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 The information provided may be out of date.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.
Limiting simulations-Dynamic Pull-in Cantilever
Posted 3 mar 2011, 11:58 GMT-5 Structural Mechanics Version 4.3 18 Replies
Please login with a confirmed email address before reporting spam
Hi everyone,
I have a quick question, Can I tell comsol to simulate till it reaches some theoretical value and the stop? I mean, what I'm trying to do is Simulate the dynamic behaviour of a cantilever beam under an applied voltage, so, when I solve the Static model, the coupled Physics(es and solid) work good cause I'm able to reach the Pull-In voltage as the value when the system fails to converge. Theoretically speaking, that's the point were the Electrostatic force equates the Elastic restoration force. Till here everything works logically, however, when I go to stationary study and simulate, there is no reaction, I mean there isn't even a convergence error or something, It just goes through the contact material, so I'm not able to get a Dynamic Pull in Voltage (which should be slightly different to the static). The point is that as comsol is not letting me obtain the dynamic pull in, I-m trying to do the following: Calculate the value of the force at which the system fails to converge in stationary study and then tell comsol in the time-dependent study to simulate just till it reaches that force value. I know it isn't the best approach but if it lts me extrapolate the dyamic pull in I'll live with it. Any help would be greatly appreciated.
Regards
I have a quick question, Can I tell comsol to simulate till it reaches some theoretical value and the stop? I mean, what I'm trying to do is Simulate the dynamic behaviour of a cantilever beam under an applied voltage, so, when I solve the Static model, the coupled Physics(es and solid) work good cause I'm able to reach the Pull-In voltage as the value when the system fails to converge. Theoretically speaking, that's the point were the Electrostatic force equates the Elastic restoration force. Till here everything works logically, however, when I go to stationary study and simulate, there is no reaction, I mean there isn't even a convergence error or something, It just goes through the contact material, so I'm not able to get a Dynamic Pull in Voltage (which should be slightly different to the static). The point is that as comsol is not letting me obtain the dynamic pull in, I-m trying to do the following: Calculate the value of the force at which the system fails to converge in stationary study and then tell comsol in the time-dependent study to simulate just till it reaches that force value. I know it isn't the best approach but if it lts me extrapolate the dyamic pull in I'll live with it. Any help would be greatly appreciated.
Regards
18 Replies Last Post 9 mar 2017, 00:52 GMT-5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
in 3.5a there was a stop cndition, I beleive its also here in V4 but havent checked it out (yet), its most probably in one of the solver sub-nodes, check the doc or help files for "stop condition"
One thing the solver might stop by an error at any point in a parametric sweep, but it mioght not store the last steps, turn on the store all calcualted steps instead of the default predefined steps. But then be aware tht the last or even few last steps might be very wrong (you see the approach of failure of the solver often a few steps in advance)
--
Good luck
Ivar
in 3.5a there was a stop cndition, I beleive its also here in V4 but havent checked it out (yet), its most probably in one of the solver sub-nodes, check the doc or help files for "stop condition"
One thing the solver might stop by an error at any point in a parametric sweep, but it mioght not store the last steps, turn on the store all calcualted steps instead of the default predefined steps. But then be aware tht the last or even few last steps might be very wrong (you see the approach of failure of the solver often a few steps in advance)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Gilbert,
I would also recommend the work-around suggested by Ivar.
By the way, are you actuating the cantilever by a DC voltage or a combination of DC+AC voltage? If it is purely DC, then you may put a stop condition if the tip displacement exceeds say 70% of the initial gap between the two electrodes (since the dynamic pull-in displacement for a cantilever is around 65% of the initial gap).
Regards,
Manish
I would also recommend the work-around suggested by Ivar.
By the way, are you actuating the cantilever by a DC voltage or a combination of DC+AC voltage? If it is purely DC, then you may put a stop condition if the tip displacement exceeds say 70% of the initial gap between the two electrodes (since the dynamic pull-in displacement for a cantilever is around 65% of the initial gap).
