evaluating robustness of signal timings for conditions of varying traffic flows

24
Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows 2013 Mid-Continent Transportation Research Symposium – August 16, 2013 – 9:30AM – 12:00PM Cameron Kergaye, PhD, PE, PMP Director of Research Utah Department of Transportation

Upload: darrin

Post on 25-Feb-2016

46 views

Category:

Documents


0 download

DESCRIPTION

Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows. 2013 Mid-Continent Transportation Research Symposium – August 16, 2013 – 9:30AM – 12:00PM Cameron Kergaye , PhD, PE, PMP Director of Research Utah Department of Transportation. Traffic Signal Optimization. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows2013 Mid-Continent Transportation Research Symposium – August 16, 2013 – 9:30AM – 12:00PM

Cameron Kergaye, PhD, PE, PMPDirector of ResearchUtah Department of Transportation

Page 2: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Traffic Signal Optimization

Should signal timings be optimized for high-than-average traffic counts?

How should signal timing optimization accommodate multiple counts?

Page 3: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Summary

Signal timing plans perform best based on average traffic flows mean, mode, and median when exposed to day-to-day traffic flow variability

Optimizing signal timings for higher traffic demand is better than for lower traffic demand Should be used only when sufficient traffic data are

unavailable.

Page 4: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Introduction Optimization of signal timings is considered to be one

of the most effective tools to improve traffic operations on urban arterials.

However, once traffic signal systems are retimed and implemented, quality of their performance largely depends on day-to-day variability of traffic flows in the field.

As soon as traffic patterns change significantly, performance of the signal timings deteriorates.

Current signal timing practice recommends development of separate signal timing plans for major day-to-day traffic patterns (weekday, weekend, special events, etc.).

Page 5: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Introduction When adaptive traffic control is implemented it is

impractical to develop plans for every traffic pattern that warrants a separate signal timing plan.

Therefore, it is important to develop signal timing plans that will minimize disbenefits of implementing signal timing plans in variable traffic conditions.  

Page 6: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Research Background Signal timing plans are based on traffic data that are

usually collected during a short term effort (e.g. 1-week).

The data includes 24-hour weekly volume profiles, turning movement counts, vehicular speeds, and travel time runs.

The data is analyzed with other sources to ensure that they are representative of common field conditions.

Page 7: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Research Background

What is a representative traffic volume pattern? It is the one that generates signal timings that work

best in a variety of traffic conditions.

Finding such a traffic volume pattern and optimizing signal timings in field experiments requires significant resources. Therefore, we use traffic simulation and other methods

that do not require field experiments.

Page 8: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Research Contribution Weekday PM-Peak hour traffic flows from the field are

modeled in microsimulation for a year. Two factors made this modeling approach possible:

Comprehensive set of field traffic flows collected during an entire year

Special tool to validate and balance traffic flows for the model.

Signal timing plans are optimized for each of the representative traffic flows resembling the process that usually occurs in practice.

Each of the signal timing plans is evaluated for the entire set of traffic flows to determine the best representative set of traffic flows.

Page 9: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Study Area

Park City, UT

14 Intersections

Long Corridor and Small Business District

N

0 0.5 1 km

Page 10: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Validation Results – Southbound

0

20

40

60

80

100

120

140

160

180

200

1 2 3 4 5 6 7 8 9 10 11 12 13Intersection Segments

Trav

el T

ime

(sec

)

2007 Field Microsimulation

Page 11: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Validation Results – Northbound

0

20

40

60

80

100

120

140

160

180

200

1 2 3 4 5 6 7 8 9 10 11 12 13

Intersection Segments

Trav

el T

ime

(sec

)

2007 Field Microsimulation

Page 12: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

MethodologyAccurately model day-to-day traffic variations in microsimulation Collect and Process Field Traffic Volumes

Traffic Volumes Recorded by SCATSTraffic Volumes Collected by Automatic Traffic RecordersManually Collected Traffic Volumes

Verify SCATS Traffic Volumes Build, Calibrate, and Validate the VISSIM Model Model Variability of Traffic Flows in VISSIM

Prepare for Modeling Variability of Traffic Flows in MicrosimulationVerify the reasonableness of SCATS traffic volumesBalance traffic flows in the networkVerify VISSIM Traffic Flows

Develop signal timing plans based on traffic flows of representative days Select scenarios of ‘representative-day traffic volumes’ Optimize signal timings in VISGAOST Evaluate Optimized Signal Timings in VISSIM

Page 13: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Scats Output with Traffic Volumes

B) Volume Store (VS) Output

A) Strategic Monitor (SM) Output

Page 14: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Variation of Traffic Flows

