....

Note that stress is the distribution of force in a material.
This discussion requires that one address the stress in
the materials, not just the applied force. 'Load' usually
refers to force, but may also refer to the stress in the
material that results from that force. I have tried to
use the terms force and stress properly, but may have
slipped up. If so, I apologize in advance.

Some of the previous discussion has addressed fastening
thin materials, like sheet metal while other parts have
addressed thicker sections like the joint of a prop to
a prop hub.

Some of the preceding folks have stated perfectly valid
examples, but of mechanically different structures.

On Aug 25, 12:28 pm, wrote:
On Aug 24, 1:38 pm, Gunny wrote:

...

... t the reason fatigue isn't much of a
problem for the rivets in the aircraft skin is because the friction
between the joined surfaces typically carries the cyclic loads from
engine vibration (See "Riveted Joints", Chris Heintz, P.E.). I won't
speak to use in aircraft, but in general construction friction is
often considered a working part of the structure.

In fact, there are many instances in steel structures where service
loads are transmitted purely by static friction - moment connections,
end restraints for slender columns, connections with slotted/oversized
holes to facilitate assembly. Bearing/shear of the bolts is obviously
checked, but day-to-day the loads in those structures are transmitted
via static friction between the members. By design. AISC references
these as slip-critical connections. HSFG (High Strength Friction Grip)
is another term. Due to construction methods and tolerances, those
connections may only have one bolt out of the whole group that is
technically "bearing", maybe none. My point is that friction as a
mechanism for transferring loads to a wood prop is not really all that
unique or unusual as an engineering concept.

To address an earlier part of the thread, however, I wouldn't count on
friction for a wood-spar attach fitting. The fittings are often made
from thin material. Out-of-plane bending prevents the fitting from
developing much friction away from the bolt holes. And you have
humidity changes constantly modifying your wood dimensions. Tried-and-
true phenolic bushings, match drilled and reamed to the fittings, cost
about a dollar per hole. In the plane I'm building, that is less than
$50, so it was an easy choice to make.


Perhaps one should also check the bolt tension frequently.

Another statement that doesn't sit well was the reasoning that a pre-
tensioned bolt has better fatigue characteristics because metal
fatigues less when the stress cycle is all in tension as compared to
stress reversal. This is a clear misunderstanding of the factors in
play. Study the S-N diagrams of these materials and you will see that
increasing the mean stress decreases the fatigue life for a given
stress cycle amplitude. The reason some pre-tensioned bolted
connections (esp. shear) have better fatigue characteristics is
because the cyclic portion of the load is transfered via friction. The
bolt actually experiences a drastically reduced or eliminated cyclic
stress, thereby extending it's fatigue life even though the mean
stress of the bolt is much higher. Tension connections see improvement
through a different mechanism, but the result is the same - reduced
cyclic stress in the bolt and increased fatigue life.


Correct me if I am mistaken but here Matt is giving us an example
of a bolt that is preloaded in tension and then stressed (cyclically)
in tension.

That is different from a bolt that is pre-loaded in tension to fasten
two surfaces and then subject to shear, right?


... Anyone reading this
thread looking for info will find the correct way to construct a wing
joint.
As to whether or not the engineers I talked to were aircraft
engineers, most definately they are.

Then I submit that there is a discontinuity in communication in
the loop from here to your friends and back to here. We are not
all discussing exactly the same things.

As to some of your comments, I need to clarify some things. If
you are a civil engineer that deals with steel structures, and you
have design and analysis standards that use friction to qualify
structure, then that is your way to do it. I don't recall seeing a
major building , bridge, etc, that wasn't either riveted or welded
together, but I don't know for sure.

Matt made it clear that bolted and riveted structures typically rely
on friction from the clamping force of the fasteners so that the
fasteners themselves typically see very little shear. I believe that
is correct. In particular, and Matt alluded to this problem, imagine
the precision required to evenly distribute the transverse shear
stress over many fasteners over a large surface, and then to
maintain that distribution over changing loads, thermal expansion,
etc.

So I will take your word for it.
I stated in my first post that friction existed and carried load, but
simply that for aerospace structures it is never counted on to carry
load. You only consider friction when it works against you. That I
know is true. In your statements about why using friction in the wood
spar joint is not a good idea, I think you have begun to uncover some
of the reasons why it is true.

I suggest that relying on the bolts to carry the entire load
without ANY load being carried by friction between the wing
attachment fittings and the wood will concentrate the stress
at the locations of the fasteners. This may well locally stress
the wood to failure, e.g. it may split. The clamping force of
the fitting distributes that load over a larger area reducing the
stress concentrations.

