How to fillet a keel or wing root

[mcg]

Here is a common problem in drawing airplanes: creating a smooth fillet at the wing root. It is similar to filleting a keel fin where it joins a yacht hull, but keel fins are symmetrical. Wings are of course airfoil shaped.

On the rc discussion groups the Rhino enthusiasts have a lively inquiry going on how to fillet a wing root.

http://www.rcgroups.com/forums/showthre ... st13631525

There are number of solutions under discussion, and workarounds, but it seems to me tsplines is the best place to start.

Is there a step by step procedure for creating a wing fillet?

Thank you for your insights.

MIchael

[mcg]

Just to illustrate this type of problem, here is NACA 2804 wing section and an elliptical section fuselage. The idea is to create a smooth fillet where the two structures are conjoined. I can send a 3dm if it would be helpful.

Michael

[JSantocono]

Hi Michael,
If you are looking for a radios fillet you wont be able to do it easilly in T-Splines........but you can get a "blended" type of fillet. I can show you if you send me the surfaces on a file. I want the surfaces because each case is different and doing it with the actual surfaces is going to be better.
Sorry for the delay response (I didnt see your first post).

Thanks,
JuanS

[mcg]

Hello Juan,

Thank you for your help.

I have uploaded a .3dm to serve as an example. The two objects are a wing and a fuselage section. The fuselage is just an extruded ellipse, so it could be made bigger or smaller as you wish. The "wing" is actually a horizontal stabilizer designed for a model airplane, but the filleting problem is essentially the same as that for a wing.

These are NURBS surfaces, I have not rebuilt nor converted them to tsplines.

Best, MIchael

[JSantocono]

HI Michael,
Probably in this case is better to do it in plain NURBS.....,but here is a File with very simplified steps to make it in T-Splines.
Check it out.

Feel free to ask. The is no explanations, just the steps.

JuanS

[mcg]

Thank you for your help.

Not quite clear how to do step 1. The 3dm file says to use tsInsertEdge and then move the wing toward the fuselage.

The insert edge command requests that I "select the edge loop to be duplicated or edge ring to be split. If I pick an edge -- any edge -- the command then requests that I identify a point on it, or enter a number between 0 and 1.

Precisely which edge do we want to select, duplicate or split? Where should the point go?

Thank you for your insights.

Michael

[JSantocono]

Hi Michael,
The tsInsertEdge command needs edges or faces selection to place the new edges. Take a look at the T-Splines manual Pag 75. It will help you to understand how to insert edges.

JuanS

[SDS]

T-Splines will make some nice aircraft shapes. For me it's easier to create the entire airplane as a T-Spline surface instead of making smaller parts such as the wing and stabilizer fairings.
Attached is a T-Spline experiment; not meant to be anything, just a generic airplane shape. It's kind of sloppy, but should give you some ideas.

[JSantocono]

Hi SDS,
Nice model. Nice topology, pretty simple and describes very good the main shape. a Spitfire?
Thanks for sharing!
JuanS

[mcg]

Beautiful work. Thank you for posting this file. There are 5 fillets here, and they are very nicely wrought. Filleting seems to me a significant, stubborn problem in both aircraft and hull design in Rhino.

I am a newcomer to tsplines and I will keep studying the program. The end results are, as ever, very impressive. The steps along the way remain kind of mysterious, but sooner or later something will click.

Was this aircraft a product of "box modelling"? Did you start from a 3-view or sketch one?

Regards, Michael

[SDS]

For this exercise I started with a Clark Y airfoil and a typical fuselage cross section, both reduced to 8-10 control points. The round radial engine type nose was used because it's easier to draw. I used mainly TSLines after linking the various sections with lines. I'm learning some of the box modeling techniques.
The goal for this drawing was just to create a TSpline airplane; here's a sample of the current TSpline project-

[mcg]

SDS, Thank you for posting this early Mustang. I especially like the wingtips and the underslung scoop. The wingtips in particular are often tricky to create.

