Discussion Forum

This sounds interesting. Note: I don't know much about material modeling, so my questions here may be a bit naive. Can you share more about the physics you are studying? Are these "grains" individually fixed in terms of their size, orientation, position, and with fixed properties, or are they perhaps dynamic (like grains of magnetization in a magnetic material, e.g., see https://en.wikipedia.org/wiki/Magnetic_domain)? If they are all fixed, then does your EBSD data correspond to (or accompany) geometry information (including the geometry, position, and orientation of each and every grain) that can all be imported into Comsol Multiphysics? If not, then it seems to me that if every grain matters individually, then you might have to draw them in detail using either Comsol Multiphysics or another tool, to include all the grain boundaries. Maybe that is practical if there aren't too many grains? Also, it would seem (since you care about crystal structure here) that your grains are anisotropic, right? If so, then it would seem to me that to model induced currents, you'll need to specify a conductivity tensor (not just a scalar) for each grain. But perhaps... if the grains are numerous and you don't really care about the details of their tiny shapes, you could simplify the grain model to just a spatially-dependent conductivity tensor, and not have to break up your actual geometry into explicitly separate grains or domains. Tell us more!

p.s. There is some info about possibly related work available. See, for example, https://www.comsol.com/blogs/strategy-designing-corrosion-resistant-materials and https://www.comsol.com/blogs/material-characterization-by-means-of-simulation among others. Or search "grain" in the knowledge base at https://www.comsol.com/support .

If you have ONLY a grain orientation distribution then you still need to know something about the grain size and grain size distribution in order to create a geometry. And I don't see an obvious way in Comsol to create a volume-filling geometry of grains with a known distribution. Only then would you be in a position to assign orientations to the various grains. (This might actually be the easiest part as (if you look up Monte Carlo simulation) there are ways to get random numbers with a known distribution. Inside Comsol I have seen a way to use methods to create a geometry of random spheres (as in Swiss cheese) but this would not be volume-filling.

I am quite certain people have done this- create artificial random geometries- but it looks like a substantial project in itself.

I suppose you might try something simpler and more ideal- say an geometry of identical cubes assigned with random orientations.

This sounds interesting. Note: I don't know much about material modeling, so my questions here may be a bit naive. Can you share more about the physics you are studying? Are these "grains" individually fixed in terms of their size, orientation, position, and with fixed properties, or are they perhaps dynamic (like grains of magnetization in a magnetic material, e.g., see https://en.wikipedia.org/wiki/Magnetic_domain)? If they are all fixed, then does your EBSD data correspond to (or accompany) geometry information (including the geometry, position, and orientation of each and every grain) that can all be imported into Comsol Multiphysics? If not, then it seems to me that if every grain matters individually, then you might have to draw them in detail using either Comsol Multiphysics or another tool, to include all the grain boundaries. Maybe that is practical if there aren't too many grains? Also, it would seem (since you care about crystal structure here) that your grains are anisotropic, right? If so, then it would seem to me that to model induced currents, you'll need to specify a conductivity tensor (not just a scalar) for each grain. But perhaps... if the grains are numerous and you don't really care about the details of their tiny shapes, you could simplify the grain model to just a spatially-dependent conductivity tensor, and not have to break up your actual geometry into explicitly separate grains or domains. Tell us more!

p.s. There is some info about possibly related work available. See, for example, https://www.comsol.com/blogs/strategy-designing-corrosion-resistant-materials and https://www.comsol.com/blogs/material-characterization-by-means-of-simulation among others. Or search "grain" in the knowledge base at https://www.comsol.com/support .

Thanks Robert for your response. To answer your question, the grains are assumed to be known beforehand and as such have fixed size, orientation, position, properties. The EBSD data that I am trying to import contains all the information about the geometry, location, properties and orientation of each grain and that is what I am trying to import into COMSOL for my simulation. In specifying the conductivity tensor, as you mentioned, I believe it is more convenient to define a spatially-dependent conductivity tensor than specifying the tensor for the grains. I am still trying to understand some of the nuances involved in the simulation and I don't have a lot of information other than the knowing that I have to find a way to import the EBSD data into COMSOL. Thanks a lot for your help.

If you have ONLY a grain orientation distribution then you still need to know something about the grain size and grain size distribution in order to create a geometry. And I don't see an obvious way in Comsol to create a volume-filling geometry of grains with a known distribution. Only then would you be in a position to assign orientations to the various grains. (This might actually be the easiest part as (if you look up Monte Carlo simulation) there are ways to get random numbers with a known distribution. Inside Comsol I have seen a way to use methods to create a geometry of random spheres (as in Swiss cheese) but this would not be volume-filling.

I am quite certain people have done this- create artificial random geometries- but it looks like a substantial project in itself.

I suppose you might try something simpler and more ideal- say an geometry of identical cubes assigned with random orientations.

Thank you Dave for your response. Currently the EBSD data that I have is supposed to contain orientation information and the of each grain in the material. For the grain size and grain distribution, we are considering a pixel-based size information which can be translate to an euclidean coordinate to represent the mean location of the grains. We have also considered distribution information in the orientation data that we have. I thought (I am not certain) by this fact, Just importing the EBSD data is a sufficient to run the simulation and all necessary information can be extracted from this data.

Hi Ridwan Olabiyi,

I think the "combined coordinate system" might be what you are looking for.

BR, Manuel