<%BANNER%>

Testing Of, and Enhancements To, Two-Lane Highway Modeling in Corsim

Permanent Link: http://ufdc.ufl.edu/UFE0042649/00001

Material Information

Title: Testing Of, and Enhancements To, Two-Lane Highway Modeling in Corsim
Physical Description: 1 online resource (143 p.)
Language: english
Creator: Hammontree, Heather
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: ats, average, based, capacity, corsim, density, directional, flow, follower, following, freedom, grade, hcm, heavy, highway, highways, identification, impeded, lanes, manual, measures, operations, overtakings, passing, percent, performance, probability, ptsf, speed, spent, splits, time, transportation, travel, truck, two, types, vehicles, zones
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Two-lane highways make up a large part of the roadway network in the United States and are important for transporting people and goods across long distances. When two-lane highways begin to operate poorly, it is important that the conditions are improved, but improvements can be costly as two-lane highways span a long distance. Therefore, it is important to have accurate tools available for analyzing two-lane highways in order to make appropriate improvements. Recently, a two-lane highway modeling capability has been implemented into the simulation program, CORSIM. CORSIM is now the only U.S.-based simulation program that is capable of modeling two-lane highways while being compatible with modern computers. The main analytical tool that has been used in the U.S. for two-lane highway analysis is the Highway Capacity Manual (HCM), but it is not yet known how the results from these two tools compare. This research provides an extensive comparison of Percent Time-Spent-Following (PTSF) and Average Travel Speed (ATS), the two primary performance measures used in the HCM analysis methodology, as estimated from the HCM and CORSIM, including the speed-flow relationship and the PTSF-flow relationship. Guidance is provided for setting up corresponding networks between CORSIM and the HCM. CORSIM is also used to estimate two-lane highway directional capacity and the effects of passing lanes are tested and discussed. Other performance measures that are not included in the HCM are critiqued and ranked based on a variety of categories. Finally, recommendations are made for improvements and enhancements to CORSIM and the HCM as well as areas for future two-lane highway research.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Heather Hammontree.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Washburn, Scott S.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042649:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042649/00001

Material Information

Title: Testing Of, and Enhancements To, Two-Lane Highway Modeling in Corsim
Physical Description: 1 online resource (143 p.)
Language: english
Creator: Hammontree, Heather
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: ats, average, based, capacity, corsim, density, directional, flow, follower, following, freedom, grade, hcm, heavy, highway, highways, identification, impeded, lanes, manual, measures, operations, overtakings, passing, percent, performance, probability, ptsf, speed, spent, splits, time, transportation, travel, truck, two, types, vehicles, zones
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Two-lane highways make up a large part of the roadway network in the United States and are important for transporting people and goods across long distances. When two-lane highways begin to operate poorly, it is important that the conditions are improved, but improvements can be costly as two-lane highways span a long distance. Therefore, it is important to have accurate tools available for analyzing two-lane highways in order to make appropriate improvements. Recently, a two-lane highway modeling capability has been implemented into the simulation program, CORSIM. CORSIM is now the only U.S.-based simulation program that is capable of modeling two-lane highways while being compatible with modern computers. The main analytical tool that has been used in the U.S. for two-lane highway analysis is the Highway Capacity Manual (HCM), but it is not yet known how the results from these two tools compare. This research provides an extensive comparison of Percent Time-Spent-Following (PTSF) and Average Travel Speed (ATS), the two primary performance measures used in the HCM analysis methodology, as estimated from the HCM and CORSIM, including the speed-flow relationship and the PTSF-flow relationship. Guidance is provided for setting up corresponding networks between CORSIM and the HCM. CORSIM is also used to estimate two-lane highway directional capacity and the effects of passing lanes are tested and discussed. Other performance measures that are not included in the HCM are critiqued and ranked based on a variety of categories. Finally, recommendations are made for improvements and enhancements to CORSIM and the HCM as well as areas for future two-lane highway research.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Heather Hammontree.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Washburn, Scott S.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042649:00001


This item has the following downloads:


Full Text

PAGE 1

1 TESTING OF, AND ENHANCEMENTS TO, TWO LANE HIGHWAY MODELING IN CORSIM By HEATHER HAMMONTREE A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Heather Hammontree

PAGE 3

3 To The University of Florida Steel Bridge Team

PAGE 4

4 ACKNOWLEDGMENTS I have many people to thank for their help in completing this thesis and graduate school. First, I would like to thank my parents and friends for their support throughout my time in school. I especially thank the friends who gave advice about the literature review, gave tips on how to expedite the testing process and gave me general guidance throughout my graduate studies I thank Tom Simmerman for making all of the necessary changes to CORSIM and Jing Li for answering questions and helping throughout my research. I would like to thank the professors at the University of Florida Transportation Research Center, especially my supervisory committee members, Dr. Yafeng Yin and Mr. Bill Sampson. Most of all, I would like to thank my advisor and supervisory committee chair, Dr. Scott Washburn, for his help and guidance throughout the course of this study.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 ABSTRACT ................................................................................................................... 1 2 CHAPTER 1 INTRODUCTION .................................................................................................... 14 Background ............................................................................................................. 14 Problem Statement ................................................................................................. 14 Research Objectives and Tasks ............................................................................. 15 Document Organization .......................................................................................... 16 2 LITERATU RE REVIEW .......................................................................................... 18 Overview ................................................................................................................. 18 2000 HCM and 2010 HCM ...................................................................................... 18 Criticisms of the HCM Methodology ........................................................................ 22 Overestimation of PTSF ................................................................................... 23 Poor Method for Estimating BFFS .................................................................... 23 SpeedFlow Relationship .................................................................................. 23 Deterministic Method for Identifying Follower Status ........................................ 24 Other Performance Measures More Applicable for LOS Determination ........... 25 Poor Method for Indicating Necessary Improvements ...................................... 26 Overestimation of Performance Measure Improvements Due to Passing Lanes ............................................................................................................ 27 Limitations During Saturated Conditions and Near Signalized Intersections .... 27 Additional Performance Measures .......................................................................... 29 Percent Impeded .............................................................................................. 29 Probability Based Follower Identification .......................................................... 31 Follower Density ............................................................................................... 33 Freedom of Flow .............................................................................................. 35 Overtakings ...................................................................................................... 36 Passing Lanes ........................................................................................................ 37 HCM and Passing Lanes .................................................................................. 38 Passing Lane Location ..................................................................................... 38 Effects on Other Performance Measures ......................................................... 39 Passing Procedure ........................................................................................... 39 Driver Decision to Pass .................................................................................... 40 Passing Sight Distance..................................................................................... 40

PAGE 6

6 Capacity .................................................................................................................. 44 Simulation Tools ..................................................................................................... 47 TWOPAS .......................................................................................................... 47 TRARR ............................................................................................................. 48 CORSIM ........................................................................................................... 48 3 RESEARCH APPROACH ....................................................................................... 53 CORSIM and 2010 HCM Comparison .................................................................... 53 Pr eliminary Experimental Designs .................................................................... 54 Existing upstream conditions ..................................................................... 54 Truck type .................................................................................................. 55 Passing zone configuration ........................................................................ 56 Speed vs. flow rate relationship and PTSF vs. flow rate ............................ 56 CORSIM/HCM Experimental Design ................................................................ 57 Flow rates and splits .................................................................................. 57 Percent heavy vehicles .............................................................................. 57 Grades ....................................................................................................... 58 Percent nopassing zones .......................................................................... 58 Passing lanes ............................................................................................. 59 CORSIM Testing Facility .................................................................................. 60 Capacity .................................................................................................................. 60 Performance Measures ........................................................................................... 61 4 RESULTS ............................................................................................................... 67 Preliminary CORSIM Test Results .......................................................................... 67 Existing Upstream Conditions .......................................................................... 67 Truck Type Distribution..................................................................................... 68 Passing Zone Configuration ............................................................................. 69 Speed vs. Flow Rate and PTSF vs. Flow Rate Relationship ............................ 70 2010 HCM and CORSIM Comparison Results ....................................................... 71 0% No Passing Zone with 0% Grade ............................................................... 71 0% No Passing Zone with 6% Grade ............................................................... 72 50% NoPassing Zone with 0% Grade ............................................................. 74 50% NoPassing Zone with 6% Grade ............................................................. 74 100% NoPassing Zone with 0% Grade ........................................................... 75 100% NoPassing Zone with 6% Grade ........................................................... 75 Passing Lanes .................................................................................................. 76 0% nopassing zones ................................................................................. 77 50% nopassing zones ............................................................................... 79 100% nopassing zones ............................................................................. 81 Passing Lane Field Observations ..................................................................... 82 Capacity Estimate ................................................................................................... 83 Performance Measures Analysis ............................................................................ 84 Percent Impeded .............................................................................................. 84 Probability Based Follower Identification .......................................................... 85

PAGE 7

7 Follower Density ............................................................................................... 86 Freedom of Flow .............................................................................................. 87 Overtakings ...................................................................................................... 87 Qualitative Analysis Results ............................................................................. 88 Quantitative Analysis Results ........................................................................... 89 5 CONCLUSIONS AND RECOMMENDATIONS ..................................................... 130 Overview ............................................................................................................... 130 Research Findings ................................................................................................ 130 CORSIM Finding s and Recommendations ........................................................... 132 Preliminary Tests ............................................................................................ 132 Passing Procedures ....................................................................................... 133 Passing Lanes ................................................................................................ 135 2010 HCM Findings and Recommendations ........................................................ 135 Recommendations for Future Research ............................................................... 137 LIST OF REFERENCES ............................................................................................. 140 BIOGRAPHICAL SKETCH .......................................................................................... 143

PAGE 8

8 LIST OF TABLES Table page 2 1 Level of service thresholds as a function of highway class ................................. 51 2 2 Threshold values for twolane highway facility LOS ........................................... 51 2 3 LOS thresholds for follower density .................................................................... 51 2 4 LOS thresholds for freedom of flow compared to PTSF and twoway flow ......... 51 2 5 AASHTO components and values of passing sight distance .............................. 52 3 1 Example inputs for HCM and CORSIM .............................................................. 64 3 2 General CORSIM inputs for lead up length tests ................................................ 64 3 3 General CORSIM inputs for truck type distribution tests .................................... 65 3 4 Scenarios for CORSIM truck type distribution test .............................................. 65 3 5 Scenarios for CORSIM no passing zone configuration test ................................ 65 3 6 General CORSIM inputs for nopassing zone configuration test ........................ 66 3 7 General input for speedflow relationship and PTSF vs. flow rate test ............... 66 3 8 Variables used in HCM and CORSIM testing ..................................................... 66 4 1 Lead up and follow up passing results .............................................................. 127 4 2 CORSIM capacity estimates 0% heavy vehicles .............................................. 127 4 3 CORSIM capacity estimates 10% heavy vehicles ............................................ 127 4 4 Performance measure rankings ........................................................................ 128 4 5 Follower density difference between 10% and 0% trucks no passing lane ...... 128 4 6 Follower density difference between 10% and 0% truck s, passing lane ........... 129

PAGE 9

9 LIST OF FIGURES Figure page 3 1 Schematic of CORSIM lead up length test for threemile facility ........................ 63 3 2 Schematic of CORSIM truck type distribution test facility ................................... 63 3 3 Two lane highway facility in CORSIM with passing lane .................................... 63 4 1 CORSIM westbound platoons ............................................................................ 91 4 2 Middle segment PTSF for passing allowed ........................................................ 92 4 3 Middle segment PTSF for passing not allowed ................................................... 92 4 4 Middle segment ATS for passing allowed ........................................................... 93 4 5 Middle segment ATS for passing not allowed ..................................................... 93 4 6 PTSF vs. facility length for passing allowed ....................................................... 94 4 7 PTSF vs. facility length for passing not allowed .................................................. 94 4 8 Truck type test PTSF vs. flow rate for passing allowed ................................... 95 4 9 Truck type test PTSF vs. flow rate for passing not allowed ............................. 95 4 10 Truck type test ATS vs. flow rate for passing allowed ..................................... 96 4 11 Truck type test ATS vs. flow rate for passing not allowed ............................... 96 4 12 Average PTSF vs. flow rate ................................................................................ 97 4 13 Average ATS vs. flow rate .................................................................................. 97 4 14 PTSF vs. flow rate .............................................................................................. 98 4 15 ATS vs. flow rate ................................................................................................ 98 4 16 PTSF vs. two way flow rate 0% grade, 0%NPZ, 0%HV, no passing lane ........ 99 4 17 PTSF vs. two way flow rate 0% grade, 0%NPZ, 10%HV, no passing lane ...... 99 4 18 Platoon s tructure for 3200 veh/h flow rate ........................................................ 100 4 19 ATS vs. two way flow rate 0% grade, 0%NPZ, 0%HV, no passing lane ........ 101 4 20 ATS vs. two way flow rate 0% grade, 0%NPZ, 10%HV, no passing lane ...... 101

PAGE 10

10 4 21 PTSF vs. two way flow rate 6% grade, 0%NPZ, 0%HV, no passing lane ...... 102 4 22 PTSF vs. two way flow rate 6% grade, 0%NPZ, 10%HV, no passing lane .... 102 4 23 ATS vs. two way flow rate 6% grade, 0%NPZ, 0%HV, no passing lane ........ 103 4 24 ATS vs. two way flow rate 6% grade, 0%NPZ, 10%HV, no passing lane ...... 103 4 25 PTSF vs. two way flow rate 6% grade, 0%NPZ, 10%HV, no passing lane .... 104 4 26 ATS vs. two way flow rate 6% grade, 0%NPZ, 10%HV, no passing lane ...... 104 4 27 PTSF vs. two way flow rate 0% grade, 50%NPZ, 0%HV, no passing lane .... 105 4 28 PTSF vs. two way flow rate 0% grade, 50%NPZ, 10%HV, no passing lane .. 105 4 29 ATS vs. two way flow rate 0% grade, 50%NPZ, 0%HV, no passing lane ...... 106 4 30 ATS vs. two way flow rate 0% grade, 50%NPZ, 10%HV, no passing lane .... 106 4 31 PTSF vs. two way flow rate 6% grade, 50%NPZ, 0%HV, no passing lane .... 107 4 32 PTSF vs two way flow rate 6% grade, 50%NPZ, 10%HV, no passing lane .. 107 4 33 ATS vs. two way flow rate 6% grade, 50%NPZ, 0%HV, no passing lane ...... 108 4 34 ATS vs. two way flow rate 6% grade, 50%NPZ, 10%HV, no passing lane .... 108 4 35 PTSF vs. two way flow rate 0% grade, 100%NPZ, 0%HV, no passing lane .. 109 4 36 PTSF vs. two way flow rate 0% grade, 100%NPZ, 10%HV, no passing lane 109 4 37 ATS vs. two way flow rate 0% grade, 100%NPZ, 0%HV, no passing lane .... 110 4 38 ATS vs. two way flow rate 0% grade, 100%NPZ, 10%HV, no passing lane .. 110 4 39 PTSF vs. two way flow rate 6% grade, 100%NPZ, 0%HV, no passing lane .. 111 4 40 PTSF vs. two way flow rate 6% grade, 100%NPZ, 10%HV, no passing lane 111 4 41 ATS vs. two way flow rate 6% grade, 100%NPZ, 0%HV, no passing lane .... 112 4 42 ATS vs. two way flow rate 6% grade, 100%NPZ, 10%HV, no passing lane .. 112 4 43 Passing lane bottleneck .................................................................................... 113 4 44 PTSF vs. distance 0% grade, 0%NPZ, 0% HV, passing lane ........................ 114 4 45 PTSF vs. distance 0% grade, 0%NPZ, 10% HV, passing lane ....................... 114

PAGE 11

11 4 46 ATS vs. distance 0% grade, 0%NPZ, 0% HV, passing lane ........................... 115 4 47 ATS vs. distance 0% grade, 0%NPZ, 10% HV, passing lane ......................... 115 4 48 PTSF vs. distance 6% grade, 0%NPZ, 0% HV, passing lane ........................ 116 4 49 PTSF vs. distance 6% grade, 0%NPZ, 10% HV, passing lane ....................... 116 4 50 ATS vs. distance 6% grade, 0%NPZ, 0% HV, passing lane ........................... 117 4 51 ATS vs. distance 6% grade, 0%NPZ, 10% HV, passing lane ......................... 117 4 52 PTSF vs. distance 0% grade, 50%NPZ, 0% HV, passing lane ....................... 118 4 53 PTSF vs. distance 0% grade, 50%NPZ, 10% HV, passing lane ..................... 118 4 54 ATS vs. distance 0% grade, 50%NPZ, 0% HV, passing lane ......................... 119 4 55 ATS vs. distance 0% grade, 50%NPZ, 10% HV, passing lane ....................... 119 4 56 PTSF vs. distance 6% grade, 50%NPZ, 0% HV, passing lane ....................... 120 4 57 PTSF vs. distance 6% grade, 50%NPZ, 10% HV, passing lane ..................... 120 4 58 ATS vs. distance 6% grade, 50%NPZ, 0% HV, passing lane ......................... 121 4 59 ATS vs. distance 6% grade, 50%NPZ, 10% HV, passing lane ....................... 121 4 60 PTSF vs. distance 0% grade, 100%NPZ, 0% HV, passing lane ..................... 122 4 61 PTSF vs. distance 0% grade, 100%NPZ, 10% HV, passing lane ................... 122 4 62 ATS vs. distance 0% grade, 100%NPZ, 0% HV, passing lane ....................... 123 4 63 ATS vs. distance 0% grade, 100%NPZ, 10% HV, passing lane ..................... 123 4 64 PTSF vs. distance 6% grade, 100%NPZ, 0% HV, passing lane ..................... 124 4 65 PTSF vs. distance 6% grade, 100%NPZ, 10% HV, passing lane ................... 124 4 66 ATS vs. distance 6% grade, 100%NPZ, 0% HV, passing lane ....................... 125 4 67 ATS vs. distance 6% grade, 100%NPZ, 10% HV, passing lane ..................... 125 4 68 CORSIM passing lane configuration ................................................................. 126 4 69 SR40 passing lane configuration ..................................................................... 126 4 70 Follower density vs. two way flow rate 50 / 50 split, 0%NPZ, no passing lane 126

PAGE 12

12 Abstract o f Thesis Presented t o t he Graduate School o f t he University o f Florida in Partial Fulfillment o f t he Requirements f or t he Degree o f Master of Engineering TESTING OF, AND ENHANCEMENTS TO, TWO LANE HIGHWAY MODELING IN CORSIM By Heather Hammontree December 2010 Chair: Scott Washburn Major: Civil Engineering Two lane highways make up a large part of the roadway network in the United States and are important for transporting people and goods across long distances. When twolane highway s begin to operate poorly, it is important that the conditions are improved, but improvements can be costly as twolane highways span a long distance. Therefore, it is important to have accurate tools available for analyzing twolane highways in order to make appropriate improvements. Recently, a twolane highway modeling capability has been implemented into the simulation program, CORSIM. CORSIM is now the only U.S. based simulation program that is capable of modeling two lane highways while being compatibl e with modern computers. The main analytical tool that has been used in the U.S. for two lane highway analysis is the Highway Capacity Manual ( HCM ) but it is not yet known how the results from these two tools compare. This research provides an extensive comparison of Percent TimeSpent Following ( PTSF ) and Average Travel Speed ( ATS) the two primary performance measures used in the HCM analysis methodology, as estimated from the HCM and CORSIM including the speedflow relationship and the PTSF flow relat ionship. Guidance is provided for

PAGE 13

13 setting up corresponding networks between CORSIM and the HCM. CORSIM is also used to estimate twolane highway directional capacity and the effects of passing lanes are tested and discussed. Other performance measures that are not included in the HCM are critiqu ed and ranked based on a variety of categories. Finally, recommendations are made for improvements and enhancements to CORSIM and the HCM as well as areas for future twolane highway research.

