Estimating the determinant factors of container dwell times at seaports

Abstract

Numerous options are being considered in practice, and have been captured in theoretical research, for increasing container terminal capacity. These include new stacking and handling technologies, optimizing yard space allocation and creating empty container depots outside of terminals to mention a few. Among these, reducing the amount of time a container spends at the terminal, container dwell time (CDT), may prove to be one of the least costly solutions. For this strategy to be successful, it is essential that terminal operators be able to define factors impacting the CDT and estimate how long a container remains in the yard. This article attempts to identify determinant factors of CDT and delineates appropriate computational tools for estimating CDT based on a set of such factors on which terminal operators typically collect relevant data. The article compares the performance of three data mining algorithms to estimate CDT: Naive Bayes, decision tree and a NB-decision tree hybrid. Using the best performing model, sample terminal data is used to measure how changes in CDT determinant factors impact CDT, yard capacity and terminal revenue. The outcomes reveal that the impacts of changes in CDT determinant factors can be fairly considerable in order to affect terminal's capacity and revenue earned from demurrage fees.