Yet another broken spoke



On 2007-09-08, jim beam <[email protected]> wrote:
> Ben C wrote:

[...]
>> Of course the only reason you can ever yield a spoke with your hands is
>> by bending it. As you yield it some more, the configuration changes and
>> the moment is reduced. So it's difficult to over-yield it.

>
> but yield /introduces/ residual stress!!!


If you (a) bend a wire round a bolt the way Carl did for his heating
experiments, and then, having bent it, you (b) give it a further good
hard pull on both ends, and then relax that to (c) just enough pull to
hold the wire in place, won't you end up with fewer regions of high
stress than if you did (a) and (c) without (b)?
 
jim beam wrote:
> Ben C wrote:
>> On 2007-09-07, [email protected]
>> <[email protected]> wrote:
>>> Ben C? writes:

>> [...]
>>>> MP Since that location has tensile residual stress, tensile applied
>>>> MP mean stress from the spoke tension and bending,
>>>> Is the _applied_ stress on the inside of the elbow from spoke
>>>> tension and bending really tensile?

>> [...]
>>>> I don't understand that. I thought when you bent a wire you got tensile
>>>> stress on the outside of the bend and compressive on the inside?
>>> These loads tend to open the elbow angle so that causes tensile
>>> stress.

>>
>> Just to recap, because I thought this was (roughly) the picture:
>>
>> 1. I put an outbound spoke in. Its natural elbow angle is a bit too wide.
>> 2. I tighten it up, the elbow bends a bit, making the elbow angle
>> smaller. It wants to spring back, but it can't, because it's
>> installed in the wheel and held in place.
>> 3. This leaves applied stress that's tensile on the outside of the elbow
>> and compressive on the inside.
>> 4. Momentary overload and relaxation leaves a spoke with reduced
>> stresses.
>>
>> Do I have this (fundamentally) wrong?
>>
>> Perhaps the point is it's the other way around for an inbound spoke,
>> whose elbow gets opened a bit by being installed in the wheel.
>>
>>> As Mike mentioned above, springback makes the stress reverse
>>> from that during forming.

>>
>> Yes, I think I understand that part. That's residual stress from spoke
>> forming, not retained stress from wheel-building, as I understand it.
>> During wheelbuilding the spoke is not able to spring back, so an
>> outbound spoke remains in tensile stress on the outside and compressive
>> on the inside until you stress-relieve.

>
> only parts of it. read this from luns tee:
> http://groups.google.com/group/rec.bicycles.tech/msg/af080b93a59cca03
>
> most notably:
> "For a more severely bent wire, the yielded layers extend deeper,
> and the residual stress pattern becomes more like:
>
> cccTCttt "
>
> so here's the problem - that [simplified but useful] depiction shows
> where the residual stress profiles would be. if residual stress were
> causing fatigue, we would observe fatigue initiating at a "T" point.
> instead, we observe it initiating at /both/ "c"'s and "t"'s.
>
> "engineers" can argue all they want about what they think should be
> happening, but if observed facts tell a different story, it's just so
> much hot air.


I performed the experiment Luns suggested on the above thread and posted
my results:

http://tinyurl.com/356ru7

I think that was an "observed fact". Yours?


>
>>
>> After stress-relieving, the stresses may be the other way round again,
>> but more importantly, reduced in magnitude.
>>
>> It seems that residual stress from forming would be mitigated and/or
>> dwarfed in magnitude by retained applied stress from the build? So
>> perhaps residual stress from forming _is_ a red herring?

>
> truth is, outside of the lab and in carefully controlled environments,
> fatigue is *always* observed to initiate at surface defects. these can
> be from processing, corrosion, or even inclusions within the material.
> addressing each of these is observed to directly affect fatigue life.
> among these, electron microscopy shows inclusion content to be a
> significant fatigue initiator. removing inclusions is _proven_ to
> extend fatigue life considerably.


Everybody knows this stuff. Lots of us have had nicked spokes break in
mid-span. So what? Stress + flaw = failure. Film at 11.

>
> that's why spoke manufacturers spend lots of money on expensive vacuum
> degassed materials. if cheap materials could offer superior fatigue
> life by way of simple stress relief, you'd better believe they'd be used.


Vacuum degassing was big news in the 50's. It's a cheap bulk process,
common as dirt. What else have you got?
 
jim beam wrote:
> Peter Cole wrote:
>> jim beam wrote:


>>> you had to dig back /four years/ to find something that predated luns
>>> tee reminding us all of residual stress profiles?

>>
>> I did a quick review of Luns' posts, I might have missed what you were
>> referring to. Do any of these ring a bell (all responses to your posts)?
>>
>> " Quit your squirming."
>> " What a load of hooey. "
>> " Translation: you're pulling stuff out of thin air,"
>> " I knew this would be your next dodge."


>> Seems like he had a typical "jim beam" experience.

