Analysis of Forecasting Models Applicable to Requirements Determination for Department of Defense Petroleum Products

摘要

The current DOD petroleum requirements determination process involves manual computation and analysis by the services in determining forecast quantities for procurement by the Defense Fuel Supply Center (DFSC). However, these forecast requirements have not been as accurate as DFSC would like. The purpose of this thesis was twofold: to provide a general description of the requirements generation process and to propose an alternative forecasting method that would more accurately represent and predict future requirements. The alternative forecasting method analyzed was a computerized program named SIBYL-RUNNER. Past JP4 and JP5 consumption data from several Air Force and Navy locations was collected and subsequently analyzed using SIBYL-RUNNER to produce annual forecasts. These computer-generated forecasts were then compared with the initial forecasts provided by the services to determine which method more accurately predicted actual consumption. The overall results indicated that SIBYL-RUNNER provided more accurate results in a majority of the locations and time periods analyzed. Several recommendations were subsequently made by the authors to enhance the services' forecasting capabilties. (Author)