Still not quite following the procedure for putting the fuse together with the wings and the horizontal stabilizer.

In Rhino, we can put surfaces together with a Boolean add, a Match & Merge Surface procedure, or do a Join.

To assemble two separated surfaces in Tsplines, I think there is a Merge command. I have also seen in tutorials surfaces being stitched together, using Weld Points. And there is also the new Bridge command, which might or might not apply.

So there is a range of different methods to stick surfaces together, either in Rhino or in Tsplines.

It appears from your post that you are proceeding kind of as follows: I put question marks where I am not quite following what to do.....It looks like you are putting the pieces together in Rhino before converting the whole unit into Tsplines?

1) draw the fuse in Rhino
2) draw the wings and stab in Rhino
3) rebuild each component (still in rhino) to radically reduce the number of control points
4) Link the various sections with lines (in Rhino?) Or does this mean link the formers, as though with stringers?
5) Maybe make surfaces from these linking lines, as for example by Rhino's network of curves?
5) Convert the already joined surfaces into Tsplines?
6) refine the watertight model in Tsplines.

rc Airplanes are my hobby but I think this particular problem of joining and filleting has wide application to a number different design fields. Many thanks for your help on this. Congratulations again on the great Mustang prototype.

Best, Michael

[SDS]

The "Generic Airplane" is not a composite of Nurbs and TSplines, it is a single TSpline surface, drawn using "TS From Lines" after first drawing separate frameworks of lines for the wing, fuselage, and empennage, then linking these together with lines. Use TSpline's "Extract Control Polygon" to examine the line frame, then use "TS From Lines" on the entire framework to recreate the surface.
I was amazed that these fairings was generated by just linking the wing and fuselage with straight lines. If you were to Weld the same vertices after first skinning each component individually, it would accomplish the same thing; I was more comfortable working with the lines. (I see now that this airfoil is obviously not a Clark Y: I must have adjusted the lower edge points to add curvature to the bottom of the airfoil.)

[Matt Sederberg]

The "Generic Airplane" is not a composite of Nurbs and TSplines, it is a single TSpline surface, drawn using "TS From Lines" after first drawing separate frameworks of lines for the wing, fuselage, and empennage, then linking these together with lines. Use TSpline's "Extract Control Polygon" to examine the line frame, then use "TS From Lines" on the entire framework to recreate the surface.
I was amazed that these fairings was generated by just linking the wing and fuselage with straight lines. If you were to Weld the same vertices after first skinning each component individually, it would accomplish the same thing; I was more comfortable working with the lines. (I see now that this airfoil is obviously not a Clark Y: I must have adjusted the lower edge points to add curvature to the bottom of the airfoil.)

SDS,

Nice work! One possible suggestion: if you use the new tsBridge command (and you might need to delete some faces) to combine the different parts, you could probably avoid needing to extract the control polygon and using the tsFromLines command.

Best,
Matt

[mcg]

SDS -- many thanks for this clear explanation, and for pioneering this whole problem. I will try it.

There is a russian aviation and modeling enthusiast who has posted a big collection of 3-views. Here is one I might try as an exercise to learn your technique, a cc 1947 Boeing 377.

http://www.airwar.ru/image/idop/aliner/b377/b377-1.gif

What makes this aircraft kind of interesting, from a tsplines point of view, is a midline crease that runs the length of the fuselage. There was a SARO flying boat from that same era that had a similar crease.

Thanks again for this method, SDS, and thanks to Matt for the Bridge concept.

Best, Michael

[SDS]

Thanks Matt, bridge would work great here. I drew that before 2.2.
MCG; that double bubble 377 would be a challenge. I discovered that site before; most of his drawings are great, some are too small and coarse.

[petertha]

SDS -- many thanks for this clear explanation, and for pioneering this whole problem. I will try it.

...my first post!...

