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 The information provided may be out of date.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.
Thermal expansion induced vibrations
Posted Mar 2, 2012, 11:41 a.m. EST Materials, Parameters, Variables, & Functions, Structural Mechanics Version 4.1, Version 4.2, Version 4.2a 6 Replies
Please login with a confirmed email address before reporting spam
Dear all,
I have a very simple question to ask of you.
I am looking to pre-strain a membrane and am exploring the possibility of using thermal expansion to do it.
The model is a membrane 30[mm] diameter and 40[um] thickness. I use fixed constraint around the edge of the membrane.
I then add a Thermal Expansion node under Linear Elastic model 1, the values I use should give rise to 5% volumetric strain.
However, when I look at the Z displacement (normal to plane of membrane) there are vibrations. The first part of the vibrations looks like its numerical problems settling down, however there is a considerable vibration in the z direction, please see the attached.
Also, the vibrations looks to be offset by around 16e-8m, I would expect it to be at 0.
Does anyone know what might be going on here? As I do not have any material damping I think these vibrations would carry on. Also, the magnitude is about the magnitude of the displacements that I would be looking at at a alter point in the model.
Have I been stupid or is this a known problem of using Thermal Expansion as a way of inducing strain into a model.
Thanks for reading this, I hope people have some interesting ideas.
I have a very simple question to ask of you.
I am looking to pre-strain a membrane and am exploring the possibility of using thermal expansion to do it.
The model is a membrane 30[mm] diameter and 40[um] thickness. I use fixed constraint around the edge of the membrane.
I then add a Thermal Expansion node under Linear Elastic model 1, the values I use should give rise to 5% volumetric strain.
However, when I look at the Z displacement (normal to plane of membrane) there are vibrations. The first part of the vibrations looks like its numerical problems settling down, however there is a considerable vibration in the z direction, please see the attached.
Also, the vibrations looks to be offset by around 16e-8m, I would expect it to be at 0.
Does anyone know what might be going on here? As I do not have any material damping I think these vibrations would carry on. Also, the magnitude is about the magnitude of the displacements that I would be looking at at a alter point in the model.
Have I been stupid or is this a known problem of using Thermal Expansion as a way of inducing strain into a model.
Thanks for reading this, I hope people have some interesting ideas.
Attachments:
6 Replies Last Post Mar 6, 2012, 6:36 a.m. EST
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
You can conceptually apply pre-strain via thermal expansion. Why not do a stationary solution instead of having to deal with these oscillations and damping?
Nagi Elabbasi
Veryst Engineering
Nagi Elabbasi
Veryst Engineering
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
or perhaps turn off the inertial terms (if possible for your model) that would drop the rho*d^2_u_/dt^2 term and work in Quasi static. Often a "Continuation sweep in stationary is easier as then you are working not only "quasi" but "fully static" to the solver resolution
--
Good luck
Ivar
or perhaps turn off the inertial terms (if possible for your model) that would drop the rho*d^2_u_/dt^2 term and work in Quasi static. Often a "Continuation sweep in stationary is easier as then you are working not only "quasi" but "fully static" to the solver resolution
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Nagi and Ivar,
Nagi.
The truth is, I am not sure how to successfully add a static step to a transient model. I know that you can add a static step and use the results for the start of the temporal study, but I am having some problems doing it.
Do you happen to know of an example of such a model? I have had a good look but I cant find one.
Ivar,
Thats a good suggestion, I will have a look if its possible and what effects it has later.
Perhaps one of you would be kind enough to create a simple static>temporal model for me? I would then create a model exchange from this with as many details as I can add as I think a lot of people have asked a similar question.
Thank you for your help with this.
Rob
Nagi.
The truth is, I am not sure how to successfully add a static step to a transient model. I know that you can add a static step and use the results for the start of the temporal study, but I am having some problems doing it.
Do you happen to know of an example of such a model? I have had a good look but I cant find one.
Ivar,
Thats a good suggestion, I will have a look if its possible and what effects it has later.
Perhaps one of you would be kind enough to create a simple static>temporal model for me? I would then create a model exchange from this with as many details as I can add as I think a lot of people have asked a similar question.
Thank you for your help with this.
