Originally Posted by Lazy

Wing draw - it's ONLY example of splitted surface. Is not a my wing...
9 meter span, 18 AR, 40 kg TOW, motoglider - that a my design.

NOW you tell us....

In that case what the other guys mentioned already is what you need.

The control surface endplates are interesting but don’t appear to lend themselves to simple analysis. The purpose of an aileron is to produce a rolling moment by changing the loading along the span. Any change in the span loading has to be accompanied by a change in the vorticity / circulation carried along the span. The presence of an end plate can’t change the fact that a change in the span loading must result in vorticity /circulation being shed into the wake (if the endplate were to completely “block” vortex shedding at the aileron tip, it would also make the aileron completely ineffective).

Given that the end plates can’t block the vorticity shedding, all they can do is redistribute it. If the redistribution they provide makes the resulting span loading closer to elliptic, they should reduce the induced drag due to aileron deflection. The question then becomes: does this benefit outweigh the profile drag cost? Given that the profile drag of an end plate is always present, and its potential induced drag savings would only come into play during aileron (or partial-span flap) deflection, their prospect for providing overall performance benefit is dubious.

My limited experience with vortex lattice codes suggests that they are poorly equipped to explore this configuration. I have had very little success trying to directly model the circulation in the vicinity of a flap or aileron break with a vortex lattice code. If I understand AVL’s implementation, control deflection is approximated by “normal vector tilting”. I have found this approach to more accuratley predict the span loading than dircet analysis of the actual control surface geometry. Unfortunatley, this approach prevents comparing the endplate case to the no-endplate case. Would useful comparison really require a full-blown experimental or Navier-Stokes analysis?

I will try to simulate problem in XFLR...

XFLR5 is not remotely capable of resolving details at this level. It does not even give good results for different planform shapes, and routinely gives span efficiencies of over 1 which is incorrect. You would need a very high order CFD program like they use for F1 cars to even have a hope.

The drag of the small opening in the trailing edge of the control surfaces with different deflections will be very small. They do not use end plates or anything else on $200k full size sailplanes, where they design to pretty small detail levels. I suspect it can be safely ignored.

Kevin

I suspect part of the reason why plates of that sort are not used is that typically the surfaces spend the big majority of their time either at neutral or deflected so little that there is no significant gap to seal.

If the plates were there for those slight times where large amounts of deflection are used then they would be producing drag of their own that would blow away any advantage they might have for some short rare times that they do help.

Originally Posted by kcaldwel

XFLR5 is not remotely capable of resolving details at this level. It does not even give good results for different planform shapes, and routinely gives span efficiencies of over 1 which is incorrect. You would need a very high order CFD program like they use for F1 cars to even have a hope.

The drag of the small opening in the trailing edge of the control surfaces with different deflections will be very small. They do not use end plates or anything else on $200k full size sailplanes, where they design to pretty small detail levels. I suspect it can be safely ignored.

Kevin

I agree the idea of control surface endplates doesn't make much sense.

Left to its own devices, a vortex lattice code (and I suspect many more "robust" panel methods) will predict wildly inaccurate span loadings in the vicinity of a control surface "break". The loading in that region appears to be sensitive to small changes in the vorticity distribution. I suspect the only way to get enough fidelity in the vorticity distribution is to solve the full viscous flow problem. I hope my intuition is wrong, because it would be convenient to be able to explore these kinds of geometry differences with a simple VL code.

I think you can have span efficiencies above 1.0 under the right conditions. The 1.0 limit only applies to cases where the wake is planar. A wing with winglets can have lower drag than a wing of the same span without winglets (and thus a span efficiency above 1.0). If you consider winglets cheating, a planar wing that sheds a non-planar wake (i.e. one with a trailing edge the is distinctly not straight) can also have a span efficiency above 1.0. Some of the problems with this kind of configuration are:
1. getting the lift distribution right.
2. not having the induced drag benefit offset by the profile drag penalty.
I wouldn't automatically dismiss a span efficiency factor slightly above 1.0.

Yes, I should have stated a planar wing. XFLR5 routinely indicates span efficiencies considerably over 1 for a planar wing. It cannot properly even sort the relative efficiency of a rectangular wing with square tips, an elliptical wing with a straight 1/4c, and a crescent shaped wing, compared to good wind tunnel testing.

Yes, non-planar wakes are interesting, but VL or LL code cannot not model that. You need pretty radical geometries, like the split tip wing proposed by Smith to perhaps have span inefficiencies over 1 for a planar wing. XFLR5 shows span efficiencies over 1 for a rectangular wing, which is not correct.

http://ntrs.nasa.gov/archive/nasa/ca...1996034550.pdf

Kevin

Originally Posted by kcaldwel

Yes, non-planar wakes are interesting, but VL or LL code cannot not model that. You need pretty radical geometries, like the split tip wing proposed by Smith to perhaps have span inefficiencies over 1 for a planar wing. XFLR5 shows span efficiencies over 1 for a rectangular wing, which is not correct.

http://ntrs.nasa.gov/archive/nasa/ca...1996034550.pdf

Kevin

Funny we were both thinking of a split tip wing. I did my thesis work side-by-side with Steve in the Ames 7x10 using the same wings. I measured the induced and profile drag of the split tip wing from wake surveys (my work is reference 24). The induced drag looked pretty good, but the profile drag at the tip junction was atrocious.

Cool! Referenced work, and you got to rub elbows with Ilan Kroo?! Do you know Steve Morris, or was he gone by then?

Is your paper available on line? I'd love to see more detail on the split tip.

The split tip is an intriguing idea, but the junction drag - and the low Re of the two tips at model sizes - look like real challenges. I keep toying with trying it out on a model.

Kevin

I know Steve Morris well, but unfortunately haven't been able to spend nearly enough time around him in the last 20 years. He help me set up my first RC plane (a Sagitta 600) a "few" years ago.

Unfortunately, the internet was quite young in 1994. The only electronic copy I have is on an ancient "SyQuest" drive that I can no longer read.