Papers of the Applied Geography Conferences (2008) 31: 234-243

CHANGING ACCESSIBILITY AND COMMUTING COSTS IN METRO LOUISVILLE: ANALYSIS OF THE HYPOTHETICAL DISRUPTION OF SHERMAN MINTON BRIDGE Wei Song Andrew McKinney Department of Geography and Geosciences University of Louisville Louisville, KY 40292 1. INTRODUCTION Transportation networks, composed of links connecting geographically dispersed communities, towns and cities, are an indispensable component of everyday life in modern society. Much of modern society has become dependent on mobility and the accessibility of important activities on the daily basis. When transportation systems operate as designed, they form the foundation upon which commerce, trade and the serviced communities’ well-being can flourish. But when the availability of these systems is jeopardized by gradual deterioration (e.g., corrosion induced deterioration) or natural hazards (e.g., earthquake induced link failure), the communities they service can likewise suffer (Birdsall et al., 2007). Thus, transportation networks, along with other physical and virtual networks (e.g., power lines, the Internet) are referred to as crucial lifelines in modern society (Miller, 2003). The August 2007 collapse of the I-35W Bridge in Minnesota has significantly impacted road-users and the Minnesota economy. The disruption in the transport system forced about 140,000 daily vehicles relying on this link to reroute. A model created by REMI Consulting and the Minnesota Department of Transportation estimated that costs to road-users due to the detours would total $400,000 per day. The daily loss of $247,000 of auto travel time was projected for commuters due to prolonged journeys to and from the work. This economic loss also has the potential to cost the state jobs throughout the economy (DEED, 2007). Many approaches have been used to measure transportation network vulnerability, but most can be grouped into two major categories (Miller, 2003). One is focused on performance-based indicators (e.g., network reliability and network performance), while the other explores user-based indicators, primarily accessibility. Network vulnerability can be viewed as its susceptibility to disruptions and the consequence of potential link degradation or complete failure (Miller, 2003; Nicholson et al., 2003). For instance, vulnerability can be measured by the increase in cost of one or more of the links between nodes. The higher the generalized cost due to the failure or degradation of the link, the higher the vulnerability of the network (Taylor and D’Este, 2007). A transportation network that concentrates flow through a small number of links (e.g., bridges and tunnels) is more vulnerable than a more fully connected network if one or more of the critical links fail. Accessibility measures the ability of individuals to participate in activities in other physical locations within a specified region using an available transportation network. The concept of accessibility recognizes that transportation is a derived demand and exists not for its own sake, but to help people accomplish other activities in space and time. Loss of accessibility greatly impacts abilities to earn a living, access critical services, or maintain social relations (Miller, 2003). Methods for analyzing network accessibility have been applied to examine infrastructure vulnerability (Chang, 2003; Chen et al., 2007; Jenelius and Mattsson, 2006; Taylor et al., 2003).


The objective of this paper is to evaluate the impact of a hypothetical network disruption on the accessibility and commuting expenditure in Metro Louisville, Kentucky. Changing accessibility for census tracts was evaluated under normal network and disrupted network conditions, with failure of the Sherman Minton Bridge. Potential economic or monetary consequences for commuters were also assessed at the level of census tracts and commuting routes. 2. STUDY AREA, DATA AND METHODS Metro Louisville, comprising Kentucky and Indiana counties on opposite sides of the Ohio River, is integrated by three main bridges (Figure 1). The CBD of Louisville is close to the intersection of Interstate I-65 and the Ohio River. The seven-lane John F. Kennedy (JFK) Bridge, the most heavily traveled transportation artery, allows access from Clark County, Indiana, via I-65 to downtown Louisville. Also allowing access from the north via Clark County is the least traveled route, four-lane Clark Memorial Bridge. The six-lane Sherman Minton Bridge carries I-64 and US 150 over the river, connecting the west side of Louisville to downtown New Albany in Floyd County, Indiana. The bridge is used for through-traffic, as well as commuters who depend on the bridge for job accessibility. On average, 93,210 cars utilize the bridge on a daily basis (Green, 2007). In a recent re-evaluation by the State of Kentucky in the aftermath of the collapse of the I-35W Bridge in Minnesota, the Sherman Minton Bridge was rated serious in deck condition and fair in the condition of superstructure and substructure. The JFK Bridge was assigned similar ratings (Green, 2007). Since the JFK and the Clark Memorial Bridges are in close proximity, about half of a mile apart, if one were to be disrupted the other could help accommodate some of the load, although with unavoidable delays and congestions. The Sherman Minton Bridge, however, is approximately 5 miles upriver. Its failure would have considerable impacts on the commuting patterns in Metro Louisville in forms of the road-user transportation detours, especially for Floyd County commuters.

