TABLE OF CONTENTS

NOTE : References for each chapter are listed at the end of that chapter.

CHAPTER PAGE

1 . INTRODUCTION 1 Project Objectives 2

2 . SURVEY OF PRACTICES Congestion 5 Freeways 5 Signalized Intersections 5 Two-Lane Roadways 5 Summary 6

3 . RURAL FREEWAYS 13

4 . RURAL TRAFFIC SIGNALS 15 Differences Among Methods 16 Differences Based on Time-of-Day 17 Differences Based on Location 17

5 . RURAL TWO-LANE ROAD PASSING LANE LITERATURE 23 Literature Review: Procedures for Justifying Climbing/Passing Lanes 23 Literature Review: Where to Install Passing or Climbing Lanes 34 Literature Review: Length and Spacing of Passing and Climbing Lanes 34 Literature Review: How to Conduct an Analysis of Passing/Climbing Lane Need 37 Literature Review: Observed Behaviors of Queued and Passing Vehicles 38

6 . SIMULATION OF PASSING AND CLIMBING LANES 43 Background of the Three Models 43 Highway Capacity Manual Approach 44 TRARR Use of Software 46 TWOPAS Use of Software 47 Simulation Test Case 49 TRARR 49 TWOPAS 51 v

Limitations of TRARR and TWOPAS 58 Summary 59

7 . RURAL TWO-LANE ROAD PASSING LANE FIELD STUDIES 65 Passing Lane Field Study Procedures 65 Passing Lane Field Study Data 66 Passing Lane Field Study Results 69 Poisson Analysis 77 Platooning Distribution 78

8 . COMPARISON OF ACCESS DESIGNS 81 Comparison of the Three Segments 81 Accident Histories 86 Travel Times 90 Observations 92 Summary 93

9 . SUMMARY AND CLOSING 95 Summary of Findings 95 Conclusion 97

APPENDIX A 99 APPENDIX B 103 APPENDIX C 107 APPENDIX D 111 APPENDIX E 143

LIST OF FIGURES

2 .01 List of responding agencies 7

2 .02 Survey results 8

5 .01 View of elevated observer collecting data 64

7 .01 View of elevated observer collecting data 67

7.02 Number of vehicles in platoons vs . total volume 80

7 .03 Number of platoons vs . total volume 80

8 .01 Photographs of each segment 82

LIST OF TABLES

1. 3 .01 Freeway Population Factors

2. 5 .02 South Africa's Traffic Volume Warrants for Climbing Lanes 27

3. 5 .03 Botswana's Traffic Volume Warrants for Climbing Lanes 27

4. 5 .04 Traffic Volume Warrants for Climbing Lanes on Two-Lane Highways . 28

5. 5 .05 Reduction in Percent Time Delay Per Unit Length of Passing Lane . 30

6. 5 .06 Level-of-Service Criteria for Two-Lane Highways 31

7. 5 .07 Optimal Passing Lane Length and Effective Distance for Given Traffic Volume 35

8. 5 .08 Optimal Passing Lane Length for Given Traffic Flow Rates 35

9. 5 .09 Passing Lanes Design Features for Canada and Australia 36

10. 6 .01 Summary of output 59

11. 6 .02 Comparison of Highway Capacity Manual, TRARR, and TWOPAS 62

12. 7 .01 List of Study Sites

7.02 Summary Speed Data

7 .03 Proportion of Vehicles by Speed

7.04 Summary Platoon Data

1. 7 .05 Proportion of Vehicles by Platoon Size

2. 7 .06 Proportion of Platoons by Size and Speed

3. 7 .07 Proportion of Vehicles Using Right Lane

4. 7 .08 Passing Attempts and Successes

5. 7 .09 Proportion Attempting to Pass for Given Speeds and Headways . . . .

6. 7 .10 Passing the Lead Platoon Vehicle as Function of Speed

7. 7 .11 Vehicle #2 Passing the Lead Platoon Vehicle

8. 7 .12 Distance to Complete a Pass Considering Speed

9. 7 .13 Distance to Complete a Pass Considering Speed and Headway . . . .

10. 7 .14 List of Sites

11. 7 .15 Sites Ranked by Volume

7.16 Platooning Data

8 .01 Description of the Three Segments 85

8 .02 Main Lane Intersection Characteristics 85

8.03 Accident Summary _ 87

8 .04 Accident Types 88

8.05 Travel Time in Minutes 91

CHAPTER 1
INTRODUCTION

Transportation agencies construct roadways to provide mobility for both people (in private automobiles and buses) and freight (in trucks) . When roadways become congested, they fail to adequately perform their intended function. Congestion is a symptom of an inadequate facility, one where demand exceeds supply. When congestion occurs, mobility has declined and delay has increased.

Congestion is somewhat a matter of expectations : if the traveling public expects a certain freedom of operation and that freedom is not met, then congestion is perceived to exist. A driver's perception of delay may result in one or more of the following events : delay, driver impatience, driver risk taking, or accidents . Congestion is not only an issue of personal frustration or economic loss due to delay of people and goods --congestion can also be a safety issue .

Roadway congestion is often perceived to be an urban problem. Examinations of urban congestion have attracted ongoing research funding (1-1) and even attention from the mass media. But traffic growth and capacity limitations outside of big cities have led to rural congestion. The resulting congestion problems can be especially acute in developing rural areas on the fringe of towns and small cities and rural tourist areas. A rural or small town local government may not have the knowledge, the experience, or the resources to adequately cope with traffic growth and congestion.

To properly manage the transportation system, transportation agencies must measure and assess the performance of various system components in order to identify needs and rank priorities . The 1991 Federal transportation legislation (ISTEA) ushered in a systems management era for transportation facilities, including the specifically-mandated Congestion Management System. To determine where congestion exists and the demand for improvements on the roadway system is greatest, planners and engineers need mathematical models that will compare traffic demand with facility supply (i .e ., capacity) .

The Highway Capacity Manual (HCM) is the standard capacity reference in the United States . It addresses such urban issues as sidewalk and transit system capacity, yet the HCM does not supply explicit information to assess certain rural capacity and congestion issues . A recent National Cooperative Highway Research Project (NCHRP) problem statement stated: "The procedures in Chapter 8 of . . . Highway Capacity Manual for analyzing two-lane highways have severe limitations. Low-cost improvements such as passing lanes . . . can be very effective in improving the operation of two-lane highways and can reduce the need to widen highways to four lanes . Chapter 8 of the HCM, however, lacks a procedure to estimate the effectiveness of these improvements" (1-2) .

Three specific rural roadway issues in need of additional study to address rural capacity/congestion problems include : PROJECT OBJECTIVES

This project was conceived as an exploratory, Phase 1 effort . The objectives of Phase 1 were to conduct a literature search and a survey to ascertain what methods had been developed and implemented at a local or state level . In addition, passing behavior on highways and small city access management controls were to be considered. To accomplish the project objectives, the following project tasks were planned and performed.

Review of Literature and Studies

Preliminary information indicated a few agencies had performed studies and prepared reports addressing one or more of these issues . Relevant literature from available journals and reports was reviewed and summarized.

Conduct a Nationwide Survey

The research team prepared and distributed a survey to U.S. state and Canadian provincial transportation departments . The purpose of the survey was to ascertain:

1. a . what existing data had been collected and/or methods had been developed at the local or state level to estimate the need for or capacity benefits from climbing lanes ;

2. b . in a rural environment, how agencies determine a need exists for more than a two-lane roadway ;

3. c . what rural-area freeway saturation flow rates had been defined by a locality or a state;

4. what rural-area signalized intersection saturation flow rates had been defined by a locality or a state;

5. e . what methods had been developed to assess overall rural-area congestion, if any; and

6. f . the degree of satisfaction with the preceding locally-developed meth

ods/rates . Respondents were queried to determine what institutional or organizational arrangements had been made to better deal with rural congestion either at the planning, design, or operations levels .

Conduct Limited Passing Field Study and Analysis

Limited field observations and video records were made of passing/climbing lane operation on US 412 east of Siloam Springs (before the new construction was completed and traffic was diverted off the "old" road) ; US 71 at Mt . Gayler; and US 71 about 10 km (6 miles) north of Alma. All three sites are in northwest Arkansas .

Assess Current Passing Simulation Models

Two computer programs, TWOPAS and TRARR, have been developed to assess passing lane operations with simulation techniques . Simulation is a flexible powerful tool for evaluating stochastic systems . Unfortunately, the current programs that have been developed for this particular application are not easy to use . An improved user interface is currently under development at University of California-Berkeley.

The team examined two currently available simulation programs that assess two-lane road operation, TRARR and TWOPAS, and compared them with real-word operation characteristics . Based on this assessment, design specifications for needed enhancements were identified. Actual work on needed enhancements would be proposed for future project phases .

Evaluate Effects of Different Access Control in a Small City

The quality of service of three arterial segments in a city with population of 40,000 (not a part of a larger metropolitan area) was compared . Each segment was four-lane with a non-traversable median. Traffic volumes on all three segments were similar . The quality of service of each segment was measured by means of travel time runs made during peak and off-peak periods and by evaluating accident histories over a three year period.

REFERENCES

this page left blank

CHAPTER 2
SURVEY OF PRACTICES

To ascertain the practices of state/province transportation agencies, a survey was mailed in December 1994 . The survey addressed issues related to rural or rural-suburban area

congestion,

freeways,

signalized intersections, and

two-lane roadways (widening with passing/climbing lanes or multilaning) .

The survey instrument stated that the focus was areas not in metropolitan planning organization (MPO) study areas . Responses were received through the first quarter of 1995 . Of the 62 surveys mailed, 31 U.S . states and 4 Canadian provinces returned responses

. Multiple surveys were returned from California, so responses from that state were tabulated separately from those of the other states . Figure 2 .01 lists the responding agencies and Figure 2 .02 presents the condensed survey with

response totals . The format of Figure 2 .02 is a listing of the questions on the left, with the response totals on the right .

CONGESTION

By a 2 to 1 margin, respondents said they did have rural and rural-urban transition areas where traffic congestion was a problem. About half of the agencies did not have methods to assess levels of congestion in rural and rural-urban transition areas .

FREEWAYS

Over 40% responded there were rural freeways in their state/province with traffic at saturation levels . Fewer than 15% had developed separate saturation flow rates or population factors (fe) to use in these situations

. Freeways with "repeat" commuters were specifically excluded from this set of questions .

SIGNALIZED INTERSECTIONS

Almost 50% of those surveyed had rural or small town signalized

intersections with traffic-heavy enough to conduct saturation flow rate and

lost time studies . Yet there appears to be almost no use of an adjusted

saturation flow rate or a population factor to incorporate habits of small

town or rural drivers in capacity or signal timing computations . A wide range

of assumed lost times (i .e ., per signal phase) appear to be in use .

TWO-LANE ROADWAYS Transportation agencies constantly face the issue of when to add passing lanes or climbing lanes to a two-lane roadway, or when to widen and convert to

a multilane facility . This series of questions addressed capacity-expansion options .

The decision to add passing is one of engineering judgement, not quantitative analysis, at about 2/3 of the responding agencies . Climbing lane analysis appears to be more widespread, with about half using quantitative methods . Only a small minority make use of computer simulation programs for

either passing or climbing lane analysis .

About half used a quantitative analysis to determine the need to widen to four lanes . Only a minority employ quantitative methods to evaluate the need for five lane (with a center two-way left turn lane) or four lane with non-

flush median designs .

SUMMARY

Based on the responses to this survey, congestion is like crime: it is no longer found only in urban areas . Freeways and signalized intersections in rural areas, rural-urban transition areas, and small towns experience heavy traffic .