PAGE 14

14 CHAPTER 1 INTRODUCTION Background Two lane highways are an important part of the roadway system in the United States and in many other countries They serve as major connectors between areas of high activity and interest and provide routes for commuters and recreational use. Twolane highways provide cost effective, reliable access to areas of low population. They usually carry low volumes and tend to have few interruptions from traf fic signals ( 1 ) Two lane highways may also have occasional passing lanes which are intended to allow faster vehicles to safely pass slower vehicles The defining characteristic of t wo lane highways is that they generally allow drivers to pass slower vehicles by going into the lane of oncoming traffic if there is a large enough gap to permit this maneuver and if the sight distance is adequate. As traffic volume increases, more vehicles are likely to want to pass. However, higher traffic volume can mean fewer gaps, which means there are fewer opportunities to pass. This leads to higher vehicle delays due to faster vehicles being stuck in platoons behind slower vehicles and causes the twolane highway to operate poorly before it is even near capacity ( 1 ) Problem Statement Although two lane highways are a very important element to roadway systems, there are few tool s available to model twolane highway traffic operations and also provide estimates of the performance measures appropriate for the basis of level of service (LOS) classification The HCM is used as the standard analysis tool in the United States, but its analysis methodology for twolane highways has been the subject

PAGE 15

15 of much debate and criticism in recent years. M any researchers believe that significant revisions to the HCMs models and performance measures are necessary One of the greatest challenges of twolane highway research is field data collection. The behavior and measures of interest for twolane highways take place over significant distances and spans of time. Thus, field data collection for twolane highways tends to be very expensive and timeconsuming, often to the point of being impractical. Consequently, the research community often must rely on simulation. However, the primary simulation tool used in past research (and in the development of the HCM analysis methodology), is based on software technology that no longer functions on modern computers In order to facilitate improvements and enhancements to the HCMs analysis procedure, a new simulation tool is required. Such a tool has recently been developed ( 2 ) More specifically, this tool was created by incorporating twolane highway modeling into the CORSIM microsimulation program (3 ) CORSIM is one of the most commonly used simulation programs in the United States due to its low cost and reliable results W hile the twolane modeling capability underwent preliminary testing as part of the implementation project, more comprehens ive testing is still required. Additionally, it is necessary to consider some of the additional performance measures that have been pr oposed in the literature for possible implementation in CORSIM. Research Objectives and Tasks The objectives of this study are to: Perform comprehensive testing of twolane highway modeling in CORSIM to verify consistency with traffic flow theory. M ak e rec ommendations for revisions, as necessary, to the twolane highway modeling methodology in CORSIM to improve its accuracy and usability.

PAGE 16

16 Use CORSIM to assess the validity of the HCM speedflow and percent timespent following flow relationships. Use CORSIM to develop more guidance on capacity values of twolane highways than currently provided in the HCM. Use CORSIM to evaluate the effects of passing lanes on performance measures and capacity. Identify other performance measures that should undergo further c onsideration for use in assessing twolane highway LOS. The following tasks were completed in order to accomplish the objectives of this research: Completed a literature review on the HCM two lane highway analysis criticisms of the HCM m ethodology, other proposed performance measures, effects of passing lanes on capacity and performance measures and simulation tools used for twolane highway analysis Developed an experimental design for testing PTSF and ATS in the HCM and CORSIM Developed an experimental design for testing capacity in CORSIM Compared current two lane highway methodology in CORSIM with the 2010 HCM methodology for PTSF and ATS for two lane highways with and without passing lanes. Analyzed the effects of passing lanes on capacity and perf ormance measures. Evaluated the appropriateness of different performance measures for analyzing two lane highways. Document Organization Chapter 2 presents a literature rev iew on current twolane highway methodologies and tools, performance measures that have been proposed for determining twolane highway LOS and how passing lanes a ffect two lane highway operations Chapter 3 explains the procedures and test scenarios used to compare different methodologies and performance measures and how to estimate capacity using CORSIM Chapter 4

PAGE 17

17 presents the results and findings of the research and provides explanations for trends and differences Chapter 5 presents conclusions and recommendations for further research for two lane highways and CORSIM improvements

PAGE 18

18 CHAPTER 2 LITERATURE REVIEW Overview This chapter describes the HCM twolane highway analysis methodology (the primary analytical method currently used in the United States ) and criticisms of the HCM methodology It also discusses other performance measures that have been proposed for use in evaluating the operations of two lane highways and methods for determining the e ffects of passing lanes on twolane highway s. This chapter also describes methods for determining twolane highway capacity and sim ulation tools capable of modeling two lane highways. 2000 HCM and 2010 HCM The 2010 HCM two lane highway methodology is very similar to the 2000 HCM version, but there are a few minor revisions One change between the 2000 HCM and the 2010 HCM is that in the 2000 edition, there are two separate procedures for finding the performance measures for a twoway analysis and for a directional analysis and the 2010 edition only has a methodology for a directional analysis. The 2010 HCM can be used for a twowa y analysis by analyzing each direction separately. Since the 2010 HCM is the most recent edition, this section does not go into detail describing the 2000 HCM two way analysis procedure ( 1 4 ). There a re slight changes in the values for certain adjustment factors such as the grade adjustment factor ( fg) and the heavy vehicle adjustment factor ( fHV) between the 2000 HCM and the 2010 HCM. The 2010 HCM displays more precise values for these adjustment factors because they are based on smaller increments of input values instead of having one adjustment factor for a broad range of input values. For example,

PAGE 19

19 in the 2000 HCM, the grade adjustment factor for ATS on rolling terrain is 0.71 for a directional flow rate between 0 and 300 pc/h whereas the 2010 HCM gives different grade adjustment factors for directional demand flow rates in increments of 100 (0.67 for a directional demand flow rate less than or equal to 100, 0.75 for a directional d emand flow rate equal to 200) ( 1 4 ). Another change is that the 2000 edit ion has only two highway classes while the 2010 edition has three. Both editions use ATS and PTSF as the performance measures to determine the LOS of class I highways and only PTSF for class II highways The LOS of class III highways is determined by Percent FreeFlow Speed ( PFFS ) ( 1 4 ). The first step in the 2010 HCM methodology is collecting accurate input information and making the necessary adjustments beginning with speed. If the speed i s measured from the field, then the average speed of the f ield data can be used to approximate the freeflow speed ( FFS ) as long as the two way flow rate is no more than 200 pc/h. If the two way flow rate exceeds 200 pc/h, then Equation 2 1 can be used to approximate the FFS : ATS HV f FMf v S FFS,00776 ( 2 1 ) where, FFS = estimated freeflow speed (mi/h) SFM = average field measured speed (mi/h) vf = flow rate observed in both directions during data collection period (veh/h) fHV ,ATS = heavy vehicle adjustment factor

PAGE 20

20 If field data is not available, then the FFS can be deter mined by estimating a base freeflow speed ( BFFS ) and making adjustments according to Equation 2 2 The BFFS is often estimated to be 10 mi/h higher than the posted speed limit A LSf f BFFS FFS ( 2 2 ) where, FFS = estimated freeflow speed (mi/h) BFFS = base freeflow speed (mi/h) fLS = lane and shoulder width adjust ment factor fA = access point adjustment factor The second step in the 2010 HCM methodology is to determine the passenger car equivalent demand flow rate based on adjustments to the hourly demand volume. This can be accomplished by using Equation 2 3 Once the FFS and demand volume are adjusted, the average travel speed can be found using Equation 2 4: ATS HV ATS g i ATS if f PHF V v, ( 2 3 ) where, i = d for analysis direction and o for opposing direction vi ,ATS = demand flow rate for i (pc/h) Vi = peak hour demand volume for i (veh/h) PHF = peak hour factor fg,ATS = grade adjustment factor fHV ,ATS = heavy vehicle adjustment factor

PAGE 21

21 ATS np ATS o ATS d df v v FFS ATS, ,) ( 00776 ( 2 4 ) where, ATSd = average travel speed in analysis direction (mi/h) FFS = free flow speed (mi/h) vd,ATS = demand flow rate in analysis direction (pc/h) vo,ATS = demand flow rate in opposing direction (pc/h) fnp,ATS = percentage of nopassing zone s adjustment factor ATS is not used to find the LOS of class II highways. PTSF which refers to the amount of time a vehicle spends in a following state because of a slow er moving lead vehicle, is found in a similar way to average travel speed, but uses different adjustment factors. Class III highways do not use PTSF as a performance measure. Equation 2 5 and Equation 2 6 are used to find PTSF ] 1 [ 100) (, b PTSF dv a de BPTSF ( 2 5 ) where, BPTSFd = base percent timespent following in analysis direction ( % ) vd,PTSF = demand flow rate in analysis direction (pc/h) a = constant from HCM Exhibit 15 20 b = constant from HCM Exhibit 15 20 PTSF o PTSF d PTSF d PTSF np d dv v v f BPTSF PTSF, ( 2 6 ) where, PTSFd = percent time spent following ( % ) fnp,PTSF = percentage of nopassing zones adjustment factor

PAGE 22

22 vo,PTSF = demand flow rate in opposing direction (pc/h) PTSF is the percentage of time that each vehicle is in a following state for its total time on the facility. The PTSF for the facility is taken as the average PTSF of all the vehicles on the facility. This performance measure cannot be accurately measured in the field because it is impossible to track every vehicles following state at every moment. Therefore, a simulation tool is required to measure the true PTSF Since PTSF is impossible to measure directly in the field, the HCM considers a vehicle to be following if it has a headway of th ree seconds or less at a chosen point along the highway The new performance measure used in the 2010 HCM for cla ss III highways is percent freeflow speed and can be found using Equation 2 7 FFS ATS PFFSd100 ( 2 7 ) where, PFFS = percent free flow speed ( % ) ATSd = average travel speed in analysis direction (mi/h) FFS = free flow speed (mi/h) ATS, PTSF and PFFS are the performance m easures used to determine the LOS of different classes of highways. Table 21 shows the LOS thresholds for each highway class. The effects of passing lanes on twolane highway performance ar e discussed in a later section. Criticisms of the HCM Methodology The HCM two lane highway methodology has been a topic of much criticism over the years. Researchers have uncovered inconsistencies between the HCM and field data, which make the results of the HCM analysis procedures less reliable. These

PAGE 23

23 problems could be leading to results that do not reflect the true state of the facility being analyzed. The criticisms and limitations of the HCM are discussed in this section. Overestimation of PTSF Luttinen ( 5 ) found that the 2000 HCM method for PTSF overestimated the PT SF for a study done in Finland. Figure 4 in Luttinens ( 6 ) study shows how the 2000 HCM method compares with the results of th e Finnish study and with a negative exponential headway distribution is discus sed in more detail The differences in PTSF between the 2000 HCM and the Finni sh model are 10% or more when the flow rate is between 1500 and 2500 pc/h. Poor Method for Estimati ng BFFS The 2010 HCM method for estimating BFFS is assuming that it is 10 mi/h higher than the posted speed limit if the analyst does not have speed data for a similar facility to use as a guide. The posted speed is used as a rough starting point because FFS is based mainly on the roadway geometry. Grade drastically a ffects traffic flow on highways and the HCM does not address those effects very well ( 7 ) The adjustment factors for grade only give a rough estimation of the geometric conditions. ATS calculations are drastically affected by the BFFS SpeedFlow Relationship The c urrent relationship between speed and flow shown in the HCM is not co nsistent with the findings of a study done in Finland. Luttinen ( 5 ) conducted a field study in Finland and found that the overall shape of the speedflow relationship matched well with a curve shape that initially has a fairly steep negative s lope for low flow rates and then the curve slope decreases to the point where it is fairly flat over the moderate to high flow rates T here was a st eeper decrease i n speed at lower flow rates and the

PAGE 24

24 effect of the opposing flow rate on speed was much small er than the analysis flow rates effect which suggests that the directional split should be accounted for when determining ATS. The HCM speedflow curves show a linear relationship that did not match up as well as the decreasing model to the field data. Figure 7 in Luttinens ( 5 ) study shows the comparison of the speedflow relationship between the Finnish field data and the HCM prediction. Brilon and Weiser ( 8 ) used field data and simulation to estimate a speedflow model. They found that the speedflow relationship follows a curve shape that initially has a fairly steep negative slope for low flow rates and then the curve slope decreases to the point where it is fairly flat over the moderate to high flow rates Equation 2 8 fits the field data of the Bri lon and We iser ( 8 ) study well. q b a car Tv ) ( ( 2 8 ) VT(car) = average passenger car travel speed (mi/h) q = two way traffic volume (veh/h) a = model parameter b = model parameter Deterministic Method for Identifying Follower Status For purposes of measuring PTSF in the field, the HCM uses a constant value of 3 seco nds for the headway threshold. That is, if the headway between successive vehicles is 3 seconds or less at a chosen point along the highway facility the trailing ve hicle is considered to be in a following state. This method does not account for drivers having different desired following headways. A stochastic approach for the

PAGE 25

25 determination of follower status has been proposed by Catabagan and Nakamura ( 9 ) and is discussed in a later section. Other Performance Measures More Applicable for LOS Determination Several researchers such as Luttinen ( 5 6 7 ) Catabagan and Nakamura ( 9 ) AlKaisy and Freedman ( 10) Van As ( 11) Polus and Cohen ( 12) and Morrall and Werner ( 13) have suggested that the HCM two lane highway analysis methodology is a very rough and potentially inaccurate standard for LOS assessment and that PTSF and ATS are not necessarily the most appropriate performance measures to use for assessing two lane highway performance. PTSF has not been used in Germany as a primary twolane highway performance measure because it mainly indicates the drivers discomfort level rather than how efficiently the highway is operating. The main performance measure used to evaluate twolane highway operations should indicate problems with safety, system operations, or the environment ( 8 ) The ATS of passenger cars as opposed to the ATS of all vehicles, has been used in Germany and Finland as the main performance measure for two lane highways because it gives a better indication of how increased volumes affect traffic operations ( 8 14) Trucks speeds are not very sensitive to increases in traffic volume, but traffic volume is the main factor affecting the ATS of passenger cars ( 8 ) Although ATS is easy to measure in the field, it is not very informative about the efficiency of the highway. Since the analysis section of a twolane highway facility is usually several miles long, there could be many changing conditions such as posted speed. This makes ATS somewhat meaningless for determining how the highway is operating ( 14)

PAGE 26

26 PFFS is meant to account for the limitations of ATS. It measures the speed reduction due to increased traffic volume, which makes it possible to compare the current conditions to the ideal conditions ( 14) Volumes in the analysis direction and opposing direction were found to be statistically significant in the PFFS models developed by Al Kaisy and Freedman ( 10) One of the limitations of PFFS is that it is almost completely unaffected by the addition of a passing lane, which indicates that it is not a useful performance measure for capturing the delay caused by platooning ( 10) The 2010 HCM only uses PFFS for determining the LOS of Class III highways, which do not include passing lanes by definition. The PFFS of passenger cars as opposed to the PFFS of all vehicles, has been analyzed as a possible performance measure. Since cars are more sensitive to increased traffic volumes than trucks, the PFFS of passenger cars gives a better indication of how traffic operations are affected by changes in traffic volume ( 14) AlKaisy and Karjala ( 14) found that speedrelated performance measures showed a higher correlation to opposing flow rate than other performance measures T he relationship between ATS and PFFS of pass enger cars and platooning variables did not prove to have a higher correlation than ATS and PFFS PFFS would be sufficient for capturing changes in platooning variables for a two lane highway facility. Other performance measures that have been proposed to replace or supplement the 2010 HCM performance measures are discussed in a later section. Poor Method for Indicating Necessary Improvements PTSF is not a good performance measure for indicating if improvements should be made to a highway that has low volumes with a high percentage of heavy vehicles and few passing sections. In this case, PTSF would be high and suggest that additional

PAGE 27

27 lanes or passing sections need to be added even though the volumes are low Another performance measure, known as follower density has been proposed to account for this shortcoming of PTSF and is discussed in a later section ( 11) Overestimation of Performance Measure Improvements Due to Passing Lanes T he results obtained from the inclusion of a passing lane lead to unrealistic improvement s to the performance measures and, consequently, the level of service. PTSF and ATS are very sensitive to changes in volume and passing lane adjustments result in large reductions to both PTSF and ATS. When solving for a service volume, which is the volume corresponding to a given LOS, based on PTSF and ATS for a facility with a passing lane, the calculated volume is unreasonably higher than the volume for the facility with n o passing lane. These results are misleading because, realistically, t he number of vehicles before and after a passing lane section should be approximately the same. A small percentage of the vehicles may change positions, but it does not drastically change the number of vehicles in the traffic stream This means a passing lane should not show major increases to the traffic carrying capability of the highway section. This issue is demonstrated in Chapter 4.1 Limitations During Saturated Conditions and Near Signalized Intersections PTSF is limited in that it is not useful in saturated conditions This is because as the traffic volume increases, spacing becomes a more im portant factor than time headways and PTSF does not incorporate spacing ( 6 ) 1 Washburn, S. S. Meeting Discussion. University of Flor ida, Gainesville, Florida, May 19, 2010.