FIGURE 1 A CLOSE VIEW OF THE OHIO RIVER BRIDGES IN METRO LOUISVILLE This research is focused on commuting impacts of the disruption of the Sherman Minton Bridge. Jefferson County, Kentucky, where the City of Louisville is located, and two Indiana counties immediately across the Ohio River−Floyd and Clark−were included in the study area. Jefferson County (population 693,604 in 2000) includes major employment centers


of the metropolitan area, such as Humana and UPS within the city, GE Appliance Park, Bluegrass Industrial & Research Park, and Ford Motor Company Truck Plant. In 2000, about 84 percent of the metropolitan area’s population, 86 percent of the total households, and 84 percent of the total employed persons, were concentrated in the three counties. According to the 2006 American Community Survey, nearly 40 percent of the workers in Clark County and 34 percent of the workers in Floyd County commuted across the Ohio River to Jefferson County (U.S. Bureau of the Census, 2006). Commuting data in this study were retrieved from Part 3 database of the Census Transportation Planning Package 2000 (BTS, 2000), which contains worker commuting flows (in persons) between areas at the selected geographic level, such as county, census tract, and traffic analysis zone (TAZ). Census tract, instead of more disaggregate TAZ, was used as the spatial unit for analysis. This is because upon analyzing changing commuting patterns and economic vulnerabilities, key demographic, social and economic variables, which are not all available at the TAZ level, were explored to uncover subsequent social and spatial inequities for population across the study area. Recent average gasoline price was obtained from the Environmental Information Administration (EIA, 2008), while fuel efficiency ratings for vehicles were gathered from the Bureau of Transportation Statistics (BTS, 2007). A linear, distance-based measure of accessibility was adopted in this study. Distance measures, in the context of commuting, view accessibility as exclusively a function of the spatial separation between two places, typically home or workplaces. Overcoming physical separation is onerous to individual commuters. Thus greater separation implies lower accessibility (Miller, 2005). Here, the accessibility of a census tract defined as the aggregate distance from its centroid to those of all other tracts along the transportation network, can be formulated as:







j ≠ i

where Ai = the measure of accessibility for census tract i, Dij = the shortest-path network distance from the centroid of census tract i to that of census tract j. This measure was once examined by Ingram (1971) to explore integral accessibility. Subsequent modified measures were used in the analysis of accessibility for metropolitan areas at various scales (Allen, et al., 1993; Berquin, 1998). The smaller the value of Ai, the more accessible a tract is in the network; the larger the value of Ai, the less accessible a tract will be. Using the centroid of each census tract as a node in the transportation network of Metro Louisville, the distance of the shortest path along the road network between tracts (Dij) were calculated with ArcGIS Network Analyst, which generated an Origin-Destination cost (i.e., distance) matrix. This O-D cost matrix is a valued graph or L-Matrix (Taaffe, Gauthier and O’Kelly, 1996). The row sum in the matrix represents the level of accessibility for each census tract. The sum of the row sums provides a single-number measure of the overall accessibility involving all the tracts in the study area along the transportation network. Two levels of accessibility were calculated for each tract, one with the normal network and the other with a disrupted network where barriers were set at each end of the Sherman Minton Bridge causing paths using the bridge/link to be rerouted. Since the majority of Jefferson County’s workers work within the county with less than 4 percent commuting outside state of Kentucky, economic impacts of the Sherman Minton Bridge’s disruption were evaluated only for commuters driving across the river from Indiana counties of Clark and Floyd to Jefferson County. The number of commuters from the tract where they lived was combined with the shortest-path network distances to the tracts where they worked, using the normal and disrupted networks separately. This process of assigning commuting flows to the network is essentially “all or nothing” (Taaffe, Gauthier and O’Kelly, 1996), which assumes that commuters choose the shortest path between a given O-D pair. Although not entirely realistic, this approach is practical and acceptable due to the fact that specific capacity restriction and performance curve corresponding to each network link are not available in the study area. The difference in total person-miles between the normal and


