A probabilistic model for U.S. Nuclear power construction times

Construction time for nuclear power plants is an important element in planning for resources to meet future load demands. Analysis of actual versus estimated construction times for past U.S. nuclear power plants indicates that utilities have continuously underestimated their power plants' construction durations. The analysis also indicates that the actual average construction time has been increasing upward, and the actual durations of power plants permitted to construct in the same year varied substantially. This study presents two probabilistic models for nuclear power construction time for use by the nuclear industry as estimating tools. The study also presents a detailed explanation of the factors that are responsible for increasing and varying nuclear power construction times. Observations on 91 complete nuclear units were involved in three interdependent analyses in the process of explanation and derivation of the probabilistic models. The historical data was first utilized in the data envelopment analysis (DEA) for the purpose of obtaining frontier index measures for project management achievement in building nuclear power plants. The DEA results were used for adjusting the raw data for the purpose of eliminating the nonfrontier units from the data set. The raw and the adjusted data, then, were analyzed graphically in the MacSpin program to find the appropriate mathematical associations between the construction times and the various potential independent variables. The graphical analysis allotted the specification of the regression equations of the raw and adjusted data sets. Both models contain the same independent variables and have the same specifications. Finally, the normal linear regression of the SYSTAT statistical package was used to calculate the parameters of both probabilistic models. It was found that it is possible to use both models to accurately estimate the construction time of a nuclear power plant project. It was also possible to explain about 87 percent of the increase and variation of the construction times on past nuclear power plant projects.