Stresses at spoke elbows
Ben C
including stresses at spoke elbows got so long it starting disappearing
off the ends of people's newsreaders. But since some people are still
interested in it it was suggested I should do a summary.
I'll try to keep it as short I can. No attributions, but most of the
material is originally from either Jobst or jim beam. Any
misrepresentations are unintentional.
Stress and stress relief
------------------------
This is a picture of a spoke sitting in a hole in a hub flange.
sHHHHHH <--- Hub
sssssssssss
sHHHHHH s
s
<-d->s <--- Spoke
s
s
s
RIM
If you pull the spoke along its axis towards the rim either the spoke
bends a bit more at the elbow, or the spoke cuts into the body of the
hub, or a bit of both.
We'll consider the first possibility in isolation first, so assume for
now that the hub hole is not deformable at all (this is pretty much the
case for a steel hub).
In the diagram the perpendicular distance from the spoke to the hub
flange has been labelled <-d->. The longer this distance, the lower the
force required to bend the spoke. The product of F and d is called a
"moment".
As this distance approaches zero, the force required to bend the spoke
absolutely flush with the flange approaches the force that would be
required to stretch a spoke plastically-- a force that's significantly
higher than spoke tension in a finished bicycle wheel.
In other words, to achieve this situation:
sHHHHHH
ssssssss
sHHHHHHs
s
s
s
s
just by pulling on the spoke would take a LOT of force.
Suppose you pull on the spoke with a force that gradually increases from
zero up to about normal tension, corresponding with screwing up the
nipple as you build the wheel. As the spoke bends towards the hub the
distance "d" gets shorter, and it gets harder to bend the spoke. Soon
you reach an equilibrium position in which the applied force holds the
bend where it is, but is unable to make it any tighter.
This applied force will hold the outside of the bend at just about the
yield stress of the material-- after all it's just been yielding (i.e.
bending) and you haven't relaxed the force, so it will be kept at yield.
So the spoke will not be quite flush with the flange, but some distance
"d" from it. How big is "d" exactly? I don't know for sure, I should
work it out, it isn't hard, and it's an important detail.
If you rode away on the bike like that you would expect rapid fatigue at
the spoke elbows since steel, like many materials, doesn't last very
long if it's cycled (repeatedly tensioned and relaxed again) at a high
mean stress.
But if you pull on the spoke a bit harder still and let go of it again,
you can plastically stretch the outside of the elbow a little bit
further, with the result that when you relax that momentary overload,
the outside of the elbow is left at a much lower stress.
This improves the life of the wheel, and is called "mechanical stress
relief".
Hub deformation
---------------
We've all seen the indentations around the exit holes of used hubs, and
we know that hubs are often made of aluminium that is softer than the
stainless steel used for spokes.
So what if the elbow doesn't bend at all, but instead just digs itself
into the hub a bit as it's pulled flush to the flange by normal
tensioning, and perhaps by some overload applied deliberately during the
build?
When you take an old wheel apart, you can often tell which spokes were
"inbound" and which were "outbound" by the different elbow bends, which
implies that the elbows _did_ bend as they were pulled towards the rim.
But the picture is confused because many builders bend the elbows
manually with their thumb or a crank-arm (which takes relatively little
force to do because of the position and direction in which the force is
applied).
We have seen inbound and outbound spokes from one poster, who advocates
against premature elbow-thumbing, that showed a barely perceptible
difference in elbow angle.
Of course if the elbow doesn't bend, or the hub is too soft to support
the force required to hold a bend in the elbow, there's no reason to
believe that the outside of the elbow will remain at yield stress after
spoke tensioning.
Both spoke bending and hub deformation
--------------------------------------
As tension is gradually applied to the spoke, at first it starts to bend
a bit, but as the spoke's exit line approaches parallel to the force
line, the moment becomes too small for it to continue to bend, and the
spoke elbow is held at yield stress.
In this scenario, as the tension is increased further, the pressure on
the inside of the hub starts to deform the exit hole, and the spoke
continues to pull closer to the flange, not by bending but by cutting
into the hub slightly.
This reduces the distance "d" even further, which means that the stress
at the spoke elbow actually reduces, even though the applied force is
still growing slowly as we wind the nipple. The moment Fd gets smaller
because although F is increasing, d is reducing more rapidly. In other
words, as the pivot around which we're twisting the spoke deforms, the
spoke unbends a bit.
We said at the start that to pull the spoke completely flush to the
flange around a non-deformable pivot would take significantly more spoke
tension than we ever get in a bicycle wheel.
So one possibility is that the spoke doesn't end up completely flush.
Another is that hub-hole deformation allows the spoke to pull flush, and
a consequence of this is that the spoke elbow does not remain at yield
after tensioning. There may still be some stress at the elbow though--
how far the bend gets relaxed depends on the extent of each of these
effects.
Spoke Line Correction
---------------------
So far we've been talking about building a wheel by just putting the
spokes in and tightening them up, followed perhaps by some momentary
overload process to relieve bending stress at the elbow.
But many builders push the spokes towards the flange earlier in the
process with their thumbs or old crank arms. If the elbow angle can be
made just right before the spoke is tensioned, then it won't need to
bend during the build, and we can expect the spoke to lie flush to the
flange when we're finished without requiring either hub-hole deformation
or yield stress at the elbow.
But is it possible to get the angle just right? In theory, no, on the
grounds that you cannot bend the spoke further than flush (the flange is
in the way), and that after you bend it it's always going to bounce back
a little bit.
In practice I'm not so sure. If the spoke is still quite loose, it may
not be sitting in quite its final position in the hub hole. If you can
hinge it back a bit before bending it, it seems at least possible that
you should be able to exceed the angle of the flange.
Which is right?
---------------
We can be fairly sure that spoke tension cannot hold the bend at yield
for the small value of "d" implied by flushness, and therefore that if
the spoke really is truly flush when we've finished, the bend is no
longer at yield.
So is the spoke truly flush? How far away does it have to be for it to
be reasonable for normal spoke tension to hold the elbow at yield?
It's hard to say. Small differences in angles and lengths is all it
takes one way or the other. Furthermore the spoke may end up bent into a
kind of question-mark shape just where it exits the flange, meaning it
can lie flat a bit further away, requiring less tension to keep it
there, but making it look flush unless you inspect very carefully around
where it exits the hole. The tighter the radius of the question-mark's
hook, the more tension is required, although the geometry gets much more
complicated at this scale. Do we see hooks? What are their radii? It
would help to have a close look at a few wheels, especially ones that we
knew hadn't undergone manual spoke line correction, but even then it
would be hard to know exactly what to conclude.
A good argument against hub-hole deformation relaxing yield stress at
the elbow (or in any way mitigating the bending of the elbow) is that we
would expect any deformation of the hub to happen sooner rather than
later in the build process. As the wire first starts to sink into the
aluminium, the contact patch is a thin line on the surface of the spoke
which sinks in like a knife blade. But as more of the wire sinks in, the
size of the contact quickly grows, and the amount of compressed
aluminium pushing it back increases. By the time the spoke is getting
near the flange, it's already sunk in enough that the apparently soft
aluminium hub is effectively no longer deformable.
My own conclusion is that theory doesn't tell us for sure whether or not
the spoke elbow is kept at or close to yield by spoke tension. More
evidence is needed. Carl puts it best: "the spokes do whatever they do
regardless of our red herrings".
It also depends on the choice of components used, in particular the
length of the elbow shanks and the orientation of the hub holes, which
varies even between different recent hubs from the same manufacturer as
people have been discussing recently. The situation may be different in
different wheels.
We have good ancedotal evidence that temporarily overloading the spokes
after tensioning produces more durable wheels. We know that this process
would relieve tensile stress at the elbow if it were there, and that
that would improve fatigue life.
On the other hand the temporary overload may have other benefits. It may
be that it helps seat the spokes into the hub, it may just be that
temporarily underloading the spokes (a sideeffect of certain
temporary-overload procedures) takes out windup.
Or it may have no positive benefits at all. The anecdotal evidence it
does something useful is good, but as time goes by the components we buy
from the manufacturers change with quite significant differences in
materials, geometry and importantly surface finish which is known to be
highly significant in fatigue failure.
