Hi,
I am trying to simulate the spherical indentation of a cylindrical film with very thin thickness (3.35 angstrom)
and a radius of um. The film is fixed at the edges.
In order to achieve this, I used solid mechanics and a stationary solver, drawed two simple bodies, cylinder and hemispherical and added two boundary conditions:
a prescribed displacement for the top of the hemispherical in the minus z direction
a fixed constraint for the relevant boundaries of the cylinder
I am using Comsol version 4.3a.
As a start, I just want to indent in the center of the film, however in a further step, I would like to extend this model by selecting other indenting points. That is why I am already working on a 3d model.
My main problem now is the following error (*):
"Failed to find a solution. The relative error (1.2e +002) is greater than the relative tolerance. Returned solution is not converged. (...)"
In order two minimize the computing time for this high aspect ratio problem, I got the clue to use the swept technique for better meshing results, however I am not quite sure how to use it, regarding source and destination faces.
I could also manage to mesh without this method, using the default free tetrahedral. However, using predefined values, I got problems with the element size:
If i choose normal, I got the error "Failed to generate mesh for domain-domain 1. Intersecting face elements. - Face 1,2,3,4,8"
If I choose extremly fine, my computing time rises inacceptable and it works only for thicker films.
But also then, finally the relative error problem (*) occurs.
Interestingly, I can also mesh with a very coarse element size, however it finally fails again with the relative error problem (*).
How can I avoid this error
"Failed to find a solution. The relative error (1.2e +002) is greater than the relative tolerance. Returned solution is not converged. (...)"
Thanks for any help
I attach the file in my next post, for any reason it is not possible in this post.
I am trying to simulate the spherical indentation of a cylindrical film with very thin thickness (3.35 angstrom)
and a radius of um. The film is fixed at the edges.
In order to achieve this, I used solid mechanics and a stationary solver, drawed two simple bodies, cylinder and hemispherical and added two boundary conditions:
a prescribed displacement for the top of the hemispherical in the minus z direction
a fixed constraint for the relevant boundaries of the cylinder
I am using Comsol version 4.3a.
As a start, I just want to indent in the center of the film, however in a further step, I would like to extend this model by selecting other indenting points. That is why I am already working on a 3d model.
My main problem now is the following error (*):
"Failed to find a solution. The relative error (1.2e +002) is greater than the relative tolerance. Returned solution is not converged. (...)"
In order two minimize the computing time for this high aspect ratio problem, I got the clue to use the swept technique for better meshing results, however I am not quite sure how to use it, regarding source and destination faces.
I could also manage to mesh without this method, using the default free tetrahedral. However, using predefined values, I got problems with the element size:
If i choose normal, I got the error "Failed to generate mesh for domain-domain 1. Intersecting face elements. - Face 1,2,3,4,8"
If I choose extremly fine, my computing time rises inacceptable and it works only for thicker films.
But also then, finally the relative error problem (*) occurs.
Interestingly, I can also mesh with a very coarse element size, however it finally fails again with the relative error problem (*).
How can I avoid this error
"Failed to find a solution. The relative error (1.2e +002) is greater than the relative tolerance. Returned solution is not converged. (...)"
Thanks for any help
I attach the file in my next post, for any reason it is not possible in this post.