Re: Spoke Failure--surface quality or stress relief?



Whatweartrack writes:

>> I stumbled across this series of pictures, diagrams, and
>> explanations about spoke fatigue and testing:


>> http://www.princeton.edu/~humcomp/bikes/design/desi_18.htm
>> http://www.princeton.edu/~humcomp/bikes/design/desi_19.htm
>> http://www.princeton.edu/~humcomp/bikes/design/desi_20.htm
>> http://www.princeton.edu/~humcomp/bikes/design/desi_21.htm
>> http://www.princeton.edu/~humcomp/bikes/design/desi_22.htm


>> It's taken from McMahon and Graham's "The Bicycle and the Walkman,"
>> Merion (1992) and indicates that testing shows that spokes crack
>> from the outside, beginning at surface flaws, not from the inside
>> where internal stresses would be located.


>> Does the testing procedure mimic real life, or does the conclusion
>> hold only for this kind of test?


> Jobst Brandt would have an authoritative answer, but my instant
> observation is that the mode of testing does not appear to resemble
> the type of stress cycle that a spoke is subjected to in actual
> usage. As such, I question whether it has any actual applicability
> to in-service failures. Personally, I don't feel that this is
> useful information.


Since spokes do not bend at this place, a bending test would not be
appropriate for such an investigation in my estimation. It would have
to be a preloaded oscillating tensile test and that takes millions of
cycles as it does on a bicycle.

>> The nipple end of a bicycle spoke in a real wheel, on the other
>> hand, undergoes tension changes, but doesn't do much bending.


> And in my experience, the few failures I've seen at the nipple end
> were clearly due to the spoke being stressed beyond the yield point
> of the threaded section; the failure was obviously caused by tension
> in excess of the spoke's load capacity. (I will note that I have
> yet to see a spoke break at that point without a spokejam being
> involved, also.)


I have seen many such failures and still do. In fact the one pictured
in #_18 above is classic. Such failures showed the greatest reduction
from stress relieving when Robergel spokes were the mainstay.

>> The elbow end of a bicycle spoke in a real wheel not only changes
>> tension, but presumably flexes a little, too.


> If properly stress-relieved, it flexes very little, and does so in a
> manner that does not approach its yield limit. There is no reason
> for it to break there absent a stress riser's presence.


Wellll, flex is a tension change except that it is a more concentrated
one than plain tension, because mainly the outer "fibers" of the bend
are affected. Stress reliving reduces the residual stress in these
fibers after the spoke has been bent into place and tensioned. I
think careful examination would show that the inside of the elbow bend
is also in tension in spite of having been bent into place.

>> I have no idea if the test was appropriate, so I'm hoping that some
>> engineers will look at it and try to explain what was going on. The
>> site and the book seem to be saying surface quality is what
>> matters, not the stress relief usually advocated here.


> From what I could see, the mode of testing that the apparatus
> appeared to be using looked like it was stressing the elbow in the
> opposite direction from what would actually be achieved in use. As
> such, if my understanding of the information is correct, the test is
> an academic exercise with no value in predicting or determining the
> cause of actual in-service failures.


I think the proper way to perform such a test is to build a flange
simulation and tensile cycle the spoke under customary operating
preload. I don't see any shortcuts.

[email protected]
 
On Sat, 09 Apr 2005 14:06:31 GMT, [email protected]
wrote:

>I think the proper way to perform such a test is to build a flange
>simulation and tensile cycle the spoke under customary operating
>preload. I don't see any shortcuts.
>
>[email protected]


Absolutely.

IMO, any test that does not closely model the in-service stress
patterns is not a test, it's a demonstration. Any test that doesn't
exactly duplicate in-service conditions isn't producing data that's
predictive of actual results that will be seen.
--
Typoes are a feature, not a bug.
Some gardening required to reply via email.
Words processed in a facility that contains nuts.
 
jim beam wrote:
> [email protected] wrote:
>> Whatweartrack writes:
>>
>>
>>>> I stumbled across this series of pictures, diagrams, and
>>>> explanations about spoke fatigue and testing:

