crystal growth

has anyone else out there tried using grasshopper to interpret crystal growth?  i have used SDR for crystal tessellation but i want something more organic and less -hedral.  we are using the naica mine in mexico as reference (see attached pictures). 

what im working with now is a rectangular plane that was divided in 10x10 then copied in the z direction.  maybe if there is a way to shift random points in the lists so that maybe 4 crystals grow and converge into one point on the top surface or vice versa.