A hybrid travel distance approximation for a GIS-based decision support system

Journal of Business Logistics, 2001 by Campbell, James F, Labelle, Alain, Langevin, Andre

This paper presents a new method to model travel distances for use in a geographic information system (GIS)-based decision support system. The widespread availability of digital roadway databases and GIS software has led to new approaches and solutions for many logistics and transportation problems. This has also sharpened the focus on how best to use such data and tools for effective decision-making. The distance approximation presented in this paper was developed to model travel distance for trucks hauling snow in an urban snow disposal decision support system. Because these trucks can have significant impacts on traffic and neighborhoods, around the clock, they may be encouraged or required to concentrate their travel on certain roadways (for example, to minimize their impact on residential neighborhoods, or reduce traffic congestion). Thus, the shortest distance or travel time path between two points may not represent the actual truck travel distance. The hybrid approximation presented in this paper combines the shortest path distance on a subset of an urban street network with an analytical distance approximation. This approach blends the accuracy and realism of shortest path travel with the simplicity and reduced computational burden of a distance approximation. The distance approximation was developed and tested using data for the City of Montreal, Canada. This approach may also be useful for logistics and transportation modeling in other contexts, especially when the majority of travel occurs on a small subset of the roadways.

This paper presents a distance approximation developed as part of a project to create a personal computer-based decision support system (DSS) for improving urban snow disposal. This DSS is an interactive tool for strategic and tactical decision making. Strategic design of a snow disposal system includes partitioning an urban region into small geographic sectors, assigning these sectors to snow disposal sites, and allocating snow hauling trucks to sectors. Because of the complexities of the operating environment, the DSS is designed to be used in an iterative and interactive design process that incorporates the expertise of the user. The DSS is also used for real-time tactical decision making such as responding to equipment breakdowns and other contingencies. To assist in all of these decisions, the DSS utilizes travel distance models since major costs in snow disposal are incurred for transporting snow to disposal sites. The goal of the distance modeling portion of the DSS development was to develop a method to approximate distances traveled by large trucks hauling snow that would be easy to use, accurate, and fast enough for use in an interactive personal computer-based decision support environment. Our approach is to model truck travel distance in two parts: first, using a simple and accurate analytical approximation for the short travel on local streets to a major roadway, then adding the shortest path distance on a reduced street network of major roadways.

This paper reports on the development of the hybrid distance approximation and presents results using data for Montreal, Canada. The paper first briefly describes snow removal and disposal operations. (For further details see Campbell and Langevin 1995a; Labelle, Langevin, and Campbell, forthcoming.) Following sections then describe relevant previous work and our approach, with an application in Montreal. The paper also addresses implementation issues and calibration of the model.

URBAN SNOW DISPOSAL OPERATIONS

Snow and ice control on roadways are essential and expensive winter maintenance operations necessary to maintain safe travel conditions. In urban regions that experience heavy snowfall, the snow plowed off the roadways will accumulate and impede pedestrian and vehicular traffic. In these cases, snow disposal operations are required to remove the snow to disposal sites (for example, water bodies, sewers, or surface lots) where it will not interfere with travel. This is usually accomplished by loading the snow into trucks which then haul the snow to assigned disposal sites. Snow hauling can occur around the clock following a storm. In Montreal, this involves an average of 300,000 truckloads per year hauling 7 million cubic meters of snow. Note that snow plowing to clear streets, which is an important and expensive component of snow and ice control, is not the focus of this work. We are interested in the snow removal and disposal operations following the initial plowing to clear streets.

Snow fighting is an expensive public service. The average annual winter maintenance budget for Montreal is approximately $60 million. (Sapporo, Japan, which receives an average of 5 meters of snow per year, has a winter maintenance budget of over $80 million!) Even cities that do not typically experience heavy persistent snowfalls, can experience very large expenses from serious storms. For example, New York City spent $57 million in 1995-1996 (the winter of the "Blizzard of '96" in January), four times their budget for winter maintenance.