process
> whenever I trued up a wheel, because :


Stress relief is entirely accurate. There are residual
stresses in spokes from the cold forming at the
factory. Some of those stresses when added to the
cyclic stress of passing through the contact patch
fatigue the metal, and the metal forms cracks then
breaks.

>
> When riding a new wheel, there is frequent "TWANGING" from the spokes
> as they change position. During the truing process, imho, not all the
> spokes reach their lowest energy state, due to friction (stiction) to
> other spokes, and a tendency for nipples and/or spoke heads to change
> seating angles as they are tightened. Spokes often do not reach
> optimal seating angles because of they must overcome the coefficient
> of friction to slide, the amount they must change position is too
> small to ever overcome the coefficient.
>
> I found in the past that I can either stress-relieve the spokes on the
> truing stand, or I can make SEVERAL returns to the truing stand,
> mounting and un-mounting the wheel each time on the bike, because
> the spokes change position during short rides around the block, with
> an associated "TWANG" sound.


The twang sound is spoke twist un-twisting. When
turning the spoke nipple turn it until the nipple
actually turns on the spoke threads, then turn the
nipple back to remove the spoke twist. Your wheels will
not twang anymore.

--
Michael Press
post #9 of 26

Re: Spoke stress relief test

Donald Gillies writes:

> I think the name "stress relieving" is a misnomer - who thought up
> this bogus name ?? I found it important to perform this process
> whenever I trued up a wheel, because :


> When riding a new wheel, there is frequent "TWANGING" from the
> spokes as they change position. During the truing process, IMHO,
> not all the spokes reach their lowest energy state, due to friction
> (stiction) to other spokes, and a tendency for nipples and/or spoke
> heads to change seating angles as they are tightened. Spokes often
> do not reach optimal seating angles because of they must overcome
> the coefficient of friction to slide, the amount they must change
> position is too small to ever overcome the coefficient.


Any sounds from a newly built wheel are indications of poor
workmanship because that means that spokes with twist (windup) were
left after adjusting spokes, something that is easily prevented. The
sounds are spokes turning in their threads to relax the twist and this
changes spoke adjustment and wheel alignment.

The sound does not come from anywhere else and cannot make the sharp
sounds that are made when wheel load induced spoke slackening allows
spokes to un-twist.

> I found in the past that I can either stress-relieve the spokes on
> the truing stand, or I can make SEVERAL returns to the truing stand,
> mounting and un-mounting the wheel each time on the bike, because
> the spokes change position during short rides around the block, with
> an associated "TWANG" sound.


I see you do not have "the Bicycle Wheel" in which this process is
explained and would have spared you the repeated truing you mention.
This is not stress relief in the sense of reducing fatigue spoke
failures.

> Therefore, I do this to get a tighter, more uniform, and straighter
> wheel, with only 1 session in the truing stand.


Just untwist spokes with the spoke wrench. It is an easy overshoot
and back twist motion that comes naturally with practice.

> I sometimes will lay the wheel on its side after truing and
> purposefully flex the rim, side-to-side, which also induces the
> pinging sound as spokes suddenly find a lower potential energy
> level...


"Potential energy" puts an odd pseudo-science spin to your problem. I
think this has been discussed here at length over the years. I'm sure
you could find it in the archives.

> I admire Carl's scientific approach to investigate stress relief of
> spokes, and I learned from this process. However, I think further
> testing to reveal the coefficient of friction between adjacent
> spokes, and between the nipple and the spoke socket, and between the
> spoke head and the hub hole, would add valuable knowledge.


Friction is an entirely new aspect in spoked wheels. What do you
perceive as a problem of friction in a wheel?

Jobst Brandt
post #10 of 26

Re: Spoke stress relief test

Carl Fogel writes:

>>> The lowest tension spoke (around 100 lbs tension) showed clear
>>> movement when heated--not nearly as much movement as the
>>> never-tensioned spoke, but more movement than the faint/possible
>>> movements of the more heavily tensioned spokes. It's the second
>>> from the left in this picture:


http://i12.tinypic.com/2j42hjt.jpg

>>> Lacking an oven that can heat even thin spokes to a glowing
>>> orange, I can only try to hold the torch at a distance and let the
>>> heat rise slowly.


>>> It's an annoying experiment. It seems to support the theory that
>>> the residual stresses are rapidly relieved as tension increases to
>>> the low end of what we use for bicycle wheels, but there are lots
>>> of potential holes in the theory.


