A public logistics network is proposed as an alternative to private logistics networks for the ground transport of parcels. Using the analogy between the packages transported in the network and the packets transmitted through the Internet, a package in a public logistics network could, for example, be sent from a retail store and then routed through a sequence of public distribution centers (DCs) located throughout the metropolitan area and then delivered to a customer’s home in a matter of hours, making a car trip to the store to get the package unnecessary. The DCs in the network, functioning like the routers in the Internet, could also be located at major highway interchanges for longer distance transport.
| Public Logistics Network | Internet |
| Packages transported | Packets transmitted |
| Distribution centers (DCs) | Routers |
| Trucks | Wire, fiber |
Currently, it is common for a single logistics firm like UPS and FedEx to handle a package throughout its transport. The such a private logistics network, much of the technology used to coordinate the operation of the network is proprietary. As a result, the principal competitive advantage that a private logistics company has is the barrier to entry due to the very large scale of operation (national or international) required in order to be able to underwrite the development of private facilities and propriety technologies. Nevertheless, a single firm, unless it becomes a monopoly, is ultimately limited in the scale of its operation, resulting in the use of a limited number of large-scale hub transshipment points that can result in packages making many circuitous hops before reaching their destinations. In a public logistics network, the different functions of the network would be separated so that a single firm is not required for coordination. This would enable scale economies to be realized in performing each logistics function since each element of the network has access to potentially all of the network’s demand. The increase in scale would make it economical to have many more transshipment points. Each transshipment point, or distribution center (DC), could be an independently operated facility that serves as both a freight terminal and a public warehouse, and could be established in small cities and towns that would never have such facilities if they were served as part of a proprietary, private logistics network.
| Public Logistics Network | Private Logistics Network |
| Each truck and DC can be operated by different firm | Single firm (e.g., UPS) handles package throughout its transport |
| Similar to mesh computer network topology, where each node is connected to many other nodes | Similar to star computer network topology, where each node is connected to a central hub |
| Decentralized control via standard coordination protocols | Centralized control via firm-specific coordination procedures |
| Availability of public DCs can increase competition to match truckload (TL) transport | Private hubs/terminals result in high barrier to entry for parcel and LTL transport |
(Left) Hypothetical public logistics network showing 36 public DCs covering the southeastern portion of the USA and connected via interstate highways. Each of the interstate DCs would serve as a hub in a sub-network of local DCs (not shown) covering the region surrounding the interstate DC. A package being transported, for example, from Jacksonville, FL (DC 4) to Richmond, VA (DC 30) would travel on different trucks (operated by different firms) between each of the DCs along the way (DCs 7, 24, 23 and 17) and could be temporarily stored at each of the DCs (e.g., to wait for lower-cost off-peak travel periods). (Right) Each interstate DC would be located next to a highway interchange in order to enable direct access to adjacent interstate DCs (with this arrangement, the DC has direct access to all eight interstate travel directions).
The two major areas of research being explored are the coordination and design of public logistics networks.
Pricing mechanism: A pricing mechanism has been developed in which each package, prior to its transport, submits a bid that reflects its desired speed of delivery. This bid is then used to pay each truck that transports the package and each DC that stores the package along the path from its origin to its destination. Thus, at each DC, all of the packages going to the same DC are competing with each other to be in the next load to be transported to that DC, and that next load is competing with the loads going to other DCs to be selected by a truck as the next load for transport. In addition, the trucks are themselves competing with each other to select each load for transport. Because of this competition, truck capacity can automatically adjust to varying transport demand along each link in the network without any type of centralized control. The mechanism consists of truck and package protocols.
Object-oriented distributed simulation: An initial simulation of a public logistics network has been developed in Matlab. Simulation is necessary since the bulk arrivals and bulk services characteristic of a public logistics network make it difficult to model analytically. An object-oriented simulation is useful because it makes it possible to encapsulate the information associated with each element of the network so that it can easily be changed to address different research issues.
Agent-based coordination. Using truck-location information together the current load bids provided at each DC, package agents determine what bids to submit for their packages and truck agents determine what loads to accept for their trucks. Both the package and truck agents can interact with other outside of the protocols and the agents can be hosted at a DC, off-site with communications via the Internet, or on each package or truck.
National public logistics network: A genetic-algorithm-based procedure has been created to determine the number and location of DCs in the network that minimizes the average transport time of a package. The procedure has been used to determine at which major highway interchanges throughout the continental United States to locate the DCs used for long-distance transport. This provides an expanded version of the 36-DC network shown in the figure above. The criterion used to guide the initial design is just the average transport time: what number and location of interstate DCs results in the minimum average transport time. The cost of establishing each DC was not used because, even if a DC were located at every interstate interchange, transshipment delays at each DC would make it preferable for packages not to stop at every DC.
DC design: In order to be cost-effective, the loading/unloading, sortation, and storage activities at each DC in the public logistics network must be highly automated since each load might visit a dozen or more DCs while it’s in transit, likely traveling on a different truck between each DC. Since existing automation technologies do not provide the flexibility needed to allow any size load to move to any location at anytime, a new DC design has been developed that would allow packages of varying size to be automatically unloaded at a DC, sorted, stored, and then loaded onto an outbound truck. Such a design would make it as cheap to ship a single package as it is to ship a larger consolidated load.
