How to plot residuals in Windows?

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

Thank you for your reply, you really helped me a lot.

I would like to know whether the reference time step is local time step? In fact I do not understand what CFL is, I just change the reference time step, the rest are the default settings, because in general the default setting is enough.

This is my mesh quality information, I think only a very small part of the cell quality is not good, but most of the cell is acceptable, I do not know if that small part will have a big impact:
mesh.PNG
I have calculated an unsteady flow, but I use the 0.000005 reference time step, because it is unsteady flow, so I use the compressible flow, density, viscosity and thermal conductivity are changing with the temperature , And I used the PISO Velocity-Pressure algorithm, calculated 25,000 times.

But its convergence results are still unsatisfactory, and the monitoring point of the speed fluctuations, but the monitoring point of density, temperature and pressure is relatively stable.

And I do not quite understand if I use an unsteady flow, how can i determine when it has reached a steady state and get the results I want? How long does it take to calculate it? Because it is changing at all times, so is the treatment of the unsteady flow the same as the steady flow?

Best Regards

XING Jian
Attachments
listing.zip
(654.16 KiB) Downloaded 185 times
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

Based on the previous calculations, I found the thickness of the boundary layer which is above the cardboard, and then I re-adjusted the monitoring points, all of them are at the nozzle and above the boundary layer a little bit. I recalculated once with the 0.0001 reference time step, iterating over 16,000 times. But the convergence result is the same as before, the residuals are very small, but the derive is very large, my listing is in attachment.

But the good news is that the pressure, the temperature and the velocity of those monitoring points are stable, only a small fluctuation.

The stability of pressure and temperature are very good:
Pressure.PNG
Temp.PNG
Best Regards

XING Jian
Attachments
listing.zip
(372.92 KiB) Downloaded 196 times
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

The velocity of X, Z direction are very stable, but the stability of Y direction is not very good nor bad:
X.PNG
Z.PNG
Y.PNG
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Omega's stability is also very good:
omega.PNG
But the stability of K is poor:
k.PNG
So I can think it is convergence?

Best Regards

XING Jian
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

Sorry it's me again :oops:

I have a problem with heat exchange, when I processed the results, I found that the model I had just calculated did not have heat exchange,but I have set the Exchange coefficient type and used the exchange coefficient of 50:
non Exchange T.PNG
And this is a model that I calculated before, is the model I used to test the Periodic Boundary:
exchange T.PNG
But I used the same parameter setting, why did the first one have no heat exchange at the cardboard, and the second had heat exchange at the cardboard? These two are both using the heat exchange coefficient of 50, k-w SST, and reference time step is 0.0001, and calculated 16000 times.

Thank you very much

Best Regards

XING Jian
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: How to plot residuals in Windows?

Post by Yvan Fournier »

Hello

This is strange. Could you post the corresponding data setup (I don't need the mesh, just the xml and user subroutines if you use them).

Regards,

Yvan
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

You said which one? The model I had just calculated and didn't have heat exchange? I don't use user subroutines, so just file xml? I will upload it Monday.

Can it be considered as convergence, according to my monitoring points?

Thank you very much

Best Regards

XING Jian
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: How to plot residuals in Windows?

Post by Yvan Fournier »

Hello,

Please post both setup files, so I can try to see why the heat exchange behavior is different.

To monitor for convergence, I recommend more probes. You frobably have a fluctuating flow, which never quite stabilizes but stays in the same area. In this case it may be useful to define time averages for postprocessing, and use those.

Regards,

Yvan
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

Thank you for your response, I will upload them Monday, thank you.

Could you remind me how to define time averages for postprocessing in Salome, I will be grateful.

Thank you very much

Best Regards

XING Jian
xingjian
Posts: 55
Joined: Thu Apr 06, 2017 12:34 pm

Re: How to plot residuals in Windows?

Post by xingjian »

Hello Yvan,

Here are the XML files, most of their settings are the same, only a small part is different. And the setting of the Periodic Boundary of my test model is wrong because it is a test. Another point, the biggest difference is that the mesh of the test model is very rough, and I do not have a boundary layer on the test model, but I have done boundary layers on all boundaries for the formal model.

I used the same settings(time step:0.0001, upwind...) to calculate the unsteady flow, and the result is worse than the steady flow.

And also I have tested the V2F turbulence model, because I thought that 7000 Re is a low Re, so I tried V2F, the result is worse than k-w.

Because I got a relatively stable result with a first-order upwind scheme last week, and on that basis I continued to restart 32,000 times(The first 16,000 times have been calculated with first-order upwind scheme) with second-order solu scheme, but the result is particularly poor, so I can only make it relatively stable, but I can not guarantee the accuracy.

So now, I decide to recalculate it with k-w model, steady flow, first-order upwind scheme in version 5.0.

Best regards

XING Jian
Attachments
Test(have exchange).xml
(11.59 KiB) Downloaded 270 times
Steady_SIMPLEC_0.0001(non_exchange).xml
(12.86 KiB) Downloaded 273 times
Post Reply