Tweet
For the past few months, I've been using FEBio with a hyperelastic model, with great results. Now, I want to extend it into a poroelastic model, but have immediately run into a problem.
As my first step in moving to poroelastic analysis, I simply replaced the facet-to-facet sliding contact with the biphasic sliding2, expecting to see no change in results. However, the problem now refuses to converge even on the first timestep, before any contact is made. When running in debug mode (-g), the plot file shows stresses on the order of 10^6 Pa, even though the surfaces are nowhere near each other.
Have you seen this happen before with sliding2? Is there any way around this behaviour? Thanks for any help.
Additional info: I can't share my files unless you've also signed the agreement to use the Visible Human Dataset, so instead I'll just describe the setup. It's a simplified knee joint, consisting only of the femur and tibia (each of which is a rigid body) and their articular cartilage (Ogden material, with parameters derived from other experiments). There is a rigid contact interface between each bone and a face of its cartilage. The two pieces of cartilage each have two additional faces defined, on the medial and lateral condyles, used for contact between them. The cartilage begin partially separated; multi-step analysis is used to start with position control, bring both condyles into contact, then switches to force control. Like I said, this all worked perfectly with facet-to-facet sliding, but fails miserably with sliding2 (I opened the feb file and replaced both instances of "facet-to-facet sliding" with "sliding2", so I know nothing else has been changed). Again, any help would be lovely!
Interesting addendum: I tried changing the contact to only apply during the second step. I thought I could have the bones position themselves, then enforce contact, then apply load. However, the results are completely unchanged: the very first timestep still experiences stresses over 10^6 Pa, even though there is no contact defined at that time (removing the contact entirely allowed the first step to behave normally). This seems to hearken back to my earlier issue with multi-step analysis, where the definition of later steps seems to propagate backwards through time, but this time I am using full-Newton in order to reduce the solving inconsistencies. I don't know what to make of this. Edit: Oh wait, the user manual says that steps can only enforce "prescribed displacement, nodal forces and rigid contact", so perhaps the sliding2 is simply being applied to all timesteps despite being specified for step 2.
As my first step in moving to poroelastic analysis, I simply replaced the facet-to-facet sliding contact with the biphasic sliding2, expecting to see no change in results. However, the problem now refuses to converge even on the first timestep, before any contact is made. When running in debug mode (-g), the plot file shows stresses on the order of 10^6 Pa, even though the surfaces are nowhere near each other.
Have you seen this happen before with sliding2? Is there any way around this behaviour? Thanks for any help.
Additional info: I can't share my files unless you've also signed the agreement to use the Visible Human Dataset, so instead I'll just describe the setup. It's a simplified knee joint, consisting only of the femur and tibia (each of which is a rigid body) and their articular cartilage (Ogden material, with parameters derived from other experiments). There is a rigid contact interface between each bone and a face of its cartilage. The two pieces of cartilage each have two additional faces defined, on the medial and lateral condyles, used for contact between them. The cartilage begin partially separated; multi-step analysis is used to start with position control, bring both condyles into contact, then switches to force control. Like I said, this all worked perfectly with facet-to-facet sliding, but fails miserably with sliding2 (I opened the feb file and replaced both instances of "facet-to-facet sliding" with "sliding2", so I know nothing else has been changed). Again, any help would be lovely!
Interesting addendum: I tried changing the contact to only apply during the second step. I thought I could have the bones position themselves, then enforce contact, then apply load. However, the results are completely unchanged: the very first timestep still experiences stresses over 10^6 Pa, even though there is no contact defined at that time (removing the contact entirely allowed the first step to behave normally). This seems to hearken back to my earlier issue with multi-step analysis, where the definition of later steps seems to propagate backwards through time, but this time I am using full-Newton in order to reduce the solving inconsistencies. I don't know what to make of this. Edit: Oh wait, the user manual says that steps can only enforce "prescribed displacement, nodal forces and rigid contact", so perhaps the sliding2 is simply being applied to all timesteps despite being specified for step 2.
Comment