General Mechanical

General Mechanical

eKill and element distortion in workbench

    • Reza Taheri
      Subscriber

      Hi everyone!

      I'm new to APDL command in workbench. Sorry if my questions are too obvious!

      First question, In each time step of my model I find the damage status. I want to kill the element that has a damage status equal one for the next time step. Does this happen automatically or it is something that I should put in manually as APDL command? 

      Second question, I get excessive element distortion error for few elements before my structure fails (I have a manual criteria for structure failure). I tried changing the mesh and load increment but no luck. Is it a good idea to use eKill to remove the element with high distortion? If so can you please guide me through it. My problem is displacement controlled and I want to go post-buckling, but error appears at buckling point.

      Cheers

    • peteroznewman
      Subscriber

       

      Ekill is not automatically applied during a solution. This topic shows ekill and other failure criteria. 

      https://forum.ansys.com/forums/topic/i-want-to-see-the-failure/

      Note that the links to the script were broken during one of the many upgrades to this website  : (

      Here is a good reference: https://simutechgroup.com/performing-ekill-element-death-in-mechanical

      If you use the Explicit Dynamics solver, elements that have failed are automatically removed from the solution, which continues along without them.

      • Reza Taheri
        Subscriber

         

        Thanks Peter. The archive file is not available. Do you mind sharing it again if you still have it. 

        I'm using surface model. I think explicit dynamic does not support surface models, so I assume APDL command in static structure is only option (?). 

        Also I have this element distortion error no matter what displacement size or substep I'm using. The stress contour shows the von mises is close to the stress limits I defined, but in a different location (element). Could this be the issue? If I increase this limit, then my damage criteria would become wrong right? If so, can you please give some insight.

        Cheers,

         

    • peteroznewman
      Subscriber

      I will look on my old computer to see if I can find the ekill script.

      It is difficult to advise you on how to proceed without seeing more details of your model. Can you insert some images of your geometry, loads and supports.

      A best practice for buckling is to seed some imperfection into the geometry. Did you do that?

      • Reza Taheri
        Subscriber

         

        Really appriciate your help Peter. Thanks.

        My model is a 3-layered sandwich structure of orthotropic material. I have defined element orientation to address the orthotropic properties (using shell elements). I’m using 1% of the first buckling mode of perfect geometry as the initial imperfection for the post-buckling analysis (image below is the full scale imperfection).

         

        Here is my BCs for post-buckling analysis:

        All BCs are direct FE conditions.

        Top: Displacement control push down. Rotation fixed. 

        Bottom: Displacement and rotation fixed.

        Sides: Displacement fixed in 2 direction, free in vertical direction. Rotation fixed.

        I have two symmetric bonded contacts between layers:

        Here is my current analysis settings:

        The idea is to kill elements with damage status of 1 and shed the load onto adjacent elements. I then have a visual failure criteria to define total structur failure. But I get the element distortion error, I think, at the buckling load (as the load vs displacement flattens out at the timestep I get the error). At this point I only have a few elements with damage status of 1.

         

    • peteroznewman
      Subscriber

      Reza, I recommend bringing the sheet bodies into SpaceClaim and using the Share button to cause the edges of the honeycomb core to imprint onto the facesheets. Then in Mechanical, the mesh will be connected between the core wall edges and the facesheets without using contact. This will result in a much cleaner simulation for ekill than using contact.

      • Reza Taheri
        Subscriber

         

         

        Hi Peter,

        I should have said earlier that the three surfaces I have are different material with diffeernt properties. So I cannot share the topology.

        I’m trying to make the code in https://simutechgroup.com/performing-ekill-element-death-in-mechanical work. Can you please tell what quantity Ansys uses for damage status (e.g. normal stress is Sx and Sy and shear is Sxy in x-y plane, I could not find the damage parameter online).

        Do you think of any thing in regards with my analysis setting?

         

         

    • peteroznewman
      Subscriber

      Hi Reza,

      In CAD, if mulitple bodies are United into a single body, then only one material assignment is possible.

      In SpaceClaim, when you use Share on multiple bodies, each body can have a different material assignment. It is only the nodes on the edges of a body that are shared so that the bodies are connected.

      I will have to look at the Simutech article and see if I can answer your other question later.

    • peteroznewman
      Subscriber

      Here is a link to the Ekill example script.

       

      • Reza Taheri
        Subscriber

        Hi Peter,

        Merry Christmas!

        Thanks for sharing the archive I appriciate it. I could not use the damage status as a kill criteria, so I’m trying a new approach.

        I tried applying kinematic hardening. My material is ortothropic, but I know the failure direction. So I use the compressive stress limit in that direction as the yield strength with 0 tangant modulus for bilinear kinematic hardening (as my case is monotonic loading, documentation says both isotropic and kinematic modes are the same). My solution now goes post-buckling (without exessive distortion) and load-displacement curve looks like this after 0.5mm total displacement:

        Total deformation also looks as should:

        But, the failure that I defined (Hashin progressive model) has a value of zero all along, for both ‘damage status’ and ‘maximum failure criteria’. The elements should start failing at the onset of the maximum deformation in above figure. I think it does not even calculate the damage parameters. 

        Do you know if damage contradicts hardening? If not, can you suggest why I don’t get any values for damage?

        Thanks a lot, and again, Merry Christmas!

         

    • peteroznewman
      Subscriber

      Hi Reza,

      I haven’t used Damage Evolution Laws in my work, where we are only concerned with staying well below failure limits.

      Have you read the ANSYS Help on Damage Evolution?

      You say you have an Orthotropic material. Is that the bulk properties of a carbon fiber composite layup? 

      I think you need to use the Layered feature of the shell elements to define the individual layers in the layup so that tensile and compressive failure limits of the fiber can be tracked for each layer as well as bond shear failure between layers.

      Merry Christmas!

       

Viewing 6 reply threads
  • You must be logged in to reply to this topic.