June 16, 2004

Dr. Robert BertiniMichael Rose

Evaluation of the “COMET”

Incident Response Program

Oregon Department of Transportation

Presentation Outline

• Research Objectives• Freeway Management in Portland• What are Incidents?• What is Incident Response?• Incident Response in Portland• Incident Data for the Region• Estimation and Costs of Incident Delay• Conclusions

Research Objectives

Demonstrate the use and display of archived data from multiple sources as a tool for evaluation and monitoring of freeway operations.

Evaluate the effectiveness of the “COMET” incident response program in Portland, Oregon

Develop tools to facilitate efficient deployment of resources and programs in other places

Transportation System Management

• 75 CCTV cameras

• 18 variable message signs

• 118 ramp meters

• 436 inductive loop detectors

• Digital archives of incident logs

• AVL Archives of COMET movements

• An extensive fiber optic communications system

In the Portland metro area ODOT currently operates an extensive

advanced traffic management system from the TMOC including:

Traveler Information

Variable Message Signs

Traffic Cameras

Traffic Reports

www.tripcheck.com

What are Incidents?

• Crashes, breakdowns and other random events on our highways• They contribute to more than half of the delay on our highways • Lead to major road closures increase drivers’ exposure to hazardous conditions • Cause secondary crashes • Divert maintenance resources and reduce overall productivity

What is Incident Response?

• A proactive strategy for dealing with incidents• Roving response vehicles • CCTV network of traffic cameras • An operations center monitoring the cameras • Communications network linking the vehicles and the operations center

Benefits of Incident Response

• IR programs are the eyes and ears of highway system

• They are proven strategy for reducing duration of incidents. They reduce delay, fuel consumption, accident exposure, air pollution and environmental impacts

• Decrease emergency vehicle response times

• Reduce secondary crashes

• Improve safety for emergency and highway maintenance personnel

• Improve relations between the driving public and the local transportation agency

Typical Delay Curve

Time

Cumulative Vehicles

Capacity Flow

Actual Incident Duration

Demand Flow

Total Delay

Incident Flow

Clearance Time

Det

ecti

on

Res

pons

e

Recovery

Capacity Flow

Inci

dent

Eff

ects

Cle

ared

Inci

dent

Cle

ared

Inci

dent

Occ

urs

Ver

ifie

d C

all R

ecei

ved

IR A

rriv

ed

Tow

Tru

ck C

alle

d

Tow

Tru

ck A

rriv

ed

Reducing incident duration by ½ reduces incident delay by ¾.

Reducing incident duration by ½ reduces incident delay by ¾.

Incident Response in Portland

The incident response program, known as “COMET”, began service in March 1997, and now covers the Portland metropolitan area nearly 24 hours a day with 11 specially equipped incident response vehicles.

Incident Response in Portland

The vehicles are equipped with:

