GB tracking in Phase Field simulations

[stnfwds]

Hello! I'm a new user to MOOSE(and simulation environments in general), and I've been getting hang of using the Phase Field module through the provided examples.

The Polycyrstal/GrainTracker systems make use of BndsCalcAux to track grain boundaries with a value threshold, but I was wondering what the best way is to narrow down the specific location of a boundary on each step for e.g, calculating GB velocity.

Relatedly, I'm also interested if there's any algorithms that could convert Diffuse Interfaces into more discrete ones, and vice versa.

[GiudGiud]

Hello

, but I was wondering what the best way is to narrow down the specific location of a boundary on each step for e.g, calculating GB velocity.

Do you need a point? A subset of the boundary?

Relatedly, I'm also interested if there's any algorithms that could convert Diffuse Interfaces into more discrete ones, and vice versa.

@dschwen @laagesen

[dschwen]

A grain boundary velocity field is best calculated using v_i = deta_i/dt / (|grad(eta_i)| + epsilon) (epsilon for regularization). This gives you a GB velocity map for the entire domain (for GB order parameter i). You can combine those from all OPs, but you need to take into account that the velocity is taken in the normal directon of teh OP gradient, and that neighboring grains cancel out unless you do (v = (|v_i| + |v_j|) / 2

[dschwen]

If however you only want the velocity of a specific point on a GB you could also use a FindValueOnLine postprocessor. I'd apply that to the grain order parameter and take 0.5 ad the value on the GB center. Together with ChangeOverTimestepPostprocessor, TimestepSize, a ParsedFunction and FunctionValuePostprocessor to make teh computed function value available as a PP output you can also compute a GB velocity.
Example here: https://github.com/idaholab/magpie/blob/devel/examples/polar_pfm/verification/velocity_theta.i

[dschwen]

It all depends on your end goal (which I encourage everyone asking questions here to state clearly to avoid the XY problem.

[stnfwds]

Thank you so much for the reply! At the time of writing the question, I was not aware of any calculation methods of the GB velocity field nor the existence of the GrainBoundaryVelocity User Object - both of which helps immensely with my immediate purposes.

The 'location of point on GB' and 'diffuse to discrete' question was related to a separate query I had about coupling the Phase Field model to other models that use discrete boundaries, and I might have unknowingly conjoined the two issues in this post. Will definitely look into the example though, it definitely looks helpful for this issue.

[laagesen]

As far as computing a discrete interface from a diffuse interface, the most common method for doing this is the marching cubes algorithm: https://en.wikipedia.org/wiki/Marching_cubes. I believe this is what is used in for example the Paraview contour filter, which allows you to extract a polygonal surface that lies at a particular value of the diffuse interface. If you use this functionality in Paraview you can perform further operations/calculations on it, such as calculating surface areas (this can also be done in MOOSE from the diffuse interface 3D data directly, but since you asked about a discrete surface representation...)