It's less clear how different processes of stress relieving could do any
harm, although I think mechanisms may have been suggested by which that
is possible. As far as I remember I think we're all pretty much agreed
that the "Mavic Method" is probably beneficial and unlikely to do any
harm.
"Residual stress"
-----------------
The bend stress that may exist at the spoke elbow after tensioning that
I've been describing is sometimes called "residual stress". I believe
this is the term used in Jobst's book _The Bicycle Wheel_.
This seems a reasonable thing to call it on the plain-English grounds
that it's residual (left-over after bending) and it's stress.
But the term has another technical sense to describe stresses inside a
material that remain after plastic bending and spring-back. When you
bend a wire, the material near the skin goes through bigger angle
changes than the stuff nearer the centre. After the wire springs back
(the bend bounces out a bit), you're left with the unyielded interior
pushing elastically against the yielded exterior.
Note that these stresses are compressive on the outside of the bend and
tensile on the inside, so if present in a spoke would actually mitigate
the tensile stress on the outside of the elbow that may remain after
bending.
The spoke-toasting experiment of Carl Fogel indicated that these
residual stresses are relieved at tensions below what you expect in a
normal wheel anyway.
But these are not the same "residual stresses" as the bend stress
remaining at the elbow that we've been talking about. However many of us
didn't realize that that was what we were talking about until quite late
in the day.
Is is correct to use the term "residual stress" for this bend stress
remaining at the elbow? I don't know, but I do know I promised Ed
Pirrero a 500-post flame war on the subject. So, gentlemen, in your own
time, start whenever you're ready.
Residual stress of a similar kind to that remaining after spring-back
may remain in spokes after fabrication, and here we get into the
descriptions of the fabrication process. It would seem that the
spoke-toasting is good evidence that these stresses, if present, are
relieved anyway by normal tensioning, but there are also arguments
people have made against that.
A simple test for presence of residual stress from fabrication is the
chloride test.
Peter Cole
> But the term has another technical sense to describe stresses inside a
> material that remain after plastic bending and spring-back. When you
> bend a wire, the material near the skin goes through bigger angle
> changes than the stuff nearer the centre. After the wire springs back
> (the bend bounces out a bit), you're left with the unyielded interior
> pushing elastically against the yielded exterior.
>
> Note that these stresses are compressive on the outside of the bend and
> tensile on the inside, so if present in a spoke would actually mitigate
> the tensile stress on the outside of the elbow that may remain after
> bending.
Only if the spoke was relaxed. If a spoke is bent and held in the bent
position, the skin forces are the same as the core forces.
> The spoke-toasting experiment of Carl Fogel indicated that these
> residual stresses are relieved at tensions below what you expect in a
> normal wheel anyway.
Carl heated a (relaxed) previously bent spoke. Heating will relieve
internal stresses, but a spoke heated with a torch will heat from the
outside in, relaxing the skin before the core. In this case, the core
forces will bend the spoke before they also relax.
> Residual stress of a similar kind to that remaining after spring-back
> may remain in spokes after fabrication, and here we get into the
> descriptions of the fabrication process. It would seem that the
> spoke-toasting is good evidence that these stresses, if present, are
> relieved anyway by normal tensioning, but there are also arguments
> people have made against that.
Carl's experiment also unbent the spoke, reducing the core elastic
forces, so smaller changes with heating would be expected.
> A simple test for presence of residual stress from fabrication is the
> chloride test.
It is neither simple nor a sure fire indicator. Stress corrosion
cracking is dependent on the alloy (none of the spoke manufacturers
reveal theirs). It involves high temperature solution exposure for long
periods (ASM G36). SCC cracks often have very low growth rates (in the
order of years per mm). The cracks are typically small intergranular and
are only visible under magnification, often only after sectioning and
polishing.
I was inspired by this article:
http://www.lanl.gov/residual/
to think about using a similar method to reveal residual stresses.
I bent a spoke sharply (small radius on a vise). I clamped the small
(~1") part in the vise so that the remainder extended vertically. I
fixed a indicator behind the spoke so I could measure deflection of the
vertical part (~10"). Using a Dremel tool with a very thin cutoff disk
at low rpm, I began to remove material from the outside of the bend. I
did this slowly and gradually over the course of about 10 min., until I
had removed about 90% of the cross section.
I found that removing material from the outside of the bend cause the
spoke to bend more (smaller angle). The change was small, perhaps 2-3mm
movement at the end. I repeated the experiment with the same result.
Next, I tried the same experiment only removing material from the inside
of the bend. This time I saw the same magnitude of change, but in the
opposite direction, a slight unbending (the angle grew larger).
Lastly, I repeated the experiment with an unused DT spoke. I clamped
just the head in the vise so the elbow would be exposed and cut a slit
on the outside of the bend. The spoke bent more (to a smaller angle),
much the same as the bends I had made myself. My conclusion is that the
DT bend has the same residual as mine.
I'd like someone else to try this to confirm my results.
For the record, I torch-heated a 90 deg. bend and saw a similar amount
of deflection as in my test (a few mm) in the same direction Carl did.
carl...@comcast.net
[snip]
Dear Ben,
Thanks for the summary.
Two missing details may be important.
The spoke holes in the flange are counterbored on both ends, so at the
scale of a spoke, the lip of the spoke hole is pre-curved to an
appreciable degree.
But I suspect that even with the help of the counterbore, the outer
spokes rarely (if ever)lie flat on the hub flange for their whole
length.
Here's your diagram modified:
s\HHHHHHH/ <--undamaged counter-bore
ssssssss
s/HHHHH\s <--some bedding-gouging
\s
\s
|s
|.s
|..s <--spoke still curves off flange
|..s . = gap, the d of Ben's diagram
|.s
|s
|s <--now bent flat against flange
|s by hand for a long stretch
|s
s
s
s
rim
That is, the spoke never lies flat on the hub flange near the elbow.
Short of using a hammer and steel mandrel, bending to adjust the spoke
line always leaves a little space where the elbow curves away from the
hub, the gap d that interests you.
Here's a diagram from p. 80 of "The Bicycle Wheel," 3rd edition,
showing the ever-present gap d:
http://i16.tinypic.com/2qdzpt4.jpg
The idealized diagram doesn't show the flange counterboring--note that
the artist has the angled spoke head sitting above the idealized
right-angle hole through the flange.
But the diagram does show the additional slight complication of the
spoke cocking in the hole through the flange, due to the slight
clearance.
Cheers,
Carl Fogel
Peter Cole
> Both spoke bending and hub deformation
> --------------------------------------
>
> As tension is gradually applied to the spoke, at first it starts to bend
> a bit, but as the spoke's exit line approaches parallel to the force
> line, the moment becomes too small for it to continue to bend, and the
> spoke elbow is held at yield stress.
>
> In this scenario, as the tension is increased further, the pressure on
> the inside of the hub starts to deform the exit hole, and the spoke
> continues to pull closer to the flange, not by bending but by cutting
> into the hub slightly.
>
> This reduces the distance "d" even further, which means that the stress
> at the spoke elbow actually reduces, even though the applied force is
> still growing slowly as we wind the nipple.
You are speculating about the "stress at the spoke elbow", and there are
several stresses at work.
The only important thing is, that after tensioning, the spoke (depending
on whether its initial angle was too big or too small) is either trying
to open up or close at the bend. In the first case (too small an initial
bend angle) the tension from bending is on the inside, in the second
case, on the outside. Tension is the only important bending stress
because it gets added to the axial tension.
If spokes break (fatigue) on the outside of the bend, it can only mean
that the initial angle was too great. That the spoke doesn't lie flat to
the (angled) flange is only a symptom of that. In the other case (too
small an angle), lying flat to the flange tells you nothing.
When you momentarily increase the spoke tension by a large amount during
stress relief, you yield the areas at or near (tensile) yield. It
doesn't matter if they are on the inside or outside of the bend. It has
the same effect (for those small regions) of briefly increasing the bend
in the direction it wants to go (to be perfect), yielding material
further in that direction and (when released) lowering the tensile
stress there. This is pretty much what would occur if you could just
bend the spoke further in the direction it wanted to go.
