Discussion Forum

Hi

Unfortunately I have not yet received my 4.3 CD, it's still in the post somewhere ;) But what you are explaining there is one of the features I'm really wanting to use and I find it important to better udnerstand what's happening, I agree it sounds slightly strange, but ...

in 4.2 if you add several sover nodes in the same study, COMSOL by default stores the intermediate solutions and link the initial values to the previous solver /sored solution case, but in v4.2 we cannot link in conditional BCs.

I'm sure COMSOl 4.3 also behave like this (when you do not touch the BCs), but its worth to check that this default behaviour remains so. On the other hand, if you change the BCs, perhaps COMSOL assumes you shoud restart from initial values by default, and you as user must go into the dependent variables nodes and link them maually to the stored solutions (2nd check to be done).

Another point is that when you switch BCs, then the final state of one solver (if assumed as new initial state of the second solver) might not be coherent with the new BCs, in this case yes I assume that COMSOL decides to restart to "0" initaila conditions "by default".
Switching the BCs is not fully trivial, as if you apply a BC load, and then remove it for the second solver, the part will tend to return to minimum stress state, and in a stationary solver case it would revert and there will be minimum difference between starting from the stressed state or from the initial state, but in a time series solver, probably the part would start to oscillate (at least if the inertial terms are considered). Then what is "logical" and how to link multiple solvers by default ?

Pls report back your findings, as this is of use for several of us ;)


--
Good luck
Ivar

Thank you very much for your reply ! You are right, Comsol V4.3 behaves same as previous versions, even with the conditional boundary conditions in different steps. As you said before, if a certain Bc is deactivated in next step, it tries to regain its min stress configuration with the new boundary conditions and of course it uses the previous solution as initial condition by default, as expected, i.e it is not using 'zero initial conditions'. Well, its fine !

However, the questions are
1. How can we apply new Bc's on the deformed geometry ?
2. How can we save the deformed geometry, and also the associated solution ? so that i can use this deformed geometry as a new initial or basic configuration for further analysis.

In my complex model (poro-hyperelastic + inelastic growth), i want to repeat it for 1000 cycles, the new Bc's are based on the solution of previous steps, deformed geometry. At the end of one cycle (which involves 3 steps), i want to save geometry, solution after every step, and if possible re-mesh it for further analysis.

Any help would be greatly appreciated !

/krishna

Hi

there are some options, that I have never tested, that proposes to remesh the deformed state and you continue with the deformed shape, check th list of the solver subnodes (right click...), but if this shape comes from an external force, the means you must keep the force load, else you are not respecting the stress equilibrium. And repeating this too many times would mean that you might well explode your disk storage, as well as RAM limit.

Probably, such cases are best treated via matlab or Java scripting. Still think well over what is happening with all the variables of the physics, how wil lthey evolve if you remove the reason for the deformation or variable change ? This is very much model dependent

--
Good luck
Ivar

Thanks Ivar ! Of course, we have to respect stress equilibrium, what i exactly mean was, the deformed geometry should be a new initial geometry for the next analysis, by clearing all the solutions; before that i needed to know all the geometric points/regions where certain variable reach the max triggering values from the solution.

The problem is simple. Initial geomtry-----> Inelastic growth--->(solution 1, Geometry 1)---->Mechanical loading----->(solution 2, Geometry 2)

I want to preserve the Geometry 1 all the time, but certain conditional boundary conditions come from solution 2 (should map the stress values from geometry 2 to geometry 1, or can be interpolated onto the standard mesh grid)

question is how to save geometry 1 (deformed due to inelastic process) ? and keep track of solution 2 /geometry 2 ?

Anyway, thanks a lot for your patient replying !

/Krishna

Hi,

I have got a similat Problem. I am working first with a solid mechanics solver and need the deformed geometry for a fluid dynamics simulation in the second study. When I try to remesh all I get is the original mesh, I then stored the mesh, but a stored geometry can't be remeshed!! When I then continue computing I get a solution that looks relatively good but is still in the non deforemed geometry!!

I attached the file!

Thanks for Helping!!
Andreas

Hi,

The remeshing works only if you include non-linear geometry into the model. You will find the option in 'study settings' under study. I hope this helps.

In my case, i wanna do it 1000 times. Still i haven't figured it out how to do it.....

/K

Hi,

I included a non linear geometry in the study and the result of this first study should be a deformed nonlinear geometry, still when I remesh it, the remeshed geometry is the input geometry with witch I started!! But I want the new geometry to be remeshed!

/Andreas Artelsmair

Hi
in your model above you have only a structural and a fluid domain, but the two are not linked. You need to add first an ALE mesh node "physics/math"and set that one up. The easiest is to use COMSOL pre-cooked FSI physics that include all three readily set-up

Then your mesh is by far too coarse for the fluid part, not to say you have no boundary layers. For the solid too, only one element in the thickness is not giving very correct results, I use at least 3 or better more. By the way you could use a structured quad mesh on the top surface and sweep it through, then add the boundary layer for the fluid, nice exercise to check the model sensitivity to the mesh density

With a pressure load, you have also an issue about absolute pressure and gauge pressure of the fluid versus structural part, but that is rather model definition dependent

--
Good luck
Ivar

Hi,

The remeshing works only if you include non-linear geometry into the model. You will find the option in 'study settings' under study. I hope this helps.

In my case, i wanna do it 1000 times. Still i haven't figured it out how to do it.....

/K

Hi, have you fixed this problem?
I have the same problem, could you share your model to me ?

Thanks

/chaoyang