>> I realize we don't have a test laboratory at hand so lets try a
>> different approach. Take a spoke in your front wheel and give it,
>> let's say a 30° over-the-thumb bend at midspan. Re-tension it to
>> the same tension as the other (tight) spokes.


>> If that spoke loses its "kink" and becomes a spoke that is straight
>> when you relax its tension, then you have your proof. If it
>> doesn't, then tighten it up again and squeeze it in the manner
>> proposed for stress relieving and see if that changes anything.


>> You may recall, that when tensile testing spokes, I noticed that a
>> fairly severely bent spoke became straight as an arrow, so to
>> speak, after taking it to yield but not significantly beyond.


>> I think we are getting close. I don't like the heat method because
>> too many things get involved and it is destructive.


> To be honest, I don't follow your idea.


> Heating a loose spoke that's been bent should relieve any residual
> tensions.


As I mentioned, flame heating relaxes residual stress from the outside
to inside and can obscure what stresses were present. That is why I
mentioned slower oven heating, which is not in the realm of this
laboratory.

> Pulling on a bent spoke could relieve residual stresses, too, but it
> also tries to make the bend yield back the other way. This approach
> seems to be guaranteed to confuse stress relief with re-bending, not
> simplify things.


If the spoke is straight after this process, it is stress relieved and
the desired result is achieved. What is re-bending if it isn't stress
relief?

Consider that curved wire coming from a reel is run through a series
of zig-zag rollers of diminishing excursion to straighten and stress
relieve it in X and Y before it is cut to length and formed into
spokes.

I think if you consider the spoke is straighter when slackened after
tensioning that it must have less residual stress.

> In any case, the bend won't go arrow-straight at any reasonable
> tension because the closer it gets to straight, the more the
> leverage is reduced.


Ah yes, but how straight the spoke becomes is an indication of the
degree of stress relief. Its bend (when slack) indicates residual
stress when held in its tensioned and straight position.

> At least, that was my experience with lots of bends in spoke
> mid-spans caused by weights hanging from ropes.


> Maybe at the yield point for the whole spoke, far above the tensions
> that we're talking about, things would go straight as the spoke
> failed.


Well? That's the problem. The closer to yield the spoke is taken,
the less residual stress remains.

> But that's still yielding in a bend and bending back the other way,
> not relieving residual stresses in a loose spoke.


This does not constitute bending the other way on a loose spoke and
is a prime example of residual stress in a spoke in place in a wheel.

> Anyway, here's what you seemed to be asking for. Maybe I've
> misunderstood things. The spoke is loose in the vise-rig with an
> angle-piece that prevents the elbow from straightening. I bent a
> roughly 30-degree bend on one side:


http://i11.tinypic.com/2zfiycg.jpg

> Here's the same spoke, which went roughly straight when tension
> reached Park mark 24, roughly 107 kgf, about 235 lbs:


http://i11.tinypic.com/4dbwfmc.jpg

> Here's the bend reappearing when tension is released:


http://i11.tinypic.com/317b4pe.jpg

> Here's the bend still reappearing after Park mark 28 tension is
> released:


http://i12.tinypic.com/2hn5p1e.jpg

> At Park mark 28, the bend is surviving roughly 179 kgf of tension,
> about 394 lbs. It has yielded considerably back the other way, but
> I don't understand what the yielding shows about residual stress
> relief.


A wire that has yielded in tension to the point where it has no
spring-back can have no residual stress. That is the only way
mechanical stress relief can have an effect. It must yield the local
high stress concentrations that in a spoke are entirely tensile, the
compressive ones being significantly reduced by spoke tension.

You see that the 30° bend is practically straight indicating that the
bend has been reduced by relaxing the stress on the inside of the bend
(the tensile portion).

> Maybe you can describe what you're looking for in some other way,
> and I can give it a try.


The reason I suggested doing this in a wheel is that the effect of
tensioning would be apparent and the subsequent additional manual
stress relief.

I tried this on a wheel and found that tightening the bent spoke
reduced its smooth bend of about 30° to a small portion (about 3/8"
from a straight line) of its original shape and that subsequent
squeezing this spoke together with its parallel kin reduced that
displacement by about half.

From this, I deduce that the remaining stress in the spoke is
primarily compression and that no manner of manual stress relief can
or needs to affect them.

