Why when the thickness increases a convergence failure occures?

munkhunurmunkhunur Member
edited November 20 in Structures

Dear expert,

I need an urgent help to wrap my simulation up as much as fast. The problem with my FE analysis is as following:

The steel bar (SOLID ELEMENT 186) with outer tube (SHELL ELEMENT) are used for this knee brace. Cyclic loading is applied in the tip of the cantilever beam. Nonlinear FE is assumed here. On the other hand, material and geometry NLs are considered.

  1. When the initial thickness of outer tube is 2 mm the convergence have done without any error. As illustrated below.
  2. If the thickness of outer tube increases to 4 mm or 6 mm etc, convergence failure occurs with high distorted element in the knee brace (inside bar/outside tube). The error with number of 99999 is pop up as well.

I did make more fine mesh for both tube and bar, decrease the initial loading step, change the normal stiffness 0.1-1, and changed the contact formulation between tube and bar. By the way a frictionless contact is used here. Unfortunately, I have still been faced with convergence failure. help me please.

UNUR

Comments

  • munkhunurmunkhunur Member
    edited November 20

    Converged solution with 2mm thickness is shown here.

    Unconverged solution with 4mm thickness of outer tube is as shown below. Why when the thickness of the bar increase solution can not converge?

    What should I do to get good results with excellent convergence in the analysis?

    Thank you so much

    UNUR

  • How are you changing the formulation of the contact between the steel bar and the outer tube? Because if you are changing the thickness of the outer tube, then the contact can get lost since there can be an initial penetration between the steel bar and outer tube. You might need to modify the geometry (so that the steel bar and outer tube are sufficiently apart) so that when you enter the thickness in the modelling, there exists a contact and there is no initial penetration.

    Also use the contact tool to check if there exists a contact between them, before the solution. If it doesn't, then you might need to change the contact settings, or use an increased pinball radius such that the contact is detected by the solver. Go with the latter approach first if the contact is far open or open and not detected by the solver.

  • munkhunurmunkhunur Member
    edited November 21

    Dear Rameez_ul_Haq 

    Thank you so much for your quick reply.

    How are you changing the formulation of the contact between the steel bar and the outer tube?

    • Please kindly see the attached picture. This is the how I made the contact formulation in between tube and bar as shown here.

    According to your suggestion I have checked the initial contact as shown below. But, I am not sure about that what does it mean the penetration in the contact 7.0485e-10 mm.

    "That when you enter the thickness in the modelling, there exists a contact and there is no initial penetration." it means initial penetration must be 0 is that right? If so how to get it to 0.

    "You might need to modify the geometry" I do not understand this. You mean it must I increase the gap between tube and bar?

    Thank you so much

    Regards

    Unur

  • Rameez_ul_HaqRameez_ul_Haq Member
    edited November 21

    Since you are using the 'shell thickness effect' as NO, then what I said about the initial penetration and the detection of the contact by the solver getting lost doesn't apply, since the change in the thickness will not effect the contact between them.

    Share clear pictures here for the location of the elements which are highly distorting, in mesh (before the solution).

    First, use default settings for the frictionless support to see if it is working or not. If not, then [for the case of thickness 4 or 6 mm) increase the diameter of the outer tube such that the gap between the inner surface of the outer tube and the outer surface of the steel bar remains the same as it was for 2 mm of thickness. Try these options.

    Also, you said cyclic loading is applied here. What kind of analysis you are doing?

  • peteroznewmanpeteroznewman Member
    edited November 21

    @munkhunur

    Is the tube over the steel rod intended to carry load, or does it have another function and will not carry significant load? If it is not intended to carry load, then I recommend you remove it from the model.

    If it is intended to carry load, what is the material and how is the tube assembled over the rod? Is there a clearance between the tube and rod that allows the tube to be slid over the rod?

  • Dear Rameez_ul_Haq

    I am doing cantilever beam with this special knee brace. I wanted to get most critical thickness of the outer tube with different gaps between tube and bar.

    As you said if I set shell thickness effect YES contact tree does not work with yellow color. Why?

    I have tried with it up to 10mm thickness but still have a convergence problem.

    Thank you so much

    Regards

    Unur

  • munkhunurmunkhunur Member
    edited November 21

    Dear @peteroznewman 

    Thank you so much for your question.

    The main purpose of using this outer tube (buckling restrainer) is to restrict the buckling of the steel bar. On the other hand, the tube is inserted to the rode freely. It means tube will slide over the rode in the initial unloading condition. After compression loading the bar must be touch to the inner surface of the tube. Yes, your right, the clearance allow the tube to be slide over the rod.

    I am just finding the critical value of the clearance and critical thickness of the tube in the analysis.

    For example: 1mm clearance with 2mm, 4mm, and 6mm etc., then 2mm and 3mm clearances with same thickness, Unfortunately, I have stuck in the beginning of my finding in analysis with convergence failure. Please help me.

    In the case of material, bilinear isotropic hardening is used with a parameter as shown in fig.

    Best regards,

    Unur

  • Dear @peteroznewman and Rameez_ul_Haq

    One more thing I have to say here is a spring connection is used to hold the tube as modeled below.


  • Dear Peter

    I appreciate your kind suggestion that you were giving really worth papers to me.

    Regards

    Unur

  • munkhunurmunkhunur Member
    edited November 25

    Hello it is me again. Anybody can you see that shared file. Where did I mistake. still i have non-convergence.

    When I set Shell thickness effect YES it shows question mark in the tree. How to set it?

    Unur

Sign In or Register to comment.