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 The information provided may be out of date.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.
Undefined value found Solvermodel
Posted 15 mars 2011, 11:52 UTC−4 Structural Mechanics Version 3.5a, Version 4.1 6 Replies
Please login with a confirmed email address before reporting spam
Hello,
I am trying to model a piece of elastomer subject to a force, to compute its deformation using comsol 3.5a.
The 3D model is extremely simple, and just to see I don't make mistakes in future more complex ones.
I defined the following constants:
thickness
side
xzero = yzero = side/2
six = siy = side/15
Force
and used some of them to define the global expression (a 2D gaussian)
forcedistr = Force * exp(((x - xzero)^2/(2*six^2))+((y-yzero)^2/(2*siy^2)))
The geometry is simply a block side*side*thickness, with base corner in the origin.
The only subdomain is of a custom material, of which i defined Young's modulus, Poisson's ratio and density (I consider the material isotropic, for now). It constraint is set to free.
As boundary conditions, I set the bottom face as fixed, and all the others are free, and I apply a face load in one direction. Initially I tried applying forcedistr as the vfalue for the load, but even when I insert a fixed value, I get the same error message:
Error: 7248
Undefined value found
Exception:
com.femlab.jni.FlNativeException: Undefined value found Solvermodel
Messages:
Undefined value found Solvermodel
I tried using a few different static solvers, with the same result.
Can someone explain to me what I might have done wrong?
Thanks,
Matteo
I am trying to model a piece of elastomer subject to a force, to compute its deformation using comsol 3.5a.
The 3D model is extremely simple, and just to see I don't make mistakes in future more complex ones.
I defined the following constants:
thickness
side
xzero = yzero = side/2
six = siy = side/15
Force
and used some of them to define the global expression (a 2D gaussian)
forcedistr = Force * exp(((x - xzero)^2/(2*six^2))+((y-yzero)^2/(2*siy^2)))
The geometry is simply a block side*side*thickness, with base corner in the origin.
The only subdomain is of a custom material, of which i defined Young's modulus, Poisson's ratio and density (I consider the material isotropic, for now). It constraint is set to free.
As boundary conditions, I set the bottom face as fixed, and all the others are free, and I apply a face load in one direction. Initially I tried applying forcedistr as the vfalue for the load, but even when I insert a fixed value, I get the same error message:
Error: 7248
Undefined value found
Exception:
com.femlab.jni.FlNativeException: Undefined value found Solvermodel
Messages:
Undefined value found Solvermodel
I tried using a few different static solvers, with the same result.
Can someone explain to me what I might have done wrong?
Thanks,
Matteo
6 Replies Last Post 16 mars 2011, 09:55 UTC−4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I would rather look for a missing material parameter or a value that is not correct (there are some 30 hits on that error message on the forum, have you had a look ?)
--
Good luck
Ivar
I would rather look for a missing material parameter or a value that is not correct (there are some 30 hits on that error message on the forum, have you had a look ?)
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello and thanks for your quick reply.
I did look around the forums, and tried already to review the possible solutions. I checked my materials parameters, and none seems to be missing, or uncorrect (should be flagged in red if it is actually wrong, right?).
Could have something to do with units of measure? The only red "sign" I found is in the unit for the global expression, but it stays there also when I had [N/cm^2] (The same units I expressed Force in) at the end of the expression for forcedistr.
I did look around the forums, and tried already to review the possible solutions. I checked my materials parameters, and none seems to be missing, or uncorrect (should be flagged in red if it is actually wrong, right?).
Could have something to do with units of measure? The only red "sign" I found is in the unit for the global expression, but it stays there also when I had [N/cm^2] (The same units I expressed Force in) at the end of the expression for forcedistr.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
A little update.
I tried an identical model on COMSOL 4.1, and I still have the same problem. I attach the model file for consideration. If anyone can help me identify the issue here, I can then solve it on the 3.5a model (on the 8-core computer we only have comsol 3.5, for some reason).
Tanks,
Matteo
I tried an identical model on COMSOL 4.1, and I still have the same problem. I attach the model file for consideration. If anyone can help me identify the issue here, I can then solve it on the 3.5a model (on the 8-core computer we only have comsol 3.5, for some reason).
Tanks,
Matteo
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
well I'm not fully astonished, that your solver is not converging ;)
1) you have nu=0.5 (Poisson) are you sure that is physical ?
Anyhow, it gives NaNs for sure. If you insist, try 0.49, but then probably you should also turn on "Nearly Incompressible material" to solve or the pressure to help the solver to evolve correctly. But anyhow, I suppose you should rather have a nu=0.3 or at most 0.4 for your material, or ? (try it out at least until your model is running)
2) have you checked your impact function ? enter your exponential into an analytical function and plot it, I'm not sure that is what you wanted, at least it was quite different from what I had expected, reading you.
