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 The information provided may be out of date.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.

Why the initial value from previous solution doesn’t work?

Please login with a confirmed email address before reporting spam

Hi, everybody,
I have a simple cube, it has a temperature T1. It undergoes two stages, first it will cool down to T2, then in the second stage it will continue to cool to T3. I want to get the final deformation when it cools to T3.
What I did is as follows:
1. I built the geometry model, used one solid mechanics, and used the linear elastic material model, under which I included a thermal expansion node. I set the reference temperature to T1, cool temperature set to T2.
2. I added two studies, each has a time dependent step.
3. In the first study, I compute the solution(deformation) when the cube cools down from T1 to T2
4. In the second study, I first change the temperature setting in the thermal expansion node in the solid mechanics, I set the reference temperature to T2, cool temperature set to T3. In the solver configuration of the second study, solver configurations -- > solver 1 -- > dependent variables -- > initial values of variables solved for and values of variables not solved for: I set the Method to Solution, and solution field to : solution 1 (using the solution from study 1), then compute and get the final deformation 1.

However, I did another trial, I built the same model, and same solid mechanics, but with only one study. In the thermal expansion node under the linear elastic material model of solid mechanics, I directly set the reference temperature to T2, and the cool temperature to T3, then I computed the study, and it got the final deformation 2, which is exactly the same as deformation 1 mentioned above. It seems that the initial values setting in study 2 which uses solution 1 in the first study don’t work at all. I was confused, why the solution from study 1 did not have any effects on the solution of study 2? Can anybody give me some advice? Thanks in advance!

5 Replies Last Post 29 juil. 2012, 00:12 UTC−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 27 juil. 2012, 04:27 UTC−4
Hi

Indeed your explanation seem logical and the result somewhat suspect. Now for us out here to better follow, it would be nice to know the version you are using and the physics (I like to state the physics tags such as solid, HT or TS ... as this tells us the dependent variables in play)

But, there is still ne point I'm not sure about, and that would have to be checked, but I'm not by my COMSOL WS for a few days. When you cool down your material, you build up some stress (if you part is constrained), when you cool it down the second step, I'm not sure that the stress form the first cooldown is fully ported over to the next solver step, and without this your deformations might not be the same for a single or a two step cooling.

You can check this by plotting the state of your material domain at the initial conditions level of the second solver step (right click the dependent variables solver sub-node and run a "compute to here")

Another way is to run a TS case with an imposed domain temperature that you "sweep" from T1 to T3 via T2

--
Good luck
Ivar
Hi Indeed your explanation seem logical and the result somewhat suspect. Now for us out here to better follow, it would be nice to know the version you are using and the physics (I like to state the physics tags such as solid, HT or TS ... as this tells us the dependent variables in play) But, there is still ne point I'm not sure about, and that would have to be checked, but I'm not by my COMSOL WS for a few days. When you cool down your material, you build up some stress (if you part is constrained), when you cool it down the second step, I'm not sure that the stress form the first cooldown is fully ported over to the next solver step, and without this your deformations might not be the same for a single or a two step cooling. You can check this by plotting the state of your material domain at the initial conditions level of the second solver step (right click the dependent variables solver sub-node and run a "compute to here") Another way is to run a TS case with an imposed domain temperature that you "sweep" from T1 to T3 via T2 -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 juil. 2012, 13:39 UTC−4
Thanks, Ivar,
I am using the Comsol 4.2, the physics tag is solid mechanics (solid), which has only one dependent variable: mod1_u,

Just as you advised, I tried to include the stress generated from the first study into the second study. After I run the first study with thermal expansion from T1 to T2, I update the reference temperature to T2, the cool temperature to T3. Besides, I added a “initial stress and strain” node under the linear elastic material model of solid mechanics, in the initial stress and strain table S0, I entered the stress as : solid.sx, solid,sy, solid.sz, and solid.sxy....,
Then in the second study solver configuration, I used the same setting as before (solver configurations -- > solver 1 -- > dependent variables -- > initial values of variables solved for and values of variables not solved for: I set the Method to Solution, and solution field to : solution 1 (using the solution from study 1))

Then I computed the second study, but got an error message says “Circular variable dependency detected: solid.sx.” , so the way I include the initial stress from the first study is not correct. I wonder how to include the stress generated from the first study into the second study?