>
>
> why don't you post from the jobstian library instead? his is much more
> lengthy. you should try quoting that idiot krygowski some time too.
> he's an engineering professor you know. or at least, it picks up an
> engineering professor's paycheck.


Your reference to Luns, not mine.

>
> as for luns, he's made some valuable contributions. but he's made some
> mistakes too, most notably on the subject of strain aging materials. he
> holds back a lot of stuff too - spoke tensiometer math for example. but
> you don't quote that.


Not relevant to the current thread. How do I quote something he "holds
back" anyway? (just curious).


why not? too intent on being a ***** to bother?

Uh oh, nasty name time. Everybody knows what this means.
 
jim beam wrote:
> Peter Cole wrote:
>> Ben C wrote:

> <snip for clarity>
>
>>> I would have thought so, yes, unless you really overdo it.

>>
>> It's pretty impossible to overdo it, your hands aren't strong enough.

>
> eh? translation: "it yields to cause 'stress relief' but it doesn't
> yield if that interferes with my ability to argue."
>
> what a crock of ****.


Mechanical stress relief is a well known technique. Go back to the Luns
post where he explains it, or to any of the cites I have provided over
the *years* -- or just keep hurling vulgarities, whatever.
 
jim beam wrote:
> Peter Cole wrote:
>> jim beam wrote:
>>> Peter Cole wrote:
>>>> jim beam wrote:
>>>>
>>>>> why don't /you/? but i forget, you don't contribute. maybe that's
>>>>> a guilt thing since you're playing this game on your employer's
>>>>> time, not your own.
>>>>
>>>> That's none of your business.
>>>
>>> it's a damned good indicator of your fundamental dishonesty though.

>>
>> That the best you got?

>
> there's more. but let's just focus on the facts.


Then why introduce (ad hominem, no less) inventions?
 
Ben C wrote:
> On 2007-09-08, Peter Cole <[email protected]> wrote:
>> Ben C wrote:
>>> On 2007-09-07, Peter Cole <[email protected]> wrote:
>>> [...]
>>>> If the spoke has no bending moment (perfectly supported, perfect path),
>>>> the applied stress from spoke tension will be tensile (uniform) across
>>>> the cross section.
>>> I think I basically agree with this, though not absolutely. If the spoke
>>> is perfectly supported, but being pulled around a corner, there will
>>> surely still be a bending moment on parts of the spoke, but the distance
>>> component of that moment will never be greater than half the diameter of
>>> the spoke.
>>>

>> It's theoretically possible to have the elbow perfectly supported such
>> that there would be no moment.

>
> If the spoke goes round a corner and you pull on it, then the corner
> will resist the pull and you will have (I think) compression on that
> side-- the inside of the bend-- and tension on the outside. So that is a
> sort of moment, depending on where you measure things from.


Think of it as a bundle of fibers, give the corner a radius.



>
> If you have a bad surface covered uniformly with defects, but a well
> supported spoke you still might see fatigue starting on the outside of
> the bend where the tensile stress in use is the highest. I think this is
> what jim beam was saying in another post.


Sure, stress + defect = failure, see it all the time with nicked spokes.


>
>> In practice, there must be a little slop to lace with.

>
> Yes, and in practice nothing's ever perfect anyway.
>
> [...]
>>> But if you correct the spoke angle, you're changing things right? It was
>>> bent in the factory by a little piston or whatever, and wants to spring
>>> back, leaving it with residual stresses. Then you go and bend it again
>>> to make the elbow a bit bigger or smaller. What happens to the residual
>>> stresses from the factory?

>> Typically, you're correcting the spoke angle on the outbound spokes,
>> closing the angle slightly. I doubt that this reduces residual
>> manufacturing stresses.

>
> But isn't the residual manufacturing stress on the skin of the spoke
> on the outside of the elbow compressive? So tightening the angle will
> put a tensile stress there, overcoming the residual compressive stress
> and then yielding the material in a tensile way?


You correct the angle before you tension. In that case, it's just like
the mfr. bending, just a little more.


>
> [...]
>>>> By stress relieving, the residual is reduced to non-fatiguing levels.
>>>> But, if the spoke has (tensile) stress levels near yield in parts of
>>>> the cross section, those will be reduced as they are forced to yield
>>>> by the momentary overload -- whatever the source. It's a "can't lose"
>>>> proposition.
>>> I would have thought so, yes, unless you really overdo it.

>> It's pretty impossible to overdo it, your hands aren't strong enough.

>
> Yes I think so, as you're only going to yield the portions that are
> close to yield already. To yield a well supported spoke takes much more
> force.
 
Ben C? writes:

>>>> If the spoke has no bending moment (perfectly supported, perfect
>>>> path), the applied stress from spoke tension will be tensile
>>>> (uniform) across the cross section.


