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.
Singularity point (corner) between two contentious BC gives convergence problem.
Posted 08.09.2015, 09:29 GMT-4 Mesh, Modeling Tools & Definitions, Parameters, Variables, & Functions, Structural Mechanics Version 5.1 3 Replies
Please login with a confirmed email address before reporting spam
Dear Comsol users,
I am trying to solve the following time dependent, 2D mechanical model:
A Visco-elasto-plastic crust of 200km x 30km is pushed from its right and bottom boundary (vx = -0.8 mm/yr) against a fixed (left) boundary (the backstop).
-=Geometry=-
---------------------------------------------------------------------
Sediment (Thickness=5km)
---------------------------------------------------------------------
Salt (T=1km)
---------------------------------------------------------------------
Basement (T=14km)
--------------------------------------------------------------------
-=BC=-
Top is an open boundary.
Left is fixed in y and x.
Right and bottom boundary: Vx= -0.8 mm/yr (compression velocity) and Vy =0.
My model is running well until a certain time, when plastic deformation just blows up at the left bottom corner. My first guess to explain this problem is the conflict between the left BC(fixed in y and x ) and bottom BC (fixed in y but moving at with a velocity of -0.8 mm/yr in x). Indeed, the left bottom node will move at the speed of 0.8 mm/yr toward the left, whereas the node just above is completely fixed in both direction. This implies that the element in between those two nodes is highly deformed and gives crazy plastic deformation to finally stop the model.
I was wondering what i should do to get around this problem at the left bottom corner, in order to allow my model to finish the computation (or at least, run for a bit longer)?
I upload the model in case you need more details.
Thank you in advance.
Rémi
I am trying to solve the following time dependent, 2D mechanical model:
A Visco-elasto-plastic crust of 200km x 30km is pushed from its right and bottom boundary (vx = -0.8 mm/yr) against a fixed (left) boundary (the backstop).
-=Geometry=-
---------------------------------------------------------------------
Sediment (Thickness=5km)
---------------------------------------------------------------------
Salt (T=1km)
---------------------------------------------------------------------
Basement (T=14km)
--------------------------------------------------------------------
-=BC=-
Top is an open boundary.
Left is fixed in y and x.
Right and bottom boundary: Vx= -0.8 mm/yr (compression velocity) and Vy =0.
My model is running well until a certain time, when plastic deformation just blows up at the left bottom corner. My first guess to explain this problem is the conflict between the left BC(fixed in y and x ) and bottom BC (fixed in y but moving at with a velocity of -0.8 mm/yr in x). Indeed, the left bottom node will move at the speed of 0.8 mm/yr toward the left, whereas the node just above is completely fixed in both direction. This implies that the element in between those two nodes is highly deformed and gives crazy plastic deformation to finally stop the model.
I was wondering what i should do to get around this problem at the left bottom corner, in order to allow my model to finish the computation (or at least, run for a bit longer)?
I upload the model in case you need more details.
Thank you in advance.
Rémi
Attachments:
3 Replies Last Post 10.09.2015, 01:47 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi
Why do you not use a symmmetry or a "rolling" boundary BC on the left vertical wall, as fixing it in X and Y is for me an overconstrain, no ?
--
Good luck
Ivar
Why do you not use a symmmetry or a "rolling" boundary BC on the left vertical wall, as fixing it in X and Y is for me an overconstrain, no ?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi Ivar,
And thank you for your interest.
I should have precised that i tried with a rolling BC on the left wall, but it didn´t solve the problem at the bottom left corner. The problem comes from the fact that the left wall is fixed in X and the bottom boundary is forced to move in X. The element at the left bottom corner is then highly deformed (see picture attached) and plastic deformation just blows up. In my opinion (and after some test), allowing the left wall to move in Y cannot solve the problem at this location.
Thank you again.
Rémi
And thank you for your interest.
I should have precised that i tried with a rolling BC on the left wall, but it didn´t solve the problem at the bottom left corner. The problem comes from the fact that the left wall is fixed in X and the bottom boundary is forced to move in X. The element at the left bottom corner is then highly deformed (see picture attached) and plastic deformation just blows up. In my opinion (and after some test), allowing the left wall to move in Y cannot solve the problem at this location.
Thank you again.
Rémi
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
9 years ago
Hi
Sorry I should have cathed that ;)
but from your image, if the black lines are your mesh then its normal you have a singularity particularly in quad mesh in such a sharp corner, it's "mathematic". And its all within your single mes element I see, also "normal"
The easiest is to simply ignore that corner or exclude it, there are different ways, one is to make a small box, of the size of your current mesh just there, mesh it somewhat finer and then exclude it from the plots, or play on the max stress rulers of the plots (tedious though)
--
Good luck
Ivar
Sorry I should have cathed that ;)
but from your image, if the black lines are your mesh then its normal you have a singularity particularly in quad mesh in such a sharp corner, it's "mathematic". And its all within your single mes element I see, also "normal"
The easiest is to simply ignore that corner or exclude it, there are different ways, one is to make a small box, of the size of your current mesh just there, mesh it somewhat finer and then exclude it from the plots, or play on the max stress rulers of the plots (tedious though)
--
Good luck
Ivar