Rob
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I'mnot behind my COMSOL Ws just now, but previously (and I hope it's still true for v4.2a, I used to load a buckling example, that is (was) a combination of a statioanry analysis + eigenfrequency (set for buckling) and then looked at the way COMSOL linked then. I have also noted, that if you chain statioanry + transient, and hen do a create default solver sequence, COMOSL set the two up for a coupled analysis, with a stored solution in the middle and transient initial conditions set to the stored solution. What remains often confusing is that you get 2 data set, and the order are not always the one you expext, but by looking at the references one quickly finds out which is the stationary (solved solution) and which is the second transient case
--
Good luck
Ivar
I'mnot behind my COMSOL Ws just now, but previously (and I hope it's still true for v4.2a, I used to load a buckling example, that is (was) a combination of a statioanry analysis + eigenfrequency (set for buckling) and then looked at the way COMSOL linked then. I have also noted, that if you chain statioanry + transient, and hen do a create default solver sequence, COMOSL set the two up for a coupled analysis, with a stored solution in the middle and transient initial conditions set to the stored solution. What remains often confusing is that you get 2 data set, and the order are not always the one you expext, but by looking at the references one quickly finds out which is the stationary (solved solution) and which is the second transient case
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
I think I better understand now.
I think I will create a very simple static-transient study with a very quick solution time. That way I hope to very quickly see whats going on with the results.
I had hoped that you could choose which loads/BC apply to which step, but that does not appear to be true.
For example,
If I have a static-transient study and the static study only looks at a prescribed displacement, it would be nice if I could select that load case (the prescribed displacement) and say ONLY SOLVE FOR STATIONARY. I did originally think that I could do this in the Equation option of each load/BC.
Or am I perhaps missing something.
It would appear that using your idea Ivar I will need to select and deselect BC's/loads for different steps. As in, only select stationary prescribed displacement, solve, deselect stationary displacement, select BC's for transient study, solve.
I will try to get to my Comsol machine later, but most of today is in a deep dark lab in the basement (I think all labs are like this!)
Thanks for keeping on helping me with this, we will get there in the end!
Rob
I think I better understand now.
I think I will create a very simple static-transient study with a very quick solution time. That way I hope to very quickly see whats going on with the results.
I had hoped that you could choose which loads/BC apply to which step, but that does not appear to be true.
For example,
If I have a static-transient study and the static study only looks at a prescribed displacement, it would be nice if I could select that load case (the prescribed displacement) and say ONLY SOLVE FOR STATIONARY. I did originally think that I could do this in the Equation option of each load/BC.
Or am I perhaps missing something.
It would appear that using your idea Ivar I will need to select and deselect BC's/loads for different steps. As in, only select stationary prescribed displacement, solve, deselect stationary displacement, select BC's for transient study, solve.
I will try to get to my Comsol machine later, but most of today is in a deep dark lab in the basement (I think all labs are like this!)
Thanks for keeping on helping me with this, we will get there in the end!
Rob
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Ivar,
You posted this some months ago....
"In V4 you add a stationary study-solving node before your time dependent node, and COMSOL will link them and use the stationary case as initial conditions (if you have tweaked your solver sequence, you should regenerate a new "default" one to get the solver sequence correctly updated by COMSOL.
Furthmore, if you use the variable "t" in some of your BC's or elsewhere, it will not be defined for the stationary solving case. My workaround is to define a Global Parameter t = 0[s] and set it to the first, or any convenient time step value."
I think the last part is the key, the global parameter t=0.
So any BC/load which has "t" will not be used in the static step? That could make life very easy indeed.
Thanks,
Rob
You posted this some months ago....
"In V4 you add a stationary study-solving node before your time dependent node, and COMSOL will link them and use the stationary case as initial conditions (if you have tweaked your solver sequence, you should regenerate a new "default" one to get the solver sequence correctly updated by COMSOL.
Furthmore, if you use the variable "t" in some of your BC's or elsewhere, it will not be defined for the stationary solving case. My workaround is to define a Global Parameter t = 0[s] and set it to the first, or any convenient time step value."
I think the last part is the key, the global parameter t=0.
So any BC/load which has "t" will not be used in the static step? That could make life very easy indeed.
Thanks,
Rob