Performance Model for a Public Logistics Network

Show full item record

Title: Performance Model for a Public Logistics Network
Author: Xiang, Ling
Advisors: Michael Kay, Committee Chair
Tao Pang, Committee Member
James Wilson, Committee Member
Russell King, Committee Member
Abstract: A public logistics network (PLN) has been proposed as an alternative to private networks for the ground transport of parcels. In this dissertation, a heuristic approach to approximate the package average waiting time in a PLN is presented; and then based on this waiting time approximation, a PLN design procedure is developed. A PLN can be viewed as a priority queuing network with bulk arrivals and bulk service. It is difficult to obtain a closed-form solution for package average waiting time in a PLN. The problem is reformulated so that trucks transport loads instead of individual packages, thereby relaxing the bulk arrivals and bulk service feature. The package average waiting time is approximated fairly accurately by Kingman’s equation when the server utilization is high. A simulation model is created to determine the parameters needed in Kingman’s equation (the coefficient of variation for package interarrival times and for truck interarrival times). A regression analysis of the results shows that the headway ratio is around 5.5. The package average waiting time is approximated by the product of the truck headway (truck average interarrival time) and the headway ratio. The PLN simulation with protocols and the package bidding process is discussed as an extension to the basic simulation model. Packages bid for their trips along the way. The highest bidder gets the highest priority for truck transport services. Results from a simulation model incorporating a series of protocols developed by Kay show that a PLN with these protocols performs better than a FIFO system. For the PLN design problem, the goal is to design a PLN that results in the minimal package average waiting time for the entire network. Potential locations (search space) for the distribution centers (DCs) are found using the U.S. network of interstate and highways. A genetic algorithm (GA) was applied to search for the optimal number of DCs, their locations, and direct arc connections between each pair of DCs.
Date: 2009-01-07
Degree: PhD
Discipline: Industrial Engineering
URI: http://www.lib.ncsu.edu/resolver/1840.16/5752


Files in this item

Files Size Format View
etd.pdf 2.483Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record