Road or Rail:

A Shortest Path Analysis of Roadway and Rail Travel Times

By: Melissa Foreman

University of Texas at Dallas

Masters of Science, Geographic Information Science

July 28, 2006

Table of Contents

I. Introduction

II. Project Objective / Purpose of Study

III. Literature Review

IV. Data Sources

V. Project Methodology and Analysis......

VI. Results and Discussion......

VII. Project Conclusion

VIII. Further Research......

IX. Terminology......

X. Bibliography and Reference Material

mforemani11/5/2018

List of Figures

Figure 1: The Study Area: 43 Rail Stations ………………………………...... 7

Figure 2: Measuring Rail Travel Time[RB1] ………………………………………………… 10

Figure 3: Bush Turnpike Station Travel Time Contour Maps ……………………….. 13

Figure 4: Roadway and Rail Travel Time Comparison for Bush Turnpike Station I … 14

Figure 5: Roadway and Rail Travel Time Comparison for Bush Turnpike Station II …15

Figure 6: Average Travel Time for all OD Pairs ………………………………………. 16

Figure 7: DallasCounty Urban and Suburban Station Locations …………………… 19

Figure 8: RandonlyRandomly Selected DallasCounty Origin Stations ………………………… 20

Figure 9: Example of Suburban and Urban Station Locations ……………….………. 21

Figure 10: Average Travel Times for DallasCountySuburban OD Pairs …………… 22

Figure 11: Average Travel Time for Urban OD Pairs ……………………………….… 23

List of Tables

Table1: Travel Time Analysis for All O/D Station Pairs ………………………………. 18

Table 2: Travel Time Analysis for DallasCounty Suburban O/D Station Pairs …...... 25

Table 3: Travel Time Analysis for DallasCounty Urban O/D Station Pairs …………. 25

mforemanii11/5/2018

Road or Rail:

A Shortest Path Analysisof Roadway and Rail Travel Times

I. Introduction

Due to sharp increases in demand for highway travel accompanied by small increases in the construction of new highway capacity, roadway congestion continues to be a critical and increasing problem. According to the U.S. Department of Transportation (USDOT) the route miles of highways increased only 1.5% from 1980 to 1999 while vehicle miles of travel (VMT) increased 76%.[1] [RB2]By 2030, the population in the North Central Texas area is expected to exceed 8.59 million. Population growth in combination with an increase of automobile ownership per household and continued urban sprawl will almost certainly increase VMT, as well as roadway travel times, at an alarming rate. Therefore, determining transportation solutions today will be crucial to meet future mobility needs.

Providing fast, affordable, reliable public transportationhas the potential to improve access to opportunities, enable economic prosperity and protect the natural environment. [RB3]The Urban Mobility Study[2] conducted by the Texas Transportation Institute (TTI) has been running studiesfor approximately twenty years.[3]The TTI studies confirmincreases in roadway congestion areaffecting individual trips with longer periods of time spent on American roadways. [RB4]

The majority of trips are drive-alone trips, or single occupancy vehicletrips,that are a major factor in roadway congestion. With roadway capacity increasing at a slower rate than population growth, these types of trips will almost certainly increase peak period roadway congestion and time spent on metropolitan roadway systems. It is assumed that an increase of current roadway networks could relieve congestion thereby relieving travel time on roadways. However, it seems that new lanes tend to fill up with new traffic within a few years, particularly if surrounding routes are at capacity.[RB5]Thus, improvingroadway capacity does not necessarily relieve roadway congestion and decrease roadway travel times, and “improving the reliability of the transportation system is an important issue. Predictable and regular travel times have a certain value for urban travelers and businesses.”[4][RB6]

Increasing the use of public transportation, such as rail transit, also does not provide a solution to roadway congestion. With the future increase of roadway congestion causing an increase in roadway travel times, perhaps it may become faster to travel by rail transit twenty-five years from now.[RB7]

Increasing the use of public transportation also may not provide a solution to roadway congestion. Since bus transit, in most cases, shares the same roadway with all other vehicles, it is impacted by congestion in the same manner as automobiles. Only transit which can operate on a separate right-of-way can potentially avoid this congestion, which includes bus transit using high-occupancy vehicle lanes (HOV) or dedicated bus ways and rail transit. Because rail transit provides the greatest degree of separation from congested highway routes, it has the greatest potential for providing future relief from roadway congestion.[5] This study will examine the relationship between road and rail travel times now, and those that can be expected in the future, to determine if rail transit can provide a faster alternative to congested highways.

