CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Meshing and divergence problem (http://www.cfd-online.com/Forums/fluent/72290-meshing-divergence-problem.html)

ghost February 1, 2010 11:12

Meshing and divergence problem
 
Hi,

I have been getting this error whenever I iterate my simulation, which has both a mass flow inlet and pressure outlet on an airfoil.

"Error: > (greater-than): invalid argument [2]: wrong type [not a number] Error Object: nan"

Is this due to divergence?

I have set courant number to as low as 0.01 and the x and y velocity residuals seem to rise after about 300 iterations.

I suspect this is due meshing problem at the pressure outlet region on the airfoil, which has mesh of high skewness (max 1 cell of 0.96).

Below are 2 pics of my meshing. Is it fine enough, and is the meshing alright? I have tried many methods but cannot remove the error at all.

Thank you.

http://img535.imageshack.us/img535/9959/69765955.th.jpg


http://img63.imageshack.us/img63/2672/88198135.th.jpg

DoHander February 1, 2010 16:52

Try to use velocity inlet and pressure outlet.

Do

DoHander February 1, 2010 16:53

About your mesh ... can you attach the *.msh file used in Fluent ?

ghost February 1, 2010 20:08

I have done a case of mass flow inlet without pressure outlet n it is fine.

And the problematic area seems to be the pressure outlet region, where there is high pressure contours at the highly skewed region shown

ghost February 3, 2010 07:38

Quote:

Originally Posted by DoHander (Post 244489)
About your mesh ... can you attach the *.msh file used in Fluent ?

The mesh file is quite big, about 20mb.

But I would really appreciate it if you could take a look and point out the errors.

http://dl.dropbox.com/u/3077709/10.msh

Thank you.

DoHander February 3, 2010 11:29

http://imagebin.ca/view/0ywLWF.html
http://imagebin.ca/view/8K24Ew.html

You mesh is particularly bad in two places (see the above links to images) at the connection between the BL mesh and exterior mesh (the cells in zone seems stretched), and at the trailing edge of the airfoil - you have a sudden change in mesh dimensions. Small cells on the airfoils and suddenly very large cells.

Do you really need the 2 internal zones in the airfoil ? (I suppose you need them if you use them).

Do

DoHander February 3, 2010 11:39

Also it seems that your topairfoil is not complete, you have two missing edges (not included in this).

The "farfield" is a wall ??? (In Gambit this has a wall boundary condition).

Try to give me more details about your simulation and maybe I can give you some advices.

Do

ghost February 3, 2010 19:50

Those cells near the wall are meshed using the "boundary layer" tool in gambit, hence they are different from the general cells. The general cells are slightly skewed due to the farfield geometry, but most are under 0.7.

I need both internal zones because I am studying the injection and suction effect. They are controlled using mass flow rate, so the internal region can be shorten, but the inlet and outlet must be constant. The 2 missing edges of the airfoil are the inlet and outlet.

I am sure the farfield boundary condition is "pressure farfield", or I will not be able to tweak its velocity in fluent. Most probably this is due to problem in gambit importing mesh files.

Last night I went through the velocity contours of failed simulations and found that the donwstream internal zones have high velocity of 600m/s. Most probably separation due to the sudden turn in flow direction. I have shorten that portion so there is no sudden turn.

ghost February 4, 2010 08:11

Quote:

Originally Posted by ghost (Post 244841)
Last night I went through the velocity contours of failed simulations and found that the donwstream internal zones have high velocity of 600m/s. Most probably separation due to the sudden turn in flow direction. I have shorten that portion so there is no sudden turn.

Even after shortening the internal region, there is still error. I am starting to wonder if Fluent is capable of solving suction problems. There seem to always be spikes in pressure and velocity at the suction region which is causing the error.

DoHander February 5, 2010 13:24

Maybe you can use a UDF to treat the suction region as a boundary condition ? Otherwise you need a much finer mesh then you currently use for the internal regions, especially in the zones where you connect the internal regions and the external flow.

Also have you checked the edges included in your topairfoil boundary condition ? I've noticed that your upper surface is made by 3 or 4 edges, but you've included in the topairfoil BC only one.

Do


All times are GMT -4. The time now is 13:17.