MCG, thanks so much for the connecting the rhino filleting Q&A's from RC-groups over here. Im a newbie T-splines (for Rhino) user, just purchased recently, slowly getting up to speed & crossing my fingers it would eventually assist with these specific applications in mind.

SDS, I'm completely blown away by the airplane renderings & TS techniques. If you ever do a step-by-step tutorial or video or white paper or whatever.. count me IN! Meanwhile, Im going to spend a lot more time following this thread & forum, its inspired me to learn these techniques with renewed enthusiasm.

Peter

[JSantocono]

Hi Peter,
Welcome to the forum.
Feel free to ask here.

JuanS

[SDS]

MCG & Peter: I added a post in the rcgroup Rhino thread, describing a technique I have used to draw wing fairings with Rhinoceros, but now that we have TSplines, fairings will smoother from now on!

[petertha]

For this exercise I started with a Clark Y airfoil and a typical fuselage cross section, both reduced to 8-10 control points.

I read this but wasnt sure how to interpret the implications, so maybe we can both elaborate. In my world, airfoil coordinates & their seemingly minor shape factor subtleties, are really, really important. So Ive always obeyed the section coordinates as a starting point to generating rhino wing surfaces because that shape ultimately dictates L/D polars & performance of the real world model. (I figure many aerodynamic eggheads spent their lives in wind tunnels & writing Fortran simulation code, there must be a reason for 6 significant figures in their x-y coordinates! ) Having said that, many airfoils are very old & they just didnt have the tools we have today. In certain instances rhino curve tools can help tweak & improve unintended deviations of many golden oldie naca coordinates. But if you oversimplify, the cad airfoil starts to deviate from the intended airfoil, and thats not good either. My recipie is higher point density in the first 30% or so nose area & less is perfectlly fine thereafter, so kind of a hybrid.

From your pic, it certainly looks like the 'reduced' control point Clark-Y is near identical to the 'standard' Clark-Y. Just wondering, if your 2 airfoils were overlayed, how 'close' are they? Or am I missing the point & the sections are more intened for localized fuselage blending into t-splines generated fuselage & fillet area etc.

When it comes to the fuselage & related features I would have to say its not nearly as critical as the flying surfaces. The t-splines stuff looks fantastic! I just cant get over how smooth & accurate the surfaces look vs real world airplanes with seeminly simple t-box skeleton. Must... continue... learning

[SDS]

I certainly agree with you concerning the critical nature of airfoils, and the curve I used is an approximation, but a good approximation. I didn't input the coordinates, but matched a curve to a somewhat low resolution jpg I found online. See attached Rhino file. The worst fit is at the trailing edge, and more careful point adjustment would help. However, in this case, we are learning TSplines, and trying to keep the model as simple as possible for that reason. When we become TSpline masters, more points can be added.
This airfoil approximation is fine for a model (which is all I would ever make from this).
For a real application, once the model was "complete", the wing minus fairings and tips (the difficult to draw "organic" parts) could be deleted and replaced with a more accurate airfoil section and surfaces matched.
Good point!

[Adam Helps]

One of the interesting properties of T-Splines, with regard to precision modeling, is that perturbations are limited to a fixed area around each control point. So the above workflow of taking a carefully calculated region and transitioning it into the rest of the shape is definitely possible, and could (at least in theory) be accomplished completely within T-Splines. It is usually possible, for example, to make a certain isocurve along the surface precisely touch a desired template curve; a similar technique could certainly be used to create airfoils that precisely and smoothly pass through specified coordinate sets.

In practice, it requires a variety of tools to actually reach this level of precision. We've been putting a few things in here and there to make certain workflows possible, but T-Splines are so new that we're still not sure what a precision/mechanical modeling environment would look like. The math is largely worked out, we just need to figure out what the user interface ought to be.

[SDS]

Wow, that makes me really look forward to future updates of TSplines!

I like the idea of matching isocurves to a curve template, provided you're working with the surface, not just vertices or edges of that surface.

