Evaluating InSync Performance in Microsimulation Aleksandar Stevanovic, PhD, PE Florida Atlantic University Transpo 2012 Bonita Springs, FL October 29,

Evaluating InSync Performance in Microsimulation

Aleksandar Stevanovic, PhD, PE

Florida Atlantic University

Transpo 2012Bonita Springs, FLOctober 29, 2012

Why Simulating InSync?

• Comprehensive evaluation of its performance

• Controlled experiments• Abundance of outputs• Better understanding of pros &

cons• Repeatability and comparability

Originality of this Study

• No comprehensive evaluation of InSync in microsimulation

• Regular operations modeled with no attention to special conditions/situations

• Evaluation of InSync for regular operations (AM, MD, and PM) +• Freeway Incident• Rail Preemption• Extra Demand• Inclement Weather

About SR 421 - Dunlawton Ave

• 12 signalized intersections (11 analyzed)

• Divided 4-6 lane facility

• Left turns• RR crossing

& school zones

• 35-50 mph speed limits

Why SR421 in Volusia County?

• A common (sub)urban corridor

• Difficult to retime

• Seasonal and annual variations in traffic flows

• Evacuation route

10,000

15,000

20,000

25,000

30,000

35,000

40,000

Vehicles/day

Williamson to Clyde Morris Blvd. Clyde Morris Blvd. to SR 5A/Nova Rd.

SR5A/Nova Rd. to Spruce Creek Rd. Spruce Creek Rd. to US 1

Data Collection for VISSIM Model

• Turning movement counts• Travel times along

mainline (GPS & video)• Signal timing sheets &

Synchro files• Other data – RR gate

timings, school zone hours, etc.

Google Map as Background

NEMA-like Ring-Barrier Controllers

Calibration

• Making sure that modeled and simulated volumes are equal

• Volume balancing• Sources and sinks• Saturation flow rates• Manual process

Calibration Details – Nova Rd

Calibration Results

R² = 0.9985

0

5000

10000

15000

20000

25000

0 5000 10000 15000 20000 25000

Sim

ula

tion

TM

C

Field TMC

Field vs Simulation AM Peak

R² = 0.9917

0

5000

10000

15000

20000

25000

0 5000 10000 15000 20000 25000

Sim

ula

tion

TM

C

Field TMC

Field vs Simulation MD Peak

R² = 0.9962

0

5000

10000

15000

20000

25000

0 5000 10000 15000 20000 25000

Sim

ula

tion

TM

C

Field TMC

Field vs Simulation PM Peak

Validation

• Verifying that segment travel times from model and field are comparable

• Speed distributions• Acceleration &

deceleration• Offsets & detectors• Understand limitations

Validation Results

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

200.0

Eastbound

VISSIM

GPS

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

Westbound

VISSIM

GPS

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

Eastbound

VISSIM

GPS

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

Westbound

VISSIM

GPS

0.0

50.0

100.0

150.0

200.0

250.0

Eastbound

VISSIM

GPS

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

Westbound

VISSIM

GPS

AM Peak MD Peak PM Peak

TOD Signal Timings

Intersection Field SS Optimized MS Optimized Field SS Optimized MS Optimized Field SS Optimized MS Optimized

Williamson 115 110 68 120 124 76 125 144 100

I-95 SB 115 110 68 120 124 76 125 144 100

I-95 NB 115 110 68 120 124 76 125 144 100

Taylor rd 115 110 68 120 124 76 125 144 100

Yorktowne 115 110 68 120 124 76 125 144 80

Clyde Morris 115 110 116 120 124 130 125 144 160

Victoria 115 110 116 120 124 130 125 144 160

Village 115 110 116 120 124 130 125 144 160

Nova A 110 116 A 124 130 A 144 160

Spruce A 110 90 A 124 74 A 144 88

US 1 140 110 90 140 124 74 145 144 88

MD PeakAM Peak PM Peak

InSync – Controllers Run Free

1) Disable all coordination (set controller into “free/uncoordinated” mode)2) Enable Detector Diagnostic Failure Mode

a. Set On failure to 250 minutesb. Set Off failure to 5 minutes

i) Diagnostic failure mode on a 170 will allow itself to come out of failure.3) Set all “Minimum Green” times to 5 sec., or as client feels comfortable4) Leave “Maximum Green” times as before5) Set “Passage Gap” or “Observed Gap” to 1 sec.6) For protected/permitted left turns, omit the left turn call when the opposing thrumovement is green.

