Page 1 of 1
Does the number of iterations matter depending upon the size of the project?
Posted: Tue Mar 20, 2018 12:43 pm
Should number of Monte Carlo iterations defined in project settings be correlated with the number of tasks in the project schedule? For example, if number of task increases, number of simulations also needs to be increased.
Re: Does the number of iterations matter depending upon the size of the project?
Posted: Tue Mar 20, 2018 12:44 pm
No, there is no correlation between the number of iterations and activities. The number of iterations is more dependent upon things such as risk probabilities and complexity of the precedent network.
For example if I have a black swan risk (low probability, high impact), I want to ensure that it is accounted for in the simulation. So if it has a 1% chance of occurring, that means it will only occur 1 in 100 iterations; therefore, I will set my number of iterations higher (e.g. 2000) to ensure it occurs enough times that I get a clear understanding of the consequences of that risk occurring.
The other case could be in schedules with many risks, lots of parallel activities, and perhaps branching (conditional or probabilistic). In these cases, there can be multiple critical paths developing depending on which risks occur, which branches are taken, etc. In these cases, I would again set a higher number of iterations to ensure that every possible outcome is accounted for.
However, in most cases enabling the convergence monitoring will be sufficient and you won’t see any significant changes to your results by running additional iterations. I would recommend that you experiment a bit with these settings to see how they impact your results and you’re your decisions based on that. Convergence monitoring is set in the Calculations options.