Fluids

Fluids

Dissimilar Materials laser welding simulation(2020R2)

    • Richard111
      Subscriber

      Hi all,

      I I've been trying this laser welding simulations for a long time, The results were very disappointing,I have set 2 phase with VOF model, air phase and mix metal phase with species model. The dome of model as showing in fig 1., The mesh Minimum Orthogonal Quality is 0.4.

    • Rob
      Ansys Employee
      Are you using the solidification and melting model?
    • Richard111
      Subscriber
      yes solidification and melting´╝î VOF and species model
    • Rob
      Ansys Employee
      What are the liquid and "solid" thermal properties? As it's a fluid model we'd expect some distortion around the free surface but what you're seeing is excessive.
    • Richard111
      Subscriber
      I tried both udf define and polynomial, Due to other parameters such as density, thermal conductivity and cp has little effect,
      like: conductivity of Ni
      if (temp >= 300 && temp < 1717)
      de2_value = 0.0212*temp + 39.439;
      else if (temp >= 1718 && temp < 3100)
      de2_value = 0.0209*temp + 16.705;
      else
      de2_value = 20;
      return de2_value

      and different value of viscosities and surface tensions in "solid" state are tried, such as very high(100000) or low(0.000001) viscosity, like:
      real viscosity;
      if (1717 <= temp <= 3100)//liquid
      viscosity = 0.88988*exp(-temp / 288.7727) + 0.01404*exp(-temp / 1014.5192) + 4.67366E-4;
      else if (3100 < temp)//gas
      viscosity = 0.00061;
      else
      viscosity = 100000.;//solid
      return viscosity
      DEFINE_PROPERTY(surf_tension, c, t)
      {
      real surf;
      real temp1 = C_T(c, t);
      if (temp1 >= 1714)
      surf = 1.0 - 0.0003*(temp1 - 1714);
      else if (temp1 >= 3100)
      surf = 0.41;
      else
      surf = 0.0000001;//
      return surf;
      }
      and U got the point, the distortion around the free surface in my model is abnormal, the mass transport between air and metal is not right, and it can spontaneous appearance.
      and warning like temperature limited to 1.000000e+00 in 12702 cells on zone 8 in domain 1, It doesn't make sense, I don't in what process cell needs to absorb energy and temperature reduce to 1K.


    • Rob
      Ansys Employee
      The limiter warnings tend to mean the solution is in the process of diverging, that can then trigger all sorts of daft results.
      Check the IF statements, what happens if T=1717.5K ? Drop the time step by a couple of orders of magnitude and see how it performs.
    • Richard111
      Subscriber
      I tried with time-step 1e-8, the temp only need 10 time-step to reach the temperature limited, I still think something wrong with free surface area, the material property of free surface between two metals is not right.

    • Rob
      Ansys Employee
      Can you post an image of the mesh? In the first images it looks OK, but in the last couple the cell aspect ratio looks high.
      Once the solution fails you may find odd artefacts in the contours as the plot range gets silly.
    • Richard111
      Subscriber
      Sorry to bother you at weekend, I tried different mesh, Structural Mesh or Unstructral Mesh, some mesh with high aspect ratio, but I tried a box mesh , and diverged and few time step(1*10^8)ÒÇé
      And, I just find out mass fraction and Liquid species mass fraction are different after patch, the liquid species fraction of metal shouldn't it be equal to 0?

    • Rob
      Ansys Employee
      It's not the weekend yet, some of us work on Friday afternoons! ;) I stopped being a student a little over 20 years ago....
      The mass fraction of the phases should make sense, but check what you initialised with and then patched. I'd initialise the domain using "standard" and all being air. I'd then patch the two "solid" regions along with the temperature. The scale on the right is 0 to 0.5 which suggests something got mixed up: you may need to patch to "zero" and then "one" for materials.
      For the mesh you want a smooth growth rate and low aspect ratio in the melt regions. You also want at least 5-10 cells across the melt region to capture the gradients.


    • Richard111
      Subscriber
      I think my patch step is right, I'm not sure that the Liquid species mass fraction at 300K is 0.5 is right or bug.
      I'm sure the cells across the melt region more 10 cellsÒÇé
      Bty, Is this really possible to simulate case like this? Including VOF, species and solidification and melting.
      Because I can find only one paper´╝î which is ÔÇ£Numerical simulation of alloy composition in dissimilar laser welding ÔÇØ byM.R. Nekouie Esfahani, they did a similar case, and I'm not sure they used species model, and the heat source is surface heat source. Some dissertation present similar case as well, by the joint mode is different´╝îwhich is lapjoint, it means that the heat source does not apply directly on both metal.
    • Rob
      Ansys Employee
      Melt region is vertical too - what do you expect to become liquid?
      The species looks odd, check what you're looking at as whilst we can model species in VOF and then melt them I'm not sure what the reporting looks like without building a model.
    • Richard111
      Subscriber
      I'm tying to get some result like this, mass transport in the welding seam.


      The computational logic should be like figure below left one, metal melting first then mix, But I think Fluent computational logic is mid one , which is mix species base on volume volume fraction or mass fraction, so is irrelevant to temperature, If I use Vof model only, then no mass transport.

    • Richard111
      Subscriber
      How to solve its problem?
Viewing 13 reply threads
  • You must be logged in to reply to this topic.