Constraint-based local search for container stowage slot planning

Due to the economical importance of stowage planning, there
recently has been an increasing interest in developing optimization
algorithms for this problem. We have developed 2-phase approach that
in most cases can generate near optimal stowage plans within a few
hundred seconds for large deep-sea vessels. This paper describes the
constrained-based local search algorithm used in the second phase of
this approach where individual containers are assigned to slots in
each bay section. The algorithm can solve this problem in an average of
0.18 seconds per bay, corresponding to a 20 seconds runtime for the
entire vessel. The algorithm has been validated on a benchmark suite
of 133 industrial instances for which 86% of the instances were
solved to optimality.