17 Dec 05 11:19
Actually this is my first project that monte carlo analysis is required. Initially I had mix feelings on the usefullness of the analysis. We had performed the analysis twice, 1st one during the basic engr design. The 2nd one, just recently during the detailed engr design.
To me, monte carlo analysis is just another tool in managing the risks. It can be schedule, cost, quality or any other risks. In our case, we used monte carlo to manage the schedule risk. I believe the process of doing this, will be similar with other risk as well.
For sharing purpose below are the activities that we performed prior to and after the analysis:
1) Extablish a level 3 project schedule (fyi we use Primavera 3.0). There are about 2000 separate activities with 1000 linkages in the schedule...from start to completion of the project.
2) The schedule must be validated by the respective discipline engineers. This is important to ensure the correct linkage has been established.
3) Run the critical path analysis. Definition of critical path is zero float or nearly zero float. From this critical path analysis, we get about 100 activities considered "critical" by P3.
4) Next is to establish the 3 scenarios (optimistic duration, most likely duration, pessimistic duration) for each of the 100 activities.
5) We believe the effective way to come up with this scenarios is to carry out a multi-discipline review about 5-10 discipline engrs. In our case, we conducted a half day review session....go activity by activity and finally agreed on the optimistic, most likely and pessimistic duration.
6) During the meeting we also developed a document called risk register. The objectives of this doc are to identify the potential risks, assess the risk (H,M,L) based on its consequence + likelihood, propose mitigations, assign the person in charge and due date. The document is a life document. The status will be reported to the management.
7) Once having the information on the scenarios, the next thing is to run the software. We used PERTMASTER. The end result is the probability of achieving the target date. This will be presented in terms of P10, P50, & P90.
How useful is the analysis? I personaly believe it's useful, however it depends on the inputs i.e. garbage in garbage out. The basic schedule must be right and the persons involved during the review session must have adequate experience to determine the optimistic, most likely and pessimistic.
Definitely having monte carlo analysis is much better than based on our guts feeling...it takes time though...