If outbound spokes show a change from obtuse (factory) to acute after
tensioning, it's only because the initial angle was too great.
Correcting the spoke line (at the hub, after tensioning) can't
completely eliminate the discrepancy; the tensile stresses will be on
the outside of the elbows, and if not stress relieved, the spokes will
fail there.
ajam...@hotmail.com
Peter Cole wrote:
> Ben C wrote:
>
>
> It is neither simple nor a sure fire indicator. Stress corrosion
> cracking is dependent on the alloy (none of the spoke manufacturers
> reveal theirs). It involves high temperature solution exposure for long
> periods (ASM G36). SCC cracks often have very low growth rates (in the
> order of years per mm). The cracks are typically small intergranular and
> are only visible under magnification, often only after sectioning and
> polishing.
>
FWIW when I first started with the Aerospace company I worked for, I
did a bunch of tests on spokes, (1) Including stress corrosion and
micro sectioning. To the best of my ability to test (no scanning
electron microscope or gas chromatograph) the DT spokes I looked at
were 304, rockwell / uts / mag-perm / sheer / acid etch etc. all
pointed in that direction and since it is the most common grade of
stainless used and relatively cheap I am reasonably confident that I am
correct.
but as it was 12+ years ago now the odds of the test pieces still
being around are slim..
As an aside and since I can't provide the test pieces or results
anymore, the grain structure and the stress corrosion results supported
the idea that residual stress existed at the elbow.. that stress wanted
the spoke to straighten.
(1) and one or two tests of hubs, welds cranks etc. as I used up the
extra bits I acquired over the years.
Ben C
> Ben C wrote:
>
>> But the term has another technical sense to describe stresses inside a
>> material that remain after plastic bending and spring-back. When you
>> bend a wire, the material near the skin goes through bigger angle
>> changes than the stuff nearer the centre. After the wire springs back
>> (the bend bounces out a bit), you're left with the unyielded interior
>> pushing elastically against the yielded exterior.
>>
>> Note that these stresses are compressive on the outside of the bend and
>> tensile on the inside, so if present in a spoke would actually mitigate
>> the tensile stress on the outside of the elbow that may remain after
>> bending.
>
> Only if the spoke was relaxed. If a spoke is bent and held in the bent
> position, the skin forces are the same as the core forces.
Yes, that was what I was trying to clarify, I think-- the distinction
between applied stress at the elbow and the sort of residual stresses
you get when you bend and allow to spring back.
>> The spoke-toasting experiment of Carl Fogel indicated that these
>> residual stresses are relieved at tensions below what you expect in a
>> normal wheel anyway.
>
> Carl heated a (relaxed) previously bent spoke. Heating will relieve
> internal stresses, but a spoke heated with a torch will heat from the
> outside in, relaxing the skin before the core. In this case, the core
> forces will bend the spoke before they also relax.
Yes isn't that the principle-- if you heat it and it unbends a bit that
means you have residual stresses, or have I misunderstood?
[snip]
>> A simple test for presence of residual stress from fabrication is the
>> chloride test.
>
> It is neither simple nor a sure fire indicator. Stress corrosion
> cracking is dependent on the alloy (none of the spoke manufacturers
> reveal theirs). It involves high temperature solution exposure for long
> periods (ASM G36). SCC cracks often have very low growth rates (in the
> order of years per mm). The cracks are typically small intergranular and
> are only visible under magnification, often only after sectioning and
> polishing.
So I guess you don't hold out much hope for the couple of samples I've
left in a jar of salt water...
[snip]
> Lastly, I repeated the experiment with an unused DT spoke. I clamped
> just the head in the vise so the elbow would be exposed and cut a slit
> on the outside of the bend. The spoke bent more (to a smaller angle),
> much the same as the bends I had made myself. My conclusion is that the
> DT bend has the same residual as mine.
Excellent result!
Actually that test could possibly even be done in an assembled wheel,
before and after stress-relief.
Peter Cole
> On 2007-01-10, Peter Cole <peter...@comcast.net> wrote:
>>> The spoke-toasting experiment of Carl Fogel indicated that these
>>> residual stresses are relieved at tensions below what you expect in a
>>> normal wheel anyway.
>> Carl heated a (relaxed) previously bent spoke. Heating will relieve
>> internal stresses, but a spoke heated with a torch will heat from the
>> outside in, relaxing the skin before the core. In this case, the core
>> forces will bend the spoke before they also relax.
>
> Yes isn't that the principle-- if you heat it and it unbends a bit that
> means you have residual stresses, or have I misunderstood?
As I said, the tensioning unbent the spoke, removing most of the
stresses that way. The spokes in wheels remain bent.
>
> [snip]
>>> A simple test for presence of residual stress from fabrication is the
>>> chloride test.
>> It is neither simple nor a sure fire indicator. Stress corrosion
>> cracking is dependent on the alloy (none of the spoke manufacturers
>> reveal theirs). It involves high temperature solution exposure for long
>> periods (ASM G36). SCC cracks often have very low growth rates (in the
>> order of years per mm). The cracks are typically small intergranular and
>> are only visible under magnification, often only after sectioning and
>> polishing.
>
> So I guess you don't hold out much hope for the couple of samples I've
> left in a jar of salt water...
>
> [snip]
>> Lastly, I repeated the experiment with an unused DT spoke. I clamped
>> just the head in the vise so the elbow would be exposed and cut a slit
>> on the outside of the bend. The spoke bent more (to a smaller angle),
>> much the same as the bends I had made myself. My conclusion is that the
>> DT bend has the same residual as mine.
>
> Excellent result!
I'm not sure it's excellent. It's a data point that seems to answer
whether spokes have residual stresses from manufacture. I'll wait for
others to comment/confirm.
> Actually that test could possibly even be done in an assembled wheel,
> before and after stress-relief.
I don't see how. I wouldn't want to cut a 90% slit through a loaded
spoke, nor do I think that would show anything.
Peter Cole
Thanks, another data point (in the same direction).
Ben C
>> Actually that test could possibly even be done in an assembled wheel,
>> before and after stress-relief.
>
> I don't see how. I wouldn't want to cut a 90% slit through a loaded
> spoke, nor do I think that would show anything.
If the outside elbow of an outbound spoke was under tension, and you cut
a little slit at the outside of the bend, would it not tear open a bit?
I mean the slit tear open, the elbow bend would become more acute. The
idea is you make the material a bit thinner so it starts to strain
visibly if it's under stress.
Ben C
This is a much better diagram. Note that "d" here doesn't work in quite
the same way as in my simplified diagram. The general principle is still
about right: that the bigger d is, the less tension you require to keep
the bend yielding, and that as d tends to zero the tension required
tends to the maximum (the amount you'd need to stretch the spoke
plastically). But straightening the hook of the question mark is rather
more difficult geometry than the simple moment F x d in the simplified
diagram.
Peter Cole
Try it and get back to me.
Tim McNamara
Peter Cole <peter...@comcast.net> wrote:
> I bent a spoke sharply (small radius on a vise). I clamped the small
> (~1") part in the vise so that the remainder extended vertically. I
> fixed a indicator behind the spoke so I could measure deflection of
> the vertical part (~10"). Using a Dremel tool with a very thin cutoff
> disk at low rpm, I began to remove material from the outside of the
> bend. I did this slowly and gradually over the course of about 10
> min., until I had removed about 90% of the cross section.
>
> I found that removing material from the outside of the bend cause the
> spoke to bend more (smaller angle). The change was small, perhaps
> 2-3mm movement at the end. I repeated the experiment with the same
> result. Next, I tried the same experiment only removing material from
> the inside of the bend. This time I saw the same magnitude of change,
> but in the opposite direction, a slight unbending (the angle grew
> larger).
>
> Lastly, I repeated the experiment with an unused DT spoke. I clamped
> just the head in the vise so the elbow would be exposed and cut a
> slit on the outside of the bend. The spoke bent more (to a smaller
> angle), much the same as the bends I had made myself. My conclusion
> is that the DT bend has the same residual as mine.
