Creating the correct profile for a wingtip


#1

I am at a rather serious loss as to how to create a profile for a wingtip. The customer wants the leading edge to slowly transition into a different curve as it sweeps around from the leading to the trailing edge. The final curve is essentially simply 2X by -1Y. If you look at the drawing what I say will make more sense.

I am unable to figure out how to create the surface so that it does not upset the proper profile of the wing. The is a point in the sweep where the wing profile starts to transition away from the proper profile and No matter what I try Zebra strip analysis always shows my profile to be wonky.

There must be a way to work out where the curve lies around the top surface of the wing where the new surface attaches to. The new surface would have to meet in the exact plane as the old surface.

Anyway, my head hurts :slight_smile:Glider rev3.3dm (733.6 KB)


#2

This is how I have come to to tips of rudders and daggerboard on my boat designs. I did this during half-time of the Ala-Tenn game so is pretty rough but you get the idea of splitting your foil at the quad of the leading edge and then cutting back a bit to Sweep2 Curves to get a top and bottom tip half. Use tangency and rebuild to do these and extra projected lines as you have started to do instead of ‘slashes’. I’ll check in with you after the game and you are right, this stuff can make your head hurt the first time or two. BTW, your foil profiles are pretty complex and inconsistent. Look for the plug-in/script here from Mitch or Pascal that will let you import real NACA foils into Rhino. Much cleaner and to a well known standard.

Glider revC3D.3dm (900.1 KB)


#3

That is not a NACA foil. I converted the foil using foil2dxf and then brought the dxf into Rhino. I know that the foil looks odd for various reasons but it is not important for this exercise.

What do you mean when you speak of “slashes” vs “extra projected lines”

I see what you did to create the profile curves. Does it mean that that one should create “best as possible” curves to make the initial surface and then get Rhino to correct the final surface using tangency ?

Just as an aside: I see that you took a nice clean screenshot of the drawing, how did you do that? My screen captures are always very res.


#4

I figured out what you did and applied it to my drawing. It looks better, what do you think.

There still seems to be a slight wonkyness in the middle of the sweepGlider rev3.3dm (644.4 KB)
but I don’t know how to fix it.


#5

Have you tried using the airfoil as the section and the leading and trailing edges (meeting at the sharp corner at the tip TE) as the rails in a 2-rail sweep?


#6

Looks like you have the hang of it. What foil section is that then?


#7

Sweep 2rail makes it look pretty. Don’t know if it will fly but it looks nice. lol

Glider rev3-2railsweep.3dm (1.5 MB)


#8

The airfoil is called JHSYM-10. I have to CNC a layup and cut the masters to create a mold to then make 1.5m carbon carbon fibre models. My customer is a frigging genius, a champion radio control pilot and a model builder of note.

I must still do the rest of the model but it is all fairly straight forward.


#9

FreekeeDeekee - Thanks but the wing profile is critical to the performance of the aircraft. My customer is not so keen on “pretty”


#10

I have no idea why your wing needs 162 cv’s. There is a ridiculous number of stacked knots that can be easily deleted without altering the shape. Your curves have 60 points which again is excessive. The lighter the surface the easier it is to get a good surface match.


#12

I take it that I should reduce the number of nodes in the initial curve and then only do the extrusion.What is the best way to get the lowest number of points and still keep the correct shape?


(David Cockey) #13

If the original imported curve was a degree 3 curve then try FitCrv or RebuildCrvNonUniform and experiment with the tolerance setting for both. Set the maximum number of points for RebuildCrvNonUniform to a large value such as 201. RebuildCrvNonUniform will use the minimum number of points needed to stay within the tolerance.

If the original imported curve was a polyline with straight sections then use FitCrv which will fit a curve to the polyline vertices.


(David Cockey) #14

Use the curvature comb to check smoothness of the rebuilt curve, and CrvDeviation to check the difference between the original and rebuilt curves.


#15

I’m with Strato about the excessive number of control points. Where was that foil section derived and in what format did you import it? Is it one of Doc James H’s foils?

When I look at the foil curve at the root I only see 103 curve points, not so bad. This is intended to be a symmetrical foil, yes? I'm going to play around with you root foil and try some of David's suggestions and see if I can clean it up. Your foil actually looks pretty good with the CurvevatureGraph turned on.

I’m just using ViewCapture to CLipboard.

BTW,

Here is a sailplane design of mine that I did using Rhino and T-Splines several years ago.


#16

You do beautiful work. I am wrestling my way through it slowly.

Here is the foil that the client sent me. It is .DATJHSYM-10.txt (2.1 KB)


#17

Thanks, I’m interested in seeing that foil in the DAT format and using the import parsing script I mentioned earlier. I will try and look up a like to it for you as well.


#18

Minor variations in airfoil profile aren’t as important as many people think. What is important is curvature discontinuities, both chord wise and span wise.

A useful technique for checking whether 2-rail sweep modifies your airfoil significantly is to cut sections at a few places along the span, scale the resulting section to the chord of the master airfoil and do the CurveDeviation as davidcockey suggested. I think you’ll be pleasantly surprised at how close they are. For the part of the wing near the tip (the portion you originally tried to develop separately) it’s much more important to get a nice continuous-curvature blend than to strictly adhere to the original airfoil.

@jodyc111: If the airfoil you show the curvature graph on is the original .DAT file data, try scaling the curvature graph in high so you can see the hairs on the main parts of the airfoil. I think you’ll be surprised at how wavy the graph is on a surface that should be smooth. I’d try it but I’m not on my Rhino computer at the minute.


#19

Here is a new wing swept from the foil root you see with the orange curve scale cranked way up as AIW suggests. I haven’t imported the actual DAT file yet, this is the curve in the original file that is used to create the new foil and tip.

The cyan extrusion is what I used to create the half tips for tangency.


#20

Thanks for all your help. The most important thing that I learned from this is to ensure that each and every curve is correct to start with. Trying to create matched surfaces with bad starting curves is a waste of time.

Those DAT files are really horrible. The trick now is to clean them up before extruding them.


#21

Good that you realise the most important thing in CAD modeling, GIGO. Garbage In, Garbage Out! Not to say I am passing judgement on the DAT file. The curve that you had at the base of the original foil was pretty good, the one that I was highlighting and using curve display on. Did you generate that based on the DAT?

Here is the name of the excellent script that I use and can be found here somewhere, Pascal or Helvetesaur I think.

ImportAirfoilDataV1.py

I’ll try it on the JH dat file and we can see.