PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Yvan Fournier
Posts: 4069
Joined: Mon Feb 20, 2012 3:25 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Yvan Fournier »

Hello,

Actually, this is due to another bug, which has been fixed. In v6.0, the default clipping for the total energy was set to 0 if you visited the numerical parameters page in the GUI while it should remain at -1e12.

So to fix the issue, you simply need to check the per variable numerical parameters page in the GUI, and set the minimum clip value for total energy back from 0 to -1e12.

Automating the fix would be a bit risky (and ignoring an incorrect value
would be easy but trick in case the user really wants to force this), so has not been done yet, since the issue is non-blocking (once you know the workaround).

I'll check for the mass flow.

Best regards,

Yvan
Yvan Fournier
Posts: 4069
Joined: Mon Feb 20, 2012 3:25 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Yvan Fournier »

Hello,

I checked for the mass flow, and here is my conclusion:

If you prescribe the boundary density as one of the two given inlet variables, this should work.

If you define a mass flow and compute the density based on other inlet variables (such as pressure and temperature), we have 2 cases:

- density from pressure and temperature, currently computed later (currently too late)
- density from pressure and total energy, whcih has a circular dependency, as total energy includes velocity, so an iterative process would be needed here.

I'll try handling the case for pressure and temperature (for v7 and above only), and keep you informed.

Best regards,

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

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Christopher »

Hello,

thank you all for your fantastic help. It is highly appreciated. After tpa's adivice to use the older version V5, I ran the simulation and I am satisfied with its results.

Besides what is running in a specific version, I wanted to add, that the results from 5.0 (160 m/s, 1,24e5 Pa) represents the real physical system much more precise than the V7 with 400 m/s. The is way to high, almost supersonic at that temperature. I did some analytic (isentropic) calculations on our maschine. The velocity is about 200 m/s, the temperature 240°C at the outlet and the pressure at the inlet is about 1,24e5 Pa, the temperature 260°C. I will see, if I can provide some messured data directly from the maschine.

Best regards,
Christopher
Yvan Fournier
Posts: 4069
Joined: Mon Feb 20, 2012 3:25 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Yvan Fournier »

Hello,

Thank you, that would be interesting. I just added a change in the master branch (v7.1) to handle the "imposed mas flow" case, but did not merge it to other versions, as it leads to even higher velocities.

It is possible that this type of inlet is better adapted to supersonic cases. Did you try using the subsonic inlet variants ?
There are also some compressible boundary condition variants not available from the graphical user interface. They could perhaps be added, but I am not familiar with the various options and their validity range.

So any validation type data, even if limited, can be very useful for this (I repeat myself, but the compressible model is not our main target, and EDF applications requiring this might need/test only a subset of the available options, so the user community is welcome to help us improve it).

Best regards,

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

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Christopher »

Dear Yvan,

I tried to install the master branch (v7.1) and re-run the simulation that already worked in v5. However, I am stuck within the installation from git. By running the automatic installation script "install_saturne.py" I get the message "FileNotFoundError: [Errno 2] No such file or directory: '/home/code_saturne/code_saturne-v7.1/code_saturne-master/configure'".

I tried to follow the instructions here: "https://github.com/code-saturne/code_sa ... INSTALL.md" below the source tree obtained through a source code repository, but till now not sure how to use it.

Maybe anyone could give me a simple advice?

I did some measurements with a pitot tube and a thermocouple (velocity and temperature under the nozzle) on our machine, but with lower inlet temperatures. So I would like to re-run the simulation with updated inlet boundaries the see, how good the numerical results will match the experimental data. Especially with v7.1.

So if I get the installation and then the simulation finished, I will let you know.

Kind regards,
Christopher
Yvan Fournier
Posts: 4069
Joined: Mon Feb 20, 2012 3:25 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Yvan Fournier »

Hello,

Compared to .tar.gz releases, if you use Git, you only need to first (just after downloading /updating from Git) :

"cd" to the downloaded directory
run "./sbin/bootstrap"

The rest is the same.

Bootstrap may fail if you are missing some tools. This page can help in that xase (https://github.com/code-saturne/code_sa ... n-packages).

In any case, if a "configure" and "Makefile.in" file is created in the top source directory, you are probably good to go for the regular install steps after that. If "configure" is missing, bootstrap probably failed (in this case could you post your error messages) ?

Best regards,

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

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Christopher »

Hi Yvan,

I made it. CS 7.1 is running. However, I didn't see anything in the Job Run window after starting the calculation??? I opened up the old .xml from CS 5.x and tried to run it in CS 7.1. There is no error message. The cursor just shows the little loading circle and the Job Run remains empty.

Have you had a problem like this before?

Kind regards,
Christopher
Attachments
CS_Job_Run.png
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Christopher »

Hi Yvan,

I guess it was because I used some too old files. I created a new case and now it begins to run. Is it correct that I cannot use more than 16 cores for calculation? Or is it possible to increase this number?

If I use 16 cores the calculation stops at the "Starting calculation" stage (see image attached). It seems that some libhdf5 libraries are missing. Do I have to install them separately? When I used the install_saturne.py I switched the add. packages ,including hdf5, to yes.

Kind regards,
Christopher
Attachments
Job_Run.png
Christopher
Posts: 36
Joined: Wed Feb 17, 2021 2:22 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Christopher »

Hi Yvan,

the simulation runs. But a little slow for a 16 core AMD Ryzen 9 processor. I will keep you informed.

Kind regards,
Christopher
Yvan Fournier
Posts: 4069
Joined: Mon Feb 20, 2012 3:25 pm

Re: PROBLEM IN THE BOUNDARY CONDITIONS USING COMPRESSABLE MODEL

Post by Yvan Fournier »

Hello,

There is a huge variability in performance based on processor types, but code_saturne is mostly memory-bound, so data-transfer to and from memory and cache are at least as important as processor speed. On any processors (and I believe this is the case for at least/most some Ryzen models), you can saturate the memory bandwith even without using all the cores. This will also depend on your mesh size (with a smaller mesh you might fit better in cache).

Do you have performance comparisons with other machines for the same case ?

Best regards,

Yvan
Post Reply