>>
>>
>>>> http://www.princeton.edu/~humcomp/bikes/design/desi_18.htm
>>>> http://www.princeton.edu/~humcomp/bikes/design/desi_19.htm
>>>> http://www.princeton.edu/~humcomp/bikes/design/desi_20.htm
>>>> http://www.princeton.edu/~humcomp/bikes/design/desi_21.htm
>>>> http://www.princeton.edu/~humcomp/bikes/design/desi_22.htm

>>
>>
>>>> It's taken from McMahon and Graham's "The Bicycle and the Walkman,"
>>>> Merion (1992) and indicates that testing shows that spokes crack
>>>> from the outside, beginning at surface flaws, not from the inside
>>>> where internal stresses would be located.

>>
>>
>>>> Does the testing procedure mimic real life, or does the conclusion
>>>> hold only for this kind of test?

>>
>>
>>> Jobst Brandt would have an authoritative answer, but my instant
>>> observation is that the mode of testing does not appear to resemble
>>> the type of stress cycle that a spoke is subjected to in actual
>>> usage. As such, I question whether it has any actual applicability
>>> to in-service failures. Personally, I don't feel that this is
>>> useful information.

>>
>>
>> Since spokes do not bend at this place, a bending test would not be
>> appropriate for such an investigation in my estimation. It would
>> have to be a preloaded oscillating tensile test and that takes
>> millions of cycles as it does on a bicycle.
>>
>>
>>>> The nipple end of a bicycle spoke in a real wheel, on the other
>>>> hand, undergoes tension changes, but doesn't do much bending.

>>
>>
>>> And in my experience, the few failures I've seen at the nipple end
>>> were clearly due to the spoke being stressed beyond the yield point
>>> of the threaded section; the failure was obviously caused by tension
>>> in excess of the spoke's load capacity. (I will note that I have
>>> yet to see a spoke break at that point without a spokejam being
>>> involved, also.)

>>
>>
>> I have seen many such failures and still do. In fact the one
>> pictured in #_18 above is classic. Such failures showed the
>> greatest reduction from stress relieving when Robergel spokes were
>> the mainstay.
>>>> The elbow end of a bicycle spoke in a real wheel not only changes
>>>> tension, but presumably flexes a little, too.

>>
>>
>>> If properly stress-relieved, it flexes very little, and does so in a
>>> manner that does not approach its yield limit. There is no reason
>>> for it to break there absent a stress riser's presence.

>>
>>
>> Wellll, flex is a tension change except that it is a more
>> concentrated one than plain tension, because mainly the outer
>> "fibers" of the bend are affected. Stress reliving reduces the
>> residual stress in these fibers after the spoke has been bent into
>> place and tensioned. I think careful examination would show that
>> the inside of the elbow bend is also in tension in spite of having
>> been bent into place.
>>>> I have no idea if the test was appropriate, so I'm hoping that some
>>>> engineers will look at it and try to explain what was going on. The
>>>> site and the book seem to be saying surface quality is what
>>>> matters, not the stress relief usually advocated here.

>>
>>
>>> From what I could see, the mode of testing that the apparatus
>>> appeared to be using looked like it was stressing the elbow in the
>>> opposite direction from what would actually be achieved in use. As
>>> such, if my understanding of the information is correct, the test is
>>> an academic exercise with no value in predicting or determining the
>>> cause of actual in-service failures.

>>
>>
>> I think the proper way to perform such a test is to build a flange
>> simulation and tensile cycle the spoke under customary operating
>> preload. I don't see any shortcuts.

>
> for thread testing, this /is/ a perfectly legitimate shortcut. the
> tread root experiences cyclic axial tension, just like in a pure axial
> load mode.


Wouldn't the 5 to 10 degrees of deviation from straight axial loading
produce an uneven force on one side of the threads, the deviation being
caused by both spoke crossings and dish?