PAGE 28

28 The HCM methodology cannot be used to accurately describe conditions for twolane highways with signalized intersections ( 1 ) Yu and Washburn ( 15) have proposed a method to addres s this situation by using a performance measure that describes the LOS of an entire highway facility instead of breaking the highway up into smaller segments Long highway facilities usually have disruptions such as twoway stop controlled intersections, driveways, and signals and the HCM does not have a methodology that describes the overall facility operations. The approach taken by Yu and Washburn ( 15) was to divide the facility into sections where conditions changed. Intersections were regarded as a segment of the highway rather than a point location and the effective influence area upstream and downstream of the signal was determined. Percent delay ( PD) was chosen as the twolane highway facility performance measure because it can be applied to all possible situations encountered on a twolane highway facility. PD is the ratio of delay to freeflowing travel time which gives more information about the operational conditions than total delay alone because it compares unfavorable time spent traveling to favorable time spent traveling. Equation 2 9 gives the formula for PD and Table 22 shows the threshold values for twolane highway facility LOS. 100 ) ( ) (, S H S S H H S H S HFFS L FFS L D D PD ( 2 9 ) where, PD = average percent delay per vehicle through the entire facility (%)

PAGE 29

29 DH = average delay per vehicle for a twolane highway segment (s/veh) DS = average delay per vehicle for a signalized intersection influence area (s/veh) FFSH = FFS for a two lane highway segment (ft/s) FFSS = FFS for a signalized intersection influence area (ft/s) LH = length of a twolane highway segment (ft) LS = length of a signalized intersection influence area (ft) Additional Performance Measures Performance measures should indicate how drivers feel about the traffic level, be useful for congested and uncongested conditions, be compatible with performance measures of other facilities, and be easy to measure in the field and estimate when it is not feasible to gather all field data ( 6 ) Different methods are us ed to determine performance measures for twolane highways and they each give different results. There are several performance measures that are described in the literature, but not included in the 2010 HCM methodology that could be useful for assessing two l ane highway operations. Perc ent Impeded AlKaisy and Freedman ( 10) have proposed the percent impeded ( PI ) measure to determine the operational status of twolane highways Speed and headway are the two main param e ters used to develop twola ne highway performance measures and PI encompasses both of these aspects. PI considers only vehicles that are unable to pass when assessing LOS. Equation 2 10 is used to calculate PI i pP P PI ( 2 10) where,

PAGE 30

30 PI = percent impeded Pp = probability that a vehicle is part of a platoon Pi = probability that a vehicle is impeded Pp accounts for the headway component and Pi accounts for the speed component of this performance measure. Pp can be estimated using a deterministic approach similar to the 2010 HCM PTSF method, which is to collect time headway data and choose a headway threshold to determine if a vehicle is in a platoon or unrestricted. Pi can be estimated by collecting speed dat a for slow moving vehicles and for vehicles that are not part of a platoon or vehicles that lead a platoon. Then the data can be used to find the percentage of vehicles with a higher desired speed than the average speed of slow moving vehicles. Figure 1 in AlKaisy and Freedmans ( 10) study shows Pi based on the desired speed distribution. PI consistently shows the greatest percent change due to the addition of a passing lane compared to percent followers, follower density, and PFFS The AlKaisy and Freedman study ( 10) also showed that volumes in the analysis direction and opposing direction were not statistically significant in PI models and percent follower models. This means an additional performance measure would need to be used, such as the volume to capacity ratio ( v/c ) to report the appropriate LOS. However, PI showed a strong correlation to other performance measures and most platooning variables and was consistent with the expected effects of the passing lane. Also, PI was the only performance measure that found the percentage of trucks in the traffic stream to be significant in the model ( 10)

PAGE 31

31 Probability Based Follower Identification A new method has been developed that can better estimate whether or not a driver is following or freely moving. Current procedures identify followers by a chosen headway, but those procedures do not account for driver variability or changes in roadway conditions. Some drivers may feel unrestricted while driving with a headway of three seconds or less which is what the 2010 HCM uses as a standard for determining follower status while others may have a headway larger than three seconds and f eel like they ar e following. Also, a certain driver may have a di fferent desi red headway based on weather conditions, pavement conditions the time of day and other factors that a ffect drivers comfort levels ( 9 ) Probability based follower identification takes a stoc h as tic approach that incorporates both speed and headway into determining when a driver is following. This procedure uses a mixed distribution model of headways known as the Semi Poisson Model (SPM) in order to separate following and free vehicles based only on the headway portion of this method. As headways increase, the probability that a vehicle is following decreases until a critical headway is reached at which point no vehicles should be considered as followers Equation 2 1 1 shows the proportion of vehicles that are following and free. ) ( ) 1 ( ) ( ) ( t h t g t f ( 2 11) where, f(t) = total observed headway distribution = proportion of constrained vehicles g(t) = constrained headway distribution function

PAGE 32

32 h(t) = u nconstrained headway distribution function Equation 21 1 can be manipulated to give the probability that a vehicle is following as shown in Equation 2 1 2 ) ( ) ( ) ( t f t g Foll Pheadway ( 2 12) where, P(Follheadway) = probability that a vehicle is followi ng given its headway The speedbased portion of this method is hindered by the difficulty in collecting data on the ever changing desired speeds of different drivers T his procedure assigns different desired speeds to certain conditions and if a vehicle s speed drops below the assumed desired speed, then it is considere d to be in the following state. The unified speed distribution method was used to approximate the desir ed speeds. Equation 2 1 3 projects the probability that a vehicle is following based only on desired speeds: iv d idv v f v Foll P ) ( ) | ( ( 2 13) where, vi = travel speed of vehicle i P( F oll | vi) = probability that vehicle i traveling at speed v is following fd(v) = desired speed distribution function Figure 1 in Catabagan and Nakamuras ( 9 ) study g raphically shows the probability that a vehicle is following based on headway and speed. Equation 2 1 4 combines the headway and desired speed models into one formula for finding the probabil ity that a vehicle is following. ) ( ) ( ) | ( v S t v t Foll Pi i i ( 2 14)

PAGE 33

33 where, i = driving condition Pi (Foll | t,v) = probability that a vehicle is following at condition i i(t) = following probability at condition i based on headway Si(v) = following probability at condition i based on speed Catabagan and Nakamura ( 9 ) found that the PTSF measure used by the HCM underestimates the number of following vehicles. As the highway approaches capacity, the percent of followers should increase and theoretically reach 100% and the probability based follower identification method is consistent with that concept This study was done in Japan, the results of which may not necessarily transfer directly to conditions in the United States particularly since Japan does not allow passing on any two lane highways. Drivers in Japan may behave differently because of the nature of their highways or because of unknown characteristics ( 9 ) Fol lower Density Van As ( 11) performed a study in South Africa and found that follower density was a useful performance measure for determining when capacity improvements need to be made. It essentially combines the percentage of followers, traffic flow, and travel speed into one performance measure. Volumes in the analysis direction and opposing direction were found to be statistically significant in the follower density model s ( 10) Follower percentage by itself is limited in that it does not capture the effects of traffic volume. The HCM uses the percentage of followers as a surrogate for estimating PTSF from field data. Follower density is calculated by multiplying the percentage of followers by density. It is best suited as a point measurement. While the measurement of follower percentage to be used in the follower density calculation is consistent with

PAGE 34

34 the surrogate measure for PTSF it should be noted that the true PTSF is intended to reflect continuous measurements of following over both space and time. AlKaisy and Karjala ( 14) found that traffic flow affected twolane highway performance measures mo re than other platooning variables such as opposing flow, percentage of heavy vehicles, percentage of nopassing zones, and speed differentials. Therefore, it is important that the performance measure used to analyze twolane highways captures the effects of traffic volume. One of the advantages of using follower density as a performance measure is that it allows two lane highway LOS to be compared easily with freeways and multilane highways, which use density as a performance measure. Follower density internally incorporates speed, which means ATS would not have to be used as a performance measure ( 7 ) The LOS for f ollower density is based on the worst conditions along the highway unlike the 2010 HCM procedure, which is based on the average conditions. The LOS is based on the segment with the highest follower density. This approach is better than the averageconditions approach because the problem area can be easily identified for improvements. T he 2010 HCM gives the LOS based on the average value of the performance measures per vehicle where follower density gives the LOS based on the value of the performance measure for the traffic stream Th e limitation of using the average per vehicle is that the average LOS may be poor, which would indicate that improvements should be made, even when there are low traffic volumes. The traffic stream approach takes traffic density into account, which gives a better indication of when highway

PAGE 35

35 improvements should be made. Equation 2 1 5 gives the calculation for follower density. Table 23 shows the proposed follower density LOS thresholds. These threshold values require further verification ( 11) Density Followers Density Follower % ( 2 15) One of the methods for finding follower density in the field is to divide the facility into different segments, making sure that each separate segment has the same characteristics such as passing or nopassing allowed Then, t he percentage of followers could be recorded at the beginning and end of each segment and the worst segment could be easily identified. It would also be possible to find the average percentage of followers along each seg ment Follower density can then be easily calculated using Equation 21 5 ( 11) Freedom of Flow Po lus and Cohen ( 12) did a study and found that freedom of flow was an effective tool for determining the LOS of a two lane highway. Freedom of flow is similar to PTSF but it gives a better representation of the drivers opportunities to pass when they are delayed by a slower lead vehicle. The variables that are needed to calculate freedom of flow can be easily estimated from existing traffic data for twolane highways with similar characteristics to the highway being analyzed. Traffic intensity is defined as the ratio of the arrival rate to the service rate. The arrival rate is the interarrival times of vehicles to the back of a platoon and the service rate is the tim e a vehicle has to wait behind an impeding vehicle until it is able to pass. For M/M/1 q ueuing, E[TD] represents the average service rate. F reedom of flow represents the ratio of travel times between platoons to the service rate. This

PAGE 36

36 performance measure c aptures the freedom that a vehicle has to move around in the traffic stream. The variables for this performance measure are all dependent on the opposing flow rate. Since freedom of flow internally incorporates the flow in both directions, it is a two directional analysis tool. F reedom of flow is given by Equation 2 16. Table 2 4 shows the projected LOS thresholds compared to thresholds for other performance measures. / ] [ /O DN T E ( 2 16) where, = freedom of flow hb = average interarrival times of fast vehicles to the back of a moving platoon (s) NO = average number of headways between platoons = traffic intensity = average travel time between platoons (s) = NO hb E[TD] = expected time between the arrival of a fast vehicle into a position behind a slo w vehicle and the time when the passing maneuver starts (s) = hb Overtakings Morrall and Werner ( 13) noticed that t he perfor mance measures used by the HCM to determine LOS do not externally show the effects of passing lanes. They proposed that an additional performance measure, called overtakings, could be used to account for this weaknes s in the HCM. The idea of measuring LOS by overtakings is based on how important passing opportunit i es are to d rivers. It takes into account the effects of passing lanes and the percentage of nopassing zones. This is a useful tool for identifying highways that need improvements. Equation 2 17 is used to determine the overtaking ratio.

PAGE 37

37 DO AO s overtaking ( 2 17) where, AO = achieved overtakings DO = desired overtakings assuming a continuous passing lane section Although a certain twolane highway may not have a continuous passing lane section, d esired overtakings are the number of overtakings the driver would make for a highway that has a continuous passing lane section and geometry similar to the highway in question. Overtakings measures how many passes a vehicle makes on a highway versus how many passes a vehicle would make if it was unrestricted. One of the limitation s of this performance measure is that it d oes not give any information about speed or delay, which is very important for determining two lane highway LOS Therefore, overtakings cannot be used as the primary measure for LOS Also, it is unclear how the overtaking ratio corresponds to how drivers f eel about the traffic conditions, which makes it difficult to assign specific LOS thresholds to overtaking ratios. Morrall and Werner ( 13) proposed that several graphs should be developed for a variety of road, traffic, and vehicle characteristics in order to find the overtaking ratio without having to do simulation for every highway that needs to be analyzed. Passing Lanes Passing lanes are a unique feature of twolane highways that provide a cost effective alternative to decreasing delay by helping platoons dissipate and providing passing opportunities when the gaps are inadequate due to high oncoming traffic volumes or when there is insufficient passing sight distance ( PSD ) They also decrease delay at locations where bottlenecks are formed due to slowmoving vehicles on

PAGE 38

38 upgrades. However, passing lanes further complicate the effort to accurately analyze a two lane highway because several more variables need to be accounted for and different roadway conditions can change how effective a passing lane is in decreasing congestion. Typical passing lane lengths are usually between 1000 feet and 3 miles ( 16) HCM and Passing Lanes According to the 2010 HCM the lengths upstream and downstream of the passing lane have a large impact on performance measures The effects of passing lanes on PTSF and ATS are shown in Exhibits 1525 and 1527 in the 2010 HCM where LU is the length upstream of the passing lane, Lpl is the length of the passing lane, Lde is the effective length downstream of the passing lane, and Ld is the noneffective length downstream of the passing lane ( 1 ) Passing Lane Location Passing lanes are often added at locations where bottlenecks are formed along upgrades due to slow moving vehicles or where general highway operations are inadequate The optimal passing lane length should be determined by using a base percent time delay. The effective length of a passing lane depends on traffic flow, vehicle composition, passing opportunities downstream of the passing lane, and the length of the passing lane itself. The passing lane is most effective if it is added before speeds decrease below tolerable levels ( 16) Passing lanes on upgrades, also known as climbing lanes, can also be effective if one short passing segment is placed near the middle of the upgrade or if two passing lane segments are placed at locations one third and two thirds of the way along the len gth of the upgrade ( 17) When a highway is operating poorly in areas besides

PAGE 39

39 upgrade segments, passing lanes should be added periodically to help improve the overall performance ( 16 ) Effects on Other Performance Measures The operational improvements due to the addition of a passing lane depend on the percentage of vehicles traveling in platoons traffic flow, and passing lane leng th. Passing lanes have a large positive impact on decreasing the PTSF and the percentage of vehicles delayed in platoons, but they do not have a large impact on the ATS ( 16) Passing lanes noticeabl y affect the segment far downstream from the actual passing lane location as shown in Exhibit 15 22 of the HCM Passing Procedure Polus et al. ( 18) did a study on quantifying values for passing maneuvers and concluded that the American Association of State Highway and Transportation Officials (AASHTO) may not be an accurate standard for determining the distances for passing components when a car is passing a truck. A car passing a truck stays in the opposing lane for a longer amount of time than a car passing another car and when the car retur ns to the original lane, there is a longer time gap between the car and the truck than between the car and another car. Polus et al. ( 18) found that the reaction time to start a passing maneuver once a gap is available is not dependent on speed. The time between the return of the passing vehicle to the original lane and the next oncoming vehicle is also not dependent on speed. Polus et al. ( 18 ) also found that the speed differential between overtaking and overtaken vehicles decreases as their travel speeds increase. AASHTO ( 19) assumes that there is no acceleration while the passing vehicle is in the oncoming lane of traffic. However, the results of Polus et al. ( 18) show that the acceleration happens mainly in

PAGE 40

40 the opposing lane. AASHTO ( 19) assumes that a vehicles acceleration is 2.1 ft/s2 before it begins to move into the opposing lane for a passing maneuver but Polus et al. ( 18) found that the average acceleration in the oncoming lane during a pass is about 3.0 ft/s2. Driver Decision to Pass There are several factors that motivate a driver to attempt a pass. Platoons caused by slow moving lead vehicles such as trucks instigate drivers to pass and the number of lane changes varies as the length of the platoon changes. As volume increases, the drivers desire to pass also increases because it is more likely that a slowmoving vehicle c ould be in the traffic stream ( 20) Passing S ight D ista nce PSD is an important parameter for deciding if drivers will attempt to pass. Several factors such as passing, impeding, and oncoming vehicle speeds, acceleration and deceleration, and vehicle lengths have an impact on the PSD required for a driver to make a safe and successful pass. These factors change depending on roadway conditions, driver type, and vehicle type. However, current design standards do not take their variability into account. Instead, they use a conservative, constant value, which means different passing scenarios cannot be modeled accurately ( 2 1 ) AASHTO ( 19) provides standard PSD guidelines for different parts of a pass and the total required PSD is calculated based on these component parts. Table 25 shows values for distances, speeds, times, and accelerations for each component of a passing maneuver according to AASHTO. Exhibit 3 4 in the AASHTO green book ( 19) illustra tes the distance values for each of these components of a passing maneuver. Descriptions of the variables that are used in Table 25 are as follows:

PAGE 41

41 d1 = initial mane uver distance traveled from start of passing maneuver until passing vehicle encroaches upon oncoming lane (ft) t1 = time required for initial maneuver (s) a = acceleration of passing vehicle when initiating passing maneuver ( ft / s /s) v = average speed of passing vehicle ( m i/h) m = difference in speed of passed vehicle and passing vehicle ( m i/h) d2 = distance traveled by passing vehicle from the point of encroachment in the oncoming lane to the point of return to the original lane (ft) t2 = t ime spent traveling in the oncoming lane (s) d3 = shortest desirable distance between the front bumpers of the passing and opposing vehicles when the passing vehicle returns to the original lane (ft) d4 = distance traveled by the opposing vehicle during the time that the passing vehicle travels from the position of being directly abreast of the vehicle being passed to the return to the original lane (ft) Glennon ( 2 2 ) developed a PSD equation that was inspired by the concept of a critical point where the sight distance requirements for a vehicle to complete or abort a pass are equal. AASHTO assumed that the passing vehicle would not have a chance to abort, which is an unreali stic assu mption. Glennons ( 2 2 ) critical sight distance equation assumes that the opposing vehicle and the passing vehicle are traveling at the design speed and that the impeding vehicle is traveling at some speed bel ow the design speed. Glennons ( 2 2 ) critical sig ht distance model is given in Equation 218 m v Sc c16 2 2 ( 2 18) where, Sc = critical sight distance (ft) v = design speed (ft/sec)

PAGE 42

42 c = critical separation between passing and impeding vehicles (ft) m = difference in speed between passing and impeding vehicles (ft/sec) El Khoury and Hobeika ( 2 1 ) did a study on the risk level associated with different PSD that helped reinforce G lennons ( 2 2 ) findings. They found that AASHTO gives a conservative design standard, while the Manual on Uniform Traffic Control Devices ( MUTCD) and Glennons ( 2 2 ) equation give reasonable estimates for the necessary PSD. AASHTO has a more conservative calculation because it measures PSD from the time the passing maneuver begins to the time it ends w hereas the MUTCD and Glennons ( 2 2 ) equation measure PSD from the time the vehicle is in the critical position to the time it completes the pass. Figure 3 in El Khoury and H obiekas ( 21) study shows the relationship between the MUTCD design standard and Glennons ( 2 2 ) equation in metric units. Glennon ( 2 2 ) also found that as vehicle length increases, PSD requirements increase, but not by a severe amount. Vehicle length may not have a drastic impact on PSD because it becomes important only for endzone passes and also because drivers are able to adapt to the conditions under which they are passing and adjust for inhibiting factors such as longer impeding vehicle lengths ( 2 2 ) Hassan et al. ( 2 3 ) built on Glennons model, incorporating a level of safety for the passing maneuvers that increases as design speed increases. The PSD model is given in Equation 2 19. ) ( 2 h t V PSDd c ( 2 19) where, PSDc = PSD required for a vehicle at the critical position to complete or abort the pass (ft)

PAGE 43

43 Vd = design speed (ft/s) t = time required for a passing vehicle to return to the original lane from the critical position for a completed pass (sec) h = minimum headway between passing and passed vehicles at the end of a completed or aborted pass and minimum headway between passing and oncoming vehicle at the end of a completed or aborted pass (sec) This model determines the critical distance based on the positions of the front bumpers of the passing and impeding vehicles. Equation 2 2 0 gives the calculation for the critical positi on. mt m V h Ld p c47 1 ) ( ( 2 20) where, c = relative positions of the front bumpers of the passing and passed vehicles at the critical position with a negative value meaning the passing vehicle is behind the vehicle being passed (ft) Lp = length of passing vehicle (ft) Vd = design speed (ft/s) m = speed differential between passing and passed vehicle (ft/s) Once the front bumper of the passing vehicle is past the front bumper of the impeding vehicle, it is not likely that the passing vehicle would abort, even if the critical position was not yet reached. To account for this, Hassan et al. ( 2 3 ) modified t in Equation 2 19 so that it provides adequate sight distance for any passing vehicle to complete the pass as long as its front bumper is side by side with the impeding vehicles front bumper. The new definition of t is the time required to complete a pass when the front bumpers of the passing and passed vehicles are side by side. Figure 16 in Hassan et al. ( 23) shows the comparison betw een the PSD models according to Hassan, Glennon, and the MUTCD.