disrupted networks then indicates the total increased commuting distances for all the workers from each tract. Using the increase in person-miles, fuel efficiency ratings for U.S. passenger vehicles, and the average U.S. gas price per gallon in late March 2008, increasing commuting expenditure due to the disruption in the road network was assessed. To a large extent, it represents the lower bound of the potential excess commuting expenditure, a useful benchmark, in a condition when commuters have knowledge of and choose the shortest paths to their destinations, and there is no restriction on the traffic that can be handled by the network link at any time. 3. RESULTS AND DISCUSSION 3.1. ACCESSIBILITY OF CENSUS TRACTS Accessibility, in miles, for each census tract under normal road network conditions is shown in Figure 2. In terms of the overall accessibility of the entire study area, the total distance traveling from the centroid of every census tract to that of every other tract along the shortest network paths would be 493,816 miles.

FIGURE 2 CENSUS TRACT ACCESSIBILITY UNDER NORMAL NETWORK Not surprisingly, the most accessible area is located in Jefferson County, where the I65 and 2nd Street bridges connect near downtown Louisville. Census tract 0062 is the most accessible, with a total of 1481.63 miles to all the other tracts. Accessibility decreases roughly in bands radiating from downtown Louisville to the outer edges of all three counties especially northern Floyd and Clark counties. As to the least accessible tracts, Clark County has five, including the top three. Jefferson County contains three, while Floyd County holds two. The central location of those most accessible census tracts within the study area is clearly behind


this observed spatial pattern of accessibility. As early as 1960, Garrison portrayed Louisville as the most accessible city due to it’s centrality to the U.S. highway network. Evidently, this network centrality also works in favor of downtown Louisville at the metropolitan scale. The least accessible census tracts are on the outer edge of the network, leading to much longer distances to all other census tracts. The least accessible census tract is Tract 0510 in Clark County with an aggregate distance of nearly 5800 miles to all the other tracts in the threecounty area. Figure 3 depicts census tract accessibilities under the disrupted network in which the Sherman Minton Bridge was blocked. Figure 4 illustrates changing accessibility in miles for each tract. Under the disrupted road network, the level of total network accessibility for the entire study area is 505,007 miles, marking a rise of 11,191 miles or 2.27 percent over that when the Sherman Minton Bridge is intact. Measured by the loss of accessibility (i.e., increase in aggregate mileage) for census tracts, Floyd County is most vulnerable to the hypothetical failure of the bridge, while the east side of Clark County is least vulnerable. The ten most vulnerable census tracts are all located in Floyd County, with each adding over 250 miles in the measure of accessibility. The worst case is Tract 0706 of Floyd County, whose total distance traveled to all the destinations in the study area would climb by more than 500 miles. This spatial pattern of vulnerability is largely due to the fact that people in western Floyd County rely on the Sherman Minton Bridge for commuting to destinations in Jefferson County along the shortest network paths. Rerouting through the John F. Kennedy Bridge or Clark Memorial Bridge would incur significant extra commuting distances. On the contrary, due to geographical proximity to the other two bridges, Clark County commuters would use the Sherman Minton Bridge least frequently. As a result, Clark County has some least vulnerable census tracts from



the hypothetical disruption of the network. There are eight Clark County census tracts that are not affected by the loss of the Sherman Minton Bridge, and many others are barely impacted by small increments in the aggregate travel distance. Jefferson County has varying levels of vulnerability, with the most vulnerable tracts on the west side of the county and a lessening degree toward the eastern edge of the county.

