Hi!
I am currently working on a contact problem and am hoping for some feedback to make it more stable.
The model concists of a 1/8th sphere contacting a rectangular box. The materials used in the sphere are all Ogden constrained, whilest the box is neo-Hookean. I am using the Paradiso (default) solver and a 'sliding2' contact type.
The model represents a part of the buttocks (spheres) that indents a cushion (the box) with a specific force (weight). I've tried a default analysis with fixed displacement aswell as a dynamic analysis, but this seemed to make little difference in stability.
The problems I have been having are:
1. the (mesh)size of the box.
I can not increase the mesh size (amount of elements/nodes) of the box without making the situation unstable. Ideally I would like a 10x10x4 element box, but this will not run in my present configuration. I have tried varying the models parameters (penalty, gaptol, dtol, tolerance, time_step)m but they all seem to end up with either high-penetration results or unstable solutions (negative jacobian).
2. Jumping nodes.
In the first and last steps of the solution the sphere 'moves' up nodes to connect them to the box. Basically all parameter variations I have tried that found a solution have ended up with these contact problems.
3. Time consuming.
Although this is less of a problem, the present model - with just 1395 nodes - takes about an hour to run. Most of this time is lost on the first and last iterations on remodeling the 'maximum gap' to obey the gap tolerance. These situations are also the situations that the jumping nodes usually occur.
Are there any ways you might be able to assist me with these problems and/or give me some pointers based on the model? I have attached the model for your convenience.
I am currently working on a contact problem and am hoping for some feedback to make it more stable.
The model concists of a 1/8th sphere contacting a rectangular box. The materials used in the sphere are all Ogden constrained, whilest the box is neo-Hookean. I am using the Paradiso (default) solver and a 'sliding2' contact type.
The model represents a part of the buttocks (spheres) that indents a cushion (the box) with a specific force (weight). I've tried a default analysis with fixed displacement aswell as a dynamic analysis, but this seemed to make little difference in stability.
The problems I have been having are:
1. the (mesh)size of the box.
I can not increase the mesh size (amount of elements/nodes) of the box without making the situation unstable. Ideally I would like a 10x10x4 element box, but this will not run in my present configuration. I have tried varying the models parameters (penalty, gaptol, dtol, tolerance, time_step)m but they all seem to end up with either high-penetration results or unstable solutions (negative jacobian).
2. Jumping nodes.
In the first and last steps of the solution the sphere 'moves' up nodes to connect them to the box. Basically all parameter variations I have tried that found a solution have ended up with these contact problems.
3. Time consuming.
Although this is less of a problem, the present model - with just 1395 nodes - takes about an hour to run. Most of this time is lost on the first and last iterations on remodeling the 'maximum gap' to obey the gap tolerance. These situations are also the situations that the jumping nodes usually occur.
Are there any ways you might be able to assist me with these problems and/or give me some pointers based on the model? I have attached the model for your convenience.
Comment