5/1/2015

1

NCHRP Report 772: Evaluating the

Performance of Corridors with

Roundabouts

Transportation Education

Series

April 28, 2015 – Juneau

April 29, 2015 – Anchorage

April 30, 2015 – Fairbanks

Ph

oto

: Le

e R

od

eg

erd

ts

Contractors

• Kittelson & Associates, Inc.

– Lee Rodegerdts - Principal Investigator

– Brian Ray, Pete Jenior, Zachary Bugg

• Institute for Transportation Research & Education (North

Carolina State University)

– Bastian Schroeder

• Texas Transportation Institute (Texas A&M University)

– Marcus Brewer

• Write Rhetoric

– Danica Rhoades

5/1/2015

2

Project Panel

• Phil Demosthenes, consultant (chair)

• Stephen Bass, Kansas DOT

• Wylie Bearup, City of Phoenix, AZ

• Sarah Bowman, Walkable Communities, Inc.

• Robert Fenton, Ohio State University

• Theron Knause, Virginia DOT

• Mark Lenters, Ourston Roundabout Engineering, Inc.

• Avijit Maji, Maryland State Highway Administration

• Dina Swires, Washington State DOT

• Rich Cunard, TRB

• Hillary Isebrands, FHWA

• Lori Sundstrom, NCHRP Program Officer

Presentation overview

• Project objective

• Project scope

• Corridors and data

collection

• Project products

Ph

oto

: Le

e R

od

eg

erd

ts

5/1/2015

3

Project Objective

• Focus on three or more roundabouts in series

Ph

oto

: Le

e R

od

eg

erd

tsProject Scope

• Document existing roundabout corridors

– Collect and evaluate field data

– Identify “lessons learned” from agencies

• Prepare methods and guidance for alternatives

evaluations

– Prepare predictive methods based on observed field data

for incorporation into key resource documents such as

HCM

– Prepare “Corridor Comparison Document” to

demonstrate broader evaluation process

5/1/2015

4

Site Selection

• 58 roundabout corridors identified in US (as of

2011)

• Nine corridors selected

• Geographically dispersed across the U.S., with some

grouping for data collection efficiency

• Mix of urban, suburban, and rural environments

• Mix of single-lane and multilane roundabouts

• Wide range of circumstances leading to each

corridor

Roundabout Corridors in US (as of 2011)

5/1/2015

5

Data Collection Corridors

• MD 216, Scaggsville, MD (pilot study site)

• La Jolla Boulevard, San Diego, CA (pilot study site)

• Old Meridian Street, Carmel, IN

• Spring Mill Road, Carmel, IN

• Borgen Boulevard, Gig Harbor, WA

• SR 539, Whatcom County, WA

• Golden Road, Golden, CO

• Avon Road, Avon, CO

• SR 67, Malta, NY

Data Collection Methods

• Travel time runs using GPS

• Bluetooth™ data collection (pilot sites only)

• Video recording of intersections (primarily for peak hour turning movement extraction)

• Spot speed samples

• Photographs and field notes

• Interviews with corridor owners/operators

• Data collection methods refined after pilot study locations (focused on GPS instead of Bluetooth)

5/1/2015

6

MD 216, Scaggsville, Maryland

Ima

ge

: ©

Go

og

le M

ap

s

La Jolla Boulevard, San Diego, CaliforniaIm

ag

e:

© G

oo

gle

Ma

ps

5/1/2015

7

Example Time-Space Trajectory

(La Jolla Boulevard SB, San Diego, CA)

Example Speed Profile for Urban Corridor

(La Jolla Boulevard SB, San Diego, CA)

5/1/2015

8

Old Meridian Street, Carmel, Indiana

Ima

ge

: ©

Go

og

le M

ap

s

Spring Mill Road, Carmel, IndianaIm

ag

e:

© G

oo

gle

Ma

ps

5/1/2015

9

Borgen Boulevard, Gig Harbor, Washington

Ima

ge

: ©

Go

og

le M

ap

s

SR 539, Whatcom County, WashingtonIm

ag

e:

© G

oo

gle

Ma

ps

5/1/2015

10

Example Speed Profile for Rural Corridor

(SR 539 NB, Whatcom County, WA)

Golden Road, Golden, ColoradoIm

ag

e:

© G

oo

gle

Ma

ps

5/1/2015

11

Avon Road, Avon, Colorado

Ima

ge

: ©

Go

og

le M

ap

s

SR 67, Malta, New YorkIm

ag

e:

© G

oo

gle

Ma

ps

5/1/2015

12

Example Speed Profile through Interchange

(SR 67 EB, Malta, NY)

Corridor Owner/Operator Interviews

• Wide variety of experiences leading to development

of corridors

• Reinforces motivation of Corridor Comparison

Document to evaluate corridors on case-by-case

basis

5/1/2015

13

Corridor Owner/Operators – Lessons Learned

• Once several roundabouts built on a corridor, more

are likely

– Increased acceptance of roundabouts

– Concerns about signal queue spillback

– Access management

– Consistency within corridor

• Traffic analysis typically analyzed roundabouts in

isolation

• Agency champion is key

Modeling

• New predictive tools for estimating operational

performance of roundabout corridors

• Created for incorporation into HCM to enable

signals-versus-roundabouts predictive comparisons

• Process of incorporating into HCM is underway with

development of HCM 2010 Major Update

5/1/2015

14

Modeling Framework

Modeling Framework

• New models developed by

this project:

– Free-flow speed (Step B)

– Roundabout Influence Area

Length (Step C)

– Geometric Delay (Step H)

– Impeded Delay (Step I) –

analogous to control delay

• Fits into existing Urban

Streets Procedure

5/1/2015

15

Segment and Sub-Segment Definitions

Operations Comparisons with “Equivalent Signalized

Corridors”

• High-level comparison to suggest trends if any

• “Equivalent” signalized corridors developed for each

roundabout corridor

• Estimate travel time (TT) performance using HCM-

based and simple simulation-based analysis

• Compare estimated TT to field-measured

roundabout performance

5/1/2015

16

Operations Comparisons with “Equivalent Signalized

Corridors” (cont.)

• Neither control option consistently results in

reduced travel time or delay for through routes

• Site-specific evaluation is key

• Roundabouts tends to improve travel time for routes

with a left turn onto or off a corridor

• Roundabouts are more likely to improve travel time

with irregular intersection spacing

Corridor Comparison Document

• Chapter 1: Introduction

• Chapter 2: Users of Arterials

• Chapter 3: Project Planning Process

– 3.1 Project Initiation (incl. understanding of context)

– 3.2 Concept Development

– 3.3 Alternatives Analysis

• Chapter 4: Performance Measures

• Chapter 5: Example Applications

5/1/2015

17

Understanding of

Context

• Select performance

metrics that are

important for the

corridor being

studied

• Develop and

evaluate reasonable

alternatives

Potential Performance

Measures

• Categories:

– Quality of Service

– Safety

– Environmental

– Costs

– Community Values

– Other

5/1/2015

18

Example Applications

• Show use of:

– Corridor Comparison Document

– Roundabout travel time model developed for this project

• Examples

– New suburban greenfield corridor

– Community enhancement on existing urban corridor

– Existing rural corridor in context sensitive, suburbanizing

area

– Existing suburban corridor undergoing operations and

safety evaluation

Questions?P

ho

to:

Lee

Ro

de

ge

rdts

5/1/2015

19

Lee A. Rodegerdts

FHWA TOPR 34: Accelerating

Roundabout Implementation in the

United States

Motivation for Research

• Roundabouts are one of FHWA’s Nine Proven

Countermeasures

• Nearly 15 years since first edition of Roundabouts:

An Informational Guide, roundabouts have yet to be

fully embraced by many agencies

• FHWA observed need for strategic undertaking to

accelerate roundabout implementation

5/1/2015

20

Goal of TOPR 34

• Uncertainty about a small number of key issues

seemed to be hampering more widespread

roundabout implementation

• FHWA well-positioned to undertake evaluations of

these key issues that will result in better guidance to

practitioner community

• Emphasis on real world, field-based data

TOPR 34 Tasks

• Assessment of Rectangular Rapid Flashing Beacons

(RRFB) for accessibility at multilane roundabouts

• Reassessment of capacity models for the Highway

Capacity Manual

• Assessing emissions characteristics of roundabouts

compared to signalized intersections

• Evaluation of fatal and severe injury crashes

5/1/2015

21

TOPR 34 Tasks (cont.)

• Evaluation of geometric parameters for trucks

• Assessment of crosswalk location and design

• Assessment of traffic control device treatments at

multilane roundabouts

Contractors

• Virginia Tech Transportation Institute

– Ron Gibbons, Contract Manager

• Kittelson & Associates, Inc.