FIGURE 4 NETWORK VULNERABILITY OF CENSUS TRACTS, MEASURED BY CHANGING ACCESSIBILITY BETWEEN THE NORMAL AND DISRUPTED NETWORKS 3.2. ECONOMIC IMPACT OF NETWORK DISRUPTION ON COMMUTERS The failure or disruption of a transportation network implies significant changes in people’s lives. In addition to exploring the physical vulnerability measured by the loss of accessibility, monetary or economic impacts the disruption imposes on individuals or communities were also analyzed. The economic impacts of commuters’ network detours were examined for a presumably extended period of time, say a year. Specifically, we focused on valuing how the unavailability of the river crossing through the Sherman Minton Bridge affects commuters, and estimated additional fuel costs incurred by the rerouting/detours for census tracts in Clark and Floyd counties where a significant share of workers commutes to Jefferson County. Network vulnerability (i.e., increase in commuting mileage between the disrupted and normal networks) along the shortest path from each census tract in Floyd (or Clark) County to a tract in Jefferson County was multiplied by the number of commuters between them to evaluate the economic vulnerability (i.e., additional fuel cost incurred) of a specific O-D commuting route or flow. The economic vulnerability of a census tract is then measured as the yearly total extra fuel costs for all the commuters from the tract. An average of 250 working


days (U.S. Census Bureau, 2004), an average passenger vehicle fuel efficiency of 22.9 miles per gallon of gasoline (BTS, 2007), as well as the average price per gallon of gasoline of $3.21 in March 2008 (EIA, 2008) were used in the assessment. The twenty most economically vulnerable commuting flows, measured by the total yearly fuel cost increase, are depicted in Figure 5. All these flows originate from Floyd County, 15 of which are from three census tracts, 0711.02, 0711.01 and 0710.04. They are destined toward major employment clusters in Jefferson County, including downtown Louisville, Enterprise Zone and Riverport Industrial Park in western Jefferson, and UPS/Louisville International Airport roughly in the middle of the county. Some of these flows reveal high level of economic vulnerability due to a large number of commuters in the flow and a relatively small increment in commuting distance from the detour. For instance, the flow from Tract 0711.02 to downtown Louisville (Tract 0049) involves 325 commuters and an increase in twoway commuting distance of 4.4 miles, resulting in a largest total yearly increase in fuel costs of almost $48,000. On the other hand, flow from the same tract to the West End of Louisville (Tract 0003) involves only 25 commuters and an escalation in two-way distance of 16 miles, leading to a $13,352 increase in total yearly fuel costs. Thus, either a large number of commuters affected, or a significantly prolonged commuting distance, or both, makes these twenty flows the most costly, in response to the hypothetical failure of the Sherman Minton Bridge.

FIGURE 5 TWENTY MOST ECONOMICALLY-AFFECTED COMMUTING FLOWS In aggregate, commuters from Floyd County would be most vulnerable economically (Figure 6). They would potentially have to cope with a large financial burden, reflected by an additional one million dollars in fuel costs for one year, if the Sherman Bridge were no longer available for travel. Not surprisingly, the census tract in Floyd County that would be hit the


hardest is Tract 0711.02. It shows the largest yearly increase in fuel cost of $256,226.59 due to the detour in daily commuting. Comparatively, Clark County would be less impacted by the bridge failure. As shown in Table 1, the total potential economic impact on Clark County workers commuting to Jefferson County would be $31,959.50, almost $950,000 less than that on Floyd County workers. The most affected census tract in Clark County is the westernmost (0508.01), which records a climb in annual fuel cost by $8,454.97 for its commuters. In sum, a rise of 28,852 person-miles per day (or over 7,213,000 person-miles per year) and a corresponding $1,011,082 in yearly fuel costs, which may represent only the lower limit, is a considerable economic burden to workers in the two counties who rely on the bridge to commute to major employment centers in Jefferson County. With more workers commuting to Louisville and the gasoline price soaring, potential excess commuting expenses people have to bear due to the bridge failure would be even more troublesome.