>
> I'd like someone else to try this to confirm my results.
Well, the next step would be to take a spoke that has been treated to
Jobst stress-relieving method and see if it behaves the same way, and to
take a spoke that has been built into a wheel but not stress-relieved
and see if that too behaves the same way. If Jobst's stress relieving
method works, it would be reasonable to predict that the angle of the
spoke would change somewhat less on the tensioned-but-not-relieved
spoke, and would change much less on the stress-relieved spoke.
jim beam
no, they're exactly the same kind of thing. the profile may be slightly
different because the bend radius is different, but the principle is
identical.
>
> Is is correct to use the term "residual stress" for this bend stress
> remaining at the elbow? I don't know, but I do know I promised Ed
> Pirrero a 500-post flame war on the subject. So, gentlemen, in your own
> time, start whenever you're ready.
>
> Residual stress of a similar kind to that remaining after spring-back
> may remain in spokes after fabrication, and here we get into the
> descriptions of the fabrication process. It would seem that the
> spoke-toasting is good evidence that these stresses, if present, are
> relieved anyway by normal tensioning, but there are also arguments
> people have made against that.
if there's residual stress at the elbow from manufacture, there's
residual stress at the elbow from manufacture. but presumption is not
any form of quantification. theory is that there's compressive [i.e.
fatigue mitigating] stress on the outside of spoke elbows, but they
still break there anyway - and they /don't/ break where tensile residual
stress is highest. since residual stress theory and observed failures
are not in accord, the role residual stress plays in spoke fatigue is at
best "unproven".
>
> A simple test for presence of residual stress from fabrication is the
> chloride test.
indeed.
all in all, from my first reading, a pretty decent summary. but i'll
have to get back to you if you want more detail. time limited.
jim beam
> Peter Cole wrote:
>> Ben C wrote:
>>
>
>> It is neither simple nor a sure fire indicator. Stress corrosion
>> cracking is dependent on the alloy (none of the spoke manufacturers
>> reveal theirs). It involves high temperature solution exposure for long
>> periods (ASM G36). SCC cracks often have very low growth rates (in the
>> order of years per mm). The cracks are typically small intergranular and
>> are only visible under magnification, often only after sectioning and
>> polishing.
>>
> FWIW when I first started with the Aerospace company I worked for, I
> did a bunch of tests on spokes, (1) Including stress corrosion and
> micro sectioning. To the best of my ability to test (no scanning
> electron microscope or gas chromatograph) the DT spokes I looked at
> were 304, rockwell / uts / mag-perm / sheer / acid etch etc. all
> pointed in that direction and since it is the most common grade of
> stainless used and relatively cheap I am reasonably confident that I am
> correct.
>
> but as it was 12+ years ago now the odds of the test pieces still
> being around are slim..
>
> As an aside and since I can't provide the test pieces or results
> anymore, the grain structure and the stress corrosion results
what did you use?
> supported
> the idea that residual stress existed at the elbow.. that stress wanted
> the spoke to straighten.
>
> (1) and one or two tests of hubs, welds cranks etc. as I used up the
> extra bits I acquired over the years.
>
good to hear - someone that actually did testing!!! it would be great
if you could share more detail.
Andrew Lee
> It is neither simple nor a sure fire indicator. Stress corrosion cracking
> is dependent on the alloy (none of the spoke manufacturers reveal theirs).
The Wheelsmith website < http://www.wheelsmith.com/index_files/wsspokes.htm
> was talked about recently here. I noticed that they also use 304
stainless steel.
carl...@comcast.net
Dear Andrew,
As a sidelight on what spokes are made of, Jobst pointed out the
change in spoke materials between the first and second edition of "The
Bicycle Wheel":
"In contrast to spokes tested for the first edition of this book,
these spokes withstood substantial elongation before failure,
indicating improved spoke materials."
Graphs in the first and second editions indicate that DT 2mm straight
stainless steel spokes tested in the 1st edition (1981) failed at ~700
lbs of stress and ~2.6 mm of strain, while DT 2mm straight stainless
steel spokes tested in the second edition (1988) failed at ~700 lbs of
stress again, but ~4.0 mm of strain.
Somehow the "stainless steel" in 2mm DT spokes became 50% stretchier.
There's also an obvious difference between the carbon steel and the
stainless steel spokes pulled to destruction in the first edition. The
carbon spokes have a sharp failure peak, while the stainless steel
spokes have much stretchier, more rounded failure curve.
Here's a crude comparison for the 2mm straight DT spokes in the two
editions:
1981 1981 1988
carbon stainless stainless
^ .. . . . . .
| / \ ' ' ' '
L / / '
B / / /
S
abcd abcdef abcdefgh
STRETCH-->
Cheers,
Carl Fogel
Ben C
> Ben C wrote:
[snip]
>> The spoke-toasting experiment of Carl Fogel indicated that these
>> residual stresses are relieved at tensions below what you expect in a
>> normal wheel anyway.
>>
>> But these are not the same "residual stresses" as the bend stress
>> remaining at the elbow that we've been talking about. However many of us
>> didn't realize that that was what we were talking about until quite late
>> in the day.
>
> no, they're exactly the same kind of thing. the profile may be slightly
> different because the bend radius is different, but the principle is
> identical.
Residual stress from fabrication seems close enough to what Carl (and
more recently also Peter Cole) tested.
But those tests tell us nothing about _applied_ stress at the elbow due
to tension in the finished wheel. That's all I was saying.
Peter Cole
> if there's residual stress at the elbow from manufacture, there's
> residual stress at the elbow from manufacture. but presumption is not
> any form of quantification. theory is that there's compressive [i.e.
> fatigue mitigating] stress on the outside of spoke elbows, but they
> still break there anyway - and they /don't/ break where tensile residual
> stress is highest. since residual stress theory and observed failures
> are not in accord, the role residual stress plays in spoke fatigue is at
> best "unproven".
If we look at residual stresses and superimpose various applied stresses
we see the following:
Residual stress, going from inside to outside of the elbow:
T C T C
If we superimpose spoke tension:
T+ C- T+ C-
If we add stress relief momentary force:
T++ C-- T++ C--
Another case where the original angle is too small:
T C T C (residual)
T+ C- T+ C- (residual + spoke tension)
T++ C-- T+- C-+ (residual + tension + bending force)
T+++ C--- T+-+ C-+- (residual + tension + bending force + stress relief)
Another case where the original angle is too large:
T C T C (residual)
T+ C- T+ C- (residual + spoke tension)
T+- C-+ T++ C-- (residual + spoke tension + bending force)
T+-+ C-+- T+++ C--- (residual + spoke tension + bending force + stress
relief)
The above illustrates only directions, not magnitudes, but if we assume
"T++" and above will result in yield followed by a reduction in T, we
can see that stress relief is beneficial in all 3 cases. It is important
to remember that these stress regions have gradients and that the large
relative change in initial absolute compression will give absolute
tension, i.e. C--- ~= T+.
ajam...@hotmail.com
jim beam wrote:
> ajam...@hotmail.com wrote:
> > Peter Cole wrote:
> >> Ben C wrote:
> >>
> >
> >> It is neither simple nor a sure fire indicator. Stress corrosion
> >> cracking is dependent on the alloy (none of the spoke manufacturers
> >> reveal theirs). It involves high temperature solution exposure for long
> >> periods (ASM G36). SCC cracks often have very low growth rates (in the
> >> order of years per mm). The cracks are typically small intergranular and
> >> are only visible under magnification, often only after sectioning and
> >> polishing.
> >>
> > FWIW when I first started with the Aerospace company I worked for, I
> > did a bunch of tests on spokes, (1) Including stress corrosion and
> > micro sectioning. To the best of my ability to test (no scanning
> > electron microscope or gas chromatograph) the DT spokes I looked at
> > were 304, rockwell / uts / mag-perm / sheer / acid etch etc. all
> > pointed in that direction and since it is the most common grade of
> > stainless used and relatively cheap I am reasonably confident that I am
> > correct.
> >
> > but as it was 12+ years ago now the odds of the test pieces still
> > being around are slim..
