In article ,
says...
In article , Ned Simmons
says...

First, like you, I recalled the NPT thread as sharp, which
is not the case. It's truncated, though obviously not
nearly as much as the UN thread. Neither requires, though
both permit, a radius in the root.

Here of course the *peaks* of the male thread, and
likewise any part of the female thread, don't count.
What only counts is the most highly stressed point,
which is the root of the male thread.

Agreed. Both the crest and root are truncated, though only
very slightly.


Socket caps over 2", and nonstandard lengths, have cut
threads. Even socket cap screws are allowed to have certain
"discontinuities" (laps and seams) in the thread, though
not below the pitch diameter.

So there is likely to be a fair amount of variation
in the thread quality. That makes sense.

For a ductile material like low carbon steel, stress
raisers have little effect on static strength.

There's a lot of substance in that last statement,
and I'm not sure I understand all of it. My impression
is that in any kind of material, including low carbon
steels, microscopic tears and fractures do reduce the
static strength. Your statement contradicts that
impression and I would be interested in following the
reasoning behind it.

And I can't claim to completely understand either.
Presumably it only applies to ductile materials because
they have the ability to deform locally, thus reducing the
stresses without fracturing.


They do have
a marked effect on fatigue, though the stress
multiplication factor used for figuring fatigue (Kf) is
often less than for static stress concentration.

Here I was thinking of the two strength-reducing
mechanisms (stress risers, and the sharp-V stress
concentrating feature) as being two separate but related
mechanisms. Maybe not. But I do know that stress
concentrations from non-optimal geometry - like sharp
inside corners or sharp V thread forms *do* reduce
the ultmate strength in all kinds of materials.

I found this out when making some bolts out of
SP-1 Vespel.

I think ceramics would be an extreme example of this. I've
never used Vespel, isn't it similar to Ultem? If so, I'm
not surprised it would be sensitive to sharp threads. I'd
expect something like nylon to be less so.


The endurance limit for carbon steels varies from 25 to 75
ksi.

Using the 3/4 pipe from before, and assuming a Kf of 3, the
joint should not fatigue below about .138 x 25000 / 3 =
1150 lbs. The 25 ksi is pretty conservative. Kf=2~3 seems
typical for cut threads in low strength fasteners. The
allowable load could be higher if you could predict the
number of cycles and/or avoid strain reversals.

Here I would try to say that the factor of 3 I was proposing
was present even in static load. That's the sort of number
I've seen in practice. Not a fatigue factor, but due
purely to stress concentration.

Determining stress concentration factors appears to involve
a lot of voodoo and empirical formulas. Most of the values
I found for the stress concentration factor for fatigue
seemed to come from experimental results and were a
function of material as well as geometry. Roark has pretty
extensive formulas for simple elastic stress concentration
for various geometries.


Failure under static load would be closer to .138 x 60000 =
8280 lb.

While the lights were still out, I did a quick experiment.
I threaded a length of 3/8 pipe from the scrap bin thru a
hollow 12 ton porta-power cylinder and put some heavy
washers over the pipe, then screwed a couple random CI
fittings onto the ends. Upon pumping up the cylinder,
initial yield occurred at about 5400 lb tension; tensile
failure was at approx 6800 lb. The failure was exactly
where we both predicted, where the threads exit the
fitting. This corresponds to 57 ksi and 72 ksi
respectively. The material spec for black pipe is 35 ksi
yield and 60 ksi tensile, minimum.

Did you apply any correction for stress concentration
to go from the 6800 lb to get to the 72 ksi number?
If not then it may be that either a) it's not as big
as I recall, b) the threads you are using are well
formed (rolled?) or c) maybe the black iron pipe was
steel?

72 ksi is simply force/cross sectional area at the root of
the thread (6800 lb/.094 in^2).

I just took a look at the threads with a magnifier and
compared them to an NPT gage. They're pretty rough looking,
obviously die cut, though they do appear to have a slightly
wider flat at the root than the gage.

Black iron is a misnomer, it really is steel.


Interesting experiment. It would be easy to duplicate it
for some home depot cut thread 3/4 inch pipe, and give
this guy a *real* number. The failure you observed
where the root of the thread crosses the top line of the
female thread is a classic, this is where all the stress
shows up. One reason why designers take pains to avoid
putting *other* stress concentration features in line
with that surface in bolted-up assemblies.

Thanks for taking the time to do this, and to post up
the results. Fascinating.

I learned a lot myself.

Ned