FIGURE 6 ECONOMIC VULNERABILITY OF CENSUS TRACTS IN FLOYD AND CLARK COUNTIES, MEASURED BY YEARLY EXTRA FUEL COSTS 4. CONCLUSIONS Transportation networks are critical infrastructure elements and face various potential threats. The vulnerability of society to the disruption; the social, economic and environmental costs of the disruption; and many other variables must be examined to correctly prioritize resource allocation for modification, response, and recovery. Using the measure of accessibility based on physical road network and origin-destination commuting data, the focus of this paper is on highlighting the spatial disparities in the loss of accessibility, and measuring the economic impacts on daily commuters, if a crucial bridge connecting counties across the


Ohio River in Metro Louisville were to fail for an extended period of time. Nonetheless, the purpose of this study is not to determine a rigid estimate of the excess commuting and the corresponding monetary costs. Rather, by examining the hypothetically disrupted network from the perspective of changing accessibility, this analysis offers a practical approach to assess user-oriented network vulnerabilities and helps reveal potential burdens imposed upon individual commuters or communities, which are of importance for urban transportation research, planning, and policy making. TABLE 1 ANNUAL EXTRA FUEL COSTS FOR WORKERS IN FLOYD AND CLARK COUNTIES
Floyd County, IN Census Tract Annual Extra Fuel Cost 0711.02 $256,226.59 0711.01 $151,384.22 0710.04 $75,964.72 0712 $72,532.14 0706 $71,676.33 0710.03 $69,378.13 0708.02 $52,288.64 0710.01 $44,584.78 0707 $41,666.72 0708.01 $37,848.47 0705 $27,633.45 0709.01 $24,223.97 0704 $13,714.68 0709.02 $13,024.24 0703.02 $12,581.61 0703.01 $9,824.81 0702 $4,569.27 Clark County, IN Census Tract Annual Extra Fuel Cost 0508.01 $8,454.97 0507.02 $6,111.34 0504.02 $4,936.44 0509.01 $2,945.02 0508.02 $2,255.94 0505.03 $1,817.42 0503.03 $1,668.44 0509.02 $1,028.66 0506.04 $716.58 0503.04 $601.59 0506.03 $412.53 0503.06 $334.06 0510 $0.00 0507.01 $0.00 0506.01 $0.00 0505.04 $0.00 0504.01 $0.00 0503.05 $0.00 0502 $0.00 0501 $0.00 County Total $31,959.50

County Total


We didn’t assign monetary values to other traffic, such as heavy commercial trucks, as well as to variable operating costs due to increased travel distance. Also, some of the dollars spending on extra fuel would have been spent on other local goods and services without this change in travel patterns. So, through multiplier effects, the costs to the local economy as a result of an extended disruption in transportation network would be even more overwhelming. This study suggests some future refinements. Travel time, instead of physical distance, can be used to derive accessibility measures for census tracts. Meanwhile, with reliable network link performance data, a more realistic traffic assignment procedure, such as user equilibrium that accounts for congestion effects can be employed to better estimate commuters’ route choices. Furthermore, levels of network and economic vulnerabilities can be examined in the context of demographic, social, and economic status of the population in a census tract, such that vulnerability in a more general sense highlighting social and spatial inequities can be pursued. 5. REFERENCES Allen, B.W., D. Liu, and S. Singer. 1993. Accessibility Measures of U.S. Metropolitan Areas. Transportation Research 27B(6):439-449.


