I've done some work with monte carlo and discrete event simulation. The runs/iterations produced utilize psuedo-random number generation. Many discrete-event simulation software platforms reproduce exactly the same numbers on a given run. Also, monte carlo simulation software like @Risk and CrystalBall enable users to select psuedo-random number generation methods. For instance, utilizing the Latin Hypercube method will sample more evenly from a Gaussian distribution rather than pulling more from the peak and tails.