Jobst Brandt
post #11 of 26
Thread Starter 

Re: Spoke stress relief test

On 04 Nov 2006 17:05:48 GMT, jobst.brandt@stanfordalumni.org wrote:

>Carl Fogel writes:
>
>>>> The lowest tension spoke (around 100 lbs tension) showed clear
>>>> movement when heated--not nearly as much movement as the
>>>> never-tensioned spoke, but more movement than the faint/possible
>>>> movements of the more heavily tensioned spokes. It's the second
>>>> from the left in this picture:

>
> http://i12.tinypic.com/2j42hjt.jpg
>
>>>> Lacking an oven that can heat even thin spokes to a glowing
>>>> orange, I can only try to hold the torch at a distance and let the
>>>> heat rise slowly.

>
>>>> It's an annoying experiment. It seems to support the theory that
>>>> the residual stresses are rapidly relieved as tension increases to
>>>> the low end of what we use for bicycle wheels, but there are lots
>>>> of potential holes in the theory.

>
>>> I realize we don't have a test laboratory at hand so lets try a
>>> different approach. Take a spoke in your front wheel and give it,
>>> let's say a 30° over-the-thumb bend at midspan. Re-tension it to
>>> the same tension as the other (tight) spokes.

>
>>> If that spoke loses its "kink" and becomes a spoke that is straight
>>> when you relax its tension, then you have your proof. If it
>>> doesn't, then tighten it up again and squeeze it in the manner
>>> proposed for stress relieving and see if that changes anything.

>
>>> You may recall, that when tensile testing spokes, I noticed that a
>>> fairly severely bent spoke became straight as an arrow, so to
>>> speak, after taking it to yield but not significantly beyond.

>
>>> I think we are getting close. I don't like the heat method because
>>> too many things get involved and it is destructive.

>
>> To be honest, I don't follow your idea.

>
>> Heating a loose spoke that's been bent should relieve any residual
>> tensions.

>
>As I mentioned, flame heating relaxes residual stress from the outside
>to inside and can obscure what stresses were present. That is why I
>mentioned slower oven heating, which is not in the realm of this
>laboratory.
>
>> Pulling on a bent spoke could relieve residual stresses, too, but it
>> also tries to make the bend yield back the other way. This approach
>> seems to be guaranteed to confuse stress relief with re-bending, not
>> simplify things.

>
>If the spoke is straight after this process, it is stress relieved and
>the desired result is achieved. What is re-bending if it isn't stress
>relief?
>
>Consider that curved wire coming from a reel is run through a series
>of zig-zag rollers of diminishing excursion to straighten and stress
>relieve it in X and Y before it is cut to length and formed into
>spokes.
>
>I think if you consider the spoke is straighter when slackened after
>tensioning that it must have less residual stress.
>
>> In any case, the bend won't go arrow-straight at any reasonable
>> tension because the closer it gets to straight, the more the
>> leverage is reduced.

>
>Ah yes, but how straight the spoke becomes is an indication of the
>degree of stress relief. Its bend (when slack) indicates residual
>stress when held in its tensioned and straight position.
>
>> At least, that was my experience with lots of bends in spoke
>> mid-spans caused by weights hanging from ropes.

>
>> Maybe at the yield point for the whole spoke, far above the tensions
>> that we're talking about, things would go straight as the spoke
>> failed.

>
>Well? That's the problem. The closer to yield the spoke is taken,
>the less residual stress remains.
>
>> But that's still yielding in a bend and bending back the other way,
>> not relieving residual stresses in a loose spoke.

>
>This does not constitute bending the other way on a loose spoke and
>is a prime example of residual stress in a spoke in place in a wheel.
>
>> Anyway, here's what you seemed to be asking for. Maybe I've
>> misunderstood things. The spoke is loose in the vise-rig with an
>> angle-piece that prevents the elbow from straightening. I bent a
>> roughly 30-degree bend on one side:

>
> http://i11.tinypic.com/2zfiycg.jpg
>
>> Here's the same spoke, which went roughly straight when tension
>> reached Park mark 24, roughly 107 kgf, about 235 lbs:

>
> http://i11.tinypic.com/4dbwfmc.jpg
>
>> Here's the bend reappearing when tension is released:

>
> http://i11.tinypic.com/317b4pe.jpg
>
>> Here's the bend still reappearing after Park mark 28 tension is
>> released:

>
> http://i12.tinypic.com/2hn5p1e.jpg
>
>> At Park mark 28, the bend is surviving roughly 179 kgf of tension,
>> about 394 lbs. It has yielded considerably back the other way, but
>> I don't understand what the yielding shows about residual stress
>> relief.

