A Simulation Study of Evaluation Heuristics for Tug Fleet Optimisation Algorithms

Abstract

Tug fleet optimisation algorithms can be designed to solve the problem of dynamically positioning a fleet of tugs in order to mitigate the risk of oil tanker drifting accidents. In this paper, we define the 1D tug fleet optimisation problem and present a receding horizon genetic algorithm for solving it. The algorithm can be configured with a set of cost functions such that each configuration effectively constitute a unique tug fleet optimisation algorithm. To measure the performance, or merit, of such algorithms, we propose two evaluation heuristics and test them by means of a computational simulation study. Finally, we discuss our findings and some of our related work on a parallel implementation and an alternative 2D nonlinear mixed integer programming formulation of the problem.