• Notes:o This will prevent a yellow trap if the controller receives calls onpermitted left turn phases when the opposing thru is green, but maynot bring on the adjacent green thru movement to allow the left turna permitted movement.

7) Enable “Soft Recall” on the mainline phases8) Disable “Yellow Lock” and “Red Lock” detector locking9) Set all “Detector Delays” to 0 sec.10) Disable all recalls: Max, Min, Hard, Vehicle, Phase, etc.11) Remove “Extensions”12) Disable “Anti-Backup” or “Left Turn Trap”13) Enable “Max Recall Inhibit”

InSync Detectors in VISSIM

InSync-VISSIM Interface

Field-like InSync Dialog Box

If a picture is used, use this layout. Pics are much preferred over bullets or other text.

Simulations

• Four signal timing scenarios - Identical traffic flows and geometric conditions• Field TOD• Optimal TOD – single section• Optimal TOD – multiple sections• InSync Adaptive – no additional adjustments

• 10 random seeds• Three peak periods (AM, MD, and

PM)• 2.5 hours (15 min for warm up and

15 min for cool off) for each peak period

Operational Scenarios

• Regular operations - 3 peak periods (AM, MD, and PM)

• Special operations (only for PM peak)• Freeway Incident - Traffic diversion

due to a freeway incident• Rail Preemption - Impact of

disruption caused by freight train operations

• Extra demand – 20% traffic growth (flat)

• Inclement Weather - Reduction in speed and sat. flows due to a heavy rain

Freeway Incident

• An 1-hour incident on I-95 freeway between SR 421 and SR 400

• One lane closed – traffic diverts to SR 421

• AADT ~ 50,000 veh/day; k factor ~ 15%; d factor ~ 55%; 4 lanes

• 1,000 veh – diverted to SR 421 during one hour

Diversion Routes

300

300

200

2001000

vph

Incident

Rail Preemption

• 4 trains modeled in 2 PM peak hours (every 30 mins)

• Anywhere between 20 – 100 cars in a train (modeled 20, 30, 70, and 100)

• Train’s speed distributed around 45 mph

• Gates take about 7 seconds to go up/down

SR 421 & FEC Railway

FEC RR

Extra Demand

• Flat increase of 20% for all of the traffic demand generators

• Turning movement proportions remain the same

• Impact on intersections – almost all intersections worsen LOS by one grade

• Oversaturated network (8 ints. at LOS D or worse; 4 ints. at LOS E or worse)

Inclement Weather

• Inclement weather conditions (heavy rain) in Florida

• PM Peak traffic demand unaltered (worst-case scenario)

• Travelling speeds reduced by ~ 15% (from - 20% to -10%)

• Saturation flows reduced by ~ 20% (~ ~ 1500 vph)

What was Evaluated?

• Intersection performance

• Main-corridor travel times

• Main street vs. Side street

• Network performance

Results Examples

Intersection Performance

AM Peak – Intersection Delay

0

5

10

15

20

25

30

35

40

Tot

al 2

hr

Inte

rsec

tion

Del

ay (s

)

Intersection

Field SS Optimized MS Optimized InSync Optimized

MD Peak – Number of Stops

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

Num

ber

of S

tops

per

Veh

icle

Intersection


PM Peak – Average Queue

0

50

100

150

200

250

Ave

rage

Que

ue (

ft)

Intersection


Intersection Field SS Optimized MS Optimized InSync Optimized

Williamson E D C DI-95 SB D D C DI-95 NB B B B ATaylor rd C C C BYorktowne C D C CClyde Morris E E E EVictoria D E E DVillage D F E BNova F F F ESpruce F D C DUS 1 D D C D

