General Mechanical

General Mechanical

Topics relate to Mechanical Enterprise, Motion, Additive Print and more

Boundary conditions in a non-linear and multi-step simulation

    • Alberto Bensi

      Hello, everyone,
      I have a problem that I can't figure out how to solve. I should insert the tool you see in the figure inside the hollow curved channel and then simulate that the tip remains locked while the base of the tool rotates on its axis to evaluate the effects of twisting. 

      The first part of the insertion into the channel works fine, the problem is when I try to twist the instrument because it takes a curved position and I can't impose total locking of the tip. 
      I have tried using a remote displacement (which is activated only in the third and fourth substeps, after the instrument is completely inside in the channel) and creating a new coordinate system that has an axis as parallel as possible to the instrument tip.

      But as soon as the third substep starts, the simulation crashes because it cannot converge. 

      How can I try to solve it?


    • Bill Bulat
      Ansys Employee

      Hello Alberto,

      Presumably, you already have large deformation effects active for an analysis like this. The first thing I would do is identify the body in which excessive element distortion occurs (I see something like "Sez_triang_Alt2.part" in your list of messages... is that the tube?). I would also verify that the surfaces involved in the contact between the tool and tube surfaces are properly selected/defined. I might try reducing contact stiffness (maybe significantly... say, by a factor of 0.05) and use the pure penalty algorithm. There are cases when, if the stiffness of the contacting materials are different, the pure Lagrange algorithm is more stable. Some experimentation with different contact detection options might also be necessary. I often have better luck with asymmetric contact... in your case I would declare with tool surfaces to be the "contact", and the tube the "target". Are the tube material properties nonlinear? If so, you might post process the last converged result and see what state that nonlinear material is in immediately before the convergence failure occurs.

      Are you using default "program controlled" convergence criteria? Themessage saying that "The reference convergence value may be less than the threshold" suggests this should be tightened so that excessive disequilibrium is reduced before proceeding to the next substep (which may be allowing the system to progress outside of the "radius of convergence"):

      One thing that's kind of a last resort (due to computational expense) would be to solve it as a "slow transient" with an appropriate level of damping. Time integration limits the motion that can occur over a time step, and damping adds further numerical stabilization.




    • Alberto Bensi

      Hello Bill,

      first of all, I would like to thank you for the time you took to help me.

      In order: 

      • Yes, I activated the large deflection option for this analysis.
      • "Sez_triang_Alt2.part" is the tool, and I'm pretty sure that problem is the way i tried to lock the tip. Because in the past when I did the first attemps with this kind analysis (with a more straight tube) I had the same problem, so as soon as the third substep (related to twisting) began, the tip of the instrument exploded, literally. It was because I initially thought that by imposing everything equal to zero, I was saying that from that moment when the remote displacement was activated would work as a fixed support and the tip would be locked. But, actually, I was imposing to bring the tip of the instrument back to the absolute origin. I am explaining this because it looks to me like something similar is happening also now. I think there is a better way to impose that portion of the tool remains locked with the  remote displacements equal to zero relative to the last position reached in the previous substep, no? Such as a coordinate system integral to the center of mass of the tip as it moves. This way I just impose everything equal to zero with respect to that coordinate system and only activate it in the third substep.
      • I already reduced stifness factor to 0.05.
      • I tried to use Normal Lagrange method and it worked well for the part of inserting the tool into the tube. I can try also with pure penalty. 
      • I already select asymmetric contact, because i know that is for case when it's clear who is the contact and who is the target. Like you said, tool is the contact and tube is the target.
      • The tube material is linear. Instead the material of tool is non lineart. Is a nickel-titanium alloy with the property of superelasticity.
      • As for the convergence criterion, on the other hand, I did not know, I will try to make the changes you suggested.

      I thank you again for the answer you have already given me, and I hope you can clarify my thoughts about the tip lock, to confirm whether what I have done is right or something else can be done.

      Best regards,


Viewing 2 reply threads
  • You must be logged in to reply to this topic.