CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Residual problem of 2D airfoil simulation (http://www.cfd-online.com/Forums/fluent/74875-residual-problem-2d-airfoil-simulation.html)

caohan April 10, 2010 11:28

Residual problem of 2D airfoil simulation
 
4 Attachment(s)
Hi all

I met a puzzle when I am doing a 2d airfoil in different AOA, the reynolds number is 30000, boundary condition is pressure far field, chord length is 1m. solver is implicit density based, viscous model is sparart-allmaras. when the simulation ran in a smal AOA, residual pattern oscillated in a small region, but the lift, drag and momentum coefficient converge easily. by increasing the AOA, result become worse, residual, lift, drag and momentum coefficient oscillated widely. can sb give me a clew for that, is there something wrong with my mesh or boundary condition. I put 4 pics into the attachment( two of them are residual and cd figure in 5 degree, the others are figures taken under the 90 degree)

Regards
Han
Attachment 2905

Attachment 2906

Attachment 2907

Attachment 2908

DoHander April 10, 2010 20:38

What is your mesh size (how many chords around the airfoil in all directions) ? I suspect you need to use a larger computational domain for large AOA.

Do

caohan April 11, 2010 05:31

Hi do
 
Quote:

Originally Posted by DoHander (Post 254114)
What is your mesh size (how many chords around the airfoil in all directions) ? I suspect you need to use a larger computational domain for large AOA.

Do

how can I check the chords around the airfoil, by the way, my computational domain is 20 times of the chord length, is it big engough? I will try a bigger domain for large AOA and see what will be happened. Thank you fou your reply

Han

Chris D April 13, 2010 01:01

You're probably getting reflections off of the boundaries. You should move your boundaries out, like DoHander said, and also make sure your really stretch the grid as you move out to the far-field. By stretching the grid, you'll damp (hopefully most of) the waves before they get to the boundary, so then they can't be reflected back to the airfoil.

edit: I should say that you might be getting reflections. There could be other problems, but this is the first thing I would try.

caohan April 13, 2010 08:20

1 Attachment(s)
Quote:

Originally Posted by Chris D (Post 254395)
You're probably getting reflections off of the boundaries. You should move your boundaries out, like DoHander said, and also make sure your really stretch the grid as you move out to the far-field. By stretching the grid, you'll damp (hopefully most of) the waves before they get to the boundary, so then they can't be reflected back to the airfoil.

edit: I should say that you might be getting reflections. There could be other problems, but this is the first thing I would try.

Hi Chris
Thank you for you reply, now I expand my domain, it is 40 times of the chord length, but still the result oscillates, does it mean the domain is still not big enough? I have nothing to do with that now. I post my mesh here, wish you can help me solve the problem.

Thank you
Han

Chris D April 13, 2010 11:55

It looks like you're not stretching the grid enough. The point is to stretch it so that the waves you are trying to push out of the domain become unresolved near the boundary. This will cause them to be taken out by the inherent dissipation of the numerical scheme. If the waves are taken out before they get to the boundary, then they won't be reflected back from the boundary.

Try a growth rate of 1.2 or more once you're sufficiently far from the airfoil, with a domain of 20c. Make sure you stretch it in a smooth manner, or the code might blow up.

Also, what AoA are you running? For high AoA's the problem becomes unsteady, which could be what the oscillations are telling you. See this youtube video:

http://www.youtube.com/watch?v=Ti5zUD08w5s&NR=1

caohan April 13, 2010 12:08

Quote:

Originally Posted by Chris D (Post 254444)
It looks like you're not stretching the grid enough. The point is to stretch it so that the waves you are trying to push out of the domain become unresolved near the boundary. This will cause them to be taken out by the inherent dissipation of the numerical scheme. If the waves are taken out before they get to the boundary, then they won't be reflected back from the boundary.

Try a growth rate of 1.2 or more once you're sufficiently far from the airfoil, with a domain of 20c. Make sure you stretch it in a smooth manner, or the code might blow up.

Also, what AoA are you running? For high AoA's the problem becomes unsteady, which could be what the oscillations are telling you. See this youtube video:

http://www.youtube.com/watch?v=Ti5zUD08w5s&NR=1

Hi Chris
Thank you so much, I would try to stretch the grid with 20c domain, by the way, my AOA is 90 degree, does the big AOA cause the oscillation and have nothing to do with the mesh?

Chris D April 13, 2010 12:55

Yeah, I would think that would be an issue. At that point, I don't think it can be solved as a steady problem anymore. However, you'll still want to stretch the grid.

How are you setting the AoA? I can't tell from the picture if the grid is with AoA = 0 or 90. If it's 0, then are you changing the velocity vector in FLUENT? If so, then, from the picture you showed, your grid won't really be aligned with the flow anymore.

caohan April 13, 2010 14:36

Quote:

Originally Posted by Chris D (Post 254453)
Yeah, I would think that would be an issue. At that point, I don't think it can be solved as a steady problem anymore. However, you'll still want to stretch the grid.

How are you setting the AoA? I can't tell from the picture if the grid is with AoA = 0 or 90. If it's 0, then are you changing the velocity vector in FLUENT? If so, then, from the picture you showed, your grid won't really be aligned with the flow anymore.

yes, I set the AOA by changing the velocity vector, if I rotate the airfoil and let it become vertically with the grid not horizontally, is it better than changing the velocity vector in boundary condition?

Chris D April 13, 2010 16:03

1 Attachment(s)
Quote:

Originally Posted by caohan (Post 254464)
yes, I set the AOA by changing the velocity vector, if I rotate the airfoil and let it become vertically with the grid not horizontally, is it better than changing the velocity vector in boundary condition?

Either way, you'll end up with the same thing. You'll get this:

Attachment 2948

Changing the velocity vector works for low AoA, but I don't think it's a good idea for high AoA. Your grid wouldn't be aligned with the flow anymore, which I think would cause you some problems with you solution. Also, from the perspective of the rotated grid, the far field boundaries are actually pretty close to the airfoil, which could explain the oscillations in the residuals.

I would suggest using a different grid for high AoA. Build a C-grid around an airfoil at a high AoA, like 80 deg or so. That way, you'll be able to align the grid with the flow. Then, if you want to run a range of AoA, changing the velocity vector by ten degrees (i.e., from either 80 deg to 70, or 80 deg to 90) won't be such a big deal.


All times are GMT -4. The time now is 23:25.