As with all Monte Carlo simulations, you need to first determine which stochastic process adequately represents the uncertainty, then calibrate the parameters of the stochastic process. After, you repeatedly perform simulations of your project using random draws of the stochastic process.
For a simple example, imagine a project that involves three activities A, B and C. Assume that activity C depends on the results from activity A and B, and the time required the activities follow the distributions D_A, D_B and D_C. Then repeatedly pick values from each of the distributions, e.g. t_A, t_B and t_C, and simulate the time required for project completion: t = max(t_A, t_B) + t_C.
Not sure if that answers your question, it was a little short on details :)
Monte-Carlo simulations can be used in dynamic project scheduling to measure the sensitivity of project activities as described in Schedule Risk Analysis or to evaluate the accuracy of forecasting methods used in Earned Value Management.
A Monte-Carlo simulation run generates a duration for each project activity given its predefined uncertainty profile, as follows:
* Generate a random number from the interval [0,1[.
* Generate a number using a cumulative distribution function.
Two important distributions that you need to be able to sample from in project scheduling simulation are the PERT distribution, based on the Beta distribution and the Triangular Distribution.
If a project is in three stages A, B, and C, you need to estimate the likely shortest duration for each step, the likely longest duration and the likely modal duration for each step. Once this is done you can build either 3 PERT distributions or three Triangular distributions in order to do a Monte Carlo simulation of the whole project.