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.

Fluid-Structure-Interaction, Model won't work

Please login with a confirmed email address before reporting spam

Hi!

I am trying to build a model with a membrane covering a water domain. A boundary load on a part of the domain should produce a 'dent' while the part of the membrane which is not subjected to the boundary load should move in the opposite direction due to the higher pressure in the water domain (assuming incompressible water).

I am interested in the time dependent solution of this problem, which unfortunatly wont work, producing this error message:

''Failed to find consistent initial values.
Segregated group 3

Attempt to evaluate real square root of negative number.
Function: sqrt
Last time step is not converged.''

The stationary solver works, though, but only with moderate boundary loads.

Can anyone help me on this topic?

Best regards, Christian


2 Replies Last Post 17 feb 2012, 08:57 GMT-5
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 30 dic 2011, 03:58 GMT-5
Hi

I can come with a few comments:
You have a system, with all "0" initial conditions (at rest) where you suddenly apply a force (no true time dependence), locally and want to see the time decay. This means a very abrupt start, the solver cannot follow any "smooth path" and diverges. I would propose to 1) use a step function to apply your force "gently". Then try first a stationary case to see if the system is OK without any time value (if you define a time dependent load and do NOT use a time solver then "t" will not be defined and you need to add it as a parameter t = 1[s]


Then you mesh, its very coarse, you membrane has only 1 element across, you will not really manage to resolve correctly what is happening therein. Furter if you want to see the flow, you need to respect the noslip conditions and have a mesh more compatible with that (boundary mesh), try the default physics mesh, fotr me its already better adapted than the one you have now.

Then, I forgot: turn on "plot while solving" or use the probe plots to analyse what the solver is doing. I'm sure once you try to solve in stationary (as a first model verification) and noice it still does not solve correctly, you will though see from the solution that you have missed something in your BC's, rather something too much ... ;) but this might not be sufficient ...

By the way, you can also define a load as a function of "r" along your boundary, to have a more realistic "smoother" load case, either with a "pulse" function or a deported gaussian or something else. Then you do not need to define your points/lines in the geometry, and you can change it's shape in a simpler way


--
Good luck
Ivar
Hi I can come with a few comments: You have a system, with all "0" initial conditions (at rest) where you suddenly apply a force (no true time dependence), locally and want to see the time decay. This means a very abrupt start, the solver cannot follow any "smooth path" and diverges. I would propose to 1) use a step function to apply your force "gently". Then try first a stationary case to see if the system is OK without any time value (if you define a time dependent load and do NOT use a time solver then "t" will not be defined and you need to add it as a parameter t = 1[s] Then you mesh, its very coarse, you membrane has only 1 element across, you will not really manage to resolve correctly what is happening therein. Furter if you want to see the flow, you need to respect the noslip conditions and have a mesh more compatible with that (boundary mesh), try the default physics mesh, fotr me its already better adapted than the one you have now. Then, I forgot: turn on "plot while solving" or use the probe plots to analyse what the solver is doing. I'm sure once you try to solve in stationary (as a first model verification) and noice it still does not solve correctly, you will though see from the solution that you have missed something in your BC's, rather something too much ... ;) but this might not be sufficient ... By the way, you can also define a load as a function of "r" along your boundary, to have a more realistic "smoother" load case, either with a "pulse" function or a deported gaussian or something else. Then you do not need to define your points/lines in the geometry, and you can change it's shape in a simpler way -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 feb 2012, 08:57 GMT-5
Hi everybody,

I have a similar simulation to run, where there is a perfusion of a fluid (say water) on an impermeable structure (say bladder patch) that cover all the diameter of the pipe, constrained to the pipe itself. In my simulation (attached) I apply a step function in order to gently increase the initial value of water speed, and I place the outlet beyond the bladder patch (since the deformed bladder is supposed to move the fluid after it) . The mesh is a fine free triangular (seems to me enough accurate), and I apply a time-dependent study in order to see how the bladder patch gets deformed (I expect a cupola shape). I don't want a stationary study since I guess the steady state in this situation is reached when the bladder patch gets broken.

It seems to me that all boundary condition are good, but still when I compute the study it says "Attempt to evaluate real square root of negative number". I guess that the fact that the water can not flow freely in the pipe because the bladder patch blocks it makes the solver crazy, but I can not find where the mistake is...

any hint?

Thanks for your kindness and for your attention
Best regards
Alberto
Hi everybody, I have a similar simulation to run, where there is a perfusion of a fluid (say water) on an impermeable structure (say bladder patch) that cover all the diameter of the pipe, constrained to the pipe itself. In my simulation (attached) I apply a step function in order to gently increase the initial value of water speed, and I place the outlet beyond the bladder patch (since the deformed bladder is supposed to move the fluid after it) . The mesh is a fine free triangular (seems to me enough accurate), and I apply a time-dependent study in order to see how the bladder patch gets deformed (I expect a cupola shape). I don't want a stationary study since I guess the steady state in this situation is reached when the bladder patch gets broken. It seems to me that all boundary condition are good, but still when I compute the study it says "Attempt to evaluate real square root of negative number". I guess that the fact that the water can not flow freely in the pipe because the bladder patch blocks it makes the solver crazy, but I can not find where the mistake is... any hint? Thanks for your kindness and for your attention Best regards Alberto

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.