LS Dyna

LS Dyna

Error in MAT_NONLOCAL

    • javat33489
      Subscriber

      Hi all.

      I'm modeling metal cutting.

      There are no chips when cutting.

      I am using the Johnson Cook model and ADD_EROSION.

      I want to apply MAT_NONLOCAL map. An error occurs during initialization:

      Why?

      What are the main parameters that must be in order for chips to appear when cutting?

      Materials:

      I have a good mesh of 100 mm by 100 mm thick, the thickness of the dies is 1 mm.

    • Reno Genest
      Ansys Employee

      Hello,

      What is L set to in *MAT_NONLOCAL? It look like you are including 302500 elements in the search which is way too much and this is why you ran out of memory. The L parameter should be set to include only a few elements. Refer to the LS-DYNA user manual for more information:

      Also, your elements have a high aspect ratio and so the quality could be improved.

      I would simplify the model for now and remvove the *MAT_NONLOCAL. To see chips, you need material failure to occur around a group of elements. If this does not occur, the elements just get deleted when they fail and you never  see chips. 

      An alternative to classical finite elements is to use particle methods to model metal cutting. With particle method, the bonds between particles break when the material fails and so  you don't have mass loss (as with regular FEM) and you can see chips right away. Have a look at the following examples:

      https://www.dynaexamples.com/efg/metal-cutting

      https://www.lstc-cmmg.org/cutting

      https://www.lstc-cmmg.org/spg

       

      Note that a higher CPU cost is associated with particle methods.

       

      Let me know how it goes.

       

      Reno.

      • javat33489
        Subscriber

        Thanks for the answer.

         >>What is L set to in *MAT_NONLOCAL?

        >>Refer to the LS-DYNA user manual for more information

        Yes, of course, I read everything and made the standard parameters

        >>To see chips, you need material failure to occur around a group of elements. If this does not occur, the elements just get deleted when they fail and you never  see chips. 

        At first I would like to solve everything by the Lagrange method, and then move on to particles. Please tell me what settings in the maps need to be done? I tried many options, nothing worked, the material is always deleted.

        >>An alternative to classical finite elements is to use particle methods to model metal cutting. With particle method, the bonds between particles break when the material fails and so  you don't have mass loss (as with regular FEM) and you can see chips right away. Have a look at the following examples

        Yes, of course I have seen them all. I looked all over youtube. I have also read many books and works. But nowhere was it said how to set up the material for chip separation.

        I can attach a k-file, can you see it? Or could you answer how to set up the maps so that the chips appear?

    • Reno Genest
      Ansys Employee

      Hello,

       

      Try setting L to a smaller value and try again. The L value is probably too large compared with the mesh.

      With the Lagrangian method, elements get deleted as failure occurs and so it will be difficult to see chips being formed and fly away. It is probably best to use the particle method if you want to see chips.

      Have a look at dynalook.com and search for metal cutting papers. We don't provide material input parameters. It is your responsibility to get them. You need parameters to model material failure. You can test the material yourself or maybe find something in the literature. 

      At last, as an Ansys employee, I am not allowed to download your model. You can still post it here so that other users can help you.

      Let me know how it goes.

       

      Reno.

      • javat33489
        Subscriber

        I will look there again. The papers on this site do not specify how to set up material failure. Can you tell me what card it's configured in? I will do the settings myself. I just need to know which cards to look at. In this case, I myself will bear the responsibility, I will do many tests and compare them.

    • Reno Genest
      Ansys Employee

      Hello,

      There are many ways to include failure. Many material models (*MAT_015 included) have material failure built-in. For metal failure, plastic strain at failure is commonly used. In *MAT_015, you can have failure strain a function of triaxiality (state of stress), strain rate, and temperature using the D parameters.

       

      You don't need to have such a complex failure to begin with. *MAT_024 is commonly used to model metals and you can define a simple plastic strain at failure using the FAIL parameter. Once you get it working, you can add complexities (triaxiality, strain rate, or temperature effects) to be more accurate in your failure prediction. Here is a good video on *MAT_024:

      https://www.dynamore.de/de/download/presentation/2020/copy_of_dynamore-express-good-old-mat_024-a-review-of-ls-dyna2019s-most-popular-material-model

       

      Also, you will find some examples of material models (*MAT_224; tabulated Johnson-Cook) for some metals used in the aerospace industry on the Aerospace Working Group:

      https://awg.lstc.com/tiki-index.php?page=Material+Parameter+Sets

      You will find the LS-DYNA parameters (.k file) and also reports on how they developed the material model. This is very instructive.

      At last, you could use *MAT_ADD_EROSION or *MAT_ADD_DAMAGE_GISSMO to add failure to your model. In the end, you want to be able to reproduce experimental test data. If you are able to achieve it, then your model should be ok. 

       

      Let me know if this helps or not.

       

      Reno.

      • javat33489
        Subscriber

        I have already tried using MAT_24 and got the same result. The material was removed. I'll try again. Thank you.

        Most of the papers and tutorials contain MAT_15. There are a lot of ready-made experimental data for him. I'd like to understand how to set it up to display shavings.

        Shavings should appear without *MAT_ADD_EROSION or *MAT_ADD_DAMAGE_GISSMO maps, they are additional to clarify the calculation.

        Thank you for your responses.

      • javat33489
        Subscriber

        My johnson cook had to work. But it does not work, there are no chips.

    • javat33489
      Subscriber

      I did a test with MAT_24, the material is not removed, only crumpled:

      MAT_24:

    • javat33489
      Subscriber

      I tested different options for materials 024, 003, 015, 018. But the material is always removed

    • javat33489
      Subscriber

    • javat33489
      Subscriber

      Result

       

      or

    • Reno Genest
      Ansys Employee

      Hello,

      With traditional FEM and failure, the elements get removed. So, most likely you will never see chips being created unless you are lucky and remove big chunks of material at a time. Element will get deleted when material failure occurs, so what you observe is what is expected.

      I would try with the SPG method:

      https://www.lstc-cmmg.org/cutting

       

      Let me know how it goes.

       

      Reno.

       

       

      • javat33489
        Subscriber

        It's a pity. I wanted to get chips in Lagrange

    • Reno Genest
      Ansys Employee

       

      Hello,

      The only way to get chips with Lagrange  mesh is to create  big chips; the tool has to be at least under the first row of elements:

      –wQJYU

      Otherwise, the particle method is a better choice.

       

      Reno.

       

       

      • javat33489
        Subscriber

        The first video is not working. The second video is an excerpt, the whole video costs 150 euros from the author.

    • javat33489
      Subscriber

      Can you tell a typical setting for any steel. For the MAT_ADD_EROSION card? I'll do some tests to see what role the card plays.

Viewing 10 reply threads
  • You must be logged in to reply to this topic.