Other Scheduling Methods - Construction Project Scheduling and Control - page 308

PERT and the Construction Industry
Even though PERT can be a good tool for predicting event dates, it is rarely, if ever,
used in the construction industry. There are four reasons for this:
1. The main area of effective application for PERT is experimental projects or
those with a great deal of uncertainty with regard to durations. In construction
projects, even if the project design is new, the basic elements of the construc-
tion operations are usually well known. Project managers usually feel com-
fortable with the estimated durations they provide. A good project manager's
estimates may be slightly over or under for the duration of individual activities,
but the overall duration of the project should be fairly accurate.
2. As shown previously, PERT focuses on a single path. The scheduler must
study different paths and analyze them for the highest uncertainty (least like-
lihood) of meeting a specific finish date. This process may be complicated and
time-consuming.
3. Project managers may have a problem providing three durations (optimistic,
most probable, and pessimistic) for each activity. During a discussion with
the author, one project manager sarcastically said, “We can barely come up
with one duration per activity, and you want me to provide you with three
durations?”
4. Many project managers adopt the practice of adjusting and correcting the dura-
tions (and possibly the logic) while updating the schedule. Even though we are
required to do such adjusting and correcting while updating, this practice must
not be an excuse for poorly prepared schedules. The project manager should
always remember that the project schedule could be a legally binding docu-
ment, as part of the contract. In such a case, it may be used in court later to
prove or disprove a delay claim.
PERT and Computer Project-Scheduling Software 3
Several computer software programs perform Monte Carlo simulations , generating
random numbers that follow the distribution of individual activities (optimistic, most
likely, and pessimistic, along with their weights). The program generates results for
each run. It is typical to do 100 runs or more. The more runs you do, the more
confidence you will have in the results. Some products run Monte Carlo simulations
in a “generic” environment. Crystal Ball (Decisioneering, Inc., Denver, Colorado)
performs risk analysis by using Monte Carlo simulations in MS Excel. Other products
3 There is more discussion on PERT (three-point estimate) and its applications in risk management in
Chapter 14.