>>> I think I basically agree with this, though not absolutely. If the
>>> spoke is perfectly supported, but being pulled around a corner,
>>> there will surely still be a bending moment on parts of the spoke,
>>> but the distance component of that moment will never be greater
>>> than half the diameter of the spoke.


>>> Moments that small (assuming the force is in the range of normal
>>> spoke tensions) are low enough not to worry about-- they're not
>>> bringing anything anywhere near dangerously high stress levels for
>>> fatigue.


>> don't forget, virtually every fatigue failure there's ever been is
>> because of "unanticipated" factors. just because stress levels
>> aren't /thought/ to be high, *observed failures* tell us that there
>> /is/ bending sufficient to cause fatigue!!!


> Yes, although doesn't it depend what you mean by "cause"? There is
> always some bending unless you have a straight pull spoke. If the
> mean stress in that bending cycle is low then you should get a long
> life unless you've got very bad surface defects.


> On the other hand if the mean stress is high (e.g. because you have
> a big bending moment), but the surface is much better, you might
> also get a short life.


> It's quite believable to me that wear resulting in corrosion that
> nucleated fatigue was a big factor in Clare's failed galvanized
> spokes, even if the stress cycle on them wasn't too big. But if
> someone's nice new stainless DT spokes in an Al hub fail after a few
> 100km then retained stress from the build or a big moment at the
> elbow look more to blame.


> In both cases there's bending and in both cases presumably the
> fatigue starts somewhere. So the same basic mechanism, but it's
> reasonable to blame or suspect the surface quality in the first case
> and the high stresses in the second.


You may recall that this thread started with someone who had a fairly
new wheel with stainless spokes (of unknown brand) that got early
spoke failures. The suggestion was that loose spokes cause failure
and that this causes higher stress. Since then, the focus has drifted
through argumentative dodges to what stresses cause failure and the
nature of stress in spoke elbows.

How low tension causes spoke failure and whether the wheel in question
was loosely spoked is still unclear, as is the brand of spokes. Just
the same, the drifting subject allowed some spoke lore to re-emerge,
such as if more than one spoke breaks, replace all of them, and this
matter of low spoke tension causing failure. The cause of either of
these has not been reasonably explained. The galvanized spokes were
not shown so the substantial wear that causes failure was visible.

As often, this thread buried logical discourse in insults and chaff.

Jobst Brandt
 
Ben C? writes:

> Good point. I was careful though to say "retained stress from the
> build", not "residual stress" [from manufacture].


> The latter I have mostly put in the red herring bucket so far but
> keep an open mind as I am not an expert on these things.


I think you should review that opinion and how you came to that
conclusion. A permanent bend in a wire causes residual stress and the
one in a spoke is increased in as many as half the spokes when
building a wheel, as the elbow angle becomes acute from on obtuse
angle.

> Retained stress from the build, which really I would lump together
> with poor spoke line (i.e. bending moment at the elbow) is what I
> suspect may be a significant factor in some failures where good
> quality spokes are used.


That cannot be readily improved by stress relief, there bing
insufficient overload manually to improve the spoke line.

> If there is high stress in the elbow after the build it follows that
> there's enough moment present for spoke tension to maintain that
> stress. I'd be inclined to say that the moment is the real problem:
> it will allow a high-stress bending cycle in use, even if you
> stress-relieve, if stress relief doesn't also have the effect of
> also reducing that moment.


That's why you improve the spoke line and then stress relieve, so that
peak stress is far enough below yield, that fatigue failures do not
occur for a long time if ever. As I said, I have wheels that went
300,000 miles with spoke failures at chain gouges (not in the elbow)
of a few outbound spokes.

> But I think stress-relief probably _does_ reduce the moment by
> bending the elbow a bit more and by conforming it to the hub (_pace_
> Peter Cole's and Jobst's well-reasoned arguments why hub conformance
> is not likely to happen much after tensioning).


Forget about the moment, just tensile changes from load bearing are
enough to cause fatigue failure if the spoke is near yield, which it
is, because the elbow angle was yielded to an acute angle in spoke
lacing, and held there by subsequent tensioning.

Jobst Brandt
 
In rec.bicycles.tech Peter Cole writes:

>> truth is, outside of the lab and in carefully controlled
>> environments, fatigue is *always* observed to initiate at surface
>> defects. these can be from processing, corrosion, or even
>> inclusions within the material. addressing each of these is
>> observed to directly affect fatigue life. among these, electron
>> microscopy shows inclusion content to be a significant fatigue
>> initiator. removing inclusions is _proven_ to extend fatigue life
>> considerably.


> Everybody knows this stuff. Lots of us have had nicked spokes break
> in mid-span. So what? Stress + flaw = failure. Film at 11.


>> that's why spoke manufacturers spend lots of money on expensive
>> vacuum degassed materials. if cheap materials could offer superior
>> fatigue life by way of simple stress relief, you'd better believe
>> they'd be used.