PAGE 44

44 Capacity The 201 0 HCM states that directional capacity is 1700 pc/h and twoway capacity cannot exceed 3200 pc/h ( 1 ) Capacity is computed by adjusting the base flow rate of 1700 pc/h for grade and heavy vehicles. Equation 2 21 and Equation 2 22 are used to calculate capacity in veh/h based on ATS and PTSF respectively. ATS HV ATS g ATS df f c, ,1700 ( 2 21) PTSF HV PTSF g PTSF df f c, ,1700 ( 2 22) Class I capacity is based on the lowest of cd,ATS and cd,PTSF Class II capacity is based only on cd,PTSF, and Class III capacity is based only on cd,ATS ( 1 ) Luttinen ( 5 ) did a study in Finland and found that capacity was reached at higher speeds and lower densities than the 2000 HCM method projected, but the 2000 HCM capacity estimate matched the field data capacity estimate well. Differences could be due to conditions that were not total ly congested and factors that were not ideal. The field data showed that capacity was reached at flow rates of 1550 veh/h and 1600 veh/h, which is close to the 2000 HCM prediction of 1700 pc /h. The 2000 HCM estimates the capacity to occur when the density is about 32 pc/h while the field study showed that capacity was reached when the density was about 29 veh/mi. T he results of the Luttinen ( 5 ) study suggest a linear relationship between opposing flow rate and directional capacity. Rozic ( 2 4 ) did a study showing that capacity depends on how the traffic flow is divided and how it is measured. The ideal twoway capacity was found by dividing traffic flow from a carfollowing aspect Ideal capacity assumes the entire traffic stream is made up of passenger cars only that are in one long platoon with 1.8 second headways

PAGE 45

45 between each successive vehicle, which is approximately the smallest realistic headway. The ideal twolane twoway highway capacity was found to be 4000 pc/h, which occurred when the traffic flow was made up of two platoons that consisted of passenger cars only. The realistic capacity accounts for trucks in the traffic stream and vehicles breaking up into several smaller platoons instead of staying in one long platoon. The realistic two lane twoway highway capacity was estimated to be about 2700 veh/h. Kim ( 2 5 ) created a simulation program called TWOSIM, which was specifically designed to estimate twolane highway capacity under different conditions. The first program, known as TWOSIM I, was used to estimate t he capacity of a 2 mile two lane highway with all passenger cars, 100% commuters, level terrain with no curves, low opposing flow rate, and 100% nopassing zones. For this two lane highway Kim ( 2 5 ) found the directional capacity to be about 1850 pc/h for and average FFS of 40 mi/h, 2000 pc/h for an average FFS of 50 mi/h, and 2100 pc/h for an average FFS between 60 and 70 mi/h. TWOSIM II was designed to analyze a twolane highway with the same base conditions as the highway analyzed in TWOSIM I but w ith a 20% nopassing zone. There were two nopassing zones of 0.2 mi at each end. TWOSIM II estimated capacity for this two lane highway configuration to be 2100 pc/h, which is the same as the TWOSIM I capacity estimate for high FFS This means that passing zones did not affect capacity. TWOSIM III was designed to estimate capacity for twolane highways with horizontal curves (radius less than 500 ft), driveways, upgrades, and trucks in the traffic

PAGE 46

46 stream. With no trucks in the traffi c stream, TWOSIM III estimated capacity reductions to be between 10% and 12% for twolane highways with a driveway, between 3% and 17% for twolane highways with a horizontal curve, and between 11% and 30% for twolane highways with an upgrade between 4% and 8%. For an increasing proportion of trucks in the traffic stream (10% 20%), TWOSIM III estimated capacity reductions to be between 10% and 23% for twolane highways with a driveway, between 3% and 26% for twolane highways with a horizontal curve, and between 11% and 40% for twolane highways with an upgrade between 4% and 8%. TWOSIM III showed that capacity increased with an increasing radius of horizontal curvature when there was a constant percentage of trucks in the traffic stream. TWOSIM III also s howed that, for 10% trucks in the traffic stream, capacity decreased as the length of the upgrade increased. Kim ( 2 5 ) found that the capacity for both directions c ould be calculated as twice the capacity of the analysis direction since the directional capaci ty was not sensitive to the opposing flow rate, which is in conflict with Luttinens ( 5 ) findings However, since both directions rarely, if ever, reach capacity at the same time, capacity should only be reported for the analysis direction. Brilon and W eiser ( 8 ) used a German simulation model, known as LASI, to estimate two lane highway capacity. The twoway capacity was estimated to be 2500 veh/h for straight two lane highways with level terrain. The directional capacity was estimated to be between 1200 veh/h and 1450 veh/h. When a twolane highway is at capacity, there are still gaps in the traffic stream, but they cannot be filled because of speed differences and few passing opportunities.

PAGE 47

47 Simulation Tools There are few simulation tools that have been developed specifically for analyz ing two lane highways TWOSIM, which was designed to estimate twolane highway capacity, is described in the capacity section of this chapter. The main software programs that were created to simulate twolane highways and to evaluate their performance are discussed in this section. TWOPAS TWOPAS is a microscopic, stochastic traffic flow simulation model that was originally created by the Midwest Research Institute ( 2 6 ) This model incorporates formulas for roadway geom etry, passing zones, driver type, vehicle type, and the direction of traffic. It uses thirteen different types of vehicles including five passenger cars, four RVs and four trucks. This program was written in the FORTRAN language and the user interface, which is known as UCBRURAL, was developed at the Universit y of California Berkeley (UCB) ( 2 7 ) The 1985 HCM was developed largely based on adjustment factors, calculations and outputs from TWOPAS ( 2 6 ) The 2000 HCM twolane highway methodology is still very similar to the 1985 HCM methodology and reflects refinements largely based on TWOPAS simulation results. Allen et al. ( 2 8 ) suggest ed that TWOPAS should include more vehicle types or that truck characteristics should have more variability to allow for bett er modeling of truck speeds on steep upgrades and downgrades. TWOPAS is limited in that it cannot i nternally evaluate downgrades for different truck types to determine which ones are at crawl speed ( 2 8 )

PAGE 48

48 TRARR The Australian Road Research Board developed a microscopic, stochastic simulation program known as TRARR to analyze twolane twoway highways. TWOPAS is recommended for use in the United States because many of the parameters in TRARR such as driver and vehicle characteristics are based on Australian conditions which causes the outputs to be considerably different from field data in the United States ( 2 6 ) CORSIM CORSIM is a microscopic, stochastic simulation program that is maintained by Mc Trans at the University of Florida ( 3 ) It uses FRESIM links to model freeways, and NETSIM links to model urban streets FRESIM, NETSIM or both models are used to simulate traffic on freeways, interchanges roundabouts, arterials, and other roadway facilities. CORSIM can also model incidents, preemption, and left sided driving Washburn and Li ( 2 ) developed the logic and algorithms for twolane highway modeling within CORSIM. Mc Trans implemented the logic and algorithms in the CORSIM code. This two lane highway modeling capability is available in TSIS 6.2. It is capable of handling passing in one or both directions and can model a twolane highway with a passing lane. T hese tasks were accomplished using FRESIM links. Washburn and Li ( 2 ) made it possible to simulate signalized intersections along twolane highways based on recommendations from the Yu and Washburn ( 15) study. This was done by connecting FRESIM and NETSIM links. Several n ew inputs and outputs were added in CORSIM to allow the user to change different factors specific to affecting the performance of twolane highways. The new inputs include passing zones, follower headway, minimum and maximum clearance

PAGE 49

49 distance between the passing vehicle and passed vehicle, impatience value, passing acceleration, and several others. PTSF ATS, f ollower d ensity and several passingrelated measures were added to the m easures of e ffectiveness (MOE) screen for the user to select as reported si mulation outputs. Most of these new outputs require further testing and validation. The twolane highway car following model used in CORSIM is the Pitt car following model and the default PSD models used are based on AASHTO values, but can be changed to PSD models based on the MUTCD values. Passing maneuvers are restrained by the distance needed to travel in the oncoming lane, the PSD, and the required clearance. CORSIM uses many different checks and equations to calculate those restraints and g enerate su ccessful and aborted passes in a stochastic manner For vehicle s passing in a section that has a passing lane, CORSIM considers a vehicles willingness to move over (WTMO) to the nonpassing lane. The WTMO is then adjusted by the driver type if the vehicl es length is less than 40 feet, which results in trucks longer than a single unit having a higher WTMO. There are also lanechanging models for determining when a vehicle wil l return to the original lane. Washburn and Li ( 2 ) tested many different combinations of inputs for both nopassing allowed and 100% passing allowed scenarios and compared the ATS, PTSF and follower density for three different volume splits, 50/50, 60/40, and 70/30, in the analysis and opposing direction. Scenarios with passing allowed had a positive impact on the major direction, which is the direction with the highest volume, for ATS and PTSF but had a negative effect on the minor direction as a result of fewer passing opportunities than the major direction. Scenarios with passing lanes added show ed

PAGE 50

50 improvement s for ATS and PTSF The results for follower density are slightly lower for 100% passing allowed compared to 100% nopassing allowed.

PAGE 51

51 Table 21 Level of service thresholds as a function of highway class Class I highways Class II highways Class III highways LOS ATS (mi/h) PTSF (%) PTSF (%) PFFS (%) A > 55 > 91.7 B > 50 55 > 35 50 > 40 55 > 83.3 91.7 C > 45 50 > 50 65 > 55 70 > 75.0 83.3 D > 40 45 > 65 80 > 70 85 > 66.7 75.0 E 40 > 80 > 85 Adapted from Highway Capacity Manual. TRB, National Research Council, Washington, D.C., 2010, (Page 157, Exhibit 153) Table 22 Threshold values for twolane highway facility LOS LOS Percent delay (%) A 7.5 B > 7.5 15.0 C > 15.0 25.0 D > 25.0 35.0 E > 35.0 45.0 F > 45.0 Adapted from Yu, Q ., and S S. Washburn. Operational Performance Assessment for Two Lane Highway Facilities. Journal of Transportation Engineering, ASCE, Vol. 135, No. 4, April 2009, pp. 197205, ( Page 202, Table 1). Table 23 LOS thresholds for follower density LOS Typical follower density Range of follower densities A 1.0 0.3 1.4 B 2.0 1.3 3.3 C 4.0 3.0 6.7 D 8.0 6.3 9.5 Adapted from Van As, C. The Development of an Analysis Method for the Determination of Level of Service of TwoLane Undivided Highways in South Africa. Project Summary, South African National Roads Agency Limited Pretoria, 2003, ( Page 18) Table 24 LOS thresholds for freedom of flow compared to PTSF and twoway flow LOS PTSF (%) Two way flow (pc/h) Freedom of flow A 0 15 0 300 B 15 30 300 700 7.1 16.5 C 30 45 700 1200 4.1 7.1 D 45 60 1200 1800 2.8 4.1 E 60 75 1800 2700 1.8 2.8 F 75 100 2700 1.8 Adapted from Polus, A., and M Cohen. Theoretical and Empirical Relationships for the Quality of Flow and for a New Level of Service on TwoLane Highways. Journal of Transportation Engineering, ASCE, Vol. 135, No. 6, June 2009, pp. 380385, (Page 384, Table 2).

PAGE 52

52 Table 25 AASHTO components and values of passing sight distance Component of passing maneuver Speed range (mi/h) 30 40 40 50 50 60 60 70 Average passing speed (mi/h) 34.9 43.8 52.6 62.0 Initial Maneuver a 1.40 1.43 1.47 1.50 t 1 3.6 4.0 4.3 4.5 d1 2 467 11 1at m v t Occupation of left lane t 2 9.3 10.0 10.7 11.3 d2 2467 1 vt Clearance length d 3 100 180 250 300 Opposing vehicle d4 2667 0 d Minimum PSD 4 3 2 1d d d d Adapted from A Policy on Geometric Design of Highways and Streets AASHTO, Washington, D.C., 2004, (Pages 120122, Exhibit 3 5, Equations 36 and 37 ).

PAGE 53

53 CHAPTER 3 RESEARCH APPROACH This chapter describes the research approach used to accomplish the objectives of this study. It describes the preliminary testing procedure and the process for comparing CORSIM outputs to the 2010 HCM outputs and gives detailed information about the files used in the testing procedure. This chapter also describes the criteria used to evaluate the usefulness of different performance measures for the assessment of two lane highway LOS It also discusses the procedure used for estimating capacity using CORSIM. CORSIM and 2010 HCM Comparison In order to compare the results between the 2010 HCM and CORSIM, it is important to have a comparable set of inputs Table 31 shows an example of common inputs for the HCM and CORSIM and also shows which inputs are not applicable for each method. The inputs in Table 31 must have the same values where applicable in CORSIM and the HCM in order to ensure that the conditions are as similar as possible for each of the corresponding HCM and CORSIM testing scenarios There are several factors that are inputs in CORSIM, but not in the HCM, that could cause majors differences in the results. These factors include existing upstream conditions truck type distribution, and passing zone configuration. A facility with a changing grade configuration could also have major effects on the results as the HCM does not have an input for specifying where the grade changes. However, all testing scenarios in this chapter have a constant grade along the entire facility, which eliminates uncertainties with changing grades. Several preliminary tests were run to see

PAGE 54

54 which CORSIM inputs for these factors should be used to give similar results to the HCM. These testing procedures are described in the next section. Preliminary Experimental Designs There are several inputs required for the HCM that are not applicable in CORSIM and conversely as shown in Table 31 The inputs that are not appl icable in CORSIM are programmed internally in the algorithms There are certain factor s that have an impact on CORSIM results that are not mentioned in the HCM such as existing conditions upstream of the facility truck type distribution, and passing zone configurations. Before any major compari son between CORSIM and the HCM was made, preliminary tests for existing conditions upstream of the facility truck type distribution, and passing lane configuration were performed. Also, a preliminary test was done that compared the speed vs. flow rate relationship and PTSF vs. flow rate b etween CORSIM and the HCM. The results of these tests were used as guidance for matching the CORSIM inputs as closely as possible to the HCM inputs Existing upstream conditions When analyzing a facility in the HCM, there are preexisting conditions upstream of the facility such as platoon structure on some unknown length of roadway. The HCM was based on TWOPAS simulation where one of the inputs was percentage of traffic flow entering as platoons. This established the incoming traffic conditions for the analysis facility. In CORSIM, vehicles are generated from an entry node and no upstream conditions have been established. This has a large impact on the entering platoon structure, especially for short facilities with low traffic volumes because the vehicles are spaced out when they are generated and can possibly move through the entire facility before coming near other vehicles This could lead to an unrealistically low

PAGE 55

55 PTSF In order t o have accurate results, there needs to be a lead up length to the facility that allows platoons to form and in cite s vehicles to interact prior to the section of roadway for which data will be collected. There should also be a follow up length that is identical in distance to the lead up length in an effort to have similar traffic conditions coming from both the eastbound and westbound directions. The results of the analysis for any facility should then only be extracted from the links that make up the analysis facility and not from the lead up or follow up segments Three different facility lengths were tested with varying lead up lengths in order to determine what the lead up length should be and how the performance measures were affected. The facilities have lengths of three, five, and ten miles. Each facility was tested with different lead up lengths ranging from one to six miles. The three facilities were tested with passing allowed and passing not allowed. Table 32 shows the general input values used in CORSIM and a sample schematic for the threemile long facility is shown in Figure 31 Truck type The HCM has an input for the percentage of trucks, but does not allow the analyst to specify what types of trucks make up that percentage. CORSIM has an input for the percentage of trucks and also allows the analyst to specify which types of trucks make up that percentage. The truck types range from 3 to 6. Type 3 indicates single unit trucks that are 35 ft long. T ype 4 trucks have a medium sized load and are 53 ft in length. T ype 5 trucks are fully loaded with a length of 53 ft, and type 6 trucks are 64 ft long doublebottom trailers ( 3 ) Trucks in the traffic stream have a large impact on ATS. Several different truck type distributions were tested in CORSIM in order to see how the performance

PAGE 56

56 measures are affected. Table 33 shows the general input values used for this test and Table 34 shows the truck type distributions that were tested. Figure 32 shows the nopassing zone configuration for this test. These tests were each performed with a lead up length based on the results of the lead up length test Also, there was a follow up length with a length based on the results of the upstream segment length test The results of this truck type test were only extracted from the links that mad e up the analysis facility and not from the lead up or follow up segments. Passing zone configuration The HCM allows the user to specify that the facility has some percentage of nopassing zones. However, CORSIM specifies passing zones on a link by link basis. In CORSIM, the user can select exactly where the nopassing zones are located. The HCM does not allow the user to select where the nopassing zones are along the facility. Therefore, even when both methods have the same percentage of nopassing zones, the configuration is ambiguous. This problem could potentially lead to differences in the performance measure results Three different nopassing zone configurations were tested in CORSIM for a 10 mile long facility with 50% nopassing zones as shown in Table 35 These configurations were all tested against a variety of traffic flow rates ranging from 100 to 4000 veh/h each with a 50/50 directional split, in order to see how the performance measures we re affected. Table 3 6 shows the general inputs for this test. Speed vs. flow rate relationship and PTSF vs. flow rate There have been claims ( 5 8 ) that the HCM speed vs. flow rate relationship does not accurately reflect what is happening in the field. Also, Luttinens ( 6 ) study show s that the HCM overestimates PTSF Before doing the major comparison between the HCM

PAGE 57

57 and CORSIM, the speed vs. flow ra te and PTSF vs. flow rate relationships between the HCM and CORSIM were tested under the conditions shown in Table 37 for both passing allowed and passing not allowed. The speed vs. flow rate and PTSF vs. flow rate r elationships for a 50% nopassing zone were tested in the passing zone configuration test. CORSIM/HCM Experimental Design The procedure that was chosen for comparing the 2010 HCM to CORSIM was to use the same inputs for both tool s, to the extent possible, and compare the outputs. A wide range of inputs was used to analyze how these two methods compared to each other for a variety of situations. A variety of combinations was used to see if ther e was a certain scenario that showed dra stically different results from the other scenarios. This experiment involves variations of the directional volumes, heavy vehicle percentages, grades, nopassing zone percentages, and the presence or absence of a passing lane. The input values that were selected for the comparison are described in this section. Flow rates and splits Six d ifferent two way flow rates were used in this experiment in order to capture possible differences with low, medium, and high volumes. The values us ed were 200, 700, 1200, 1700, 2200, and 3200 veh/h. These six volumes were each analyzed under three different splits, 50/50, 60/40, and 7 0/ 3 0. The base conditions we re 0% heavy vehicles, 0% nopassing zones, and 0% grade. Percent heavy vehicles T wo different percentages of heavy vehicles were used in this experiment in order to analyze the changes in performance measures due to heavy vehicles and to analyze the effects of heavy vehicles on attempted passes The heavy vehicl e percentages that

PAGE 58

58 were used were 0% and 10% These values were chosen because the performance measures are expected to show a noticeable change for an increase from no trucks to 10% trucks in the traffic stream. All heavy vehicles are assumed to be trucks and not recreational vehicles. The base conditions were modified to incorporate the increased truck percentage in the traffic stream. The truck type distribution is based on the results of the preliminary truck type distribution test. Grades T wo different grades were used in this experiment because upgrades can have a large impact on truck speeds and this variation in grade is expected to have a large effect on the number of platoons, attempted passes, and following vehicles The values that were chosen for grade were 0 % and 6% The slope of 6% was added as a test variable by building on the previous cases, which included the base conditions and the cases that incorporated the percentage of heavy vehicles. Percent nopassing zones The percentage of nopassing zones is another variable that affects twolane highway performance measures. T he 2010 HCM and CORSIM have different methods for inputting the percent age of nopassing zones. The differences between these two tools make it impossible to ensure that the twolane highway analyzed in the 2010 HCM is the same as the twolane highway created in CORSIM In the 2010 HCM, the percentage of nopassing zones is accounted for in the percent nopassing zone adjustment factor, fnp, which is different for ATS and PTSF calculations However, the 2010 HCM does not specify which section of the highway is a nopassing zone. It only specifies that, somewhere along the highway, a certain p ercentage is a nopassing zone.

