Optimization Modeling For Container Loading And Unloading Operations At Maritime Terminals

Modeling and optimizing loading and unloading operations at maritime terminals have increasingly become more important to container terminal operators if they need to remain competitive. This trend is fuelled by the recent growth in international waterborne trade and developments in shipping, cargo handling and hinterland transportation. In addition, the rising of inter-port and intra-port competition has put substantial pressure on terminal operators to improve the container terminal performance. A well managed maritime terminal strives to improve terminal production, terminal productivity, terminal utilization and terminal service quality measures. Nonetheless, of all the terminal performance measures, terminal service quality measures (i.e. ship turnaround time, operational dwell time) are key factors for customer service. In most terminals, the vast portion of the ship turnaround time is spent on unloading and loading containers.Nevertheless, a container terminal represents a complex system with highly dynamic interactions among the three operations of the container terminal; quay operation, transfer operation, and yard operation. The very large number of decisions involved in planning the terminal, the multi-objective nature of the problem, uncertainty, and the complexity of decisions, render it impossible to make informed decisions without the assistance of optimization modeling.This thesis presents an optimization model for loading and unloading operations at maritime terminals (i.e. HFSPLUOMT). The HFSPLUOMT is a6-stage hybrid flowshop scheduling problem with identical parallel machines in each stage, set-up time constraints, blocking constraints and precedence constraints. The objective of the HFSPLUOMT is to minimize the makespan (i.e. the completion time of handling the last container in a planning horizon). It is well known that multistage hybrid flowshop scheduling problem with makespan based criteria is NP-hard in the strong sense and even small size problems are tediously solvable. The HFSPLUOMT also is strongly NP-hard and hence heuristic procedures are employed to achieve near-optimal solutions. In this thesis the solution procedure, Global Schedule Construction Algorithm (GSCA), is composed of seven heuristics and Genetic Algorithm (GA). More specifically, the GSCA decomposes the HFSPLUOMT into five scheduling sub- problems and solving one at a time to get the initial container sequence. Lastly, GA is employed to improve the secured initial container sequence.In order to validate the developed model and determine the effectiveness of the GSCA, a lower bound is developed. This lower bound is the supremum of the two lower bounds, operation-based lower bound and lower bound with arrival and tail times. The parameters employed in numerical experiments are based on the data from Dar es Salaam Port-Container terminal, Tanzania. Numerical tests reveal that optimization modeling can solve the scheduling problem for container loading and unloading operations efficiently.