> Vacuum degassing was big news in the 50's. It's a cheap bulk
> process, common as dirt. What else have you got?


Let me add that DT does not make its spoke wire which is made in
Sweden. DT cannot stress relieve their spokes for you because that
must be done after wheel tensioning. Manufacturers of spokes
anticipate spoke yielding in wheel building and for that reason make
elbows with slightly obtuse angles. Still we hear of riders who break
spokes, some of which are from reputable sources. This is not from
lack of vacuum degassing or cheap materials.

Jobst Brandt
 
On 2007-09-08, [email protected] <[email protected]> wrote:
> Ben C? writes:

[...]
>> It's quite believable to me that wear resulting in corrosion that
>> nucleated fatigue was a big factor in Clare's failed galvanized
>> spokes, even if the stress cycle on them wasn't too big. But if
>> someone's nice new stainless DT spokes in an Al hub fail after a few
>> 100km then retained stress from the build or a big moment at the
>> elbow look more to blame.

>
>> In both cases there's bending and in both cases presumably the
>> fatigue starts somewhere. So the same basic mechanism, but it's
>> reasonable to blame or suspect the surface quality in the first case
>> and the high stresses in the second.

>
> You may recall that this thread started with someone who had a fairly
> new wheel with stainless spokes (of unknown brand) that got early
> spoke failures. The suggestion was that loose spokes cause failure
> and that this causes higher stress.
>
> Since then, the focus has drifted through argumentative dodges to what
> stresses cause failure and the nature of stress in spoke elbows.


We can never flog that one enough.

> How low tension causes spoke failure and whether the wheel in question
> was loosely spoked is still unclear, as is the brand of spokes. Just
> the same, the drifting subject allowed some spoke lore to re-emerge,
> such as if more than one spoke breaks, replace all of them,


Is that bad lore? If a few spokes break because of lack of stress relief
(which according to you I think is the most likely reason) then isn't it
reasonable to suspect the wheel wasn't stress-relieved properly and so
other spokes will soon fail too?

> and this matter of low spoke tension causing failure. The cause of
> either of these has not been reasonably explained. The galvanized
> spokes were not shown so the substantial wear that causes failure was
> visible.


Two reasons were given why loose spokes might lead to failure: wear and
bending.

And yes, fair point, we haven't actually seen the worn galvanized spokes
but Clare provided a convincing description of them. I think Gene may
have reported similar problems in the past and has also described
movement in the spoke holes of loose spokes leading to wear.

The bending theory remains a theory.
 
On 2007-09-08, [email protected] <[email protected]> wrote:
> Ben C? writes:
>
>> Good point. I was careful though to say "retained stress from the
>> build", not "residual stress" [from manufacture].

>
>> The latter I have mostly put in the red herring bucket so far but
>> keep an open mind as I am not an expert on these things.

>
> I think you should review that opinion and how you came to that
> conclusion. A permanent bend in a wire causes residual stress and the
> one in a spoke is increased in as many as half the spokes when
> building a wheel, as the elbow angle becomes acute from on obtuse
> angle.


Isn't it the other way round? The residual stresses are compressive on
the exterior on outside of the bend (I thought) and tensile on the
exterior of the inside. So it's the elbows that become more obtuse for
which the building-bend adds to the manufacturing residual stress?

>> Retained stress from the build, which really I would lump together
>> with poor spoke line (i.e. bending moment at the elbow) is what I
>> suspect may be a significant factor in some failures where good
>> quality spokes are used.

>
> That cannot be readily improved by stress relief, there bing
> insufficient overload manually to improve the spoke line.
>
>> If there is high stress in the elbow after the build it follows that
>> there's enough moment present for spoke tension to maintain that
>> stress. I'd be inclined to say that the moment is the real problem:
>> it will allow a high-stress bending cycle in use, even if you
>> stress-relieve, if stress relief doesn't also have the effect of
>> also reducing that moment.

>
> That's why you improve the spoke line and then stress relieve, so that
> peak stress is far enough below yield, that fatigue failures do not
> occur for a long time if ever. As I said, I have wheels that went
> 300,000 miles with spoke failures at chain gouges (not in the elbow)
> of a few outbound spokes.


300K miles without the brake track wearing out? That is outstanding.

I don't think I've had anything last as long as 300K miles in my whole
life.

>> But I think stress-relief probably _does_ reduce the moment by
>> bending the elbow a bit more and by conforming it to the hub (_pace_
>> Peter Cole's and Jobst's well-reasoned arguments why hub conformance
>> is not likely to happen much after tensioning).

>
> Forget about the moment, just tensile changes from load bearing are
> enough to cause fatigue failure if the spoke is near yield, which it
> is, because the elbow angle was yielded to an acute angle in spoke
> lacing, and held there by subsequent tensioning.


