Multi-level spatial models are proposed to approximate large size
logistics network systems. The linkage across different levels of spatial models
is specified through estimating equations to utilize information considered at different
stages. The multi-level models are applied in decision analysis of logistics systems
such as facility location allocation and demand forecasting problems. System
reliability of logistics service is introduced to characterize the uncertainty of
supply chain disruptions. In particular, a sum-of-disjoint-product (SDP) method
is presented to evaluate the system reliability. Real-life examples show the potential
use of the models in characterizing logistics networks, assisting logistics planning
processes and evaluating network designs for dealing with unexpected supply chain
disruption. (Joint work with J.-C. Lu and P. H. Kvam).