Most agencies expressed satisfaction with their current method of analyzing the need to add climbing or passing lanes, and the need to widen and convert a two-lane road to a multilane facility. Most agencies were likewise satisfied with current freeway and signal saturation flow rates, and signalized intersection lost time values used in rural, rural-urban transition, and small town areas . Most transportation agencies have not developed factors separate and distinct from urban factors to evaluate saturation flow limits outside of the larger urbanized areas .

If in fact rural area driver characteristics (saturation flow rates, lost

time at traffic signals) are more relaxed than those in larger urban areas and reflected in the current Highway Capacity Manual, it is probable that agencies using the default Highway Capacity Manual factors for freeway or signalized intersection saturation flow are overestimating the ability of these

facilities to accommodate traffic . As a result, congestion management systems may be driven with faulty assumptions and inputs . Ranking systems based on volume-to-capacity (v/c) ratios may be weighed against rural areas . Traffic signal timing may also be affected .

Realistically, many agencies have neither the funding nor the inclination

to develop tailored saturation flow and signalized-intersection lost time values . The Highway Capacity Manual already includes adjustment factors for varying lane and shoulder widths, or driveway densities, without expecting each user to perform an independent study to develop tailored factors. It

seems equally reasonable to develop a sample of rural area population factors

(fp) or saturation flow rates, and include them so practicing planners and engineers can be better equipped to design and operate rural roadways .

CANADA

UNITED STATES

1. 1 . Arkansas

2. 2 . California (four responses)

3. 3 . Colorado

4. 4 . Connecticut

5. 5 . Georgia

6. 6 . Hawaii

7. 7 . Idaho

8. 8 . Iowa

9. 9 . Kansas

10. 10 . Maine

11. 11 . Maryland

12. 12 . Massachusetts

13. 13 . Michigan

14. 14 . Minnesota

15. 15 . Mississippi

16. 16 . Nebraska

17. 17 . New Hampshire

18. 18 . North Carolina

19. 19 . North Dakota

20. 20 . Oklahoma

21. 21 . Oregon

22. 22 . Pennsylvania

23. 23 . Rhode Island

24. 24 . South Carolina

25. 25 . South Dakota

26. 26 . Tennessee

27. 27 . Texas

28. 28 . Vermont

29. 29 . Virginia

30. 30 . Wisconsin

31. 31 . Wyoming

Note : District of Columbia reported "no rural areas;" totals

FIGURE 2 .01 --List of responding agencies

not included in survey

The purpose of this survey is to find what warrants, procedures, or practices

the various states and provinces use to evaluate the following rural roadway

situations . In your responses, please state to what degree you rely on material in the current (1985 as revised) Highway Capacity Manual

.

_QUESTIONS RESPONSE TOTALS

1. I . RURAL OR RURAL-SUBURBAN AREA CONGESTION YES NO

2. Do you have rural areas (permanent population under 50,000 ; areas not in a Metropolitan Planning Organization-MPO study area) in your state/province where traffic congestion is a problem? CAN 1 3

US22 8

D YES D NO CAL 1 2

B . Do you have rural-suburban transition areas in your

state/province where traffic congestion is a problem? CAN 2 2 US22 8

D YES D NO CAL 0 3

C . Does your agency have any method to assess overall rural or rural-suburban transition area congestion levels? CAN 1 3

US 1811 D YES, do have method 0 NO, do not have method CAL 0 3

II . RURAL FREEWAYS -- For locations where the drivers are YES NO not repeat commuters, but perhaps tourist areas or areas near special events such as fairs, athletic events,

concerts .

A. Do you have any rural freeway locations where traffic is sometimes heavy enough to conduct rural freeway saturation flow rate studies? CAN 1 3

US 1515

D YES, do have location(s) D NO, do not have location CAL 0 3

FIGURE 2 .02 --Survey results

What lost time per phase (sec.) values do you use for cities or areas over 50,000 population?

FIGURE 2 .02 con't --Survey results

YES NO

fp :
MD :1 .1,1 .0,0 .9

sec . : # states

CAL-

2to3 :1

US-

1to1.5 : 1

2: 1 2to 3: 2

3: 16 3+ALL RED : 1 3to4: 2 3to 5:2

4: 1 4to 5: 1

sec. : # states

US-

1.5to 2: 1

2 :3 2to3 :1

2.5to 3.5 : 1

3: 14 3+ALL RED : 1 3to4: 2 3to 5: 3

4: 1 10

IV. RURALTWO-LANE ROADWAY When you have a two-lane roadway in a rural or rural-urban fringe area, how do you determine when to . . .

A. . . . have passing lanes added? Was the procedure that you use to evaluate each roadway developed by "engineering judgment" or was it developed through a more rigorous quantitative analysis?

CAL US CAN • our method based on engineering judgement 3 18 2

• our method based on a more rigorous quantitative 1 9 2 analysis

• YES, do use computer simulation 1 3 1

• NO, do not use computer simulation 3 19 2

B . . . . have climbing lanes added? Was the procedure that you use to evaluate each roadway developed by "engineering judgment" or was it developed through a more rigorous quantitative analysis?

CAL US CAN • our method based on engineering judgement 2 14 2

• our method based on a more rigorous quantitative 1 16 2 analysis

• YES, do use computer simulation 0 3 1

• NO, do not use computer simulation 2 20 0

C. . . . widened to four lanes (no median)? Was the procedure that you use to evaluate each roadway developed by "engineering judgment" or was it developed through a more rigorous quantitative analysis?

CAL US CAN • our method based on engineering judgement 2 12 2

• our method based on a more rigorous quantitative 0 14 2 analysis

D . . . . widened to four through lanes plus a continuous two-way left turn lane? Was the procedure that you use to evaluate each roadway developed by "engineering judgment" or was it developed through a more rigorous quantitative analysis?

CAL US CAN • our method based on engineering judgement 2 ^{17 0}

• our method based on a more rigorous quantitative 0 9 0 analysis

FIGURE 2 .02 con't --Survey results

E . . . . be widened to four through lanes plus a raised or depressed median? Was the procedure that you use to evaluate each roadway developed by "engineering judgment" or was it developed through a more rigorous quantitative analysis?

CAL US CAN

• our method based on engineering judgement 2 15 1

• our method based on a more rigorous quantitative 0 10 2 analysis

V. SATISFACTIONWITHCURRENTMETHODSFORRURALAREAS

A. How satisfied are you with your current method for US :S -23 determining when a two-lane road (rural or rural-SD-5 suburban fringe) needs to have climbing or passing lanes D -2 added?

CAL:S -1

• Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied SD-2

CAN : S-2

B. How satisfied are you with your current method for US :S -20 determining when a two-lane road in a rural or rural-SD-7 suburban fringe area needs to be widened to four or more D- 3 lanes?

CAL: S-2

• Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied D-1

CAN: S-2

C . How satisfied are you with your current value for rural US :S- 19

area freeway saturation flow rate? SD- 9 Sat. Flow:

• Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied 1800- 2 2000- 8

What value do you use? 2200- 3 HCM - 9

CAL:S- 1 SD-1 Sat. Flow: 2000- 1

CAN :S-1 Sat. Flow : HCM-1

FIGURE 2 .02 con't --Survey results

12

D. How satisfied are you with your current method for US :S -24 determining rural freeway levels-of-service and/or SD-6 capacity? D - 0

0 Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied CAL :S-2

CAN:S-3

E . How satisfied are you with your current value for rural US:S-20 or small-town (permanent population under 50,000) SD-9 signalized intersections saturation flow rate and lost time values? Sat . Flow: 700 - 1

D Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied 1500-1 1700-1

What values do you use? 1800-8 1900-3 HCM - 8

Lost time : 3+ALL RED : 1 3TO 5: 2 3TO 4: 1

2: 1

3: 5

4: 1

CAL:S - 1 SD-1 D -1

F. How satisfied are you with your current method for US :S -25
determining rural or small-town signalized intersection SD-4
levels-of-service and/or capacity?
CAL:S-2
0 Satisfied 0 Somewhat Dissatisfied 0 Dissatisfied
CAN:S-2

FIGURE 2 .02 con't --Survey results

CHAPTER 3
RURAL FREEWAYS

The literature search found few published examinations of rural or rural-suburban area reduced capacities . The Bureau of Public Roads published the first highway capacity manual in 1950 . It was followed by the ^Highway

Research Board's 1965 version. The 1985 version included a "driver population _{factor" (f}p), based on the belief that commuter or other regular-user

populations drive the freeway more efficiently than do others, such ^as recreational drivers . The fp value is a multiplier used to decrease the standard saturation flow rate to better estimate the expected saturation flow rate for a given situation .

The 1985 manual reported that although data were sparse, capacities as low as 1500-1600 passenger cars per hour per lane (pcphpl) have been reported _onweekends, particularly inrecreationalareas. A fp range of 0.75-0.9 was

suggested (3-1) . The Transportation Research Circular 319 of 1987 (3-2) stated "There may also be regional differences in driver behavior that cause

significant variation in capacity."

Concepts from Sharma, Wyatt, and Werner were utilized to ^classify _roadways according to both trip length and driver population (related to trip purpose), leading to a methodology for estimating fp values . Using this

approach, roads in Alberta and Saskatchewan were assigned one of 7 classes and fp values, shown in Table 3 .01 .

TABLE 3 .01 --Freeway population factors

Class fp

urban commuter 1.0

regional commuter 0 .95

regional recreational/commuter 0 .90

interregional 0.85

long distance 0 .85

long distance/recreational 0 .80

highly recreational 0 .75

.

These values were based on "engineering judgement", not any reported tests

If evaluations of rural freeway capacity are based ^{on unreduced}

saturation flow rates, rural capacities may be overestimated ^{and levels of}

congestion may be underestimated . When comparing needs for funding, rural freeways will be at a disadvantage with urban area freeways, unless adjusted bases that reflect the saturation flow of each area-type are used.

REFERENCES

CHAPTER 4
RURAL TRAFFIC SIGNALS

When calculating signalized intersection capacities and traffic signal timings, transportation engineers and planners employ values for "saturation flow rate" and for "lost time" . Saturation flow rate describes the number of vehicles per unit time that can pass through an intersection, and lost time describes a certain portion of the signal cycle during which vehicles do not

move.

Traffic on any approach should proceed only during that part of the cycle when a green indication is displayed to that movement . Traffic accumulates during the red indication, and the resulting queue then discharges during the following green. When a queue of vehicles is released by the onset of green, the departure flow rate quickly increases until a uniform average departure rate is reached. This uniform departure rate is called the saturation flow rate of the intersection approach. Because the total flow for any single movement at a signalized intersection is controlled by the amount of green time allotted to that movement, saturation flow under these conditions is defined as the flow rate that would result if there were a continuous queue of vehicles and they were given 100% green time . Saturation flow is generally expressed in vehicles per hour of green time (vphg) (4-1) . Traffic signal timings and intersection performance are evaluated on the basis of saturation flow . Most current signalized intersection design methods rely on the critical lane technique, in which the critical lane, or group of movements, is determined from the relationship between the volume it carries and its saturation flow (4-2) .

The concept of lost time accounts for an observable phenomena : when a signal changes to green, a small amount of time usually elapses before vehicles begin to move in response to the new green. In addition, when the phase ends with the sequence of yellow becoming red, there may be a brief instant during the end of the yellow or an all-red phase when traffic on those approaches "losing the green" has stopped moving . In any event, there is a short period when the right-of-way is changing during which no traffic is moving. Thus, a fraction of the time allocated to each signal cycle is effectively wasted or lost .

The Highway Capacity Manual and other reference publications contain

standard values for saturation flow and lost time . However, guidance in

selecting the appropriate adjusted value for specific situations is lacking

. Transportation Research Circular 319 (4-3) noted there may also be regional differences in driver behavior that cause significant variation in capacity

.

If signalized intersection saturation flow rates in rural areas are less than

those in urban areas, then using flow rates derived from a predominately urban

area database will result in overestimating capacity and in setting ^green

times at traffic signals that are too short .

