On 7/11/2018 11:43 PM, Jeff Liebermann wrote:
On Wed, 11 Jul 2018 17:24:40 -0400, Frank Krygowski
wrote:


Exactly. Strong enough is strong enough.

OK, so let's pretend that the tube with the Rivnut bent at 10% less
tension. Is that "strong enough"? There's no way to tell without the
original design calculations, or reverse engineering the frame with an
FEA model. Too bad Autodesk killed their online ForceEffect web app.
http://blogs.autodesk.com/inventor/2017/01/17/autodesk-forceeffect-family-retirement/
I think I could have modeled the problem using the program.

Maybe this will work:
http://structural-analyser.com
Nope. Won't do tubing.

And BTW, the test you're describing would be much, much easier to do in
a proper tensile testing machine. Find an engineering student, get him
interested, have him get permission to do it as a class project, and
your data would be much better.

Yep, but my use of UCSC equipment has turned into a complex hassle.
I'll spare you the details, but at this time, it's not an easy
options.

I was wondering how I would do it on a proper machine. Probably
support the tube at the ends and push in the middle behind the Rivnut.
The problem with that is I'll probably crimp the tube where it's
pushing. It's likely I'll tear the tube before it bends. So, I'm
stuck with securing one end, and pulling (or pushing) on the other.

But on the other hand, tensile strength of the tube isn't really the
concern. The concern would be fatigue strength, and if we're talking
about the down tube, it would be under repeated, reversing torsional
stresses.

Good point. However, it will take too many tubing samples to test all
the possible combinations of forces available. Shall we keep it
simple and just bend a tube or two?

Well, it depends on which you want: Test results? Or good and applicable
data?


I strongly suspect that you'd find no significant difference. One
feature of the Rivnut is that its clamping action on the parent metal
applies compressive stress. Fatigue cracks start in regions of tensile
stress. The Rivnut may even make the object stronger.

Clamping action requires equal compression at all points around the
Rivnut hole. That's not going to happen in tubing where the Rivnut is
being crimped onto a curved surface. At the peak of the curve, there
will probably be plenty of compression force holding the Rivnut in
place. 90 degrees to either side, there may be an air gap with zero
compression force. In order to make it stronger on the curved surface
of the tubing stronger, the Rivnut would need a matching curve.

I agree there would be variations in compression around the
circumference of the hole. I doubt that they would be important.

Here's the situation with dozens of details of bicycle design: Because
of the geometric complexities, plus the uncertain and variable loads, we
can't be precisely sure of the stress levels or safety factors. We could
ensure nice high safety factors only by adding considerable weight, but
that's usually undesirable.

So what has happened in practice over the last 150 years? Effectively,
it's been evolution by trial and error. A framebuilder may try a new way
of fabricating (say) the connection of the seat stays to the main frame
tubes, one that saves two ounces. Others notice and wonder if it will
break. If it doesn't break, others copy it. If it breaks, it's not used
again. It's survival of the fittest designs.

These days, some can streamline the trial and error process by use of
FEA. But A) that usually happens only in big firms like Trek, Cannondale
or Specialized; and B) it's still normally done only for main design
features, not for details like Rivnuts.

How does this evolution process apply to Rivnuts? They're used a lot.
They work. They're acceptable. Anything more is Scharfian nonsense or
navel gazing.


--
- Frank Krygowski