Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
2009年10月16日 GMT-4 17:36
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
2009年10月30日 GMT-4 06:09
How do you reinitialize at every time step? Isn't this a process being done only at the very beginning? I am trying to do a two-phase phase field case .. but that doesn't converge either.
How do you reinitialize at every time step? Isn't this a process being done only at the very beginning? I am trying to do a two-phase phase field case .. but that doesn't converge either.
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
2009年11月1日 GMT-5 03:56
Hi
I have understood (but I have only followed one training course , hav'nt worked on these solutions extensively) that these kind of problems needs first an "initialisation phase" (short time step) to set up correctly the physics, then you turn on longer time steps analysis.
In your model there is only the initialisation
You can automatise this in the Solver Manager Sequence tab, by first saving a "Transient Initialisation" sequence for a short start up time, save the solution, and then follow by a standard "transient" calculation, but based on the saved values (set up your Solver Parameters correctly for this second phase).
There are a few examples in the doc, and others in the standard models. You need to get aquinted with the solver settings, multiple sequenced solving and the secondary tabs, this takes some training.
It will all be shown in another sequence logic in V4. I'm not sure it will be easier for the new-beginners, but certainly more logical for those of us feeling more familiar with COMSOL and its way to do physics, which is quite different from classical FEM methods limited to a single physics were everything is predefined ready.
Good luck
Ivar
Hi
I have understood (but I have only followed one training course , hav'nt worked on these solutions extensively) that these kind of problems needs first an "initialisation phase" (short time step) to set up correctly the physics, then you turn on longer time steps analysis.
In your model there is only the initialisation
You can automatise this in the Solver Manager Sequence tab, by first saving a "Transient Initialisation" sequence for a short start up time, save the solution, and then follow by a standard "transient" calculation, but based on the saved values (set up your Solver Parameters correctly for this second phase).
There are a few examples in the doc, and others in the standard models. You need to get aquinted with the solver settings, multiple sequenced solving and the secondary tabs, this takes some training.
It will all be shown in another sequence logic in V4. I'm not sure it will be easier for the new-beginners, but certainly more logical for those of us feeling more familiar with COMSOL and its way to do physics, which is quite different from classical FEM methods limited to a single physics were everything is predefined ready.
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
2009年11月1日 GMT-5 10:38
Yes, and I have been following these initialization steps closely. I have a uniform grid, and the initialization sesms to work fine - but the simulations simply doesn't converge. And this is a pretty simple 2D-case.
Well, I attach my test-case here. Feel free to have a look - it is a fairly simple two phase inlet-outlet flow with 20% volume fraction gas in. And that's it. Still, it doesn't converge at all ....
Yes, and I have been following these initialization steps closely. I have a uniform grid, and the initialization sesms to work fine - but the simulations simply doesn't converge. And this is a pretty simple 2D-case.
Well, I attach my test-case here. Feel free to have a look - it is a fairly simple two phase inlet-outlet flow with 20% volume fraction gas in. And that's it. Still, it doesn't converge at all ....
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
2009年11月4日 GMT-5 09:56
I reply to my own post, after having confronted the Comsol-support desk, and I am now instead using bubbly flow, and I am at least one step closer to getting my 3D-case to work. The phase field does not seem to be the right solver for my case anymore ...
I reply to my own post, after having confronted the Comsol-support desk, and I am now instead using bubbly flow, and I am at least one step closer to getting my 3D-case to work. The phase field does not seem to be the right solver for my case anymore ...