[MrCaplin]

Hello,

I'm new to this forum and also new to the vast field of cfd simulations.
I would be really thankful if someone could help me or give me some advice.

I'm currently working on the set up of a flow reactor and part of the characterisation phase is to perform flow simulations to determine its behaviour.
The reactor has several gas inlets at the top (6). Those streams are combined in a mixing volume (the smaller cylinder) and arrive into the outlet tube through a perforated plate.
The biggest problem I'm having is caused by the perforated plate. My supervisor wants me to use the exact dimensions which means 469 holes with diametres of 0,15 mm. This causes problems during the meshing process because of the computational time and thus flexibility in performing alterations.

So far I've only succeeded in creating a mesh by applying the automatic method through the global mesh controls. The reactor was meshed as a single body part. This results in a very high number of elements (up to 8*10^6), high max. skewness values (~0,89) and relatively low min. orthogonal quality values (~0,1-0,15).
I created the single body by freezing a multibody part and unfreeze it afterwards.

My next step is to mesh the geometry as a multibody part in order to gain more control over the meshes of individual parts. The latest problem I'm facing is that the meshing process seems to take forever as soon as I try to mesh it as a multibody part. I guess this caused by the number of holes in the plate. Another question I have is:
Which is the most appropriate meshing method for my problem?
Structured/Unstructured, Tetra/Hex, MultiZone/Sweep etc...?

I've included several pictures with this post. Two of them show the geometry while the other pictures show the mesh I was able to create so far.

If you need further information please let me know.
Thanks in advance!

Edit: I just realised that my thread title sounds misleading as I'm looking for a solution not offering one. Can a mod change it?

[FJSJ]

There are many factors to keep in mind in order to build your mesh, time, accuracy, numbers of elements, etc.

What I would suggest first, is use a Hex Meshing method, particularly, Sweep or Multizone since Hex Dominant is not recommended for CFD. Try to use sweepable bodies tool and get some clues. At the same time you could use a Body of Influence close to your holes.

Then, depending on holes distribution, you could think of fully structured mesh with ICEM (but this would be a HUGE job, and again, depends on the holes distribution).

Have a look here (this would be a simple example of the approach):
Multiple Pipe In Cylinder Blocking

[MrCaplin]

Originally Posted by FJSJ

There are many factors to keep in mind in order to build your mesh, time, accuracy, numbers of elements, etc.

What I would suggest first, is use a Hex Meshing method, particularly, Sweep or Multizone since Hex Dominant is not recommended for CFD. Try to use sweepable bodies tool and get some clues. At the same time you could use a Body of Influence close to your holes.

Then, depending on holes distribution, you could think of fully structured mesh with ICEM (but this would be a HUGE job, and again, depends on the holes distribution).

Have a look here (this would be a simple example of the approach):
Multiple Pipe In Cylinder Blocking

Hi,

first of all, thank you for your reply!
Well, our main goal is to derive the main flow characteristics and later on residence times. Especially near the mixing volume and downstream of the plate.
I simplified my model again by cutting a piece of the inlet region, reducing the number of inlets from 6 to 3. I will post new images on monday. I could also upload the CAD-model.

About ICEM: I don't think we have the license at our institute. At least I can't find it anywhere on my pc/in workbench. If possible I'd like to do it with the Ansys Meshing tool.
The holes are distributed in a polar arrangement.

Yeah, I was already thinking about generating a Hex mesh but this caused several problems. One of those problems was the fact that only the plate holes are shown as sweepable bodies. The lower tube wasn't shown. Has this something to do with advanced sizing functions and the mesh being conformal? Multi Zone meshing failed when I tried to apply it to certain bodies.

I tried a First order Upwind simulation with the automatic mesh but the continuity residual reached a plateau around 5-9*10^-3.

[FJSJ]

Hi MrCapling,

If you could share it It would be nice to have a look at the real geometry

[MrCaplin]

Hi FJSJ,

I iploaded the model as an .agdb-file. If you prefer a different format just let me know.
This is the simplified version of our model which I'm trying to mesh. The inlet region was simplified and the number of holes in the plate was reduced.

https://mega.nz/#fm/P0s2mSJL

Note that I had to "build" the interior of the geometry because fill-operations didn't work. Some interior parts were modelled by using boolean operations. I guess I have to work with virtual topologies regarding those bodies.

Thank you for your help!

[MrCaplin]

Ok here's an update:

In the first few pictures you can see how the flow reactor ist designed and how it is supposed to work. It's mostly carrier gas (N2) and the two inner inlets are for the N2-H2O/N2-H2SO4 mixtures.
Mischvolumen = mixing volume
Lochplatte = perforated plate

As it was recommend I tried using a hex (sweep) mesh on the geometry. Even though most shapes are cylindrical not every body was sweepable. Is it because of the geometry being a single multibody and hence forcing a patch conformal mesh onto the bodies?

I tried slicing up bodies in order to create more sweepable bodies. Is this recommended?
The problem is that the bodies that use an automatic tetra mesh increase the skewness above 0.9.

You can find the geometry here:
https://www.dropbox.com/s/smb1cr5eew...rmal.agdb?dl=0

Any kind of help is really appreciated!
Thank you in advance!

Best regards!

[MrCaplin]

Hi,

I've got another quick question:

Is it actually possible to run a simulation with a geometry like this in 2D?
Or is it not possible due to the positioning of the holes etc.?

Thank you in advance!