interface mobility and energy

ripening phenomena, dislocations, grainboundary topology
Post Reply
khajezade
Posts: 17
Joined: Fri Feb 10, 2017 7:20 am
anti_bot: 333

interface mobility and energy

Post by khajezade » Tue May 01, 2018 12:36 am

Hi there,

I am doing simulation on anisotropic grain growth where I use misorientation to describe interface mobility and surface energy. For now it is 2D and everything is fine. I want to go toward the 3D simulations but I just want to estimate the amount of physical memory that I need for the 3D simulation ( to choose a feasible domain size and number of grains). I was wondering how do you store misorientations between pairs of grains inside the code. Do you make a look-up table for each grain or use function calls or just store grains and its neighbours in each time increment.

Thanks,

Ali

Bernd
Posts: 1148
Joined: Mon Jun 23, 2008 9:29 pm

Re: interface mobility and energy

Post by Bernd » Tue May 01, 2018 6:32 pm

Dear Ali,

Essentially there is no specific problem with 3D and misorientation. Given the fact that in MICRESS orientations of the grains typically don't change during simulation, it is only necessary to calculate and store once the model parameters for each grain pair. For example, whether two grains form a low angle grain boundary does not make a difference in 2D or 3D . This condition needs to be calculated only once, and there is only one (boolean) value which needs to be stored.

Another problem is when the number of grains is very high. Above about 10000 grains, there are so many pairwise interactions that memory could be exceeded. Then, categorization (i.e. restriction to a smaller number of discrete orientation values) can be the solution. However, the problem also exists without misorientations and is only slightly aggravated.

Bernd

Post Reply