Optimising quay crane operations based on data-driven cycle time prediction
A case study at APM Terminals MVII
L.V. Gerlach (TU Delft - Civil Engineering & Geosciences)
Bilge Atasoy – Graduation committee member (TU Delft - Transport Engineering and Logistics)
Jaap Vleugel – Graduation committee member (TU Delft - Transport and Planning)
R. Negenborn – Graduation committee member (TU Delft - Transport Engineering and Logistics)
F.A. Dekker – Graduation committee member (APM Terminals)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Quay cranes on modern container terminals have the possibility to lift two 40ft. containers side by side. In order to be able to perform this type of lift, the crane must be equipped with a tandem spreader. As this type of spreader cannot be used for performing certain lifts, such as the handling of hatch covers, switches between single spreader and tandem spreader are inevitable. However, there is no method in current literature nor in practice to determine when the spreader changes should be performed. Therefore, the aim of this research is to develop a model that calculates the optimal moments to switch spreaders. As the performance of tandem lifting depends on the cycle times, a cycle time prediction model is developed first. This Artificial Neural Network model predicts the cycle time based on the type of lift, the actual position of the container(s) on the ship, the weight of the load and the current wind speed. The predictions yield a Mean Absolute Percentage Error of less than 11%. Afterwards, the cycle time prediction model is used to develop an optimal spreader switching strategy model in the form of a Mixed Integer Linear Programming model. A case study, in which several different bay layouts of a container ship need to be handled, shows a decrease in handling time of up to 22% compared to the traditional spreader switching strategy.