Mathematical modelling and efficient algorithms for autonomous straddle carriers planning at automated container terminals

Publication Type:
Issue Date:
Full metadata record
Files in This Item:
Filename Description SizeFormat
01front.pdf128.56 kBAdobe PDF
02whole.pdf3.89 MBAdobe PDF
In the past several decades, automation of handling equipment has been a worldwide trend in seaport container terminals. Increasing automation of yard handling vehicles not only reduces the cost of terminal operation, but also increases the efficiency of container transport. However, the primary loss of performance in the transhipment process is caused by the uncoordinated allocation and scheduling of quay cranes, yard vehicles and land-side operations. Hence, integrating transhipment processes is imperative for a fully automated container terminal. This thesis aims to study an integrated process and develop practically deployable strategies and algorithms, with the practical example of the Patrick AutoStrad container terminal, located in Brisbane, Australia. The thesis first formulates two mathematical models: The Comprehensive Model is an analytical optimisation model which integrates the quay-side, yard and land-side operational sub-problems of the Patrick AutoStrad container terminal. Derived from the comprehensive model, the Job Scheduling Model is formulated to focus on the optimisation of job scheduling, as job scheduling plays a more important role than path planning, and resource utilisation and port operation are more dependent on job scheduling. To solve the Comprehensive Model, a job grouping approach is proposed for solving the integrated problem, and experimental results show that the job grouping approach can effectively improve the time related performance of planning container transfers. Solving the Job Scheduling Model using a global optimisation approach is expected to provide higher productivity in automated container terminals. Hence, a modified genetic algorithm is proposed for solving the job scheduling problem derived from the integrated mathematical model of container transfers. Moreover, the live testing results show that the proposed algorithm can effectively reduce the overall time-related cost of container transfers at the automated container terminal. Last but not least, a new crossover approach is proposed in order to further improve the solution quality based on the modified genetic algorithm, and it can also be directly applied in solving the generic multiple travelling salesmen problem using the two-part chromosome genetic algorithm. The experimental results also show that the proposed crossover approach statistically outperforms the existing approaches when solving the job scheduling problem and the standard multiple travelling salesmen problem.
Please use this identifier to cite or link to this item: