|
|
|
This research focuses on large-scale dynamic resource management
problems faced by freight transportation service providers. We focus
on two main themes: (1) fast solution of large-scale resource control
problems, and (2) practical methods for incorporating future uncertainty
into dynamic resource planning problems.
|
Robust Optimization for Empty Repositioning Problems
to appear, Operations Research, 2007.
(w/ J.C. Morales and M.W.P. Savelsbergh)
This paper develops a robust optimization approach
for dynamic empty resource repositioning problems with demand
uncertainty. Designed as a pragmatic approach, the methodology
is labeled "robust" since it focuses on finding a repositioning
plan that can be transformed into a feasible solution for each
uncertain realization in some user-specified set. The paper
provides (1) the introduction of a transformable robust problem
(TRP) framework, similar to two-stage stochastic optimization;
(2) the application of the framework to repositioning problems,
and development of a tight set of sufficient conditions characterizing
feasible TRP solutions; and (3) the development of practical recovery
strategies based on local resource sharing that lead to computationally-tractable
integer programs for solving large-scale TRP repositioning problems.
|
|
A Dynamic Driver Management Scheme for Less-than-Truckload Carriers
to appear, Computers and Operations Research, 2007.
(w/ B. Karacik and M.W.P. Savelsbergh)
This paper describes a scheme for the dynamic management of linehaul
drivers developed for the one of the largest U.S. LTL carriers. Most
scheduling problems faced by transportation service providers
are complicated by time-constrained vehicle operators that can
only be renewed by resting. Further, LTL driver scheduling is
difficult since trucking moves are usually unscheduled. The driver
management scheme proposed in this paper combines greedy search with
enumeration of time-feasible driver duties, and is capable of
generating cost-effective, high-quality feasible driver schedules for typical
three-day planning horizons with 15,000-20,000 loads in a matter of minutes.
|
|
Locating Drivers in a Trucking Terminal Network
submitted, 2006.
(w/ B. Karacik, M. Hewitt, and M.W.P. Savelsbergh)
This paper applies a dynamic driver scheduling scheme for LTL trucking
systems to the problem of determining driver domiciles and union
bids. We develop an incremental scheme for locating drivers and
determining bids, and apply the approach to scenarios derived from a
major LTL carrier. Cost implications of changes to driver policies are analyzed.
|
|
Global Intermodal Tank Container Management for the Chemical Industry
Transportation Research, Part E, 2005.
(w/ J.C. Morales and M.W.P. Savelsbergh)
International logistics is especially crucial to the high-value
chemicals industry, since raw materials sources, production
facilities, and consumer markets are distributed globally.
Fluctuating demand, imbalanced trade flows, and expensive
transportation equipment necessitate dynamic asset management.
This paper focuses on problems faced by tank container operators,
formulating and solving an operational tank container management problem,
integrating container routing and repositioning decisions. Results
indicate that cost savings of 10-15% can be realized by using such
an integrated approach.
|
|
Dynamic Operational Planning for Regional LTL Carriers
working paper, 2006.
(w/ P. Warier and M.W.P. Savelsbergh)
This paper develops a set of daily operational planning models for
regional LTL carriers. Regional carriers provide high levels of
service over smaller regions than national carriers. Daily
planning models are developed that determine best daily flow
plans through the consolidation terminal network, and then
determine a driver schedule serving the load plan.
|
|
|
|