II.Project Objective / Purpose of Study

In most cases,today it is faster to travel by roadway thanto travel by passenger rail. The purpose of this study is to answer the question – Under the best circumstances for individuals living near or at passenger rail, will the increase in congestion caused by population growth over the next 25 years result in increased road travel times such that it becomes faster to travel by rail. This question will be examined by comparing the travel times of the current 2005 roadway and the proposed 2030 roadway networks with those of the rail transit network. The Dallas Area Rapid Transit (DART) rail network operating in the City of Dallas and surrounding suburban cities provides the basis for the study.

III. Literature Review

A report submitted to the Federal Highway Administration, October 2003, addressed the impact of congestion on bus operations. The purpose of this study was “to quantify the impact of traffic congestion on bus operations and costs, and to forecast the future impacts of congestion on operations and costs.”[6] This study found the increment increase in bus travel vehicle hours due to the increase in roadway traffic time over time has a small, but compelling adverse impact on bus travel time. As expected, there was an increase of bus roadway travel time during the peak hours. This would seem evident since buses and cars travel on the same roadway network and share the same congestion. The difference between studying the increment increase of bus travel time to that of rail travel time is that both utilize different transportation networks. Unlike buses, passenger trains travel on rail lines with minimal interaction with roadway congestion. However, many buses travel on fixed routes with a predetermined schedule of departure times and arrival times, as well as having special lanes designated on the roadways. Though the increase in travel time was not at the same rate as automobile travel times, the travel time of buses, with special conditions,were still slightly increased by roadway congestion. Therefore, a study addressing the impact of congestion on bus operations provides some support for a study on how the increase in roadway congestion may impact road travel time relative to rail travel times, but does not provide a truly comparable answer.

Rail Transit has only been seriously considered in mobility congestion studies within the last five years. Perhaps its exclusion hitherto is due to the differences between what constitutes congestion and free-flow travel time on roadways as opposed to rail transit. As asserted in the Urban Mobility Report: Vol2, “a significant potential source of confusion is in translating the roadway mobility concept of congestion into urban transit operations.”[7] For example, rail transit can be dependent on speed, station spacing, and the amount of interface with general roadway traffic. Automobiles can be dependent on roadway volumes, speed of the traffic, and the occasional roadway obstacles such as construction, wrecks or other forms of blockage. The different characteristics between roadway and rail congestion would affectthe travel times of these two modes.

IV. Data Sources

The majority of thedata in this study was obtained from the transportation department of North Central Texas Council of Governments (NCTCOG) which serves as the metropolitan planning organization (MPO) for the Dallas - Fort Worth (DFW) area.“The MPO works with state and local governments, the private sector, and the region’s citizens to plan coordinated transportation systems designed to move goods and people affordably, efficiently, and safely.”[8] NCTCOG maintains a regional travel model (RTM) for the DFW area. This model provided the basis for the travel time estimates used in this study.

RTMs can have limitations in their ability to replicate actual public transportation use, especially at the individual route level. Limitations can be derived from characteristics such as: Transportation Survey Zones (TSZs) that may be too large toadequately predict transit accessibility;vagueknowledge of demographic data; trip distribution that maynot fully account for mode decision factors; and transit assignment that couldmisrepresent a passenger’s choice of route.[RB8]Regardless of these limitations,approximate forecast values of a population’s travel behavior can be obtained from RTMs. Indeed, these are the foundation for all transportation planning in the DFW region and these were the basis for the travel time estimates used in this study.

The study was limited to individuals living near rail stations to maintain a controlled environment. The current year 2005 rail station locations, as shown in Figure 1, were used as the origination and destination points.

mforeman- 1 -11/5/2018

Figure 1: The Study Area: 43 Rail Stations

V. Project Methodology and Analysis

If congested roadways are related to slower roadway travel times, then an increase in future population should lead to an increase in roadway congestion and continued slower roadway travel times. Therefore, the difference between roadway travel times and passenger rail travel times should be larger in 2030 than in 2005. This study compares 2005 road travel times and projected 2030 road travel times relative to constant rail travel times between 43 existing rail stations in the Dallas - Fort Worth metropolitan area.

Shortest Path algorithms areis used to make this comparison. Shortest Path solutions involve finding a path with the smallest possible length (or cost) by summing the individual arcs in this path. In this study, travel time will represent the cost of a path. Roadway Travel Times (RTT) will be found for 2005 and for 2030. Rail Travel Time (RLTT) will remain constant under the assumption that rail technology does not change from 2005 standards. [RB9]

The shortest path methodology incorporated in Caliper’s TransCAD software and incorporated ina part of the NCTCOG RTMwas utilized to measure travel times. The methodology includes, butis not limited to, Dijkstra’s shortest path algorithm[9] with non-negative arc lengths, as well as a labeling method to find all shortest paths from a source to selected vertices.[RB10] For this study, all possible OD pairs for the forty-three rail stations were used to generate a dataset of 1,806 possible shortest path routes. After the shortest path algorithm was run in TransCAD, the data was exported to ESRI ArcView for some additional analysis on travel times using proportional measures.