While it is essential that the fasteners be sized to safely carry
the entire load, flying with wing attachment fittings that are so
loosely clamped that they actually DO carry the load is a
likely route to inclusion in an NTSB report.

Wood is anisotropic in its properties. Whatever their other
merits, woods commonly used in aircraft construction for the
most part do not include those with interlocking grain, meaning
that they split easily. To avoid splitting, you want to minimize
tensile across or shear stress along a grain boundary. If you
drill a hole in a piece of wood and apply a load to something
sitting loosely in that hole has much the opposite effect.

Since most airframes are thin shell
material, most of these reasons apply just as well to metal as wood.

It is precisely BECAUSE most airframes are thin shell material that
rivets and bolts seldom carry all of the shear at a typical joint.

Imagine two long flat strips of sheet metal laid end-to-
end, but overlapped at their ends. Now drill (and ream!) through
both and bolt or rivet them together. Now pull on the ends.
Simple stress analysis assumes infinite stiffness, that is
it assumes that deformation of the part does not redistribute
the applied loads. For thin materials like aircraft that assumption
is often inapplicable (e.g. bucking is important). But for
this example we assume the sheet metal strips do not bend.

So, the sheet metal strips are both loaded in pure tension.
Does the bolt or rivet now carry the entire load in shear?
Not if it is properly installed! The bolt or rivet clamps the two
pieces of sheet metal together so that the friction between
them does not allow them to move relative to each other.
Since they do not slide accross each other, they carry
some of the load. The shear stress in the joint is equal
to the force in tension divided by the area of overlap.

Now as we increase the tension in the sheet metal from
zero to some higher value, the shear accross the bolt
increases, from zero to some higher value,
but only slightly because much of that load is carried
in the sheet metal. If the cross-sectional area of the
bolt is only one-tenth of the area of overlap, then the
bolt only carries one tenth of the shear stress.

Now if we relax the assumption of infinite stiffness then
the clamping force the fastener applies to the sheet
metal is maximum under the bolt head and rapidly drops
off away from the bolt. But the bolt will still share quite
a bit of the shear with the material being clamped. The
sheet metal will also bend (buckle) near the bolt putting
some additional tension on it but let's continue to ignore
that.

Now, let's also relax the pre-loading in the bolt :-)

The bolt now carries the entire shear load.
The sheet metal will also bend (buckle) near the bolt
putting some tension on it but let's continue to ignore that.

The bearing stress on the inside of the hole in either
piece of sheer metal is zero over 180 degrees of its
cirrcumfrence, and rises to a maximum at a point
centered opposite the unstressed arc. Since this is
thin sheet metal it is easy to see that the force needed
to raise that bearing stress above yield is small relative
to the force needed to yield any part of the properly
fastened joint. IOW, if the bolt is not tight enough,
the hole will become elongated. Not good. A hole
drilled in a thicker but softer material would also
elongate.

Using bushings in a hole drilled in wood helps to reduce
that elongation by spreading that bearing stress over
a larger area in the wood, and is a lighter approach than
simply using a larger bolt. But it is still not a substitute
for maintaining the proper tension in the bolts.

I am far from clear on what constitutes 'proper' for
a metal piece bolted to wood. If the wood is clamped
too tightly, an increase is humidity may overstress it
causing it to split. If too loose, a decrease in humidity
may cause the joint to loosen too much.

As to the statement that I clearly don't understand the factors
involved, you clearly do not understand what I said, the nature of
preloaded bolts, or even the S-n curves themselves. Improved fatigue
life due to preloading has nothing to do with friction. Friction may
improve fatigue life in the real world by spreading load over a larger
area, but the benefit of preloading on fatigue life is due primarily
to an effect that exists even if no friction is present at all.

Could you please elaborate on the theory of that effect?
E.g. is this a result of the superposition of stresses?

It has been twenty years since I did any serious stress analysis
so I'm not about to elaborate on the superposition problem but
I will point out that as a purely practical matter any properly
torqued bolt will share shear loading with at lest the material
clamped between the bolt head and the nut, and if that material
is not strong enough to bear a significant load then it will
fail before the bolt does.

Why
you think I need it pointed out that higher stress levels result in
shorter fatigue life is puzzling. Of course the higher the load you
place on a structure, the fewer cycles it will survive before failure.
What is hard to understand about that? What you apparently don't
understand is what constitutes a load cycle, how much is the load, and
what preload does to that. Preloading the bolt reduces the cyclic load
that it sees, since the load in a preloaded bolt only increases about
10% until the applied load exceeds the preload. When the prop bolts
are allowed to lose their preload, the full applied load becomes the
amount of cyclic load that causes fatigue. This is best demonstrated
by giving an example. Take two identical bolts, having a breaking
strength of 5,000 lbs each, and preload one to 2000 lbs, and none to
the other.