Yes, but as I said before, if the spoke is near yield under normal spoke
tension _there must be a moment_. I don't see how you can have one
without the other.

That's why I think the moment is key: it's where big stresses at the
elbow, on the exterior, come from.

I'm not denying that residual stresses from manufacture will add to
those stresses in places.
 
On 2007-09-08, Peter Cole <[email protected]> wrote:
> Ben C wrote:

[...]
>> But isn't the residual manufacturing stress on the skin of the spoke
>> on the outside of the elbow compressive? So tightening the angle will
>> put a tensile stress there, overcoming the residual compressive stress
>> and then yielding the material in a tensile way?

>
> You correct the angle before you tension. In that case, it's just like
> the mfr. bending, just a little more.


Not according to my understanding of Jobst's earlier explanation. His
point was that the spoke line correction bend doesn't get a chance to
spring back because the spoke is installed in the wheel and the elbow
can't be overbent because the flange is in the way. So its stress will
remain tensile on the outside of the bend (outbound spoke).

Mfr bending on the other hand does get to spring back, which is why
you'd get compressive outside the bend and tensile inside.
 
Ben C? writes:

>>> It's quite believable to me that wear resulting in corrosion that
>>> nucleated fatigue was a big factor in Clare's failed galvanized
>>> spokes, even if the stress cycle on them wasn't too big. But if
>>> someone's nice new stainless DT spokes in an Al hub fail after a
>>> few 100km then retained stress from the build or a big moment at
>>> the elbow look more to blame.


>>> In both cases there's bending and in both cases presumably the
>>> fatigue starts somewhere. So the same basic mechanism, but it's
>>> reasonable to blame or suspect the surface quality in the first
>>> case and the high stresses in the second.


>> You may recall that this thread started with someone who had a
>> fairly new wheel with stainless spokes (of unknown brand) that got
>> early spoke failures. The suggestion was that loose spokes cause
>> failure and that this causes higher stress.


>> Since then, the focus has drifted through argumentative dodges to
>> what stresses cause failure and the nature of stress in spoke
>> elbows.


> We can never flog that one enough.


>> How low tension causes spoke failure and whether the wheel in
>> question was loosely spoked is still unclear, as is the brand of
>> spokes. Just the same, the drifting subject allowed some spoke
>> lore to re-emerge, such as if more than one spoke breaks, replace
>> all of them,


> Is that bad lore? If a few spokes break because of lack of stress
> relief (which according to you I think is the most likely reason)
> then isn't it reasonable to suspect the wheel wasn't stress-relieved
> properly and so other spokes will soon fail too?


From my experience it is without merit. I stumbled onto stress relief
when after a crash on a tour, I had to switch to a set of wheels that
broke spokes often, having large flange 36-spoke Campagnolo record
hubs. They promptly broke two spokes and having spare spokes in my
touring bag, I disassembled the Regina freewheel at a gas station that
lent me a hammer and punch to remove the outer cone to give access
to the spokes. (lots of loose bearing balls and two ratchet pawls)

After replacing the broken spokes and fearing that others would soon
fail, I "stress relieved" the wheels, breaking two more spokes that I
then replaced before repeating the spoke breaking exercise. It
occurred to me weeks later that no more spokes had broken and that the
wheels were performing reliably.

It was this experience that got me to thinking about stress relief and
to perform this operation on all wheels that I built. After that,
spoke failures were rare even though we were using Robergel and Stella
brand spokes, both having a poor finish and known to fail readily.

I did not replace all the spokes and got good service from the wheels.

>> and this matter of low spoke tension causing failure. The cause of
>> either of these has not been reasonably explained. The galvanized
>> spokes were not shown so the substantial wear that causes failure
>> was visible.


> Two reasons were given why loose spokes might lead to failure: wear
> and bending.


From the evidence reported, I do not understand the bending theory of
loose spokes. Bending under essentially no load is like holding a
spoke by one end and waving it back and forth manually.

> And yes, fair point, we haven't actually seen the worn galvanized
> spokes but Clare provided a convincing description of them. I think
> Gene may have reported similar problems in the past and has also
> described movement in the spoke holes of loose spokes leading to
> wear.


I don't believe that they were worn half way through or even more than
polishing the plating.

> The bending theory remains a theory.


Jobst Brandt
 
Ben C? writes:

>>> Good point. I was careful though to say "retained stress from the
>>> build", not "residual stress" [from manufacture].


>>> The latter I have mostly put in the red herring bucket so far but
>>> keep an open mind as I am not an expert on these things.


>> I think you should review that opinion and how you came to that
>> conclusion. A permanent bend in a wire causes residual stress and
>> the one in a spoke is increased in as many as half the spokes when
>> building a wheel, as the elbow angle becomes acute from on obtuse
>> angle.


> Isn't it the other way round? The residual stresses are compressive
> on the exterior on outside of the bend (I thought) and tensile on
> the exterior of the inside. So it's the elbows that become more
> obtuse for which the building-bend adds to the manufacturing
> residual stress?