PAGE 59

59 CORSIM does not include a direct input for the percentage of nopassing zones. Instead, the u ser specifies what the center line striping condition is for each link (passing allowed in onedirection, passing allowed in both directions, or passing not allowed in either direction). For example, to specify a 50% nopassing zone on a tenmile long highway segment with ten onemile links, the user could select no passing allowed on the first five links and passing allowed on the last five links. The user could choo se any five links to allow no passing and it would still be a 50 % nopassing zone. The advantage in CORSIM is that the user can chose the passing zone configuration or wh ere the passing zone section is located along the highway. The preliminary test for passing configuration was also used to c ompare the two tools sensitivity to this factor. The other passing zone scenarios that were tested in this experiment besides 0%, were 50% and 100% nopassing zones. These changes were made in the existing files of the previously tested cases, which all consisted of 0% nopassing zones. In CORSIM, the seg ments that were changed from 0% nopassing zones were the ones closest to the eastbound entry node. For example, the 5 0% nopassing zone case would have five miles of nopassing allowed closest to the eastbound entry node and would continue with five miles of passing allowed Links with passing allowed in one direction were not tested in this study. Passing l anes This experiment also tested the effects of passing lanes on twolane highway performance measures All the s cenarios described previously had no passing lanes Changes were made to those scenarios to incorporate a passing lane with a length of 5280 feet

PAGE 60

60 There are 432 different trial s based on all of the different combinations of four volumes, three directional splits, three percentages of heavy vehicles, three percentages of nopassing zones, two grades, and two passing lane length scenarios. Table 38 shows a summary of all the variabl e combinations being used for this study. The 2010 HCM methodology was programmed into a numerical calculations worksheet and every scenario was analyzed. The numerical calculations work sheet allowed quick and easy changes to be made to the inputs for each testing scenario. CORSIM Testing Facility The CORSIM two lane highway testing facility used in this experiment is ten miles long. The peak hour demand volume for the analysis direction is generated from node 8100 and travels eastbound. The opposing demand volume is generated from node 8200 and travels westbound. The passing lane scenarios have one passing lane that is one mile in length. The passing lane runs along the eastbound direction of the highway and is the fifth link away from node 8100. There are four onemile upstream links and five onemile downstream links. Figure 33 shows the CORSIM two lane highway schematic. Ten runs were executed in CORSIM for each testing scenario and the average PTSF and ATS for the ten runs was recorded. Capacity CORSIM was used to estimate the capacity of twol ane highways, for combinations of the following variables : FFS of 55, 60 and 65 mi/h Heavy vehicle percentages of 0% and 10% Splits of 50/50, 60/40, and 70/30 No passing lanes passing lanes

PAGE 61

61 The general approach for each of these conditions wa s to specify an input volume that clearly exceeded capacity T hen, the throughput was measured at the end of the facility. The highest average throughput from ten runs in CORSIM that was measured wa s taken as the facility capacity. All test scenarios had 0% nopassing zones and the minimum entry headway was specified as 1.6 seconds. The twoway capacity may not necessarily be equal to twice the directional capacity. For twoway capacity, there are few passing opportunities if any, because of the high traffic volumes in both directions. For directional capacity, the volume in the opposing direction could be close to zero, which means there will be plenty of opportunities for vehicles to pass and reconfigure themselves into new platoons as long as there are appropriate gaps for the passing vehicles to move back in to in the original lane. With a low volume in the opposing direction, oneway capacity could be higher than half of the twoway capacity because impeded vehicles can potentially join faster platoons, which increases the throughput. Performance Measures Another task of this study was to determine if other performance measures as proposed in the literature, could reasonably be used to evaluate the LOS of twolane highways. A co nceptual evaluation of different performance measures that have been proposed in the literature was conducted. The performance measures included in the qualitative analysis were PTSF (with a deterministic and probability based method for follower identific ation) ATS, PI follower density, freedom of flow, and overtakings. The procedure for this qualitative analysis was to compare advantages and disadvantages for each performance measure as discussed in the literature. T he ease of calculation

PAGE 62

62 and field meas urement for each performance measure were ranked along with other categories The highest ranked performance measure was analyzed quantitatively.

PAGE 63

63 1-6 mi 3 mi facility upstream segment Figure 31 Schematic of CORSIM lead up length test for three mile facility 5% 5% 5% 5% 5% 1 mi 1 mi 1 mi 1 mi 1 mi Figure 32 Schematic of CORSIM truck type distribution test facility Figure 33 Two lane highway facility in CORSIM with passing lane

PAGE 64

64 T able 31 Example inputs for HCM and CORSIM Inputs HCM CORSIM Geometric d ata facility length (mi) 10 10 lane width (ft) 12 N/A shoulder width (ft) 6 N/A access point density (point s/ mi) 0 N/A grade (%) 0 0 radius of curvature (ft) N/A 0 superelevation (%) N/A 0 p ercent age of no passing zones (%) 0 N/A passing lane length (ft) (if applicable) 5280 5280 highway class 1 N/A Demand d ata length of analysis period (h) 1 1 PHF 1 N/A base FFS (mi/h) 60 N/A FFS (mi/h) 65 65 heavy vehicle percent age (%) 0 0 directional split 50/50 N/A two way flow rate ( veh/h ) 1200 N/A e astbound flow rate (veh/h) N/A 600 we stbound flow rate (veh/h) N/A 600 Adapted from Transportation Research Board (TRB). Highway Capacity Manual. TRB, National Research Council, Washington D.C. 2010. Table 32 General CORSIM inputs for lead up length tests Inputs Values g rade (%) 0 radius of curvature (ft) 0 FFS (mi/h) 60 heavy vehicle percentage (%) 0 eastbound flow rate(veh/h) 600 westbound flow rate(veh/h) 600

PAGE 65

65 Table 33 General CORSIM inputs for truck type distribution tests Inputs Values Values Geometric data facility length (mi) 10 10 grade (%) 0 0 number of links with passing allowed (mi) 0 10 p ercentage of no passing zones (%) 100 0 Demand data p ercentage of passenger car s (%) 90 90 p ercentage of trucks (%) 10 10 FFS (mi/h) 65 65 Table 34 Scenarios for CORSIM truck type distribution test Test % T ype 3 % T ype 4 % T ype 5 % T ype 6 1 50 50 0 0 2 50 0 50 0 3 50 0 0 50 4 25 25 25 25 5 100 0 0 0 6 0 0 0 100 7 50 25 25 0 Table 35 Scenarios for CORSIM nopassing zone configuration test Facility A Facility B Facility C Link length (mi) Passing allowed Link length (mi) Passing allowed Link length (mi) Passing allowed 5 N 2.5 N 1 N 5 Y 2.5 Y 1 Y 2.5 N 1 N 2.5 Y 1 Y 1 N 1 Y 1 N 1 Y 1 N 1 Y

PAGE 66

66 Table 36 General CORSIM inputs for nopassing zone configuration test Inputs Values Geometric data facility length (mi) 10 n umber of links 20 l ink length (mi) 0.5 grade (%) 0 p ercentage of no passing zones (%) 5 0 Demand data p ercentage of passenger car s (%) 100 p ercentage of trucks (%) 0 FFS (mi/h) 60 Table 37 General input for speedflow relationship and PTSF vs. flow rate test Inputs HCM CORSIM Geometric d ata facility length (mi) 5 5 lane width (ft) 12 N/A shoulder width (ft) 6 N/A access point density (point s/ mi) 0 N/A grade (%) 0 0 radius of curvature (ft) N/A 0 superelevation (%) N/A 0 Demand d ata PHF 1 N/A FFS (mi/h) 65 65 directional split 50/50 N/A heavy vehicle percent age (%) 0 0 Table 38 Variables used in HCM and CORSIM testing Flow rate s (veh/h) Splits % HV % NPZ % Grade Passing l ane l ength (ft) 200 50/50 0 0 0 0 700 60/40 10 50 6 5280 1200 70/3 0 100 1700 2200 3200

PAGE 67

67 CHAPTER 4 RESULTS Preliminary CORSIM Test Results Three main preliminary tests were done in CORSIM for guidance on how to properly compare the 2010 HCM with CORSIM. The CORSIM values that best reflected what the HCM most likely assumes were chosen as the inputs for the main CORSIM and HCM comparison after analyzing the results of these tests. This section discusses the results of each preliminary test. Existing Upstream Conditions For the case with a lead up length that allows passing and a follow up length that does not allow passing, the analysis direction showed a significantly higher number of total passes on the lead up segment and along the whole facility. Most of the passes were happening on the lead up segment. Although the opposing direction had many platoons by the time the vehicles reached the lead up segment, the analysis direction vehicles were generated with spacing that led to inadequate passing gaps for the opposing direction vehicles Therefore, the opposing direction traffic could not pass even though most of them were in platoons. The analysis direction vehicles were generally not in long platoons on the lead up segment because they had just been generated from the eastbound node. However, even when one vehicle was following for a short amount of time, the vehicle was quickly able to pass because of the long gaps between the westbound platoons. Figure 41 shows a screenshot of this phenomenon. In order to prevent lopsided passing, both the lead up and follow up segments were specified as nopassing zones. The performance

PAGE 68

68 measures were almost the same between the two directions for both the lead up and follow up segments being specified as nopassing zones as shown in Table 41 The PTSF results we re almost the sam e for all three facilities for lead up lengths between two and five miles long for passing allowed as shown in Figure 42 For nopassing allowed, the plots of PTSF for each facility follow the same general pattern, but the PTSF increased as facility length increased for all of the lead up lengths that were tested as shown in Figure 43 However the difference in PTSF between the three facilities decreased as the lead up length increased ATS was unaffected by whether or not there was a lead up segment as shown in Figure 44 and Figure 45 A lead up length of five miles was chosen because it led to a small difference in the PTSF results for both passing allowed and passing not allowed. This allowed the fivemile lead up segment to be used for all passing zone scenarios. Five miles was also chosen as the follow up segment length in order to maintain symmetry in the facility. It is necessary to have a lead up segment in CORSIM because existing traffic conditions need to be establ ishe d prior to the data collection section Figure 46 shows the difference in PTSF for facilities with a five mile lead up segment and facilities with no lead up length when passing is allowed and Figure 4 7 shows the results for facilities where passing is not allowed. The short facilities are affected more than the long facilities by the presence or absence of a lead up segment for both passing zone scenarios. Truck Type Distribution Seven different truck type splits were tested and each scenario that had some percentage of type 6 trucks had results that were considerably different from the

PAGE 69

69 scenarios with no type 6 trucks. These scenarios included test 3, 4, and 6. The type 6 trucks cause the PTSF to be high and ATS to be low compared to the HCM results Tests 1, 2, 5, and 7 match ed up closely to the HCM results for PTS F for both passing allowed and passing not allowed as shown in Figure 48 and Figure 49 All seven tests either underestimated or overestimated the HCM results for speed. The HCM shows a linear relationship between speed and flow while the CORSIM truck type tests all showed a curve shape that initially has a fairly steep negative slope for low flow rates and then the curve slope decreases to the point where it is fairly flat o ver the moderate to high flow rates which supports Luttinens ( 5 ) and Brilon and Weisers ( 8 ) findings about the speedflow relationship on twolane highways Tests 1, 2, 5, and 7 showed ATS results that were higher than the HCM estimates and tests 3, 4, and 6 showed results that were lower than the HCM estimate s for both passing allowed and passing not allowed as shown in Figure 410 and Figure 411. The test 1 and test 5 values are very close to the test 2 and test 7 values Therefore, the test 1 and test 5 curves are somewhat overlapped in the figures by the test 2 and test 7 curves. Ultimately, the test 7 truck percentage split was chosen to be used for the major comparison of the HCM and CORSIM because it has the most realistic distribution of truck type s and matched better than the ot her tests for all four of the figures shown. Passing Zone Configuration The average PTSF for all three facilities matched up well with the HCM estimates as shown in Figure 412. This contradicts Luttinens ( 6 ) findings, which showed that the HCM overestimated PTSF As the flow rate increases, the PTSF increases rapidly for the lower flow rates, but then levels off to a gentl er slope for the higher flow rates The average ATS was similar between all three facilities, but did not match up well with the

PAGE 70

7 0 HCM estimate for ATS The HCM shows that the ATS follows a linear decreasing line, while all three facilities showed that the ATS followed a curve shape that initi ally has a fairly steep negative slope for low flow rates and then the curve slope decreases to the point where it is fairly flat over the moderate to high flow rates as shown in Figure 413. The curves began to level off after the flow rate reached 1200 veh/h. Figure 413 is consistent with other sources ( 5 8 ) where the speedflow curve was found to decrease sharply at first, then level off to a gentler slope rather than follow a linear trend as the HCM predicts. Since the results for PTSF and ATS were similar between all three facilities, facility A was chosen as the 50% nopassing zone configuration for the major HCM and CORSIM comparison. Facility A was chosen because it is the simplest of the three facilities with the first five miles being specified as a nopassing zone and the last five miles being specifie d as a passing zone. Any of the facilities would have been adequate for the major comparison since the passing zone configuration does not have an impact on the average of the performance measure results. Speed vs. Flow Rate and PTSF vs. Flow Rate Relation ship The HCM and CORSIM show similar results for PTSF vs. flow rate for their respective passing zone specifications as shown in Figure 414. The HCM does not consistently overestimate PTSF for either passing zone configuration. The results ar e inconsistent with Luttinens ( 6 ) claim. The speed vs. flow relationship between the HCM and CORSIM is drastically different as shown in Figure 415. The HCM shows that ATS decreases in a linear fashion as flow rate increases while CORSIM shows that ATS has a curve shape that initially has a fairly steep negative slope for low flow rates and then the curve slope decreases to the point where it is fairly flat over the moderate to high

PAGE 71

71 flow rates These results are consistent with Luttinens ( 5 ) fi ndings and Brilon and Weisers ( 8 ) findings. 2010 HCM and CORSIM Comparison Results Each combination of variables described in Table 38 were tested in CORSIM and the 2010 HCM. The PTSF and ATS were plotted against twoway flow rate for the three different splits to allow for easy comparison of the HCM and CORSIM. This process was repeated for all three passingzone cases both grade cases and both percentages of heavy vehicles for scenarios with no passing lane. For the cases with passing lanes, performance measures were plotted against facility length for the 60/40 split only. The results of this test are discussed in this section. 0% No Passing Zone with 0% Grade The 2010 HCM and CORSIM followed the same increasing trend for the PTSF plot s for both 0% and 10% heavy vehicles as shown in Figure 416 and Figure 417. When the twoway flow rate reached 3200 veh/h, the HCM showed a PTSF estimate of 98.6% for the 70/30 split while CORSIM showed a PTSF estimate of 91.1% It is unrealistic that the PTSF would be nearly 100% because the traffic stream always breaks up into sever al platoons due to slow truck s. Figure 418 shows the platoon structure for a two way flow rate of 3200 veh/h under a 70/30 directional split. Although t he directional flow rate is 2240 veh/h, there are still some gaps between vehicles. The ATS plots are consistent with t he preliminary speed flow tests. The HCM has a linear decreasing tr end and CORSIM has a curve shape that initially has a fairly steep negative slope for low flow rates and then the curve slope decreases to the point where it is fairly flat over the moderate to high flow rates for both 0% and 10% heavy vehicles. T he ATS values are similar between 0% and 10% heavy vehicles. The ATS plots are

PAGE 72

72 shown in Figure 419 and Figure 4 20. The plots show that f low rate may not have such a large effect on speed as the HCM shows. 0% No Passing Zone with 6% Grade For 0% heavy vehicles, CORSIM showed higher PTSF results than the H CM for flow rates lower than 1700 veh/hr. For fl ow rates higher than 1700 veh/h, the HCM showed higher values for PTSF than CORSIM. The trend for PTSF is similar between the two tool s. For 10% heavy vehicles, the HCM shows the same general trend, but the values for PTSF are higher because grade has a large effect on truck speeds and slow moving trucks cause platoons to form. The HCM PTSF value at 3200 veh/h for the 70/30 split is 99.9%, which is unrealistic. The reason may be that the HCM equations are not valid for two lane highways after breakdown. I n CORSIM, the PTSF results at 200 veh/h are about 50% higher than the HCM results for a 50/50 directional split and about 60% higher than the HCM results for the 60/40 and 70/30 directional splits for 10% heavy vehicles For flow rates higher than 7 00 veh/h, the PTSF is consistently near 95% for the CORSIM curve while the HCM trend is similar to the 0% heavy vehicle case. The values given by CORSIM seem unreasonably high, especially for the lower flow rates The PTSF values are high because long platoons form on the upgrade and the vehicles travel at high speeds on the opposingdirection downgrade. Therefore, there are few gaps for passing opportunities because the opposing vehicles arrive frequently. The PT SF plots are shown in Figure 421 and Figure 4 22. For 0% heavy vehicles, CO RSIM and the HCM had the same trends and values for ATS as with the 0% grade. For 10% heavy vehicles, the HCM ATS values generally followed a linear decreasing path. The directional splits showed slight differences in

PAGE 73

73 ATS for values before the flow rate of 1700 veh/h. At a flow rate of 3200 veh/h the speed drops to an unreasonably low value of 4.7 mi/h. The ATS plots are shown in Figure 423 and Figure 424. The CORSIM results for ATS for 10% heavy vehicles followed the same trend as the 0% grade scenario, but the values for ATS we re much lower. These low ATS values are caused by slow moving trucks in the traffic stream due to the grade. Even for low traffic volumes, there are few opportunit ies to pass because the vehicles in the opposing direction are on a 6% downgrade. They are traveling at high s peeds so vehicles cannot be in the opposing lane very long before the oncoming vehicle approaches and, as a result, vehicles in the major direction have inadequate time to pass For a flow rate of 200 veh/h, the CORSIM ATS for a 50/50 split wa s higher than the CORSIM ATS for a 60/40 split and a 70/30 split. In order to further investigate the CORSIM PTSF and ATS results for 10% heavy vehicles on a 6% grade, two additional splits, 40/60 and 30/70, were analyzed. The results are shown in Figure 425 and Figure 4 26. There was a greater difference between the split s for the flow rates of 200 and 700 veh/h. Then, the curves for all splits converged to about 96% for PTSF and about 23 mi/h for ATS. The performance measures are more sensitive to splits for lower flow rates because the number of vehicles in the traffic s tream is major factor for vehicle interactions, which affects platooning. Vehicles are generated more frequently for a 70/30 split than for a 30/70 split. One slow truck in the 70/30 split traffic stream impedes all subsequent vehicles. There are fewer tot al vehicles to be affected by a slow truck in the 30/70 split.

