University Bus Systems: Network Flow Demand Analysis

University Bus Systems: Network Flow Demand Analysis

By Craig YannesUniversity of ConnecticutOctober 21, 2008

Introduction

University bus systems are an important form of transportation around university campuses

Research has shown that these systems generate more ridership than their counterparts (Daggett and Gutkowski, 2005)

University enrollment has also been expanding

Problem

The sharp increase in demand has put a strain on university resources to provide an effective bus system

Many routing design decisions have been made subjectively

This results in an inefficient system, which leaves demand centers unidentified, underserved or unserved

Solution

Network flow theory, in conjunction with spatial analysis (GIS), can help remove the subjectivity from university bus system design

Focus of this project is the selection of optimal stopping locations considering operational cost while serving the maximum passengers (demand)

Goals

Create a simple and efficient model framework to analyze the coverage of a university bus system

Analyze the effects of stopping service areas (walk distance to stop) on the selection of optimal stopping locations

Objectives

Collect and generate transportation link and transit demand data

Generate and analyze demand data to create centers which will serve as potential bus stops

Create a network flow model representation and use an appropriate solution technique, yielding the optimal demand centers to be served by the bus system at differing service areas

Background

The following have utilized network flow algorithms and theory to solve transit design (routing, scheduling, frequency, etc.) problems: Ceder and Wilson (1986) Chu and Hobeika (1979) Ranjithan, Singh and Van Oudheusden (1987) LeBlanc (1988) Kocur and Hendrickson (1982) Kuah and Perl (1985)

Background

Overall transit design involves two interest groups (passenger and operator)

Depending on the component being designed, the focus shifts between these groups

This proposed research will focus solely on operator costs while attempting to serve the maximum amount of passengers

Background

Furth, Mekuria and SanClemente (2007) created and used GIS applications to analyze the spacing of transit stops based on the street network and parcel data

The study evaluated walking, riding and operating costs when stops were reconfigured from the existing placement

Similar to this research except that no network flow theory was used

Network Representation

The following design was used to create a network representation of the bus system:

Arcs leading to accessible nodes based on current location

Cost on Links is a function of distance and demand

Nodes representing potential stopping locations

Supply Node(Bus Depot)Supply = Number of demand nodes

Destination nodes which are spread throughout the exterior of the network to pull flow in all directions

Demand = 1 for each node

Data

Connecticut Department of Environmental Protection: Connecticut Street Network Shapefile (1:100,000)

2006 UConn personal geodatabase which included street, building and parking lot feature classes

Transportation Network

0 0.50.25Miles

Demand

Calculated using ITE Trip Generation Manual

Produces auto trips based on the building/area purpose and attributes such as area, number of seats and number of units

Although this research focuses on transit trips, the auto trips can be used to determined relative demand between locations

Building and Parking Lot Demand

0 0.50.25Miles

Potential Stopping Locations

32 Locations were selected at points along the roadway near key intersections and large generators

The demand at each one of these stops is equal to the sum of demand of the buildings and parking lots within a particular distance (1/8, 1/4, 1/2 mile) of the stopping location

5 destination locations were also selected such that the flow would be spread around the network evenly

Potential Stopping Locations

0 0.50.25Miles

Link Generation between Stops

Links represent the transportation path between two stops though it does not follow street layout directly

The following rules were used to create the links between the stops: Links must proceed in a forward progression

(must be getting closer to a destination) Links cannot pass through stops Stops on the exterior must connect with sink

locations

Link Generation

Link Cost

Combination of two factors Distance between stops Demand at the ending stop

Because higher demand should incur less cost the inverse demand was used

This requires a scaling factor so that demand values are comparable in magnitude to distance

1000*Demand

1 Distance CostLink

jijij

Network Representation

Solution Technique

A geometric network was created in GIS weighted with the calculated link cost

The Network analyst toolset in ArcGIS was used to find the shortest path between the source node and each of the 5 sink nodes

The nodes that lie on these shortest paths are the optimal stopping locations

1/8 Mile Service Area

1/4 Mile Service Area

1/2 Mile Service Area

Results

Path

1/8 mile 1/4 mile 1/2 mile

StopsDistance(miles) Stops

Distance(miles) Stops

Distance(miles)

1 3-5 2.54 3-5 2.54 3-5 2.54

2 13-12-11-8 1.14 13-12-11-22 0.99 13-29-10 0.79

3 13-12-11-22 0.99 13-12-11-22 0.99 13-12-11-22 0.99

4 30-23-32-20 1.32 30-23-32-20 1.32 30-23-32-20 1.32

5 30-23-28 0.99 30-23-28 0.99 30-23-28 0.99

Total17 stops

(12 unique)6.98

17 stops(11 unique)

6.8316 stops

(13 unique)6.63

Conclusions

A model framework has been created which analyzes a bus network yielding optimal stopping locations

Increasing the service area, reduces the distance traveled and increases that amount of unique stops served by the system

Planners must be cautious when trying to balance ridership and efficiency

Analyzing the system for different service areas can help quantify this relationship and create a more efficient system

Future Research

Expand the analysis area to include other locations surrounding the campus

Acquire or generate more accurate demand data

Comparison / application to the existing bus system

Incorporate the effect of larger service areas on demand

Create similar network frameworks that focus on the other aspects of bus system design

Comments / Questions?

References Daggett J. and Gutkowski R. (2003). University Transportation Survey: Transportation

in University Communities. Colorado State University. Sutton J. C. GIS Applications in Transit Planning and Operations: A Review of Current

Practice, Effective Applications and Challeneges in the USA. Transportation Planning and Technology, Vol. 28, 2005, pp. 237-250.

Furth P. G., Mekuria M. and SanClemente J. Stop-spacing Analysis Using GIS Tools with Parcel and Street Network Data. Presented at 86th Annual Meeting of the Transportation Research Board, Washington, D.C., 2007.

Ceder A. and Wilson N. H. M.. Bus Network Design. Transportation Research Part B, Vol. 20B, 1986, pp. 331-344.

Chu C. and A. G. Hobeika. In Transportation Research Record: Journal of the Transportation Research Board, No.730, Transportation Research Board of the National Academies, Washington, D.C., 1979, pp. 7-13.

Institute of Transportation Engineers (ITE). Trip Generation, 6th ed., Washington, D.C., 2003. Kocur G. and Hendrickson C. Design of Local Bus Service with Demand Equilibrium.

Transportation Science, Vol. 16, 1982, pp. 149-170. Kuah G. K. and Perl J. A methodology for feeder bus network design. In Transportation

Research Record: Journal of the Transportation Research Board, No.1120, Transportation Research Board of the National Academies, Washington, D.C., 1985, pp. 40–51.

LeBlanc L. J. (1988). Transit system network design. Transportation Research Part B, Vol. 22B, 1988, pp. 383-390.

Ranjithan S., Singh K. N., and Van Oudheusden D. L. The Design of Bus Route Systems – An Interactive Location-Allocation Approach. Transportation, Vol. 14, 1987, pp. 253-270.

References

GIS Data Connecticut Department of Environmental Protection

http://www.ct.gov/dep/cwp/view.asp?a=2698&q=322898

Richard Mrozinski, University of Connecticut Department of Geography

Images Title: http://www.park.uconn.edu/ Problem:

http://www.news.com.au/dailytelegraph/story/0,22049,2 1265300-5011906,00.html

Background: www.caliper.com/UK/transcad.htm Questions:

http://www.pritchettcartoons.com/newcar2.htm http://home.fuse.net/ard/jandress/transfuture3.jpg