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.

Help with Microcantilever displacement under load

Please login with a confirmed email address before reporting spam

Hey,

You guys have been very helpful in the past, and I've tried to solve this on my own but I keep hitting an error and can't figure out why. I'm hoping you can offer some advice.

I'm trying to find the spring constant of a microcantilever that is 300um long by 45um wide and .75um thick. Right now I'm not doing anything fancy--just the cantilever and a block it's attached to. I'm applying a 1uN force to the end to find the displacement using a stationary solver.

I've applied boundary conditions, set initial values, and meshed everything. And each time when I click "solve" it works for about 30 seconds to a minute and then errors out saying "Error: Failed to find a solution" and when I click on the actual error it reads:

[Very ill-conditioned preconditioner.
The relative residual is more than 1000 times larger than the relative tolerance.
Returned solution has not converged.]

While I understand that it needs to converge and all that, I can't figure out why it's struggling or where the error is coming in. This is a very basic model that I'm planning to build up into a more realistic one for our research, but I'm not able to get a solution.

I know Comsol can easily solve this; what am I missing?

Thank you in advance for any help; you guys are very much appreciated.


7 Replies Last Post 1 set 2016, 12:17 GMT-4
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 10 feb 2011, 03:11 GMT-5
Hi

first of all get updated to 4.1 with latest patch, really its much smoother than the early 4.0 version. And in v4.0 you need regularly to delete your solver case and rebuild it, as if you edit it a few times it get corrupte (which emans you must also redefine all your postprocessing, this is "repaired " in the newer version)

For your canteliever, why do you block all sides of your large block ? you can just ignore it, your model will be as precise but just simpler to solve. But if you want to see the effet of the base, then block only one side of the base, and you can simulate the influence of the fixation of your device.

You can also, for such simple devices perform a few simplications, and certainly you should check the results w.r.t. an analytical solution, several excellent books give you te formulas to evalaute the modes and stiffness of a cantilever.

Yo ucan also apply a gravity "body load" = -g_const*solid.rho [N/m^3] and check tat your first eigenfrequency

2*pi*fo[Hz] = sqrt(k[N/m]/m[kg])=sqrt(g_const[m/s^2]/maxop1(w)[m])

where maxop1(w) is the maximum tip sag under gravity load in the gravity direction assumed here along "-Z"

its also fun to prove the equation above ;)


--
Good luck
Ivar
Hi first of all get updated to 4.1 with latest patch, really its much smoother than the early 4.0 version. And in v4.0 you need regularly to delete your solver case and rebuild it, as if you edit it a few times it get corrupte (which emans you must also redefine all your postprocessing, this is "repaired " in the newer version) For your canteliever, why do you block all sides of your large block ? you can just ignore it, your model will be as precise but just simpler to solve. But if you want to see the effet of the base, then block only one side of the base, and you can simulate the influence of the fixation of your device. You can also, for such simple devices perform a few simplications, and certainly you should check the results w.r.t. an analytical solution, several excellent books give you te formulas to evalaute the modes and stiffness of a cantilever. Yo ucan also apply a gravity "body load" = -g_const*solid.rho [N/m^3] and check tat your first eigenfrequency 2*pi*fo[Hz] = sqrt(k[N/m]/m[kg])=sqrt(g_const[m/s^2]/maxop1(w)[m]) where maxop1(w) is the maximum tip sag under gravity load in the gravity direction assumed here along "-Z" its also fun to prove the equation above ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 feb 2011, 03:42 GMT-5

Hi

first of all get updated to 4.1 with latest patch, really its much smoother than the early 4.0 version. And in v4.0 you need regularly to delete your solver case and rebuild it, as if you edit it a few times it get corrupte (which emans you must also redefine all your postprocessing, this is "repaired " in the newer version)

Upgrading to 4.1 is on my list of things to do on tomorrow. I'll get that done asap.

For your canteliever, why do you block all sides of your large block ? you can just ignore it, your model will be as precise but just simpler to solve. But if you want to see the effet of the base, then block only one side of the base, and you can simulate the influence of the fixation of your device.

The reason I put the large block in is because what I need to model is how the displacement changes with different types of undercut edges. In our actual research cantilevers the base is far less uniform, so I'm going to make it more complex and asymmetric once I get the basic model working.

I blocked all sides of the large block because I didn't want it going anywhere. No other reason.

You can also, for such simple devices perform a few simplications, and certainly you should check the results w.r.t. an analytical solution, several excellent books give you te formulas to evalaute the modes and stiffness of a cantilever.

Yo ucan also apply a gravity "body load" = -g_const*solid.rho [N/m^3] and check tat your first eigenfrequency

2*pi*fo[Hz] = sqrt(k[N/m]/m[kg])=sqrt(g_const[m/s^2]/maxop1(w)[m])

where maxop1(w) is the maximum tip sag under gravity load in the gravity direction assumed here along "-Z"

its also fun to prove the equation above ;)


--
Good luck
Ivar

I've checked it with some basic equations and Comsol wasn't giving me the same deflection. It was giving me the "error in solver" I described above and then stopping wherever, which did not give the same results.

