Turbomachinery Module

[AndrewH]

Hello,

I have been trying to use the turbomachinery module in v3.3.2 on a rotating cylinder at Re=200 as a test case, but my simulation stops in middle of the time step without any error being outputted. The error seems to occur always around the 1700th time iteration. I noticed in the changelog for v3.3.3 that it fixes a memory leak for the turbomachinery module, which I assume is this problem. I tried to install v3.3.3 on both a local Ubuntu machine and on a Cray HPC, but I received a bft memory error at the start of my calculations. I tried several different setups and geometries, but I received the same error. I also used the same installation process I used for previous versions. I attached my listing file if it proves to be helpful.

From there, I jumped to v4.0-r7183. I tried to use manually setup the turbomachinery module in the SRC folder, but I received an error when cs_turbomachinery_add_rotor function was called. However, if I defined my turbomachinery module in the xml file instead, everything works fine. I also attached my listing and turbomachinery file for this case version too.

Please let me if these are known bugs or if it is mistake on my part.

Thank you,
Andrew

[Yvan Fournier]

Hello,

Could you also post your xml file ? The backtrace seems to indicate the issue could be due to some interaction between the GUI and SRC settings (I did not see any obvious mistake in your source file).

Regards,

Yvan

[AndrewH]

Hello,

I'm having a hard time logging back into the HPC server to retrieve my xml file. However, I quickly re-ran my case locally and I think I might have figured out my problem. It seems that the turbomachinery module line in the xml has to be present, but set equal to OFF if the turbomachinery module is specified in the SRC folder. If I remember correctly, I think I used a xml file that I generated from v3.3.2. Before I tried running a rotating cylinder, I ran a stationary cylinder in v4.0 to check my installation and I used a xml file that I generated from v3.3.2. I probably came to bad conclusion that the xml from v3.3.2 was competely compatible with v4.0. However, I also noticed that if I turned on the turbomachinery module in the GUI in v4.0 and then specified my turbomachinery module in the SRC folder, I received a similar error.

Thank you,
Andrew

[Yvan Fournier]

Hello,

The XML file should be (mostly) updated from 3.2 if you open it with the version 4.0-beta GUI, but it is not "directly" compatible.

I'll try to reproduce the issue and make the reader more robust.

Regards,

Yvan

[AndrewH]

Hello,

Attached is the listing file that I was originally using if it proves to be helpful.

On a somewhat related topic, is periodic bc compatible with the turbomachinery module? I originally attempted to use periodic bc with my case, however I received the attached error. Is this a limitation of the code? The periodic bc works if I use a simple mesh joining instead of the turbomachinery module.

Thank you,
Andrew

[Yvan Fournier]

Hello,

Periodicity should be compatible with turbomachinery if there is no intersection between periodic boundaries and rotor/stator boundaries.

It is not compatible with the turbomachinery module if it includes the rotor/stator boundaries (such as meshing only part of the rotor and stator and then trying to use periodicity).
This is because to match rotor / stator boundaries in this case, we would need to mix joining (where mesh faces match in the current position) and periodicity (where they do not match in the current position, but would map if some of the faces were transformed by a given number of times the periodicity transformation). This is quite tricky, and is not handled (it is not considered a priority, as in most real turbomachinery applications, rotors and stators do not have the same periodicity, implying that even if the geometry is periodic by subsets, the flow near the boundary can not be so).

Regards,

Yvan

[vgm567]

Periodicity should be compatible with turbomachinery if there is no intersection between periodic boundaries and rotor/stator boundaries.

Hi,

I too have difficulty regarding a similar situation. For example I want to simulate a Taylor-Couette type setup (two concentric cylinders with inner-rotating/outer-fixed with fluid inside the annular gap). I normally have the circular side faces (on either end of the rotation axis) as inflow and outflow. However I want to place periodic BC on these faces also. I am also faced with the same error as the previous poster. Please could you advise me how to deal with this.

Many thanks

Vipin

[Yvan Fournier]

Hello,

For true (implicit) periodic BCs, a specific development would be required.

If you only need a pseudo-periodic BC, for example reinjecting the output profile at the inlet to simulate an infinite profile, you may use the new "mapped inlet" functionnality of Code_Saturne-4.0 to handle this (the previous "auto-inlet profile" from version 3.0 migh do also).

So I recommend checking those user examples (either sources or Doxygen documentation) to see if they might be applicable to your case.

Regards,

Yvan