PAGE 74

74 50% No Passing Zone with 0% Grade The 2010 HCM and CORSIM followed the same trend for 50% nopassing zones as for the corresponding 0% nopassing zone plots for PTSF for both 0% and 10% heavy v ehicles as shown in Figure 427 and Figure 428 However, the HCM 70/30 split has more separation from the other splits for CORSIM and the HCM than in the 0% nopassing zone case. This is especially true at the flow rate of 2200 veh/h. The HCM and CORSIM followed the same trend for 50% nopassing zones as for the corresponding 0% nopassing zone plots for ATS for both 0% and 10% heavy vehicles as shown in Figure 429 and Figure 430. 50% No Passing Zone with 6% Grade The HCM and CORSIM PTSF curves have a more similar slope for 0% heavy vehicles than for t he 0% no passing zone case. The HCM curve for the 7 0/30 directional split shows the greatest values for PTSF and is slightly separated from the other curves. Overall, the PTSF values are higher for 50% nopassing zones than the PTSF values for 0% nopassing zones because there are fewer passing opportuniti es in the 50% nopassing zone case. For 10% heavy vehicles, the HCM and CORSIM followed the same trend for 50% nopassing zones as for the corresponding 0% nopassing zone plots. However, the HCM curves have more separation between themselves and the CORSIM 60/40 split is more separated from the CORSIM 70/30 split at the flow rate of 200 veh/h. The CORSIM PTSF values for the 50/50 and 70/30 split are higher than for the 0% nopassing zones case and the 60/40 split is lower for the flow rate of 200 veh/h. T he PTSF plots are shown in Figure 431 and Figure 432.

PAGE 75

75 The ATS plots for 0% and 10% heavy vehicles are similar to the 0% nopassing zones c ase. For 10% heavy vehicles, the ATS value for the CORSIM 60/40 split is about 5 mi/h higher at the flow rate of 200 veh/h than for 0% nopassing zones. The ATS plots are shown in Figure 433 and Figure 434. 100% NoPassing Zone with 0% Grade The HCM and CORSIM PTSF curves are much closer together for 100% nopassing zones for bot h 0% and 10% heavy vehicles than for the 50% nopassing zone case. The HCM 50/50 split shows the highest separation from the other curves. The HCM curves level off at lower PTSF values than the other two nopassing zone scenarios. The highest PTSF value reached for both heavy vehicle percentages is near 95% rather than 100%. PTSF shows improvement for 100% nopassing zones because the value for the nopassing zone adjustment factor must be extrapolated from Exhibit 1521 in the 2010 HCM when the twoway flow rate is greater tha n the highest value in the table. This could potentially lead to a negative nopassing zone adjustment factor, which would cause the PTSF to improve. The PTSF plots are shown in Figure 435 and Figure 436 The HCM and CORSIM followed the same trend for 100% nopassing zones as for the 0% and 50% nopassing zone cases for ATS for both 0% and 10% heavy vehicles The ATS plots are shown in Figure 437 and Figure 438. 100% NoPassing Zone with 6% Grade The HCM and CORSIM PTSF curves a re closer together for the 100% nopassing zone case for 0% heavy vehicles than for the 0% and 50% nopassing zone cases. Overall the PTSF values a re higher, but the maximum values reached for any curve are lower than the values reached by curves in the other two nopassing zone cases The plot is shown in Figure 439. For the 10% heavy vehicle case, The HCM 50/50

PAGE 76

76 directional split curve has more separation from the other two HCM curves than in the 0% and 50% nopassing zone cases. Also, the CORSIM 60/40 curve overlaps the CORSIM 70/30 split curve from the flow r ate of 200 veh/h to the flow rate of 700 veh/h as in the 0% nopassing zone case. For the lower flow rates, the PTSF values were all slightly higher than the values given for the other two no passing zone cases. The plot is shown in F igure 440. The ATS curves for CORSIM and the HCM have trends and values that are similar to the other nopassing zone cases for both percentages of heavy vehicles. However, the CORSIM 60/ 40 split overlaps the CORSIM 70/30 split curve from the flow rate of 200 veh/h to the flow rate of 700 veh/h as in the 0% nopassing zone case. The plots are shown in Figure 441 and Figure 442. Passing Lanes The ATS average for all passing lane scenarios is slightly higher than the corresponding no passing lane scenarios in both CORSIM and the HCM, but within about one mi/h. The PTSF values were affected more than ATS by the passing lane in both tools. CORSIM showed improvements as high as 17%, which were higher than the HCMs greatest improvements at 13%. The CORSIM results showed that the total PTSF improvements increased as the percentage of nopassing zones increased and that the improvements were greater for lower flow rates. The performance measures for the passing lane scenarios were analyzed using CORSIM based on the position along the highway in order to evaluate how a passing lane affects PTSF and ATS. Sinc e the no passing lane scenarios did not show much variation between splits, only the 60/40 split scenarios were plotted. This split was

PAGE 77

77 chosen to be plotted because it is the median between the perfectly even distribution of 50/50 and the biased distribution of 70/30. T he ATS values decreased at the passing lane link for all nopassing zone cases for the higher volumes This is counterintuitive, but there is a logical explanation. For the higher volumes such as 2200 veh/h and 3200 veh/h, most of the vehicles are in platoons throughout the entire facility because of a slower truck. Most of the vehicles in the platoons are passenger cars, which are capable of traveling at much higher speeds than trucks, especially when there is a 6% grade. As soon as the platoon reaches the passing lane section, most of the slow trucks move to the outside lane and the cars that have been following are able to travel through at their desired speeds. Eventually, the trucks reach the end of the passing lane section and have to merge back into the original lane. When a truck merges back in from the passing lane and continues onto the next link the cars are trapped again in a platoon behind the truck as shown in Figure 443 The cars are almost at a complete stop and back up onto the passing lane link. T he l ink directly after the passing lane link act s as a bottleneck. The plots for each nopassing zone case are discussed in the following sections. 0% no passing zones For 0% grade, the PTSF values increased as flow rate increased for both heavy vehicle cases The separation between the curves decreased as flow rate increased with the 200 veh/h curve having the greates t separation from the others. There was a large PTSF drop at the passing lane location. This happens because the slow trucks move into the right hand lane and the cars are able to travel through at their desired speeds. The effects of the PTSF reduction lasted furthe r downstream of the passing

PAGE 78

78 lane location for the lower flow rates. Overall, the PTSF values were higher for the 10% heavy vehicles case. The plots are shown in Figure 444 and Figure 445. The ATS curves are bunched close together for 0% grade. The passing lane does not have a large impact on speed. There wa s a slight increase in the ATS values after the passing lane for the f low rate of 700 veh/h. The ATS values were basically unchanged after the passing lane for the flow rate of 200 veh/h. All other flow rates had an ATS drop at the passing lane location. This happens at the higher flow rates because, when trucks merge back i nto the original lane, the cars are forced to slow down and become part of a slow moving platoon again. Since the passenger cars are able to travel fast on the passing lane link, they arrive quickly and frequently at the end of that link, but they get back ed up because of the slow moving trucks at the beginning of the next link. The ATS plots are shown in Figure 446 and Figure 447. For a 6% grade, the PTSF trend is similar to the 0% grade trend. The curves are closer together and the values are slightly higher. The 200 veh/h curve still has the greatest separation from the other c urves. The PTSF drop is larger for the higher flow rates but the downstream effects last longer for the lower flow rates. For 10% heavy vehicles, the flow rate curves are all close together except for the 200 veh/h curve. The PTSF has a large drop at the passing lane location for the flow rate of 3200 veh/h. The plots are shown in Figure 448 and Figure 449. The ATS values are slightly lower on the 6% grade for the 0% heavy vehicles scenario compared to the 0% grade scenario. For 10% heavy vehicles, the speeds jump at the passing lane location and decrease again for all flow rates except 2200 veh/h and 3200 veh/h. For the flow rates of 2200 veh/h and 3200 veh/h, the ATS jumped at the

PAGE 79

79 link before the passing lane link. This is because the slow trucks leading the platoons move to the right hand lane in the passing lane section, which creates a shockwave upstream. The passenger cars are suddenly free to travel at their desired speeds. Passing lanes are a more effective highway improvement method for low flow rates. The plots are shown in Figure 450 and Figure 451. 50% nopassing zones For the 0% grade, the trends are almost the sa me as for 0% nopassing zones. However, the PTSF values increased for the first four links in the 50% nopassing zone case whereas the values decreased in the 0% nopassing zone case until the vehicles reached the passing lane. The values increase because the platoons grow larger as the vehicles move through the facility. In the 0% nopassing zone case, the values decreased before the passing lane because the opposing flow rate was low enough that the following vehicles could make use of the passing zones. For the 50% case, passing was not allowed on the links leading up to the passing lane section. Therefore the PTSF values grew larger and larger until the passing lane relieved the following vehicles. There was no major difference between the 0% and 10% heavy vehicles cases as shown in Figure 452 and Figure 4 53. For the 0% grade, the ATS values increased more sharply due to the effects of the passing lane for the flow rates of 200 veh/h and 700 veh/h than in the 0% nopassing zone case. This is because the l inks prior to the passing lane link were specified as nopassing zones. Slow trucks on those links caused all the following vehicles to travel at lower speeds until they were finally able to pass on the passing lane link. All flow rates showed a small impr ovement after the passing lane section for both heavy vehicle percentages. The plots are shown in Figure 454 and Figure 455.

PAGE 80

80 For the 6% grade, the curves for the PTSF plot for 0% heavy vehicles are spaced out with the spacing between them getting smaller as the flow rate increases as shown in Figure 456. The PTSF values are much higher just before the passing lane. The 200 veh/h curve reached 36.2%, but in the 0% nopassing zone case it only reached 23.8%. The passing lane effects last longer for this no passing zone case The PTSF values did not return to the values they were at prior to the passing lane section as quickly. For 10% heavy vehicles, the values are the same as the corresponding 0% nopassing zone case except the 200 veh/h curve only returns to 67.4% by the end of the facility rather than 73% as it did for 0% nopassing zones. The plot is shown in Figure 457. For the 6% grade, the ATS plots have the same general trend as in the 0% nopassing zone case for 0% heavy vehicles. The curve for the flow rate of 3200 veh/h drops to a lower value at the passing lane than it did in the 0% nopassing zone case and the 200 veh/h flow curve increases aft er the passing lane as shown in Figure 458. For 10% heavy vehicles, the ATS values reached the highest point on the link after the passing lane for the flow rate curv e for 200 veh/h. That link is a passing zone and the vehicles that come from the passing section have just gotten away from slow trucks. The combination of these two factors is the cause of the ATS highpoint happening on link 6. For the flow curves for 220 0 veh/h and 3200 veh/h, the highest point occurs on link 4, which is the link just before the passing zone section. The traffic is extremely congested prior to the passing lane because of the few trucks in the traffic stream impeding all of the other vehic les As soon as a slow truck reaches the passing lane, the other vehicles are unimpeded and begin traveling at high speeds. This creates a

PAGE 81

81 shockwave that propagates through the links prior to the passing lane and link 4 is affected as shown in Figure 459. 100% nopassing zones For the 0% grade, the PTSF plots for both heavy vehicle percentages have the same trends and values as the 50% nopassing zone case until after the passing lane. The flow curves return more quickly to the values they were at prior to the passing lanes for this nopassing zone case as shown in Figure 460 and Figure 4 61. The passing lane effects do not carr y on to the downstream links when passing is not allowed. The ATS trends are also similar to the 50% nopassing zone case for both percentages of heavy vehicles except the values return quickly to the values they were at just before the passing lane. The A TS plots are shown in Figure 462 and Figure 463. For the 6% grade, the PTSF results have the same patterns as the 0% grade for 0% heavy vehicles. The 10% heavy vehicles case has the same trends as the 50% nopassing zone case, but for the 100% nopassing zone scenario, the PTSF values after the passing lane for the flow rate of 200 veh/h return quickly to the values they were at before the passing lane section. The plots are shown in Figure 464 and Figure 465. For 6% grade, The ATS trends and values are about the same as for the 50% nopassing zone case for 0% heavy vehicles. For 10% heavy vehicles, t he curve for the flow rate of 200 veh/h decreases more sharply after the passing lane section than it does in the 50% nopassing zone case and it decreases to a lower speed by the end of the facility than in the 50% nopassing zone case. The ATS plots are shown in Figure 466 and Figure 467.

PAGE 82

82 Passing Lane Field Observations In order to compare the CORSIM passing lane procedure to the true field procedure, observations were made on SR 40, where there are several passing lane sections. CORSIM incorporates a passing lane configuration where the slower vehicles make a lane change so faster vehicles can continue through as shown in Figure 468. SR40 had a different passing lane configuration where slow vehicle s continue on the original lane and fas ter vehicles pass to the left. The original lane merges back in later along the road as shown in Figure 469. The field observations were taken on Tuesday, October 5, 2010 and the traffic volume was low Typically, there were three vehicles in a group entering the passing lane section, including the observation vehicle. The other vehicles treated the passing lane section as if it were two full lanes. They passed to the right if the slower vehicle was not in the right most lane. There was a case where the observation vehicle was going about 10 mi/h under the speed limit. The observation vehicles stayed in the right most lane and two vehicles went by in the left most lane. Then, the second vehicle to go by changed lanes again and stayed in the slow vehicle lane until it was time to merge. This vehicle w as not going slower than the other vehicle that passed. The reason for this second lane change is most likely becaus e that vehicle wanted to be a free vehicle while it was on the two lane section. In order for the passing lanes to be used by slow vehicles as they were intended, the speed differential needs to be somewhat high. Otherwise, the faster vehicles do not have enough time to pass before the passing lane ends unless they go well over the speed limit.

PAGE 83

83 Capacity Estimate CORSIM was also used to estima te the capacity of twolane highways with and without passing lanes. Capacity on a twolane highway without a passing lane, according to the HCM, is estimated to be 1700 pc/h in one direction and 1500 pc/h in the opposing direction. CORSIM was used to esti mate the directional capacity for three different splits and freeflow speeds and two heavy vehicle percentages. Each case was tested with and without a passing lane. Table 42 shows the results of the capacity tests for 0% heavy vehicles and Table 43 shows the results for 10% heavy vehicles. For 0% heavy vehicles, the split had a slight impact on the capacities as well as speed. The passing lane did not change the capacity in any case. The passing lane does not improve capacity because the number of vehicles moving through the system is the same even though the passing lane may change how they interact with each other and allow some platoons to reconfigure themselves For 10% heavy vehicles, the capacity worsens for all cases. The capacity gets worse in CORSIM when there is a passing lane along with trucks because there is congestion at the merge area when vehicles in the passing lane try to move back over The passing lane makes capacity worse for 10% heavy vehicles but the results are the same with or without a passing lane for 0% heavy vehicles. The passing lane link ess entially detains some of the vehicles and the downstream throughput is affected. When the same number of vehicles moves through the facility without a passing lane, there is nothing causing the vehicles to slow down as the merge section does. Therefore, when there is no passing lane, the last link in the facility shows a higher throughput than when there is a passing lane.

PAGE 84

84 Performance Measures Analysis A qualitative analysis was performed in order to evaluate the effectiveness and usefulness of each performance measure that was discussed in the literature review. After being analyzed, t hey were given a score in several categories with 1 being the best and 5 being the worst score attainable. The points were averaged and the performance measures were ra nked with 1 being the best ranking and 5 being the worst ranking. The performance measure with the highest ranking was analyzed quantitatively. The advantages and disadvantages of each performance measure that was discussed in the literature review are sum marized in this section. Percent I mpeded PI shows a high change for the addition of a passing lane. Therefore, it is a useful performance measure for reflecting the effects of highway improvements. Also, PI is useful for capturing changes due to heavy vehic les in the traffic stream which is a major cause of platooning. PI showed a strong correlation to most platooning variables. This performance measure is limited in that it shows similar results for the analysis direction volume and the opposing direction volume, regardless of the split. This means an additional performance measure, such as v/c may have to be used to determine the LOS. PI is relatively easy to measure in the field because Pp can be estimat ed using a minimum headway threshold value and Pi can be determined by collecting speed data for slowmoving and free vehicles ( 10) These measurements can also be found with a simulation program. The PI calculation is simply Pp multiplied by Pi.

PAGE 85

85 Probability B ased F ollower I dentification Probability based follower identification is unique in that it determines follower status in a stochastic manner, which is better for capturing driver variability and changes in roadway conditions Probability based follower identification is consistent with the expectation that when the highway is near capacity it should approach 100% This performance measure is limited in that it was tested in Japan, which does not allow passing in the oncoming lane. This method may not be accurate for determining the LOS of two lane highways in the United States because of differences in driver characteristics and other unknown differences ( 9 ) Probability based follower identification is relatively difficult to measure in the field because there are challenges with finding the proportion of constrained vehicles based on headway without using a deterministic method. It is easier to find the following probability based on headway with theoretical models. Also, it is impossible to collect field data on the desired speeds of drivers because they are constantly changing. Some drivers may want to travel at higher speeds than the FFS and some drivers may want to travel at lower speeds than the FFS but there is no way for the analyst t o collect this data accurately Probability based follower identification is also somewhat difficult to calculate even with theoretical values obtained from models. The headway portion of this performance measure uses a mixed distribution model known as t he SPM to differentiate between following and free vehicles. This performance measure uses a critical headway, which is the largest headway that a vehicle can still be considered following, to determine following and free vehicles based on the headway dis tribution. The desired speeds are estimated by assigning different desired speeds to certain roadway conditions. If a

PAGE 86

86 vehicles speed falls below the assumed desired speed, then the vehicle is considered to be in a following state T he unified speed distribution method was used to estimate the desired speeds for this performance measure ( 9 ) Simulation could be used to track each individual vehicles speed at every second, which could lead to more accurate results for desired speed. Follower D ensity F ollower density is a useful performance measure for determining when capacity improvements need to be made ( 11) Volumes in the analysis direction and opposing direction were found to be statistically significant in the follower density models which is important because t raffic flow affects two lane highway performance measures more than other platooning variables such as opposing flow, percentage of heavy vehicles, percentage of nopassing zones, and speed differential ( 10, 14) F ollower density allows two lane highway LOS to be compared easily with freeways and multilane highways, which use density as a performance measure. Follower density internally incorporates speed, which means ATS would not have to be used as a performance measure ( 7 ) Follower density is based on the worst conditions along the highway rather than the average conditions This makes it easy to pinpoint where improvements need to be made, which can lead to sav ings in highway improvement costs. F ollower density gives the LOS based on the value of the performance measure for the traffic stream rather than for individual vehicles This eliminates having a situation where the average LOS is poor even when there are low traffic volumes leading to unnecessary highway improvements The traffic strea m approach takes traffic density into account, which gives a better indication of when highway improvements should be made ( 11)

PAGE 87

87 Follower density is relatively easy to measure in the field because it is based on the percentage of follower s, which is measur ed at a point The percentage of followers can be found at different sections of the facility in a deterministic manner. The density can be found based on speed and flow data. Follower density can be easily calculated by multiplying the percentage of follo wers by density ( 11) Freedom of F low Freedom of flow gives a good representation of drivers passing opportunities on two lane highways. This performance measure is closely related to the risk that drivers are willing to take, which means that it captures information about safety. Freedom of flow is limited in that it is a two way analysis tool. Therefore, it is not reliable for directional analyses. Freedom of flow factors can be easily estimated from existing traffic data and f reedom of flow is easy to c alculate. The most difficult part of calculating this performance measure is making correct assumptions about the variables used in the formula. Overtakings O vertakings focuses on drivers passing opportunities, which is important in twolane highway analysis because passing in the oncoming lane is a major twolane highway characteristic Overtakings account s for the effects of passing lanes and the percentage of nopassing zones. This performance measure is useful for identifying highways that need improvements ( 13) Overtaking is limited in that it is not informative about speed or delay. Since speed and delay are very important components of twolane highway operations overta kings should only be used as a secondary performance measure for LOS determination ( 13)

PAGE 88

88 The number of achieved overtakings is easy to count in the field with the appropriate equipment. The number of desired overtakings assuming a continuous passing lane section, is moderately difficult to determine because it is not possible to gather information about each drivers desire to pass. Desired overtakings can be estimated with a deterministic approach that consists of choosing a maximum time threshold that a vehicle is following before it should always have a desire to pass. The calculation for overtakings is simple. The challenge of this calculation is assuming reasonable and accurate values for the variables. Qualitative Analysis Results Probability based follower identification was given the worst ranking because it is very cumbersome to measure and calculate. Therefore, it is unlikely that professionals could readily apply this method of follower identification to projects. Also, it is based on a specific distribution of headways, which means if the headways on a twolane highway do not follow that distribution, then the method may not be applicable to that highway. The Catabagan and Nakamura ( 9 ) study showed that there were many cases where both follower identification methods led to the same results for follower status. For the results that led to differences between the two methods, there was no way to tell which one was correct. Both methods gi ve rough estimates of how the driver feels. Therefore, the more complicated, stochastic method is not well justified for follower identification compared to the simple, deterministic method. Freedom of flow was assigned the second worst ranking mainly bec ause of its difficulty to measure in the field. PI and overtakings received fours in the category for usefulness in determining LOS This is because they are meant to be secondary performance measures and should not be used independently to determine LOS.

