why don't you
use a random number generator function to determine if a cb will proc, pick from 0 to 99, if number < %cb then successful check for cb
to the same thing for the range of physical damage except use the number as % of range of damage variation
calculate cb damage accurately and add that to actual damage
iterate to next step until hp<=0
calculate time
have this calculated 10 or 20 or 100 times.
then when you have that done
post: min, max, mean, and std of the results
so at that point you'll have reliable, believable results
and there still won't likely be a considerable difference or anyone who cares