LOS

ED – Level of Service

Main-Corridor Travel Times

IW Travel Times EB

0

50

100

150

200

250

300

350

Tra

vel T

ime

(s)

Segments


IW Travel Times WB

0

50

100

150

Tra

vel T

ime

(s)

Segments


IW Travel Times All

0

100

200

300

400

500

600

700

800

900

Tra

vel T

ime

(s)

Direction


IW Travel Times - Summary

Percent improvement

No. From To TT (s) Speed (mph) LOS TT (s) Speed (mph) LOS TT (s) Speed (mph) LOS TT (s) Speed (mph) LOS InSync vs Field InSync vs SS Synchro InSync vs MS Synchro

1 Williamson I-95 SB 25.4 19.0 D 26.3 18.3 D 22.6 21.3 D 24.6 19.6 D -3.36 -6.54 8.572 I-95 SB I-95 NB 9.3 35.0 A 9.2 35.2 A 13.5 24.1 C 10.0 32.5 B 7.60 8.35 -25.973 I-95 NB Taylor 23.4 21.3 D 22.2 22.5 C 20.9 24.0 C 18.6 26.8 C -20.53 -16.15 -10.694 Taylor Yorktowne 28.7 38.5 A 31.1 35.5 A 34.8 31.7 B 33.5 33.0 B 16.61 7.67 -3.865 Yorktowne Clyde Morris 44.0 27.6 C 68.7 17.7 D 66.8 18.2 D 45.0 27.0 C 2.21 -34.44 -32.596 Clyde Morris Victoria 51.2 28.4 B 48.9 29.7 B 54.2 26.8 C 44.0 33.0 B -13.92 -9.92 -18.727 Victoria/City Swallowtail 72.2 20.4 D 57.7 25.6 C 58.0 25.4 C 57.1 25.8 C -20.91 -1.10 -1.648 Swallowtail Nova 198.7 9.2 F 300.9 6.1 F 295.2 6.2 F 93.3 19.5 D -53.05 -69.00 -68.409 Nova Spruce 101.8 32.2 B 108.7 30.2 B 114.6 28.6 B 88.6 37.0 A -12.94 -18.46 -22.7010 Spruce US 1 148.8 17.3 D 98.0 26.2 C 120.8 21.3 D 120.8 21.3 D -18.83 23.33 -0.04

11 US 1 Spruce 134.6 19.2 D 78.2 33.0 B 104.7 24.6 C 81.8 31.5 B -39.19 4.63 -21.8212 Spruce Nova 134.9 24.2 C 141.3 23.1 C 143.3 22.8 C 124.8 26.1 C -7.49 -11.65 -12.8913 Nova Swallowtail 67.7 26.9 C 53.7 33.9 B 52.9 34.5 B 56.2 32.5 B -17.06 4.49 6.1514 Swallowtail Victoria 53.8 27.4 C 57.0 25.9 C 57.5 25.6 C 62.4 23.6 C 16.12 9.41 8.5215 Victoria/City Clyde Morris 55.2 26.5 C 61.3 23.9 C 59.9 24.4 C 50.3 29.1 B -8.89 -17.93 -16.0816 Clyde Morris Yorktowne 40.2 30.8 B 59.7 20.7 D 48.8 25.4 C 42.1 29.4 B 4.75 -29.51 -13.7917 Yorktowne Taylor 33.7 32.8 B 32.8 33.7 B 34.2 32.3 B 30.3 36.4 A -9.99 -7.59 -11.4118 Taylor I-95 NB 23.3 19.8 D 20.8 22.2 C 28.3 16.3 E 17.2 26.8 C -26.13 -16.98 -39.1619 I-95 NB I-95 SB 14.0 23.2 C 14.8 21.8 D 21.5 15.0 E 14.1 23.0 C 0.88 -5.04 -34.5520 I-95 SB Williamson 49.9 9.7 F 32.2 15.0 E 30.5 15.8 E 21.4 22.6 C -57.17 -33.74 -30.1021 Williamson US 1 681.8 20.9 D 763.5 18.6 D 802.2 17.7 D 548.5 25.9 C -19.55 -28.16 -31.6222 US 1 Williamson 585.0 24.3 C 548.0 25.9 C 559.3 25.4 C 521.6 27.2 C -10.83 -4.81 -6.74