Since in the solver configuration setting of the second study, I set the initial values solved for and not solved for to use the solution from study 1, what does the solution from study one contain? Does it contain results of stress, or ....?? I assume it already contain all the results calculated from the first study, including the stress and displacement, is that right? Thanks!
Thanks, Ivar, I am using the Comsol 4.2, the physics tag is solid mechanics (solid), which has only one dependent variable: mod1_u, Just as you advised, I tried to include the stress generated from the first study into the second study. After I run the first study with thermal expansion from T1 to T2, I update the reference temperature to T2, the cool temperature to T3. Besides, I added a “initial stress and strain” node under the linear elastic material model of solid mechanics, in the initial stress and strain table S0, I entered the stress as : solid.sx, solid,sy, solid.sz, and solid.sxy...., Then in the second study solver configuration, I used the same setting as before (solver configurations -- > solver 1 -- > dependent variables -- > initial values of variables solved for and values of variables not solved for: I set the Method to Solution, and solution field to : solution 1 (using the solution from study 1)) Then I computed the second study, but got an error message says “Circular variable dependency detected: solid.sx.” , so the way I include the initial stress from the first study is not correct. I wonder how to include the stress generated from the first study into the second study? Since in the solver configuration setting of the second study, I set the initial values solved for and not solved for to use the solution from study 1, what does the solution from study one contain? Does it contain results of stress, or ....?? I assume it already contain all the results calculated from the first study, including the stress and displacement, is that right? Thanks!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 juil. 2012, 14:02 UTC−4
Hi, Ivar,
I check the initial condition of the material domain by plotting its state, yes, the plot does show the solution results from the first study, but why after computing the second study, the final results turn to be the one without including the initial values from the first study? Since in the comparison trial, I only use one study, with the temperature setting in the thermal expansion node the same as in the second study (reference temperature is T2, cool to T3), and get the same results as in the two study trial. I was confused. Thanks.
Hi, Ivar, I check the initial condition of the material domain by plotting its state, yes, the plot does show the solution results from the first study, but why after computing the second study, the final results turn to be the one without including the initial values from the first study? Since in the comparison trial, I only use one study, with the temperature setting in the thermal expansion node the same as in the second study (reference temperature is T2, cool to T3), and get the same results as in the two study trial. I was confused. 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 27 juil. 2012, 14:59 UTC−4
Hi

As I'm out of office and a few tousand km from my COMSOL WS I cannot test it out myself ;)
indeed if you redefine an stress node by the current variabes solid.sx etc you will get a circular vaiable error, but as you state I'm neither not sure the stress is fully coupled, but is this the true or only issue ? I cant tell like that.
So I believe we are both missing some simple setting here, but not sure what it is, perhaps someone else has some good hints ;)

--
Good luck
Ivar
Hi As I'm out of office and a few tousand km from my COMSOL WS I cannot test it out myself ;) indeed if you redefine an stress node by the current variabes solid.sx etc you will get a circular vaiable error, but as you state I'm neither not sure the stress is fully coupled, but is this the true or only issue ? I cant tell like that. So I believe we are both missing some simple setting here, but not sure what it is, perhaps someone else has some good hints ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 juil. 2012, 00:12 UTC−4
Hi, Ivar,
I also think I missed some simple settings, but I don't know how. Can anybody help?

By the way: to verify my problem, I did another trial, using the “layered_plate” model in the Structural_Mechanics_Module model library.

What I did is as follows:
1. Make a copy of file “layered_plate.mph”, naming it “layered_plate 1.mph”, in the file “layered_plate 1.mph”, I deleted the physics “Two Layers (solid)” , deleted “Initial stress and strain 1” node under the Linear elastic material model of the “Three Layers (solid)”, and deleted “Step1: Stationary” node and “Solver Configurations” node under the “Study 1”. Then I compute the study 1 and in the “Stress (solid2)” node under the results root, I plot the total displacement using the expression: solid2.disp. the final deformation shape is denoted as deformation 1.
2. In the original file “layered_plate.mph”, I plot the same total displacement using the expression: solid2.disp in the “Stress (solid 2)”, the final deformation is denoted as deformation 2.
3. Compare deformation 1 and 2, they are exactly the same!!

I wonder why there isn’t any difference, since in the original file “layered_plate.mph”, the physics Three layers (solid2) is based on the physics Two layers (solid2). Why the result of physics Two layers (solid2) don’t have any effects on the final deformation of physics Three layers (solid2)??

Can anybody give me some explanations? Thanks!
Hi, Ivar, I also think I missed some simple settings, but I don't know how. Can anybody help? By the way: to verify my problem, I did another trial, using the “layered_plate” model in the Structural_Mechanics_Module model library. What I did is as follows: 1. Make a copy of file “layered_plate.mph”, naming it “layered_plate 1.mph”, in the file “layered_plate 1.mph”, I deleted the physics “Two Layers (solid)” , deleted “Initial stress and strain 1” node under the Linear elastic material model of the “Three Layers (solid)”, and deleted “Step1: Stationary” node and “Solver Configurations” node under the “Study 1”. Then I compute the study 1 and in the “Stress (solid2)” node under the results root, I plot the total displacement using the expression: solid2.disp. the final deformation shape is denoted as deformation 1. 2. In the original file “layered_plate.mph”, I plot the same total displacement using the expression: solid2.disp in the “Stress (solid 2)”, the final deformation is denoted as deformation 2. 3. Compare deformation 1 and 2, they are exactly the same!! I wonder why there isn’t any difference, since in the original file “layered_plate.mph”, the physics Three layers (solid2) is based on the physics Two layers (solid2). Why the result of physics Two layers (solid2) don’t have any effects on the final deformation of physics Three layers (solid2)?? Can anybody give me some explanations? Thanks!

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.