General Mechanical

General Mechanical

Trouble with Convergence on a parametric non linear simulation

    • thefrog
      Subscriber

      Hi,

      I'm new to Ansys and trying to do a contact analysis of a power plug and after some intial struggle i've got it to run quite nice but now I need to do a parametric study where I'm decreasing L1 until I've got the desired Insertion force. Due to this the green spring will compress against the Compression Limit.

    • 1shan
      Ansys Employee
      Bring the bodies in contact rather than with a initial gap, add 1000 initial substeps and try solving the model. Also, the bodies look sweepable. You could create a good hex mesh with 3 elements across the thickness rather than using a tetrahedral mesh.
      Regards Ishan.
    • thefrog
      Subscriber
      Hi Ishan,

      Thanks for the help.
      I've moved the component and closed the initial gap and added the substeps. Even tought I've had no issue with initial contact rather with the squeezing of the spring between the 2 parts. And I already had the substep maximum at 10000 but i've changed the initial value to the one you suggested anyway. Now these two thing didn't change anything ... it still won't converge and fails at roughly the same point as before.

      With the sweep mesh I have some trouble getting it to work. If I show the sweepable bodys it only shows me these 3 but I've also should be able to sweep the 2 that are marked (I split the body with a plane so they are sweepable and then shared topology).
      The Problem is that even tough the 3 bodys are shown as sweepable. The mesher can't sweep them :(


    • Emperor
      Subscriber

      What error does the solver output show you (you can do crtl+f in the solver output and search for 'error')? Did you look at the Newton-Raphson residuals? After how many bisections does the calculation stop?
    • thefrog
      Subscriber
      Hi brivael
      I only have the error that the solution did not converge.
      For the Residuals ... I save the last 3 and they link different depending on the specific simulation setting. In this instance they are all on the spring at a face with coarse mesh.
      It said in the output that it failed after EQUIL ITER 25 ... is that the bisection count? For this run the inital step size was 100 (range 20-1000).


    • thefrog
      Subscriber
      I think I've found the bisection count. It says: BEGIN BISECTION NUMBER2NEW TIME INCREMENT=0.10000E-03
      At that bisection it fails because the residual forces are only increasing.
    • peteroznewman
      Subscriber
      This model will have a much easier time converging (and solve faster) if you midsurface the solid body of the spring contact and replace the solid elements with shell elements and assign the thickness of the part as a property. When defining contact, turn the Shell Thickness effect On so that the same contact occurs as it did on the solid elements.
Viewing 6 reply threads
  • You must be logged in to reply to this topic.