PAGE 89

89 Follower density was given the highest ranking because of its ease of field data collection and calculation and also because of its usefulness in indicating when and where capaci ty improvements need to be made. Also, it can be easily compared to performanc e measures for freeways and multilane highways Table 44 shows the rankings of each of the performance measures that were analyzed. Quantitative Analysis Results Follower density was tested in a quantitative analysis because it was the highest ranked performance measure in the qualitative analysis The follower density methodology was programmed into CORSIM and is available in the run properties measure of effectiv eness screen. The 432 tests that were used to compare PTSF and ATS between the 2010 HCM and CORSIM were repeated in CORSIM for follower density. Figure 470 shows the general trends for all four combinations of grade and percentage of trucks under a 50/50 directional split for 0% nopassing zones and no passing lane. The curves all follow a linear increasing trend. The results of the follower density tests were somewhat counterintuitive for several scenarios. For 0% grade, the follower density was higher for 0% heavy vehicles than for 10% heavy vehicles, especially for the higher flow rates as shown in Table 45 and Table 46 There were also some cases where the follower density was higher for 0% heavy vehicles than for 10% heavy vehicles for the lower flow rates. The differences were small for the scenarios that had a higher follower density for 0% heavy vehicles than for 10% heavy vehicles. One explanation for the resul ts is that it was found that heavy vehicles do not perform much differently than passenger cars when there is a 0% grade. The 6% grade scenarios all showed the follower density being higher for 10% heavy vehicles in the

PAGE 90

90 traffic stream than for 0% heavy vehicles. This is because the 6% grade has a large impact on heavy vehicle performance. Another reason for these differences may be attributed to traffic streams with heavy vehicles having a lower density than corresponding traffic streams without trucks. Follower density is directly proportional to density. Unlike the HCM, CORSIM deals only in units of vehicles for the traffic stream, not an equivalent number of passenger cars. Trucks occupy more space than passenger cars Therefore, for the same flow rate, the density of the traffic stream will be lower (considering units of veh/mi/ln) for a traffic stream with a higher percentage of trucks than for a traffic steam with a lower percentage of trucks. Thus, increases in heavy vehicle percentages (consequently reducing the traffic stream density) may, for the most part, offset increases in follower percentage that the trucks usually create.

PAGE 91

91 Figure 41 CORSIM westbound platoons

PAGE 92

92 Figure 42 Middle segment PTSF for passing allowed Figure 43 Middle segment PTSF for passing not allowed 55 60 65 70 75 1 2 3 4 5 6PTSF (%)Lead up segment length (mi) 3 mi 5 mi 10 mi 55 60 65 70 75 80 85 1 2 3 4 5 6PTSF (%)Lead up segment length (mi) 3 mi 5 mi 10 mi

PAGE 93

93 Figure 44 Middle segment ATS for passing allowed Fig ure 45 Middle segment ATS for passing not allowed 40 45 50 55 60 1 2 3 4 5 6ATS (mi/h)Lead up segment length (mi) 3 mi 5 mi 10 mi 40 45 50 55 60 1 2 3 4 5 6ATS (mi/h)Lead up segment length (mi) 3 mi 5 mi 10 mi

PAGE 94

94 Figure 46 PTSF vs. facility length for passing allowed Figure 47 PTSF vs. facility length for passing not allowed 0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10PTSF (%)Facility length (mi) five mile lead up length no lead up length 0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10PTSF (%)Facility length (mi) five mile lead up length no lead up length

PAGE 95

95 Figure 48 Truck type test PTSF vs. flow rate for passing allowed Figure 49 Truck type test PTSF vs. flow rate for passing not allowed 0 20 40 60 80 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)Flow rate (veh/h) test 1 test 2 test 3 test 4 test 5 test 6 test 7 HCM 0 20 40 60 80 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)Flow rate (veh/h) test 1 test 2 test 3 test 4 test 5 test 6 test 7 HCM

PAGE 96

96 Figure 410 Truck type test ATS vs. flow rate for passing allowed Figure 411 Truck type test ATS vs. flow rate for passing not allowed 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)Flow rate (veh/h) test 1 test 2 test 3 test 4 test 5 test 6 test 7 HCM 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)Flow rate (veh/h) test 1 test 2 test 3 test 4 test 5 test 6 test 7 HCM

PAGE 97

97 Figure 412 Average PTSF vs. flow rate Figure 413 Average ATS vs. flow rate 0 20 40 60 80 100 0 500 1000 1500 2000 2500 3000 3500 4000PTSF (%)Flow rate (veh/h) Facility A Facility B Facility C HCM 0 10 20 30 40 50 60 0 500 1000 1500 2000 2500 3000 3500 4000ATS (mi/h)Flow rate (veh/h) Facility A Facility B Facility C HCM

PAGE 98

98 Figure 414 PTSF vs. flow rate Figure 415 ATS vs. flow rate 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500 4000PTSF (%)Two way flow rate (veh/h) 0% no passing zone HCM 0% no passing zone CORSIM 100% no passing zone HCM 100% no passing zone CORSIM 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500 4000ATS (mi/h)Two way flow rate (veh/h) 0% no passing zone HCM 0% no passing zone CORSIM 100% no passing zone HCM 100% no passing zone CORSIM

PAGE 99

99 Figure 416 PTSF vs. two way flow rate 0% grade, 0%NPZ, 0%HV, no passing lane Figure 417 PTSF vs. two way flow rate 0% grade, 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 100

100 Figure 418 Platoon structure for 3200 veh/h flow rate

PAGE 101

101 Figure 419 ATS vs. two way flow rate 0% grade, 0%NPZ, 0%HV, no passing lane Figure 420 ATS vs. two way flow rate 0% grade, 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 102

102 Figure 421 PTSF vs. two way flow rate 6 % grade, 0%NPZ, 0%HV, no passing lane Figure 422 PTSF vs. two way flow rate 6 % grade, 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 103

103 Figure 423 ATS vs. two way flow rate 6 % grade, 0%NPZ, 0%HV, no passing lane Figure 424 ATS vs. two way flow rate 6 % grade, 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 104

104 Figure 425 PTSF vs. two way flow rate 6 % grade, 0%NPZ, 1 0%HV, no passing lane Figure 426 A TS vs. two way flow rate 6 % grade, 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) CORSIM 3070 CORSIM 4060 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) CORSIM 3070 CORSIM 4060 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 105

105 Figure 427 PTSF vs. two way flow rate 0% grade, 5 0%NPZ, 0%HV, no passing lane Figure 428 PTSF vs. two way flow rate 0% grade, 5 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 106

106 Figure 429 ATS vs. two way flow rate 0% grade, 5 0%NPZ, 0%HV, no passing lane Figure 430 ATS vs. two way flow rate 0% grade, 5 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 107

107 Figure 431 PTSF vs. two way flow rate 6 % grade, 5 0%NPZ, 0%HV, no passing lane Figure 432 PTSF vs. two way flow rate 6 % grade, 5 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 108

108 Figure 433 ATS vs. two way flow rate 6 % grade, 5 0%NPZ, 0%HV, no passing lane Figure 434 ATS vs. two way flow rate 6 % grade, 5 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 109

109 Figure 435 PTSF vs. two way flow rate 0 % grade, 10 0%NPZ, 0%HV, no passing lane Figure 436 PTSF vs. two way flow rate 0 % grade, 10 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 110

110 Figure 437 ATS vs. two way flow rate 0 % grade, 100%NPZ, 0%HV, no passing lane Figure 438 ATS vs. two way flow rate 0 % grade, 100%N PZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 111

111 Figure 439 PTSF vs. two way flow rate 6 % grade, 10 0%NPZ, 0%HV, no passing lane Figure 440 PTSF vs. two way flow rate 6 % grade, 10 0%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500PTSF (%)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 112

112 Figure 441 ATS vs. two way flow rate 6 % grade, 100%NPZ, 0%HV, no passing lane Figure 442 ATS vs. two way flow rate 6 % grade, 100%NPZ, 1 0%HV, no passing lane 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030 0 10 20 30 40 50 60 70 0 500 1000 1500 2000 2500 3000 3500ATS (mi/h)flow rate (veh/hr) HCM 5050 HCM 6040 HCM 7030 CORSIM 5050 CORSIM 6040 CORSIM 7030

PAGE 113

113 Figure 443 Passing lane bottleneck

PAGE 114

114 Figure 444 PTSF vs. distance 0% grade, 0%NPZ, 0% HV, passing lane Figure 445 PTSF vs. distance 0% grade, 0%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 115

115 Figure 446 ATS vs. distance 0% grade, 0%NPZ, 0% HV, passing lane Figure 447 ATS vs. distance 0% grade, 0%NPZ, 1 0% HV, passing la ne 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 116

116 Figure 448 PTSF vs. distance 6 % grade, 0%NPZ, 0% HV, passing lane Figure 449 PTSF vs. distance 6 % grade, 0%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 117

117 Figure 450 ATS vs. distance 6 % grade, 0%NPZ, 0% HV, passing lane Figure 451 ATS vs. distance 6 % grade, 0%NPZ, 1 0% HV, passing lane 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 118

118 Figure 452 PTSF vs. distance 0% grade, 5 0%NPZ, 0% HV, passing lane Figure 453 PTSF vs. distance 0% grade, 5 0%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 119

119 Figure 454 ATS vs. distance 0% grade, 5 0%NPZ, 0% HV, passing lane Figure 455 ATS vs. distance 0% grade, 5 0%NPZ, 1 0% HV, passing lane 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 120

120 Figure 456 PTSF vs. distance 6 % grade, 5 0%NPZ, 0% HV, passing lane Figure 457 PTSF vs. distance 6 % grade, 5 0%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 121

121 Figure 458 ATS vs. distance 6 % grade, 5 0%NPZ, 0% HV, passing lane Figure 459 ATS vs. distance 6 % grade, 5 0%NPZ, 1 0% HV, passing lane 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 122

122 Figure 460 PTSF vs. distance 0% grade, 100%NPZ, 0% HV, passing lane Figure 461 PTSF vs. distance 0% grade, 100%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 123

123 Figure 462 ATS vs. distance 0% grade, 100%NPZ, 0% HV, pass ing lane Figure 46 3 ATS vs. distance 0% grade, 100%NPZ, 1 0% HV, passing lane 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 124

124 Figure 464 PTSF vs. distance 6 % grade, 100%NPZ, 0% HV, passing lane Figure 465 PTSF vs. distance 6 % grade, 100%NPZ, 1 0% HV, passing lane 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 20 40 60 80 1000 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800PTSF (%)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 125

125 Figure 466 ATS vs. distance 6 % grade, 100%NPZ, 0% HV, passing lane Figure 467 ATS vs. distance 6 % grade, 100%NPZ, 10% HV, p assing lane 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h 0 10 20 30 40 50 60 700 2640 5280 7920 10560 13200 15840 18480 21120 23760 26400 29040 31680 34320 36960 39600 42240 44880 47520 50160 52800ATS (mi/h)Distance (ft) 200 veh/h 700 veh/h 1200 veh/h 1700 veh/h 2200 veh/h 3200 veh/h

PAGE 126

126 Slow vehicles move over Figure 468 CORSIM passing lane configuration Slow vehicles continue in original lane and merge later Figure 469 SR40 passing lane configuration Figure 470 Follower density vs. two way flow rate 50 / 50 split, 0%NPZ, no passing lane 0 5 10 15 20 25 30 35 0 500 1000 1500 2000 2500 3000 3500Follower density (veh/mi)flow rate (veh/hr) 0G 0HV 0G 10HV 6G 0HV 6G 10HV

PAGE 127

127 Table 41 Lead up and follow up passing results L ead up segment pass allowed F ollow up segment pass allowed Number of EB passes Number of WB passes Difference in number passes EB PTSF EB ATS WB PTSF WB ATS N N 252 291 39 68.4 54.7 69.0 54.6 Y Y 471 668 197 63.5 55.2 65.3 55.0 Y N 589 201 388 62.0 55.4 70.8 54.3 N Y 126 692 566 70.9 54.3 62.8 55.4 Table 42 CORSIM capacity estimates 0% heavy vehicles Speed (mi/h) 50/50 split 60/40 split 70/30 split 65 Passing lane 2245 2240 2230 No passing lane 2245 2240 2230 60 Passing lane 2240 2230 2225 No passing lane 2240 2230 2225 55 Passing lane 2240 2225 2215 No passing lane 2240 2225 2215 Table 43 CORSIM capacity estimates 10% heavy vehicles Speed (mi/h) 50/50 split 60/40 split 70/30 split 65 Passing lane 2205 2205 2195 No passing lane 2235 2230 2240 60 Passing lane 2 215 2200 2200 No passing lane 2220 2215 2200 55 Passing lane 2225 2210 2 20 5 No passing lane 22 30 22 15 2220

PAGE 128

128 Table 44 Performance measure rankings Performance measure 1 2 3 4 5 6 7 Ease of conceptual understanding 1 3 1 2 1 1 1 Ease of calculation 1 4 1 2 1 2 2 Ease of field measurement 2 5 1 4 3 1 1 Usefulness in identifying improvement needs 3 3 1 3 1 3 3 Usefulness in capturing effects of improvements 1 2 1 3 4 2 3 Ease of adaption to different highway conditions 1 4 2 2 4 2 2 Usefulness in determining LOS 4 1 1 3 4 1 2 Average ** 1.9 3.1 1.1 2. 7 2.6 1.7 2. 0 Final ranking 3 7 1 6 5 2 4 1 Percent impeded, 2Probability based follower identification PTSF 3 Follower density, 4 Freedom of flow 5 Overtakings, 6Deterministic headway PTSF 7 ATS. ** 1 best average, 5 worst average Table 45 Follower density difference between 10% and 0% trucks no passing lane 0% Grade 6% Grade Flow 5050 split 6040 split 7030 split Flow 5050 split 6040 split 7030 split 0% NPZ 0% NPZ 200 0.04 0.07 0.07 200 0.78 1.93 1.84 700 0.54 0.72 1.27 700 4.19 5.22 5.71 1200 0.66 1.06 1.30 1200 5.49 6.51 6.84 1700 0.12 0.60 1.02 1700 6.57 7.12 8.26 2200 0.21 0.00 0.42 2200 6.86 6.95 8.77 3200 0.45 0.17 0.50 3200 8.64 9.05 7.00 50% NPZ 50% NPZ 200 0.02 0.04 0.06 200 0.85 1.57 1.65 700 0.17 0.22 0.37 700 4.44 5.60 6.27 1200 0.22 0.42 0.33 1200 6.77 7.90 8.65 1700 0.07 0.13 0.20 1700 8.23 10.34 10.69 2200 0.07 0.04 0.19 2200 9.26 10.87 13.37 3200 0.18 0.39 0.57 3200 12.85 14.39 11.94 100% NPZ 100% NPZ 200 0.01 0.01 0.05 200 0.96 1.94 1.84 700 0.03 0.06 0.20 700 3.32 4.26 4.00 1200 0.11 0.03 0.27 1200 4.86 5.07 4.65 1700 0.04 0.08 0.15 1700 5.31 6.36 6.92 2200 0.18 0.07 0.07 2200 6.23 6.69 6.92 3200 0.07 0.43 0.23 3200 7.85 8.19 4.82

PAGE 129

129 Table 46 Follower density difference between 10% and 0% trucks passing lane 0% Grade 6% Grade Flow 5050 split 6040 split 7030 split Flow 5050 split 6040 split 7030 split 0% NPZ 0% NPZ 200 0.01 0.02 0.03 200 0.35 0.73 0.77 700 0.15 0.30 0.41 700 4.07 5.23 6.17 1200 0.20 0.33 0.65 1200 7.03 8.58 10.85 1700 0.28 0.27 0.73 1700 9.45 11.75 13.72 2200 0.17 0.08 0.45 2200 11.71 14.63 17.56 3200 0.42 0.81 0.03 3200 15.33 15.85 12.95 50% NPZ 50% NPZ 200 0.00 0.01 0.02 200 0.41 0.69 0.98 700 0.06 0.13 0.25 700 3.75 4.72 5.68 1200 0.01 0.24 0.18 1200 6.58 7.64 9.52 1700 0.03 0.38 0.42 1700 9.07 10.67 12.67 2200 0.22 0.08 0.18 2200 10.88 13.00 15.21 3200 0.36 0.29 0.21 3200 15.08 14.89 11.24 100% NPZ 100% NPZ 200 0.00 0.01 0.01 200 0.46 0.85 1.25 700 0.04 0.01 0.18 700 3.77 4.62 5.07 1200 0.04 0.05 0.01 1200 6.50 7.33 8.29 1700 0.07 0.06 0.25 1700 8.99 10.10 11.26 2200 0.07 0.18 0.21 2200 10.92 12.41 13.79 3200 0.22 0.44 0.57 3200 14.75 15.31 10.29

PAGE 130

130 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS Overview The res ults of this study provide an assessment of the reasonableness of the two lane highway modeling algorithms used in CORSIM and an assessment of the validity of the two lane analysis methodology of the HCM. Input differences between the two tools were addressed and the outputs of each tool were compared. This study tested how different values of input variables and the presence of passing lanes affect the performance measures. The preliminary CORSIM tests provide users with guidance on setting up an analysis network This study gives i nsight on two lane highway capacity under a variety of traffic conditions and also provides an assessment of different performance measures that have been proposed for analyzing two lane highways. During this study, several potential enhancements and improvements to CORSIM were identified. This chapter presents the main findings of this research and discusses recommendations for CORSIM and HCM improvements as well as areas for fut ure research. Research Findings The preliminary tests in CORSIM showed that a lead up length is necessary to establish existing upstream conditions and that type 6 trucks in the traffic stream have a major impact on performance measures. The passing zone c onfiguration test showed that the average of the performance measures was not greatly affected by the passing zone configurations. There were differences at points along the highway between the three facilities, but the average of the performance measures for each facility was similar.

