Simulating 6 sigma CMOS device variations in Cadence Spectre

Hi all,

Could anyone comment (or share references) on simulating device variations in Cadence Spectre, particularly using the "VTA adder" methodology ? Thanks ~!