[samurai_01]

Hello! I need help with meshing in ICEM after banging my head against the wall for long.

I am trying to mesh two concentric spheres in ICEM using blocking. I used two methods:

First: I created a 3D sphere, by making a quarter-circle and then doing the steps (2d-Blocking, mirror, Extrusion by rotate) shown in images titled Method1_step

Second: I tweaked the steps of the sphere-cube tutorial to mesh the two spheres. I don't do any association, except I snap vertices for the 3D blocks for outer and inner spheres. Also, I create an O-grid around the inner-sphere. The images have been attached as Method2.

In both methods, I encounter the negative volume error. For method 2, the error is shown in the 3x3x3 determinant quality.
When I try to smoothen pre-mesh using the Pre-Mesh Smooth option, and although it smoothens the quality of determinant to above 0, the change is not reflected in the pre-mesh and if I do any other operation on the mesh, it reverts to previous quality.

Can someone please help me with this issue? I can try some other method, if anyone can suggest.

Thanks !

[samurai_01]

These are the images for method 2 I've used. I can post the project files/, more images and explanation if required. The unstruct mesh check gives NO error/possible problems, and yet the solver tells me that there are negative volumes in mesh. This is one of the worst problems I've faced in ICEM meshing, please help!

[bluebase]

Hi samurai_01,

Have a look at the blocks whose corners show low quality elements. Just the respective block.
You might see that that corner with the bad element is almost squashed to a flat plate...

Imagine a dice. Look at one corner where you can see all three attached edges. Now try to imagine how the cube/dice would have to look when all three edges are found in same flat/2d plane. What does this mean for the other faces? And what volume would this object have?

Now, when we investigate the surface of the sphere, we can assume that it is locally flat, since the curvature is low at a large radius.

Another observation is, that you created an o-grid around the inner sphere. But now, the block containing that o-grid is a cuboid again. And that cuboid is now inside a sphere again. This leaves you with a case which is similar but in a way inverse to the o-grid around the inner sphere. This requires the same solution strategy, an ogrid. In method 1 you correctly extended the o-layer to the outer sphere, but method 2 does not have this merit

Bottom line is, you need to fix your blocking in a way that keeps every block as close as possible to a perfectly rectangular cuboid.

I see two solutions straight forward solutions.

The first, create an ogrid with a single layer just as in the method1 where the single layer spans across the radius from the inner to the outer sphere.
The advantage is, the block structure is very simple. The disadvantage is the huge difference in mesh density between regions of small radii and regions of large radii.

The second approach is to use the blocking you have from method2 and just introduce another o-grid all around. That way you can reshape the degenerated blocks at those 'corners'. I imagine just selecting all blocks and hit create in the o-grid feature should do the trick. Though, you might need to add the block in the void of the inner sphere, or select the boundary faces on the inner sphere.
The advantage is you can continue with what you have. Moreover, the mesh density distribution is narrower. The disadvantage is, you will have to do some tedious manual vertex moving.

Best,
Sebastian

[samurai_01]

Dear Sabastian,

Thank you for the reply, let me just quickly attempt these and get back to you. I

The first, create an ogrid with a single layer just as in the method1 where the single layer spans across the radius from the inner to the outer sphere.

Can you please elaborate this a little more? I did not understand, I should create a 2D o-grid and then revolve it, right?

[samurai_01]

Dear Sebastian,

It worked!! I got rid of all the negative volumes by method2 you suggested. I have attached the image of the quality check and a cut plane!

Thanks a ton!! I can't thank you enough