[Bollonga]

Originally Posted by Far

2. check domain extents

I've made the domain larger in the y direction (pic 1). I've changed slightly the node distribution but divergence appear at just the beginning in Fluent.

Originally Posted by Far

1. Open up mesh in far field. In other words do not carry forward mesh sizing and growth ratio in farfield. Instead make spacing uniform and use match edges.

I've also checked the mesh in thread http://www.cfd-online.com/Forums/ans...take-mesh.html
The problem is I had the "copy to all parallel edges" option enabled and now when I select "copy to selected edges" it doesn't work. If I change the bunching in the farfield vertical edges (in red circles in pic 2) the rest of parallel edges change too (blue circles).
Also, when I try to match one of the farfield edges with the other ones in the farfield it displays the message:

Warning: edges {103 25 -1} and { 109 26 -1 } don't match

[Bollonga]

I've managed to improve the mesh at the farfield but divergence problems are still there.

For a steady case with k-omega SST, TI=0.5% and turbulent length scale of 0.07m (7% of chord length) reverse flow appear from the first iterations until Turbulent Viscosity Ratio is limited to 1e5 in too many cells and divergence occures. The same happens for the transient case, starting at a timestep of 1e-8s.
Is it a domain extent problem?
Is it a mesh density problem?
Is it an initialization problem?

I upload prj, tin and blk files.

I'm gonna open a new thread just for this issue. I'll keep posting here for the blade tip and wind turbine related issues.

Thanks!

[Far]

turn off Nodes locked

turn off parameters locked and try

domain extents : upstream 15 C and downstream 25 C Try and come back.

What is your angle of attach?

Reverse flow and turbulence intensity occurs in Fluent. No problem if it not creating problems for solution convergence. In my expereince this happens in very high aspect ratio cells in far field. For that matter, I recommended to open mesh.

[Bollonga]

Hi again!

Returning to the blade tip blocking, I've managed to do a blocking but I still need to improve quality of some elements as it's under 0.2 in some corners and leading edge.
I upload the files.

Any comment or suggestion is well appreciated! Thanks!

[Bollonga]

Hi again!

I managed to solve the blade tip blocking, now I get quality over 0.2 in that area.

Now I'm facing problems with the lower blade part, the cilindrical union with the shaft. I've made a blocking and been moving vertices but I cannot get rid of some low quality elements (less than 0.01%) (blue elements in the picture)
Here you are the files so anybody please could help me to correct that low quality elements.

Thanks a lot!

[Bollonga]

Okay, solved. Just try and error moving vertices and changing nodes distribution.

[Far]

You have done a wonderful job. I would like to see when you combine blocking of this block with upper block where trailing edge is sharp.

[Bollonga]

Thanks Far!

Now I've merged this blocking around the blade with the rest of the domain. The issue now is I have two or more faces of different blocks for just one surface. How can I fix that, cause it's messing it up when I mesh that surfaces. Is there an option to match surface meshes?

Thanks!

[Far]

Better option would be to give finite thickness at trailing edge and apply full o-grid around lower and upper blocks.

[Bollonga]

Originally Posted by Far

Better option would be to give finite thickness at trailing edge and apply full o-grid around lower and upper blocks.

Sorry, I don't see what you say.
Once I have my prism around the blade blocked and meshed (picture 1), how do I complete the cylindrical 120º sector (picture 2)?

[Bollonga]

I've already solved that issue by deleting the internal surfaces.

Now I have low quality elements in some areas.
One of them is a low angle sector, lines go from a vertex (line) to a circle arch (cylindrical surface) and near the vertex/line there are skew elements which I don't know how to fix (red circles in the picture). How can I avoid that?
There are also some random low Q elements (yellow circles) that I can't fix either. Which can be the cause for that ones? How can I avoid them?

[Bollonga]

I'm considering to import different meshes for each part of the blade prism and domain sectors.
Is merging different meshes recommended even if contact surface mesh would not match?
I know fluent can handle that, but is it more expensive than just a unique mesh?

[Far]

why not a single mesh? I would start with one block and then using splits will capture wind turbine.

[Bollonga]

Originally Posted by Far

why not a single mesh? I would start with one block and then using splits will capture wind turbine.

Well, I already have the prism blocking around the blade and it was a lot of work so I don't want to repeat all that.

I would also like to use the domain mesh with different blade configurations so I would just need to change that blocks.

What are the drawbacks of using different meshes?

Besides that, I have some low quality elements in the corners of some sector block in the domain (pics), I don't know how to improve their quality . Any suggestion?

[Bollonga]

Merging different meshes or blocks is not giving good results so now I'm trying to expand the prism blocks and then split them as Far suggested.

The problem is that some vertex seem to be linked somehow so I cannot move one without moving the other one. This avoids an appropiate block modification. I have deleted all associations but they still are linked! See nodes 960 and 970 in the picture.

Why does that happen? How can I fix it? I guess it can be a rather simple issue but I'm stuck!

Thanks

[Bollonga]

Originally Posted by Bollonga

Merging different meshes or blocks is not giving good results so now I'm trying to expand the prism blocks and then split them as Far suggested.

The problem is that some vertex seem to be linked somehow so I cannot move one without moving the other one. This avoids an appropiate block modification. I have deleted all associations but they still are linked! See nodes 960 and 970 in the picture.

Why does that happen? How can I fix it? I guess it can be a rather simple issue but I'm stuck!

Thanks

I've just had a happy idea, vertices were periodic, removing periodicity just solved my problem.

[Bollonga]

Hi again.

Once I've finished my mesh (1.7 Gb) I've started the simulation in Fluent.
I've started with the steady laminar case. I've chosen coupled pressure-based solver but I have some questions as I've posted in the thread:

http://www.cfd-online.com/Forums/flu...onal-cost.html

It takes too long (45 min aprox) to do each iteration for Courant=200 and cont and mom URF=0.75 (default values)
There's also reverse flow in the outlet, it was decreasing but in the end it's starting to increase.
Should I change Courant number and URF? Is there a better solver option for this case?

Thanks a lot!

[Far]

Mesh size ?

[Bollonga]

Originally Posted by Far

Mesh size ?

Total elements : 8250475
Total nodes : 8118741

[Far]

can you make mesh under 1 million size? That is better starting point to set important parameters.