Simple Wing Meta Surface

[Jörg 29]

Please find attached an example for a very simple (parametric!) wing made from a NACA 4-digit profile, modeled in CAESES.

 

For more information about aerodynamic optimization, see this aerospace section.

What I did: I first created the section in the object tree and then introduced a set of parameters. From the result, I created a feature definition from selection (remember: do not select the parameters in order to have them as input arguments). This definition is the input for a curve engine.

I added simple bspline functions to this example, feel free to exchange them. If you like, introduce design variables for these functions (i.e. their control vertices) for automatic changes.

You can also control the wing in x- or y- direction by adding new input arguments for the feature definition (e.g. call them dx and dy and set them at the translation object in the feature definition). Of course, create your own functions for dx and dy, too, and set them at the curve engine.

Cheers
Joerg

simplewing.fdb

[Mr. Marius Piiksi Dahli 0]

Thank you very much!

[Mr. Marius Piiksi Dahli 0]

Is there anyway to change the Naca4D definitions, so the start and end positions meet at y=0? I do not want the gap at the trailing edge of the wing.

 

Thanks!

[Jörg 29]

See the attached file for an example where fsplines are used to close the trailing edge.

 

Cheers

Joerg

trailingedgeclose.fdb

[Mr. Marius Piiksi Dahli 0]

Thank you Joerg!

[Mr. Marius Piiksi Dahli 0]

Hello!

 

I have another question about closing the airfoil this way. When I have finalized the wing, and exporting it to the CFD software, the wingshape looks quite nice. The problem is that there is a strange mesh along the wing, and it seems to be connected this way of closing the trailing edge. When I do the CFD calculation, this line effects the result. There is an abbrupt change of pressure gradients along this line, and I cant understand why. Do anybody have any suggestions?

 

Is there another way of creating a closed trailing edged NACA0012 without using this method?

 

Best regards,

Marius

[Jörg 29]

Hi Marius,

 

Do you have a screenshot of your mesh? I am just interested in what it looks like. Another method to close the trailing edge might be to introduce a simple trailing edge fillet curve (something like a circular fillet etc). See the attachment for an example. This should at least avoid the undesired pressure distribution. What do you think?

 

Joerg

trailingedgeclose2.fdb

[Mr. Marius Piiksi Dahli 0]

Hi Joerg!

 

Thank you for helping. I tried adding a new feature definition, based on the symmetric NACA00XX foil, as mentioned here: http://en.wikipedia....it_NACA_airfoil

I think that is a good option too. I added the thickness as a variable, and used 12% thickness as default (naca0012). As said on wikipedia, I changed the last variable (to -0.1036) to get a closed trailing edge.The only problem then is I cant use any refinements, and the curve is very rough. The leading edge is like a square. Any way I can modify how many points it should use? I tried using an image curve, but I need to set the resolution to 10000 to get a smooth curve. I have to have a similarly high res on the coons patch, and I believe it is a bit too much?

 

I added photos from the last wing, with the strange phenomenon. The trailing edge is on the lower left on all the pictures. I also added a picture from the symmetric foil distribution I created, with the rough resolution.

 

Thank you so much, Joerg!

 

Best regards,

Marius

 

edit: It doesnt look like the CFD software should have problems. There isnt a visible "bend" at all? The last picture shows the root of the wing, with the seperating line..

[Jörg 29]

Hi Marius,

 

What do you export, IGES or STL? For more "density" at the LE you could try to change the curve's parametrization using the command

 

curve.setParametrization("dense inner")

 

in your feature definition. This works well for symmetric profiles, I think. Also, you could split the entire profile (or even later on the wing) by using images. I guess you use a coons patch as the final wing surface, right? Split the coons patch again with image surfaces by using the image surfaces' domains. See also the Getting Started tutorial "Basic Preprocessing" (last page).

 

Joerg

[Mr. Marius Piiksi Dahli 0]

Hi Joerg!

 

I am exporting to STL->StarCCM. Iges doesnt look good on the curved part of the wing. It gets very messy, so STL is a better option it seems.

 

I am definitely using a Coons surface. Only using half the naca curve, both in the wingroot and the wingtip. I will try to make a new image surface, and see if that helps. Allthough, last time, it seemed to bring the closing trailing edge function, looking at the mesh, even though I used a new image curve to represent the naca profile. Not sure why it does that.