Here I presume the preload to which you refer is 2,000 lbs of axial
tension in the bolts. If these are 1/4" bolts that imposes a stress
of about 41,000 psi, implying that their ultimate strength is about
100,000 psi which, IIRC, is in the achievable range for high strength
steel.

If we now begin to subject both bolts to the same cyclic
loading of 1500 lbs, where the applied load is increased from 0 up to
1500 and then reduced to zero again, the bolt with the 2000 lb preload
will see a cyclic load of only about 150 lbs, whereas the un-preloaded
bolt will see a cyclic load of 1500 lbs, and will obviously fail much
sooner.

Here you temporarily lost me because you have not told us
HOW the bolt is loaded. If the load consists of additional
tension, then plainly the bolt will see cyclical stress over the
range of 3500 lb to 2000. That is clearly the type of loading Matt
was discussing. If I make the unremarkable assumption that y
ou are familiar with addition then clearly you are NOT assuming
that the load is applied in the form of additional tension.

However, the clamping force will still cause the shear to be
distributed over the surface area being clamped and not just
through the bolts. The superposition of stresses is not
the total story.

Same bolts, same loads. The meaning of this is that if you
keep the prop bolts properly preloaded or torqued as it is, then BOTH
the bolts and the prop hub see a much smaller cyclic fatigue load than
if you allow them to become loose, thereby greatly increasing the
cyclic load that they see, and increasing likelyhood of failure.

If the bolts are allowed to become loose, then all of the shear is
carried by the bolts. If they clamp the prop to the hub, then
almost all of the shear is carried by the friction between the
prop and the hub. A tractor will add a small about of tension
to the bolts, since it pulls in that direction. A pusher will
actually reduce the pre-loading in the bolts by a small amount,
but increase the clamping force between the prop and
the hub. IN neither case do I suppose that the friction between
the prop and the hub makes an insignificant contribution to
the integrity of the joint.

Now. please consider two examples, neither of which is a good way to
make something, but which do allow us to isolate the phenomenum
to pure superpositon of stress.

Lets assume nice thick stiff bars fastened like the sheet metal strips
together as in the first example. But in this case the bars are so
thick and strong that it is the bolt that will fail. Again, we apply
a cyclic
load to the bars, alternately pulling on them and relaxing.

In the first case, the bolt is slipped into the hole and not
tightened.
As the bolt is not tightened, all of the cyclic stress in the bolt
is transverse shear.

In the second case, we coat the underside of the bolt head with
a lubricant and turn the nut up against the head pre-loading the bolt
in pure tension with no material at all clamped in between the nut
and the head. Now we enlarge the hole in the bars so that the
nut nd head will fit inside and align them so that the nut bears
on one bar and the head on the other. ALL of the shear is being
carried by the pre-loaded shank of the bolt.

If a cyclic force is applied to those bars, which bolt fails
first?

As for S-n curves, there are more than one type. The one
relating to what I am talking about are the ones that show S vs N for
different stress ratios. The stress ratio is the fraction equivalent
of the maximum to minimum load. For example, something that is loaded
in tension to 25000 psi, followed by being loaded in compression to
25000 psi back and forth, will have a ratio of -1.0 ( +25000 tension/
-25000 compression). Something loaded to 25000 psi tension that is
reduced to 10000 psi tension and back and forth will have a stress
ratio of .4 (10000 tension/ 25000 tension). The S-n curves show that
the amount of cyclic load that structure loaded with a ratio of -1
will fail far sooner than one with a ratio of .4, even though the
maximum stress level is the same. You can look in Mil-Hnbk-5 or
elsewhere for S-n curves to verify that.

The peak-to-peak stress difference in the first case, (ratio -1)
is 5,000psi, for the second case (.4) it is 1500 psi. So it is
no surprise that the first case fails earlier!

Now suppose two cases in which the magnitudes of
the stress cycles are equal:

In the first case the bolt is pre-loaded to 2500 psi then
subjected to an alternating load of an additional +/- 1500 psi,
(e.g. from 4000 to 1000 both in tension) while a second,
otherwise identical but not prestressed bolt is cycled
from 1500 psi in tension to 1500 psi in compression.
Both bolts see the same peak-to-peak stress difference.
The ration in the first case (preloaded bolt) is 4, in the
second case it is -1. Which bolt fails first?

The best book to explain all this is "Mechanical Engineering
Design" by Joseph Edward Shigley, Professor at the University of
Michigan, chapter 8, "Design of Screws, Fasteners, and Connections".
It is THE most widely used text on the subject in the top engineering
schools of the country, and has been for many years.


Barring misprints I am confident that any engineering test used
in US engineering schools will correctly address the subject.

--

FF