Tugboat resting location optimization using AIS data analysis

In this thesis we investigate how well Automatic Identification System data can describe the nautical chain of the Port of Rotterdam and we use this data analysis to optimize the positioning of tugboat resting locations in the port. We find that the quality of AIS data is not flawless, however, it can accurately describe certain parts of the nautical chain, such as locations where pilot tenders and tugboats meet deepsea vessels to start or end service and service durations. Because of missing data, we find that the AIS data cannot be used to describe interarrival times of nautical service requests accurately. Using the findings of this analysis, we then model the problem of positioning tugboat resting locations in the Port of Rotterdam as a two-stage stochastic problem with uncertainty in the demand and return parameters. We are able to find upper and lower bounds on the optimal objective value of the stochastic model and we find the optimal solution within reasonable computation time. We test the effect of changing to the optimal locations with a discrete-event simulation model, where we find that a yearly saving of almost e65,000 or 3.8% of total costs is achieved.