– Lee Rodegerdts, Principal Investigator

• Institute of Transportation Research and Education

(ITRE) at North Carolina State University

– Bastian Schroeder, Co-Principal Investigator

• Task Leaders from all three organizations

5/1/2015

22

Reassessment of Roundabout

Capacity Models for the HCM

Ph

oto

: Le

e R

od

eg

erd

ts

Motivation for Research

• New capacity models for roundabouts added to

HCM 2010 based on NCHRP Report 572 (2007)

• Concern throughout user community that capacities

are lower than currently being observed

• Results of perceived capacity underestimation is

either oversizing roundabouts or avoiding them

altogether

5/1/2015

23

Purpose of Research Effort

• Collect new set of national field data (2012-2013)

– NCHRP Report 572 data: 2003

• Determine fit of HCM 2010 model to new data

• Determine best course of action to improve fit as

needed within constraints of time and budget:

– Calibrate using critical headway and follow-up time

– Develop new exponential or linear regression model

– Identify flow-based or geometric-based factors if

beneficial in improving model fit

Site Selection

• 23 intersections across the United States:

– Colorado (5)

– Indiana (7, all in Carmel)

– New York (2)

– Virginia (1)

– Vermont (1)

– Washington (7)

• Each approach studied at a given intersection is

considered a “site” for this analysis

5/1/2015

24

Data Collection

• Video recording of 48 hours at each site (2012-2013)

• Cameras focused on entering-circulating-exiting area

• Back of queue not always observed

– Camera angles

– Saturated but slowly rolling queues

Conditions for Data to be Usable

• Looking only for data periods that represent

capacity conditions: continuous queuing over the

entire study interval (1 min)

• Two criteria examined:

– Minimum queue recorded (where possible)

– Maximum move-up time <= 6 s

• Camera angle prevented visibility of queue, or

• Conditions saturated but rolling queue

• Generally consistent with NCHRP Report 572

methods

5/1/2015

25

Usable Data for Capacity Analysis

• Single-lane sites: 876 minutes

• Multilane sites: 1,285 minutes (all types)

– Multilane 1x2: 56

– Multilane 2x1: 231 right lane, 288 left lane

– Multilane 2x2: 365 right lane, 345 left lane

• NCHRP Report 572: single-lane 318 minutes, multilane 383 minutes (all types)

• Significantly larger dataset than for NCHRP Report 572

���� � 216

�2010:

Fit of HCM 2010 Model: Single-Lane

5/1/2015

26

Fit of HCM 2010 Model: 2x2 Right Lane

���� � 183

�2010:

Fit of HCM 2010 Model: 2x2 Left Lane

���� � 218

�2010:

5/1/2015

27

Fit of HCM 2010 Model: 2x1 Right Lane

���� � 255

�2010:

Fit of HCM 2010 Model: 2x1 Left Lane

���� � 224

�2010:

5/1/2015

28

Modeling Techniques

• Basic model forms analyzed:

– Exponential: �� � ������

– Linear: �� � � � ���

• Gap acceptance parameters for HCM 2010

calibration (measured under queued conditions)

– Critical Headway (��)

– Follow-Up Time (��)

• Model parameters set to minimize Root Mean

Square Error (RMSE) when allowed to vary

���� � 182

���. �� !�""#$%:

Example of Regression: Single-Lane Sites

Desire anchor for intercept

���� � 186

&#%�'!�� !�""#$%:

5/1/2015

29

Follow-Up Time Field Measurements

• Measured under queued conditions

• Outliers greater than mean+ 3 s.d. removed

• Intercept A = 3600/tf

State Number

Observations

Mean Standard

Deviation

Intercept

Single-Lane 2,647 2.6 1.0 1,380

2x2 Right Lane 1,964 2.5 1.0 1,420

2x2 Left Lane 1,563 2.7 1.2 1,350

2x1 Right Lane 886 2.3 0.9 1,560

2x1 Left Lane 948 2.1 0.6 1,710

1x2 318 2.5 0.9 1,440

Single-Lane Sites: Calibrated to National Follow-Up

Time

���� � 190

���. �� !�""#$%:

���� � 223

&#%�'!�� !�""#$%:

5/1/2015

30

0

200

400

600

800

1000

1200

1400

1600

1800

0 200 400 600 800 1000 1200 1400 1600 1800 2000

En

teri

ng

Flo

w (

pc/

h)

Conflicting Flow (pc/h)

