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.
Moving Mesh - Thermal stress coupling
Posted 2 mai 2013, 09:44 UTC−4 Mesh 5 Replies
Please login with a confirmed email address before reporting spam
Hello to all,
I need to perform an RF frequency domain analysis on a waveguide which has been previously deformed by a Thermal Stress stationary analysis.
In order to do this, I need to calculate the new coordinate x,y,z (after the thermal stress induced displacement) using a moving mesh module.
I configured a Thermal stress module with all the needed conditions, to compute temperature T and deformation u,v,w.
In order to verify the used settings, I computed this step and the the deformation has been well calculated so that the waveguide has been deformed by the heating (excluding the fixed constraints boundaries) showing deformation and, using a boundary probe, I verify the computation of u, v, and w.
Then I inserted a Moving Mesh module, in which I set "Prescribed mesh displacement" on all boundaries, with dx=u, dy=v and dz=w (in order to use the previous step results). I insert a "free deformation" block in which I set to zero the initial values: dx0=0 dy0=0 and dz0=0 for all domains. In the default "fixed Mesh" block, all domains have been overridden.
This module has been assigned to a Stationary step in the same study.
After few seconds, it generates the following error:
Failed to evaluate expression.
- Expression: X+mod1.u-x
Failed to evaluate variable.
- Variable: mod1.u
- Geometry: 1
- Boundary: 3 10
Have you any suggestions to help me configuring the right coupling between these modules?
Thanks in advance.
Alberto
I need to perform an RF frequency domain analysis on a waveguide which has been previously deformed by a Thermal Stress stationary analysis.
In order to do this, I need to calculate the new coordinate x,y,z (after the thermal stress induced displacement) using a moving mesh module.
I configured a Thermal stress module with all the needed conditions, to compute temperature T and deformation u,v,w.
In order to verify the used settings, I computed this step and the the deformation has been well calculated so that the waveguide has been deformed by the heating (excluding the fixed constraints boundaries) showing deformation and, using a boundary probe, I verify the computation of u, v, and w.
Then I inserted a Moving Mesh module, in which I set "Prescribed mesh displacement" on all boundaries, with dx=u, dy=v and dz=w (in order to use the previous step results). I insert a "free deformation" block in which I set to zero the initial values: dx0=0 dy0=0 and dz0=0 for all domains. In the default "fixed Mesh" block, all domains have been overridden.
This module has been assigned to a Stationary step in the same study.
After few seconds, it generates the following error:
Failed to evaluate expression.
- Expression: X+mod1.u-x
Failed to evaluate variable.
- Variable: mod1.u
- Geometry: 1
- Boundary: 3 10
Have you any suggestions to help me configuring the right coupling between these modules?
Thanks in advance.
Alberto
5 Replies Last Post 2 juil. 2013, 06:03 UTC−4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
it's not clear for me if you have 2 models, or if you do everything in the same model with coupled or uncoupled physics
one thing COMSOL complains about is mod1.u, probably it would like better just "u" but then which "u", if you have solved tor the TS separately, and run the RF as a separate solver sequence, then you must ensure that you load in the variables not solved for from the previous solver step so that the fields (as u is a field of the type u(x,y,z,t) ) can map correctly
If the heating is done separately and you are not interested in the stress, only the deformed geometry, you can also save the deformed mesh as a mesh2 and continue to work from there
Another way is to work full coupled, I believe the est is then to look at the model library example of the bending coupler, or was it an isolator, some time since I was digging into that domain ;) Have a look
--
Good luck
Ivar
it's not clear for me if you have 2 models, or if you do everything in the same model with coupled or uncoupled physics
one thing COMSOL complains about is mod1.u, probably it would like better just "u" but then which "u", if you have solved tor the TS separately, and run the RF as a separate solver sequence, then you must ensure that you load in the variables not solved for from the previous solver step so that the fields (as u is a field of the type u(x,y,z,t) ) can map correctly
If the heating is done separately and you are not interested in the stress, only the deformed geometry, you can also save the deformed mesh as a mesh2 and continue to work from there
Another way is to work full coupled, I believe the est is then to look at the model library example of the bending coupler, or was it an isolator, some time since I was digging into that domain ;) Have a look
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
thank you for responding!
I'm using only one model, I have tried to use both the form "mod1.u" and "u", but I received the same error.
In the solver sequence:
- in the first step (in which u, v, w are computed) all variables are set to "Method: Initial expression", "Solution: zero" by default, and I left it unchanged.
In this step, the stationary solver uses Advanced and Fully coupled feature (the Direct is disable by default).
- in the second step (where x, y, x are computed) I set the variables solved for to "Method: Initial expression", "Solution: zero" and the variable not solved for to "Method: Solution", "Solution: solver sequence 1" (which is the current and the only sequence in the study) and "Use: Store Solution 2" (that is the solution stored by the step 1 in which u,v,w are computed) .
In this step the stationary solver uses Direct, Advanced and Fully coupled feature.
I am interested in the stress, I will try to reproduce the settings of the models that you have suggest.
thank you for responding!
I'm using only one model, I have tried to use both the form "mod1.u" and "u", but I received the same error.
In the solver sequence:
- in the first step (in which u, v, w are computed) all variables are set to "Method: Initial expression", "Solution: zero" by default, and I left it unchanged.
In this step, the stationary solver uses Advanced and Fully coupled feature (the Direct is disable by default).
- in the second step (where x, y, x are computed) I set the variables solved for to "Method: Initial expression", "Solution: zero" and the variable not solved for to "Method: Solution", "Solution: solver sequence 1" (which is the current and the only sequence in the study) and "Use: Store Solution 2" (that is the solution stored by the step 1 in which u,v,w are computed) .