DIFFERENCES AMONG METHODS

Standard reference documents from Australia, Canada, and the United States recommend the use of measured saturation flows for the design and analysis of signalized intersections (4-2) . But in their use of saturation flow, the Highway Capacity Manual (HCM), the Canadian Capacity Guide for Signalized Intersections (CCG), and the Australian Road Research Board (ARRB) Report 123 differ with respect to the following attributes :

The HCM describes the saturation flow rate as the flow, in vehicles per hour per lane, that can be accommodated by the lane assuming that the green phase is always available to the approach. The CCG defines saturation flow as the rate of queue discharge from the stop line of an approach lane, expressed in passenger-car units per hour of green (pcu/hr green) . ARRB Report 123 defines saturation flow as the maximum constant departure rate from the queue during the green period, expressed in through-car units per hour (tcu/hr) .

All three documents agree on the traditional concept of saturation flow .

This concept assumes that, after an initial hesitation immediately following the beginning of the green interval, traffic discharges at a constant rate

(the saturation flow rate) until the queue is exhausted or until shortly after

the beginning of yellow, when a sharp drop in the flow occurs . The documents also agree on the great variability in saturation flows caused by different urban conditions and various geometric and traffic configurations

.

Two major categories of saturation flow surveys are techniques based on the successive times of vehicle discharge at a specified reference line, and techniques which count the number of vehicles passing the reference line during short portions of the green interval . The CCG uses the passage of the front bumper over the stop line as the time of discharge because it is consistent with the usual definition of a headway. The HCM and ARRB Report 123 are combinations of the two basic groups of surveys because they are based on the determination of the average headways during a specifically defined portion of the green interval . This portion starts with the passage of the fourth vehicle in the HCM method and after 10 sec of green in the ARRB technique .

To compare the three methods, 10 saturation flow surveys were conducted during the summer of 1989 in Edmonton, Alberta, Canada . These locations were carefully selected to conform with the general definition of basic saturation

flow as designated by the CCG. To have a generic basic saturation flow value to compare with the values found using the three methods, the computer program PCU was used. The PCU program finds values of the pcu equivalents using a

least squares optimization technique to minimize the variations in the number

of pcu discharging for green intervals of equal duration

. The difference between the results of the PCU program and the CCG was found to be negligible; on average, the CCG method yielded flows only 0 .13%

higher than those derived from the PCU program. The ARRB method yielded saturation flows from 5 to 23% higher than the PCU flows . The HCM values were in most cases higher than the PCU results (a range of 0 .3 to 11%) .

DIFFERENCES BASED ON TIME-OF-DAY

Differences reported in measured saturation flow rates may be in part a result of collecting data during different times . Leonard (4-4) studied saturation flows at triple left turns in a metropolitan area . He observed variations in saturation flow by time-of-day and by day of the week. Midday traffic streams, composed of non-commuting drivers, had saturation flow rates 3% lower than those of the evening peak period and 7% lower than those of the morning peak period . Weekend traffic exhibited a saturation flow rate 7% under the weekday rate .

While this study was not conducted in a rural area, the principle of variation in saturation flow by time of day may have wide-ranging applications . Supposed differences in measured saturation flow values from one roadway to another may in part be due to not collecting data at the same time of day.

DIFFERENCES BASED ON LOCATION

In the early 1980s, Agent and Crabtree (4-5) investigated a number of

factors affecting signalized intersection capacity. Two of the factors considered were location within a city, and city population . They referred to an Australian study by Miller showing no evidence of a relationship between

saturation flow and city size, and a Canadian study by Teply indicating

saturation flows were dependent on the size of a community only when local

traffic behavior was reflected . Local traffic behavior was concluded to be

associated with other community characteristics that were not necessarily

reflected in the population size .

Agent and Crabtree studied intersection saturation flow rates in Kentucky cities of varying population size . Data were collected only for vehicles that were part of a queue when the signal turned green or that had become part of the queue before reaching the stop bar. The location of each studied intersection was classified as central business district (CBD), fringe area, outlying business district, or residential area . The CBD was defined as the portion of a municipality in which the dominant land use is intense business activity . The fringe area was defined as the portion of a municipality immediately outside the CBD in which there is a wide range in type of business activity, generally including small commercial, light industrial, warehousing, automobile service activities, and intermediate strip development, as well as some concentrated residential areas . The outlying business district was defined as the portion of a municipality or an area within the influence of a municipality, normally separated geographically by some distance from the central business district and its fringe area, in which the principal land use is for business activity . The residential area was defined as the portion of a municipality or an area within the influence of a municipality in which the

dominant land use is residential development, but where small business areas may be included .

Due to the strong relationship between pedestrian activity and location in the city, these two factors were combined for analysis . In order to control for effects of other significant variables, the analysis excluded cities with populations under 20,000, and included through and left-turning

passenger cars, lane widths from 10 to 15 feet, grades from minus 3 .0 to plus 3 .0%, speed limits of 35 to 45 mph, and approaches with no parking within 200 feet of the stop bar .

Table 4 .01 shows saturation flow rates for various combinations of location in the city and level of pedestrian activity. There was little difference between values for fringe areas, outlying business districts, and residential areas . There also was little difference between values for light and moderate pedestrian activity levels . Locations with heavy pedestrian activity had saturation flow levels about four percent lower than other locations, as did locations in the CBD.

TABLE 4 .01 --Saturation Flow Rates (vphg, per lane) for location and for pedestrian activity levels

Location in city Level of pedestrian activity

To determine effects of city size on saturation flow, the eight cities for which data had been collected were classified into four population categories . The method of classification, the average population for each category, and the saturation flow rates observed for each category are shown in Table 4 .02 . The summary was also limited to through and left-turning passenger cars Data taken in the central business district and at locations with heavy pedestrian activity were excluded .

Saturation flow increased with increasing population, with levels in Louisville 21% higher than in the lowest population category. It appeared that for populations of about 20,000 to 500,000, population had a moderate effect on saturation flow. However, for populations under 20,000, saturation flow values decreased substantially.

TABLE 4 .02 --Effect of city size on saturation flow

Agent and Crabtree (4-6) also examined appropriate values to use for lost times at signalized intersections . The lost times at the beginning and ending of the phases were analyzed . Each intersection investigated was classified as being in a central business district (CBD), fringe area, outlying business district (OBD), or residential area. This classification used definitions given in the previous study (4-5) .

Analysis showed that lost time at the beginning of the cycle was highest for the CBD and lowest for the residential areas, with a relative difference of about 29%. The highest values for ending lost time occurred in the CBD, with values decreasing as locations got farther from the CBD. However, later analysis indicated those results were affected substantially by cycle length, which is closely related to location in the city. The results obtained after controlling cycle length in the study are shown in Table 4.03. This additional restriction virtually eliminated the difference in beginning lost time between CBD and fringe areas . The difference between the CBD and OBD locations was reduced from 17% to 10%. It was apparent that effects first observed for location in the city were at least partially due to the effect of cycle length. The ending lost time values continued to be highest in the CBD, indicating an adjustment factor was needed to account for location in the city.

TABLE 4 .03 --Effect of location in city on lost time

Location in city Lost time (seconds)

Beginning Ending

CBD 1.55 2.70 Fringe 1 .53 2 .18 OBD 1.40 1.78

Residential NA NA

The effects of city size on beginning lost time are shown in Table 4 .04 . The analysis was limited to through passenger cars at locations outside the central business district (CBD) with grades from minus 3 .0 to plus 3 .0% and speed limits of 35 to 45 mph . Results showed that, as city size increases, lost time at the beginning of the phase decreased, although the lost time was fairly constant for all cities with populations under 50,000 . Lost time for the largest city was only slightly more than one-half that for the smallest cities .

Ending lost time data summarized by city size category are also given in Table 4 .04 . Compared to Lexington data, ending lost time increased for the smaller cities and decreased for the largest city (Louisville) . The same conditions were used for analyzing ending lost time as were used for analyzing beginning lost time . The data suggested that adjustment factors were needed to decrease ending lost time from the base value in large cities (over 250,000 population) and to increase ending lost times in the smaller cities (under 20,000 population) .

TABLE 4 .04 --Effect of city size on lost time

2 Lexington 204,200 1 .28 1 .67

3 Bowling Green 40,400 1 .97 1 .79* Paducah 29,800 Frankfort 26,000 Richmond 21,700

4 Winchester 15,200 1 .95 2 .04 Somerset 10,600 Hazard 5,400

Note : Ending lost time study did not include locations in Frankfort

Zegeer compiled an extensive database to verify the adjustment factors and the basic saturation flow rate values so that the accuracy of the signalized intersection capacity analysis technique would be improved (4-7) . Saturation flow headways for more than 20,000 observations across the United States were collected for a series of 12 geometric, traffic characteristic, and environmental factors and compared with baseline saturation flow headways for various signal cycle length and phase combinations . The following saturation flow rate adjustment factors were proposed to reflect the effect of metropolitan area size :

population of more than 100,000 -- 1 .00 ;
population of between 20,000 and 100,000 --0 .91; and
population of less than 20,000 -- 0 .83 .
Hall and Ofori-Awuah examined saturation flow rates in eight New Mexico

municipalities of varying population (4-8) . Weekday data at 17 nearly level intersections with little pedestrian activity were collected . The traffic streams included a limited number of heavy vehicles . They found saturation

flow rates in smaller communities to be less than those recommended in the Highway Capacity Manual (4-9) . Longer headways at the front of discharging queues at a signalized intersections were observed in smaller towns. When population was less than 60,000, saturation flow rate was approximated by

SFR = 1245 + 11 .0 * population (in 1000s)

A comparison of saturation flow rates derived from each of the three preceding studies (Agent and Crabtree, Zegeer, Hall and Ofori-Awuah) yields the values in Table 4 .05 . The Zegeer values were computed from the 1985 Highway Capacity Manual "base" saturation flow rate of 1800 pcphpl .

TABLE 4 .05 --Comparison of signalized intersection saturation flow rates

Area Population

It should be noted that the reported values may be somewhat "dirty" ; that is, the effects of trucks or lane widths may not have been factored into the reported saturation flow rate values . However, from the original literature one can infer that the data were relatively "clean" .

REFERENCES

4-1 Robert W. Stokes, Vergil G. Stover, and Carroll J. Messer. "Use and Effectiveness of Simple Linear Regression To Estimate Saturation Flows at Signalized Intersections", Transportation Research Record 1091, 1986, pp . 95-101 . Transportation Research Board, Washington, D.C.

22

^4-2 S . Teply, A.M. Jones . "Saturation Flow: Do We Speak the Same Language?", Transportation Research Record 1320, 1991 . Transportation Research Board, Washington D.C.

4-3 Transportation Research Circular 319, June 1987. Transportation Research Board, Washington, D .C .

4-4 John D . Leonard II . "Operational Characteristics of Triple Left Turns ." Presented at 73rd Annual Transportation Research Board meeting, Washington, DC, January 1994 .

4-5 K.R. Agent and J.D . Crabtree . Analysis of Saturation Flow at Signalized Intersections, May 1982 . Kentucky Transportation Program, University of Kentucky, Lexington, KY .

4-6 K.R. Agent and J.D . Crabtree . Analysis of Lost Times at Signalized Intersections, 1983 . Kentucky Transportation Research Program, University of Kentucky, Lexington, KY.

4-7 John D . Zegeer . "Field Validation of Intersection Capacity Factors", Transportation Research Record 1091, 1986, p. 67-77. Transportation Research Board, Washington, D .C.

4-8 J. W. Hall and K. Ofori-Awuah. "Saturation Flow Rate variation with Population ." University of New Mexico, Albuquerque, NM .

4-9 Highway Capacity Manual, Special Report 209, 1985 . Transportation Research Board, Washington, D .C.