Birdsall, J.D., R. Hajdin, A. Erath, and K. Axhausen. 2007. Assessing Infrastructure Vulnerability to Sudden Events. 6th Conference on Applied-Infrastructure Research. Berlin, 5-6 October. Berquin, P. 1998. Measuring City Accessibility by Its Public Transport System. Université Catholique de Louvain. Bureau of Transportation Statistics (BTS). 2000. Census Transportation Planning Package (CTPP) 2000 – Part 3: Journey-to-Work (IN) CD. Washington, DC: U.S. Department of Transportation. Bureau of Transportation Statistics (BTS). 2007. National Transportation Statistics. Washington, DC: U.S. Department of Transportation. Chang, S.E. 2003. Transportation Planning for Disasters: An Accessibility Approach. Environment and Planning A 35(6): 1051-1072. Chen, A., C. Yang, S. Kongsomsaksakul, and M. Lee. 2007. Network-Based Accessibility Measure for Vulnerability Analysis of Degradable Transport Networks. Networks and Spatial Economics 7(3):241-256. Department of Employment and Economic Development (DEED). 2007. Economic Impacts of the 1-35W Bridge Collapse. Minnesota Department of Transportation. Energy Information Administration (EIA). 2008. Short-Term Energy Outlook. Last accessed 2 April 2008 Garrison, W.L. 1960. Connectivity of the Interstate Highway System. Papers and Proceedings of the Regional Science Association 6:121-137. Green, M. 2007. Kennedy Bridge Reports Cite Some ‘Deficiencies.’ The Courier-Journal. August 12, 2007, 1A. Jenelius, E., and L. Mattsson. 2006. Developing a Methodology for Road Network Vulnerability Analysis. Nectar Cluster 1 Seminar, Molde University College, Molde, 12-13 May. Miller, H.J. 2003. Transportation and Communication Lifelines Disruption. In: The Geographical Dimensions of Terrorism, pp. 145-152. S. Cutter, D. Richardson and T. Wilbanks, eds. New York and London: Routledge. Miller, H.J. 2005. Place-Based versus People-Based Accessibility. In: Access to Destinations, pp. 63-89. D. Levinson and K. J. Krizek, eds. London: Elsevier. Ingram, D.R. 1971. The Concept of Accessibility: A Search for an Operational Form. Regional Studies 5(2):101-107. Nicholson, A., J. Schmöcker, M.G.H. Bell, and Y. Iida. 2003. Assessing Transport Reliability: Malevolence and User Knowledge. In: The Network Reliability of Transport, pp. 122. M. Bell and Y. Iida, eds. Oxford, UK: Pergamon. Taaffe, E.J., H.L. Gauthier, and M.E. O’Kelly. 1996. Geography of Transportation. Upper Saddle River, New Jersey: Prentice Hall. Taylor, M.A.P., S.V.C. Sekhar, and G.M. D’Este. 2006. Application of Accessibility Based Methods for Vulnerability Analysis of Strategic Road Networks. Networks and Spatial Economics 6(3/4):267-291. Taylor, M.A.P., and G.M. D’Este. 2007. Transport Network Vulnerability: A Method for Diagnosis of Critical Locations in Transport Infrastructure Systems. In: Critical Infrastructure: Reliability and Vulnerability, pp. 9-30. A. Murray and T. Grubesic, eds. New York, NY: Springer. U.S. Bureau of the Census. 2004. New York Has Longest Commute to Work in Nation, American Community Survey Finds. Release/www/ releases/archives/american_community_survey_acs/001695.html. Last accessed 2 April 2008. U.S. Bureau of the Census. 2006. 2006 American Community Survey, Table S0801Commuting Characteristics by Sex, Clark County and Floyd County, IN. Washington, DC: Bureau of the Census.


Sign up to vote on this title
UsefulNot useful