Strategic Maritime Preparedness Under Uncertainty - Using Sample Average Approximation to Locate Tugboats Along the Norwegian Coast

Abstract

The Norwegian Coastal Administration (NCA) is responsible for the maritime preparedness in Norwegian waters. By operating tugboats and monitoring the vessel traffic, the NCA aims to be well positioned against future drifting accidents. This requires extensive planning at the strategic, tactical and operational levels. Even though operation research and optimization models have been important tools for decision making within preparedness planning a long time, to our knowledge the NCA has not applied optimization techniques as strategic decision support.

The problem studied in this thesis is referred to as the strategic Tugboat Positioning Problem (TPP). It aims to find the optimal number of tugboats and the location that minimize the total cost. The total cost consists of tugboat investment costs and expected penalty costs due to insufficient preparedness. Furthermore, the problem is considered from the NCA's point of view.

A two-stage recourse model is developed to solve the strategic TPP. To ensure a minimal level of preparedness along the coast regardless of the realized scenarios, a minimal cover constraint is included in the model. The model is solved using the Sample Average Approximation (SAA) combined with a self-developed search heuristic. For a constrained version of the problem, a near-optimal solution is found.

The NCA's current strategy is compared with the best stochastic solution. Results show that the stochastic solution performs much better than the current NCA strategy. Further comparison of the two solutions reveals that the solutions have several similarities, as both solutions have the same number of tugboats and use similar port locations. However, the stochastic solution suggests a tugboat investment cost that is far grater than the current NCA strategy. This imply that the total cost is reduced significantly if the NCA invests in tugboats with higher bollard pull and sailing speed.

Even though several assumptions have been applied to develop the model, it is able to capture some important aspects. Hence, we believe that the model gives valuable information regarding strategic preparedness planning that should be further analyzed by the the NCA.