When the bend is made, the main stress is tension on the outside of
the bend and that bend is increased when the outbound spokes get their
spoke line corrected, either by spoke tension or manually. In any
event, the outside of the spoke elbows are in tensile stress on those
spokes.

>>> Retained stress from the build, which really I would lump together
>>> with poor spoke line (i.e. bending moment at the elbow) is what I
>>> suspect may be a significant factor in some failures where good
>>> quality spokes are used.


>> That cannot be readily improved by stress relief, there bing
>> insufficient overload manually to improve the spoke line.


>>> If there is high stress in the elbow after the build it follows
>>> that there's enough moment present for spoke tension to maintain
>>> that stress. I'd be inclined to say that the moment is the real
>>> problem: it will allow a high-stress bending cycle in use, even if
>>> you stress-relieve, if stress relief doesn't also have the effect
>>> of also reducing that moment.


>> That's why you improve the spoke line and then stress relieve, so
>> that peak stress is far enough below yield, that fatigue failures
>> do not occur for a long time if ever. As I said, I have wheels
>> that went 300,000 miles with spoke failures at chain gouges (not in
>> the elbow) of a few outbound spokes.


> 300K miles without the brake track wearing out? That is
> outstanding.


I didn't say with the same rims. I replaced those several times and
at least once because the rim was damaged. I did not remove the
spokes from the hub in these rim changes. Not doing so is an
important feature so that the spokes remain in the places to which
they have come to fit.

> I don't think I've had anything last as long as 300K miles in my
> whole life.


That's 30 years of at least 10,000 miles per year. As I said, I
retired that rear wheel in good condition after I broke a right rear
dropout for the second time. Old 4-speed Campagnolo Record hubs have
a lot of axle overhang and break axles. In doing so, the dropout gets
flexed and finally breaks. I'm using a Shimano 7-speed hub now with
bearings at the outer ends.

>>> But I think stress-relief probably _does_ reduce the moment by
>>> bending the elbow a bit more and by conforming it to the hub
>>> (_pace_ Peter Cole's and Jobst's well-reasoned arguments why hub
>>> conformance is not likely to happen much after tensioning).


>> Forget about the moment, just tensile changes from load bearing are
>> enough to cause fatigue failure if the spoke is near yield, which
>> it is, because the elbow angle was yielded to an acute angle in
>> spoke lacing, and held there by subsequent tensioning.


> Yes, but as I said before, if the spoke is near yield under normal
> spoke tension _there must be a moment_. I don't see how you can
> have one without the other.


A spoke that is fully supported in the aluminum flange is loaded
purely in tension as it wraps around a curved bore that matches its
inner radius. The spoke head should rest solidly on the entrance to
the flange and the spoke pulled around the corner. Such a spoke will
last a long time. The straight pull spoke came along because people
were not seeing the effect of reasonable flange bores with reasonable
flange thickness and stress relieving.

> That's why I think the moment is key: it's where big stresses at the
> elbow, on the exterior, come from.


There is no moment in a fully supported elbow and if you look at a
flange bore after removing the spoke you should see a smooth curve
that matches the spoke elbow shape.

> I'm not denying that residual stresses from manufacture will add to
> those stresses in places.


Spoke elbows should not be operating in bending mode but rather
tensile mode.

Jobst Brandt
 
Ben C? writes:

>>> But isn't the residual manufacturing stress on the skin of the
>>> spoke on the outside of the elbow compressive? So tightening the
>>> angle will put a tensile stress there, overcoming the residual
>>> compressive stress and then yielding the material in a tensile
>>> way?


>> You correct the angle before you tension. In that case, it's just
>> like the mfr. bending, just a little more.


> Not according to my understanding of Jobst's earlier explanation.
> His point was that the spoke line correction bend doesn't get a
> chance to spring back because the spoke is installed in the wheel
> and the elbow can't be overbent because the flange is in the way.
> So its stress will remain tensile on the outside of the bend
> (outbound spoke).


> Mfr bending on the other hand does get to spring back, which is why
> you'd get compressive outside the bend and tensile inside.


By pressing outbound spokes against the flange the space between the
elbow bend and the flange is reduced to a small gap that is closed by
tension. There is no cyclic motion any more as there is with a
stand-off spoke and the stress is lowered by stress relief.

Jobst Brandt
 
On 2007-09-09, [email protected] <[email protected]> wrote:
> Ben C? writes:

[...]
> After replacing the broken spokes and fearing that others would soon
> fail, I "stress relieved" the wheels, breaking two more spokes that I
> then replaced before repeating the spoke breaking exercise. It
> occurred to me weeks later that no more spokes had broken and that the
> wheels were performing reliably.