CHAPTER 5
RURAL TWO-LANE ROAD PASSING LANE LITERATURE

The greatest amount of project effort was expended examining passing/climbing lanes on two-lane rural highways . The effort had three components : literature review of passing/climbing lane issues, examination of TWOPAS and TRARR simulation packages, and field studies to collect and analyze passing lane data .

Passing and climbing lanes are auxiliary lanes added to the highway, usually over a relatively short distance . A passing lane allows faster vehicles to, for a short duration, get around slower vehicles . Climbing lanes are similar to passing lanes, except that climbing lanes are associated with upgrade sections where heavier vehicles slow down due to gradient . A passing lane is usually one added to a two-lane road, whereas a climbing lane may be added to a two-lane or to a multilane road. When added to a two-lane road, both types allow traffic queues which have formed behind slower vehicles to dissipate in a safer manner than by passing in an oncoming lane. Either on a two-lane or a multilane road, both allow a greater number of vehicles to proceed down the road without being impeded .

Passing and climbing lanes are built to provide an intermediate level of improvement on highway sections which may not meet warrants for adding additional lanes, but exhibit deteriorating levels-of-service in terms of reduced speeds, increased time spent following in platoons, and a demand for overtaking which exceeds the available opportunities (5-1) .

Two-lane rural highways compose about 970 of the total rural system and

800 of the national highway system but carry only 35% of the total annual vehicle-miles of travel (5-2) . More than two-thirds of the two-lane mileage

is in mountainous or rolling terrain characterized by steep grades and sharp curves . This system is responsible for over 48% of all fatal motor vehicle accidents each year (5-3) . The head-on collision is the second most common type of fatal accident on the rural two-lane system, causing 5,100 fatalities annually (5-4) . An important safety issue on two-lane rural roads relates to crashes involving passing maneuvers (5-5) .

LITERATURE REVIEW: PROCEDURES FOR JUSTIFYING CLIMBING/PASSING LANES

Explaining the need for passing and climbing lanes is much easier than quantifying the need. Various general guidelines have been devised for determining when passing and climbing lanes should be installed. Developing a widely accepted criteria based on a robust analytical procedure has proven to be an elusive goal . A review of literature found passing and climbing lane

justification based on : safety/accidents, lack of sight distance, traffic volume and composition, gradient, speed and speed reduction,

delay or platooning,

level-of-service (LOS), and

economics .

Based on Safety/Accidents

The passing maneuver is one of the most common and potentially dangerous two-lane operational maneuvers . However, the passing maneuver also has the capability to substantially reduce rural, two-lane travel time and delay. Therefore, on the rural two-lane system there exists a need to improve safety performance by reducing severe accidents while maintaining or improving

traffic operational performance (5-6) .

Analyses using highway safety information system data from 3 states found the proportion of passing accidents ranged from 0 .76°% to 2 310 of all accidents on all two-lane roads, and 1 .43% to 2 .63% of all accidents on two-lane rural roads (5-5) . Passing accidents were somewhat more severe than non-passing accidents . Sideswipe passing accidents were found more often than head-on passing accidents, indicating that drivers will "opt for almost anything to avoid a head-on crash" (5-5) .

Safety evaluations have shown that passing lanes and short four-lane sections reduce accident rates below the levels found on conventional two-lane highways (5-7) . Rinde conducted a before-and-after study of accident rates at 23 California widening projects in level, rolling, and mountainous terrain. Rinde concluded that passing lane installation reduced accident rates by 11 to 27%, depending on road width. The accident rate reduction effectiveness at the 13 study sites with level or rolling terrain was 42% (5-8) . In data from 22 sites in four states, Harwood et al . found that passing lanes reduced all accidents by 9% and fatal and injury accidents by 17% (5-9) . The combined data from both studies indicates that the use of passing lanes reduced accident rates 25% below those of conventional two-lane highways (5-7) .

Although simulation and field results provide convincing evidence of the

traffic flow benefits provided by auxiliary lanes, concerns regarding the safety performance of such methods remain . These concerns arise from the high intensity of traffic maneuvers on auxiliary lane sections, particularly the merge at the section end (5-10) . Harwood et al . compared accident rates in the diverge and merge portions of overtaking lanes with rates in the mid-portion. A slightly greater proportion of accidents occurred in the transition areas than would be expected from their relative length alone, but

the differences were not large . Thus, there was no indication of any marked

safety problem in the diverge and merge transition areas of passing ^{lanes (5}

9) .

Homburger specifically addressed the issue of accidents in the merge area by reviewing individual accident reports in the merge areas of 21 climbing lanes in California . These sites were chosen as being likely problem locations on the basis of geometry (particularly merge on up-grade), accident rates, and the opinion of highway agency personnel . Of the total 157 accidents reported for the 21 merge sites over a five-year period, only 11

were associates with a merge maneuver. It was concluded that merging _{at the}

end of climbing lanes was not a major safety problem (5-11) .

Based on Lack of Sight Distance

Passing opportunities on a two-lane highway depend on both the availability of adequate sight distance and the availability of gaps in the opposing traffic stream (5-12) . If adequate sight distance is not available, a highway section should be striped as a no-passing zone .

when passing opportunity supply falls below passing demand, platooning occurs (5-13) . Platoon sizes are proportional to delay and associated with an

inadequate LOS ; therefore, if inadequate sight distance exists, auxiliary

lanes should be constructed to provide passing opportunities .

Auxiliary lanes should not be constructed where there is adequate sight distance, unless the AADT is greater than approximately 5,000 . In situations where the annual average daily traffic (AADT) is less, a considerable amount of passing can already occur (5-14) . The Manual on Uniform Traffic Control Devices (MUTCD) requires a sight distance of 305 meters (m) (1,000 feet - ft) for 97 kilometers/hour (km/hr) (60 mph) travel before allowing passing (5-15) . In Oregon, passing is permitted where the sight distance exceeds 610 meters

(2,000 feet) (5-16) .

The current design criteria for passing sight distance on two-lane highways are set forth in the American Association of State Highway and Transportation Officials' (AASHTO) "Green Book" and are based on field studies conducted between 1938 and 1941 (5-17) . Based on these studies, the AASHTO policy defines the minimum passing sight distance as the sum of the following

four distances (5-17) :

dl: distance traveled during perception and reaction time and during initial

acceleration to the point of entering the left lane. d2: distance traveled while the passing vehicle occupies the left lane . d3: distance between passing vehicle and opposing vehicle at the end of the

passing maneuver (clearing distance), and d4: distance traveled by a vehicle in the opposing lane for two-thirds of d2 .

Certain assumptions were stated in the AASHTO "Green Book" in order to develop design values for the four distances described above . First, the passed vehicle travels at a uniform speed . Second, the passing vehicle reduces speed and trails the passed vehicle as it enters the passing section. Third, when the passing section is reached, the passing driver requires a short period of time to perceive the clear passing section and to begin to accelerate . Also, the passing vehicle accelerates during the maneuver, and its average speed during the occupancy of the left lane is 10 mph higher than that of the passed vehicle . Lastly, when the passing vehicle returns to its lane, it is assumed that there is a suitable clearance length between it and any oncoming vehicle in the other lane. Table 5 .01 lists the AASHTO design values for passing sight distance at various design speeds (5-17) .

TABLE 5 .01 --AASHTO passing sight distance requirements

Design Speed (mph)

40 5060 70

Assumed Speed of Passed Vehicle (mph) 34 41 47 54 Assumed Speed of Passing Vehicle (mph) 44 51 57 64 Total Sight Distance (ft) 1,500 1,800 2,100 2,500

Source : D .W. Harwood, et al . Truck Characteristics for Use in Highway Design and Operation, Volume I, FHWA-RD-89-226, 1990 . FHWA, Washington D .C .

Harwood et al . found the current MUTCD passing sight distance requirements for a passenger car passing another passenger car to be in close agreement with a model recently developed by Glennon (5-18), but determined that the AASHTO criteria for passing sight distance are much more conservative. Harwood also found that passing scenarios involving a passenger car passing a truck, a truck passing a passenger car, and a truck passing a truck require progressively more passing sight distance than a passenger car passing a passenger car . Trucks also require longer vertical curves if a passing zone is to be maintained over a crest (5-17) .

There are no current criteria for passing zone lengths, except for a default 400-ft guideline set by the MUTCD. For all design speeds above 30 mph, the distance required for one vehicle to pass another is substantially longer than 400 ft, indicating the need for longer passing zones . The required passing distances are increased substantially when the passing vehicle, the passed vehicle, or both, are trucks (5-17) .

A 1970 study evaluated two passing zones with lengths of 400 and 640 ft . Very few passing opportunities were accepted in these zones and, of those that were accepted, more than 70% resulted in a slightly forced or very forced

return to the right lane in the face of opposing traffic (5-19) .

Based on Traffic Volume and Composition

According to AASHTO, either the upgrade traffic flow rate must be in excess of 200 vehicles per hour (vph) or the upgrade truck flow rate must be in excess of 20 vph in order to justify the construction of a climbing lane

(5-20) . Traffic volume warrants for climbing lanes in South Africa, shown in

Table 5 .02, are based on different grades and various truck percentages in the traffic stream (5-21) . Botswana determines the need for climbing lanes based on the traffic volume warrants summarized in Table 5 .03 (5-22) .

TABLE 5 .02 --South Africa's traffic volume warrants for climbing lanes

Gradient (%) Traffic volume in design hour

5% Trucks 10% Trucks

4 632 486 6 468 316 8 383 243 10 324 198

Source : National Institute for Transport and Road Research. TRH 17: Geometric Design of Rural Roads. 1985 . CSIR, Pretoria.

TABLE 5 .03 -- Botswana's traffic volume warrants for climbing lanes

Percent trucks Minimum design hourly volumes in design hour Major Routes Other Roads

5 450 500 10 300 400 20 200 300 30 150 200

Source : Botswana Road Design Manual . 1982 . Botswana Ministry of Works, Gaberone .

Sananez and May developed a macroscopic computer model (RURAL1) for rural highways . The passing lane submodel of RURAL1 can calculate traffic performance for the direction that has the passing lane . The submodel can determine service volume given the ideal capacity, terrain, percentage of heavy vehicles, and lane width of a two-lane highway . The basic assumption of the passing lane submodel is that lane distribution is a function of volume and percentage of large vehicles in the traffic stream (5-23) .

Based on Gradient

Trucks and other heavy vehicles have a difficult time climbing hills. As they climb, they often impede the flow of traffic, reduce the highway's capacity to carry traffic, and create possible hazards to other vehicles . Studies have indicated that accidents happen more frequently and are more severe as speed differentials increase . These studies support the conclusion that hill-climbing lanes make highway travel safer (5-24) . A climbing lane is added to the upgrade side of a two-lane highway, and the center and downgrade outer lanes operate as a conventional two-lane highway.

28

AASHO (1965) expressed the capacity criterion for climbing lanes in terms of "practical capacity" terminology . AASHO stated that, provided the length of grade is greater than the critical length, a climbing lane is warranted if the design hour volume exceeds the service volume by 20% on the up-grade section for the design level-of-service of the highway. The allowable 20% traffic overload was based on the observation that drivers expect, and will tolerate, more congested operating conditions on up-grade sections relative to the average conditions over an extended length of highway. Table 5 .04 gives representative design hour volumes for which climbing lanes would be warranted according to the AASHO criteria. McClean claims that there was a logical flaw in the 1965 AASHO consideration of traffic capacity adequacy for extended lengths of two-lane highway: inadequate overtaking opportunity over the extended length could not be used as justification for climbing lanes on sections that did not meet the specific grade criteria . In this situation AASHO required construction of four-lane sections, or full upgrading to a four-lane highway (5-10) .