2x2 Right Lane: Calibration to tf

2x2 Right Lane Data

Exp Regr

Linear Regr

HCM 2010

2x2 Right Lane: Calibrated to National Follow-Up

Time

���� � 164

���. �� !�""#$%:

���� � 192

&#%�'!�� !�""#$%:

0

200

400

600

800

1000

1200

1400

1600

1800

0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200

En

teri

ng

Flo

w (

pc/

h)

Conflicting Flow (pc/h)

2x2 Left Lane: Calibration to tf

2x2 Left Lane Data

Exp Regr

Linear Regr

HCM 2010

2x2 Left Lane: Calibrated to National Follow-Up Time

���� � 214

���. �� !�""#$%:

���� � 243

&#%�'!�� !�""#$%:

5/1/2015

31

0

200

400

600

800

1000

1200

1400

1600

1800

0 200 400 600 800 1000 1200 1400 1600

En

teri

ng

Flo

w (

pc/

h)

Conflicting Flow (pc/h)

2x1 Combined Right + Left Lanes: Match intercept to 2x2 right lane

2x1 Right+Left Lane Data

Exp Regr

Linear Regr

HCM 2010

2x1 Sites: Use 2x2 Right Lane Intercept

���� � 153

���. �� !�""#$%:

���� � 161

&#%�'!�� !�""#$%:

Geometric Effects

• Explored relationships between follow-up time and

key geometric parameters:

– Inscribed circle diameter

– Entry lane width

– Entry angle

– Splitter island width (for exiting effect)

• Conclusion: Trends are apparent but not strong

enough to include in the capacity model

5/1/2015

32

Geometric Effects (cont.)

R² = 0.0026

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0 50 100 150 200 250 300

Fo

llo

w-U

p T

ime

, tf

(s)

Inscribed Circle Diameter, D (ft)

Follow-Up Time vs. Inscribed Circle Diameter

R² = 0.0345

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

8 10 12 14 16 18 20 22 24

Fo

llo

w-U

p T

ime

, tf

(s)

Entry Lane Width (ft)

Follow-Up Time vs. Entry Lane Width

ICD

Entry Lane Width

R² = 0.025

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0 10 20 30 40 50 60 70

Foll

ow

-Up

Tim

e,

tf (

s)

Splitter Island Width (ft)

Follow-Up Time vs. Splitter Island Width

R² = 0.0586

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0 10 20 30 40 50 60 70

Fo

llo

w-U

p T

ime

, tf

(s)

Entry Angle (degrees)

Follow-Up Time vs. Entry Angle

Geometric effects (cont.)

• Trends in intuitive direction but too weak to use

Correlation = 0.24

R2 = 0.06

Splitter Island

WidthEntry Angle

5/1/2015

33

Trends over Time: Sites in common with NCHRP

Report 572 study

0

200

400

600

800

1,000

1,200

1,400

1,600

0 100 200 300 400 500 600 700 800

En

teri

ng

Flo

w (

pc/

h)

Conflicting Flow (pc/h)

SR 166/Mile Hill Drive/Bethel Avenue (WA04) -

North Leg

2012 Data 2003 Data

0

200

400

600

800

1,000

1,200

0 200 400 600 800 1,000 1,200 1,400 1,600

En

teri

ng

Flo

w

(pc/

h)

Circulating Flow (pc/h)

SR 9/US 5 (VT03) - West Leg - Right Lane

2012 Data 2003 Data

No apparent change in

capacity between

2003 and 2012

Summary of Findings

• Calibration to national follow-up time generally

produces plausible results with means for

calibration

• Adjustment made to 2x1 model for better fit

• Exponential form fits same or better than linear

form in all cases

• Separate investigation found that calibration to local

follow-up time produces best fit of all models to

date

5/1/2015

34

Recommended Candidate Models

• Single-lane model: ve = 1380 exp(-0.00102 vc)

• 2x2 right lane: ve = 1420 exp(-0.00085 vc)

• 2x2 left lane: ve = 1350 exp(-0.00092 vc)

• 2x1 both lanes: ve = 1420 exp(-0.00091 vc)

• 1x2: Use 2x2 right lane model

• Calibrate using local follow-up time where possible

• NOTE: Updated HCM chapters anticipated for adoption by the TRB Committee on Highway Capacity and Quality of Service in June 2015

Questions?

• Lee Rodegerdts, lrodegerdts@kittelson.com

Ph

oto

: Le

e R

od

eg

erd

tsP

ho

to:

Lee

Ro

de

ge

rdts