> >
> > As an aside and since I can't provide the test pieces or results
> > anymore, the grain structure and the stress corrosion results
>
> what did you use?
The NAS accelerated Stress corrosion test used for stainless
fasteners.. I can't remember the number for the life of me.. spokes
were pre-stressed in a fixture then underwent repeated immersion in
salt water solution at elevated temp. I attempted to compensate for the
hub rim offset..
micro-sections were done and samples were Acid etched (Oxalic?
Hydrofluoric? don't remember) / microscope showed inter-granular
corrosion (grain boundary attack what ever the correct term is)
concentrated under the spoke elbow but present as well under the head
and in the threads.
> >
> > (1) and one or two tests of hubs, welds cranks etc. as I used up the
> > extra bits I acquired over the years.
> >
>
> good to hear - someone that actually did testing!!! it would be great
> if you could share more detail.
I don't want people to get me wrong here I'm not an engineer I was a
technician ( a damned good one, but still) I did this for personal
interest and as practice when I was new in the job. My sample sizes
were small and could very easily have been non-representational, but
they are consistent with what other anecdotal evidence seems to show.
Peter Cole
> spokes
> were pre-stressed in a fixture then underwent repeated immersion in
> salt water solution at elevated temp. I attempted to compensate for the
> hub rim offset..
Unfortunately, this makes these tests not say much about residual
manufacturing stress.
ajam...@hotmail.com
certainly not conclusive of but it is I think indicative of.. I have a
hard time picturing a fixture that would hold the spoke in tension for
a stress corrosion test that would not impart some sort angular stress.
Ron Ruff
Peter Cole wrote:
> The above illustrates only directions, not magnitudes, but if we assume
> "T++" and above will result in yield followed by a reduction in T, we
> can see that stress relief is beneficial in all 3 cases. It is important
> to remember that these stress regions have gradients and that the large
> relative change in initial absolute compression will give absolute
> tension, i.e. C--- ~= T+.
That makes more sense to me now, Peter...
Do we know anyhing about the magnitudes? In other words, can we say how
the magnitude of residual stresses compares to the loaded stresses?
Maybe T+-+ = T, etc.
Would it be safe to say that over-stressing a built wheel as high as
possible... which I guess would be to the yield stress of the "weakest"
link in the body of the spoke; the threads or butted section... would
be ideal? At least that would even out the stresses in the elbow as
much as possible.
Michael Press
<1168457385....@i56g2000hsf.googlegroups.com>,
"ajam...@hotmail.com" <ajam...@hotmail.com> wrote:
The last sentence is difficult to translate. Seems to
me the residual stress balances out to a static
configuration.
Removing material from the outside of the elbow will
change the balance of the residual stresses, so that
the bend will open up until all stresses are balanced
again. Similarly, removing material from the inside of
the bend will make the bend close a bit.
--
Michael Press
Peter Cole
You left out an important piece:
"spokes were pre-stressed in a fixture then underwent repeated immersion
in salt water solution at elevated temp"
It is the "pre-stressing" that may mask out residual manufacturing
stresses. The skin stresses in the elbow could go either way depending
on whether the fixture loaded with inward or outward bending forces.
Even pure tension would hide the outside elbow skin compression which we
would expect to find. The only sure-fire way to determine skin stress
from manufacturing residual stress would be to run the test on an
unloaded, never loaded, spoke.
> Removing material from the outside of the elbow will
> change the balance of the residual stresses, so that
> the bend will open up until all stresses are balanced
> again. Similarly, removing material from the inside of
> the bend will make the bend close a bit.
According to the article I cited, doing this precisely while measuring
strains, when combined with fairly hairy math will yield a stress
profile. What I did was very crude in comparison, but I think enough to
show the presence of residual manufacturing stresses.
Peter Cole
> Peter Cole wrote:
>> The above illustrates only directions, not magnitudes, but if we assume
>> "T++" and above will result in yield followed by a reduction in T, we
>> can see that stress relief is beneficial in all 3 cases. It is important
>> to remember that these stress regions have gradients and that the large
>> relative change in initial absolute compression will give absolute
>> tension, i.e. C--- ~= T+.
>
> That makes more sense to me now, Peter...
>
> Do we know anyhing about the magnitudes? In other words, can we say how
> the magnitude of residual stresses compares to the loaded stresses?
> Maybe T+-+ = T, etc.
I think we know some things. We know that the stress from tensioning the
spoke is roughly 1/3 yield. I don't know about the residual
manufacturing stress, my guess is that its peak would be similar in
magnitude. The wild card is the bending stress. If the angular mismatch
is great enough it could swamp out the other stresses. It is certainly
possible for it to go to yield -- in either direction.
> Would it be safe to say that over-stressing a built wheel as high as
> possible... which I guess would be to the yield stress of the "weakest"
> link in the body of the spoke; the threads or butted section... would
> be ideal? At least that would even out the stresses in the elbow as
> much as possible.
I think so. More is better, at least to the point where you have to
worry about buckling the rim or snapping spokes.
jim beam
problem. that's testing for applied stress, not residual. the spoke
needs to be free, straight from the package and unmounted in any way.
> then underwent repeated immersion in
> salt water solution at elevated temp. I attempted to compensate for the
> hub rim offset..
>
> micro-sections were done and samples were Acid etched (Oxalic?
> Hydrofluoric? don't remember) / microscope showed inter-granular
> corrosion (grain boundary attack what ever the correct term is)
> concentrated under the spoke elbow but present as well under the head
> and in the threads.
>
>
>
>>> (1) and one or two tests of hubs, welds cranks etc. as I used up the
>>> extra bits I acquired over the years.
>>>
>> good to hear - someone that actually did testing!!! it would be great
>> if you could share more detail.
>
> I don't want people to get me wrong here I'm not an engineer I was a
> technician ( a damned good one, but still) I did this for personal
> interest and as practice when I was new in the job. My sample sizes
> were small and could very easily have been non-representational, but
> they are consistent with what other anecdotal evidence seems to show.
be very careful with the anecdotal here on r.b.t.
jim beam
> Peter Cole wrote:
>> The above illustrates only directions, not magnitudes, but if we assume
>> "T++" and above will result in yield followed by a reduction in T, we
>> can see that stress relief is beneficial in all 3 cases. It is important
>> to remember that these stress regions have gradients and that the large
>> relative change in initial absolute compression will give absolute
>> tension, i.e. C--- ~= T+.
>
> That makes more sense to me now, Peter...
>
> Do we know anyhing about the magnitudes?
the most important point of the whole matter!!!...
> In other words, can we say how
> the magnitude of residual stresses compares to the loaded stresses?
> Maybe T+-+ = T, etc.
>
> Would it be safe to say that over-stressing a built wheel as high as
> possible... which I guess would be to the yield stress of the "weakest"
> link in the body of the spoke; the threads or butted section... would
> be ideal? At least that would even out the stresses in the elbow as
> much as possible.
>
there is a logical disconnect between this announcement that one phase
of yielding creates residual stress, and the supposition that yet more
of the same magically removes it again.
testing does of course show that it can happen to some degree in some
conditions, but the "lab" conditions under which those are observed are
limited in both degree and duration - factors not controlled in the
outside world. we also have to ask ourselves about the extent to which
this really is a quantifiable real-world problem. pressed steel wheels
used on cars for example are highly cold worked, have "residual stress"
from manufacture, are subject to the same old cycles of cyclic elastic
deformation due to high loads and extended mileage, yet hardly ever
evidence any fatigue at all, let alone fatigue that can be directly
attributed to residual stress. same for drawn tube, pressed chassis,
nervex lugs... presumptive nonsense that residual stress is the single
most important factor, and that it can be effectively managed in an
inconsistent and arbitrary fashion, betrays a deep and fundamental
ignorance of the whole subject.
Michael Press
<FPOdnTwXiKSSHDrY...@comcast.com>,
Peter Cole <peter...@comcast.net> wrote:
Yes, I missed that part.
> It is the "pre-stressing" that may mask out residual manufacturing
> stresses. The skin stresses in the elbow could go either way depending
> on whether the fixture loaded with inward or outward bending forces.