>
>A wire that has yielded in tension to the point where it has no
>spring-back can have no residual stress. That is the only way
>mechanical stress relief can have an effect. It must yield the local
>high stress concentrations that in a spoke are entirely tensile, the
>compressive ones being significantly reduced by spoke tension.
>
>You see that the 30° bend is practically straight indicating that the
>bend has been reduced by relaxing the stress on the inside of the bend
>(the tensile portion).
>
>> Maybe you can describe what you're looking for in some other way,
>> and I can give it a try.

>
>The reason I suggested doing this in a wheel is that the effect of
>tensioning would be apparent and the subsequent additional manual
>stress relief.
>
>I tried this on a wheel and found that tightening the bent spoke
>reduced its smooth bend of about 30° to a small portion (about 3/8"
>from a straight line) of its original shape and that subsequent
>squeezing this spoke together with its parallel kin reduced that
>displacement by about half.
>
>From this, I deduce that the remaining stress in the spoke is
>primarily compression and that no manner of manual stress relief can
>or needs to affect them.
>
>Jobst Brandt


Dear Jobst,

What I see is a spoke that has been bent one way being bent back the
other way.

The main process seems to be yielding under enormous tension, far
greater than can be achieved by spoke squeezing, which I've repeatedly
measured as about 60 lbs of tension rise on actual wheels for a 60 lb
squeeze force.

Park mark 24 is roughly 107 kgf / 235 lbs and roughly what most wheels
are tensioned to.

Park mark 28 is roughly 179 kgf / 394 lbs, about 160 lbs more tension.

If you hire a gorilla to squeeze two spoke pairs hard enough in the
middle to raise tension 160 lbs on a 32-spoke wheel, the spokes will
probably touch in the middle and will certainly retain impressive,
visible bends.

I know because I hung weights from spoke midspans from 0 to 100 pounds
in 5-lb increments. Tension rose only 90 pounds with a 100-lb squeeze
force, from an initial 250-lb spoke tension to 340 lbs, because thhe
rim invariably distorts from the inevitably unbalanced loads of
staggered spokes into a Z or N shape. The midspan gap between the
285mm exposed spoke sections narrowed from 82 to 47 mm:

http://home.comcast.net/~carlfogel/d...d/newspok2.jpg

A similar tension rise of about 100 lbs for a 100 lb squeeze occurred
when initital tension was lower, starting around 150 lbs on one spoke
and around 180 lbs on the other spoke of the pair, with tensions
converging and rising about 80 / 110 pounds to around 260 lbs with 100
lbs of squeeze force:

http://home.comcast.net/~carlfogel/d...poke_graph.jpg

Again, the result was permanently bent spokes that had yielded in the
midspans. The typical poster on RBT will be lucky to produce a 60 lb
squeeze force with each hand, doing 4 spokes 8 or 9 times around a 32
or 36 spoke wheel.

Cheers,

Carl Fogel
post #12 of 26

Re: Spoke stress relief test

Luns Tee wrote:
> Recall that a bent spoke has four layers of residual stress,
> alternating in compression and tension. For a spoke that's been bent
> slightly, only a thin surface layer yields. The stress profile would
> be something like:
>
> CtttcccT
>
> The yielded layers of CT hold the bend in place, while the tttccc
> layers inside (some yielded material, and all the unyielded) try to
> straighten out.
>
> For a more severely bent wire, the yielded layers extend deeper, and
> the residual stress pattern becomes more like:
>
> cccTCttt


I was thinking about that a few days ago... What if one were able to
remove the outer layer of the bend *without* overheating the entire
structure? Would the resulting change in the bend be a better indicator
of the residual stress? For example, maybe you could remove some of the
c's or C's from outside of the bend, causing the angle to open up.

As I mentioned in another post, I don't know enough about the steel used
for spokes to predict any sort of phase transformations at the
temperatures produced by Carl's torch. I was trying to think of another
way to test for residual stress without heating the rest of the bend
significantly. Feel free to point out the flaws in this experiment.