Thanksgiving Holiday 2009

Thanksgiving Holiday 2006

Christmas 2009

Memorial Day 2007

Memorial Day 2009

Martin Luther King Day 2009

Christmas 2006

Mean = 1185

St. Dev. = 40.74

Mean = 1135

St. Dev. = 249.49

Page 15: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Balancing Traffic Flows

Page 16: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Peak-Hour ATR & Scats Data

A) ATR 605 Soutbound

D) ATR 606 NorthboundC) ATR 606 Soutbound

B) ATR 605 Northbound

100

200

300

400

500

600

700

800Tr

affic V

olum

e (v

eh/h

our)

Days

SB Field SB SCATS

200

400

600

800

1000

1200

1400

Traffi

c Vol

ume

(veh

/hou

r)

Days

NB Field NB SCATS

200

400

600

800

1000

1200

1400

1600

Traffi

c Vol

ume

(veh

/hou

r)

Days

SB Field SB SCATS Data

400

600

800

1000

1200

1400

1600

1800

Traffi

c Vol

ume

(veh

/hou

r)

Days

NB Field NB SCATS

Page 17: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Verifying Reasonableness of Scats Traffic Volumes

0

200

400

600

800

1000

1200

14001 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101

105

109

113

117

121

125

Traffi

c Vol

umes

(veh

/hou

r)

Turning Movements (in Ascending Order based on Hourly Traffic Flows)

2005 Manually Counted Traffic Volumes 2009 Average SCATS Traffic Volumes

2009 Minimum SCATS Traffic Volumes 2009 Maximum SCATS Traffic Volumes

Page 18: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Verifying Match of Scats & Vissim Traffic Flowsy = 0.949x + 9.1946

R² = 0.9474

0

200

400

600

800

1000

1200

1400

1600

1800

0 200 400 600 800 1000 1200 1400 1600 1800

Aver

age

VISS

IM Tu

rnin

g M

ovem

ents

Cou

nts [

veh/

hour

]

Average Field Turning Movements Counts [veh/hour]

Page 19: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Optimizing Signal Timings In Visgaost

Split[1,8]=[[10.0,23.0,10.0,23.0,10.0,23.0,10.0,23.0]];LeadPhase[1,8]=[[1,0,0,1,1,0,1,0]];CycleLength[1]=[66.0];Offset[1]=[30.0];

VISSIM InputSignalGroups[8]=[1.0,2.0,3.0,4.0,5.

0,6.0,7.0,8.0];

Simulation time: 600 to 4200Parameter ValueTotal travel time[h] 835.8Total delay time[h] 159.2Number of stops 21828Stopped delay[h] 84.2

Network Performance

VISSIM Output

VISSIM VISGAOST

Signal Timings

PerformanceMeasures

Page 20: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

“Representative Days” of Traffic Volumes

• AVERAGE• MAX• MEDIAN• MIN• MODE• 75th PERCENTILE• 85th PERCENTILE

Page 21: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Reduction of PI During Visgaost Optimizations

1000

1100

1200

1300

1400

1500

0

100

200

300

400

500

600

0 20 40 60 80 100

Perfo

rman

ce In

dex

-MAX

Perfo

rman

ce In

dex

-All

othe

rs

Number of Generations

MIN AVERAGE MODE MEDIAN 75th PERCENTILE 85th PERCENTILE MAX

Cycle Lengths & Offsets

Cycle Lengths, Offsets & Splits

Page 22: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Variations in Performance Indices

4500

5500

6500

7500

8500

9500

10500

11500

12500

200

300

400

500

600

700

800

900

1000

1100

Net

wor

k Tr

affic F

low

[veh

/hou

r]

Perf

orm

ance

Inde

x

DatesMAX MIN EXISTING

AVERAGE MODE 85th PERCENTILE

75th PERCENTILE MEDIAN Total Network Traffic Flow

Page 23: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

Performances Based on Various ‘Representative-Days’

2000

4000

6000

8000

10000

12000

14000

0

100

200

300

400

500

600

Net

wor

k Tr

affic T

hrou

ghpu

t [ve

h/ho

ur]

Perf

orm

ance

Inde

x

Optimization Scenarios

Performance Index Traffic Demand

Page 24: Evaluating Robustness of Signal Timings for Conditions of Varying Traffic Flows

ConclusionsSignal timings optimized for median traffic flows were best. Similar results were found for other average traffic flows (i.e. mean and mode).

Findings show that signal timings developed for traffic flows that most frequently occur in the field bring more benefits than those that are developed for less frequent but higher traffic flows.

Basing signal timings on higher-than-average traffic demand still generates better results than those developed for lower traffic flows.

This is justified where there is a shortage of reliable traffic flow data from the field and when demand is expected to grow significantly.