|
[Sponsors] |
June 6, 2016, 03:27 |
strategy to enhance of time steps
|
#1 |
New Member
faraz
Join Date: Nov 2015
Posts: 8
Rep Power: 10 |
Hi all,
I try to simulate an unsteady multiphase issue with vof model, but time step size for continuity convergence is order of e^10-6, and flow time is about an hour, so i can't simulate it because of very high computational expense. Are there a strategy to enhance time step size? thank you for answers. |
|
June 7, 2016, 16:04 |
|
#2 |
Senior Member
Matt
Join Date: Aug 2014
Posts: 947
Rep Power: 17 |
Can you describe how you arrived at your time step and why the total simulation time needs to be one hour?
|
|
June 7, 2016, 17:34 |
|
#3 |
New Member
faraz
Join Date: Nov 2015
Posts: 8
Rep Power: 10 |
I use variable time step with courant number=2, time step automatically change to near 10^-6. i simulate a big tank fill process that have about one hour filling time.
|
|
June 7, 2016, 17:50 |
|
#4 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,764
Rep Power: 71 |
Quote:
what about the physical dimensions of the tank, the size of the computational domain and the velocity? |
||
June 7, 2016, 18:03 |
|
#5 |
New Member
faraz
Join Date: Nov 2015
Posts: 8
Rep Power: 10 |
||
June 7, 2016, 18:27 |
|
#6 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,764
Rep Power: 71 |
Quote:
well, consider that in each direction ci = ui dt/ hi but, depending on the scheme you are using, the numerical stability does not depend only on the courant number but is it a combination with the diffusive constraint. Are you using an explicit time integration? |
||
June 7, 2016, 18:53 |
|
#7 |
New Member
faraz
Join Date: Nov 2015
Posts: 8
Rep Power: 10 |
No, i am using implicit formulation, because of pressure-based solver.
|
|
June 8, 2016, 03:06 |
|
#8 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,764
Rep Power: 71 |
Quote:
But with your small time step the solution is physically acceptable after a certain amount of steps? I always suspect a bug in the solver. |
||
June 8, 2016, 06:45 |
|
#9 |
Senior Member
Michael Prinkey
Join Date: Mar 2009
Location: Pittsburgh PA
Posts: 363
Rep Power: 25 |
It is not clear to me if this is a commercial code, community code, or something you've developed. The time step will be determined by the smallest cell in the mesh. If you have a inlet tube and some transition cells feeding into the tank or, if you have applied boundary layer stretching near walls, some of those cells could have a very small length scale with a typical velocity value, and that makes the CFL number time step very small.
|
|
June 8, 2016, 07:08 |
|
#10 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,764
Rep Power: 71 |
Quote:
I agree ... but he wrote of being using an implicit formulation and that makes me having some doubts...for that reason I suggested to write the resulting max cfl in the computation |
||
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
[Other] Contribution a new utility: refine wall layer mesh based on yPlus field | lakeat | OpenFOAM Community Contributions | 58 | December 23, 2021 02:36 |
Multiple floating objects | CKH | OpenFOAM Running, Solving & CFD | 14 | February 20, 2019 09:08 |
High Courant Number @ icoFoam | Artex85 | OpenFOAM Running, Solving & CFD | 11 | February 16, 2017 13:40 |
Star cd es-ice solver error | ernarasimman | STAR-CD | 2 | September 12, 2014 00:01 |
AMI interDyMFoam for mixer nu problem | danny123 | OpenFOAM Programming & Development | 8 | September 6, 2013 02:34 |