--
Dave
dvt at psu dot edu

Everyone confesses that exertion which brings out all the powers of body
and mind is the best thing for us; but most people do all they can to
get rid of it, and as a general rule nobody does much more than
circumstances drive them to do. -Harriet Beecher Stowe, abolitionist and
novelist (1811-1896)
post #13 of 26
Thread Starter 

Re: Spoke stress relief test

On Thu, 02 Nov 2006 14:47:20 -0700, carlfogel@comcast.net wrote:

>When bent at the elbow, spokes are expected to have residual stresses.
>
>It's often claimed that squeezing spoke pairs together will relieve
>these residual stresses by raising the spoke tension above normal.
>
>Unfortunately, before and after stresses have never been shown to
>confirm or refute this theory, nor has any formal testing ever been
>published to show whether squeezed spokes outlast unsqueezed spokes.
>Fatigue testing for spokes is incredibly tedious, taking months on
>expensive rigs, so there's little hope in that direction.
>
>A further complication is that squeezing may also improve the wheel
>build in other ways, such as seating spoke heads and nipples,
>releasing spoke wind-up, and simply encouraging care in other steps.
>So even if squeezed spokes do last longer, it may have nothing to do
>with stress relief.
>
>Neutron diffraction images are supposed to reveal stresses, but the
>neutron diffractor at Fogel Labs has been on the blink for a long
>time.
>
>However, Fogel Labs has a propane torch that works just fine. Heat is
>generally agreed to relieve residual stresses, and a propane torch
>will heat a spoke bend to a cheery orange glow in a few seconds,
>relieving any residual stresses. The heated spoke visibly bends to a
>new angle as the stresses are relieved.
>
>Alas, initial testing showed an odd problem when old-fashioned carbon
>spokes were sacrificed on the altar of stress-relief. Bent in the
>middle and toasted, the carbon spokes bent impressively to a new
>angle. But they turned out to bend the wrong way, opposite of the
>direction in which modern stainless steel spokes bend when toasted.
>
>Hmmm . . . the general opinion in materials circles was that the
>carbon steel went through a phase change that overwhelmed the change
>in residual stresses. Stainless steel, on the other hand, is not
>expected to suffer such phase changes, so testing proceeded with
>modern spokes.
>
>The next problem was that factory elbow bends are too tiny, too short,
>and too awkward for any useful testing. If they bent after toasting,
>the angle was too small to be seen, even with camera magnification.
>
>So a spoke-bending rig was cobbled together:
>
>http://server5.theimagehosting.com/i..._spoke_rig.jpg
>or http://tinyurl.com/y9kmws
>
>A U-bend is made in the middle of the spoke by spreading the jaws of
>the vise. The vise tension will easily draw the elbow bend straight:
>
>http://server5.theimagehosting.com/i...rig_165kgf.jpg
>or http://tinyurl.com/ycnr2f
>
>After tensioning, the elbow end must be cut off to remove the spoke:
>
>http://server5.theimagehosting.com/i..._elbow_cut.jpg
>or http://tinyurl.com/yenfz9
>
>At first, the tensioned spoke was squeezed with pliers to mimic the
>typical hand-squeezing. But pliers are so powerful that they left
>bends in the spokes that were already at about 165 kgf:
>
>http://server5.theimagehosting.com/i...eeze_bends.jpg
>or http://tinyurl.com/tdkf3
>
>It turned out to be much simpler to skip the pliers and just use the
>vise and tension gauge to raise spoke tension to the desired levels.
>
>Here are four straight 2mm stainless steel Sapim Leader spokes, each
>bent in the vise rig and faintly marked at their ends and one side on
>2mm graph paper:
>
>http://i12.tinypic.com/2zgtfs5.jpg
>
>From left to right:
> Hand-bent, no tension 0 kgf
> Elbow straightened, ~76 kgf / 167 lbs (Park gauge mark 21)
> Tensioned more, ~121 kgf / 266 lbs (Park gauge mark 25)
> Tensioned like crazy, ~179 kgf / 394 lbs (Park gauge mark 28)
>
>Previous experiments found no wheels where tension rose more than
>55~65 lbs when opposite pairs of spokes were squeezed with 60 lbs
>of force. The spokes bend impressively, giving rise to mistaken
>calculations of huge tension increases, but the rim is merely
>bending into a faint N-shape that provides slack for the spokes.
>
>After the U-bends were all heated to an orange glow with a propane
>torch, all residual stresses were presumably relieved. The faint marks
>on the graph paper were enhanced with a red Sharpie:
>
>http://i11.tinypic.com/2qsc507.jpg
>
>The hand-bent spoke shows a noticeable change. In fact, it had to be
>set a little to one side to avoid hitting the next spoke. Obviously,
>the residual stresses were relieved by heating.
>
>The other three spokes show no significant change after heating.
>
>The test suggests that practically all residual stress at the spoke
>elbow is mechanically relieved long before the spokes are brought to
>ordinary tension.
>
>That is, just 76 kgf is enough to eliminate residual stresses in 2mm
>stainless steel spokes bent into a U-shape.
>
>I'll try to accomodate any suggestions for other tests.
>
>Anyone who wants to pursue such tests can do so with some spokes (the
>longer the better), a propane torch, a vise, bolts to stick in the
>vise-jaw-plate holes, wire-cutters, and graph paper. A tension gauge
>is nice, but not necessary.
>
>Here's the vise, which needed 8x1.25mm bolts:
>
>http://www.harborfreight.com/cpi/cta...temnumber=5655
>
>It's on sale for about $40, and Harbor Freight often offers discounts
>in newspaper ads and fliers.
>
>Here's a nice site with a bad name for printing graph paper pdf's at
>any size:
>
>http://incompetech.com/beta/linedGraphPaper/easy.html
>
>Here are long Sapim spokes without nipples, $6.50 per 20:
>
>http://www.biketoolsetc.com/index.cg...id=SA-LSG14304
>
>BikeToolsEtc also sells nipples.
>
>Cheers,
>
>Carl Fogel


