Monte Carlo Analysis
Monte Carlo Analysis follows on nicely from my blog on PERT Analysis last week. As the title suggests, we are going to take many trips through our critical path to see how lucky we are. Monte Carlo Analysis uses the range of durations for tasks with three duration estimates. It picks a random number to determine the calculated duration. How lucky do you feel today?
How is Monte Carlo Analysis used?
Monte Carlo is a Risk Analysis method. In particular, it helps to assess the risk of a project not meeting it’s scheduled end date.
Estimating the duration of any task is difficult. The PERT Analysis blog suggested estimating three durations:
- The Optimistic
- Most Likely and
- The Pessimistic durations
Three estimates can give an idea of the confidence in the estimate based on the range between the optimistic and pessimistic estimates. If these are close, then there must be some confidence. Perhaps they are far apart, then this shows a lack of confidence.
Special software is then used to run through the project with a random set of durations (between the optimistic and pessimistic limits). After running through at least 50 times (a 1000 times provides a smoother (but not more accurate!) result) a distribution of project end dates can be presented.
In the above example, the project end date varies between 31st March and early September. We can use the cumulative probability to make a forecast:
- We are 90% confident of finishing before the end of July.
Monte Carlo Analysis can also be used to model the costs of a project.
Problems with Monte Carlo Analysis
It can often be difficult to get one accurate estimate. If the Project Manager is not careful the estimate may be based on a simple guess. If getting one accurate estimate is difficult, why should getting three estimates deliver a better result?
Special software is required. Although the three point anlaysis was available in Microsoft Project up to and including Project 2007, it was removed in Project 2010. You can use the custom fields in Microsoft Project to create your own columns for Optimistic and Pessimistic duration, but you need to know how!
All Project 2007 (and before) provided was the option of showing a pessimistic, most likely and optimistic Gantt chart. (The pessimistic to be given to the customer perhaps.) Full Monte Carlo Analysis has always needed special software.
In addition to the software, the method requires a project where the range of durations can be estimated accurately. And a project scheduler who can operate the software and also interpret the results.
Real life examples are hard to come by. My picture above comes from a 2018 APM publication. However note that the example is over 30 years old! I recently supervised a student on a research project looking at the practical applications of Monte Carlo in Small and Medium Enterprises (SME’s). Their conclusion was that although it is known about, the special software and skills to use it make it impractical for SME’s.
Monte Carlo Analysis might be a nice to have detail on a large project. However for smaller projects, the time spent might better be used in communication, talking to the project team, or stakeholders.