http://ozreport.com/forum/viewtopic.php?t=34442

Esoterica-- self-sustaining yaw-roll oscillations, keel-fixed
versus keel free:

Spectrum 144

Normal glider configuration:

http://vimeo.com/77807574

Slo-motion version of above:

http://vimeo.com/77870634

Keel temporarily fixed tight to crossbar:

http://vimeo.com/77804817

Slo-motion version of above:

http://vimeo.com/77873468

There is much more trailing-edge movement in keel-free case-- naturally.
Allowing greater "shedding" of the loads imposed by aerodynamic damping in roll,
and much greater roll rates.

(link to thread on fixing keel to crossbar:
http://ozreport.com/forum/viewtopic.php?t=34422 )

Once I got the oscillation going, I did my best to hold myself completely fixed
on the bar. I don't believe that the slight movement of my body that you see
especially in the keel-free case, where the oscillation was so large and
involved lots of slip and sideforce, was significantly driving the oscillation.
Occasionally you can see me give a significant input in which case the
oscillation has damped out for some reason and I am getting it restarted.

Commentary:

Oscillations involving roll, yaw, and pitch. After some control inputs to
initiate the oscillation, pilot is holding body fixed in center of base bar as
much as possible, with bar well pulled-in. (Arms fully extended, but pilot is
not tucked head-down or knees-over bar into extreme dive, which actually would
yield a steep dive without these oscillations.)

The pilot's body does move a bit, especially in the test with the keel free to
move, where the roll was so dramatic and the resulting adverse yaw and
aerodynamic sideforce on the wing was significant. Even as the pilot's hands
grip the base bar tightly, his CG is slightly aft of his hands and tends to
swing in the same direction that a slip-skid ball would deflect-- in the
"upwind" direction, as signalled by the yaw strings. The pilot's arms and hands
are not in a good position to generate a great deal of yaw torque and completely
prevent his body from twisting. This creates a slight movement of the hang
strap. On the whole I believe the effect of this is negligible: a similar
maneuver where the pilot pulled in the bar but tried to let himself move freely
from side to side on the bar, yielded essentially the very same results. The
glider's roll dynamics are dominated by the wing's powerful anhedral response to
sideslip, not by very small movements of the pilot's body. To the extent that
the pilot fixes himself in place on the base bar, he lowers the CG of the whole
system and this shifts the glider's "effective dihedral" to be slightly less
negative. To the extent that the pilot swings freely, he pulls the keel to the
side during sideslip, which has the opposite effect. In either case, the
glider's "effective dihedral" is still strongly negative at this low
angle-of-attack that results from the bar being well pulled-in.

Even in the test where the keel is fixed relative to the crossbar, note that the
wing does have some flexibility to distort under load and shed some of the loads
imposed by aerodynamic damping in roll. Watch the trailing edges carefully. This
flexibility helps to increase the roll rate. The keel pocket on the Spectrum is
a very tight sleeve, and the changes in trailing edge shape are not the result
of the keel pocket shifting from side to side. A glider with a looser keel
pocket would should much more flexibility in the trailing edge, even if the keel
were not free to move relative to the crossbar.

In the test where the keel is free to move, note how much the trailing edges
move, as the wing "unloads" some of the aerodynamic damping force created by the
rolling motion. This allows for a much greater roll rate, which in turns
increases the bank angle, slip angle, and pitch excursions that we see in the
oscillation.

Note the yaw string deflections. Their deflections are almost synchronized to
the roll rate, deflecting toward the rising wing, but with a slight lag--
lagging by 1/8 or less of the complete oscillation cycle.

As the roll rate slows, the yaw strings are still strongly deflected. When the
direction of roll reverses, the yaw strings are still slightly deflected from
the previous rolling motion. At the instant the direction of roll reverses, the
yaw strings are still deflected toward the wing that was formerly rising, which
is still the high wing.

(This is much more visible when the video is watched at 1/10 speed or slower!)

With the bar pulled in, the glider's "effective dihedral" is strongly negative--
anhedral dominates over sweep. When there is a sideways component in the
airflow, anhedral creates an "upwind" roll torque. At the moment that the
direction of roll reverses, with the yaw strings still deflected toward the high
wing, this "upwind" roll torque will tend to keep increasing the bank angle.
Therefore, the aircraft's yaw/ slip dynamics cannot be responsible for the
reversal in roll direction. '

This is very different from a classic "Dutch roll" oscillation, in an aircraft
with enough sweep or dihedral to have significant positive "effective dihedral"
at the angle-of-attack of interest.

What is responsible for the reversal in roll direction? I believe a strong
rolling-out torque is generated by the fact that the wingtips are negatively
lifting, and the outboard or high wingtip is moving at a higher airspeed than
the inboard or low wingtip.

A related observation-- when we fly in a constant-banked turn and then pull in
the bar while exerting no roll torque on the bar, the glider will roll into a
turn in the opposite direction. Again, when we pull in the bar, I believe a
strong rolling-out torque is generated by the fact that the wingtips are
negatively lifting, and the outboard or high wingtip is moving at a higher
airspeed than the inboard or low wingtip.

It's a dihedral-like effect, tending to roll the glider toward wings-level, but
it responds to yaw rate, not sideslip angle.

A wing with less washout could not show these dynamics.

On the other hand, once the direction of roll has reversed, the glider's yaw/
slip dynamics are surely helping the glider reach and sustain a high roll rate.
The interaction between aerodynamic adverse yaw and yaw rotational inertia
probably causes the nose to swing "too far" toward the rising wing compared to
what we would see if the aircraft had less yaw inertia, and then stay strongly
deflected toward the rising wing for some seconds after the maximum roll rate
has passed. The resulting "upwind" roll torque created by anhedral, acting to
roll the glider toward the descending wing, helps sustain a high roll rate
longer into the oscillation than would otherwise be the case. Decreasing the
aerodynamic adverse yaw torque associated with rolling would certainly decrease
the peak roll rate that we see in this oscillation. Decreasing the aircraft's
yaw rotational inertia might also decrease the peak roll rate that we see in
this oscillation.

A related observation: adding a vertical fin to the keel greatly diminishes the
oscillation that we are seeing in this video.