CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   [GAMBIT] What's your "favourite" 2D airfoil mesh. (

Seba January 18, 2011 18:50

[GAMBIT] What's your "favourite" 2D airfoil mesh.
3 Attachment(s)
My first post here, mechanical engineering student, CFD beginner.

In my current project, I have to make some CFD analysis, both 2D (airfoil) and 3D (wing), however despite I'm quite happy with Cl results, pressure distributions, etc, I'm not happy with Cd, after reading some threads here it seems that it's a common problem special for newbies. The problem is that I need cl/cd chart, instead of cl/AOA, and proper cd values are essential for me.

I have to make analysis for 3 different airfoils NACA 64A212, GA37A212, NACA 747A315, (previously I was analysing turbulent airfoils usually with S-A model), but all three current airfoils have in at least 30% laminar distribution, after reading some threads here, it seems that I have to try other model for analysis. But before I'll make it, I want to improve my mesh, so for this reason I started this thread, I was testing probably most types of different approaches, but I'm wondering about your experience.

First I was using structured grid very similar to that in Cornell tutorial, next I was using let's call it "mixed approach", boundary prism layers and unstructured grid(Tri/Pave), like this in 1 attachment, currently I'm examining structured grid (very similar to that which I found in Ansys tutorials), second and third attachments are my current examples.

I want also to introduce my approach, maybe someone can give me some directions, first I choose proper thickness of the first prism layer for this I use this webpage, for overall boundary thickness I use this webpage, do you guys also using those pages or do you have other ones?

What is your "favorite" type of mesh, any recommendations, I know that in theory structured mesh should be "faster" in calculations and a little bit more accurate, personally I noticed some differences, but were marginal.

I have also one question about Gambit, for example using pages mentioned above I found out that for my current analysis (Re 6x10^6) my first prism layer should be around 5.2e-006 m thick, there is no problem with creating it using with boundary layer tool, but I have problems when I try to create it with Quad/Map elements, I'm getting skew elements (I tried dozens different settings without much success), however there is no problem when I set thickness of the first layer to around 5.2e-005 m, I know that probably I should change some settings in Edit>Defaults>Mesh, but I'm not sure which ones, does anyone know, what should I looking for? I know that I can solve similar problems for boundary layer with edit>defaults>blayer>angle_smooth/smooth_continuous_sides or for 3D boundary edit>defaults>vertex>max_end_dangle, etc.

I would be grateful for any recommendations, directions, links, etc.

DoHander January 18, 2011 21:32

5.2e-006 m ... hmmm try to see your problem in different way like for example using centimeters instead of meters, so basically your first layer will have a height of 5.2e-004 now.

(The problem with Gambit is that he has tolerances based on simple precision so you will have about 7 exact decimals. At least I think the code works only in simple precision, I didn't have a chance to see the actual source code.)


All times are GMT -4. The time now is 18:53.