TABLE 5 .04 --Traffic volume warrants for climbing lanes on two-lane highways

Gradient Length of Two-way design hour volume (%) (km) (veh/hr) for given truck percentage

5% 10% 15%

1. 4 0.8 * 550 450

1. 1 .6 640 470 370

2. 3 .2 590 420 340

1. 5 0 .8 620 460 370

1. 1 .6 510 360 270

2. 3 .2 480 330 250

1. 6 0 .8 470 330 250

1. 1 .6 420 280 210

2. 3 .2 410 270 200

1. 7 0 .8 320 210 160

1. 1 .6 280 180 140

2. 3 .2 260 160 120

Note : * Four lanes warranted Source : John R . McLean . Two-Lane Highway Traffic Operations : Theory and Practice, pp 349-374, 1989 . Gordon and Breach Science Publishers, New York.

A 1979 California study analyzed the sustained speeds of more than 14,000 trucks and of more than 2,600 recreational vehicles, pickup trucks, vans, and other vehicles on various grades along rural freeways . The study showed that these speeds were much faster than the truck speeds which were for many years recommended by the Highway Capacity Manual and AASHTO (5-25) .

Traffic volume warrants for climbing lanes in South Africa, shown in Table 5 .02, are based on different grades and various truck percentages in the traffic stream (5-21) . Passing bays are considered in Australia when truck speeds drop and the following circumstances exist : grades of over 8t, a high proportion of heavy vehicles, low overall traffic volumes, and high construction costs (5-26) . The United Kingdom requires climbing lanes on grades steeper than 2% and longer than 500 meters when they are also economically or environmentally justified (5-27) .

Wolhuter and Polus studied the current warrants for and usage of climbing lanes . The relationships among flow, gradient, and speed were analyzed to create a simulation program that can associate delay and flow with various gradients . The delay per vehicle was found to be relatively constant regardless of the gradient (5-28) .

Based on Speed and Speed Reduction

AASHTO states that a minimum truck speed reduction of 16 km/hr (10 mph) justifies the addition of climbing lanes (5-20) . South African rural road design practices require a reduced truck speed of 20 km/hr (12 mph) for

climbing lane construction. Traffic volume warrants for the climbing lanes, shown in Table 5 .02, are based on different grades and various truck percentages in the traffic stream (5-21) . Botswana considers the use of climbing lanes when the capacity is restricted by sight distance and by truck speed reductions of 25 km/hr (15 mph) . Botswana determines the need for climbing lanes based on the traffic volume warrants summarized in Table 5 .03

(5-22) . Australia justifies climbing lanes based on traffic volume, percentage of slow vehicles, and the availability of overtaking opportunities along the route section . Speed reductions of 40 km/hr (24 mph) are required and should exist along the full length of grade to justify climbing lanes . Passing bays are considered in Australia when truck speeds are reduced and the following circumstances exist : grades of over 8%, a high proportion of heavy vehicles, low overall traffic volumes, and high construction costs (5-26) .

Based on Delay or Platooning

Delays are experienced on the two-lane highway when drivers are prevented from traveling at their desired speeds by slower vehicles, turning traffic, highway alignment, or roadside development . Delays result in lost time, vehicle conflicts, and driver frustration. Even if time delays are relatively small, driver perception magnifies the problem when speed changes are frequent or when a large portion of travel time is spent following other vehicles . Delay problems are usually caused by a lack of passing opportunities, and may be localized at a grade, curve, or intersection, or extended over a section of

30

highway . Delays are greater in a traffic stream of mixed vehicles or trip purposes (5-7) . Emoto and May found that the percent of delay time for travel on a

highway without a passing lane was on the order of 65 to 90% . The delay was reduced to a range of 30 to 45% when the highway contained a passing lane (C) . Using a computer simulation model (TWOPAS), Harwood, St . John, and Warren

concluded that passing lanes decrease the percentage of vehicles platooned, increase the rate of passing maneuvers, and have a small effect on mean travel speeds . The percent time delay reductions for various passing lane lengths and flow rates are summarized in Table 5 .05 (5-7) .

TABLE 5 .05 --Reduction in percent time delay per unit length of passing lane

One-Way Flow Rate Passing Lane Length (mi)* (veh/hr)

0.25 0.50 0.75 1.00 1.50 2.00

100 2.8 8.2 8.1 8.1 6.8 6.2
200 11 .1 13.1 14 .0 11.7 10.6 9.5
400 2 .8 8.2 13 .1 9.0 8.1 9.5
700 2.8 8.2 8.1 9.0 8.7 9.0

Note : * Unit length of passing lanes increased by 600 ft to account for cost
of constructing lane addition and lane drop tapers .
Source : D .W. Harwood, C .J . Hoban. Low-Cost Methods for Improving Traffic
Operations on Two-Lane Roads : Informational Guide, FHWA-IP-87-2, 1987 . FHWA,
Washington, D .C .

A comprehensive study of the operational and safety effectiveness of overtaking lanes in the USA is reported by Harwood, et al . In the study, traffic speeds, platooning, and overtaking rates were observed on 12 passing lane sections and 3 short four-lane sections, representing 18 directional overtaking lanes in all . Passing lanes were determined to decrease vehicle platooning and provide various other benefits . Averaged over all sites, the proportions of following vehicles were 35% at entry, 29% at exit, and 32% about 1 .6 km downstream from the exit of the overtaking lane (5-9) .

Based on Level-of-Service

AASHTO states that climbing lane construction is justified if LOS E or F exists on the grade or if a LOS reduction of two or more is experienced moving from the approach segment to the grade (5-20) . The two main factors causing a drop in LOS are delays and hazards (5-13) .

Polus suggests that a climbing lane should be justified if the directional hourly volume (DHV) exceeds the specific grade service volume for

a LOS one level lower than that adopted for the design in level terrain (530) .

Research by Messer (5-31) has found vehicle platooning to be more sensitive to traffic flow rate than mean speed . Messer proposes that the percentage of time spent following in platoons should be the primary criterion for defining level-of-service on two-lane highways in the 1985 revision of the Highway Capacity Manual . Table 5.06 illustrates the criteria for levels-ofservice, as defined by the 1985 Highway Capacity Manual (5-32) .

TABLE 5 .06 --Level-of-service criteria for two-lane highways

Level of Percent time delay Average upgrade speed service on general segments (mi/hr) on specific grades

A s 30 a55

• s45 s50 C s60 a45

• s75 a40

• > 75 a25 -40 F 100 < 25 -40

Source : Transportation Research Board, Highway Capacity Manual, TRB Special

Report 209, 1985 .

In 1984, Hoban reviewed a number of studies dealing with various aspects of level-of-service on two-lane roads . Hoban noted that, in addition to inaccurate or obsolete values in the existing Highway Capacity Manual, there were four major conceptual problems in the two-lane road analysis procedures . These were :

1. 1 . reliance on capacity to establish service volumes ;

2. 2 . the vague definition of operating speed;

3. 3 . inability to take account of driver expectations ; and

4. 4 . the assumption of spatial equilibrium, so that upstream and downstream
   effects of particular road features (e .g . auxiliary lanes) cannot be
   included.
   

Hoban also noted two problems in the use of a volume/capacity ratio for

determining level-of-service . First, the majority of two-lane roads operate

well below capacity, and high-capacity events often occur only when influenced

by specific bottlenecks . Bottlenecking has little effect on overall traffic

operations at lower volumes . The second difficulty is that a volume/capacity

ratio is difficult to visualize, and cannot be related to any directly-observed traffic characteristics . Capacity is thus difficult to measure, and not very relevant to two-lane rural highway design (5-33) .

Based on Economics

When a large amount of interaction between slow and fast vehicles occur, it is more cost effective to provide passing and climbing lanes than to reconstruct the existing highway. Improvement strategies that consider passing lanes are economical alternatives to complete road reconstruction (57) . A primary objective in choosing the location for a passing lane should be minimizing construction costs. Depending upon terrain, the typical cost of constructing a passing lane can vary from $200,000 to $750,000 per mile (mi) . A 1987 publication stated that climbing lanes in mountainous terrain can cost up to $1,800,000 per mile (5-7) .

Khan noted that passing lanes, despite their substantial capital cost, appear to be cost effective and economically feasible for higher traffic volume levels . In general, passing lanes become economically feasible for DHVs higher than 500 (5-34) .

Botha found the most cost-effective strategy for improving traffic operations on various upgrade sections which have similar traffic composition is to provide a minimum length climbing lane at the midpoint of each grade . Botha also stated that cost effectiveness is more sensitive to gradient and least sensitive to truck proportion (5-35) . Rooney also found that if all other factors are constant, then the most cost-effective location for constructing one auxiliary lane is at the halfway distance of the highway section (5-14) .

Staba determined a combination of shorter passing lanes is generally

better than one long passing lane . Shorter passing lanes are most cost

effective when considering the operational effectiveness because the overtaking rates per length of lane are higher (5-13) .

In Hoban's study, effectiveness was defined as the percent travel time saved relative to the basic two-lane highway case . This percent is divided by the provided length of the four lane section and its construction cost in order to determine the cost effectiveness of the auxiliary lane . In consecutive passing lane sections, Hoban found that the first overtaking lane section is most cost effective, while for each additional section of an overtaking lane, the cost effectiveness decreases (5-36) .

Kaub and Berg studied the conditions under which the construction of auxiliary passing lanes on two-lane highways are economically justified based on a passing conflict model . The passing conflict model estimates the number of passing conflicts that will occur on a two-lane highway . A benefit-cost model and nomograph for the critical average daily traffic (ADT) were created to economically analyze the use of passing lanes . The nomograph was created as a function of construction cost of passing lane per mile, percent of existing passing, cost of conflicts, and total passing lane length . It is used to calculate the ADT that must be exceeded on the two-lane highway to economically justify a passing lane . This nomograph can also be used when the current ADT is known to determine the required length of the passing lane (5

6) .

LITERATURE REVIEW: WHERE TO INSTALL PASSING OR CLIMBING LANES

Those who have studied passing/climbing lanes operations have suggested optimum locations for installation. They have also identified less-thandesirable locations .

Desirable Locations

The location chosen for a passing lane should be logical to the driver . Passing lanes should be located where passing sight distance is already restricted, not on long tangent sections where passing opportunities are already adequate . The passing lanes can be located on a sustained grade or on a level section, depending on where the delays occur (5-37) . McLean found the most effective locations for auxiliary lanes are where a high level of platooning at the lane entry exists, where platooning would continue to increase in the absence of an overtaking lane, and where there is considerable variation in vehicle free speeds to enable the achievement of a high overtaking rate on upgrades . McLean also stated that a climbing lane located at the midsection of the grade provides the best balance of platooning before and after the climbing lane (5-10) . Rooney states that the construction of a passing lane in downhill sections of certain highway segments produce safety benefits (5-14) .

Harwood et al . noted climbing lanes should usually begin before speeds are reduced to unacceptable levels and should continue over the vertical curve crest so that vehicles may gain speed before merging. Research showed that single short climbing lanes of 460 m (1,500 ft) near the midpoint of the grade, or two sections at the one-third and two-third points on the grade are cost-effective methods for providing passing on sustained grades (5-37) .

Undesirable Locations

McLean stated that climbing lanes are underutilized if placed early on the grade because of relatively low level of platooning at the lane entry. As a result, a high level of platooning occurs after the end of the climbing lane . If a climbing lane is located late on the grade, it has a high level of utilization, but the traffic has operated with a relatively high level of platooning over the grade midsection (5-10) .

