Effect of Assumptions about Cost Element Probability Distributions on Operating and Support Cost Risk Analysis.

摘要

This thesis sought to determine the effects of assumptions about cost element probability distribution on Operating and Support cost estimates. Currently, these assumptions are being made arbitrarily, without regard for their effect on the risk associated with a total cost estimate. The experiment was performed with a Monte Carlo simulation. Three cost elements, (fuel costs, maintenance personnel costs, and depot maintenance costs) were used as the sample space. Historical data from 41 fighter aircraft was used to determine the low, high, and modal values of each cost element. The Triangular, Beta and Gamma distributions were selected as candidate distributions for the cost elements. The low, modal and high values provided a means by which to parameterize the distributions for specific cost elements. The distribution for each cost element was varied over the set of candidates with all other factors held constant, samples were drawn and then summed to provide total cost estimates. The result was an array of 18 total cost distributions containing 1000 data points each. The variance of the 18 resultant distributions was the focus of the analysis. That analysis indicated that the cost element distribution chosen does have an effect on the risk of the total cost estimate. Further, distributions with infinite upper bounds result in consistently higher risk than those with finite upper bounds.