A notable exception, which has been curiously
ignored, apart from a passing citation by Ronen9, is
that of Brown et al.10, whereby a ship multiple speed
level mix is selected a priori, whilst defining the time spent at each
selected speed level as a decision variable, resulting in set of
decision variables and a linear programming model. Effectively, in this
reformulation of the speed optimisation problem, a linear programming
model replaces the traditional nonlinear programming model, which
clearly is a significant advantage, both from a mathematical standpoint
and a computational point of view. However, there does not exist a
quantitative assessment of the results obtained for the same speed
optimisation problem variant instance. It is precisely the objective of
the work presented in this paper to provide such an
assessment.