LES with conjugate heat transfer

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Post Reply
xiaoxue123
Posts: 29
Joined: Fri Mar 02, 2018 11:50 am

LES with conjugate heat transfer

Post by xiaoxue123 »

Hello,

I'm running a LES with conjugate heat transfer. The fluid-solid coupling is defined in cs_user_parameters.c. The version I'm using is 6.0.0. The LES is restarted from a RANS (k-omega SST) simulation using the same mesh.

I found the temperture filed in the LES is "silent". In the run_solver.log, N_iter for temperature is 1 at the first time step, while it becomes 0 for the time steps after the first one. It seems like the temperature field is not changing at all.

The settings and the output files are attached as follows. Can anyone help me out?

Best regards,
Xiaoxue
Attachments
cylinder.xml
(17.63 KiB) Downloaded 145 times
cs_user_physical_properties.f90
(15.45 KiB) Downloaded 146 times
cs_user_parameters.c
(6.86 KiB) Downloaded 155 times
xiaoxue123
Posts: 29
Joined: Fri Mar 02, 2018 11:50 am

Re: LES with conjugate heat transfer

Post by xiaoxue123 »

Hello, the log files are attached here.

Xiaoxue
Attachments
setup.log
(26.77 KiB) Downloaded 148 times
run_solver.log
(68.8 KiB) Downloaded 146 times
performance.log
(21 KiB) Downloaded 144 times
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: LES with conjugate heat transfer

Post by Yvan Fournier »

Hello,

Does the same conjugate heat transfer setup work OK with the initial k-omega setup, or did you only test it in LES ? That would help know where to look.

Also, I seem to recall a similar issue recently on this forum, that I did not look into yet (this might indeed be a bug). Just in case, did you check recent posts ?

Also, could you post a visualization showing the temperature field (and the fluid/solid zones).

Best regards,

Yvan
Luciano Garelli
Posts: 280
Joined: Fri Dec 04, 2015 1:42 pm

Re: LES with conjugate heat transfer

Post by Luciano Garelli »

Hello,

I did a test in a small case using CS 6.0.4 and the temperature field in celsius. I used LES dyn model restarting from K-omega SST and I don't get this behaviour in the temperature iterations. Maybe it was a bug that was fixed in the subsequent versions.


Regards,
Luciano
run_solver.log
(534.85 KiB) Downloaded 171 times
xiaoxue123
Posts: 29
Joined: Fri Mar 02, 2018 11:50 am

Re: LES with conjugate heat transfer

Post by xiaoxue123 »

Hello Yvan,

Sorry about the late reply.

I've tried RANS simulations, that is OK. I've attached a picutre of the temperature field (LES restarting from a RANS, but there is no iteration for temperature in LES). The geometry is axisymmetric, and the picture shows a 2D slice of the domain. The domain consists of an inlet pipe, a cylinder and an outlet pipe. The inlet is at the bottom of the domain, the outlet is at the top of it. The solid domain is an inner plate in the cylinder. Cold fluid flows from the bottom and get mixed with the hot fluid in the cylinder, then flows upwards to the inlet.

I've checked recent posts. Maybe the similar issue you mentioned was mine. viewtopic.php?f=2&t=2699&p=14692&hilit= ... les#p14692
I've made some modifications of the subroutines for internal coupling following your advice last time, but the issue of the temperature field still exist.

Best regards,
Xiaoxue
Attachments
temperature field
temperature field
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: LES with conjugate heat transfer

Post by Yvan Fournier »

Hello,

Did you upgrade from version 6.0.0 to the latest 6.0.z version (6.0.5) ?

Regards,

Yvan
xiaoxue123
Posts: 29
Joined: Fri Mar 02, 2018 11:50 am

Re: LES with conjugate heat transfer

Post by xiaoxue123 »

Hello,

I've installed a version 6.0.5. But it appears the same that the iteration for temperature is 0. Maybe there was something wrong with the settings in my simulation?

Best regards,
Xiaoxue
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: LES with conjugate heat transfer

Post by Yvan Fournier »

Hello,

Looking at the previous log, I wonder whether this may be due to solver convergence criteria. Given a small time step, if the residue normalization is high enough, solution steps you may be "under" the minimum error required to call the solver, leading to only the explicit part of increments being computed.

To test this, you may reduce the convergence criteria for temperature (for example set it to 1e-10) and check if this makes a difference. If this is the explanation, then you can go back to the initial residue, and the computation should be ok. An option might be to force at least 1 iteration in the sources, or add an option for this, but depending on the case, I am not sure it should be done or not.

Best regards,

Yvan
Post Reply