Regards,
Manish
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Gilbert,
you can add a stop condition with V4.1 in your Time-Dependent Solver (sub-node of Solver) by right-click -> Stop Condition.
There you can add your condition, let's say "force_max-force_calc". The solver will stop, if this condition will be negative.
Ralf
you can add a stop condition with V4.1 in your Time-Dependent Solver (sub-node of Solver) by right-click -> Stop Condition.
There you can add your condition, let's say "force_max-force_calc". The solver will stop, if this condition will be negative.
Ralf
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Thank you all guys,
I managed to solve the dynamic model combined with the moving mesh and I'm able to see the sudden increase in electrostatic force finally. I tried to solve with the Stop condition (just for curiosity) , but it kept me saying that the expression couldn't be evaluated. From what I know, Comsol stops if the expression goes negative, so I typed something as (40e-9)-v, as v is the deformation or deflection, I tried also with mod1.v, but I got the same message. If anyone knows how to make this one to work I would like to know, just for the sake of curiosity. Ahh, and just two questions that just came to my mind: Is there anyway to use conditioned Forces, like for example, tell comsol to deactivate certain force after certain condition is achieved?, I think I can do it for time with a displaced step multiplying that force, but I was thinking more of other parameters condition. The second question is if you know by chance What where the changes made in 4.1 related to 4.0a, specially in terms of contact pairing? I ask this cause I was not able to simulate contact succesfully in 4.0a. However in 4.1 I tried a simple model and it worked like a charm. Any answers?
Regards
I managed to solve the dynamic model combined with the moving mesh and I'm able to see the sudden increase in electrostatic force finally. I tried to solve with the Stop condition (just for curiosity) , but it kept me saying that the expression couldn't be evaluated. From what I know, Comsol stops if the expression goes negative, so I typed something as (40e-9)-v, as v is the deformation or deflection, I tried also with mod1.v, but I got the same message. If anyone knows how to make this one to work I would like to know, just for the sake of curiosity. Ahh, and just two questions that just came to my mind: Is there anyway to use conditioned Forces, like for example, tell comsol to deactivate certain force after certain condition is achieved?, I think I can do it for time with a displaced step multiplying that force, but I was thinking more of other parameters condition. The second question is if you know by chance What where the changes made in 4.1 related to 4.0a, specially in terms of contact pairing? I ask this cause I was not able to simulate contact succesfully in 4.0a. However in 4.1 I tried a simple model and it worked like a charm. Any answers?
Regards
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
check the release notes, v4.0 was really not ironed out enough, v4.1 is far better. I expect an even as important jump with v4.2 soon here I have heard, with this I believe we can say the v4 transition is mainly done, and we can get back to business again, but still, I need to verify everything in v4.2 when it arrives before I really let it go for your tough projects
--
Good luck
Ivar
check the release notes, v4.0 was really not ironed out enough, v4.1 is far better. I expect an even as important jump with v4.2 soon here I have heard, with this I believe we can say the v4 transition is mainly done, and we can get back to business again, but still, I need to verify everything in v4.2 when it arrives before I really let it go for your tough projects
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Ivar,
I am Rakesh,also working on the same but I have completed static as well as dynamic still I want to do damping(viscous damping) analysis.Can u guide me how can I do it in v3.5.I am thinking in this way I can put the beam in some fluid(viscous).In this regard I am unable to do so.Help me out how can I model it.
I am Rakesh,also working on the same but I have completed static as well as dynamic still I want to do damping(viscous damping) analysis.Can u guide me how can I do it in v3.5.I am thinking in this way I can put the beam in some fluid(viscous).In this regard I am unable to do so.Help me out how can I model it.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
coupling in a fluid is not trivial for the eigenfrequency analysis, as there are the damping and the added mass (of the fluid) and COMSOL does not seem to have yet a coupled physics set up. It's probably difficult to do in all generality.
I'm neither not sure what is the easiest way, I have been looking into this for some time.
