Unfortunately this area is severely undocumented. I read once that windows uses the SSTF algorithm. But thats probably an outdated info. I only know this because I've been actually involved in programming such simulation for disk arm scheduling.
By the way, even if you find accurate, up to date info. They are NOT going to reflect the real algorithm applied on the disk, because the physical geometry (cylinders,sectors per track,heads) is hidden from the OS because of BIOS limitations. The OS is given a logical geometry.
