Increasing international maritime transport drives the need for efficient container terminals. The speed at which containers can be processed through a terminal is an important performance indicator. In particular, the productivity of the quay cranes (QCs) determines the performance of a container terminal; hence QC scheduling has received considerable attention. This article develops a comprehensive model to represent the waterside operations of a container terminal. Waterside operations comprise single and twinlift handling of containers by QCs, automated guided vehi- cles and yard cranes. In common practice, an uncoordinated scheduling heuristic is used to dispatch the equipment operating on a terminal. Here, uncoordinated means that the different machines that operate in the container terminal seek optimal productivity solely considering their own respective stage. By contrast, our model provides a coordinated schedule in which operations of all terminal equipment can be considered at once to achieve productivity closer to the QC optimal. The model takes the form of a hybrid flow shop (HFS) with novel features for bi-directional flows and job pairing. The former enables jobs to move freely through the HFS in both directions; the latter constrains certain jobs to be performed simultaneously by a single machine. We solve the coordinated model by means of a tailored simulated annealing (SA) algorithm that balances solution quality and computational time. We empirically study time-bounded variants of SA and compare them with a branch- and-bound algorithm. We show that our approach can produce coordinated sched- ules for a terminal with up to eight QCs in near real time.@en

The range of applications that require processing of temporally and spatially distributed sensory data is expanding. Common challenges in domains with these characteristics are sound reasoning about uncertain phenomena and coping with the dynamic nature of processes that influence these phenomena. To address these challenges we propose the use of causal Bayesian Networks for probabilistic reasoning and introduce the Logical OR gate in order to combine them with dynamic processes estimated by arbitrary Markov processes. To illustrate the genericness of the proposed approach, we apply it in a wildlife protection use case. Furthermore we show that the resulting model supports modularization of computations, which allows for efficient decentralized processing.@en