As reference I find the book of Axsia very interesting (fluid structure interactions) but I have not yet found enough time to implement a few examples
Vol3 Modelling of Mechanical Systems: Fluid-Structure Interaction, F. Axisa & J. Antunes, 2007, B&H-Elsevier, ISBN 978-0 750-66847-7
--
Good luck
Ivar
coupling in a fluid is not trivial for the eigenfrequency analysis, as there are the damping and the added mass (of the fluid) and COMSOL does not seem to have yet a coupled physics set up. It's probably difficult to do in all generality.
I'm neither not sure what is the easiest way, I have been looking into this for some time.
As reference I find the book of Axsia very interesting (fluid structure interactions) but I have not yet found enough time to implement a few examples
Vol3 Modelling of Mechanical Systems: Fluid-Structure Interaction, F. Axisa & J. Antunes, 2007, B&H-Elsevier, ISBN 978-0 750-66847-7
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
I will be working with MEMS shunt switch using Comsol 4.3, whether Comsol can simulate the pull in voltage with dynamic process, and hysteresis behavior like what CoventorWare did?
Thanks.
I will be working with MEMS shunt switch using Comsol 4.3, whether Comsol can simulate the pull in voltage with dynamic process, and hysteresis behavior like what CoventorWare did?
Thanks.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I do not know the specificities of Coventor, but from my understanding COMSOL can do about everything, but it does not have specific scripts for specific type of models, only for general physics (of specific type). Therefore, you need to know exactly what Coventor is doing as BC sequence etc, and then do the same with COMSOL
The advantage is then that you do know what your are really modelling, something one easily forget with "black box solutions"
--
Good luck
Ivar
I do not know the specificities of Coventor, but from my understanding COMSOL can do about everything, but it does not have specific scripts for specific type of models, only for general physics (of specific type). Therefore, you need to know exactly what Coventor is doing as BC sequence etc, and then do the same with COMSOL
The advantage is then that you do know what your are really modelling, something one easily forget with "black box solutions"
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
HI, Ivar,
My questions are almost similar to Mr. Gilbert above. I am new to Comsol, I have a RF MEMS shunt switch which has gap distance of 1.5um and pull in at 5V, simulated under Coventor. However when I model it with Comsol (Trial copy provided by imath-asia), it takes 12V to move the switch by 0.475um, when I further increase the voltage level, Comsol turn out error(no stable solution). I know there is something wrong with the multiphysics setup. May I know which modules and which solver are required to make my simulation successful?
By the way, I read some Comsol documents and some thesis paper found that MEMS devices pull in at 2/3 of gap distance, is that the error “No stable solution” means the switch already been actuated? I wish to simulate its dynamic behavior until the switch makes contact with bottom electrode. Would you provide me info on how to achieve that using Comsol 4.3?
Thanks.
My questions are almost similar to Mr. Gilbert above. I am new to Comsol, I have a RF MEMS shunt switch which has gap distance of 1.5um and pull in at 5V, simulated under Coventor. However when I model it with Comsol (Trial copy provided by imath-asia), it takes 12V to move the switch by 0.475um, when I further increase the voltage level, Comsol turn out error(no stable solution). I know there is something wrong with the multiphysics setup. May I know which modules and which solver are required to make my simulation successful?
By the way, I read some Comsol documents and some thesis paper found that MEMS devices pull in at 2/3 of gap distance, is that the error “No stable solution” means the switch already been actuated? I wish to simulate its dynamic behavior until the switch makes contact with bottom electrode. Would you provide me info on how to achieve that using Comsol 4.3?
Thanks.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
it could be indicating you have pulled in, but it depends also on your solver and the steps used, and how you set-up your model. COMSOL is probably less "black box" than Coventor, so you need to understand a bit more of what is really happening, at least for the first study, to get good convergence
--
Good luck
Ivar
it could be indicating you have pulled in, but it depends also on your solver and the steps used, and how you set-up your model. COMSOL is probably less "black box" than Coventor, so you need to understand a bit more of what is really happening, at least for the first study, to get good convergence
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
dear Ivar
It will be very helpful if the tutorials for after pull in effects are made available. Dynamic pull in effect is also very important. Amplitude dependent frequency softening of MEMS beams under electrostatic actuation is also missing till now.
