A commercial project management tool,
populated with a complete set of activity-related data, was used to
estimate a deterministic baseline project completion time for the PD-337
activity network. Low (optimistic) and high (pessimistic) activity duration
inputs were applied to the critical path method to generate a more “bounded”
duration range. Stochastic methods were applied to the low, most likely,
and high PD-337 activity duration inputs to estimate project completion
times as continuous probability distributions. The baseline approach,
called “the dependent version of the Project Evaluation Review
Technique (Dependent PERT),” enables an estimator to measure dependence
among time-related uncertainty distributions. Dependent PERT is similar
to the Extended Version of PERT (PERT), a method commonly used in today’s
commercially available software packages. Although activity durations
from both the dependent and independent methods are sampled using Monte
Carlo simulation techniques, dependent PERT estimates schedule uncertainty
based upon some prescribed measures of dependencies within an activity
network while PERT estimates schedule uncertainty only as a sum of statistically
independent uncertainty distributions.
The results and sensitivity analysis provided
evidence that Dependent PERT accounts for duration uncertainties better
than PERT. Unlike PERT, Dependent PERT showed that variations in the
statistical dependence of activities and variations in the uncertainty
of activities can have significant impacts on the uncertainty of project
completion times. An example was finally provided on how the ERAM group
might use the methods demonstrated in this study to make decisions early
on in the system evaluation process. Dependent PERT results provide
additional information to assist decsion-makers in assessing the potential
risks ERAM has on production schedule duration. Understanding the implications
of risk in a ship's production phase assists commercial shipyards in
improving strategies to become more competitive in the world market.
<<
Back Page 2 of 2