> Even pure tension would hide the outside elbow skin compression which we
> would expect to find. The only sure-fire way to determine skin stress
> from manufacturing residual stress would be to run the test on an
> unloaded, never loaded, spoke.
>
>
> > Removing material from the outside of the elbow will
> > change the balance of the residual stresses, so that
> > the bend will open up until all stresses are balanced
> > again. Similarly, removing material from the inside of
> > the bend will make the bend close a bit.
>
> According to the article I cited, doing this precisely while measuring
> strains, when combined with fairly hairy math will yield a stress
> profile. What I did was very crude in comparison, but I think enough to
> show the presence of residual manufacturing stresses.
--
Michael Press
Peter Cole
> Ron Ruff wrote:
>> Peter Cole wrote:
>>> The above illustrates only directions, not magnitudes, but if we assume
>>> "T++" and above will result in yield followed by a reduction in T, we
>>> can see that stress relief is beneficial in all 3 cases. It is important
>>> to remember that these stress regions have gradients and that the large
>>> relative change in initial absolute compression will give absolute
>>> tension, i.e. C--- ~= T+.
>>
>> That makes more sense to me now, Peter...
>>
>> Do we know anyhing about the magnitudes?
>
> the most important point of the whole matter!!!...
>
>> In other words, can we say how
>> the magnitude of residual stresses compares to the loaded stresses?
>> Maybe T+-+ = T, etc.
>>
>> Would it be safe to say that over-stressing a built wheel as high as
>> possible... which I guess would be to the yield stress of the "weakest"
>> link in the body of the spoke; the threads or butted section... would
>> be ideal? At least that would even out the stresses in the elbow as
>> much as possible.
>>
> there is a logical disconnect between this announcement that one phase
> of yielding creates residual stress, and the supposition that yet more
> of the same magically removes it again.
No "disconnect", only stress relief.
>
> testing does of course show that it can happen to some degree in some
> conditions, but the "lab" conditions under which those are observed are
> limited in both degree and duration - factors not controlled in the
> outside world.
OK, you've gone from rejecting the presence of residual stress in spokes
to now claiming it's there but unimportant.
> we also have to ask ourselves about the extent to which
> this really is a quantifiable real-world problem. pressed steel wheels
> used on cars for example are highly cold worked, have "residual stress"
> from manufacture, are subject to the same old cycles of cyclic elastic
> deformation due to high loads and extended mileage, yet hardly ever
> evidence any fatigue at all, let alone fatigue that can be directly
> attributed to residual stress. same for drawn tube, pressed chassis,
> nervex lugs... presumptive nonsense that residual stress is the single
> most important factor, and that it can be effectively managed in an
> inconsistent and arbitrary fashion, betrays a deep and fundamental
> ignorance of the whole subject.
If you take the trouble to read up on this, you'll find that fatigue and
residual stress are factors in steel automobile wheels.
ajam...@hotmail.com
jim beam wrote:
> ajam...@hotmail.com wrote:
> > jim beam wrote:
> >> ajam...@hotmail.com wrote:
> >>> Peter Cole wrote:
> >>>> Ben C wrote:
> >
> > The NAS accelerated Stress corrosion test used for stainless
> > fasteners.. I can't remember the number for the life of me.. spokes
> > were pre-stressed in a fixture
>
> problem. that's testing for applied stress, not residual. the spoke
> needs to be free, straight from the package and unmounted in any way.
>
I wasn't specifically testing for residual stress, just looking for a
better personal understanding of what was going on. However I would
like to propose the following as a basement chemists attempt at a
reasonable test..
Assuming that the material is 304, immersing the spoke elbow in hot
oxalic acid should (after some time) etch the grain boundaries,
microscopic examination should show a greater degree of etching in
areas where the compressive stress is less. Additionally as the grain
boundaries on the surface are attacked some small change in the angle
should be noted, though it might require a Shadowgraph to measure.
Agree? Disagree? Comments?
Since I no longer work anywhere I have access to either a Shadowgraph
or Oxalic acid I volunteer to provide the spokes to anyone who can do
this test..
jim beam
> jim beam wrote:
>> ajam...@hotmail.com wrote:
>>> jim beam wrote:
>>>> ajam...@hotmail.com wrote:
>>>>> Peter Cole wrote:
>>>>>> Ben C wrote:
>
>>> The NAS accelerated Stress corrosion test used for stainless
>>> fasteners.. I can't remember the number for the life of me.. spokes
>>> were pre-stressed in a fixture
>> problem. that's testing for applied stress, not residual. the spoke
>> needs to be free, straight from the package and unmounted in any way.
>>
>
> I wasn't specifically testing for residual stress, just looking for a
> better personal understanding of what was going on. However I would
> like to propose the following as a basement chemists attempt at a
> reasonable test..
>
> Assuming that the material is 304, immersing the spoke elbow in hot
> oxalic acid should (after some time) etch the grain boundaries,
> microscopic examination should show a greater degree of etching in
> areas where the compressive stress is less. Additionally as the grain
> boundaries on the surface are attacked some small change in the angle
> should be noted, though it might require a Shadowgraph to measure.
>
> Agree? Disagree? Comments?
that's just etching. yes, dissolution rate is/can be affected by
lattice strain, but stress corrosion cracking, what you want to induce
when testing for residual stress, is a different ball of wax. for 304,
as with many other stainless alloys, chlorides are the way to go.
depending on alloy/scc system, crack progression can be transgranular as
well as intergranular, and /very/ rapid. one of the reagents used for
mild steel can open up about a foot of drawn steel tube inside 5 minutes.
jim beam
the logical disconnect is the vast gulf of unexplained uncertainty
between the two. want to put numbers to strain that induces relief and
strain that induces an increase?
>
>>
>> testing does of course show that it can happen to some degree in some
>> conditions, but the "lab" conditions under which those are observed
>> are limited in both degree and duration - factors not controlled in
>> the outside world.
>
>
> OK, you've gone from rejecting the presence of residual stress in spokes
> to now claiming it's there but unimportant.
the key word is "significant". read again and look for it.
>
>
>> we also have to ask ourselves about the extent to which this really
>> is a quantifiable real-world problem. pressed steel wheels used on
>> cars for example are highly cold worked, have "residual stress" from
>> manufacture, are subject to the same old cycles of cyclic elastic
>> deformation due to high loads and extended mileage, yet hardly ever
>> evidence any fatigue at all, let alone fatigue that can be directly
>> attributed to residual stress. same for drawn tube, pressed chassis,
>> nervex lugs... presumptive nonsense that residual stress is the
>> single most important factor, and that it can be effectively managed
>> in an inconsistent and arbitrary fashion, betrays a deep and
>> fundamental ignorance of the whole subject.
>
> If you take the trouble to read up on this, you'll find that fatigue and
> residual stress are factors in steel automobile wheels.
not in any significant way. i've seen many many thousands of these
things - not a single failure that can be ascribed to residual stress.
Peter Cole
> Peter Cole wrote:
>> jim beam wrote:
>>> there is a logical disconnect between this announcement that one
>>> phase of yielding creates residual stress, and the supposition that
>>> yet more of the same magically removes it again.
>>
>> No "disconnect", only stress relief.
>
> the logical disconnect is the vast gulf of unexplained uncertainty
> between the two. want to put numbers to strain that induces relief and
> strain that induces an increase?
Who's talking "strain"? I think you still don't get it. Mechanical
stress relief is simple, straightforward, and common in practice. It
should be obvious that a nominal 50% increase in stress will only affect
those regions that are already that close to yield -- whether those
existing stresses are residual manufacturing stresses or static load
stresses or a combination -- but this is just beating a dead horse.
>>> testing does of course show that it can happen to some degree in some
>>> conditions, but the "lab" conditions under which those are observed
>>> are limited in both degree and duration - factors not controlled in
>>> the outside world.
>>
>>
>> OK, you've gone from rejecting the presence of residual stress in
>> spokes to now claiming it's there but unimportant.
>
> the key word is "significant". read again and look for it.
You're just hand waving. The residual stresses are there. They will bias
the fatigue curve. End of story. You want quantification? Quantify it
yourself. It doesn't matter.
