Dear all,
I have performed the tutorial "Turbulence simulation in a mixing tee" available in the documentation folder of the Code_Saturne site using the 64 bit Window 5.0.9 version of CS. The velocitiy calculation is OK, I get a very comparable result (see annex "Velocity.png"). Unfortunately, the temperature calculation is wrong (see annex "Temperature.png"). This result is more related to a stationary heat conduction problem with no flow, rather than a convective thermal flow problem. I have performed the same tutorial set-up and calculation using the Linux-version of CS 5.0.9 and the results are all OK. I then used the previouw Windows version of CS (4.0.8) to run the same tutorial and the results are also OK. So, only the Windows version of CS 5.0.9 is not giving the expected results. Does anyone else also experienced this problem? If not, can somebody help me figuring out why the temperature calculation is wrong using the latest Windows version of CS?
I thank you very much for your kind help!
Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
- Attachments
-
- Velocity plot in the axial plane of the model
-
- Temperature plot in the axial plane of the model
-
- RANS.xml
- Code_Saturne set-up file of the simulation model
- (10.53 KiB) Downloaded 174 times
Re: Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
Hello.
Sorry for offtopic, but I would not recommend to use Windows as the operating system for the free CFD setup (Saturne ParaView, OpenFoam, Salome). It's usual thing that Windows versions are less reliable or new versions are not available. Nowadays, it's not a problem to have different OSes for different needs. The default Saturne OS, as I know, is Ubuntu (I use Kubuntu and CentOS). If you need the text editor etc, OpenOffice/LibreOffice is there and absolutely free, you can open it's files in MS office then to continue with your report. Yes, you will have some playing with forums and console commands, but the result is usually well.
Regarding this Windows version. Try an absolutely new case, use first order (upwind) scheme for velocity, steady flow, adaptive timestep with target Fourier and Courant numbers of around 10. Set the velocity at inlets to 5...10 m/s. Run it, let it to converge (let it run for 100...300 iterations). Then check the temperature field. That on you picture is absolutely wrong, it must not be so. If you get the same problem it's likely that there is a bug in this particular version for Windows. It's likely that Ivan will help you to cope with it but it may be not simpler than to create a Linux setup and be able to run any version including newest. And there may be other issues. I recommend to switch to Linux (on separate SSD or hard drive).
Sorry for offtopic, but I would not recommend to use Windows as the operating system for the free CFD setup (Saturne ParaView, OpenFoam, Salome). It's usual thing that Windows versions are less reliable or new versions are not available. Nowadays, it's not a problem to have different OSes for different needs. The default Saturne OS, as I know, is Ubuntu (I use Kubuntu and CentOS). If you need the text editor etc, OpenOffice/LibreOffice is there and absolutely free, you can open it's files in MS office then to continue with your report. Yes, you will have some playing with forums and console commands, but the result is usually well.
Regarding this Windows version. Try an absolutely new case, use first order (upwind) scheme for velocity, steady flow, adaptive timestep with target Fourier and Courant numbers of around 10. Set the velocity at inlets to 5...10 m/s. Run it, let it to converge (let it run for 100...300 iterations). Then check the temperature field. That on you picture is absolutely wrong, it must not be so. If you get the same problem it's likely that there is a bug in this particular version for Windows. It's likely that Ivan will help you to cope with it but it may be not simpler than to create a Linux setup and be able to run any version including newest. And there may be other issues. I recommend to switch to Linux (on separate SSD or hard drive).
Re: Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
Dear Antech,
Many thanks for your reply. I had also tried the other settings you suggest, but the calculated temperature profile was also wrong. So I guess there is really a bug in the latest Windows version. Regarding Linux, I know that this OS is much better for performing these calculations using EDF software, but the research lab I am working at is relying on Windows OS. So I hope the developers working on the Windows port will solve this bug in the near future.
Again many thanks for your help and advise.
Many thanks for your reply. I had also tried the other settings you suggest, but the calculated temperature profile was also wrong. So I guess there is really a bug in the latest Windows version. Regarding Linux, I know that this OS is much better for performing these calculations using EDF software, but the research lab I am working at is relying on Windows OS. So I hope the developers working on the Windows port will solve this bug in the near future.
Again many thanks for your help and advise.
-
Yvan Fournier
- Posts: 4077
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
Hello,
Unfortunately, to make things clear, check this post: viewtopic.php?f=4&t=2034
Basically, there are no developers working on the Windows version to my knowledge. I tried updating it a bit some time ago, but as I explain in the post, this would require much more manpower than what is available. Which is why we do not build newer versions (5.2 and 5.3) for Windows... Especially, the current approach for our Windows build makes things quite difficult on the debugging side.
On windows 10, you best solution is to use the Linux subsystem for Windows, and install the code in a manner similar to that on Linux. For the GUI, an X server is needed but seems to work (I may try building just the PyQt GUI directly on Windows ans use the Linux subsystem for Windows or a Docker container for the non-graphical parts, which could be a nice alternative, but do not expect anything more unless someone in the open source community decides to restart work on the Windows version).
Best regards,
Yvan
Unfortunately, to make things clear, check this post: viewtopic.php?f=4&t=2034
Basically, there are no developers working on the Windows version to my knowledge. I tried updating it a bit some time ago, but as I explain in the post, this would require much more manpower than what is available. Which is why we do not build newer versions (5.2 and 5.3) for Windows... Especially, the current approach for our Windows build makes things quite difficult on the debugging side.
On windows 10, you best solution is to use the Linux subsystem for Windows, and install the code in a manner similar to that on Linux. For the GUI, an X server is needed but seems to work (I may try building just the PyQt GUI directly on Windows ans use the Linux subsystem for Windows or a Docker container for the non-graphical parts, which could be a nice alternative, but do not expect anything more unless someone in the open source community decides to restart work on the Windows version).
Best regards,
Yvan
Re: Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
stichko
Hello. I supposed that something like this may be the case... I use Saturne at home on Linux workstation and on Linux laptop at work, while company's machines run Windows (except the small Linux cluster). If you don't need large meshes you can use your personal laptop (install Linux as second OS) or virtual machine with Linux in Windows on company's machine. You will only need the calculation stage in Linux if you have ready meshes made with some Windows mesher (I use Ansys mesh export to CGNS). If you need relatively large meshes, I don't see the direct solution. If it's important for your company, you can talk to your boss for installing Linux as the second OS on a workstation.
Hello. I supposed that something like this may be the case... I use Saturne at home on Linux workstation and on Linux laptop at work, while company's machines run Windows (except the small Linux cluster). If you don't need large meshes you can use your personal laptop (install Linux as second OS) or virtual machine with Linux in Windows on company's machine. You will only need the calculation stage in Linux if you have ready meshes made with some Windows mesher (I use Ansys mesh export to CGNS). If you need relatively large meshes, I don't see the direct solution. If it's important for your company, you can talk to your boss for installing Linux as the second OS on a workstation.
Re: Temperature calculation problem in tutorial "Turbulence simulation in a mixing tee" using CS 5.0.9 Win 64bit
Dear Ivan, dear Antech,
Again many thanks for your help. I have already spoken with my boss and I will install a second HD in my computer with Linux installed on it. This will allow me to use the Linux/Unix version of Code_Saturne (much better!).
Kind regards,
Serge
Again many thanks for your help. I have already spoken with my boss and I will install a second HD in my computer with Linux installed on it. This will allow me to use the Linux/Unix version of Code_Saturne (much better!).
Kind regards,
Serge