Luns Tee wondered about the effects of bending the spoke around a
tighter-radius bolt.

In the middle is a used bending bolt, thinned with chainsaw file to
about 0.155 inches wide. It's flanked by new 8x1.25 bolt, threads
about 0.310 inches, and by the previously used 8x1.125 bending bolt,
whose threads are mashed down to about 0.280 inches:

http://i14.tinypic.com/2vkmiwp.jpg

Here are four unrelieved 2mm stainless steel spokes with the tighter
bend radius:

http://i14.tinypic.com/2dchi10.jpg

From left to right
No tension, 0 kgf, over-bent by hand and a bit twisted
Tensioned, Park mark ~17, ~51 kgf, ~112 lbs
Tensioned, Park mark ~24, ~107 kgf, ~235 lbs
Tensioned, Park mark ~28, ~179 kgf, ~394 lbs

The same four spokes after bends were heated orange:

http://i13.tinypic.com/44rg4kg.jpg

From left to right:

moved tension
~7 mm 0 lbs no relief from stretching
~7 mm ~112 lbs very low
~2 mm ~235 lbs bike wheel
xx xx ~300 lbs effect of a ~60-lb spoke-squeeze
~2 mm ~394 lbs much higher than spoke squeeze

The camera angle slightly exaggerates the apparent expansion on
rightmost spoke, while slightly minimizing it on leftmost spoke.

Again, an 80% increase over high normal ~235-lb spoke tension to ~394
lbs made no apparent difference in how much the spoke bend opened when
freed and heated orange.

Hand-squeezing the same ~235-lb spoke would be unlikely to raise the
tension to over 300 lbs:

http://home.comcast.net/~carlfogel/d...d/newspok2.jpg

The graph shows the tension rise for a pair of ~250-lb tension spokes
as they're squeezed together with 0-100 lbs of force at 5-lb
increments. Count over from 0 to 60, and you'll be at around 310 lbs,
a 60-lb rise. The impressive bend angle of the squeezed spoke makes
people think that the tension must be rising enormously, but what
actually happens is that the rim distorts into an N or Z shape as the
two pairs of spokes are squeezed, providing the slack needed for the
spoke to bend outrageously. In lengthy tests, no wheels were found to
provide measured tension increases significantly different than the
squeeze force.

Anyway, despite the smaller bend radius, the effects of increasing
tension appear to occur well below normal bike wheel tension.

Dave Van Tol has wondered about the effect of the heating on stainless
steel and whether a phase change might affect the bend change. This
seems unlikely, since ordinary tension seems to kill most of the
effect of the same glowing orange heat.

Luns Tee has wondered about the effect of a tighter radius bend,
relative to the spoke thickness. The tighter bend in this test didn't
seem to change things.

Jobst Brandt has wondered about the effect of slower heating in an
oven. Slower heating with the propane torch hasn't changed things, and
it's hard to see how the rate of heating would matter when the whole
bend reaches the same glowing orange.

However, a time factor could be involved. Possibly keeping the bends
heated for hours would release more residual stresses

This raises another question--does it matter how long a spoke's
tension is raised?

The vise rig not only allows raising spoke tension much higher than
hand spoke-squeezing, but also involves leaving the spoke at tension
much longer, so the tests are if anything more likely to relieve
tension.

For large pressure vessels, Peter Cole has quoted a standard of two
hours at tension per inch of metal for mechanical stress relief:

http://groups.google.com/group/rec.b...ed271ca9254270

This works out to about 11-12 minutes for a 2 mm thick spoke, but I
suspect that the large-pressure-vessel standard doesn't scale down
linearly.