--
Phil, Squid-in-Training
 
Phil, Squid-in-Training wrote:
> jim beam wrote:
>
>>[email protected] wrote:
>>
>>>Whatweartrack writes:
>>>
>>>
>>>
>>>>>I stumbled across this series of pictures, diagrams, and
>>>>>explanations about spoke fatigue and testing:
>>>
>>>
>>>>>http://www.princeton.edu/~humcomp/bikes/design/desi_18.htm
>>>>>http://www.princeton.edu/~humcomp/bikes/design/desi_19.htm
>>>>>http://www.princeton.edu/~humcomp/bikes/design/desi_20.htm
>>>>>http://www.princeton.edu/~humcomp/bikes/design/desi_21.htm
>>>>>http://www.princeton.edu/~humcomp/bikes/design/desi_22.htm
>>>
>>>
>>>>>It's taken from McMahon and Graham's "The Bicycle and the Walkman,"
>>>>>Merion (1992) and indicates that testing shows that spokes crack
>>>>>from the outside, beginning at surface flaws, not from the inside
>>>>>where internal stresses would be located.
>>>
>>>
>>>>>Does the testing procedure mimic real life, or does the conclusion
>>>>>hold only for this kind of test?
>>>
>>>
>>>>Jobst Brandt would have an authoritative answer, but my instant
>>>>observation is that the mode of testing does not appear to resemble
>>>>the type of stress cycle that a spoke is subjected to in actual
>>>>usage. As such, I question whether it has any actual applicability
>>>>to in-service failures. Personally, I don't feel that this is
>>>>useful information.
>>>
>>>
>>>Since spokes do not bend at this place, a bending test would not be
>>>appropriate for such an investigation in my estimation. It would
>>>have to be a preloaded oscillating tensile test and that takes
>>>millions of cycles as it does on a bicycle.
>>>
>>>
>>>
>>>>>The nipple end of a bicycle spoke in a real wheel, on the other
>>>>>hand, undergoes tension changes, but doesn't do much bending.
>>>
>>>
>>>>And in my experience, the few failures I've seen at the nipple end
>>>>were clearly due to the spoke being stressed beyond the yield point
>>>>of the threaded section; the failure was obviously caused by tension
>>>>in excess of the spoke's load capacity. (I will note that I have
>>>>yet to see a spoke break at that point without a spokejam being
>>>>involved, also.)
>>>
>>>
>>>I have seen many such failures and still do. In fact the one
>>>pictured in #_18 above is classic. Such failures showed the
>>>greatest reduction from stress relieving when Robergel spokes were
>>>the mainstay.
>>>
>>>>>The elbow end of a bicycle spoke in a real wheel not only changes
>>>>>tension, but presumably flexes a little, too.
>>>
>>>
>>>>If properly stress-relieved, it flexes very little, and does so in a
>>>>manner that does not approach its yield limit. There is no reason
>>>>for it to break there absent a stress riser's presence.
>>>
>>>
>>>Wellll, flex is a tension change except that it is a more
>>>concentrated one than plain tension, because mainly the outer
>>>"fibers" of the bend are affected. Stress reliving reduces the
>>>residual stress in these fibers after the spoke has been bent into
>>>place and tensioned. I think careful examination would show that
>>>the inside of the elbow bend is also in tension in spite of having
>>>been bent into place.
>>>
>>>>>I have no idea if the test was appropriate, so I'm hoping that some
>>>>>engineers will look at it and try to explain what was going on. The
>>>>>site and the book seem to be saying surface quality is what
>>>>>matters, not the stress relief usually advocated here.
>>>
>>>
>>>>From what I could see, the mode of testing that the apparatus
>>>>appeared to be using looked like it was stressing the elbow in the
>>>>opposite direction from what would actually be achieved in use. As
>>>>such, if my understanding of the information is correct, the test is
>>>>an academic exercise with no value in predicting or determining the
>>>>cause of actual in-service failures.
>>>
>>>
>>>I think the proper way to perform such a test is to build a flange
>>>simulation and tensile cycle the spoke under customary operating
>>>preload. I don't see any shortcuts.

>>
>>for thread testing, this /is/ a perfectly legitimate shortcut. the
>>tread root experiences cyclic axial tension, just like in a pure axial
>>load mode.

>
>
> Wouldn't the 5 to 10 degrees of deviation from straight axial loading
> produce an uneven force on one side of the threads, the deviation being
> caused by both spoke crossings and dish?
>

absolutely. another reason why this is a legit shortcut.
 

Similar threads

M
Replies
63
Views
1K
D