For now, the eight points of the simplified airfoil are the template, and as long as editing remains beyond two points or edges from the wing section defined by that airfoil, the airfoil section is not affected by editing adjacent areas.

[petertha]

One of the interesting properties of T-Splines, with regard to precision modeling, is that perturbations are limited to a fixed area around each control point...

If you would like to try an example just for kicks, I happen to be familiar with this airfoil & happen to have it handy. It's probably representative of the toughest curve ball one would see, non-symmetrical + camber + tight tolerance required. The author's original x-y data points are imported on one layer, my fitted rhino curve on another. Its been a while, but I recall using Curve Interpolate Points through the x-y's, smoothening the curve a bit with FairCurve, reducing the point count to simplify & some minor diddling around the LE & TE. Airfoil 'shapes' come from a variety of sources, but generally the format is similar to what's presented here. http://www.mh-aerotools.de/airfoils/index.htm

So what would be the TS procedure to rebuild this & how whould you be able to 'see' the resultant TS curve variation vs the so called 'stock' shape? I would think this 'TS meets pre-defined generic part shape' capability would be powerful feature in many applications.

/pt

[sgreenawalt]

SDS -

Wow, this model you posted is truly fun to play around with, and I really like how you created the topology. I was really blown away with how simple the wing root fillet was to accomplish, as it seems you were as well. I've been mostly playing around with box modeling, but I'm going to do a lot more with tsfromlines. This is really wonderful and has given me a lot of ideas!

-Sky

[sgreenawalt]

What with the interest in aircraft on here lately, I decided to begin what I've been meaning to do for my first "real" project with t- splines - a nice model of Mike Arnold's record setting AR-5. If you're not familiar with it, he has a nice website here:

http://www.ar-5.com/

There's a really bad 3-view on his site in one of the articles, but it's enough to get the ball rolling. I started with a cylinder and then mostly point edited from there. No flying surfaces as of yet - just the fuselage. I'll work on the wings and empennage over the break. The part that's giving me the most trouble so far is the front portion of the transition between the canopy and fuselage. Any suggestions would be appreciated!

-Sky

[SDS]

SGreenawalt; Thanks for posting the AR-5 link. The AR-5 is an incredible accomplishment; Mike Arnold is not only a genius, but a practical genius with the tenacity to complete such a complex project. I especially enjoyed the construction photos; I've built quite a few things using the Rutan "sculpt foam and skin with fiberglass" technique, but nothing to compare with the AR-5.
The AR-5 is an ideal TSplines project; nice curves and very flowing shape. My first "real" TSpline airplane was the P-51B I posted, and I learned a lot from that project. I'm looking forward to see your AR-5 model.
Petertha: That is a very strange airfoil- my initial impression is the reverse camber would reduce the downwash thus lift, like flying with both ailerons up. Do you have any wind tunnel data for this shape (L/D, pitching moments)? I would approximate this airfoil like the example I posted earlier but more points would be needed.

[sgreenawalt]

SDS -

The AR-5 is truly a work of art. I was really into sheet metal planes, and then I watched Mike's videos, and ever since then I've been head over heels for composites. I'm slowly designing an airplane of my own, and I see the AR-5 as the logical departure point for what I want to do, so for me making a nice AR-5 model in T-Splines is the first modeling step in realizing my dream. If you ever get to the Hiller Aviation Museum in Redwood City, you can see Mike's AR-5 in person - well at least the bottom of it since it's hanging from the ceiling. Hanging adjacent to it is Rutan's personal Defiant, and the difference in craftsmanship is just staggering. The Defiant looks like the product of a high school composites course next to the AR-5, which looks like....well a masterpiece.

