[pulplus]

Hi to everyone! I'm new to ICEM and I'm using it for my thesis project.
I have to mesh a stepped diffuser (a conical one followed by a sudden expansion). I did it using a double ogrid in the sudden expansion zone as you can see in the pictures.

img1.jpg
img2.jpg
img3.JPG

I have a small first cell height along all the pipe (2*10^-6 m). The problem with this blocking strategy is that in the sudden expansion block I have the same nodes distribution of the previous blocks near the axis. So I have small cells (like the ones near the wall) in the middle of the flow and a bad growth ratio, as you can see.

img4.jpg

I tried to use the block refinement option, but I read on this forum that it is managed by fluent for 1/2 or 1/3 levels, which are not usuful in my case.I should go with a 1/40 level in my case, but I get uncovered faces issue and can't read the mesh in fluent.

Another way I tried is to not propagate the ogrid in the sudden expansion zone in this way:

img5.jpg

But in this way I get very skewed elements in the small block at the interface between the conical diffuser and the sudden expansion pipe.

So I'm looking for any advice about a better blocking strategy... maybe in order to get a single oblock for each component of the pipe.

Thanks for your help!

[bluebase]

Your initial blocking structure is the best. You only need to refine the internal edges so that size difference between different layers is small.
Small element sizes in the stream is not inherently bad. This is only an issue if your run time becomes too large.

That way, you'll end up with nicely flow-aligned elements at that critical region at the edge of the expansion. Downstream you could slightly increase the element size, so that the former boundary layer fans out to more even element sizes.

Other methods likely create bad elements in important regions. For example, you could let the o-grid/boundary layer blocks follow all the outer surface, not just the inlet pipe's surface. This however, would also result in some misaligned interfaces right at the edge and increase numeric diffusion.

[pulplus]

Firstly, thanks for your reply!

Originally Posted by bluebase

Other methods likely create bad elements in important regions. For example, you could let the o-grid/boundary layer blocks follow all the outer surface, not just the inlet pipe's surface. This however, would also result in some misaligned interfaces right at the edge and increase numeric diffusion.

How can I let the ogrid/boundary layer blocks follow all the outer surfaces?

Thanks!

[bluebase]

Ideally you find yourself a copy of "Simon's tips and tricks" on ICEM. It contains lots of useful information on blocking.

In your case, start from the first blocking you presented. In the o-grid menu, select all blocks in that domain. Then use select faces to select the inlet's and outlet's faces. Apply.
You should get an additional o-grid, which extrudes from all the geometry's walls but the inlet and outlet.