McLean further found that auxiliary lanes should not be located where they are a disadvantage to opposing traffic . An auxiliary lane placed in a zone where adequate sight distance is already provided can reduce the overtaking opportunity of the opposing traffic stream (if the auxiliary lane section is striped with double yellow), which can create a platooning problem in that direction (5-10) . Auxiliary lanes should not be placed where turning movements are frequent, nor where physical constraints restrict the construction of a continuous shoulder (5-37) . Bridges and culverts should be avoided if they restrict the provision of a continuous shoulder (5-7) . Passing zones for opposing traffic should be prohibited where passing lanes with high-volume intersections and driveways are present (5-37) . However, low-volume intersections and driveways do not usually create problems in passing lanes . Where higher-volume intersections cannot be avoided, special provisions for turning vehicles should be considered . The prohibition of passing by vehicles traveling in the opposing direction should also be considered on passing lane sections with higher-volume intersections and driveways (5-7) .

In some cases, a passing lane on a long tangent may encourage slow drivers to speed up, thus reducing the passing lane effectiveness . At the other extreme, highway sections with low speed curves should be avoided, since

they may not be suitable for passing (5-7) . The construction of a passing lane should be avoided in sections with low design speeds relative to the rest of the considered stretch of the highway (5-14) .

LITERATURE REVIEW: LENGTH AND SPACING OF PASSING AND CLIMBING LANES

The American Association of State Highway and Transportation Officials (AASHTO) states that auxiliary, passing, and climbing lanes should be considered where the achievement of adequate passing sight distance is not

practical . Passing lanes should be considered where the roadway can be expanded to four lanes in the future . These lanes should be to the right of the normal traffic lane and 3 .6 m (12 ft) wide or as wide as the through lanes . Highway sections with these auxiliary lanes should also include a minimum 1 .2 m (4 ft) shoulder (5-20) .

Passing lanes should be provided at regular intervals to reduce delays caused by inadequate passing opportunities over lengthy sections of highway. Passing lanes should be placed where a minimum 305 m (1,000 ft) sight distance exists at the approach and lane drop-off tapers . The minimum length of a passing lane should be 305 m (1,000 ft), although passing lanes of 0 .4 km

(0 .25 mi) or less in length are not very effective in reducing traffic platooning . The optimal length of a passing lane to reduce platooning is 0 .8 to 1.6km (0.5to 1.0mi) (5-7) .

Rooney noted that the maximum distance between diverge and merge of auxiliary lanes should not be so excessive that not much passing per mile of the lane length would occur. The distance between diverge and merge varies depending on grade and traffic volume (5-14) . Khan noted that longer passing lanes have a somewhat higher effect on reduction of platooning than shorter lanes . However, when viewed in term of benefits, a 1 to 2 km lane is more effective than a 3 km lane . Khan created effective passing lane lengths and effective distances for given ranges of traffic volume shown in Table 5 .07 (534) . Emoto et al . also developed a table of optimum passing lane length as a function of traffic volume (Table 5 .08) (5-29) . Botha and May stated that the most cost-effective climbing lane is 1,500 feet located at the midpoint of the uphill grade (5-35) .

The rural traffic simulation model TRARR was developed by the Australian Road Research Board (ARRB) to evaluate various road improvement options on specific two-lane road sections . The three-12 km road segments modeled by TRARR varied from an ideal straight and level road to a segment of rural highway with low geometric standards and few opportunities for overtaking due to lack of sight distance and frequent, steep grades . Among the road improvement options studied were auxiliary lanes, widening to four lanes, and reconstruction on an improved alignment . A benefit-cost analysis was also conducted on the road improvement options for a range of average annual daily traffic volumes . The results from the TRARR simulation model suggest that short auxiliary lanes offer a low-cost road improvement option that can be warranted at low traffic volumes (5-38) .

TABLE 5 .07 --Optimal passing lane length and effective distance for given traffic volume

Traffic volume Optimal passing lane length Effective distance (DHV) (miles) (miles)

500-750 1 .0-1 .6 13 800-1000 2 .0 10 Over 1000 2 .0 8

Source: A.M. Khan, et al . Cost-effectivenessofPassingLanes:Safety,Level of Service, and Cost Factors, 1991 . The Research and Development Branch, Ontario Ministry of Transportation, Ontario .

TABLE 5 .08 --Optimal passing lane length for given traffic flow rates

One-way flow rate Optimal passing lane length (vph) (miles)

100 0.5 200 0 .5-0 .75 400 0 .75-1 .0

700 1 .0-2 .0

Source : A.D. May, T.C. Emoto . Operational Evaluation of Passing Lanes in Level Terrain : Final Report, UCB-ITS-RR-88-13, 1988 . Institute of Transportation Studies, University of California, Berkeley, California .

The location of the climbing lane is typically based on the 10 mph speed reduction criteria . The lane should begin where the truck speed drops to 10 mph below other traffic speed for a given truck weight/horsepower ratio. A

25 :1 approach taper, or minimum 150 ft, should precede the climbing lane . The lane drop-off taper length measured in feet should be the lane width in feet times the design speed in miles per hour. The approach taper length should be one half to two-thirds the length of the drop-off taper (5-20) . It is desirable for the lane to extend beyond the crest of the hill, where the speed differential is 10 mph, or where an LOS D is reached . However, it is most practical to end a climbing lane where a truck can return to the normal lane without undue hazard. This is preferably an additional 200 feet beyond the point of sufficient sight distance . Finally, a taper of 50 :1 or a minimum 200 feet should be provided at the end of the climbing lane (5-20) .

Auxiliary lanes should have sufficient length to disperse most slow-

leader platoons and reduce delay . Early practice was a minimum design length of 1 .5 kilometers . Current minimum length is set as 400 meters (5-10) . AASHTO guidelines state a minimum of 1,000 feet should be used with typical

lengths of 0 .5 to 1 .0 miles (5-20) . In a sensitivity analysis by Staba and May, 0 .25-0 .75 miles appears to be most beneficial length in California from a perspective of cost effectiveness, number of passes per length of passing

lane, and travel time saved per mile of passing lane (5-13) .

Australian auxiliary lane lengths are dependent on road design speed. For example, the design guidelines recommend lengths of 600 to 1200 m for a design speed of 100 km/hr, and 400 to 850 m for a design speed of 80 km/hr. The recommended length for overtaking lanes in Canada is dependent upon provincial highway regulations, but is generally 2 km. Table 5 .09 presents a comparison of geometric design features of overtaking lanes for Canada and Australia (5-1) .

TABLE 5 .09 --Passing lanes design features for Canada and Australia

Highway Length Taper Length Lane Width Shoulder Width agency (m) Diverge Merge

Ontario 1,500 -200 m 200 m Desired min : Equal to approach 2,000 3 .4 m shoulder; min . lm

B. C. Min :800 20:1 25:1 3.6 m 1.8 m Desired Min : 1,000 Alberta 2,000 25 :1 50 :1 3 .5 m 1 .5 m sealed (exc . tapers) Parks 2,000 100 m 200 m 3 .65 m 1 .2 m seal plus (inc . tapers) gravel

Australia Function of VW/3* VW/2* 3 .5 m Min: 1 m design speed; normal max: 1,200

Note : * S = speed limit (km/h) ; V = 85th percentile approach speed ; W = lane

width or amount of pavement widening

Source : J .F . Morrall, C .J . Hoban. "Design Guidelines for Overtaking Lanes,"

Traffic Engineering and Control . Vol 26, no.10, 1985 . pp 476-484 .

Rural highway design in Alberta is based on the objective of providing at least 50% net passing opportunities in both directions of travel in the peak period traffic during the design year . Where traffic volumes are low and available sight distances adequate to achieve this level, a two-lane highway will suffice . Where 50% net passing opportunities are not available in the design year, additional passing lanes or four-lane sections should be added. Passing lane and four-lane sections provide 100% net passing opportunities throughout their length and can raise the net passing opportunities over an extended section of highway to the level desired for design. It should be noted that most of Alberta contains relatively level terrain and good sight distance . Highway agencies in more restrictive terrain may find it economically infeasible to provide 50% net passing opportunities (5-7) .

Auxiliary lanes should be spaced according to the desired reduction in traffic platooning or the desired increase in passing opportunities . Normal practice is to initially install consecutive passing lanes in the same direction at a spacing of 10 miles or more (5-7) . As traffic volume and the need for passing opportunities increase, additional passing lanes are added, resulting in a spacing of 2 to 5 miles (5-13) .

LITERATURE REVIEW: HOW TO CONDUCT AN ANALYSIS OF PASSING/CLIMBING LANE NEED

Field studies of traffic platooning give the engineer useful local information on passing problems and needs . For a length of road, spot platooning is much easier to measure in the field than time delay over a length of road .

Field studies of platooning are performed by manually calculating the percentage of vehicles traveling with headways less than a specified time. Depending upon the desired precision, the intervals between vehicles can be timed with a stopwatch or estimated visually. Some commercially available traffic counting equipment can also count both the total traffic volume and the number of vehicles following at small headways (5-7) .

Spot studies of traffic platooning performed at several locations can be averaged to estimate the overall level-of-service (i .e ., percent time delayed), while locations of higher platooning levels can identify potential improvement sites and improvement needs . In this assessment, the following points may be noted : What Data Should be Collected

An extensive study from valley to crest of a mountain range must be performed to optimize the effectiveness of an auxiliary lane . Data should include speed profiles for average trucks, the approximate maximum safe speeds of automobiles, locations where there is passing sight distance greater than 1,500 feet and where passing is legal, an approximate elevation profile, locations where there are two or more lanes per direction, and annual average daily traffic (5-14) .

Defining Headway

Different opinions exist concerning the headway time used to define delay. Spot platooning or percent following is defined by the 1985 Highway Capacity Manual as the percentage of vehicles with headways (or time gaps) of 5 seconds (sec) or less as they pass a given point (5-7) . Morrall (5-31) used

a platoon definition based on a 6 sec headway, and Hoban (5-39) recommended a 4-sec headway criterion . Guell and Virkler claim that using 5 sec as the definition for being delayed produces an inconsistency in the level-of-service between certain general terrain segments and specific grades . Guell and Virkler further state that the selection of 3 .5 to 4 .0 seconds as the definition for delay would alleviate this problem. For example, at 60 mph the head-to-head spacing between two vehicles traveling at 5-seconds headway is

440 feet . This represents a per lane density of 12 vehicles per mile, the

value associated with level-of-service A on freeways . Therefore, a level-of

service lower than LOS A would demand that the headway be less than the delay

definition of 5 seconds . At 4 and 3 .5 seconds, the spacings are 352 and 308

feet, and the per lane densities are 15 and 17 vehicles per mile. Both of

these conditions are associated with LOS B on a freeway (5-40) .

Study Segment Length

Passing lane analysis should be based on an evaluation of a significant length of road (say 10 to 20 km) to take full account of upstream and downstream effects . The effect, for example, of a four-lane section immediately upstream or downstream on the amount of bunching can significantly influence the effectiveness, of a passing lane (5-1) .

LITERATURE REVIEW: OBSERVED BEHAVIORS OF QUEUED AND PASSING VEHICLES

McLean states that there are two main methods for studying driver

behavior . One method employs a limited number of experimental-subject drivers

and studies their behavior in detail, providing much information on a few

drivers . However, uncertainty arises as to how the information relates to the

total driver population. The second method relies on observing a large sample

of real drivers in the field, providing a little information on a large

sample . Similarly, uncertainties arise as to how the information relates to

individual behavior . This issue can be resolved by combining the ^behavioral

and traffic analysis approach to obtaining driver behavior data . ^Overtaking

gap-acceptance data for traffic analysis have traditionally been collected by driving an observation vehicle in a traffic stream, and recording gap-

acceptance data from drivers who overtake it. This can easily be supplemented

by subject-driver studies providing information on variance in gap-acceptance

behavior for individual subjects . The combination of observation-vehicle and

subject-driver studies would enable a probability model for gap-acceptance

behavior covering both sources of variance (5-10) .