PAGE 131

131 The speedflow relationship between the HCM and CORSIM did not match up well. Other sources ( 5 8 ) indicate that the speedflow relationship is not linear and CORSIM further supports those claims. The HCM and CORSIM compari son showed very different results for the nopassing lane cases with 6% grade and 10% heavy vehicles. CORSIM had much higher values for PTSF and much lower values for ATS than the HCM for this condition. The ATS for the facility was not greatly affected by the addition of a passing lane, but the PTSF showed considerable improvements for lower flow rates Passing lane effects last further downstream for low flow rates than for high flow rates and they are more effective for upgrades. For high flow rates, the best solution for improving performance measures is adding a full lane because passing lanes only pr ovide a modest improvement at the passing lane location. The effects do not extend downstream after a certain flow rate depending on the other roadway and traffic conditions. The ATS graphs for passing lanes show that average speed drops at the passing lan e link for higher flow rates due to the congestion at the merge area. Therefore, although the total average speed for the facility increases slightly, by about 1 mi/h, the benefit to cost ratio w ould likely not support building a passing lane. This researc h provides capacity estimates for certain geometric and traffic scenarios including with and without a passing lane. CORSIM estimated the capacity to be about 500 veh/h higher than the HCM estimate. The CORSIM capacity estimate matches more closely with Ki ms ( 25) estimate than with the HCM estimate. This study also provides a qualitative analysis of several performance measures that could be useful for analyzing twolane highways. Follower density was ranked as the best

PAGE 132

132 performance measure for analyzing tw o lane highways because of its ease of field measurement and conceptual understanding as well as its usefulness in determining when improvements should be made. This performance measure was also analyzed in a series of quantitative tests. CORSIM Findings and Recommendations There are several areas where CORSIM could be improved. The preliminary tests showed some limitations within CORSIM that should be addressed. The passing procedures and passing lane results also sh ould be improved. This section discusses some of the changes that should be made in CORSIM and describes further testing that should be done in future research. Preliminary Test s The preliminary tests exposed some of the user friendly weaknesses in CORSIM. There should be an easier way for the user to establish existing upstream traffic conditions than by adding a lead up segment. The reason for using a lead up length is to avoid biased results, but there is no guarantee that the lead up length eliminates the possibility of biased results. Howev er, the lead up length is necessary for the vehicles to develop a platoon structure similar to what would be observed in the field before they enter the analysis section. The first recommendation is to add a record type in the CORSIM input file that allows the user to specify a percentage of vehicles entering in platoons Vehicles would be generated from an entry node as part of a platoon and t he user would no longer have to extract data only from the desired links The percentage of entering platoons was an input used in TWOPAS to establish upstream conditions ( 26) The two main challenges with this recommendation are that the user would have to measure the percentage of vehicles entering as platoons in the

PAGE 133

133 field for real world projects and a new algorithm would have to be developed for vehicle generation in CORSIM. The truck type distribution test showed that type 6 trucks have a major effect on the performance measures, even when only 2.5% of the traffic consists of type 6 trucks. Type 6 truck speeds should not be extremely different from the other truck type speeds on a 0% grade. The performance characteristics of the type 6 truck as contained in CORSIM, should be revis ited to see if they are still consistent with real world type 6 trucks. Passing Procedures CORSIM is programmed so that a vehicle that is following another vehicle will not pass unless there is a gap ahead of the impeding vehicle. T he impeding vehicle may initially ha ve a large enough gap in front of it for a vehicle to pass and move back into the traffic stream. However, i t is possible that during the pass the impeding vehicle reaches the back of another platoon. If the platoon is large enough, then the passing vehicle could be left in the opposing lane with no available gap in the original lane. There are two ways that CORSIM could be improved to address this shortcoming. The current passing procedure in CORSIM makes a check to see how far ahead the next vehicle is in front of the impeding vehicle before the pass begins CORSIM could implement this check for every time step during the pass and if the gap in the original lane becomes unacceptable, then the passing vehicle should abort the pass. Another sol ution could be to implement a methodology for driver cooperation for vehicles in the platoon. If a passing vehicle has no available gap to move back into as it initially had when it began the pass, then one of the vehicles in the platoon should adjust its speed and create a gap for the passing vehicle to fit into.

PAGE 134

134 Currently in CORSIM, vehicles only have a desire to pass when they are in a following mode (i.e., headway 3 sec). Realistically, when a faster vehicle approaches a slower vehicle, it is possible that the faster vehicle will pass the slower vehicle before it technically becomes a following vehicle. This is referred to as passing on the fly, and the passing algorithm in CORSIM should be modified to allow for this type of passing maneuver. The tests that compared the HCM to CORSIM showed high PTSF results for low flow rates when the grade was 6% with 10% heavy vehicles in the traffic stream The HCM results were much lower than the CORSIM results for the 6% grade. This large difference in results bet ween the two tools gives another reason for the truck type performance characteristics in CORSIM to be investigated further Speed data should be collected for different types of trucks on different upgrades so the proper changes can be implem ented into CO RSIM. There were instances in CORSIM where vehicles were following behind a slow vehicle and were unable to pass because the gaps in the opposing lane were unacceptable. The desire to pass increases with the time a vehicle spends in a following state acco rding to the CORSIM algorithm Eventually, it is likely that vehicles would be willing to go well above the posted speed limit for a short period of time in order to get around a slow vehicle. The slower the lead vehicle goes, the more likely following vehicles are willing to speed in order to get around. This is another reason that a passing on the fly algorithm should be incorporated into CORSIM. Vehicles should accelerate as they approach a vehicle that is going too slow so they can prepare for the pass If a vehicle accelerates to a high speed before the pass begins, it could potentially

PAGE 135

135 spend less time in the opposing lane. This could help increase the number of passes, which would consequently result in more platoon dispersion. Passing Lanes There are two types of passing lanes along twolane highways as shown in Figure 4 68 and Figure 469. CORSIM is only capable of modeling the passing lane type shown in Figure 468. The passing lane type shown in Figure 469 should be implemented into CORSIM so users can model twolane highways as accurately as pos sible. The passing lane algorithm should be modified accordingly because slow vehicles are expected to move over in CORSIMs current passing lane algorithm, whereas the slow vehicles continue on the original lane in the other configuration. Heavy vehicles typically take more time to change lanes than passenger cars. Once the other passing lane type is implemented into CORSIM, the two passing lane types should be compared in order to see if they have an impact on the results. 2010 HCM Findings and Recommendations The ATS trends between CORSIM and the HCM were very different. Other sources ( 5 8 ) support the CORSIM trend. Therefore, t he HCM methodology should be modified so that the speeds level off as the flow rate increases. Also, the HCM curves show that the PTSF reaches values just short of 100% for the 70/30 split i n the 0% and 50% nopassing zone cases. Vehicles naturally break up into many smaller platoons. They do not travel in one or two long platoons. The HCM PTSF methodology should be investigated further so it reflects accurate results for different conditions It is possible that the PTSF could be near 100% for heavy traffic in both directions. However, the flow rate of 2200 veh/h under a 70/30 split gives the directional flow rate of 1540 veh/h,

PAGE 136

136 which is less than the HCMs estimated capacity. For normal conditions where the traffic is not approaching capacity the PTSF should not be 100%. CORSIM provides a true estimate of PTSF in that it records every vehicles follower status at every time step and finds the percentage of time that each vehicle was in a following state. Then, for the average facility PTSF CORSIM takes the sum of the following time across all vehicles divided by the sum of the time spent in the network by all vehicles. The HCM method uses a regression equation for finding PTSF that is based on TWOPAS simulation results Although the two tools have different procedures for calculating the PTSF the difference between the results was small. The TWOPAS simulation PTSF calc ulation is most likely similar to the one used in CORSIM, which is the reason for small PTSF differences between the HCM and CORSIM. The HCM generally provides a good estimate for PTSF and does not consistently overestimate i t. This contradicts Luttinens ( 6 ) findings. For grades of 6%, the HCM ATS values go as low as 4 mi/h for the flow rate of 3200 for all directional splits when there are 10% heavy vehicles in the traffic stream The HCM gives the directional capacity to be 1700 pc/h. The flow rate of 3200 veh/h under a 50/50 split has 1600 veh/h in both directions, which means the major direction is operating at a directional flow rate that is less than the capacity according to the HCM Therefore, the HCM equations should be applicable to this case. The ATS calculations and adjustments for grade should be modified to give more reasonable results. Follower density has been implemented into CORSIM and basic tests were analyzed, but the CORSIM results could not be compared to the HCM as the HCM does

PAGE 137

137 not hav e a calculation methodology for follower density. The qualitative analysis done in this study showed that follower density has several merits as a performance measure for analyzing twolane highways. Follower density should be given strong consideration fo r inclusion in the HCM as a performance measure for evaluating LOS for twolane highways. This would allow for more comprehensive analyses of two lane highways. CORSIM could be used to develop an analytical procedure for follower density calculation, as well as appropriate LOS thresholds. Recommendations for Future Research The passing zone configuration tests showed that the three 50% nopassing zone configurations did not have a large effect on the results even though the two tools have different methods for inputting this variable. Different highway lengths and percentages of nopassing zones should be tested in order to confirm that the passing zone configuration does not have a large effect on the average PTSF and ATS for any specific facility There is a similar input difference between the HCM and CORSIM for two lane highways with a grade change. The HCM does not directly allow the user to specify where there is a grade change along the hi ghway. The user can split the highway into segments depending on where the grade changes and calculate the results based on the weighted averages. In future research, a test that is similar to the passing zone configuration test should be designed to test the difference in results between CORSIM and the HCM for twolane highways with a grade change. If there are major differences based on this variable, then a procedure should be implemented into the HCM that allows the user to specify where the grades are located along the highway.

PAGE 138

138 The graphs that were created based on the HCM and CORSIM comparison showed several areas where there were major differences between the two tools. The grade effects between CORSIM and the HCM were very different. Field data on tw o lane highways with grades should be used to give guidance on how the two tools should be adjusted to better reflect how vehicles are affected by grade. In the f uture, more extensive research should be conducted on how capacity is affected by the placement and frequency of passing lanes along a twolane highway. This study does not support the concept of capacity being increased because of the presence of a passing lane. The results of this study show that the passing lane hinders capacity when there are heavy vehicles in the traffic stream. This could be due to friction at the merge area, which is plausible in the real world. However, there is not conclusive evidence to support these results. Future research should be done on the merge behavior of vehicles especially under high flow rates. Further testing should be done to show whether or not the rearrangement of vehicles into faster moving platoons on passing lanes affect s overall highway capacity. One passing lane may not allow the vehicles enough opport unities to form faster platoons, but multiple passing lanes placed strategically along the highway may increase the capacity. Also, capacity estimates should be tested for different grades, percentages of heavy vehicles, and FFS in order to develop an accepted capacity standard for all types of conditions. Other countries have other techniques besides passing lanes to help platoons disperse. South Africa uses wide shoulders to allow slower vehicles to move over so that faster vehicles can go by in the regul ar lane. Finland uses wide lanes, (18 ft in each

PAGE 139

139 direction, for a total lane cross section width of 36 ft) in some locations to accommodate passing. Vehicles are able to pass by using the middle of the cross section, as the opposing vehicle and the vehicle being passed are expected to stick close to the right side of their respective lanes The HCM currently lacks sufficient adjustment factors to account for wide lanes and wide shoulders. CORSIM currently does not account for changes in lane and shoulder wi dth. The user has to make assumptions about what variables are affected by a change in lane or shoulder width and by how much. Then, the user must enter different values for those inputs according to the assumed effects. For example, if a twolane highway has lanes that are wider than 12 ft, the user may enter the FFS at a higher value than for 12 ft lanes. Wide lane and wide shoulder capabilities should be considered for inclusion in CORSIM to make it more usable to the international community. For future research, methodologies should be developed so that more of the performance measures in Table 44 can be quantitatively tested in CORSIM. After CORSIM is c apable of generating the results for the proposed performance measures, they can be compared with each other and then additional LOS methodologies could be developed for the HCM as appropriate.

PAGE 140

140 LIST OF REFERENCES 1. Highway Capacity Manual TRB, National Research Counci l, Washington, D.C. 2010. 2. Washburn, S. S. and J. Li. Development of a Simulation Program for TwoLane Highway Analysis University of Florida, Transportation Research Center, 2010. 3. University of Florida. Traffic Software Integrated System (TSIS CORSIM) Version 6.2. Mc Trans Gainesville, FL, 2010. 4. Highway Capacity Manual TRB, National Research Counci l, Washington, D.C., 2000. 5. Luttinen, R. T. Level of Service on Finnish TwoLane Highways. In Transportation Research Circular E C018: Fourth International Symposium on Highway Capacity TRB, National Research Council, Washington, D.C., 2000, pp. 175 187. 6. Luttinen, R. T. Percent TimeSpent Following as Performance Measure for TwoLane Highways. In Transportation Research Record 1776, TRB, National Research Council, Washington, D.C., 2001, pp. 52 59. 7. Luttinen, R. T., M. Dixon, and S. Washburn. TwoLane Highway Analysis in HCM 2000. Draft White Paper Presented at Transportation Research Board 84th Annual M eeting, Washington, D. C., 2005. 8. Brilon W., and F. Weiser. TwoLane Rural Highways, The German Experience. In Transportation Research Record 1988, TRB, National R esearch Council, Washington, D. C., 2006, pp. 3847. 9. Catbagan, J. L. and H. Nakamura. Probability Based Follower Identification in TwoLane Highways. 88th TRB Annual Meeting, 15p, DVD ROM. 2009. 10. AlKaisy A. and Z. Freedman. Estimating Performance on TwoLane Highways: Case Study Validation of a New Methodology. Pres ented at the Transportation Research Board 89th Annual Meeting, Washington, D.C., January 1014, 2010. 11. Van As, C. The Development of an Analysis Method for the Determination of Level of Service of TwoLane Undivided Highways in South Africa. Project Summary, South African National Roads Agency Limited Pretoria, 2003. 12. Polus A. and M Cohen. Theoretical and Empirical Relationships for the Quality of Flow and for a New Level of Service on TwoLane Highways. Journal of Transportation Engineering, A SCE, Vol. 135, No. 6, June 2009, pp. 380385. 13. Morrall J. F. and A. Werner. Measuring Level of Service of TwoLane Highways by Overtakings. In Transportation Research Record 1287, TRB, National Research Council, Washington, D.C., 1990, pp. 6269.

PAGE 141

141 14. AlKaisy, A., and S. Karjala. Indicators of Performance on TwoLane Rural Highways, Empirical Investigation. In Transportation Research Record 2008, TRB, National R esearch Council, Washington, D. C., 2008, pp. 8797. 15. Yu, Q ., and S. S. Washburn. Operational Performance Assessment for TwoLane Highway Facilities. Journal of Transportation Engineering, ASCE, Vol. 135, No. 4 April 2009 pp. 197205. 16. Harwood, D. W., C. J. Hoban, and D. L. Warren. Effective Use of Passing Lanes on Two Lane Highways. In Transportation Research Record 1195, TRB, National Research Council, Washington, D. C., 1988, pp. 7991. 17. Botha, J. L. and A. D. May. A DecisionMaking Framework for the Evaluation of Climbing Lanes on TwoLane, TwoWay Rural Roads. Report FHWA/CA/T080, University of California, California Department of Transportation, July 1980. 18. Polus, A., M. Livneh, and B. Frischer. Evaluation of the Passing Process on TwoLane Rural Highways In Transportation Research Record 1701, TRB, National Research Council, Washington, D.C., 2000, pp. 53 60. 19. A Policy on Geometric Design of Highways and Streets AASHTO Washington, D.C., 200 4 20. Kaub, A. R. and W. D. Berg. Design Guide for Auxiliary Passing Lanes on Rural Two Lane Highways. In Transportation Research Record 1195, TRB, National Rese arch Council, Washington, D. C., 1988, pp. 92100. 21. El Khoury, J. a nd A. G. Hobieka. Assessing the Risk in the Design of Passing Sight Distances. Journal of Transportation Engineering, ASCE, Vol. 133, No. 6, June 2007, pp. 370377. 22. Glennon, J. C. New and Improved Model of Passing Sight Distance on TwoLane Highways. In Transportation Research Record 1195, TRB, National Research Council, Washington, D. C., 1988, pp. 132137. 23. Hassan, Y., S. M. Easa, and A. O. Abd El Halim. Passing Sight Distance on TwoLane Highways: Review and Revision. Transportation Research Part A Vol. 30, No. 6, 1996, pp. 453 469. 24. Rozic P. Capacity of TwoLane, TwoWay Rural Highways: The New Approach. In Transportation Research Record 1365, TRB, National Research Council, Washington, D. C., 1992, pp. 1929. 25. Kim, J. A Capacity Estima tion Methods for Twolane Twoway Highways Using Simulation Modeling. Ph.D. dissertation of The Pennsylvania State University, 2006.

PAGE 142

142 26. Botha, J. L., X. Zeng, and E. C. Sullivan. Comparison of Performance of TWOPAS and TRARR Models When Simulating Traffi c on TwoLane Highways with Low Design Speeds. In Transportation Research Record 1398, TRB, National R esearch Council, Washington, D. C., 1993, pp. 7 16. 27. Harwood, D. W., A. D. May, I. D. Anderson, L. Leiman, and A. R. Archilla. Capacity and Quality of Service of TwoLane Highways. Final Report, NCHRP Project 355(3). Midwest Research Institute. University of California, Berkeley, November 1999. 28. Allen, R. W., D. W. Harwood, J. P. Christos, and W. D. Glauz. The Capability and Enhancement of VDANL and TWOPAS for Analyzing Vehicle Performance on Upgrades and Downgrades within IHSDM. Report No. FHWA RD00078, Federal Highway Administration, Washington, D. C., August 2000.

PAGE 143

143 BIOGRAPHICAL SKETCH Heather Hammontree grew up in Canton, OH where she graduated from Jackson High School in 2005. She completed her undergraduate studies at the University of Florida and graduated Cum Laude in the fall of 2009 with a Bachelor of Science in civil engineering. At the beginning of 2010, she began graduate studies at the University of Florida and completed a Master of Engineering in transportation engineering in the fall of 2010.