The GIS Approach towards Development of a Facility Location Model

The present age is an era of dynamism, growth and efficiency, where the demands of people are more oriented towards service commitment. To meet the challenges of 21st Century, FedEx is always striving towards new technology by expanding its existing facilities and adding new improved facilities worldwide. Currently, there is no automated system or methodology for locating facilities such as stations (sorting, courier loading/unloading facility), dispatch centers, etc. This paper focuses on an ongoing project, " Based Facility Location Model", which determines the optimum location for a given number of stations and also allocates service areas to a station. This project will encompass a creation of a GIS application using ArcView/Avenue, system interfaces, database conversion, and a communication interface to interact with selected existing systems. The proposed integrated system will allow users to define/redefine AOI (Area of Interest), provide user interface to edit modeling parameters, and formulate and analyze "what if" scenarios. this system will also provide the ability to query database, produce maps and reports, and include spatial analysis procedures to evaluate site suitability.

INTRODUCTION

The proposed Facility Location Model (FLM) is very specific to FedEx operations. Some literature is available for different types of FLM; however, none of them meet the requirements of FedEx FLM. The FedEx service commitment is the basis for this model. Depending on the building leave time and the service commitment, the model will be able to locate a station from where the ZIP codes assigned to that station can be serviced efficiently and economically. The model will also have the ability to allocate ZIP codes to the station(s) as well as check the demands of the assigned ZIP codes against the capacity of the station.

MODEL OBJECTIVES

The objectives of this model are:

1. To determine the optimum locations for a given number of stations so that ZIP codes within a defined AOI can be serviced efficiently and economically.
2. To allocate ZIP codes within an AOI to a given number of station(s).
3. To determine number of couriers required to service a ZIP code.

METHODOLOGY

The methodology involves a mathematical programming approach. Thus expressing the formulation in words:

Minimize total number of couriers

Subject to:

1. Total courier hours spent for servicing a ZIP code from a station is less than or equal to time available for a service commitment (e.g. 10:30 - Building Leave Time).
2. Total inbound/outbound volume serviced to a ZIP code from a station is equal to the demand of the ZIP code.
3. Total inbound/outbound volume serviced from a station must be less than or equal to that station capacity.

The solution will be accomplished by writing a program in C/Avenue and can be accessed through ArcView 3.0.

ASSUMPTIONS AND CONSTRAINTS

The following assumptions were made in the proposed FLM:

1. The number of packages (volume) that a courier can carry in the van is not a constraint.

2. The computation time for a courier to reach the first stop in the ZIP code from the station is calculated by dividing the two point measure distance (from the station to the ZIP code) by the average speed of the courier, and assumed to be close to the real world situation.

The model computation time will depend on size of the problem (i.e. number of ZIP codes and stations to be modeled) and should not be a problem in modeling effort.

DATA INPUT REQUIREMENTS

The following geodatasets will be used for background reference and defining AOI.

1. ZIP codes
2. Stations
3. Regions and Districts
4. Roads
5. Landmarks
6. Water Features
7. Railroads

Other data input elements are inbound/outbound volumes and the productivity rate by ZIP code. An user interface will be provided for the FLM users to input following model parameters:
1. Number of stations and their capacity (in terms of volumes).
2. An option to keep existing station(s) fixed and make the rest (if any) floating
3. Building leave time
4. Average speed of the courier

SUMMARY

The project is scheduled to be completed in May 1997. A few test cases with different scenarios will be presented at the time of paper presentation. The following functions will be provided to the users:

1. Allow users to specify the AOI by one or combinations of the followings:
Station name(s)
Zip codes
Graphic pick of ZIP codes/station(s)
District name
2. Provide the ability to exclude/include certain ZIP codes to redefine AOI.
3. Provide ability to move point ZIP codes and/or stations to redefine AOI.
4. Import ZIP code forecast volumes into GIS environment and provide graphic support for better analysis.
5. Provide user interface to edit modeling parameters.
6. Develop an optimization routine in GIS environment using Avenue for ArcView 3.0 for determining optimum locations of the given
number of stations.
7. Provide ability to query database and produce reports and maps. These reports and maps may contain modeling input/output.
8. Display roads, landmarks, water features, railroads, etc. in the background for references.
9. Provide tools to formulate and analyze "what if" scenarios.

POTENTIAL FUTURE EXTENSIONS

The following potential future enhancements are recommended:

1. Incorporate additional modeling parameters such as square foot leasing cost, etc., in the optimization routine to enhance modeling capabilities.
2. Use historical stops to derive well representative point ZIP code locations for defining AOI.
3. Include procedures to calculate distance between the station and a ZIP code from the street network and consider this calculated distance as one of the optimization parameters.
4. Include procedures to register/rectify aerial photographs, satellite images, etc., and display these images in the background.
5. Include spatial analysis procedures to evaluate site suitability.

REFERENCES

1. Surya N Sahoo, Functional Requirement Document: GIS Based Facility Location Model, FedEx, Operations Research & Spatial Applications, December 1996.
2. Surya N Sahoo, Methodology & Data Input Requirements: GIS Based Facility Location Model, FedEx, Operations Research & Spatial Applications, January 1997.
3. Frederick S. Hillier, Gerald J. Lieberman: An Introduction to Mathematical Programming, McGraw-Hill publication, 1990.