These papers exclusively study empty repositioning of loading
units. However, none of them includes uncertainty or disruption
(at locations or on the routes) into account. Di Francesco, Lai, and
Zuddas (2013) study the effect of partial or complete port disruption
in empty container repositioning in a liner shipping system.
They model it as a time–space representation and consider a set
of different disruption scenarios. They also include some nonanticipativity
conditions to equalize the here-and-now decision
variables over all scenarios. Di Francesco et al. (2013) test all combinations
of a problem including 5 locations (2 hubs), a 50 period
rolling horizon scheme, 2 (normal and disrupted) scenarios, and 2000 customer orders. They show that in case of a normal scenario,
the optimal deterministic solutions are the best, but in case of a
disruption, the multi-scenario model produces the most effective
results. Di Francesco et al. (2013) show their model is able to handle
12 disruption scenarios with up to 20 location in less than one
hour.