Mixed Models of Single-Berth Interarrival Time Distributions

Abstract

Modeling and analysis of the statistical patterns of vessel interarrival time is a crucial research area for port planning and is also the basis of port capacity analysis and simulation-related works using theoretical queuing methods. This study seeks to identify probabilistic models of vessel interarrival time for a single berth. Two generalized queuing models (GQMs), gamma-GQM and lognormal-GQM, were proposed and tested by a novel hybrid parameter estimation method. To test the proposed models, detailed records of vessels entering Tianjin Port between 2012 and 2014 were utilized; nine berths consisting of three types—bulk cargo, container, and crude carrier—were chosen and studied. The results of the study show that (1) single models are unsatisfactory for single-berth interarrival time modeling in most cases, whereas the proposed mixed models provide much better overall goodness of fit; (2) the statistical properties and the proportion of congested interarrival time are highly related to traffic intensity; (3) the lognormal distribution model performed the best of all tested single models, and the lognormal-GQM performed the best of all tested models.