LEVEL SET: PROBLEM WITH MESH

Numerical methods and mathematical models of Elmer
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: LEVEL SET: PROBLEM WITH MESH

Post by kevinarden »

That is strange, I have only received that error when using umfpack and have only associated with the umfpack direct solver.
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

kevinarden wrote: 02 Jun 2021, 19:18 That is strange, I have only received that error when using umfpack and have only associated with the umfpack direct solver.
So, when you use the case.sif and the mesh that I have shared, it works on your pc?
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: LEVEL SET: PROBLEM WITH MESH

Post by kevinarden »

To ensure I have the right one can you share the sif and mesh again?
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

kevinarden wrote: 02 Jun 2021, 19:25 To ensure I have the right one can you share the sif and mesh again?
I share the link of the used mesh: https://we.tl/t-sVMmcZGE6l
Attachments
case_levelset.sif
(5 KiB) Downloaded 129 times
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: LEVEL SET: PROBLEM WITH MESH

Post by kevinarden »

As written I did not get the error, but the iterative solution did diverge.
I divided the timestep size and velocity by 100 and it converged.
case.sif
(5.01 KiB) Downloaded 132 times
Attachments
velocity.png
velocity.png (16.33 KiB) Viewed 1779 times
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

kevinarden wrote: 03 Jun 2021, 01:49 As written I did not get the error, but the iterative solution did diverge.
I divided the timestep size and velocity by 100 and it converged.

case.sif
Thanks a lot Kevin for your work, unfortunately, I have to use a velocity between 0.5 and 1 m/s for my project.
Regards,
Andrea
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

kevinarden wrote: 03 Jun 2021, 01:49 As written I did not get the error, but the iterative solution did diverge.
I divided the timestep size and velocity by 100 and it converged.

case.sif
Hi Kevin,
I was running your code and a problem appeared in the results visualization in Paraview, apparently, I could see the velocity evolution up to the 2-time step, next it couldn't be visualized longer. I corrected the mistake and I attach the new case.sif. The velocity profile formula is:

Code: Select all

Real MATC "if (tx<=h*h1) -0.5*(tx/h1)^2+ h*(tx/h1) -Up; else 0;"

Thanks a lot for your help.
Regards,
Andrea
Attachments
case.txt
(5.21 KiB) Downloaded 128 times
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

I have the following question: If I have a problem with the Level set solver, It would be better to use an iterative method or a direct method for the linear system? The bandwidth of the Level set on what depends? Actually, I have obtained better results with a direct method like umfpack and Mumps.
Regards,
Andrea
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: LEVEL SET: PROBLEM WITH MESH

Post by kevinarden »

For any linear problem a direct method, if available, is preferred over an iterative method. It is computationally more efficient and you don't have to worry about divergence.
Andrea_P
Posts: 176
Joined: 22 Mar 2021, 18:39
Antispam: Yes
Location: POLYTHECNIC UNIVERSITY OF TURIN

Re: LEVEL SET: PROBLEM WITH MESH

Post by Andrea_P »

kevinarden wrote: 03 Jun 2021, 12:25 For any linear problem a direct method, if available, is preferred over an iterative method. It is computationally more efficient and you don't have to worry about divergence.
So, in my case it is better a direct method, for this reason, the original one used was umfpack.
POLYTECHNIC UNIVERSITY OF TURIN-DIMEAS
Post Reply