Deduced traffic congestion: Local traffic congestion will be reduced because many less-than-truckload deliveries to stores could be replaced by fewer full-truckload deliveries, and individual car trips to stores could be replaced with same-day parcel deliveries from the stores to homes (where each parcel might visit several local DCs as it is transported to the home).
Widespread parcel delivery: It will be possible for small cities and towns to have access to the same low-cost parcel delivery services (e.g., same-day and next-morning) that are currently only available in larger cities.
In-transit trade: A unique capability of a public logistics network is that a third party can search the network for any type of package in transit. Once located, negotiations can take place and the package might be resold to the third party and redirected to a new destination.
Dynamic inventory: Some finished-goods inventory and safety stocks would become mobile, since they can be located and re-routed to where they are needed only when they are needed. A unique capability of the public logistics network pricing mechanism is that packages can be transported at almost no cost by using what would otherwise by unused space aboard trucks, paying only for their storage at each DC. In this manner, inventory can be positioned in regions where it is likely to be needed in advance of an actual order, thereby reducing the resulting transportation costs when the order occurs.
Kay, Michael G., and Parlikad, Ajithkumar N., Material Flow Analysis of Public Logistics Networks, in Progress in Material Handling Research: 2002, R. Meller et al., Eds., Charlotte, NC: The Material Handling Institute, 2002, pp. 205–218 (presented (Slideshow) at the 7th International Material Handling Research Colloquium, June 1–5, 2002, Portland, ME). Compares the average transport time of a hypothetical public logistics network covering the southeastern U.S. to the times of a hub-and-spoke and a point-to-point network covering the same region. The public logistics network provided the minimum average transport time when the time required for loading/unloading at each transshipment point in the network was short. This result is robust with respect to a range of different transport demands and truck capacities considered in the analysis.
Parlikad, Ajithkumar N., Performance Analysis of Intelligent Supply Chain Networks, Master’s Thesis, Dept. of Industrial Eng., North Carolina State Univ., Raleigh, NC, 2002.
Gandlur, Karthik S., Implementation of Adaptive Routing in Public Logistics Networks, Master’s Thesis, Dept. of Industrial Eng., North Carolina State Univ., Raleigh, NC, 2002.
Kay, Michael G., and Jain, Ashish, Issues in Agent-based Coordination of Public Logistics Networks, Tech. Rep. 02-01, Dept. of Industrial Engineering, North Carolina State Univ., Raleigh, NC, October 2002, describes some of the issues involved in research that is just starting to design an agent architecture and protocols that can be used to coordinate the operation of this type of network in order to facilitate adaptive routing and in-transit trade.
Bansal, Amogh, Designing a Public Logistics Network, Master’s Thesis, Dept. of Industrial Eng., North Carolina State Univ., Raleigh, NC, 2004. Initial work on the public logistics network design problem. A GA is used to design a PLN that covers the entire continental U.S.
Kay, Michael G., and Jain, Ashish, Pricing a Public Logistics Network, extended abstract of presentation at the Industrial Engineering Research Conference, May 15-19, 2004, Houston, TX. Initial ideas on protocol requirements.
Kay, Michael G., Protocol Design for a Public Logistics
Network, in Progress
in Material Handling Research: 2004, R. Meller et al., Eds., Charlotte,
NC: The Material Handling Institute, 2004, pp. 181–188
(presented at the International Material Handling Research Colloquium,
June 13–17, 2004, Graz, Austria). Initial specification of truck and
package protocols.
Jain, Ashish, Protocol Design for a Public Logistics Network, Master’s Thesis, Dept. of Industrial Eng., North Carolina State Univ., Raleigh, NC, 2004. Simulation of a public logistics network in order to estimate the waiting time at each DC, and implementation of a portion of the truck and package protocols specified in [7]. Using a simulation of a public logistics network, weighted average waiting times with and without the use of the protocol were determined and it was found that there was a statistically significant decrease in the waiting time associated with the use of the protocol.
Kay, Michael G., and Jain, Ashish, Implementing a Pricing Mechanism for Public Logistics Networks, in Proceedings of the Industrial Engineering Research Conference, Atlanta, GA, May 14–18, 2005. This paper describes an implementation of the pricing mechanism specified in [7] and summarizes some of the results of [8].
The data used for the hypothetical 36-DC public logistics network, above, is available as a Matlab file plnex36.mat (8 KB). The file contains the 3-column arc list matrix IJD of arc beginning and ending nodes and distance (see Matlog toolbox for functions that work with arc lists), and a 36-element vector w containing the population of the region covered by each DC (see Table 1 in Publication 1, above, for more details).
The simulation software used in the research will be made available when it is completed.
This research is supported, in part, by the National Science Foundation under Grant CMS-0229720 (NSF/USDOT: Agent-based Coordination in Public Logistics Networks, PI: Michael G. Kay)
Michael G. Kay, Department of Industrial Engineering, North Carolina State University, Raleigh, NC.
Last changed: Thursday, April 06, 2006 04:54 PM