(click most images to enlarge)
A major problem with deturbulators to date has been the ventilation system. With changing altitudes, comes changing static pressures that make the membrane either balloon up or press down onto the substrate. Both actions diminish or eliminate beneficial flow-surface interaction modes.
The solution has been to ventilate the panels with a small hole at each end. (see picture on right) However, this fails due to long time lags in equalizing static pressures above and below the membrane. For example, it is obvious that with decreasing altitude and increasing pressure, the membrane will be pressed down onto the substrate with no appreciable time delay. This condition is relieved by air entering the vent holes and working its way down the substrate channels to the center of the panel. Of course, with the membrane pressed into the channels, the open cross sectional area of the channels is reduced and the correction rate decreases.
Ventilation lag was thought to explain why a time-moving average of Johnson's first measurement of extreme performance (blue curves below) and my repeat measurement a year later (red curves below) demonstrated the same characteristic behavior while holding 50 KIAS. I thought that the faster rate at which my performance ran through the pattern (red curve) was due to the deeper channels in the substrates used for my deturbulators compared to those tested by Johnson. I was thinking that the deeper channels allowed air entering the vent holes to run to the center of the deturbulator panels faster. This raised the question whether a better ventilation arrangement might sustain optimal deturbulator performance beyond 1.2 minutes, perhaps even indefinitely?
Additional problems with the vent hole method are (1) that it traps humidity below the membrane and (2) that flow dynamics around the vent holes may have detrimental effects. The moisture issue explains why subsequent flights each day in the Johnson tests yielded better performance. Each tow to 13,000 feet, outgassed the some of the moisture under the membrane.
For a few years, it has been clear to me that these problems might be eliminated entirely by using a porous membrane. The membrane itself would become a surface ventilation system. This would eliminate ventilation lag. It would allow quick equalization of moisture levels above and below the membrane. And, it would eliminate potentially detrimental flow dynamics around the old vent holes. On April, 25 2009, I ran my first test of this idea.
For that test, I chose a readily available membrane material, parachute Nylon. The porosity level is not critical, as the surface only needs to be leaky for pressure disturbances below say 1 Hz and essentially opaque to disturbances above a few kHz. So, parachute fabric seemed like a reasonable material for a first try. However, the weave pattern presents problems. First, although the material does not stretch when pulled along the line of the threads, it readily stretches when pulled along a bias direction. This raises the question of how to orient the material. Should the threads be aligned in the flow/span directions or should they be aligned a say 45 degrees to these directions? Answering this question was the purpose of this first test.
Some insight into this issue may be gleaned from competer visualizations of possible flow-surface interaction modes. Following are AVI files for two initial stabs at the problem of modeling deturbulator flow-surface interaction modes. They were developed by Jari Hyvärinen of ANKER–ZEMER Engineering AB in Karlskoga, Sweden.
I decided to deturbulate the span between the airbrakes and the ailerons, with the fabric oriented flow-wise on the left wing and biased on the right wing. Then I would see which way the glider yaws when holding the magic airspeeds from past tests. That was the purpose of this first test.
Earlier, after determining that last year's fiber reinforced Mylar membranes were no longer working, I removed them but left the channelled substrates in place. It only remained to attach the parachute fabric over the old substrates. By maintaining the previous installation geometry, I could expect the old performance airspeeds to repeat. The fabric was attached to leading and trailing edges of the substrate by means of thin Scotch transfer tape. Glossy Scotch tape was then applied over the fabric edges to secure the installation. This is very critical on the leading edge of the deturbulator, as even a little squareness at the leading edge will trip the flow, turning the deturbulator into a turbulator.
For this test flight, I only needed reasonably calm air. Performance testing conditions were not necessary. Conditions were locally overcast with a mild wind. I towed to 2000 feet agl, released and used 1,000 feet to see what happened at 50 and 70 KIAS, and then at near-stall speeds.
The route pictured on the right shows the yawing of the glider while holding 50 and 70 KIAS and while testing straight ahead stalls (35-38 KIAS). It can be seen that the glider consisently yaws to the right, away from the wing with the deturbulator panels oriented flow- and span-wise. Perhaps this means that diagonal traveling surface waves, which would be encouraged by that fabric orientation, were working better than whatever was happening on the right wing.
The surprising thing was the high yaw rate near stall, even though the wings were level and the rudder was straight. In addition, the stall speed was reduced by a full knot and the normal agressive dropping of the left wing was completely gone. No rudder was needed on the edge of the stall. The altitude vs. time profile during this time (left) shows surprising performance for speeds just above a stall. I do not read much into that, as there is not information indicating that convection was not present. However, perhaps we can conclude that performance was not suffering greatly and so the yawing was not due to high drag on the right wing, but reduced drag on the left wing. Also, the absence of the usual left wing drop may indicate improved lift.
At first I was encouraged by the near stall performance, but upon further testing with tightened fabric membranes (5/22/09) the stall performance returned to normal and other traditional performance speeds worked well. That, and the fact that the membranes were very loose after this flight plus evidence from subsequent oil flow visualizations with taut and loose fabric membranes, have led to the conclusion that in this flight the deturbulators were in fact making additional turbulence which had the effect of reducing top-surface trailing edge separation at the high angles of attack just before a stall and that is the thing that changed the stall behavior.
A post-flight look at the deturbulator membranes (below) was revealing. On the left wing, where the threads were aligned with the flow, stretched areas of the fabric often fell into cellular patterns, three deep in the flow direction and a little over an inch span-wise. This may indicate something about the type of wave that predominated when the deturbulators were working. On the right wing, where the fabric was stretchy in the flow and span directions and deturbulator action was weak or nonexistent, structured stretch patterns are not visible.
|© Copyright 2003-2012 Jim Hendrix||Disclaimer|