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 The information provided may be out of date.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.

"maximum number of segregated iterations reached", soln not converged

Please login with a confirmed email address before reporting spam

Can someone give me a tip on this.

I have an axisymmetric model that runs fine on 3.5a (attached), no issues.

However, when I re-create this same model in 4.2 (as best I can re-create it, with a contact pair), there is a problem getting it to run for the same level of prescribed z-displacement of the "pin".

The 3.5a model allows the pin to be displaced downward (-ve z) by -0.15 mm, and the model runs fine. There is a contact pair along the interface of the pin and the medium. But in 4.2, I can only get a solution for pin displacement of up to about -0.01 mm in the z-direction. Beyond that value of -0.01mm, I get the attached error message about "maximum number of segregated iterations reached". I used default solvers etc. in 3.5a -- nothing special. Is there some adjustment I should be making to the solvers in 4.2 to get this to run?

Thanks very much.


8 Replies Last Post 2011年12月7日 GMT-5 17:10
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 2011年11月29日 GMT-5 02:21
Hi

I cannot even open your files, I get the message "Boolean geometry Operation failed" ;)
so something must be funny therein. (I have 4.2a)

--
Good luck
Ivar
Hi I cannot even open your files, I get the message "Boolean geometry Operation failed" ;) so something must be funny therein. (I have 4.2a) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年11月29日 GMT-5 13:00
Thanks for taking a look, Ivar.

Try the attached model. This model runs to a solution (after about 11 iterations) for a negative z displacement of 0.02mm, but then starts to give non-convergence/no solution-type messages if the prescribed z displacement is increased to -0.025 mm and beyond -- which seems pretty odd to me at least. As I mentioned, this problem was able to run fine in 3.5a with a prescribed negative z displacement of -0.150 mm. I'm perplexed at what I am messing up in replicating the problem within 4.2.

Your comment about the "Boolean geometry operation failing...." is interesting. Maybe the following is a related issue: I initially created the contact pair 1 by doing an assembly of the pin (having the two ridges) together with the surrounding material (which was my method in 3.5a). The contact pair showed up under the "Model 1 definitions" list, which seemed fine.

However, I also (evidently) needed to define the same contact pair 1 within the "solid mechanics" listing, and in any case, this is the way the model is running right now. But on this point, I'm not fully clear why the boundaries making up contact pair 1 are also appearing within the "Free 1" boundaries under the solid mechanics list. (Terminology-wise, I'm not sure why boundaries involved in the contact pair 1 would also be listed as "free".)

Anyway, thanks for seeing if you can find the glitch.

(I do not yet have 4.2a installed due to a short delay in license delivery within Stanford's system.)
Thanks for taking a look, Ivar. Try the attached model. This model runs to a solution (after about 11 iterations) for a negative z displacement of 0.02mm, but then starts to give non-convergence/no solution-type messages if the prescribed z displacement is increased to -0.025 mm and beyond -- which seems pretty odd to me at least. As I mentioned, this problem was able to run fine in 3.5a with a prescribed negative z displacement of -0.150 mm. I'm perplexed at what I am messing up in replicating the problem within 4.2. Your comment about the "Boolean geometry operation failing...." is interesting. Maybe the following is a related issue: I initially created the contact pair 1 by doing an assembly of the pin (having the two ridges) together with the surrounding material (which was my method in 3.5a). The contact pair showed up under the "Model 1 definitions" list, which seemed fine. However, I also (evidently) needed to define the same contact pair 1 within the "solid mechanics" listing, and in any case, this is the way the model is running right now. But on this point, I'm not fully clear why the boundaries making up contact pair 1 are also appearing within the "Free 1" boundaries under the solid mechanics list. (Terminology-wise, I'm not sure why boundaries involved in the contact pair 1 would also be listed as "free".) Anyway, thanks for seeing if you can find the glitch. (I do not yet have 4.2a installed due to a short delay in license delivery within Stanford's system.)


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 2011年11月30日 GMT-5 12:04
Hi

first of all your uploaded file appears as "0 bytes" when I try to fetch it, probbaly there was an issue with the site the other day. Try up loading it again.

2nd, I do not trust the translations, specially with contact pairs, I always recreate my models, this gives me the opportunity to learn the "new way" too. The fact that you can copy past from one to the other module helps, apart for the new variables names to adapt ;)

--
Good luck
Ivar
Hi first of all your uploaded file appears as "0 bytes" when I try to fetch it, probbaly there was an issue with the site the other day. Try up loading it again. 2nd, I do not trust the translations, specially with contact pairs, I always recreate my models, this gives me the opportunity to learn the "new way" too. The fact that you can copy past from one to the other module helps, apart for the new variables names to adapt ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年11月30日 GMT-5 12:31
Thanks. The model I attached most recently (in the second posting) is the one I developed new in 4.2 (as opposed to bringing it in from 3.5a).
Thanks. The model I attached most recently (in the second posting) is the one I developed new in 4.2 (as opposed to bringing it in from 3.5a).

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 2011年11月30日 GMT-5 13:21
Hi

but the latest one has a size of "0" bytes, you need to reload it ;)

--
Good luck
Ivar
Hi but the latest one has a size of "0" bytes, you need to reload it ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年11月30日 GMT-5 18:01
OK, try this one.
OK, try this one.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月6日 GMT-5 13:08
Hi Ivar:

Any luck downloading that last file?

Maybe the problem could be resolved with a coupled solver? Or perhaps a step by step increase in displacement, using a parameter?

Thanks,

John
Hi Ivar: Any luck downloading that last file? Maybe the problem could be resolved with a coupled solver? Or perhaps a step by step increase in displacement, using a parameter? Thanks, John

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月7日 GMT-5 17:10
Ivar:

I seemed to have resolved the issue I originally asked you about.

If you look at the two attached images, one shows a contact pair along the interface only, and the other shows a contact pair that extends above when the "pin" exits out of the surrounding material. I'm not sure why my assembly step in 4.2 created the latter type of contact pair, but that seems to be the problem. I say that because when I used the geometry and contact pair as created in Comsol 3.5a (first image attached), the model ran OK in 4.2 -- although only with a fully coupled solver, not the segregated solver. (This 4.2 solution agreed with what I had found in Comsol 3.5a previously.)

In any case, thanks for the tips. I learned a lot by looking thru a lot of the discussion pages. For example, along the way I tried various other "solutions" to this issue, including parametric sweeps -- external as well as continuation sweeps. But these only worked for small displacements of the pin into the surrounding material.
Ivar: I seemed to have resolved the issue I originally asked you about. If you look at the two attached images, one shows a contact pair along the interface only, and the other shows a contact pair that extends above when the "pin" exits out of the surrounding material. I'm not sure why my assembly step in 4.2 created the latter type of contact pair, but that seems to be the problem. I say that because when I used the geometry and contact pair as created in Comsol 3.5a (first image attached), the model ran OK in 4.2 -- although only with a fully coupled solver, not the segregated solver. (This 4.2 solution agreed with what I had found in Comsol 3.5a previously.) In any case, thanks for the tips. I learned a lot by looking thru a lot of the discussion pages. For example, along the way I tried various other "solutions" to this issue, including parametric sweeps -- external as well as continuation sweeps. But these only worked for small displacements of the pin into the surrounding material.

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.