Monte Carlo


Standard factorial designs (Tornado, Plackett-Burman, Box-Behnken, Central Composite Face etc.) limit each of the uncertainty variables to having a small number of distinct values, typically three levels of low, middle and high values (or -1,0,+1).

In a Monte Carlo design, rather than using a small number of distinct values, we assign a probability distribution (and related parameters) to each uncertainty variable.

We then decide on the number of experiments (or in our case simulation decks) we are going to generate. For each of these Rezen then randomly selects a value for each uncertainty variable based on the assigned probability distributions.

Also look at the Latin Hypercube algorithm which is a constrained Monte-Carlo algorithm.

Note that because the generated variable values are combined randomly there is an implicit assumption that there are no or at least minor interactions between variables, e.g. a higher value of one variable can be combined realistically with a lower value of another.