Meshing a s809 wing in Gambit
 

Hello everybody,

Im new in CFD so my questions may seem a bit dumb but i cant help it so here it goes!Im currently trying to simulate a 3d multiphase (airstream+rain drops) flow over a s809 wing.So i attempted to mesh the wing and the farfield and the result was a mesh that kept the solution from converging in Realizable k-e model of 2nd order.After reading many answers in similar situations in these forums i tried to make a 2nd mesh in order to get convergence.The thing is that although the solution did converge,the results seem to be a bit off the experimental ones plus there is a highly skewed element in my wing which i dont know how to smooth it.
I am simulating a S809 wing (initially in single phase flow-air-and then in multiphase flow) in Re=2x10^6 using k-w sst, realizable k-e and spalart allmaras turb models.

Here are the 2 meshes i used for this simulation:

I would appreciate it very much if someone could give me a hint on how to improve the 3.mesh because of its highly skewed element at the trailing edge.Here are the vertices that i used for the airfoil

post picture of your mesh, and especially where you get skewed elements

thanks for the reply Max,I attach below some pictures

from your picture 7, it seems you meshed also your wing (solid part).
why?

I am simulating the 3d flow over the wing so i thought that i should have the wing unmeshed for the part where there is the actual wing and have the wing meshed where there should only be the flowfield,in order to study the effects due to induced drag

if you delete the wing volume, do you see your wing as hollow?
You should have substracted your outer domain with your wing.

Yes Max i deleted the volume of the meshed wing (and the mesh in the face of the wing where z=5/I meshed that face because i got a warning when i tried to export the mesh,concerning boundaries) and theres a hole in my mesh.As for substracting,i didnt substract the wing initially (in 2d mesh) i just created the hole out of its adjacent edges.

then delete the wing volume, and compute on the rest of the grid

Initially i had my wing unmeshed for 0<z<5(theres the actual wing in the calculations) and had it meshed for 0>z>-2(theres only air in calculations).If i delete the wing volume (which is -2<z<0) then theres a hole in the mesh and then we assume that the wing extends from the left to the right side of the flowfield,or?Excuse me for those dumb questions :)

what im trying to say is that if i calculate the grid just like that,dont we assume that we have got a wing all over the width of the flowfield and as such we neglect the 3d effects?

ok I understood
The problem comes from your wing extension.
So you have to fix this skew element.
You can try to split the trailing edge (sharp angle), and mesh this surface with only one tri element.
Then propate this split in the z-direction
Attachment 23700

So i split the upper and lower edges at the second-to-last node and i create a triangle as you show in your picture.Then i mesh the 2 parts of the wing separately (the whole wing and the last triangle-like part which i have created with the split).For the latter i use 1 tri element.Is that correct?

yes, it is.
But you will need another split on the right (next to the triangle), especially if you want hexa mesh. Hence the triangle "lock" the thickness of your "wing" with one element. With this second split, you can generate a transition from x elements to one element.
Then, as you refined your mesh on the wing, you will also do this refinement on this area

I didnt understand the second split next to the triangle :confused:

on my picture I represented this volume with a rectangle (inside this rectangle you can mesh with tri or pave)

So the wing is split into 3 parts:the first one to the left(the biggest),the 2nd one(the trapezoid one) and the third one (the triangle one)?

yes:
hexa > tri or pave (transition) > tri (one element)

Then another split along (xy) for refinement at the end of the wing

After trying to split the right most area (the triangle one) my geometry changes and the edges seem to be converging :confused:

wait your angle will be to sharp and will generate skewness.
Question, is your wing section constant? on your picture 7 , the trainling edge is represented in light blue, right?

yes it is the light blue.I have attached the airfoil data in a previous post if you need the vertices

I have divided the airfoil into 4 sections

ok then move your split points far from angle (ie: links)

I have deliberately put the split points in those exact spots because it was at those points that the skewed elements appeared in the first place.I will try moving them (to the left or to the right of the vertices?does it matter?)

far from trailing edge

So the triangle section of the wing will get bigger now(it will be like the outer triangle of ur pic) and i need a transition area aswell in front of it again?

Attachment 23710
Something like that :rolleyes: (I alreddy see next problem, if you have refinement at the trailing edge)
Maybe someone with airfoils experience can give you more tips.
At the other hand, if you are not interested in this extremity, you can simply delete the adjacent volume

I have done something like that and im working on the final section, the triangle part the one i should 1 tri element mesh,but i dont know how to do so.The first section was meshed using quad pave and the other 3 using quad map.Till now the most skewed element is in the first section with a value of 0.39 or so.But i m wondering how can i choose only one element to mesh my section with

if you mesh the last section, the skewness will be around 0.9
But you won't be able to mesh it since you didn't create the transition.
To your question, you can try to sweep all this surfaces in z-direction. Gambit will generates the volume (and you will have to connect all surfaces to be sure, that everything is ok)
But your last split in trailing edge is to close from the angle, that is causing the skewness trouble
Else on your picture 5 you can delete the triangle surface, and you forget your trailing edge. Since your split is very close to the angle, it won't affect the solution

So basically even if i create that transition area and even if i mesh the last section,it will be of 0.9 skewness regardless the transition area?By transition area you mean the part before lthe triangle section meaning part 4 but it has to be meshed differently than what i did in order to be an actual transition section i presume?On your suggestion about dumping the triangle part,is it more accurate to just dump it than keep it along with its skewness?

it will give you less work ;)

Here is the topology without meshes

Heres a new mesh

Worst Aspect Ratio element