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.
Transient analysis in COMSOL
Posted Nov 9, 2012, 12:27 p.m. EST 2 Replies
Please login with a confirmed email address before reporting spam
When doing a transient analysis (Let’s say heat transfer or fluid flow), you may sometime need very small time step size, though you don’t need the solution for every time step. Recording the solution for every time step makes the file huge and hard to process.
I’m not sure if COMSOL has such a feature when you are doing a transient analysis. For example to set the time step to 0.1 but save the solution each 10 seconds.
I’m sure such feature is essentially needed for transient analysis.
Could anyone help me with that?
I’m not sure if COMSOL has such a feature when you are doing a transient analysis. For example to set the time step to 0.1 but save the solution each 10 seconds.
I’m sure such feature is essentially needed for transient analysis.
Could anyone help me with that?
2 Replies Last Post Aug 29, 2016, 11:48 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
Comsol, in default mode selects the time steps automatically, and saves those you define in the list, so you should rather define the time steps you want and let COMSOL do the intermediate depending on how it analysis the solution.
On the other han , by default, COMSOL is set up for diffusion type solutions, and if you have oscillating BC, the solver might decide to take too large steps, after some while, and might skip your detailed BC variantions. To avoid this you should then set up a strickt or intermediate instead of free-automatic stepping scheme in the lower solver node.
For diffusion type analysis, I often use power series time stepping of the type 2^{range(-4,0.5,3)} to avoid storing too many intermediate points and make the files unecesarilly larges
--
Good luck
Ivar
Comsol, in default mode selects the time steps automatically, and saves those you define in the list, so you should rather define the time steps you want and let COMSOL do the intermediate depending on how it analysis the solution.
On the other han , by default, COMSOL is set up for diffusion type solutions, and if you have oscillating BC, the solver might decide to take too large steps, after some while, and might skip your detailed BC variantions. To avoid this you should then set up a strickt or intermediate instead of free-automatic stepping scheme in the lower solver node.
For diffusion type analysis, I often use power series time stepping of the type 2^{range(-4,0.5,3)} to avoid storing too many intermediate points and make the files unecesarilly larges
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
8 years ago
Hi Ivar,
In 2D, I am using a COMSOL, transient solver to simulate a Gaussian pulse plane wave (wavelength=1.0um, pulse width=10fs) propagation in free space. Scattering boundary condition and Perfect magnetic conductor condition are used for boundaries so that the pulsed plane wave propagates well in free space. Near the right boundary, a cutline/cutpoint is put to grab the complex data.
In order to simplify the problem, I use uniform mesh and dx=dy=wavelength/10. However, I find the time step will influence the accuracy of simulation result, like pulse envelope phase, peak arrival time. Hence, how can I set the optimal time step in advance to get an accurate pulse envelope phase? Thanks.
Best,
Shengli
In 2D, I am using a COMSOL, transient solver to simulate a Gaussian pulse plane wave (wavelength=1.0um, pulse width=10fs) propagation in free space. Scattering boundary condition and Perfect magnetic conductor condition are used for boundaries so that the pulsed plane wave propagates well in free space. Near the right boundary, a cutline/cutpoint is put to grab the complex data.
In order to simplify the problem, I use uniform mesh and dx=dy=wavelength/10. However, I find the time step will influence the accuracy of simulation result, like pulse envelope phase, peak arrival time. Hence, how can I set the optimal time step in advance to get an accurate pulse envelope phase? Thanks.
Best,
Shengli