THESIS
2016
xi, 150 pages : illustrations ; 30 cm
Abstract
This thesis addresses three issues arising in the ocean transport industry. The first essay
studies the cost-speed conundrum which confronts seasonal-product shippers. We consider
a newsvendor-type shipper who transports and sells seasonal products to an overseas market
where the selling price declines over time. Multiple shipping services are available with
different (uncertain) arrival times and freight rates. Our analysis reveals that a portfolio
of shipping services has twofold benefits in mitigating both demand and arrival-time uncertainties. We first derive the optimal portfolio by exploiting the special structure when
vessels’ arrival sequence is deterministic, and then propose an iterative approximation procedure
to solve the general problem with uncertain arrival sequenc...[
Read more ]
This thesis addresses three issues arising in the ocean transport industry. The first essay
studies the cost-speed conundrum which confronts seasonal-product shippers. We consider
a newsvendor-type shipper who transports and sells seasonal products to an overseas market
where the selling price declines over time. Multiple shipping services are available with
different (uncertain) arrival times and freight rates. Our analysis reveals that a portfolio
of shipping services has twofold benefits in mitigating both demand and arrival-time uncertainties. We first derive the optimal portfolio by exploiting the special structure when
vessels’ arrival sequence is deterministic, and then propose an iterative approximation procedure
to solve the general problem with uncertain arrival sequence. In each iteration of our procedure, we only need to minimize a cost function approximated by a deterministic
arrival schedule and the portfolio generated can converge to the optimal one under mild conditions.
In the second essay, we analyze the joint decision of empty container repositioning and
pricing from the perspective of ocean liners. We develop a stylized Markovian decision
model consisting of two locations with stochastic and endogenous shipping demand. With
the help of L
♮-concavity, we explicitly characterize the structure of optimal control policies,
which sheds light on the interplay between the pricing decision and empty container
repositioning. As opposed to single-location inventory systems, we find that the allocation
of empty containers may oscillate between two different base-stock levels, depending on
the direction of demand imbalance.
In the third essay, we extend the classical berth allocation problem (BAP) by incorporating
vessels’ uncertain arrival and service times. A distributionally robust optimization
framework is developed, which enables terminal operators to optimize the berth schedule
based only on partial information about random variables. The stochastic BAP problem
under our framework can be exactly reformulated into a mixed integer conic programming.
In particular, given the mean and the variance of each random variable, our reformulation
reduces to a mixed integer second-order conic program, which can be further approximated
by a mixed integer linear program with the same number of binary variables as the deterministic
counterpart. Hence, our framework is able to address stochastic BAPs without
imposing much additional complexity compared to the deterministic model. In addition, an
extensive numerical study based on a real dataset from Hongkong International Terminal is
reported.
Post a Comment