 

I tried the curve.setParametrization("dense inner") command, but it didnt do anything. Same problem at the tip. Ideally the curve starts with tangent 90.. There must be some way to change the density of the control points, without using an image curve?

 

Thanks Joerg!

 

Edit: I added another image of the curve defined from the definition. Why isnt it possible to change these variables? (approximation)

[Mr. Marius Piiksi Dahli 0]

I changed to "dense outer", and that worked very nice. :D

 

Thanks a lot Joerg, great tip! Is it another command, that allows me to adjust how many control points there should be?

 

Edit: Will changing this variable make the curve smoother? (in the pic). Also, is it possible to change in the feature definition?

 

edit: I tried making an image curve of the new curve with "dense outer". I used 200 render solution, and it looks very nice. I believe this is very good, and I wont need any better than this.

 

Thanks a lot Joerg! Really appreciate it.

 

Best regards,

Marius

[Jörg 29]

Hi Marius,

 

Sounds good to me ... Perfect. Just for your information: We added the approximation category to the generic curve mostly for performance reasons. The generic curve is rather slow with regards to position evaluation so you can approximate it with a NURBS curve which is very fast.

 

Note that you can also introduce your own density function instead of using "dense outer" etc. - all you need is a strictly monotonically increasing function in the xy-plane in the range [0,1] x [0,1], starting at x=0 with y=0 and ending up at x=1 with y=1. Just in case you need more control ...

 

Cheers

Joerg

[Mr. Marius Piiksi Dahli 0]

Hi Joerg,

 

I will have to try that out. I am making a reference wing, which has a semi-span of 5m. Naca0012. For that wing, the curve I used worked perfectly. I am also making a crescent shaped wing, and for that wing, it didnt worked. The tip is the same shape, but chord only 1cm. With the coons patch, it looked more like a flying saucer than a crescent shaped naca0012. I am a bit frustrated, since I have tried many different ways now, but they all seem to fail. Is it possible to "lock" the curve, so the tangent starts out "90"? That could help. I will also try to make my own density function, to see if that helps. I have an exam tomorrow, but will try it out tomorrow afternoon. I really need this to work.

 

Thanks a lot for all the help!

 

Cheers!

Marius

[Jörg 29]

Hi Marius,

 

Can you attach your attempt (fdb file) of modeling the crescent wing, including your problematic resulting surface? I could have a closer look at it....

 

Joerg

[Mr. Marius Piiksi Dahli 0]

Yes, I definitely can. In 15 mins I will upload the file!

 

Thanks a lot!

[Mr. Marius Piiksi Dahli 0]

The parameter "TipPos" is where I adjust how much the wing should bend, in the x-axis. The z-axis is fixed to 5m.

 

The "OLDconfiguration" is the one that bugs in CFD,because of the way I closed the foil. Something strange occurs because there is a denser mesh, but I already showed that to you. I still dont understand why the CFD software gets a sudden change of pressure gradients along that line, but that foil looks a lot better along the leading edge..

 

Thanks!!

NACACrescentShaped.fdbc

NACACrescentShapedOLDConfiguration.fdbc

[Jörg 29]

Hi Marius,

 

Please find attached my version of the wing. I used a meta surface (see also the provided tutorials in CAESES/FFW). Finally, I created a trimesh which has an adaptive triangulation (this covers your density issues). For export of the wing, just select the final trimesh wing and choose file > export > STL ... Note that I made the other scope invisible (just click on them again in order to visualize everything that is in there).

 

I hope this helps ...

 

Cheers

Joerg

NACACrescentShaped.fdbc

[Mr. Marius Piiksi Dahli 0]

Joerg, that looks beatiful! I can't thank you enough. I will redo the steps, to learn how to do it.

 

Really, thank you so much!

 

Cheers,

Marius

[Jörg 29]

Perfect... Good luck!

 

Joerg

[Mr. Marius Piiksi Dahli 0]

Joerg, I need to thank you again. You really helped me deepen my understanding of how and what this program can do. I am so happy with the result! I will send you a PDF copy of the project when I am done, if you are interested. I am running the CFD code now, from 2-6 degree angle of attack, and will compare to an elliptical wing by the end of the week. Getting help from a local CFD company, both with licenses and servers to solve it in StarCCM. The setup looks totally sweet, and I can't wait to see the results!

 

Thanks a lot for your help, sir!

Best regards,

Marius

[Jörg 29]

Perfect! And thanks in advance to send us a project summary soon...

 

Good luck,

Joerg