Still, most hand spoke squeezing will be several orders of magnitude
less than 600 seconds, only a second or so.

Cheers,

Carl Fogel
post #14 of 26

Re: Spoke stress relief test

dvt wrote:
>
>
> I was thinking about that a few days ago... What if one were able to
> remove the outer layer of the bend *without* overheating the entire
> structure? Would the resulting change in the bend be a better indicator
> of the residual stress? For example, maybe you could remove some of the
> c's or C's from outside of the bend, causing the angle to open up.


Deformation as a result of removing some material does indicate
residual stress. I'm aware of two situations where this is important.
First, machinists know that metal parts with residual stresses can
distort when, say, a surface is milled or ground. The stress balance
(so to speak) is changed, and the part warps.

Second, there are (or used to be?) resistance strain gages designed to
measure residual stresses, but in a semi-destructive way. IIRC, the
strain gages sort of encircled the area to which they were applied.
After gluing down the gages and balancing, a hole would be drilled and
new readings taken. The resulting strain could be used to determine
the amount of residual stress.

The problem with spokes, of course, is that they are so tiny. But
perhaps a fixture could be constructed to grip the spoke firmly at the
head and just beyond, leaving the curve and long blade free to move.
Then a small amount of the bend's outside could be removed using a
Dremel, being careful to keep everything cool. Assuming the outside of
the bend had compressive residual stresses, the spoke would be seen to
straighten a bit. This could be measureable at the free end, using
Carl's graph paper.

It shouldn't be hard to come up with a way to grip the head properly to
allow this test. Then the amount of motion (if any) of "stress
relieved" spokes could be compared with virgin spokes.

- Frank Krygowski
post #15 of 26
Thread Starter 

Re: Spoke stress relief test

On Thu, 02 Nov 2006 14:47:20 -0700, carlfogel@comcast.net wrote:

