[Bollonga]

I've been moving vertices and changing nodes number, negative volumes change in number and distribution (see pics) but they are always there.
I've been checking http://www.youtube.com/watch?v=tYrbScUH9RE for some blocking ideas but every blocking strategy I try to imitate gives negatives volumes for me.
Is there any base problem?
Isn't it a common mistake in Icem practice? Cause I'm facing it too often!
Any suggestion? Thanks!

[Far]

please attach .tin

[Bollonga]

Here you are

[Far]

blocking 1 is perfect. No problem of negative volume

[Bollonga]

I've converted one of the meshes to unstructured and then exported it to fluent and it doesn't detect any negative volumes! I had tried that before with no success but now it works...
How does the pre-mesh to unstructured step work?
Apparently Negative Volume indicator doesn't mean too much...
Which quality parameters should I pay attention to? Quality, angle, volume change?

Thanks!

[Far]

Qaulity, min angle and volume change. Also dihedral angle (should be less than 165)

[Bollonga]

Originally Posted by Far

Qaulity, min angle and volume change. Also dihedral angle (should be less than 165)

Which values would be appropiate for that? I've read min quality > 0.2, min angle >18 and max volume change??

[Far]

Min quality > 0.3, min angle > 18 amd volume change <20

[BrolY]

Originally Posted by Bollonga

I've converted one of the meshes to unstructured and then exported it to fluent and it doesn't detect any negative volumes! I had tried that before with no success but now it works...
How does the pre-mesh to unstructured step work?
Apparently Negative Volume indicator doesn't mean too much...
Which quality parameters should I pay attention to? Quality, angle, volume change?

Thanks!

You won't see negative volume unless you do a check mesh on the unstructured mesh. But if you look at the quality, you would find a lot of elements with a quality around 0. Those elements are your negative volume elements

About the values of quality, Far is right, but quality, > 0.2 is good too.

[Bollonga]

Originally Posted by BrolY

You won't see negative volume unless you do a check mesh on the unstructured mesh. But if you look at the quality, you would find a lot of elements with a quality around 0. Those elements are your negative volume elements

For the same mesh I get negative volumes at the pre-mesh and quality over 0.6 at the unstructured mesh. I don't understand why. (attached files)

Then in fluent it doesn't show any negative volumes.

However, for this mesh, my fluent case diverges from the beginning. I'm using 33.87 m/s, k-ep standard with TI=1% and TLS=1e-4m at the inlet. Adaptive time-step starting at 1e-8. I've tried reducing under-relax factors one order of magnitude but it doesn't avoid divergence.

[BrolY]

No negative volume in this project, the pre-mesh quality > 0.8.

[Bollonga]

But if I select Volume in the quality menu, select histogram form -1 to 0 you can see there are negative volumes. In the image there's the histogram from 1e-15 to 0 showing many elements, even if quality histogram is >0.6...

[Far]

check " only visible index range"

[Bollonga]

Originally Posted by Far

check " only visible index range"

Nothing changes

[Bollonga]

I've discovered that just changing the min distance of the edge near the wall from 1e-5 to 5e-5 it improves quality a lot, from having near zero quality elements to a minimum quality of 0.6. It's pretty sensitive to node distribution!

[Bollonga]

Okay, so negative volumes is solved so far. Now I want to make some test with the mesh but it diverges no matter what I do.

Originally Posted by Bollonga

However, for this mesh, my fluent case diverges from the beginning. I'm using 33.87 m/s, k-ep standard with TI=1% and TLS=1e-4m at the inlet. Adaptive time-step starting at 1e-8. I've tried reducing under-relax factors one order of magnitude but it doesn't avoid divergence.

I've tried ke-epsilon standard (enhanced wall treatment) and k-om SST aswell for steady and transient case. Mesh quality is good, volume change is not that good I guess (see histogram and element distribution in picture 2).

Could it be the problem for not reaching convergence?
Could I reach convergence with a coarser mesh?

Thanks!

[Bollonga]

Originally Posted by Bollonga

volume change is not that good I guess (see histogram and element distribution in picture 2).

Could it be the problem for not reaching convergence?
Could I reach convergence with a coarser mesh?

Thanks!

I thought an appropiarte volume change was 20% so I understood that higher than 1.2 was not good. But as I've read in the user guide, less than 20 means 20 times the maximum neighbour volume. So my volume change dsitribution is pretty good! So I don't know what can be wrong

[Far]

Originally Posted by Bollonga

I thought an appropiarte volume change was 20% so I understood that higher than 1.2 was not good. But as I've read in the user guide, less than 20 means 20 times the maximum neighbour volume. So my volume change dsitribution is pretty good! So I don't know what can be wrong

Just few thoughts:

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.

2. check domain extents

[Bollonga]

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 haven't understood it completely. You mean I shouldn't keep edges parallel until far field but to open them?

And use uniform spacing instead of hyperbolic or geometric or whichever?

Which edges should I matche to which ones?

Thanks!

[Far]

Download file from first post in following thread and look at the meshing in farfield.

http://www.cfd-online.com/Forums/ans...take-mesh.html