REFERENCES

5-1.J.F.MorrallandC.J .Hoban. "DesignGuidelinesforOvertakingLanes," Traffic Engineering and Control, Vol . 26, no. 10, pp . 476-484, 1985 . 5-2 . Highway Statistics, 1984 . U.S . Department of Transportation, Washington, DC . 5-3 . Fatal & Injury Accident Rates, 1984 . U.S. Department of Transportation, Washington, DC . 5-4 . Fatal Accident Reporting System, 1984 . U.S. Department of Transportation, Washington, DC. 5-5 . The Magnitude and Severity of Passing Accidents on Two-Lane Rural Roads, FHWA-RD-94-068, November 1994 . Federal Highway Administration, McLean, Va.

5-6 . Alan R. Kaub and William D . Berg. "Design Guide for Auxiliary Passing Lanes on Rural Two-Lane Highways," Transportation Research Record 1195, pp. 92-100, 1988 . Transportation Research Board, Washington, D.C.

5-7 . D.W . Harwood and C.J. Hoban . Low-Cost Methods for Improving Traffic Operations on Two-Lane Roads: Informational Guide, Report No. FHWA-IP87-2, 1987 . Federal Highway Administration, Washington, D.C.

5-8 . E .A. Rinde . "Accident Rates vs . Shoulder Widths : Two-lane Roads, Two-lane Roads with Passing Lanes," Report No . CA-DOT-TR-3147-1-77-01, 1977 . California Department of Transportation . 5-9 . D.W. Harwood, A.D . St . John, and D.L . Warren. "Operational and Safety Effectiveness of Passing Lanes on Two-Lane Highways," Transportation Research Record 1026, pp . 31-39, 1985 . Transportation Research Board, Washington, D .C . 5-10 . John R. McLean. Two-Lane Highway Traffic Operations : Theory and Practice, pp . 349-374, 1989 . Gordon and Breach Science Publishers, New York. 5-11 . W .S . Homburger. "An Analysis of Safety at Upgrade Terminals on Climbing Lanes on Two-Lane Highways," Transportation Research Record 1122, pp . 27-36, 1987 . Transportation Research Board, Washington, D.C. 5-12 . A. Werner and J.F . Morall. "A Unified Traffic Flow Theory Model for Two-lane Rural Highways," Transportation Forum, Vol 1, no 3, pp. 79-87,

1985 . 5-13 . Gail R . Staba, Adolf D . May, and Hao 0. Phung . Development of Comprehensive Passing Lane Guidelines Volume I: Final Report, UCB-ITSRR91-1, 1991 . University of California, Berkeley, California .

40

5-14 . F .D . Rooney. Speeds and Capacities on Grades, Climbing Lanes, Passing Lanes, and Runaway Trucks, 1985 . CALTRANS, Division of Transportation Operations . 5-15 . Manual on Uniform Traffic Control Devices, 1988 . Federal Highway Administration, Washington, D .C .

5-16. D.W. Harwood and A.D. St. John . PassingLanesandOther Operational Improvements on Two-Lane Highways, FHWA/RD-85/028, pp. 11-54, 1985 . Federal Highway Administration, Washington, D .C .

5-17. D.W. Harwood, et al . Truck Characteristics forUseinHighwayDesign and Operation, Volume I, Report No. FHWA-RD-89-226, 1990. Federal Highway Administration, Washington D.C .

5-18. J.C. Glennon. "A New and Improved Model of Passing Sight Distance on Two-Lane Highways," presented at the 67th Annual Meeting of the Transportation Research Board, January 1988 .

5-19. J.R. Jones. "An Evaluation of the Safety and Utilization of Short Passing Sections," M .S . Thesis, Texas A & M University, 1970 .

5-20 . A Policy on Geometric Design of Highways and Streets, pp 240-250, 499501, 1990 . American Association of State Highway and Transportation Officials, Washington D.C.

5-21 . National Institute for Transport and Road Research. TRH 17: Geometric Design of Rural Roads. 1985. CSIR, Pretoria, South Africa . 5-22 . Botswana Road Design Manual, 1982 . Botswana Ministry of Works, Gaberone .

5-23. J.C. Sananez and A.D. May. "Development and Application of a Macroscopic Model for Rural Highways," Transportation Research Record 971, pp. 49-58. Transportation Research Board, Washington, D.C.

5-24 . M . Firestine, et al . New Methods for Determining Requirements for Truck-Climbing Lanes, FHWA-IP-89-022, 1989 . Federal Highway Administration, Washington D .C .

5-25 . P.Y. Ching. Truck Speeds on Grades in California, FHWA-CA-TO-79-1, 1979 . Federal Highway Administration, Washington D .C.

5-26 . Guide to the Geometric Design of Rural Roads, 1985 . National Association of Australian State Road Authorities, Canberra .

5-27 . United Kingdom Department of Transport . TD 9/81, 1981 . Highway Link Design London, London, England .

5-28. K.M. Wolhuter and A. Polus. "Uniform Delay Approach to Warrants for Climbing Lanes," Transportation Research Record 1195, pp. 101-110, 1988 . Transportation Research Board, Washington, D .C.

5-29. A.D. May and T.C. Emoto . Operational Evaluation of Passing Lanes in Level Terrain: Final Report, UCB-ITS-RR-88-13, 1988. Institute of Transportation Studies, University of California, Berkeley, California .

5-30. A. Polus, et al. "Applying the Level-of-Service Concept to Climbing Lanes," Transportation Research Record 806. Transportation Research Board, Washington, D .C .

41

5-31 . C .J . Messer and J .F . Morrall . Two-Lane, Two-Way Rural Highway Capacity. Final Report, NCHRP Project 3-28A, 1983 . Texas Transportation Institute, College Station, TX.

5-32 . Highway Capacity Manual, Special Report 209, 1985 . Transportation Research Board, Washington, D.C.

5-33 . C .J . Hoban . "Measuring Quality of Service on Two-lane Rural Roads ." Proceedings, 12th ARRB Conference, Vol . 12, no. 5, pp . 117-130, 1984 .

5-34. A.M.Khan,etal . Cost-effectivenessofPassing Lanes:Safety,Levelof Service, and Cost Factors, 1991 . The Research and Development Branch, Ontario Ministry of Transportation, Ontario.

5-35 . J.L . Botha and A.D . May. "A Decision-Making Framework for the Evaluation of Climbing Lanes on Two-lane Two-way Rural Roads," Proceedings of the 8th International Symposium on Transportation and Traffic Theory, pp . 91-120, 1981 . University of Toronto Press, Toronto, Canada .

5-36 . C.J. Hoban. Overtaking Lanes on Two-Lane Rural Highways, Ph.D. thesis . 1980 . Monash University, Melbourne, Australia .

5-37. Douglas W. Harwood, Chris J. Hoban, and Davey L. Warren. "Effective Use of Passing Lanes on Two-Lane Highways," Transportation Research Record 1195, pp. 79-91, 1988. Transportation Research Board, Washington, D.C. 5-38. C.J. Hoban. "Simulation Study of Guidelines for Rural Road Improvements," Transportation Research Record 971, pp . 93-95 . Transportation Research Board, Washington, D .C .

5-39 . C .J . Hoban . "Towards a Review of the Concept of Level of Service for Two-Lane Roads, Australian Road Research, Vol . 13, no . 3, pp . 216-218, 1983

5-40. D.L. Guell and M.R. Virkler. "Capacity Analysis of Two-Lane Highways," Transportation Research Record 1194, pp . 199-203 . 1988 . Transportation Research Board, Washington, D .C.

this page left blank

CHAPTER 6
SIMULATION OF PASSING AND CLIMBING LANES

Approximately 80 percent of the national highway system and 97 percent of the total rural system is composed of two-lane highways (6-1) . Over two-thirds of the two-lane mileage is in mountainous or rolling terrain (6-2) . This means it would be desirable for a high proportion of rural roadways to offer drivers the chance to pass, but terrain may often limit passing availability.

The 1985 Highway Capacity Manual (HCM) Chapter 8 (6-4) contains procedures for the analysis of capacity and level of service of two-lane highway operations . However, many users of the manual feel that the chapter lacks important information and that some existing procedures need to be revised. The Two-Lane Roads Subcommittee of the Transportation Research Board Committee on Highway Capacity and Quality of Service conducted a review of both user suggestions and the existing procedures in the HCM (6-3) . The Subcommittee found several areas in need of improvement .

The focus of this chapter of the report is assessing two simulation packages, TRARR and TWOPAS, associated with passing and climbing lane operations, and assessing their ability to evaluate auxiliary lanes

.

Furthermore, this chapter includes a comparison of the basic traffic models given in Chapter 8 (chapter on two lane highways) of the HCM with the two

software packages, gives a brief background of both simulation models,

addresses the ease of use of the two models, discusses limitations encountered during the evaluation, and makes recommendations based on those limitations . Additionally, data collected from the old U. S. Highway 412 was used to

simulate a test case with both software packages .

BACKGROUND OF THE THREE MODELS

Highway Capacity Manual Background

The Transportation Research Board (TRB) does not endorse any specific

software packages, but it does recognize the "Highway Capacity Software" (HCS)

as a widely used computerized implementation of the procedures in the current version of the Highway Capacity Manual . This software was originally developed by the Federal Highway Administration and is currently maintained by _the Center for Microcomputers in Transportation at the University of Florida.

TRARR Background

TRARR is a microscopic, stochastic simulation model of the traffic

_associated with two-lane highway systems (6-5) . Developed by Dr. Geoff

Robinson for the Australian Road Research Board in the late 1970's, the

current model in use is the 1994 version TRARR 4 . The name TRARR is an

_acronym for TRAffic on Rural Roads . The model is written in FORTRAN and can

_{be run on an} IBM compatible system running DOS or on a mainframe computer using FORTRAN 77 . Since use of the current version is significantly modified

from the 1991 version 3 .2, this report will include concepts from each version .

The model is used to study the effects of changes to rural two-lane road design and variation in traffic characteristics (6-5) . It is also capable of simulating auxiliary lanes on either side of the road (such as passing lanes) . By controlling certain variables and modifying others, one can investigate the behavior of traffic on existing highway systems or use the program to assist in design of future highways .

TWOPAS Background

TWOPAS, like TRARR, is a microscopic model for simulation of rural traffic systems (6-4) . Originally known as TWOWAF, it was developed by the Midwest Research Institute (MRI) during the early 1970s . The initial version was designed to run on a Control Data Corporation (CDC) computer and was later modified to be compatible with IBM operating systems . TWOPAS incorporates modifications and additions made in the National Cooperative Highway Research Program (NCHRP) Project 3-28A.

TWOPAS has the ability to simulate both two-lane highways and two-lane highways with added passing or climbing lanes . The model simulates traffic by reviewing the position, speed, and acceleration of each vehicle at one second intervals on a simulated roadway. The program uses geometrics (design features) of the road, traffic controls (passing zones and speed limits), vehicle characteristics, driver characteristics, and information obtained from entering traffic to simulate traffic operations .

HIGHWAY CAPACITY MANUAL APPROACH

The 1994 HCM presents ideal conditions for two-lane highways (6-4) . A two-lane highway containing no restrictive geometric, traffic, or environmental conditions is considered ideal . The conditions are listed as :

Under these ideal conditions, the total capacity in both directions of two-lane rural highways is 2,800 passenger cars per hour (pcph) . Using this ideal capacity, the 1985 HCM presents basic models for rural highways . One such model describes the general relationship between the service flow rate and different adjustment factors for a given terrain segment . The model is presented as :

SF, = 2800 (v/c) ; fd fR, _fby where : SF, = total service flow rate (vehicles per hour-vph) in both directions for a

LOS; (level of service), where the level of service is defined as a qualitative measure describing operational condition within a traffic stream

2800 = ideal capacity flow rate (vph) (v/c) ; = ratio of vehicle flow rate to ideal capacity for the LOS; ranging from 0 to 1 (obtained from Table 8-1 in the HCM) fd = directional distribution adjustment factor ranging from 0 to 1 (HCM Table 8-4) fq, = adjustment factor for narrow lanes and restricted shoulder width ranging from 0 to 1 (Table 8-5 in the HCM) fhv = adjustment factor (ranging from 0 to 1) for the heavy vehicles in the traffic stream. This factor is obtained from the formula:

fhv = 1 / [1 + PT(ET -1) + PR(ER-1) + PB(EB-1) ]

where PT, PR, and PB are the respective fractions of trucks, RV's, and buses in the traffic stream. ET , ER , and EB (obtained from Table 8-6 in the HCM) are the respective passenger-car equivalents for trucks, RV's, and buses .

