SimScale CAE Forum

Time step continuity error in a turbulent two-phase flow

All,

I am continuously receiving error from a two phase flow in a wye. According to the log file, I get a “time step continuity error” in my simulation run. One phase is natural gas and the other phase is brine. Could you check it and let me know how I can remove the problem?

https://www.simscale.com/workbench/?pid=4149221339654153189#tab_1-0

1 Like

Hi @YB128,

Ah I see you have closed up and simplified the geometry.

Seems like your boundary conditions and mesh are alright. Hopefully one of the @PowerUsers_CFD with more experience in Multiphase flow can help you.

Cheers!

Regards,
Barry

2 Likes

@YB128,

Looks like the there is numeric instability here, which has thrown auto time stepping off making really small timesteps leading to this error.

I would start by increasing the number of correctors (1 isn’t enough, I usually go with 15 to 20 and if it converges within that it will finish early) and also there is no region in the domain that is allocated to the other phase, this can cause problems. I would create a primitive so that you can subdomainly initialise the solution with the other fluid also. In the past this has helped me. Another thing to consider is looking at how the walls are treated in other multiphase simulations.

Looking at your mesh I think you should consider reducing the refinement and inflating boundary layer cells this would make your simulation more efficient, for a transient simulation I would really aim for a mesh of 100k for the first attempt to ensure setup is solid.

good luck,
Darren

3 Likes

Hi @1318980

I have increased the number of correctors to 15, and I still receive error. I also decreased the length of the time steps. However, I receive this error in “RUN2”:

“The Courant number (CFL) exceeded the limit of 1. You may experience either instability or bad temporal accuracy. It is recommended to keep the CFL number below 0.7. In order to achieve this you need to decrease the time step.”

Blockquote
I would create a primitive so that you can subdomainly initialise the solution with the other fluid also.

Is there any link that can show me how to do it?

1 Like

Hey @YB128
did you have succses with the simulation ?

Greetz
Kai

1 Like

@Kai_himself

Not yet, unfortunately! I have tried different settings, but I still can’t make it.

1 Like

Hi @YB128, really sorry for the delay, only noticed this when it was bumped by @Kai_himself. I think most multiphase simulation projects on the public projects have subdomain based initialisation for phase fraction. The general workflow is to make a geometry primitive in the simulation setup, and then enable subdomain initialisation from the drop-down box, then assign your second fluid in the region and set the default fraction to the primary fluid.

The error about CFL is detailed in the log where it says it was in the order of e34, this is unrealistic and so you have likely diverged for another reason, try the above subdomain based initialisation and see if this goes away otherwise I will look into the simulation for you.

Hope this helps and isn’t too late,
Darren

2 Likes

Hey @1318980,
i have to interrupt once, sorry.
What do you mean with a subdomain based initialisation. Should it be a point at the Input or a Cartesian Box to envelop the Mesh ?
Sorry for the stupid question.

So this is a general idea:

The default phase is 0 whereas the phase assigned to the primitive is 1. The primitive could be any shape and is used to initialise the domain within it. Hence the ‘subdomain’ part. So here the channel that inlets phase 1 will already have some phase 1 fluid in there making stability much better and will also reduce the time it takes for that phase to reach the Y section. If however, you want to see its path simply make the domain much smaller just to initialise the first few cells of the pipe.

Hope this helps you and sorts out your stability issue.
Darren

3 Likes

Hey @1318980,
i try it too, i dont get the result you get.
Under Geometry Primitives i can choose Cartesian Box and Box, but only the Box can be rotatet. Furthermore the Box cant be choose at the Phase fraction/subdomain 1…

How do you get the Box rotatet or choose the normal Box at the subdomain ?

Greetz Kai

Hi @Kai_himself, not sure why box woudnt work however I looked at how I setup the above and I used a cylinder with these settings:

Hope this works :slight_smile:
Darren

1 Like

Ah sorry it was my fault. I think it is a rectangle on your first picture.

Thank you

Kai