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.
Applying boundary stresses
Posted Jun 14, 2010, 2:42 a.m. EDT 1 Reply
Please login with a confirmed email address before reporting spam
Hi all,
I'm trying to simulate fluid flow and heat transfer of molten metal. Strong thermal gradients cause Marangoni boundary conditions on upper and lower surfaces of the domain. Due to convergence difficulties the idea was to ramp up the Marangoni boundary condition with an additional factor.
I'm not completely sure how to understand the COMSOL workflow in this point. Is it appropriate to use the parametric sweep with the conditions for the ramp factor with initial conditions set to "Solution"? There also exists the possibility to enable the "continuation" checkbox in the stationary tab in the study settings. The stationary tab in the solver sequence unfortunately comes without this checkbox.
Now I use a combination of both: continuation checkbox on and parameter defined in the first stationary tab under "study" and parametric sweep with the same parameter defined again. I guess this is not the way to do it??
Moreover there is no convergence yet for realistic Marangoni stress conditions, maybe because of wrong solver setup?
Thanks in advance!
I'm trying to simulate fluid flow and heat transfer of molten metal. Strong thermal gradients cause Marangoni boundary conditions on upper and lower surfaces of the domain. Due to convergence difficulties the idea was to ramp up the Marangoni boundary condition with an additional factor.
I'm not completely sure how to understand the COMSOL workflow in this point. Is it appropriate to use the parametric sweep with the conditions for the ramp factor with initial conditions set to "Solution"? There also exists the possibility to enable the "continuation" checkbox in the stationary tab in the study settings. The stationary tab in the solver sequence unfortunately comes without this checkbox.
Now I use a combination of both: continuation checkbox on and parameter defined in the first stationary tab under "study" and parametric sweep with the same parameter defined again. I guess this is not the way to do it??
Moreover there is no convergence yet for realistic Marangoni stress conditions, maybe because of wrong solver setup?
Thanks in advance!
1 Reply Last Post Jun 14, 2010, 9:57 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I'm too fighting with the parametric set-up, I have noticed that when you have a "stationary" under the study tree, you should set the continuation here (highest in the tree, BEFORE you generate the sequence) this will add a "parametric sweep under the "Solver Sequence Stationary" that inherits the values you entered above. (you can also add it by hand at this node level, but then its not regenerated automatically the next time you compute except of you "use existing sequence")
You can also mix parametric sweep and internal continuation, it's slightly tricky I find, and you must add some "store solutions" to have the fresh starting points for secondary analysis.
I learned quite a lot by studying the structural buckling analysis, that automatically adds a linked stationary and eigenfrerquency/buckling sequence
have fun Comsoling
Ivar
I'm too fighting with the parametric set-up, I have noticed that when you have a "stationary" under the study tree, you should set the continuation here (highest in the tree, BEFORE you generate the sequence) this will add a "parametric sweep under the "Solver Sequence Stationary" that inherits the values you entered above. (you can also add it by hand at this node level, but then its not regenerated automatically the next time you compute except of you "use existing sequence")
You can also mix parametric sweep and internal continuation, it's slightly tricky I find, and you must add some "store solutions" to have the fresh starting points for secondary analysis.
I learned quite a lot by studying the structural buckling analysis, that automatically adds a linked stationary and eigenfrerquency/buckling sequence
have fun Comsoling
Ivar