Wind tunnel flow field visualization confirmed the horizontal tail was immersed in the wake of the stalled wing over a range of high angles of attack. Local flow speeds near the aft fuselage and empennage were a modest fraction of flight speed at these conditions, greatly reducing pitch control effectiveness.
Many configuration changes were developed and tested (possibly in the GALCIT Wind Tunnel at Cal Tech or at NASA-Ames). No single change was completely effective.
A larger horizontal tail (focused on increased span) provided significant, but inadequate, improvement.
A vortex-generating pylon on the undersurface of the wing and just aft of the leading edge also provided significant, but inadequate, improvement. At cruise, the pylon was attached to the lower wing just aft of the local wing leading edge stagnation line, where local flow speed was low, so that its cruise drag was less than nominal. At high angles of attack, the stagnation streamline moved well aft of the pylon leading edge, with local flow moving forward past the pylon. In this locally-reversed flow field, the pylon produced a vortex that swept forward along the lower wing surface, up past the wing leading edge, aft over the wing, and into the horizontal tail flow field. The vortex was thought to improve both the separation pattern on the wing and downwash at the tail, and possibly provide an increase in local flow speeds near the tail. This pylon extended aft for more than half of the local wing chord.
A short-chord fence was also used outboard of the vortex-generating pylon. This fence was conventional, as it extended around the wing leading edge on both upper and lower surfaces.
There might have been other fixes but I do not recall them. In the end, the only change that proved adequate was a combination of fixes, including the three described above.
A meeting had been called to report on Deep Stall progress with the President of Commercial Airplane Division, Donald W. Douglas, Jr. Costs estimates had been prepared for each of the individual changes.
There was a lot of nervousness about "Junior's" reaction to spending as much as $10M on a combination of fixes, a large amount of money in the 1960s. He had not been President very long, and was about to make a huge decision for the future of the Company. Some anticipated a decision to delay, with direction to find a lower-cost fix, resulting in probable overall program delay, and all of the expense that would entail.
The meeting with Mr. Douglas was short. When the options had been presented, he directed immediate implementation of the combined changes. He was quoted as saying, "We have no other choice." A lot of people in Engineering gained a huge respect for Mr. Douglas on that day.
Someone did not like the name, "vortex-generating pylon." It became the "vortilon." I believe this was the origin of both the name and the device, which is used on business jets today.
The vortilon was used on all of the DC-9 models, and the MD-80. I do not know if it was used on either the MD-90 or the MD-95.
The original flight test program had focused on the DC-9-10, which had the shortest fuselage, and hence, the shortest tail arm for effecting pitch stability and control. The DC-9-15 and DC-9-20 also had this same fuselage length. However, these later models did not have the wing fence, according to:
http://www.airlinercafe.com/page.php?id=396. I do not know the reason(s) for eliminating it.
These shortest-fuselage models were all able to demonstrate acceptable stall recovery (certifiable) at aft CG locations with the Deep Stall modifications, but such recovery was less than ideal and could not have been described as robust. I presume "acceptable" meant that stalls could be entered and a pitch attitude higher than that for maximum lift could be obtained, followed by a normal, if slow, stall recovery.
The DC-9-30, -40, and -50 each had increasingly longer fuselages than their predecessors. The MD-80 and -90 family fuselages were also longer than that of the DC-9-10, -15, and -20. All thus had longer tail arms than the initially certified DC-9-10, favorably affecting both pitch stability and control, and very possibly producing slightly weaker downwash at the tail, as the tail was further aft from the wing. Tail entry into a stalled-wing wake would occur at a higher angle of attack than for the shorter fuselage models.
Thus, all models of the DC-9 and MD series were, and are, capable of recovery from a stall, provided the pilots identify the initial stalling condition and respond appropriately.
One interesting aspect of the DC-9 Deep Stall effort was the pitch control system. Douglas Aircraft used trim tabs to drive the control surfaces on all large commercial products through the DC-9. This was investigated as a possible contributor to the DC-9 Deep Stall, but was not a factor.
Douglas had used trim tabs to actuate control surfaces since the DC-3, and had not changed this practice, even with the DC-8 and DC-9 jet transports, probably as a result of "technology inertia." This use was questioned again after DC-9 drag had been determined in flight test.
My Fluid Mechanics supervisor, Ed Rutowski, explained the DC-9 drag issue and asked for suggestions to resolve it. The issue, as Ed put it to me, was that "We lucked out on drag. Our basic drag level is high by about 5% but our compressibility drag at cruise is low by about 5%, so we are OK for cruise drag."
Ed then asked for ideas for the source of the high basic drag level. I mentioned several possibilities. All but one had been considered. The exception was gap drag. I asked about the control surface gaps and how they were sealed. Ed told me they were not sealed, as the control surfaces were driven by trim tabs. The control surface gaps needed to be open for aerodynamic balancing, so that trim tabs could provide moments adequate to drive the primary surfaces. I suggested those gaps were a likely source for at least part of the drag, and mentioned that Art Mooney sealed the control surfaces gaps on his later, highly efficient light planes with strips of fabricate to prevent flow through those gaps and the resulting drag. Others people might also have identified this candidate drag source.
Ultimately, a senior aerodynamicist, Frank Lynch, managed two high Reynolds number wind tunnel tests to measure gap drag. A 6% scale DC-9 horizontal tail was tested in the North American Rockwell Tri-sonic Wind Tunnel.
Frank concluded the DC-9 control surface gap drag was about 5% and control surface float drag was about 1%. For the DC-8, these drags were 4% and 2%, respectively.
As a result of these tests, the DC-10 became the first Douglas commercial transport going back to the DC-3 to use powered controls and in all three axes. The DC-10 gap drag was estimated as "less than ½ %."
I still have a copy of Frank's memo in my files.