Non Convergence due to material failure or problematic contacts? Please HELP!!!

zoirallizoiralli Member Posts: 14

Hi everyone,

I am trying for over 6 months to model the post-cracking behavior of an ultra high performance concrete subjected to bending. After trying various suggestions that I found here in this community (solid65, microplane with elastic damage, menetrey-willam) I ended up using Drucker-Prager Model for Concrete as I can input stress-strain data obtained from experiments to describe the constitutive law of strain hardening ultra high performance concrete. From a quick look on the solver output I think the problem lies on the material ( "The material solution failed for element 21503 with material 1"). However, when I plot the Newton-Raphson Residual Forces it seems that the contacts between the prism and the supports or the supports and the steel bases are problematic. I changed set the stiffness to be updated after each iteration, but still nothing works. At the end of the day I want to be able to obtain the plateau and the descending branch on the Force-Midspan Displacement curve. Could someone please have a look on the attached model and help me? My deadline is approaching and I am in desperate need of help!!! 





  • MarcinXMarcinX Member Posts: 7
    edited October 2019

    Why don't you try to model all supports and loads more simple? I see a lot of non-linearities here. This is hard to converge such model. Try to make simple supports like "remote displacement" with the rotation-free setting applied directly to the beam (you have to prepare geometry for this). Forces can be applied also direct to the beam geometry. After that, you can try to play with additional blocks but I don't see the need.

  • zoirallizoiralli Member Posts: 14
    edited October 2019
    Thank you for your suggestions! I want to study the effect that different types of contact between the experimental set up and the beam havd on the flexural strength. That is why I have modeled the supports and impactors.
  • QuiqueQuique Member Posts: 8
    edited April 2020

    Dear Zoiralli,

    Recently I was working in a similar problem comparing different methods to study concrete using FEM. I don´t know if you are solved the problem. Anyway, at the beguining I had defined the Drucker-Prager concrete model with dilitancy parameters but I got the same problem as you. After several tests, I omited dilatancy factors and my FE model converged.

  • akhemkaakhemka Forum Coordinator Posts: 676

    Hi @zoiralli ,

    We have seen this happen when the material is experiencing softening and for a small load increment there is a large strain associated. This large strain goes to the routine which does not "expect" to "see" these large values and then outputs the error. There is not a way to post-process this issue specifically. What is suggested is to post-process the residuals and locate the areas where the issue occurs but you have already done it.


    Ashish Khemka

  • jjdoylejjdoyle PittsburghMember Posts: 97

    Assuming the DP material model is correct and the convergence failure you are experiencing is caused by material losing stiffness in tension, the only two tools I can suggest to take the model further down the negative slope of force-deflection curve are nonlinear stabilization or Arc-Length Method. You cannot use them both together. See documented APDL commands: STABILIZE and ARCLEN for more details.

Sign In or Register to comment.