-Sky

[JSantocono]

This plane is beautifull! ........and its perfect for T-Splines.
I would like to give it a try....but, any chance you guys have the 3-views blueprints?
Thanks,
JuanS

[sgreenawalt]

Juan -

The best you're going to get in the way of 3-views is the super low res one on this page:

http://www.ar-5.com/condrag94.html

That's the one I used to get my model started. There's also a bunch of dimensions on that page, that you could use to cross check the 3-view. Mike was concerned with liability, and so he never released any plans or drawings of the plane. If you look around on his site, you'll also see some links and pictures for the AR-6, which is his latest version of that plane. The 6 is a Reno racer, and so it has a horizontally opposed Continental O-200 engine instead of the inline Rotax the 5 had. Check out the cowl on the 6, it's a work of art in itself!

http://ar-5.com/gallery/AR-6

-Sky

[JSantocono]

Thanks Sky for the info........
hmmm......I think the R6 is the one to model!
I wish I could get those R6 blueprints.
JuanS

[sgreenawalt]

Juan -

It's worth emailing him and asking him if he would be willing to provide any documentation. I know that a radio control model was built by someone (there's some pics of it on his site, and it's very accurate) and so I imagine he might have helped them with some documentation. It's also possible that model was made from direct measurements of the actual plane by the modeler. Anyway, I've emailed with him a few times about his videos, and he's a reasonable fellow so it's worth a shot. If that doesn't work, keep in mind that he started with the AR-5 dimensions to make the AR-6. He reduced the wetted area of the rear fuselage, changed the fin/keel arrangement of the tail, and modified the cowl for an O-200. So, it's not all that far off from the AR-5.

-Sky

[JSantocono]

Hey Sky, I already did it
I contact him and ask for some 3 views blueprints and he give me this great "hand writing" blueprints. (Thanks Mike Arnold for sharing!)
Here they are........its good information to model it........Someone is going to try? I will probably give a try.
Have fun.

JuanS

[sgreenawalt]

Oh that's a fantastic resource! Kudos to Mike for sharing!

-Sky

[JSantocono]

Hi all,
here is my first attempt to model the AR6 in a single T-Splines surface.
JuanS

[sgreenawalt]

Juan - I think you nailed it! -Sky

[petertha]

That is awesome Juan! If you ever create a tutorial on how you did that, I'd love to see it!

[aeronautica]

Juan,

Yes, a tutorial like the one you did of the couch would answer a million questions.
You have done in less than a day what I have been trying to achieve for weeks.

This looks great.

Larry

[Adam Helps]

That looks great! I'd love to see this one speed-modeled. I think I watched the couch speed modeling five times .

How exact is the airfoil shape on the wings here? It seems like that would be the hard part to nail.

[SDS]

Nice job, Juan! We're waiting for the tutorial!
There 's better (or maybe just bigger) three view of the AR-5 here; http://ar-5.com/gallery/AR-5constructio ... iew?full=1 It was on page three of the construction photos.

[JSantocono]

Hi all,
Thanks!
Here is the file to play with. (last model of 2009
I mainly use polygonal modeling. Started from the tip with a circle and extrude each section from there to get the fusselage. A tutorial to make this would be kind of hard because its mostly control points movement to get the correct shape..........but, lets see if I can make some kind of more detailed explanation of this model next week. For now, check the model, maybe can answer some questions.
Hope it helps to bring some ligth!

Happy new year for all! And thanks all of you for making this forum a good place to learn and share.

JuanS

[sgreenawalt]

SDS - Man how did I miss that 3 view?!?!?! That one is WAYYYY better than the one I was using. It's not like I haven't drooled over Mike's site a million times or been made fun of by my wife for watching his DVD's over and over and over.....

Juan - Thanks so much for posting your model for all to see. Actually, the fact that you're doing lots of point editing would be really interesting to me - it's something that I find takes a lot of practice to make it look just right, and it's helpful to see how others approach point manipulation. Happy New Year to all, I made 2009 the year for me to really learn Rhino, and I'm going to make a very solid understanding of T-Splines my 2010 goal. I really love the fact that this forum is small but attended by others who are so willing to share and help.

-Sky