InSync Optimized % Change

EA

STB

OU

ND

WE

STB

OU

ND

S

Field SS Optimized MS Optimized

-19.55 -28.16 -31.62

-10.83 -4.81 -6.74

Main Street vs. Side Street

RP Main vs. Side Street T & L

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

Agg

rega

te D

elay

per

Veh

icle

(s)

Street / Movement


RP Main vs. Side Street

0.0

10.0

20.0

30.0

40.0

50.0

60.0

Agg

rega

te D

elay

per

Veh

icle

(s)

Street


Network Performance

FI Total Network Delay

0

100

200

300

400

500

600

700

800

900

Tot

al N

etw

ork

Del

ay (h

)

Signal Timings


FI Total Number of Stops

47000

48000

49000

50000

51000

52000

53000

54000

55000

56000

57000

Tot

al N

um

ber

of S

top

s

Signal Timings


FI Total Travel Time

1450

1500

1550

1600

1650

1700

1750

Tot

al T

rave

l Tim

e (h

)

Signal Timings


FI Network Summary

Parameter Field SS Optimized MS Optimized InSync InSync vs Field InSync vs SS Opt InSync vs MS Opt

Average delay time per vehicle [s] 90.0 93.8 88.4 74.0 -17.8 -21.1 -16.3 Average number of stops per vehicles 1.77 1.80 1.79 1.64 -7.7 -9.1 -8.7 Average speed [mph] 20.5 20.1 20.7 22.3 8.9 11.1 8.0 Average stopped delay per vehicle [s] 61.7 63.0 57.7 48.2 -21.8 -23.5 -16.4 Total delay time [h] 771.3 804.4 756.9 633.4 -17.9 -21.3 -16.3 Total Distance Traveled [mi] 34419.8 34384.4 34360.2 34380.3 -0.1 0.0 0.1 Latent delay time [h] 2.3 2.4 2.2 2.2 -6.9 -9.2 -3.5 Latent demand 14.0 14.3 11.6 12.7 -9.3 -11.2 9.5 Number of Stops 54737.7 55652.4 55293.7 50461.7 -7.8 -9.3 -8.7 Number of vehicles in the network 627 629 598 593 -5.5 -5.8 -1.0 Number of vehicles that have left the network 30218 30244 30231 30216 0.0 -0.1 0.0Total number of vehicles (network throughput)30845 30873 30830 30809 -0.1 -0.2 -0.1 Total stopped delay [h] 528.4 540.6 494.1 412.8 -21.9 -23.6 -16.5 Total travel time [h] 1678.1 1710.3 1661.8 1539.0 -8.3 -10.0 -7.4

% change

Summary of Results

Reduction [%] of Delays & Stops

Field TOD

SS Optimized

MS Optimized

Regular Operations

Total Delay

-17.9 -21.3 -16.3

Stops -7.8 -9.3 -8.7

Freeway Incident

Total Delay

-24 -30.5 -32.4

Stops -25.8 -26.1 -32.6

Rail Preemption

Total Delay

-18.2 -21.3 -16.6

Stops -9.4 -10.7 -10.5

Extra Demand

Total Delay

-25.5 -30.3 -27.5

Stops -14.7 -20.5 -19.1

Inclement Weather

Total Delay

-23.5 -18.1 -15.4

Stops -11.6 -7.4 -8.7

Average improvements: Delay ~ 22%; Stops ~ 15 %

Summary – All Scenarios

Field TOD SS Optimized

MS Optimized

PM Peak

Intersections

Travel times

Main vs. Side

Network

InSync better than a respective TOD signal timingResults are not definite or TOD is better

Questions, feedback, comments?

[email protected]

Evaluating InSync Performance in Microsimulation Aleksandar Stevanovic, PhD, PE Florida Atlantic University Transpo 2012 Bonita Springs, FL October 29,

Documents