Thanks
Saurabh
It will be very helpful if the tutorials for after pull in effects are made available. Dynamic pull in effect is also very important. Amplitude dependent frequency softening of MEMS beams under electrostatic actuation is also missing till now.
Thanks
Saurabh
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Saurabh
I am also using COMSOL to simulate MEMS device after pull-in and encounter lots of problems.
Have you found some methods to simulate these behaviors by COMSOL?
I am also using COMSOL to simulate MEMS device after pull-in and encounter lots of problems.
Have you found some methods to simulate these behaviors by COMSOL?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Contact in comsol has issues, and all other commercial softwares also have the same issues. Since simulation after pull in requires contact, Comsol is ineffective.
I have my own code which does beam contact problem without using contact mechanics explicitly, and that works well.
WE can collaborate! if you want.
I have my own code which does beam contact problem without using contact mechanics explicitly, and that works well.
WE can collaborate! if you want.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Saurabh,
Thanks for your information. Surely, I am glad to collaborate with you.
I have tried to simulate the pull-in phenomenon with contact analysis in geometry assembly mode. The electrostatics force in assembly mode should be calculated by equations . But the solution is not converged if the bias voltage approachs to pull-in voltage.
What is your code developed for? COMSOL or some other FEM tools?
Thanks for your information. Surely, I am glad to collaborate with you.
I have tried to simulate the pull-in phenomenon with contact analysis in geometry assembly mode. The electrostatics force in assembly mode should be calculated by equations . But the solution is not converged if the bias voltage approachs to pull-in voltage.
What is your code developed for? COMSOL or some other FEM tools?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Actually, the beam falls beyond the substrate without detecting the contact. You can define contact surface of the substrate so that contact forces are applied whenever pull in occurs and the beam does not penetrates the substrate. Please revert if defining contact works.
As I mentioned, I have tried in different commercial software and the contact is a problem in all of those software.
I am sorry I cannot publicly discuss the scheme I devised to simulate results after pull in. It is a FEM technique , so it will work in any FEM software. Wait till we publish the results!
As I mentioned, I have tried in different commercial software and the contact is a problem in all of those software.
I am sorry I cannot publicly discuss the scheme I devised to simulate results after pull in. It is a FEM technique , so it will work in any FEM software. Wait till we publish the results!
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Yes. I have defined the contact pairs and they work well in the case that external force applied on the boundary.
However in the case that using the electrostatics force in EMI interface the solution couldnot converged.
This problem has been confirmed by COMSOL support. As a result, I have to apply the external force to realize the contact shape of a plate which should be caused by electrostatics force.
In additional, I found that some variables could not be gotten in assembly mode by COMSOL, since these variables in the adjacent boundaries were counteracted automatically. I think these cases should be corrected by COMSOL because these variables coupling menus are selectable but they don't work in the assembly mode.
However in the case that using the electrostatics force in EMI interface the solution couldnot converged.
This problem has been confirmed by COMSOL support. As a result, I have to apply the external force to realize the contact shape of a plate which should be caused by electrostatics force.
In additional, I found that some variables could not be gotten in assembly mode by COMSOL, since these variables in the adjacent boundaries were counteracted automatically. I think these cases should be corrected by COMSOL because these variables coupling menus are selectable but they don't work in the assembly mode.
Please login with a confirmed email address before reporting spam
Posted:
7 years ago
Hello sir,
I have been trying to model the electrostatically actuated bistable beam but unable to find it. As I also encountered same contact detection problem . I tried to use the contact pair, but this disables the boundaries for application of voltage.
Please help.
Regards,
Vasudha
I have been trying to model the electrostatically actuated bistable beam but unable to find it. As I also encountered same contact detection problem . I tried to use the contact pair, but this disables the boundaries for application of voltage.
Please help.
Regards,
Vasudha