Did you try to run the solver? Did it error out?
[QUOTE] Hi first of all get updated to 4.1 with latest patch, really its much smoother than the early 4.0 version. And in v4.0 you need regularly to delete your solver case and rebuild it, as if you edit it a few times it get corrupte (which emans you must also redefine all your postprocessing, this is "repaired " in the newer version)[/quote] Upgrading to 4.1 is on my list of things to do on tomorrow. I'll get that done asap. [quote] For your canteliever, why do you block all sides of your large block ? you can just ignore it, your model will be as precise but just simpler to solve. But if you want to see the effet of the base, then block only one side of the base, and you can simulate the influence of the fixation of your device.[/quote] The reason I put the large block in is because what I need to model is how the displacement changes with different types of undercut edges. In our actual research cantilevers the base is far less uniform, so I'm going to make it more complex and asymmetric once I get the basic model working. I blocked all sides of the large block because I didn't want it going anywhere. No other reason. [quote] You can also, for such simple devices perform a few simplications, and certainly you should check the results w.r.t. an analytical solution, several excellent books give you te formulas to evalaute the modes and stiffness of a cantilever. Yo ucan also apply a gravity "body load" = -g_const*solid.rho [N/m^3] and check tat your first eigenfrequency 2*pi*fo[Hz] = sqrt(k[N/m]/m[kg])=sqrt(g_const[m/s^2]/maxop1(w)[m]) where maxop1(w) is the maximum tip sag under gravity load in the gravity direction assumed here along "-Z" its also fun to prove the equation above ;) -- Good luck Ivar [/QUOTE] I've checked it with some basic equations and Comsol wasn't giving me the same deflection. It was giving me the "error in solver" I described above and then stopping wherever, which did not give the same results. Did you try to run the solver? Did it error out?

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 10 feb 2011, 05:37 GMT-5
Hi

no I did not runmit ("my" COMSOL licence is used by a colleage ;) but I ahve made many of those examples and thay have always worked very nicely so I'm sure you must have a little flaw in a BC or in a selection, that is quickly done. But also if you fix all sides of your "fixed" part your material does not have much freedom to behave, that could perhaps overconstrain somewhat the solver (just a guess)

--
Good luck
Ivar
Hi no I did not runmit ("my" COMSOL licence is used by a colleage ;) but I ahve made many of those examples and thay have always worked very nicely so I'm sure you must have a little flaw in a BC or in a selection, that is quickly done. But also if you fix all sides of your "fixed" part your material does not have much freedom to behave, that could perhaps overconstrain somewhat the solver (just a guess) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 feb 2011, 15:48 GMT-5

Hi

no I did not runmit ("my" COMSOL licence is used by a colleage ;) but I ahve made many of those examples and thay have always worked very nicely so I'm sure you must have a little flaw in a BC or in a selection, that is quickly done. But also if you fix all sides of your "fixed" part your material does not have much freedom to behave, that could perhaps overconstrain somewhat the solver (just a guess)

--
Good luck
Ivar

Oddly enough, even though the simulation error'd out every time saying it was unable to converge, it did give a displacement value which, when I checked it against an analytical solution, gave an accurate result.

I don't know what the error was for, but it is working (although it still says it errors out).

Thanks for the advice!
[QUOTE] Hi no I did not runmit ("my" COMSOL licence is used by a colleage ;) but I ahve made many of those examples and thay have always worked very nicely so I'm sure you must have a little flaw in a BC or in a selection, that is quickly done. But also if you fix all sides of your "fixed" part your material does not have much freedom to behave, that could perhaps overconstrain somewhat the solver (just a guess) -- Good luck Ivar [/QUOTE] Oddly enough, even though the simulation error'd out every time saying it was unable to converge, it did give a displacement value which, when I checked it against an analytical solution, gave an accurate result. I don't know what the error was for, but it is working (although it still says it errors out). Thanks for the advice!

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 10 feb 2011, 16:00 GMT-5
Hi

I believe its rather one of the weaknesses in the 4.0, often if you changed a few time yxour solver settings, you need to delete it and rebuild it from scratch as it got corrupt. Really consider to upgrade, you will be less frsutrated, chekc it out with your rep or sysmanager

--
Good luck
Ivar
Hi I believe its rather one of the weaknesses in the 4.0, often if you changed a few time yxour solver settings, you need to delete it and rebuild it from scratch as it got corrupt. Really consider to upgrade, you will be less frsutrated, chekc it out with your rep or sysmanager -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 8 years ago 30 ago 2016, 08:53 GMT-4
hi Ivar,
how do you know which abreviation should be added to the beginning of definition?
I mean like adding ''solid.'' to ''rho''.


cheers,
Shirin
hi Ivar, how do you know which abreviation should be added to the beginning of definition? I mean like adding ''solid.'' to ''rho''. cheers, Shirin

Please login with a confirmed email address before reporting spam

Posted: 8 years ago 1 set 2016, 12:17 GMT-4
Apologies as this was mentioned already, but this error in my experience typically results from the system being mechanically unconstrained: applying a force without sufficient zero-displacement boundaries to prevent acceleration of the entire structure.
Apologies as this was mentioned already, but this error in my experience typically results from the system being mechanically unconstrained: applying a force without sufficient zero-displacement boundaries to prevent acceleration of the entire structure.

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.