Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Implicit versus explicit method in the CDF module

Please login with a confirmed email address before reporting spam

When using implicit method (BDF: Backwards differentiation formula) in the CFD module in COMSOL, should I assume that it works with the same theory of Backwards Euler in structures? And that I can choose larger time steps and my solution will always be numerically stable and convergent, without the need to consider Courant number limit.

I am running a time dependent simulation in the Rotating machinery-laminar flow module, and it is not clear to me how to assure stability and convergency in my mesh based on time step versus mesh size.

Thanks

3 Replies Last Post 30.01.2012, 01:21 GMT-5
Jim Freels mechanical side of nuclear engineering, multiphysics analysis, COMSOL specialist

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29.01.2012, 21:37 GMT-5
Yes. COMSOL is using fully implicit time stepping throughout; hence, large Jacobian matrix is formed and used in the solution.

One word of caution that we ran into with the fluid/structure coupling (FSI) type simulations: If there is a frequency within the system that you want to capture in a time-dependent solution, you need to keep the time step small enough to capture that frequency. This would not be an issue for the steady-state. For example, if the fundamental frequency of the structure was 10 hz, you would need to keep the time step at less than 1/10 sec or smaller in order to visualize the results accurately in the time response.
Yes. COMSOL is using fully implicit time stepping throughout; hence, large Jacobian matrix is formed and used in the solution. One word of caution that we ran into with the fluid/structure coupling (FSI) type simulations: If there is a frequency within the system that you want to capture in a time-dependent solution, you need to keep the time step small enough to capture that frequency. This would not be an issue for the steady-state. For example, if the fundamental frequency of the structure was 10 hz, you would need to keep the time step at less than 1/10 sec or smaller in order to visualize the results accurately in the time response.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29.01.2012, 23:01 GMT-5
Thank you so much.

I wonder if you can help with other difficulty I've been going through. I am try to find the time in which the velocity of my model stabilizes. I ran three different simulations with a time step of 0.1 during 5 s, 10 s and 20s . However, I am obtaining the same answers for the three models, because COMSOL is giving me just the first 50 time step solutions, which means the answers for the first 5 seconds.

Is there something should I modify in COMSOL to obtain more data, or does it mean that after 5 seconds the model is stable and for this reason COMSOL is not giving me any more solutions?

Thanks
Thank you so much. I wonder if you can help with other difficulty I've been going through. I am try to find the time in which the velocity of my model stabilizes. I ran three different simulations with a time step of 0.1 during 5 s, 10 s and 20s . However, I am obtaining the same answers for the three models, because COMSOL is giving me just the first 50 time step solutions, which means the answers for the first 5 seconds. Is there something should I modify in COMSOL to obtain more data, or does it mean that after 5 seconds the model is stable and for this reason COMSOL is not giving me any more solutions? Thanks

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30.01.2012, 01:21 GMT-5
Hi

What kind of time stepping do you use ? if its the default "free" COMSOL assumes a diffusion type problem and the ime increments will increase regulraly (provided the slopes of the solving system remains more or less stable). If you expect oscillating BC's or responses, its worth to define a rahter short time step range (many steps to resolve the oscillations) and use a time stepping "intermediate or strict" this will certainly take longer to solve but you will be sure COMSOl considers the true response at your time steps and not giving you the interpolation between two loarge steps

--
Good luck
Ivar
Hi What kind of time stepping do you use ? if its the default "free" COMSOL assumes a diffusion type problem and the ime increments will increase regulraly (provided the slopes of the solving system remains more or less stable). If you expect oscillating BC's or responses, its worth to define a rahter short time step range (many steps to resolve the oscillations) and use a time stepping "intermediate or strict" this will certainly take longer to solve but you will be sure COMSOl considers the true response at your time steps and not giving you the interpolation between two loarge steps -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.