>>> we also have to ask ourselves about the extent to which this really
>>> is a quantifiable real-world problem. pressed steel wheels used on
>>> cars for example are highly cold worked, have "residual stress" from
>>> manufacture, are subject to the same old cycles of cyclic elastic
>>> deformation due to high loads and extended mileage, yet hardly ever
>>> evidence any fatigue at all, let alone fatigue that can be directly
>>> attributed to residual stress. same for drawn tube, pressed chassis,
>>> nervex lugs... presumptive nonsense that residual stress is the
>>> single most important factor, and that it can be effectively managed
>>> in an inconsistent and arbitrary fashion, betrays a deep and
>>> fundamental ignorance of the whole subject.
>>
>> If you take the trouble to read up on this, you'll find that fatigue
>> and residual stress are factors in steel automobile wheels.
>
> not in any significant way. i've seen many many thousands of these
> things - not a single failure that can be ascribed to residual stress.
Well I suppose that endorsement will be very bad news to the industry
that makes fatigue testing equipment for steel automotive wheels and the
designers who deliberately introduce residual stress to improve fatigue
characteristics while reducing weight.
Ben C
> Peter Cole wrote:
>> jim beam wrote:
>>> Ron Ruff wrote:
>>>> Would it be safe to say that over-stressing a built wheel as high as
>>>> possible... which I guess would be to the yield stress of the "weakest"
>>>> link in the body of the spoke; the threads or butted section... would
>>>> be ideal? At least that would even out the stresses in the elbow as
>>>> much as possible.
>>>>
>>> there is a logical disconnect between this announcement that one phase
>>> of yielding creates residual stress, and the supposition that yet more
>>> of the same magically removes it again.
>>
>> No "disconnect", only stress relief.
>
> the logical disconnect is the vast gulf of unexplained uncertainty
> between the two. want to put numbers to strain that induces relief and
> strain that induces an increase?
The logical connection is made by requiring subsequent relaxation after
the momentary overload.
The idea is that if you pull a wire such that it bends, and hold it
there, you must be holding the outside of the bend at somewhere near
yield. If you pull it a bit further, you'll bend it some more, keeping
it at yield. But if you then relax the overload again, the parts you
just yielded can sink back to a bit below yield.
So the key thing is that you relax again after the momentary overload
used for stress relief.
The small amount of strain on the outside the elbow is what corresponds
to it being left at a lower stress when you relax the overload.
This is my understanding of the theory. Whether it's really that simple
I'm not so sure-- especially given the changes in geometry that are
happening all the time as the hub-hole deforms (unless it doesn't,
etc.).
And there are still some geometry questions even if you disregard
hub-hole deformation. Suppose you have ended up with question-mark
shaped hook in the spoke where it leaves the hub, as described earlier.
The momentary overload might be expected to reduce the hook-radius a
bit, and we're going to get some changes in strain around the hook. Will
we necessarily be sure to be left with no part of the hook at yield
stress? I'm not sure.
Tim McNamara
Peter Cole <peter...@comcast.net> wrote:
> jim beam wrote:
> > Peter Cole wrote:
> >> jim beam wrote:
>
> >>> there is a logical disconnect between this announcement that one
> >>> phase of yielding creates residual stress, and the supposition
> >>> that yet more of the same magically removes it again.
> >>
> >> No "disconnect", only stress relief.
> >
> > the logical disconnect is the vast gulf of unexplained uncertainty
> > between the two. want to put numbers to strain that induces relief
> > and strain that induces an increase?
>
> Who's talking "strain"? I think you still don't get it. Mechanical
> stress relief is simple, straightforward, and common in practice. It
> should be obvious that a nominal 50% increase in stress will only
> affect those regions that are already that close to yield -- whether
> those existing stresses are residual manufacturing stresses or static
> load stresses or a combination -- but this is just beating a dead
> horse.
Yes, the equine has been an ex-equine for many, many posts now. But jim
is sadly more interested in casting aspersions than offering substantive
proof. For some reason people keep trying to convince him, but it's
right up there with teaching pigs to sing.
jim beam
> jim beam wrote:
>> Peter Cole wrote:
>>> jim beam wrote:
>
>>>> there is a logical disconnect between this announcement that one
>>>> phase of yielding creates residual stress, and the supposition that
>>>> yet more of the same magically removes it again.
>>>
>>> No "disconnect", only stress relief.
>>
>> the logical disconnect is the vast gulf of unexplained uncertainty
>> between the two. want to put numbers to strain that induces relief
>> and strain that induces an increase?
>
> Who's talking "strain"?
you should be. it's strain [deformation] that's supposed to relieve
residual stress.
> I think you still don't get it. Mechanical
> stress relief is simple, straightforward, and common in practice. It
> should be obvious that a nominal 50% increase in stress will only affect
> those regions that are already that close to yield -- whether those
> existing stresses are residual manufacturing stresses or static load
> stresses or a combination -- but this is just beating a dead horse.
the theory is perfectly simple. what seems to be the conceptual hurdle
for engineers [who don't seem to study materials theory in any depth] is
the blind "belief" that it's something magical that always works to cure
ills, but never causes problems. /if/ it relieves stress, it's all well
and good, but if it increases it, it's not. data is required, not
supposition. it's like supposing that elasto-hydrodynamic separation
"protects" bike bearings. reality is, they rarely rotate fast enough -
data explodes the supposition.
>
>>>> testing does of course show that it can happen to some degree in
>>>> some conditions, but the "lab" conditions under which those are
>>>> observed are limited in both degree and duration - factors not
>>>> controlled in the outside world.
>>>
>>>
>>> OK, you've gone from rejecting the presence of residual stress in
>>> spokes to now claiming it's there but unimportant.
>>
>> the key word is "significant". read again and look for it.
>
> You're just hand waving. The residual stresses are there. They will bias
> the fatigue curve. End of story.
to repeat, relative to production quality and metallurgy, it's way down
the list. presumption isn't proof and /definitely/ not grounds on which
to disregard known fact.
> You want quantification? Quantify it
> yourself.
within my abilities, i have. i've stress corrosion tested. i have no
cracking. repeating misconceptions from a source that doesn't
understand the fundamentals of deformation or fatigue doesn't fly.
> It doesn't matter.
!!!
>
>>>> we also have to ask ourselves about the extent to which this really
>>>> is a quantifiable real-world problem. pressed steel wheels used on
>>>> cars for example are highly cold worked, have "residual stress" from
>>>> manufacture, are subject to the same old cycles of cyclic elastic
>>>> deformation due to high loads and extended mileage, yet hardly ever
>>>> evidence any fatigue at all, let alone fatigue that can be directly
>>>> attributed to residual stress. same for drawn tube, pressed
>>>> chassis, nervex lugs... presumptive nonsense that residual stress
>>>> is the single most important factor, and that it can be effectively
>>>> managed in an inconsistent and arbitrary fashion, betrays a deep and
>>>> fundamental ignorance of the whole subject.
>>>
>>> If you take the trouble to read up on this, you'll find that fatigue
>>> and residual stress are factors in steel automobile wheels.
>>
>> not in any significant way. i've seen many many thousands of these
>> things - not a single failure that can be ascribed to residual stress.
>
> Well I suppose that endorsement will be very bad news to the industry
> that makes fatigue testing equipment for steel automotive wheels and the
> designers who deliberately introduce residual stress to improve fatigue
> characteristics while reducing weight.
show service failures. post pics on the net so they can be seen by all.
show analysis that directly attributes these failures to residual
stress as opposed to the other more likely initiators.
jim beam
> On 2007-01-16, jim beam <spamv...@bad.example.net> wrote:
>> Peter Cole wrote:
>>> jim beam wrote:
>>>> Ron Ruff wrote:
> [snip]
>>>>> Would it be safe to say that over-stressing a built wheel as high as
>>>>> possible... which I guess would be to the yield stress of the "weakest"
>>>>> link in the body of the spoke; the threads or butted section... would
>>>>> be ideal? At least that would even out the stresses in the elbow as
>>>>> much as possible.