If a spoke isn't stress relieved, and run at a high mean stress stress
cycle for some time, and then stress-relieved just in the nick of time
before it actually breaks, what is its life time from that point
expected to be? In theory?

Not rhetorical question.

[...]
>> Two reasons were given why loose spokes might lead to failure: wear
>> and bending.

>
> From the evidence reported, I do not understand the bending theory of
> loose spokes. Bending under essentially no load is like holding a
> spoke by one end and waving it back and forth manually.


The rim is pushing the spoke towards the hub. First it (a) wobbles a
bit, and then, (b) not being a stiff rod, it may flex one way or
another. (c) It's supported at the elbow so it might bend there.

jim beam has also suggested the spoke crossings may be involved although
I'm less clear on the details so won't try to explain that.

(b) may not happen depending on the magnitude of (a)-- I believe that's
what you said. So I don't know if (b) or (c) happens.

That's all there is to it. That's all the reasoning I know of. There's
no evidence.

>> And yes, fair point, we haven't actually seen the worn galvanized
>> spokes but Clare provided a convincing description of them. I think
>> Gene may have reported similar problems in the past and has also
>> described movement in the spoke holes of loose spokes leading to
>> wear.

>
> I don't believe that they were worn half way through or even more than
> polishing the plating.


Isn't a galvanized spoke coated with a kind of zinc stuff to stop it
rusting? If that wears away the steel underneath starts to rust a
bit. A small rust spot is enough to nucleate fatigue. So the spoke
doesn't wear all the way through, or rust all the way through, but a bit
of wear is enough to make it fatigue quickly.
 
On 2007-09-09, [email protected] <[email protected]> wrote:
> Ben C? writes:

[...]
>> Isn't it the other way round? The residual stresses are compressive
>> on the exterior on outside of the bend (I thought) and tensile on
>> the exterior of the inside. So it's the elbows that become more
>> obtuse for which the building-bend adds to the manufacturing
>> residual stress?

>
> When the bend is made, the main stress is tension on the outside of
> the bend


Yes but that bend springs back, leaving residual compressive stress on
the exterior on the outside?

> and that bend is increased when the outbound spokes get their
> spoke line corrected, either by spoke tension or manually. In any
> event, the outside of the spoke elbows are in tensile stress on those
> spokes.


Yes, the applied stress is on the outside of the bend on the exterior on
the outbound spokes.
[...]
>> Yes, but as I said before, if the spoke is near yield under normal
>> spoke tension _there must be a moment_. I don't see how you can
>> have one without the other.

>
> A spoke that is fully supported in the aluminum flange is loaded
> purely in tension as it wraps around a curved bore that matches its
> inner radius. The spoke head should rest solidly on the entrance to
> the flange and the spoke pulled around the corner. Such a spoke will
> last a long time. The straight pull spoke came along because people
> were not seeing the effect of reasonable flange bores with reasonable
> flange thickness and stress relieving.
>
>> That's why I think the moment is key: it's where big stresses at the
>> elbow, on the exterior, come from.

>
> There is no moment in a fully supported elbow and if you look at a
> flange bore after removing the spoke you should see a smooth curve
> that matches the spoke elbow shape.


That's what one would hope for in a good wheel, certainly.

But if there's no moment there's no stress there from the spoke line
correction bend. Because there's nothing there to hold that bend in
place.

The last time we went through this it went slightly differently.

Outbound spoke
--------------

1. You said we bend the elbow to correct the line, it can't spring back,
so the bend is held there, and so the stress remains high on the
outside. I call this "retained build stress" to avoid confusion with
manufacturing residual stress.
2. I said for spoke tension to hold the bend in place needs sufficient
moment. If the spoke is flush to the flange you don't have that
moment.
3. We repeated (1) and (2) to each other approximately 300 times.
4. Eventually you said yes but what do you mean by "flush"? It's never
100% completely totally and utterly flush. We get a little
question-mark shaped hook, the details of which are complicated.
5. I said OK so there might be a moment, retained stress might be there,
and stress-relief might get rid of it.

But now you're saying "there's no moment". In that case there is no
retained build stress on the outside of the bend. 100kgf or so _with no
moment_ is not enough to hold a bend in a spoke.

Really they're the same thing. Residual stress (the kind that remains
inside a wire after it's been bent and sprung back) requires knowledge
of history. But retained build stress is just applied stress. It doesn't
matter how it got there: if spoke tension is 100kgf and there's a
too-big radius, you have a moment, and too-high stress on the elbow
outside. No moment, nothing to worry about. If you stress-relieve and it
makes the stress go away then it necessarily also made the moment go
away.

So saying stress-relief is a process of helping things gain intimate
conformity of some kind (as jim beam has said) and saying that it yields
the bits of the cross section that are still close to yield from spoke
line correction (as you have said) pretty much amount to the same thing.

