Underestimation of velocity, compressible model

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hi Yvan,

anything new from this side?

Kind regards,
Christopher
Yvan Fournier
Posts: 3647
Joined: Mon Feb 20, 2012 3:25 pm

Re: Underestimation of velocity, compressible model

Post by Yvan Fournier »

Hello,

Sorry, I had forgotten. Thanks for reminding me.

The attached user parameters sets up cv0 where needed, though I used the same value for CV that you set for Cp (just for checking). With that value, I have a negative pressure after 2 iterations, so I'll let you check and set the correct value.

If things still fail, I'll need to check with colleagues who know the compressible module better...

Best regards,

Yvan
Attachments
cs_user_parameters.c
(5.66 KiB) Downloaded 55 times
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hi Yvan,

thank you. Right now a calculation is running. After that I will give it a try.

By the way, is there a limit to the calculation steps in CS7.1? I am running a compressible hot air flow simulation on a locally very refined mesh (smallest mesh size = 0,1 mm). With a max CFL <= 0,4, the achievable time step is quite small. So it goes up to more than one million time steps to be calculated. However, the solver hangs up after about 1.500.000 steps and stops responding. The desired physical time is 0,15 s, but it reaches only 0,8 s before it hangs up.

Kind regrads,
Christopher
Yvan Fournier
Posts: 3647
Joined: Mon Feb 20, 2012 3:25 pm

Re: Underestimation of velocity, compressible model

Post by Yvan Fournier »

Hello,

No, there is no theoretical time step limit.
Or at least, the limit would be ar 2^32, so a little over one billion.

Did you try using a computation restart ? What happens if you increase the reference time step a bit. You might not lose stability immediately (though you can't use a too high CFL, you can test a bit higher).

Best regards,

Yvan
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hi Yvan,

I will try that as well and inform you about the results.

So far, I haven't tried a computation restart. Might be a good idea to calculate to 1.000.000 steps and then restart from there to continue. Unfortunatly, I can't increase the reference time step because of negativ energy encountered in a few cells. The max. t <= 1e-07.

However, the hang up could be caused by Ubuntu and not CS.

Kind ragards,
Christopher
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hello Yvan,

so far, I tried the computation restart. It works. I can go above 1.500.000 steps by splitting the caluculation in parts with checkpoints at each ends.

However, one more question.

I wanted to start the calculation with a coarse mesh for 600,000 iterations and a larger time step and then restart with a fine mesh for the rest using the "Different mesh" option in CS. However, the restart aborts with an error message. It says that the calculation cannot start with different cell counts.

Is there an assignment (mapping) option to transfer the results of the first calculation with the coarse mesh to the second with the finer mesh with more cells? Or what is the option for different meshes for?

Kind regards,
Christopher
Yvan Fournier
Posts: 3647
Joined: Mon Feb 20, 2012 3:25 pm

Re: Underestimation of velocity, compressible model

Post by Yvan Fournier »

Hello,

Yes, in v7.0, there is an option to restart on a different mesh. You have a box to check for this in the restart definition section in the GUI, and must provide a link to the original mesh there (many users get confused and invert the initial and new mesh, so if things do not work when checking the box, try inverting the 2).

Best regards,

Yvan
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hi Yvan,

alright. So to do it right, I have to replace the old (original) mesh with the new one under the mesh menu and then place a link to the old mesh under restart => different mesh, right?

Kind regards,
Christopher
Yvan Fournier
Posts: 3647
Joined: Mon Feb 20, 2012 3:25 pm

Re: Underestimation of velocity, compressible model

Post by Yvan Fournier »

Hello,

Yes, if I remember correctly, that is how it works. In any case, there are not so many possible combinations, and only one of them can work :-).

Best regards,

Yvan
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: Underestimation of velocity, compressible model

Post by Christopher »

Hi Yvan,

so far I can say that your cs_user_params works. I can't see any negative pressure. The difference between cp and cv is just the isentropic exponent k.

Next I will compare the calculation results with our experimental results. I will make two runs. One with the stiffend gas model and another with a strong mesh refinement at the nozzle outlet. It could be that a millimeter mesh size and a bit below is still too coarse to correctly capture all these physical properties. Therefore, this could be another reason for the underestimation. But for now it is hard to say.

Once the calculations are complete, I will post the result here and let you and other participants know.

I wish everyone a Merry Christmas and a Happy (and healthy) New Year.

Yours sincerely,
Christopher
Attachments
cp_cv.png
Post Reply