>>>>>
>>>> there is a logical disconnect between this announcement that one phase
>>>> of yielding creates residual stress, and the supposition that yet more
>>>> of the same magically removes it again.
>>> No "disconnect", only stress relief.
>> the logical disconnect is the vast gulf of unexplained uncertainty
>> between the two. want to put numbers to strain that induces relief and
>> strain that induces an increase?
>
> The logical connection is made by requiring subsequent relaxation after
> the momentary overload.
that can't tell us whether relief has occurred or not.
>
> The idea is that if you pull a wire such that it bends, and hold it
> there, you must be holding the outside of the bend at somewhere near
> yield. If you pull it a bit further, you'll bend it some more, keeping
> it at yield. But if you then relax the overload again, the parts you
> just yielded can sink back to a bit below yield.
of course, but that's applied stress...
>
> So the key thing is that you relax again after the momentary overload
> used for stress relief.
make that "stress relief". there's no presumption that residual stress,
that strain which remains trapped in the lattice after relaxation, is
relieved. we have fogel's experiments showing that it can be relieved
from recently bent spokes that are shown to have residual stress, but so
far, we only have presumption that factory spokes are delivered in this
condition, and also presumption that it is relieved.
>
> The small amount of strain on the outside the elbow is what corresponds
> to it being left at a lower stress when you relax the overload.
that's the /applied/ stress...
>
> This is my understanding of the theory. Whether it's really that simple
> I'm not so sure-- especially given the changes in geometry that are
> happening all the time as the hub-hole deforms (unless it doesn't,
> etc.).
i think we can all agree that /applied/ stress is reduced by "stress
relief". the problem remains with presumption that arbitrary process
can magically mitigate something whose existence is unproven. "there's
dragons behind that rock, but if you say the spell, they'll disappear".
so, you look behind the rock. were there dragons? unless you know
that, you have no idea whether the spell worked.
>
> And there are still some geometry questions even if you disregard
> hub-hole deformation. Suppose you have ended up with question-mark
> shaped hook in the spoke where it leaves the hub, as described earlier.
> The momentary overload might be expected to reduce the hook-radius a
> bit, and we're going to get some changes in strain around the hook. Will
> we necessarily be sure to be left with no part of the hook at yield
> stress? I'm not sure.
that's the kind of question to ask. presumption makes no sense.
Tim McNamara
jim beam <spamv...@bad.example.net> wrote:
> the theory is perfectly simple. what seems to be the conceptual
> hurdle for engineers [who don't seem to study materials theory in any
> depth]
As opposed to metallurgists who don't study engineering in any depth?
> data is required, not supposition.
So provide data not supposition.
Peter Cole
> Peter Cole wrote:
>> jim beam wrote:
>>>> If you take the trouble to read up on this, you'll find that fatigue
>>>> and residual stress are factors in steel automobile wheels.
>>>
>>> not in any significant way. i've seen many many thousands of these
>>> things - not a single failure that can be ascribed to residual stress.
>>
>> Well I suppose that endorsement will be very bad news to the industry
>> that makes fatigue testing equipment for steel automotive wheels and
>> the designers who deliberately introduce residual stress to improve
>> fatigue characteristics while reducing weight.
>
> show service failures. post pics on the net so they can be seen by all.
> show analysis that directly attributes these failures to residual
> stress as opposed to the other more likely initiators.
I really don't have the time to prepare a reading list for you. As I
said, if you take the trouble to look, there is a huge amount of
information available. There are large industries devoted to the
analysis and control of residual stress in many applications. The
principles are the same. I have seen nothing inconsistent in the
literature I have read with the descriptions of bicycle wheels in
Jobst's book. I find the subject matter interesting, so I have perused
the literature to get a deeper understanding. I encourage you to do the
same. There's a big world out there.
A couple of interesting recent finds:
"Intro" to residual stress concepts
http://www.protoxrd.com/pdf/rsintro.pdf
http://www.shotpeener.com/library/pdf/1952001.pdf
"Title: Effects Of Residual Stress On Rolling Bodies, Appendix I
Author: Almen, J. O.
Source: Elsevier Publ.,Roll.Contact Phenom., p. 400
Publication year 1952
Document number: 1952001
Number of pages: 25
Abstract: It is well known that almost all dynamically loaded parts in
our modern high-duty machines suffer from or profit by residual
stresses. Residual stresses are developed by processing operations such
as hot or cold forming, heat treatments, machining, grinding, polishing,
rolling, tumbling, shot peening, straightening and many more, including
local plastic yielding in the course of normal service. The causes of
residual stresses and their effects on fatigue failure and other
brittle-type failures are rapidly becoming better understood in academic
circles as well as in industry. The purpose of this paper is to show how
residual stresses are developed and how they affect the functioning of
compressively loaded rolling bodies, such as ball and roller bearings. D
uring the last thirty years I have repeatedly found convincing evidence
that rolling bodies, including those made of the hardest steel, undergo
extensive changes during normal service because of local plastic flow of
the compressively loaded metal. These changes necessarily caused
residual stresses most of which indicated deterioration of the part."
This one is particularly interesting, first because it's over 50 years
old, and second because it uses a method of cutting slots & measuring
strain to experimentally measure residual stress. I did the same
approximate thing (much more crudely) on a spoke elbow. There is no
doubt in my mind that there are significant residual manufacturing
stresses in spokes. There is also an interesting comparison between
railroad wheels and bicycle wheels.
If you look into automobile wheels, you'll see that the steel industry
has been working very hard to regain lost market share from the aluminum
industry. Weight reduction and large decorative/functional cutouts are
necessary to compete. Creative use of high strength steels and control
of design/manufacturing process to both exclude harmful residual
stresses and include beneficial ones is the focus. Fatigue lifetimes are
the limiting factor. I have not found a database of steel auto wheel
failures. There is plenty of literature on truck, railroad car, aircraft
and race car failures, analysis and testing methods.
jim beam
very [un]skillful deception peter. residual stress is well known and
understood. the point, and this is something you avoid repeatedly is
THE PRESUMPTION THAT IT'S A FACTOR IN BIKE SPOKE FATIGUE.
when i see a spoke fatigue fracture surface whose crack initiated from
the /compressive/ residual stress side of the spoke, i cannot *presume*
residual stress is a factor in initiation. and nor can you or anybody
else. doesn't anybody around here ever look at evidence from actual
failures? what's next, presumption that anodizing causes rim cracking -
when the rim crack is tangential to the anodizing crack?
Peter Cole
No need to shout, I thought you wanted to discuss steel automobile wheels.
> when i see a spoke fatigue fracture surface whose crack initiated from
> the /compressive/ residual stress side of the spoke, i cannot *presume*
> residual stress is a factor in initiation. and nor can you or anybody
> else. doesn't anybody around here ever look at evidence from actual
> failures?
As I sketched out earlier in this thread, residual manufacturing
stresses are one of 3 principal stresses at the elbow. The other 2 are
spoke tension and bending stress. The magnitude of the tensioning stress
is known, the residual manufacturing stress can be estimated, the only
wild card is the bending stress. If your spokes break predominantly at
the outside of the elbow it is highly suggestive that you have large
residual bending stresses. You need to correct your spoke line. The
simple superposition I sketched out shows that stress relief can only be
beneficial -- whether you have residual bending stresses or not, it
doesn't matter which direction they're in, either. Of course that's just
for the elbows, the threads & heads are even more straightforward.
All this is completely consistent with Jobst's description of stress
relief in his book. You should read it sometime.
> what's next, presumption that anodizing causes rim cracking -
> when the rim crack is tangential to the anodizing crack?
For someone who's so concerned about the effect of surface finish on
fatigue, I'd expect you to be much less tolerant of surface cracking.
For materials where most of the fatigue life is spent in the
"initiation" phase, it really doesn't matter if cracks, once formed,
preferentially propagate along extrusion weaknesses, it's the initiation
process that's important, something a hard brittle coating only accelerates.
Joe Riel
> what's next, presumption that anodizing causes rim cracking -
> when the rim crack is tangential to the anodizing crack?
Do you really mean tangential? Seems like orthogonal
fits better.
--
Joe Riel