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.

Errors, errors, errors...??

Please login with a confirmed email address before reporting spam

Hi everyone

I'm trying to build a groundwater model using matlab and I have two problems so far and I would appreciate any help

1) After I build all the features of my model I run the geomcsg command. For some reason sometimes the command works fine and for the same model sometimes get the following error "Could not resolve subdomain relations". Is there any explanation for that? Does the command geomcsg depends on any random "seed" so that sometimes works and other not?

2) After i build the model when I use very coarse mesh (i.e. meshinit(fem,'hauto', 10)) the problem can be solved, however when I use finer mesh I receive the message

"Singular Matrix"
and also
"there are 16 void equations for the variable p at coordinates (......)"
I've read in an other forum that if the minimum element quality(MEQ) is less than 0.05 this may happen. However in my model for hauto=10 ->MEQ=0.07 and worked. For hauto=1(which corresponds to the finest mesh) ->MEQ=0.1994 and get the above msg.
Any help on that??


Some more info about the model: is just a steady state groundwater flow with quite complex geometry

Thank you

5 Replies Last Post 7 mars 2011, 07:15 UTC−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 3 juin 2010, 02:37 UTC−4
Hi

I do not believe the mesh quality as such is giving singular matrix, but there can be some related effects.

I would rather suspect your geometry to have some voids or very fine features. Then when the mesh is too coarse you might get overlapping of parts (I have had a few examples where the coarse meshing on cylindrical parts "soldered the parts together" as the approximation of a circel by a polygone makes some shortcuts and that linked the concentric cylinders, not respecting some thin gaps I had. As this is often in the middle of your model these kind of errors are not easy to detect, before you analyse the funny results you get (after a couple of hours calculateions)

So I would suggest to study in detail your geometry and meshing graphically, why not import it into the COMSOL GUI and observe it therein ?

Good luck
Ivar
Hi I do not believe the mesh quality as such is giving singular matrix, but there can be some related effects. I would rather suspect your geometry to have some voids or very fine features. Then when the mesh is too coarse you might get overlapping of parts (I have had a few examples where the coarse meshing on cylindrical parts "soldered the parts together" as the approximation of a circel by a polygone makes some shortcuts and that linked the concentric cylinders, not respecting some thin gaps I had. As this is often in the middle of your model these kind of errors are not easy to detect, before you analyse the funny results you get (after a couple of hours calculateions) So I would suggest to study in detail your geometry and meshing graphically, why not import it into the COMSOL GUI and observe it therein ? Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 juin 2010, 16:49 UTC−4
Thank you Ivar for your reply.
It seems that I solved this for my case. All I had to do, is to assign resonable initial conditions, although my problem is steady state.
Thank you Ivar for your reply. It seems that I solved this for my case. All I had to do, is to assign resonable initial conditions, although my problem is steady state.

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 21 juin 2010, 13:51 UTC−4
Hi

Certainly, choosing a good starting point, helps a lot on convergence, and time to results ;)

Usually if you want to travel, it helps heading in the right direction, OK 40[Mm] further straigh ahead and you end up more or less at the same place, but that is a long travel, no ?

But I must admit that I mostly do not bother define initial conditions, oters than "zero", Comsol usually finds them rapidly, but not always

Have fun Comsoling
Ivar
Hi Certainly, choosing a good starting point, helps a lot on convergence, and time to results ;) Usually if you want to travel, it helps heading in the right direction, OK 40[Mm] further straigh ahead and you end up more or less at the same place, but that is a long travel, no ? But I must admit that I mostly do not bother define initial conditions, oters than "zero", Comsol usually finds them rapidly, but not always Have fun Comsoling Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 7 mars 2011, 06:35 UTC−5
hi
wath is "the COMSOL GUI "?

thanks
hi wath is "the COMSOL GUI "? thanks

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 7 mars 2011, 07:15 UTC−5
Hi

GUI = en.wikipedia.org/wiki/GUI

The graphical User Interface Window (often refers to the fill-in fields for variables in COMSOL graphics interface, is also used to oppose the line (punch card type, for those who remember what that was) input of a matlab script

--
Good luck
Ivar
Hi GUI = http://en.wikipedia.org/wiki/GUI The graphical User Interface Window (often refers to the fill-in fields for variables in COMSOL graphics interface, is also used to oppose the line (punch card type, for those who remember what that was) input of a matlab script -- Good luck Ivar

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.