Computation of a Simplified Centrifugal Blood Pump

[SRIB]

Hello Yvan,

i refined my meshes and now both meshes have 1,500,000 volume elements. Until now i was not able to get a rotation. I don know whats wrong. One question: which turbulence model would you suggest for my case. I read that k-epsilon is worst in case of roattion. k-omega would be slightly better. What do you think?

Regards,

Sebastian

[Yvan Fournier]

Hello,

If you do not obtain a rotation, could you post a "listing" file and your user subroutines and xml file ?

As for the turbulence model, I am not sure. I know an option (irccor) was added in 3.0 for turbulence terms in rotation mode, for both k-epsilon and k-omega, but I do not know more.

Regards,

Yvan

[SRIB]

Hello Yvan,

attached you can find the listing file, XML file and the subroutine. I am just using cs_turbomachinery.c and your cs_join_util.c whereas i did not changed anything in cs_join_util.c.

In terms of the irccor-turbulence model. Do you know wheather thecs_turbomachinery set this model or if i have to use furhter subroutines?

Thank you.

Regards,

Sebastian

[Yvan Fournier]

Hello,

I'll need to check why the turbomachinery rotation and joining do not seem to be called (I see no obvious error there).

But you still need to change your mesh: you need a different group of cells for the rotor ans stator, so only the rotor will turn.

As for irccor, I am not sure, but its value should appear somewhere in the listing.

Regards,

Yvan

[SRIB]

Hello,

ok. Maybe i did not understand it.The Groups are as following:

Rotor:

Groups of Faces:
boundary group for the joining
wall (the inner rotor wall)

Groups of Volumes:
all tetrahedrons of the rotor mesh

Stator:

Groups of Faces:
Inlet
Outlet
boundary (for the joining)

Do i need in addition a volume group for the stator? Are there any groups listed above that i do not need?

Regards,

Sebastian

[Yvan Fournier]

Hello,

No, you do not need a group for the stator.

Sorry, in the listing, I was mislead by the generic "cells" group name, but you have abour 95000 cells in that group, and 1629817 total, so things seem OK here.

It is strange that you obtain no rotation. I'll try to take a look at this within a day or two.

Regards,

Yvan

[SRIB]

Hello,

that would be very nice. Thank you very much!

Regards,

Sebastian

[Yvan Fournier]

Hello,

Running with your Case and renaming groups/adding volume groups to your previous meshes to match this setup, I obtain rotation (in my "listing", I have the part about joining that you are missing).

What system are you running on ? We have (very rarely) encountered cases where user subroutines did not seem to be taken into account correctly. If this is the case, compiling a static build (--disable-shared configure option) might be a workaround.

Also, for rotor/stator, performance will be bad with k-omega as computation of the wall distance to each cell need to be recomputed at each time step.

Regards,

Yvan

[SRIB]

Hello,

ok. In principle this is good. This means no mistakes in the subroutine or Meshes. I am running on Ubuntu 12.04 LTS 64-bit. What do i have to do for such a compile?

PS: i installed 3.2.1 next to 3.0.3. Could this be a problem? On what system are you running? I could maybe also change my one. Most important is that my rotor rotates as soon as possible.

Thank you and regards,

Sebastian

[Yvan Fournier]

Hello,

I ran a test on Ubuntu 13.10.

Installing versions side by side should work fine.

The only issues we have had were with "packaged" versions of the code, where the reference versions of the user subroutines seemed to gain priority over the user-modified versions.

To check, add something like :

Code: Select all

bft_printf("calling %s\n", __func__);

To your user functions, and run the case again, to check if the matching message appears in your "listing" file. If they do not appear, de-install Code_Saturne (completely removing your /home/caelinux/opt/code_saturne-3.2.1/arch/Linux_x86_64 directory), and re-install it adding "--disable-shared" to your configure command.

Regards,

Yvan