• Variable message sign

• Basic traffic control equipment

• Gasoline and automotive fluids

• Basic automotive tools

• Communications system

• Automatic Vehicle Location system

Patrol Region

Data Sources

• Incident Database

• Automated Vehicle Location

• Automatic Traffic Recorders

• Inductive Loop Detectors

• Archived Weather Data

2001 Incident Data

70,976 incident records

21,708 incidents

6,334 Incidents on I-5

860 Crashes on I-5

Removal of 49,286 duplicate records

3 week sample for delay

estimation

660 incidents

3,188 Crashes11,078 Stalls

Incident Frequency Map

Low

Medium

High

2001 Incidents

Stall50%

Debris16%

Crash15%

Tag & Tow11% Other

7% Construction, Congestion & Other Closure

1%

Incident Types

N=21,728

0 Lanes66%

1 Lane31%

3, 4 or More1%

2 Lanes2%

Lane Blocking Incidents

N=18,920

2001 Incident Location

Right Shoulder59%

Right Lanes14%

Left Lanes10%

Left Shoulder6%

Center Lanes4%

All Lanes3%

Gore Area3%

Off Road1%

N=13,464

Crashes & Stalls

Account for 65% of incidents in the region

Average Number of Lanes Blocked

0.64

0.31

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Crashes Stalls

Average Number of Lanes Blocked

0.64

0.31

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Crashes Stalls

Average Duration in Minutes

65

47

01020304050607080

Crashes Stalls

Average Duration in Minutes

65

47

01020304050607080

Crashes Stalls

Crashes by Time of Day and Day of Week

Day of Week

Fridays had the highest crash frequency in 2001

198

451

519488

445

530

230

0

100

200

300

400

500

600

Day of WeekN

um

be

r o

f C

ras

he

s N=2860

5437

51 4319

74

127

174

207

147126

137 142

179 182

240

272

236

147

6448

61 5934

0

50

100

150

200

250

300

Time

Fre

qu

en

cy

N=2860Time of Day

The highest crash frequency was during the evening peak

Incidents by Month

215 208 236 269 222 234 186 242 195304 293 257

789 764803 803 917

714919

892

745

937 1009

774

497445

425462

621

444

585 514

423

547538

492

0

200

400

600

800

1000

1200

1400

1600

1800

2000

Month

Nu

mb

er

of

Inc

ide

nts

Crash Stall OtherN=18920

1501 15341464

1417

1760

1392

16901648

1363

1788 1840

1523 Mean1576

Rainy Days and Crashes

6550

79

144

50

101

36 4335

174 181194

150158 157

172

133150

199

160

112

66

130125

0

50

100

150

200

250

Month

Nu

mb

er

of

Cra

sh

es

0

5

10

15

20

25

Nu

mb

er

of

We

t D

ay

s

Crashes on Wet DaysCrashes on Dry DaysWet Days1152 Crashes on Wet Days

1712 Crashes on Dry Days113 Wet days

Ongoing Incidents 1 Day

12:00AM

2:00AM

4:00AM

6:00AM

8:00AM

10:00AM

12:00PM

2:00PM

4:00PM

6:00PM

8:00PM

10:00PM

12:00AM

Time of Day

1

11

21

31

41

51

61

71

81In

cid

en

t C

ou

nt

Incident Time and Duration on November 18th 2002

N=83

0

2

4

6

8

10

12:00AM

1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM 9:00 AM 10:00AM

11:00AM

12:00PM

1:00 PM 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00PM

11:00PM

Time

Nu

mb

er

of

Inc

ide

nts Incidents Scheduled IR Vehicles

Ongoing Incidents - One Year

Average Ongoing Incidents by Day of Week vs. IR Vehicles

0.00

1.00

2.00

3.00

4.00

5.00

Time

Num

ber

of In

cide

nts/

IR V

ehic

les

Sunday Monday TuesdayWednesday Thursday FridaySaturday IR Vehicles Weekday IR Vehicles Weekend

3 Week Incident Delay Estimation

3 week sample for delay

estimation

660 incidents

164 delay causing

incidents

Filtering

Manually matched incidents to ATRs

Determined the capacity reduction using incident data and tables from the Highway Capacity Manual

Average flow data for the specific hour of the incident were drawn from the 2001 ATR database.

The result was 112,000 vehicle hours of delay caused by incidents during the 3 week period

Estimated cost of additional fuelconsumption due to delay $150,000Estimated cost of lost time due to delay $2,800,000Total Estimated cost of delay for 3 weeks $2,950,000

Estimation of Annual Incident Delay 2001

-Extrapolated from 3 week sample-

• Estimated vehicle hours of delay: 1,940,000 hours

• Estimated cost of additional fuel consumption due to delay: $2,500,000.

• Estimated cost of lost time due to delay: $48,500,000.• Total Estimated cost of incident delay: $51,000,000.

• If each incident were to increase in duration by an average of 1 minute the annual cost of delay increases by: $1,400,000

• The cost to operate COMET for one year is about $750,000

For COMET to be cost effective the duration of each incident needs to be reduced by an average of 32 seconds.

For COMET to be cost effective the duration of each incident needs to be reduced by an average of 32 seconds.

Efficiency Curve

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 1 2 3 4 5 6 7 8 9 10

Average Reduction of Incident Duration in Minutes

Pe

rce

nta

ge

of

Inc

ide

nts

As

sis

ted

Benefits are Greater than Costs

Costs are Greater than Benefits

Conclusions

• Comet is clearly beneficial. The responders only need to reduce the duration of each incident by just a few minutes to have a positive effect on the flow of traffic.

• It is impossible to measure and assign a dollar value to the numerous other environmental and public relations benefits of the program.

• Archived data is a rich and useful source of information.

• Ongoing improvements to database entry and dispatching need to be made as traffic volumes and patterns change.

Thank You

•U.S. Department of Transportation•Transportation Northwest (TransNow)•Oregon Department of Transportation•Portland State University Department of Civil & Environmental Engineering

•Dennis Mitchell, Jack Marchant, Richard Santa Ana and Eric Anderson of ODOT

•Barnie Jones, Rob Edgar, Galen McGill and Edward Anderson at ODOT for their support

The full report is available online at:http://www.its.pdx.edu/opbenefits.html

Michael Rose: mwrose@pdx.edu

Dr. Robert Bertini: bertini@pdx.edu