Constraining/Tying Thermal DOF of two surfaces in ABAQUS/Explicit

[mechdrive1]

Hello Everyone!

I am facing element distortion issue due to thermal conductance (high magnitude) between aluminum-steel surfaces.

The set-up of tool and workpiece is similar to friction stir-spot welding wherein tool is positioned on the top of two layers (aluminum & steel). The tool drills through the aluminum layer and interacts with the steel layer. Explicit thermo-mechanical step has been used in ABAQUS. C3D8RT element are used for all parts. JC damage law has been used for removing highly distorted elements from the aluminum layer. Contact pressure depended thermal conductance is specified between the two layers.

I have noticed in my simulations that using thermal conductance is causing element distortion at the interface, with the distortion being extremely random i.e. sometimes it occurs in aluminum layer and other times it occurs in steel layer. I was able to complete the simulation without specifying thermal conductance till the end but this has led to loss of thermal interaction at the interface of two layers which is crucial from thermal effect point of view.
I have two ideas to come around element distortion issue and also include thermal interaction:

1) Ensure that the two surfaces have the same temperature throughout the process by using a constrain.
2) Specify heat flux from aluminum layer to steel layer.

With respect to option 1, I have tried using 'tie' constrain which ties translational/rotational/thermal DOFs but severe element penetration and subsequent distortion has been observed. I have looked at documentation for some other way to tie the thermal DOF but there's nothing regarding Explicit procedure. I could only find the 'equation' constrain for abaqus/standard procedures.

I would request you to provide any suggestions regarding both of these options as to how I can either tie only temperature DOF of the two surfaces and/or how can heat flux, which is already calculated by ABAQUS, be specified from one surface to another?

Thanks,
Ankit

 

[Mustaine3]

Take out the expansion coefficient and you'll see if the distortion really comes from thermal effect.

 

[mechdrive1]

Hello Mustaine3,

I've submitted a model as per your suggestion and will update with the results once available but I don't get the relation between coeff. of expansion and element distortion? Can you please explain the same?

Thanks,
Ankit

 

[Mustaine3]

You said that deformation/distortion is the problem. This is a mechanical effect. Only few thermal behavior affect that - expansion is one of them.
Without expansion you'll see if the issue still exists. If yes, then the thermal behavior is not responsible.

 

[mechdrive1]

Element distortion is my primary concern. I've read in some research paper, although no proper supporting evidence is there, that high magnitude of thermal conductance can cause hourglassing and element distortion issues. I reduced the magnitude of thermal conductance and noticed that the model could run for longer duration. Having completely removed the thermal conductance, the model is able to run till the final step.

I am just wondering if the very small magnitude of expansion coeff. can cause such an effect on an element.

 

[mechdrive1]

Mustaine3,

I submitted a model without expansion coeff. and the result was similar. Element distortion occurred in the model and this one couldn't complete the first step itself. (other models were able to complete until the middle of 2nd step).

Any suggestions?