[HuifengGong]

Hi all,

I was using a quarter sector geometry to do my simulation, and was encountered such errors as described in the title (not the exact error words).
I used periodic boundary condition for the cutting planes, and because the cutting planes belong to different regions, I set these two periodic boundaries as ‘dependent’ for region association.
Can anybody give me some comments for my case setup? Thank you!

[tmburton]

Huifeng,

You probably don't want to use Dependent region assignment in this case. If you take a look at any of our sector example cases in Studio, you'll see that the periodic boundaries are assigned to the same region as the piston and the head (or if you think about a quarter-pipe simulation, the same region as the inlet and the outlet).

Best regards,

Tristan

[HuifengGong]

Originally Posted by tmburton

Huifeng,

You probably don't want to use Dependent region assignment in this case. If you take a look at any of our sector example cases in Studio, you'll see that the periodic boundaries are assigned to the same region as the piston and the head (or if you think about a quarter-pipe simulation, the same region as the inlet and the outlet).

Best regards,

Tristan

Hi Tristan, thank you for your reply.

Actually, I tried to get some ideas from the example cases, but all the sector example cases have only one region for the whole domain, which are different than my case. I divide the domain into several different regions along the flow direction, as the attached sketch shows (not the real case geometry), and the periodic boundaries cover all the regions.

After getting your reply, I assigned the periodic boundaries to the same region as the inlet boundary, and the simulation can continue but the errors still exist:
"error, the separation between region 0 and 1 is not in a continuous loop
not enough edges intersecting at the vertex located at 3.195493e-004 -3.195479e-004 -6.619079e-004"
There are many similar error messages regarding to different regions and vertex locations.

Besides, the events about region connection also confused me. After I assigned the periodic boundaries to region0, there were connection events between regions that were not physically connected (ie region 0 and region 3 as shown in attached sketch), and even if I closed these connections, the error still existed.

That is my situation now. The simulation can continue with such errors, but is it correct?

Huifeng

[tmburton]

Huifeng,

Each periodic face should be broken into four segments so that you can assign the relevant sections to the appropriate region. Looking at your diagram, you should have the following in each region:

Region 0: inlet, 2 periodic face segments, outer wall segment
Region 1: 2 periodic face segments, outer wall segment
Region 2: 2 periodic face segments, outer wall segment
Region 3: 2 periodic face segments, outer wall segment, outlet

In your events setup, you need permanent OPEN events between the following regions:

0->1
1->2
2->3

Let me know if this is unclear.

Best regards,

Tristan

[HuifengGong]

Originally Posted by tmburton

Huifeng,

Each periodic face should be broken into four segments so that you can assign the relevant sections to the appropriate region. Looking at your diagram, you should have the following in each region:

Region 0: inlet, 2 periodic face segments, outer wall segment
Region 1: 2 periodic face segments, outer wall segment
Region 2: 2 periodic face segments, outer wall segment
Region 3: 2 periodic face segments, outer wall segment, outlet

In your events setup, you need permanent OPEN events between the following regions:

0->1
1->2
2->3

Let me know if this is unclear.

Best regards,

Tristan

Hi Tristan, thank you for your reply! It helps. I used some edge to make fences, and segmented the periodic boundary into several small boundaries, and the errors disappeared. Did I do it right?
But, there are some other problems.
Firstly, there are some warnings that the areas of paired periodic boundaires do not match, but the difference is very small (at the level of 10^-7). I think this is due to the accuracy of the geometry. I refined the geometry when I exported stl files, but it didn't work. In some cases, there were no differences when measuring the areas of paired periodic boundaries in Studio, but the log file still showed a tiny difference when running the simulation. Is it possible to ignore this tiny difference? i.e. making CONVERGE less sensitive to the geometry accuracy.
Secondly, regarding to the segmenting method, is it also useful when applying moving boundaries? When the wall boundary is moving, the association between boudaries and regions might change, can this method account for this situation?

Thank you for your kind help.

[tmburton]

Huifeng,

If you want the areas of the periodic faces to match exactly, you can often delete one of them and copy/rotate the other periodic face into the same position as the deleted face, and then reconnect to nearby non-periodic triangles. We also have the make engine sector surface tool in Studio that may be useful. It is important that the triangulations on the periodic faces match. If Studio reports that they match then you should probably be ok with your solution.

Can you clarify your moving boundary question with an example?

Best regards,

Tristan

[HuifengGong]

Originally Posted by tmburton

Huifeng,

If you want the areas of the periodic faces to match exactly, you can often delete one of them and copy/rotate the other periodic face into the same position as the deleted face, and then reconnect to nearby non-periodic triangles. We also have the make engine sector surface tool in Studio that may be useful. It is important that the triangulations on the periodic faces match. If Studio reports that they match then you should probably be ok with your solution.

Can you clarify your moving boundary question with an example?

Best regards,

Tristan

Hi Tristan,

My geometry is not axisymmetric but quarterly symmetric, I think it is not possible to use the make surface tool. I used Geometry->Repair->Surface->Periodic boundaries to try to match the areas, but the log file still gave warnings:
"area of periodic boundary 6: 0.000004713593
area of periodic boundary 7: 0.000004713593
periodic boundary area diff: 2.559733917213e-010
WARNING: areas of periodic boundaries 6 and 7 do not match. This may cause mass loss."
PS. The periodic boudary ID in my case setup were 7 and 8, I don't know why they were 6 and 7 here.
The difference is very small, I don't whether this difference will affect the simulation results or not, can I just ignore this warning?

The moving boundary I mean is the one like the polyline part in my diagram. When it moves along the flow direction, how will the seperated periodic boudaries change? My confusion comes from the example case 'Pintle injector - VOF'. There are two regions in that case: the innozzle flow region and the chamber space region, and the moving pintile tip boundary is set as 'Dependent', which I think means that different parts of the tip belong to different regions when it moves. How will the code divide this boundary into two regions in a 'dependent' case? Where is the separation? And in my periodic boundary case, I 'manually' assign a fence to separate the boundary into different regions, how will this manual separation change under a moving boundary condtion? Why can't I set it as dependent and let the code automatically determine the separation?

Hope I have made my question clear, but maybe I'm too confused about this part to point it out clearly.

Thank you very much!

Huifeng