Analysis of Time Series Forecasting Methods for the Airlift of Palletized Sustainment Cargo

摘要

In the Department of Defense (DOD) supply chain for sustainment cargo, over 45,000 pallets were transloaded at Incirlik AB, Turkey in from April 2008 to December 2009 with ultimate destinations in the USCENTCOM AOR, making Incirlik the largest transshipment node for airlift in the DOD by over two-fold. Current methods of forecasting follow-on transportation requirements are based on the amount of pallets on hand at the aerial port as well as short- term visibility of pallets currently in the channel system destined for Incirlik AB. This process yields a forecast horizon of only 2-3 days for planning subsequent airlift missions. An analysis several time series forecasting methods using post-sample pallet data from the Global Air Transportation Execution System (GATES) was accomplished to examine their ability to forecast different sustainment scenarios at least 7 days in advance of arrival at Incirlik AB, Turkey. Recommendations on how to implement forecasting methodologies in a general framework were proposed.