[jeroendemto]

Hey everyone!

I've got a vasculair system that is symmetrical, however the results are not. CFD-DEM is used to model the flow of particles through the vasculair system. The flow seems symmetrical, however the DEM particle distribution is not. More than 10000 particles per second are used, so the distribution should be symmetrical. I think the problem arises due to the mesh not being symmetrical. I thought I could make my mesh symmetrical by mirroring the mesh on the symmetry axis. But, I do not see any way to do so.

How can I make my mesh symmetrical? Added a picture of the mesh.

[me3840]

To make a mesh perfectly symmetric, you need to mesh half of the geometry, so cut your part in half.

Once meshed, duplicate the region, mirror it, and then combine the two regions back together.

[jeroendemto]

Hey me3840,

I have cut in half my geometry and meshed it. Duplicated the region, however I'm not able to mirror it. This is due to the boundary conditions imposed on the in and outlet. If I don't put any boundary conditions on the in and outlet it works. How can I impose BC while still be able to mirror the region?

With kind regards,
Jeroen

[JBeilke]

Do you consider people to be 100% symmetric? I would be very suspicious when I see a symmetric solution.

[jeroendemto]

Originally Posted by JBeilke

Do you consider people to be 100% symmetric? I would be very suspicious when I see a symmetric solution.

Hello Joern,

Al through this is a vasculair system, it is a simplified one. The geometry and boundary conditions are completely symmetric, so this would mean that the solution is also symmetric.

Kind regards,
Jeroen

[JBeilke]

Not for a turbulent flow !!

[jeroendemto]

Originally Posted by JBeilke

Not for a turbulent flow !!

Hey Joern,

Sorry I should have said that I assumed it to be laminar because the flow speed is around 0.3 m/s, the width is ~4mm and the liquid is blood. (non newtonian generalized Carreau-Yasuda Fluid) If there is a turbulent blood flow inside someones vasculair system there is something wrong.

Kind regards,
Jeroen

[JBeilke]

When you expect a symmetric solution you can calculate just a half and mirror the results. (Scene-Displayer-Transform)



Anyway, I dont see a reason why copy/reflect and merge of the mesh should not work.

[jeroendemto]

Originally Posted by JBeilke

When you expect a symmetric solution you can calculate just a half and mirror the results. (Scene-Displayer-Transform)



Anyway, I dont see a reason why copy/reflect and merge of the mesh should not work.

I thought about the same solution, however the next step in this experiment is to change the pressure of one outlet. Changing the outlet on one side will break the symmetry, which means I can not mirror the solution.

The copying and reflecting part is where it goes wrong. This is my procedure: (1)I split my model so I have 1/2 of the model --> (2)I split by patch to name my in and outlets --> (3)assign parts to region and create a boundary for each part surface --> (4)I copy my region. Now I need to transform the copied region, however the transform option is grey and unclickable. If I don't do step 2 but go from 1 to 3 I'm able to transform and combine the regions and thus the mesh. However step 2 is necessary to be able to assign boundary conditions on my in and outlets.

I hope this helps understanding the problem, if not feel free to ask.

Kind regards,
Jeroen Reus

[me3840]

Mirroring and combining the region should not be influenced by the boundary types, what problem were you getting when attempting that procedure?

[jeroendemto]

The problem with the procedure without naming my surfaces is that I'm unable to name my surfaces after mirroring of the mirrored region. Because only the region and mesh is mirrored while (I think) only part surfaces can be named and thus imposed boundary conditions.

So I can only name the surfaces of the original part and not on the mirrored region.

The problem with the procedure where I do name my surfaces before making the region is that I'm unable to mirror the region. Because, the option is grey and unclickable, I don't know why this is the case.

[me3840]

I'm not quite sure I understand. You can rename boundaries in a region independently of the part surface names, the two have nothing to do with each other. Once the second region is generated you should be able to rename the boundaries easily.

If the transform/mirror option in the region is greyed out that probably means you haven't generated the mesh yet. Nothing to transform if there's nothing meshed.

[jeroendemto]

Originally Posted by me3840

I'm not quite sure I understand. You can rename boundaries in a region independently of the part surface names, the two have nothing to do with each other. Once the second region is generated you should be able to rename the boundaries easily.

If the transform/mirror option in the region is greyed out that probably means you haven't generated the mesh yet. Nothing to transform if there's nothing meshed.

Ah yes I forgot to mesh in the second procedure before trying to mirror, that was quite stupid. Thank you for your patience, it is working now.