Schedule Risk Management - Construction Project Scheduling and Control - page 373

into one of the common probability distributions of activity duration. One of these
probability distributions that is commonly seen and used is the normal distribution ,
which graphs into a bell-shaped curve, with the most likely duration occurring at
the highest peak of the curve and the less likely probabilities occurring as the curve
diminishes in both directions. An understanding of statistics is important in order to
use Monte Carlo analysis for risk management, especially when choosing the proba-
bility distribution and knowing how to evaluate the three-point estimates for use in
the analysis. There are several features in the Monte Carlo computer simulation that
produce interesting and useful results with the analysis, such as the ability to provide
global or filtered duration estimate spreads and the ability to adjust risk by activity
code. The Monte Carlo analysis provides statistically significant “confidence levels”
for the probabilistic prediction of completion dates and, since schedules deal with
unknowns, allows the schedules to have higher probabilities of meeting the chosen
completion date.
The Monte Carlo simulation analysis is available in numerous software packages,
such as PertMaster (acquired by Primavera Systems, which was later acquired by Oracle
and now is called Primavera Risk Management), which can be linked to the CPM
software by the same entity. There are also spreadsheet versions of Monte Carlo tech-
niques, such as @Risk, published by the Palisade Corporation. The use of three-point
estimates for risk analysis of uncertainties in duration originated in the late 1950s
during the U.S. navy's Polaris missile program. That program was called PERT (Pro-
gram Evaluation and Review Technique). 3 PERT was developed independently of,
and concurrently with, CPM methodology and used the three-point estimate system
to provide a weighted average duration for use in network calculations.
One important point to recognize about duration uncertainties is that the risk
from pessimistic durations' stacking and causing a delay during a project's execution is
greatly reduced when there is a robust and highly technical project controls effort used
on the project. When there is this technical controls effort, as soon as activities start
to miss their durations, the analysis and reporting raises a red flag about the problems
related to durations. With this strong reporting and identification of problems, the
project management team is in a stronger position to mitigate the time overruns from
missing durations before the time overruns stack up. So, the industry is moving away
from the need to analyze uncertainties in original durations and toward the analysis of
risk drivers, which are treated similarly to the specific event risks in the next section.
However, there are important and valuable results that come from a Monte Carlo
simulation. These results provide insight into the project and how the activities will
be performed in a variety of situations. The simulation can provide a list of activities
that most likely will appear on any or all the calculated critical paths and will display
the results in a prioritized list that often is called a tornado chart . Then the topmost
likely to be critical activities can be monitored more carefully, recognizing that many
3 Explained in detail in Chapter 11.