Schedule Risk Management - Construction Project Scheduling and Control - page 378

Another important practice is to limit the use of relationships among activities to
mostly finish-start relationships, with minimal use of lags. 5 This improves the ability of
the Monte Carlo analysis software to provide accurate results from those calculations.
Some Monte Carlo software programs do not calculate accurately with negative lags,
so those should be minimized in any schedule that is intended to be adjusted for risk
through a Monte Carlo analysis.
In addition, the level of detail in the schedule is an important requirement for
using risk analysis. There must be enough detail in the schedule to isolate each activity
by responsibility. This means that, for a schedule to be used in risk analysis, an activity
should only model one trade. A rough-in activity should be broken down into separate
activities for mechanical, electrical, plumbing, and fire sprinkler work. The rough-in
work should also be separated into wall rough-ins and ceiling rough-ins, and by area
designation plans if those areas tend to progress differently. This allows an evaluation
of each trade for different assumptions of risk, as well as an evaluation of areas for
different risks. For example, if a plumbing subcontractor shows up on the risk register
after the workshop as a high risk because of his or her poor reputation, then a different
factor for risk or a different spread of duration estimates can be used to account for
that risk. On a very simple level, without Monte Carlo analysis, a time contingency
could be added to the plumbing activities to allowmore time than predicted, especially
on those activities that are vital to opening up other significant areas of work.
The use of constraints can affect the validity and accuracy of risk calculations
because constraints may sequester float or prevent the network from calculating cor-
rectly. It may be necessary in the risk analysis to remove constraints in order to see the
impact of those constraints on the project's completion. The fewer constraints there
are, the easier it is to incorporate risk management. In addition, the use of manda-
tory constraints that do not allow the network to calculate float accurately in either
direction, delay, or acceleration is problematic and might render the schedule an inap-
propriate candidate for a risk analysis.
Just as in any analysis of a schedule, the more calendars that exist in the schedule,
the harder it is to analyze the schedule, whether it is for delay or risk. Multiple calendars
with many shifts from calendar to calendar along critical paths will amplify or reduce
the total float values. This can affect the risk analysis and, at the least, can require
additional work on the analysis.
The risk management plan is based on CPMnetwork calculations, whether you are
using Monte Carlo analysis or what-if scenarios. So, if the network does not calculate
correctly, the value from risk management is severely reduced. This means that one of
the generally accepted scheduling best practices, the limitation of open-ended activities
to the first and last activities, is vital to risk management analysis.
With a Monte Carlo analysis, the iterations are run with multiple values of activity
durations, so those durations have to be reasonable, comparable, and ideally limited to
5 Overlapping activities can be accurately represented by splitting the predecessor. See the discussion in
Chapter 5 and Figure 5.8.