It is not clear how TRARR relates to the HCM models . TWOPAS, however,

was used to generate the basic level of service and capacity analysis in the 1994 HCM (6-8) . From the input values required for both software packages, it is obvious that some aspects of the HCM models are incorporated into their logic . For example, both simulation models, TRARR and TWOPAS, require user-specified values for directional flow rates and vehicle mix (giving a fraction of heavy vehicles), but neither model lets the user directly input values dealing with narrow lanes or shoulder width . TWOPAS utilizes the user-specified distribution of desired speeds to indirectly characterize the lane width, shoulder width, and pavement quality (6-5) . Version 3 .2 of TRARR (6-6) uses a road speed index to reflect reductions in speed due to horizontal curves, narrow pavements, speed limits or other road characteristics (6-6) . Version 4 of TRARR (6-7) uses a steady state speed model adopted from "Vehicle Speeds and Operating Costs" (6-9) that incorporates road roughness along with curve radius and gradient .

Both simulation models also allow the user to input regions of percent gradient of a road segment . The HCM points out that since passenger-car and heavy vehicle operations are affected by gradient, adjustments need to be made to the v/c ratio (described earlier) to reflect the hindrance . The adjustment for the operation of passenger-cars is made by the equation :

fg= 1 / [1 + (PPIP)I where :

P, = fraction of passenger cars in the upgrade traffic stream Ip = impedance factor for passenger cars, calculated by Ip = 0.02 (E-E(,)

E = the base passenger-car equivalent for a given percent grade, length of

grade, and speed, and E0 = the base passenger-car for 0% grade and a given speed (obtained from

Table 8-9 in the HCM) .

Passing sight distance critically impacts capacity and service flow rate on basic two-lane highways . The 1985 HCM states "the opportunity to pass, given a constant volume, is a function of the availability of passing sight distance" (6-4) . Both TRARR and TWOPAS require the user to directly input sight distance at certain points in each direction along the simulated road. The programs then calculate sight distance by linearly interpolating from the sight distance specified at the beginning of a region to the next point where sight distance is specified .

Other than the basic vehicle flow models, Chapter 8 of the HCM does not include any further equations related to rural two-lane highways . It does include a discussion of the design and operational treatment of rural highway systems . This discussion is strictly subjective, however, and does not include any models related to passing sight distance, auxiliary lanes, or other design characteristics . Currently, there is a request for a proposal from the National Cooperative Highway Research Program (NCHRP) that addresses the improvement of methods and procedures for capacity and quality-of-service analysis of two-lane highways (6-10) . One of the tasks described in the request involves the enhancement and calibration of a selected simulation package .

TRARR USE OF SOFTWARE Input Requirements

The input requirements for TRARR 4 and TRARR 3 .2 are significantly different . Version 3 .2 of TRARR requires the creation of extensive files involving traffic information, vehicle and driver characteristics, road design, and observing requirements whereas version 4 of TRARR uses a set of default data files with which the user makes simple modifications. TRARR 4 is run from a control menu, but if detailed or complex changes are needed for a simulation, the control menu cannot be utilized and TRARR 3 .2 must be used (66) .

With TRARR 4, modifications of the road and traffic files are all that are needed to run the simulation . To modify or create a new road file, the user chooses a file from the control menu and inputs the necessary parameters . These parameters include the length of road to be simulated, barrier lines in both directions of travel, the presence or absence of auxiliary lanes, sight distance at different points along the simulated road, gradient (only needed for direction 1), 85th percentile speed (which is calculated by the model using the other parameters), and curve radius at certain points along the road. TRARR 4 automatically sets up a data file that takes these observations at 100 meter intervals . It is important to realize that the road file does not build an exact replica of the road being simulated. Instead, an approximation of the design is created by entering values at certain intervals along the road segment. Since TRARR and TWOPAS both operate on a time based simulation rather than an event based simulation, this has no great effect on the results .

To modify a traffic file in TRARR 4, the user selects a file from the control menu and makes the necessary changes to the flow rate, the percent platooning (called bunching in the TRARR 3 .2 and TRARR 4 manuals), and the percentage of trucks in either direction (6-6) . Simulation time, random number seed, and warmup time are also specified at this point .

After modifying the road and traffic files, the execute option is selected from the control menu and the simulation is run. Graphical interpretation of the results are viewed from the control menu, but the actual output file must be viewed from a text editor .

Description of Output

Version 4 of TRARR saves the output to a file named OUT1, OUT2, etc . These files may be viewed from a text editor . The output begins by summarizing the input parameters from the traffic file . It then lists point observations which include the number of passes between successive points, the mean speed (observed) between those points, standard deviation of speeds,

percent following during successive road segments, and the number of vehicles between the two points at a given time . A summary is then given for each direction of travel that includes mean speed and standard deviation by user-specified category (cars and trucks), the mean travel time over a given distance (journey speed) and standard deviation, the percentage of time those categories spent in platoons, the number of passes accomplished by each category, and fuel consumption data for each category. The output then combines both directions of travel and presents similar results . Finally, the output presents free speed (vehicles traveling without interacting with other vehicles) distributions for each user-specified category.

TWOPAS USE OF SOFTWARE Input Requirements

The input data for TWOPAS is entered into a file from a text editor (64) . The format of the model uses a unidirectional coordinate system expressed in feet from point 0 to the end of the road. For example, the direction 2 input values do not begin with point 0 on the simulated road, rather, they

begin with the last point (in feet) encountered in direction 1 of the road segment . The data is entered in what the TWOPAS User's guide calls a "data deck." This deck contains "cards" (or fields) presenting specific types of information. The deck consists of the following items :

D. First Card of the next deck . When multiple simulations are to be run, the first card of the next deck is entered in order to let the program know to continue .

Description of Output The output consists of several pages presenting the information gathered during the simulation. The sections are as follows :

1. A version of the input data set ;

1. 2 . Summary of user-specified times, speeds, flow rates, and vehicle characteristics;

2. 3 . Summary of the influential road characteristics on traffic operations ;

3. 4 . Summary of the representative desired speeds, travel times, and reference overall speeds;

4. 5 . Snapshot of the status of traffic at the user-specified points in time ;

5. 6 . Summary of space-averaged data and operating speeds ;

6. 7 . Summary of travel times and delays;

7. 8 . Overall speed histograms for both directions ;

8. 9 . Time margins of passes and pass aborts ;

9. 10 . Summary of passing and pass abort rates, platoon leaders, and percent of time a category of vehicles is unimpeded;

10. 11 . Headway and platooning data ;

11. 12 . Overtaking event summary classified by speed differences ;

12. 13 . Summary of overtaking events classified on initial acceleration and acceleration noise;

13. 14 . Summary of the operations at user-specified locations with both single lanes and then with auxiliary lanes ;

14. 15 . If the user specified subsections on the road, the traffic operations are presented for both single lane and auxiliary lane subsections .

This output, as confusing as it may seem, will be clarified in a later example .

Included with the simulation package is a copy of TWOSUM, a program that condenses the output into one or two pages of summary data. However, the user's manual strongly suggests that the operator review the TWOPAS output for errors before relying totally on the TWOSUM output .

SIMULATION TEST CASE

A sample run, using TRARR and TWOPAS, is presented using data obtained from a road segment on old U . S. Highway 412 between Siloam Springs and Tontitown. The road segment is an 1 .11 mile two-lane highway containing a short passing lane in both directions . Videotaped data gathered at the site of interest shows that the estimated average directional flows are 335 vehicles/hour in the east bound direction and 277 vehicles/hour in the west bound direction. Both TRARR and TWOPAS use the terms "direction 1" and "direction 2" to specify directions of travel . Therefore this report will replace east bound and west bound with direction 1 and 2, respectively. Percent platooning is estimated (from the videotape data) to be 70 percent for direction 1 . The road is level (with small rolling hills) with two curves of fairly large radius . The estimated average of observed passes in direction 1 is approximately 58 per hour . Passing information and percent platooning were not gathered for direction 2, therefore, the same averages were assumed.

Input and output files for TRARR 4 are found in Appendix A and B, respectively . The TWOPAS input file is found in Appendix C. Appendix D and Appendix E contain the output generated by the TWOPAS and TWOSUM programs

. The following sections present brief explanations of each program's input and output .

TRARR

As previously explained, the ways that the user creates the input files for TRARR 3 .2 and TRARR 4 are different . With version 3 .2 of TRARR, the user actually modifies the traffic, drivers' characteristics, road, and observing requirements files in a text editor. With TRARR 4, the user only modifies the road and traffic files . The TRARR 4 user-interface does not display the

actual files to be modified . Instead, the program only requires the input of

information needed to modify certain important values in the default files

. The rest of the input data (not required as input in TRARR 4) is taken from the default values given in the files . Since TRARR 4 was used herein, only the road and traffic input files are presented for the reader .

Input Files The TRARR 4 input data (user-specified and all default values) are found in Appendix A. The first file presented is the road input file The first

. parameter shown is the DSS . This represents the total length of the simulated road. DENDS is the length of road that a simulated vehicles travels before being removed from the simulation (the user's manual explains that this is

50

done to avoid unrealistic end effects in the model) . DUR represents the

interval between each point where data is entered by the user (TRARR 4 uses

100 meters as the default) . The desired 85th percentile speed is shown next,

and is followed by the degree of bendiness (curviness) on the simulated road.

The value and the degree of bendiness and of IRI are not explained in the

user's manual . Following the first line of input data, the road is presented in 100 meter intervals (specified as "chainage") . At each point, the following road characteristics are specified.

1. 1 . Barrier lines for directions 1 and 2 (-1 for no-passing zone, 1 for
   passing zone) .
   

2. 2 . The presence of auxiliary lanes in directions 1 and 2 (True or False) .

3. 3 . Road Speed Indices . This value represents the effects of horizontal curves, narrow pavements, speed limits, or other road characteristics on the basic desired speed of the simulation driver . This value is computed by the program in TRARR 4, but otherwise found on page 32 of the TRARR

4. 3 .2 User's Manual (6-6)) .

5. 4 . Sight distance for each direction.

6. 5 . Percent gradient .

7. 6 . Curve Radius in meters (9999 represents a straight road segment) .

8. 7 . 85th Percentile Speed in km/hr (calculated by TRARR 4) . This data, when saved, is automatically copied by the TRARR 4 interface to a

* .rod file in the TRARR directory.

The second file included in Appendix A is the traffic file . In this file, traffic characteristics are stored. Using the TRARR 4 interface, the user specifies the warm-up (settling down time) for the simulation, time for the simulation, directional flows, percent of passenger cars, and percent platooning in each direction . TRARR 4 takes the rest of the information from default values . User-specified and default values are saved as a * .trf file under the TRARR directory. The traffic generation categories contain the fraction of vehicle types in the simulated traffic stream. There are up to eight streams of vehicle mixes used. Within these streams, different probabilities are specified for each of the eighteen vehicle types . The proportion of traffic in each of the eight categories of vehicle mixes is shown for both directions . Then, the two-directional traffic volume (veh/hr), the mean desired speed (km/hr), the standard deviation of desired speeds

(km/hr), speed distribution indices (described in the user's manual), and the platooning-flow distribution are shown for each of the eight categories .