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 The information provided may be out of date.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.
Time dependent geometry change- problem of convergence
Posted 11 juin 2012, 08:58 UTC−4 Geometry, Studies & Solvers Version 4.2a 3 Replies
Please login with a confirmed email address before reporting spam
Im trying to simulate a geometry change by time, I studied the previous questions regarding this topic and also run the sample model from Ivar for parameter time sweep1.mph.
to explain the problem, Im trying to simulate the breathing process in human lungs. to do so I have a rather complex 3d model but to find out how it would work, I try to have a simple cylinder inside a sphere. in addition to that we have a parametric curve which contains a current, in magnetic field physics, first study is a frequency domain,
the second study is a time dependent one, I want to simply change the radius of the mentioned cylinder in time steps.
and the error :
Failed to find consistent initial values.
Matrix has zero on diagonal.
Last time step is not converged.
appears.
I tried to change the time stepping to intermediate. no change, the problem is disapeared when Im disabling the parametric curve. ( I mean with no current and as a consequence no result)
anyone has any idea?
Thanks in advance.
Please login with a confirmed email address before reporting spam
I'm not sure you can drive a geometry with the variable "t", in fact you cannot use any Variables, only Parameters and you need to use the Parameric Sweep solver node to loop around the geometry and the mesh nodes. A time stepping is like a parametric continuation sweep, it does not loop around the geometry neither.
On the other hand, if you do not need the inertial terms (d^2_u_/dt^2) you could mimic a quasti-static time sweep by a Parameter you can call "tt", and a Parametric sweep node on that variable
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
and one more thing,
Imagine that we want to have a frequency sweep, we have an array of 7 frequencies, in each frequency all of the conductivities would change and as far as I know comsol will show all the possible combinations which in fact is not correct and not necessary.
for example in 1 MHz we can have a specific array of conductivity which would not be the same for 10 MHz..
Is there any way to define like : If freq=x then sigma=y. ( note that there is no formula or exact model for the relation between conductivity and frequency in our case)
I'm not sure if I express my problem correctly.
Thanks again
Hadis
Please login with a confirmed email address before reporting spam
what you are saying is that the conductivity is depending on the frequency, then you can change sigma in the material data node and make it depend on the frequency.
now why does sigma change with "f", probably because of skin effects, this is not a true material effect but a complex EM field effect, and depending on the model and physics you are using COMSOL will already include this behaviour in your model, so you should perhaps, not add it twice also via the material parameters, and in a n empirical way.
that is up to you to decide , analyse carefully your model
--
Good luck
Ivar