In this step the stationary solver uses Direct, Advanced and Fully coupled feature.
I am interested in the stress, I will try to reproduce the settings of the models that you have suggest.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
one important issue: be sure you place the ALE above the TS or solid if you solve them together, as Solid and TS have the spatial frame representation (which is an ALE type) these solid domains should not be included into the ALE part, and this will be automatically taken into account if ALE is before SOLID as then solid will override similar features defined before, here in the ALE (so even if ALE includes ALL domains, the solid after will override the ALE for the domains included in the solid section.
Now normally you can add the RF physics too, then add a frequency domain solver after the stationary case under the same study. Delete first all solver sequences before you add he second solver in the study, then generate the default
solver sequence. You should have now stationary (pls select only for TS and ALE physics) then COMSOL writes a "store solution" node to save the data (actually the one with id 2) set the frequency domain solver only for the RF and check that COMSOL has linked the initial values solved for and not solved for to the physics stationary results
When you solve this (turn on the plots while solving) you should get the deformation of the TS and he RF solving on the deformed structure. But this is a one way coupling, as you do not consider any heat from the RF.
For that you must make a new study add a single frequency stationary or frequency - transient solver and run all 3 physics together. Also be sure the domain/boundary heat sources of TS is linked to the dissipated heat sources of the RF physics
--
Good luck
Ivar
one important issue: be sure you place the ALE above the TS or solid if you solve them together, as Solid and TS have the spatial frame representation (which is an ALE type) these solid domains should not be included into the ALE part, and this will be automatically taken into account if ALE is before SOLID as then solid will override similar features defined before, here in the ALE (so even if ALE includes ALL domains, the solid after will override the ALE for the domains included in the solid section.
Now normally you can add the RF physics too, then add a frequency domain solver after the stationary case under the same study. Delete first all solver sequences before you add he second solver in the study, then generate the default
solver sequence. You should have now stationary (pls select only for TS and ALE physics) then COMSOL writes a "store solution" node to save the data (actually the one with id 2) set the frequency domain solver only for the RF and check that COMSOL has linked the initial values solved for and not solved for to the physics stationary results
When you solve this (turn on the plots while solving) you should get the deformation of the TS and he RF solving on the deformed structure. But this is a one way coupling, as you do not consider any heat from the RF.
For that you must make a new study add a single frequency stationary or frequency - transient solver and run all 3 physics together. Also be sure the domain/boundary heat sources of TS is linked to the dissipated heat sources of the RF physics
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
In few words: I need all domains of my model in the thermal stress module, since I need to evaluate the deformation and the heat diffusion in the waveguide.
If I insert the ALE first, all domains result not applicable in ALE.
The RF analysis will be employed to compute only the Scattering parameters of the structure, after thermal deformation, by using the Frequency domain analysis.
So I have two questions:
1) Is there any way to put ALE after TS in the model tree?
2) Can I avoid to include ALE in my model ?
So, if in the same study I use only TS (solved by stationary) and RF (solved by frequency domain) without ALE in the model, is the RF study based on the mesh previously deformed by TS? or is it based on the initial configuration of the mesh?
Thanks in advance.
In few words: I need all domains of my model in the thermal stress module, since I need to evaluate the deformation and the heat diffusion in the waveguide.
If I insert the ALE first, all domains result not applicable in ALE.
The RF analysis will be employed to compute only the Scattering parameters of the structure, after thermal deformation, by using the Frequency domain analysis.
So I have two questions:
1) Is there any way to put ALE after TS in the model tree?
2) Can I avoid to include ALE in my model ?
So, if in the same study I use only TS (solved by stationary) and RF (solved by frequency domain) without ALE in the model, is the RF study based on the mesh previously deformed by TS? or is it based on the initial configuration of the mesh?
Thanks in advance.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hy all,
I solved the MM coupling by inserting the following boundary conditions:
- Domain Prescribed deformation:
Applied to the Domains which undergoes thermal stress - deformations.
The displacement vectors (u, v, w) computed by the TS module are employed to specify mesh displacement.
Prescribed Mesh displacement is set to dx = u, dy = v, dz = w.
- Domain Free deformation:
Applied to the domains which are not subjected to any structural formulation by the TS analysis aimed to displacement computation.
The domains are free to move, in order to follow the deformation of the adjacent domains.
The initial deformation is set to dx0 = 0, dy0 = 0, dz0 = 0.
- Boundary Prescribed Mesh Displacement:
Applied to the Boundaries which undergoes thermal stress - deformations.
I use this condition to specify that these boundaries are deformed by the thermal stress computation, though are connected to the free-deformation domains.
The displacement vectors (u, v, w), computed by the TS module, are employed to specify this superficial displacement:
dx = u, dy = v, dz = w.
Best regards to all!
I solved the MM coupling by inserting the following boundary conditions:
- Domain Prescribed deformation:
Applied to the Domains which undergoes thermal stress - deformations.
The displacement vectors (u, v, w) computed by the TS module are employed to specify mesh displacement.
Prescribed Mesh displacement is set to dx = u, dy = v, dz = w.
- Domain Free deformation:
Applied to the domains which are not subjected to any structural formulation by the TS analysis aimed to displacement computation.
The domains are free to move, in order to follow the deformation of the adjacent domains.
The initial deformation is set to dx0 = 0, dy0 = 0, dz0 = 0.
- Boundary Prescribed Mesh Displacement:
Applied to the Boundaries which undergoes thermal stress - deformations.
I use this condition to specify that these boundaries are deformed by the thermal stress computation, though are connected to the free-deformation domains.
The displacement vectors (u, v, w), computed by the TS module, are employed to specify this superficial displacement:
dx = u, dy = v, dz = w.
Best regards to all!