报告摘要：
Current business trends towards longer and leaner supply chains
increase risks in the whole chain.  The supplier selection decision
has become an important strategic level decision.  We develop a
two-stage stochastic programming (SP) model and a chance constrained
programming (CCP) model to incorporate uncertainty of demand and
supplier capacity.  We use a probabilistic distribution of supplier
capacity to represent issues of timely delivered supplies due to long
transportation routes and production disruption at the suppliers.  The
demand distribution represents uncertainty in future demand.  Both
models include several objectives and determine a minimal set of
suppliers and optimal order quantities with consideration of business
volume discounts.

The developed models aim to balance the benefit from a small number of
suppliers with the risk of not being able to meet demand.  A
deterministic mixed integer programming (MIP) model is also formulated
for comparison purposes.  Both the SP model and the CCP model improve
on the MIP model in terms of providing a robust selection of suppliers
and give the decision maker a more complete picture of tradeoffs
between cost, system reliability and other factors.  We present
Pareto-optimal solutions for a sample problem to demonstrate the
benefits of the probabilistic models. In order to describe the
tradeoffs between costs and risks under alternative Pareto-optimal
supplier selection solutions in an analytical form, we develop
multiparametric programming techniques to more completely analyze the
multiple selection criteria and probabilistic constraints on demand
and supplier capacity in the CCP model. We found from the experimental
results that the CCP model with the multiparameteric analysis gives
the decision maker a complete picture of tradeoffs between alternative
solutions. However, the CCP model assumes a probability distribution
for the random demand and supplier capacity whereas the SP model is
scenario-based, which also has advantages.

报告人简介：
Dr. Zelda B. Zabinsky is a Professor in Industrial Engineering at the
University of Washington.  She started at the University of Washington
in 1985, after she completed her Ph.D. in Industrial and Operations
Engineering from the University of Michigan.  Professor Zabinsky was
awarded a Benton Fellowship for her graduate studies in Operations
Research at the University of Michigan.  In 1992, while at the
University of Washington she received a Research Initiation Award from
the National Science Foundation (NSF) to develop global optimization
algorithms for engineering design. She currently has a second NSF
grant to continue developing optimization in engineering design.
Software developed, in part, under this effort is currently in use at
the Boeing Co. to design fuselage panels with advanced composite
materials.

Professor Zabinsky has published numerous papers in the areas of
global optimization, algorithm complexity, and optimal design of
composite structures.  She recently received an Erskine Fellowship to
continue work in global optimization at the University of Canterbury,
NZ, and she is writing a book on Stochastic Methods in Global
Optimization. Her work has appeared in Mathematical Programming,
Journal of Global Optimization and the international journal Composite
Structures.  Her research has been funded by Boeing Commercial
Airplane Company, NASA-Langley, FAA, and NSF.  Dr. Zabinsky is also
involved in applications of Operations Research to air traffic
management and airspace design, as well as other applications in
transportation, health care, forestry, structural optimization and
manufacturing.