Then I yould use upper case X and Y to use the initial Material frame as reference and then I would adapt the formula to give some more single gaussian impact. You could in fact use the "Gaussian" function build into COSMOL v4, check your functions
3) If you fix the entire underneath surface, is that how it is ? or should it be a membrane ? attached to it's borders?
--
Good luck
Ivar
well I'm not fully astonished, that your solver is not converging ;)
1) you have nu=0.5 (Poisson) are you sure that is physical ?
Anyhow, it gives NaNs for sure. If you insist, try 0.49, but then probably you should also turn on "Nearly Incompressible material" to solve or the pressure to help the solver to evolve correctly. But anyhow, I suppose you should rather have a nu=0.3 or at most 0.4 for your material, or ? (try it out at least until your model is running)
2) have you checked your impact function ? enter your exponential into an analytical function and plot it, I'm not sure that is what you wanted, at least it was quite different from what I had expected, reading you.
Then I yould use upper case X and Y to use the initial Material frame as reference and then I would adapt the formula to give some more single gaussian impact. You could in fact use the "Gaussian" function build into COSMOL v4, check your functions
3) If you fix the entire underneath surface, is that how it is ? or should it be a membrane ? attached to it's borders?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello,
truth be told, I was hoping that solver would not converge, as that would help me correct my 3.5 model. Anyway...
0.5 is actually the value for Poisson's ratio I found in my references. However, if, as you say, that gives numerical issues, I'll run the simulations with a different one, for instance 0.45.
I can't, however, set the Nearly Incompressible Material, as it doesn't make much sense while trying to simulate the behaviour of an elastomer (I know I should probably set some different elastic model rather than the linear one, but I couldn't find the values for all the required parameters)
You are quite right. I made the typical high school mistake of forgetting the minus sign.
The idea is that the elastomeric slab is on a surface, and I look at how it deforms when someone touches it.
Anyway, I made the modifications and suggestions you recommended, and I came up with new Errors.
The most common is from the solver (but I get this same error upon opening the model) (I thought this might be a graphics error, so I tried changing the preferences to use directX, but to no avail):
but sometimes I get this one (I can't point out how to get it at all times):
Searching around the forum, I found a few threads about something similar, and tried resetting solutions, meshes and history (this latter step often prompts me with an "unexpected error"), I also tried removing the solver nodes and adding them anew, but with no result.
truth be told, I was hoping that solver would not converge, as that would help me correct my 3.5 model. Anyway...
1) you have nu=0.5 (Poisson) are you sure that is physical ?
0.5 is actually the value for Poisson's ratio I found in my references. However, if, as you say, that gives numerical issues, I'll run the simulations with a different one, for instance 0.45.
I can't, however, set the Nearly Incompressible Material, as it doesn't make much sense while trying to simulate the behaviour of an elastomer (I know I should probably set some different elastic model rather than the linear one, but I couldn't find the values for all the required parameters)
2) have you checked your impact function ?
You are quite right. I made the typical high school mistake of forgetting the minus sign.
3) If you fix the entire underneath surface, is that how it is ? or should it be a membrane ? attached to it's borders?
The idea is that the elastomeric slab is on a surface, and I look at how it deforms when someone touches it.
Anyway, I made the modifications and suggestions you recommended, and I came up with new Errors.
The most common is from the solver (but I get this same error upon opening the model) (I thought this might be a graphics error, so I tried changing the preferences to use directX, but to no avail):
Invalid property value.
- Property: edges
'Plot data set edges' is a string.
- : "off", "on"
but sometimes I get this one (I can't point out how to get it at all times):
Found a model entity with a name null that does not belong to a model
Searching around the forum, I found a few threads about something similar, and tried resetting solutions, meshes and history (this latter step often prompts me with an "unexpected error"), I also tried removing the solver nodes and adding them anew, but with no result.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
for the last error there might be a new empty Definition integration coupling variable that has appeared (depends on the version you have). If you have a phantom one, that is none of yours, try to delete it and resolve, or save and reload COMSOL and the file, then delete the phantom. (Comsol support have told me they are working on it, it has not happen to me in he last patch so far though)
--
Good luck
Ivar
for the last error there might be a new empty Definition integration coupling variable that has appeared (depends on the version you have). If you have a phantom one, that is none of yours, try to delete it and resolve, or save and reload COMSOL and the file, then delete the phantom. (Comsol support have told me they are working on it, it has not happen to me in he last patch so far though)
--
Good luck
Ivar