The standard procedure for finding shortest paths involves selecting two nodes, an OD pair, on a point layer. The cost of travel between the two points is then calculated based on a variety of measures such as peak or off-peak travel times, distance, number of turns, and other factors important to the researcher. This study uses peak travel time as a measure for the shortest path costs.

The standard method in TransCAD limits the analysis of routes to one shortest path route at a time. Based on the number of routes under evaluation, the standard method for shortest path evaluationcan be a tedious and time consuming process. Therefore, a preferred method was developed to find the shortest path. A macro was written to solve for multiple shortest path routes from any number of origination points to any number of destination points. After the shortest path macro was run, a text report of the origin and destination (OD) points, along with the travel time for roadway and rail was posted to a text report. The report produced in this study is useful for a variety of planning studies such as multi-modal corridor research and rail transit alternative analysis.

Prior to solving shortest path routes, several pieces of information had to be entered into the North Central Texas Council of Government’s Dallas Fort Worth Regional Transportation Model (NCTCOG DFW RTM) which, as indicated earlier, provided the basis for the travel time estimates used in this study. Roadway travel time components entered into the model were arterial speed, capacity and distance. [RB11]Rail travel time components required by the RTM to calculate travel times are more complex than roadway travel time components. They include speed, distance, acceleration and deceleration rates, and rail transit headway times. Additionally, there are several different types of passenger rail equipment such as Light Rail Trains (LRT), Commuter Rail Trains (CRT), and Bus Rapid Transit (BRT). The RLTT formula and assumptions used are shown in Figure 2 and relate to LRT and CRT since these are the two equipment types in the DART system studied here.

mforeman- 1 -11/5/2018

Figure 2: Measuring Rail Travel Time[RB12]

After data entry, the RTM was run to create an output dataset to use for travel time analysis. The RTM output data included roadway capacity volumes per link based on time-of-day, ridership and transfer data for public transportation, type of mode usage, and various demographic and employment characteristics. Geographic and spatial analysis was performed on the RTM output data using spatial joins in Caliper’s TransCAD and ESRI’s ArcView software. The roadway and rail transit networks were joined to time-of-day trip volume matrices resulting in the data used to examine the travel time differences between road and rail. The spatial analysis provided information as to whether the location of a station, urban or suburban, has an effect on roadway travel times relative to rail travel times.

When interpreting the results of this study, several aspects of the methodology employed should be kept in mind. Travel timesare measured from the point of origin to the point of destination. Since the intent of this study is to compare road and rail travel times, rail stations represent originand destination points for both road and rail travel times. This is realistic for travelers whose residence is close to the railroad stations, such as would be the case in transit-oriented residential developments. It is less realistic for persons more distant from rail stations who in reality would be expected to incur additional road-based travel times to access the rail network. Since these access times were not included,it means that rail travel times are being given the greatest possible advantage relative to road travel times in the comparisons presented here. The importance of this point, and the fact that it is not as problematic as it might first appear, will become evident when the results are discussed. Additionally, it shouldbe noted that special generator rail stations,such as Ameritech Field in Arlington, are not be included in this study. Lastly, it should be emphasized that the comparison between road and rail is based strictly on travel times. Other issues potentially relevant to a road/rail comparison such as engineering and operating costs, environmental impacts, and air quality effects, are purposefully not included.

mforeman- 1 -11/5/2018

VI. Results and Discussion

Travel Time Contours[RB13]

A brief review of theroadway congestion’s effect on roadway travel times can be beneficial prior to presenting the results of the shortest path data analysis. Travel Time Contour maps are a useful device for this purpose. They illustrate how the increases in roadway congestion from 2005 to 2030 effect roadway travel times.

Travel Time contours graphically display the time required to travel from a single origin to any destination. They can be created in Caliper’s TransCAD software using the network analysis tool. The process begins by selecting anorigin point, maximum travel time impedance, and defining the intervals based on a variety of measurements such as time or distance. An illustration of how Travel Time Contour Maps can be used to show the difference in roadway travel times near the I635/I35E intersection in 2005 and 2030 can be seen in Figure 3; the Bush Turnpike station is usedas the origin.The maximum impedance was set at thirty minutes with five minute intervals, and peak hour travel times were used as the system of measurement. Due to the increase inroadway congestion, travel times to the area circled in both maps can be an indication of higher roadway travel times in 2030 than in 2005. An individual traveling by road in 2005 could reach the IH35/I635 interchange within thirty minutes. By 2030, that same individual will take more than thirty minutes to reach the same area.