As for residual manufacturing stress-- it sounds like that might add a
bit to the tensile skin stress on the inside of the bend of the inbound
spokes. So it may have a contribution to failures of inbound spokes that
begin on the inside. But there's no reason to believe it's significant:
Even if we accept the evidence that stress-relief practice reduces
failures we don't know how much of that is to do with reducing retained
build stress as opposed to reducing manufacturing residual stress.

Since we're bending the elbow again anyway I'm inclined to think
manufacturing residual stress is basically ancient history by the the
time we've finished the wheel.
 
[email protected] wrote:
> In rec.bicycles.tech Peter Cole writes:
>
>>> truth is, outside of the lab and in carefully controlled
>>> environments, fatigue is *always* observed to initiate at surface
>>> defects. these can be from processing, corrosion, or even
>>> inclusions within the material. addressing each of these is
>>> observed to directly affect fatigue life. among these, electron
>>> microscopy shows inclusion content to be a significant fatigue
>>> initiator. removing inclusions is _proven_ to extend fatigue life
>>> considerably.

>
>> Everybody knows this stuff. Lots of us have had nicked spokes break
>> in mid-span. So what? Stress + flaw = failure. Film at 11.

>
>>> that's why spoke manufacturers spend lots of money on expensive
>>> vacuum degassed materials. if cheap materials could offer superior
>>> fatigue life by way of simple stress relief, you'd better believe
>>> they'd be used.

>
>> Vacuum degassing was big news in the 50's. It's a cheap bulk
>> process, common as dirt. What else have you got?

>
> Let me add that DT does not make its spoke wire which is made in
> Sweden. DT cannot stress relieve their spokes for you because that
> must be done after wheel tensioning. Manufacturers of spokes
> anticipate spoke yielding in wheel building and for that reason make
> elbows with slightly obtuse angles.


no, the angle is set in anticipation of hub geometry, not deformation by
the wheelbuilder. indeed, sapim specifically advise /not/ bending their
spokes.

http://www.flickr.com/photos/38636024@N00/327722444/

basic geometry gives the spoke exit angle at ~95 degrees - exactly what
spoke manufacturers use.

> Still we hear of riders who break
> spokes, some of which are from reputable sources. This is not from
> lack of vacuum degassing or cheap materials.
>


and cheapo chinese spokes break in a few hundred miles. quality spokes
with decent materials, many thousands. /that/ is the effect if
materials and manufacture, not "stress relief".
 
[email protected] wrote:
> Ben C? writes:
>
>>>> But isn't the residual manufacturing stress on the skin of the
>>>> spoke on the outside of the elbow compressive? So tightening the
>>>> angle will put a tensile stress there, overcoming the residual
>>>> compressive stress and then yielding the material in a tensile
>>>> way?

>
>>> You correct the angle before you tension. In that case, it's just
>>> like the mfr. bending, just a little more.

>
>> Not according to my understanding of Jobst's earlier explanation.
>> His point was that the spoke line correction bend doesn't get a
>> chance to spring back because the spoke is installed in the wheel
>> and the elbow can't be overbent because the flange is in the way.
>> So its stress will remain tensile on the outside of the bend
>> (outbound spoke).

>
>> Mfr bending on the other hand does get to spring back, which is why
>> you'd get compressive outside the bend and tensile inside.

>
> By pressing outbound spokes against the flange the space between the
> elbow bend and the flange is reduced to a small gap that is closed by
> tension.


only on heads-in spokes, and only if the spoke/flange geometry permits.
heads-out spokes have no such gap closing.

> There is no cyclic motion any more as there is with a
> stand-off spoke and the stress is lowered by stress relief.


yes there is - by definition. the spoke elbow is offset from its axis
and the spoke shaft is not rigidly attached to the hub flange, even if
it touches. the elasticity that you happily cite in other situations as
causing fretting is still at work here - and elastic distortion means
stress cycling at the bend - by definition. stress cycling at the bend
means fatigue. and examination of the fracture surface tells you the
whole story. you should try it some time.
 
Ben C wrote:
> On 2007-09-08, jim beam <[email protected]> wrote:
>> Ben C wrote:

> [...]
>>> Of course the only reason you can ever yield a spoke with your hands is
>>> by bending it. As you yield it some more, the configuration changes and
>>> the moment is reduced. So it's difficult to over-yield it.

>> but yield /introduces/ residual stress!!!

>
> If you (a) bend a wire round a bolt the way Carl did for his heating
> experiments, and then, having bent it, you (b) give it a further good
> hard pull on both ends, and then relax that to (c) just enough pull to
> hold the wire in place, won't you end up with fewer regions of high
> stress than if you did (a) and (c) without (b)?


maybe, maybe not. it's easy to introduce _more_ residual stress. and
if "stress relief" is not precise and controlled, what hope does the
average joe wheelbuilder have of getting it right? absent data [which
of course will never be provided because it poses a risk of destroying
the legend], we'll never know.