>When bent at the elbow, spokes are expected to have residual stresses.
>
>It's often claimed that squeezing spoke pairs together will relieve
>these residual stresses by raising the spoke tension above normal.
>
>Unfortunately, before and after stresses have never been shown to
>confirm or refute this theory, nor has any formal testing ever been
>published to show whether squeezed spokes outlast unsqueezed spokes.
>Fatigue testing for spokes is incredibly tedious, taking months on
>expensive rigs, so there's little hope in that direction.
>
>A further complication is that squeezing may also improve the wheel
>build in other ways, such as seating spoke heads and nipples,
>releasing spoke wind-up, and simply encouraging care in other steps.
>So even if squeezed spokes do last longer, it may have nothing to do
>with stress relief.
>
>Neutron diffraction images are supposed to reveal stresses, but the
>neutron diffractor at Fogel Labs has been on the blink for a long
>time.
>
>However, Fogel Labs has a propane torch that works just fine. Heat is
>generally agreed to relieve residual stresses, and a propane torch
>will heat a spoke bend to a cheery orange glow in a few seconds,
>relieving any residual stresses. The heated spoke visibly bends to a
>new angle as the stresses are relieved.
>
>Alas, initial testing showed an odd problem when old-fashioned carbon
>spokes were sacrificed on the altar of stress-relief. Bent in the
>middle and toasted, the carbon spokes bent impressively to a new
>angle. But they turned out to bend the wrong way, opposite of the
>direction in which modern stainless steel spokes bend when toasted.
>
>Hmmm . . . the general opinion in materials circles was that the
>carbon steel went through a phase change that overwhelmed the change
>in residual stresses. Stainless steel, on the other hand, is not
>expected to suffer such phase changes, so testing proceeded with
>modern spokes.
>
>The next problem was that factory elbow bends are too tiny, too short,
>and too awkward for any useful testing. If they bent after toasting,
>the angle was too small to be seen, even with camera magnification.
>
>So a spoke-bending rig was cobbled together:
>
>http://server5.theimagehosting.com/i..._spoke_rig.jpg
>or http://tinyurl.com/y9kmws
>
>A U-bend is made in the middle of the spoke by spreading the jaws of
>the vise. The vise tension will easily draw the elbow bend straight:
>
>http://server5.theimagehosting.com/i...rig_165kgf.jpg
>or http://tinyurl.com/ycnr2f
>
>After tensioning, the elbow end must be cut off to remove the spoke:
>
>http://server5.theimagehosting.com/i..._elbow_cut.jpg
>or http://tinyurl.com/yenfz9
>
>At first, the tensioned spoke was squeezed with pliers to mimic the
>typical hand-squeezing. But pliers are so powerful that they left
>bends in the spokes that were already at about 165 kgf:
>
>http://server5.theimagehosting.com/i...eeze_bends.jpg
>or http://tinyurl.com/tdkf3
>
>It turned out to be much simpler to skip the pliers and just use the
>vise and tension gauge to raise spoke tension to the desired levels.
>
>Here are four straight 2mm stainless steel Sapim Leader spokes, each
>bent in the vise rig and faintly marked at their ends and one side on
>2mm graph paper:
>
>http://i12.tinypic.com/2zgtfs5.jpg
>
>From left to right:
> Hand-bent, no tension 0 kgf
> Elbow straightened, ~76 kgf / 167 lbs (Park gauge mark 21)
> Tensioned more, ~121 kgf / 266 lbs (Park gauge mark 25)
> Tensioned like crazy, ~179 kgf / 394 lbs (Park gauge mark 28)
>
>Previous experiments found no wheels where tension rose more than
>55~65 lbs when opposite pairs of spokes were squeezed with 60 lbs
>of force. The spokes bend impressively, giving rise to mistaken
>calculations of huge tension increases, but the rim is merely
>bending into a faint N-shape that provides slack for the spokes.
>
>After the U-bends were all heated to an orange glow with a propane
>torch, all residual stresses were presumably relieved. The faint marks
>on the graph paper were enhanced with a red Sharpie:
>
>http://i11.tinypic.com/2qsc507.jpg
>
>The hand-bent spoke shows a noticeable change. In fact, it had to be
>set a little to one side to avoid hitting the next spoke. Obviously,
>the residual stresses were relieved by heating.
>
>The other three spokes show no significant change after heating.
>
>The test suggests that practically all residual stress at the spoke
>elbow is mechanically relieved long before the spokes are brought to
>ordinary tension.
>
>That is, just 76 kgf is enough to eliminate residual stresses in 2mm
>stainless steel spokes bent into a U-shape.
>
>I'll try to accomodate any suggestions for other tests.
>
>Anyone who wants to pursue such tests can do so with some spokes (the
>longer the better), a propane torch, a vise, bolts to stick in the
>vise-jaw-plate holes, wire-cutters, and graph paper. A tension gauge
>is nice, but not necessary.
>
>Here's the vise, which needed 8x1.25mm bolts:
>
>http://www.harborfreight.com/cpi/cta...temnumber=5655
>
>It's on sale for about $40, and Harbor Freight often offers discounts
>in newspaper ads and fliers.
>
>Here's a nice site with a bad name for printing graph paper pdf's at
>any size:
>
>http://incompetech.com/beta/linedGraphPaper/easy.html
>
>Here are long Sapim spokes without nipples, $6.50 per 20:
>
>http://www.biketoolsetc.com/index.cg...id=SA-LSG14304
>
>BikeToolsEtc also sells nipples.
>
>Cheers,
>
>Carl Fogel


An email asked me about the effect of heating unused spokes.

I went through a dozen 298 mm Sapim 14 gauge spokes from the same bag,
laying them against straight steel rulers, and found four whose
threaded ends curved very slightly away from the ruler when the elbow
ends were hooked over the end of the ruler:

elbow
ends
______________________threaded end curves up
|
| XXXXXXXXXXXXXXXXXXXXXXX
-------ruler----------
| XXXXXXXXXXXXXXXXXXXXXXX
|______________________threaded end curves down


After marking them against 1 mm graph paper, I hung them one by one
from the threaded end and slowly heated them from end to end to
glowing orange with a propane torch, twice, skipping the last inch at
each end.

Four unrelieved spokes, all curving very slightly away from their
elbows, which are hooked around the far ends of the rulers:

http://i14.tinypic.com/2w3ampy.jpg

Same four spokes after heating twice to orange from end to end:

http://i14.tinypic.com/2cpxcwj.jpg

As the 1 mm graph squares show, the change in bend is negligible.

The top spoke appears unchanged.

The middle two spokes moved slightly, curving perhaps 1 mm further
away from the ruler.

The bottom spoke bent the most, curving about 2 mm further away.

This suggests that the straight sections of unused spokes have very
little residual tension to relieve and and are not warped by heating
to glowing orange.

It doesn't seem worth trying to subject straight, unused spokes to
high tension to see if the faint bending lessens, since the change is
already so slight.

Cheers,

Carl Fogel
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Cycling Equipment
Cycling Forums › Forums › Bikes › Cycling Equipment › Spoke stress relief test