In my spinal model, I would like to use facet joints and FDK. The facet joints model I'm using is 'FacetJointsModel', where they are represented as a conditional force element with the class AnyGeneralMuscle.
The spine model works with FDK without the facets, but when I activate them I get the following error:
Failed to resolve force-dependent kinematics. Too small steps for GSS search causes looping of Newton method (final force error = 3.792912E+01). Continue anyhow...
To understand better what causes this error, I tried changing the strength value assigned to the facets and observed that the final force error also changes, however is still present and quite high. I checked the wiki section on how to trouble shoot FDK models and tried to play with commands like MaxNewtonStep, ForceTol and MaxIteration, but still have not found the cause of this error. My hypothesis is that the problem is related to the addition of these "facet-muscles" that somehow interfere with the FDK. So my question is: how are they considered within the FDK problem?
To my knowledge, without FDK, the strength (= 1000) of the facets is deliberately high so that they do not interfere with the muscle recruitment optimization, so maybe it needs to be tuned when switching FDK on.
Could you kindly direct me to the cause of this problem?
I will very much appreciate your suggestions. Thank you!
You practically answered your own question. The strength is very high and basically what that means is that the force in the contact muscles is a subject to the muscle recruitment optimization and can get rather high / not sensitive enough to provide the equilibrium to the system. May I ask why do you want to use the muscle-based approach rather than an actual boney contact? Are you using a patient geometry or generic AnyBody one? Keep in mind that you could approximate articular surfaces with geometric primitives to make it more efficient for FDK.
I'm not using a patient-specific geometry, just generic AnyBody thoracolumbar spine model. The reason why I am using the muscle-based approach is because in this way I can represent pain in the facet joints (penalizing the recruitment of these muscles).
My questions now are:
Since the muscle-based approach requires muscles with rather high strength, will this approach ever work with FDK? Does it clash with the FDK method or is there a way to make it work?
With "you could approximate articular surfaces with geometric primitives to make it more efficient for FDK", are you referring to the example with the fixation device in the AMMR? In that example there are spheres, but the contact force is still defined via an AnyForce and not as AnyForceSurfaceContact.
