General Mechanical

General Mechanical

Contact conditions of welded region

    • Didar Mei
      Subscriber

      Hello! I am doing a static structural analysis of an assembly consisting of supports (half-square plates), bolts, a column, and a base, as seen in the photo below. I applied displacement to the column's top to identify the load-bearing capacity of the structure. The base is fixed. I conducted an experiment, where the failure occurred due to the crack in the welded region as shown in the next photo. My main concern is in the contact. I assigned a bonded connection between the supporting plate and base (due to the welding), and a frictionless connection between the column and base (since the column and base are not connected to each other, I used it to avoid penetration). However, my results are much higher than the experimental results (3rd photo). From the analysis of the Ansys results, I figured out that since the bonded connection is inserted between the supports and base, it has infinite stiffness and it will be impossible to simulate the crack of the weld (4th photo). What can be done to create a crack in the welded region? Is it more related to the simulation of weld or it is a concern of contact normal stiffness? Because I am decreasing the normal stiffness of bonded contact to simulate the opening of the weld.  

    • Erik Kostson
      Ansys Employee

       

       

      Hi

       

      The failure in Ansys is due to local buckling/failure and not due to cracking like in the experiments so as you say that is the main difference and hence the different results (exp. vs fea).

      See if an explicit or LS-Dyna analysis can capture the crack via element erosion (removing elements that have failed), but that is something to investigate as I have not seen an analysis on this before. Also the bottom base plate seems to have more flexibility in the experiment so that is also something to consider (probably other things to).

      Hope this helps 

       

      Erik

       

       

Viewing 1 reply thread
  • You must be logged in to reply to this topic.