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 The information provided may be out of date.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.

Plasticity problems

Please login with a confirmed email address before reporting spam

Good evening everybody,

I've got some problems with my solid mechanics module. I use a plasticity node with hardening function. I have two study steps. In the first study step a temperature distribution is computed. During the second study step the thermal stress shall be computed. For the second study step, the settings in dependent variable node are set to "Inital Value" from the previous study step.
So my problem is the following. When I do not add initial values under the plasticity node different from zero I get the following error:

Tried to evaluate negative power of zero.
Function ^

Could not compute expression:
d(sqrt(2*(d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SX)^2+0.5*d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SXY)^2+d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SY)^2+d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SZ)^2)/3),mod1.solid.epX)d(sqrt(2*(d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SX)^2+0.5*d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SXY)^2+d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SY)^2+d(1.5*((mod1.solid.SX-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SXY^2+2*mod1.solid.SXZ^2+(mod1.solid.SY-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2+2*mod1.solid.SYZ^2+(mod1.solid.SZ-(mod1.solid.SX+mod1.solid.SY+mod1.solid.SZ)/3)^2),mod1.solid.SZ)^2)/3),mod1.solid.epX)

I use the german version, so the translation may not be exactly the same what Comsol shows in the english version.
To avoid this error I added an initial strain of 1e-10 to the plasticity node. The first study step works quite well with it, but the second study step fails to be computed with an error message, that the elasto plastic variables could not be computed.

Does anyone have any suggestions how I can avoid these errors?

5 Replies Last Post 28.02.2012, 09:54 GMT-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 28.02.2012, 02:28 GMT-5
Hi

these are tricky error NaNs and 1/0^n are not well accepted.
But often they come from number under/overlow elsewhere such as too large differences in scales,
or also mesh sampling issues, do you have /expect high gradients in certain regions ? and have you used a fine ehough mesh in these regions ? (not everywhere).

I add often boundary mesh in critical regions, but this all depends on the model and physics

--
Good luck
Ivar
Hi these are tricky error NaNs and 1/0^n are not well accepted. But often they come from number under/overlow elsewhere such as too large differences in scales, or also mesh sampling issues, do you have /expect high gradients in certain regions ? and have you used a fine ehough mesh in these regions ? (not everywhere). I add often boundary mesh in critical regions, but this all depends on the model and physics -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.02.2012, 04:00 GMT-5
Good morning,

I am expecting high gradients due to the hight gradients of the temperature distribution, but refining the mesh did not solve the problem.. Interesting thing is the error message arrises at the very beginning of the solving process if initial values for plasticity are zero. If they are slightly different to zero does its work, but then at the start of the second study step I get the couldn't compute elasto plastic strain variables error.
Good morning, I am expecting high gradients due to the hight gradients of the temperature distribution, but refining the mesh did not solve the problem.. Interesting thing is the error message arrises at the very beginning of the solving process if initial values for plasticity are zero. If they are slightly different to zero does its work, but then at the start of the second study step I get the couldn't compute elasto plastic strain variables error.

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 28.02.2012, 07:57 GMT-5
Hi

nextt step is cetainly to consider initial values, all "0" is a, but often not the best, starting point.

what about setting some initial conditions closer to what you expect for the BC starting values.
Or to ensure your BC loads start at corresponding "0" initialconditions and then build up as a step response with a smooth transition. But then one must consider if the inertial terms should be kept or not, or in fact do you need a stationary parametric load change instead of a true transient response ?

Do you want the dynamic response or the effect of a time varying laod (time varying slowly w.r.t. danmic solid response, or any acoustic or shock wave response ?

--
Good luck
Ivar
Hi nextt step is cetainly to consider initial values, all "0" is a, but often not the best, starting point. what about setting some initial conditions closer to what you expect for the BC starting values. Or to ensure your BC loads start at corresponding "0" initialconditions and then build up as a step response with a smooth transition. But then one must consider if the inertial terms should be kept or not, or in fact do you need a stationary parametric load change instead of a true transient response ? Do you want the dynamic response or the effect of a time varying laod (time varying slowly w.r.t. danmic solid response, or any acoustic or shock wave response ? -- Good luck Ivar

Nagi Elabbasi Facebook Reality Labs

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.02.2012, 09:19 GMT-5
If you post your model maybe someone on the Forum can provide more specific help. I have one more suggestion though in addition to what Ivar mentioned. For the first study step disable the Solid Mechanics physics. That way COMSOL will only solve the thermal problem, and not spend time (and error checking) on the plasticity.

Nagi Elabbasi
Veryst Engineering
If you post your model maybe someone on the Forum can provide more specific help. I have one more suggestion though in addition to what Ivar mentioned. For the first study step disable the Solid Mechanics physics. That way COMSOL will only solve the thermal problem, and not spend time (and error checking) on the plasticity. Nagi Elabbasi Veryst Engineering

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28.02.2012, 09:54 GMT-5
I'm not sure, if I am allowed to post the model. I tried the suggestion of Nagi Elabbasi and deactivated the solid mechanics module in the first study step physics selection interface. Courisly the same error message is shown. If I deactivate the whole solid mechanics node, the error message does not appear, but then I cannot use it during the second study step.

@Ivar:
I am not sure, if I really understand what you mean. The only loads I have are temperature induced deformations from a gaussian heat source.
I'm not sure, if I am allowed to post the model. I tried the suggestion of Nagi Elabbasi and deactivated the solid mechanics module in the first study step physics selection interface. Courisly the same error message is shown. If I deactivate the whole solid mechanics node, the error message does not appear, but then I cannot use it during the second study step. @Ivar: I am not sure, if I really understand what you mean. The only loads I have are temperature induced deformations from a gaussian heat source.

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.