What if lost of contact / separation with AnyForceSurfaceContact?

As I mentioned in other posts, I’m writing final modifications to my paper to describe the shoulder joint model with FDK and surface contact between two prosthetic components.

I received a very interesting question about contact algorithm and would like to share it to get your opinion.

I used the AnyForceSurfaceContact function to compute the glenohumeral joint reaction force (Gh-JRF). I understand that the FDK is highly dependant of this force since the humeral head positions depend on this GH-JRF (and vice-versa, till an equilibrium, for a given step). My studied motion and set-up ensure keeping a contact between humeral head and glenoid (i.e. between the 2 prosthetic components).

But: what if it was not the case?! I mean, how the loss of GH-JRF would be (computationally) taken into account if there was a separation between the two components? Would the system be still stable or would it introduce such computational instability that it would be impossible to converge due to non-smoothness?

Of course, I read the reference manual several times; but wasn’t able to find a clear and precise answer for this question.

I really thank you for your answers and the elements you could add to answer my questions!

Hi Lauranne,

The loss of contact does not necessary mean an unsolvable instability as long as there are enough elastic elements that keep everything together (like in real life). For example, sometimes we use 6dof or less springs that keep an engineering joint together + surface contacts. This kind of setup allows for separation - effectively the surface contact is just a spring of some kind.

I believe it is best to make a test and have some solid base for the answer you could quickly design a 1dof problem, where two springs are balancing a ball, a load/unload force is applied to the ball, and a surface comes in play at some point. I will let you do this one :slight_smile:

Kind regards,

Hi Pavel,
Thank you for your reply!
I understand what you explained. I think that I already played, during the development of my shoulder model, with the concept of springs that you described. In fact, it’s my main assumption: it’s possible to reproduce soft tissues action using linear or non-linear springs model. It’s the reason why I used the FDK. However, I didn’t need to introduce springs in the medio-lateral direction and I suppose it’s linked to the arm basic movement I studied.

Your idea of designing a very easy one DoF problem is great; I’ll probably follow it to get a better understanding of what happens :wink: