1 Intelligent Transportation Systems: Saving Lives, Time and Money PORTAL: Transportation Data Archive Intelligent Transportation Systems Laboratory Deena.

1Intelligent Transportation Systems: Saving Lives, Time and MoneyIntelligent Transportation Systems: Saving Lives, Time and Money

PORTAL: Transportation Data PORTAL: Transportation Data ArchiveArchive

Intelligent Transportation Systems LaboratoryIntelligent Transportation Systems Laboratory

Deena Platman, MetroDeena Platman, Metro

Dr. Kristin Tufte, Portland State UniversityDr. Kristin Tufte, Portland State University

PORTAL: Transportation Data PORTAL: Transportation Data ArchiveArchive

Intelligent Transportation Systems LaboratoryIntelligent Transportation Systems Laboratory

Deena Platman, MetroDeena Platman, Metro

Dr. Kristin Tufte, Portland State UniversityDr. Kristin Tufte, Portland State University


• PORTAL Background–Developed with CAREER grant from

National Science Foundation with additional financial support from FHWA

–Large investment in developing regional transportation archive

• Funding Situation–Current funding has run out, archive will

wither –Need sustainable funding source

• PORTAL Background–Developed with CAREER grant from

National Science Foundation with additional financial support from FHWA

–Large investment in developing regional transportation archive

• Funding Situation–Current funding has run out, archive will

wither –Need sustainable funding source


What’s in the PORTAL Database?What’s in the PORTAL Database?What’s in the PORTAL Database?What’s in the PORTAL Database?

Loop Detector DataLoop Detector Data20 s count, lane occupancy, speed from 20 s count, lane occupancy, speed from

500 detectors (1.2 mi spacing) 500 detectors (1.2 mi spacing)

Incident DataIncident Data140,000 since 1999140,000 since 1999

Weather DataWeather Data VMS DataVMS Data19 VMS since 199919 VMS since 1999

Data ArchiveData Archive

DaysDaysSince July 2004Since July 2004About 700 GBAbout 700 GB

4.2 Million 4.2 Million Detector IntervalsDetector Intervals

Bus DataBus Data1 year stop level data1 year stop level data

140,000,000 rows140,000,000 rows


PORTAL Web SitePORTAL Web SitePORTAL Web SitePORTAL Web Site

• Graphical display of archived data – Speed, Weather, Incidents, TriMet AVL

• Performance Reports, Traffic Counts, Freight Data, …

• Graphical display of archived data – Speed, Weather, Incidents, TriMet AVL

• Performance Reports, Traffic Counts, Freight Data, …


Speed Plot & Incident ReportsSpeed Plot & Incident ReportsSpeed Plot & Incident ReportsSpeed Plot & Incident Reports

Incident on NB I-205, Incident on NB I-205, log truck rear-ended a log truck rear-ended a

nursery truck, two nursery truck, two cars also involved, cars also involved,

duration over 4 hours.duration over 4 hours.

11/15/2005 Northbound I-20511/15/2005 Northbound I-205

20,000 20,000 reported reported

incidents/yearincidents/year

92 Database 92 Database fieldsfields

4 Entries per 4 Entries per incidentincident


Performance Report - ReliabilityPerformance Report - ReliabilityPerformance Report - ReliabilityPerformance Report - Reliability


Uses of PORTALUses of PORTAL

• Resource for local transportation professionals

• Metro RTP (?) • Projects

– SWARM– Travel Time– Bottleneck Identification– Data Quality Evaluation– Gap Filling– TriMet Data Analysis– Freight Data Display– Incident Autopsy

• Resource for local transportation professionals

• Metro RTP (?) • Projects

– SWARM– Travel Time– Bottleneck Identification– Data Quality Evaluation– Gap Filling– TriMet Data Analysis– Freight Data Display– Incident Autopsy

A

B

C

Bottleneck

Estimated Propagation Speed

A – 25 mphB – 22 mphC – 21 mph

Activation

Deactivation

90% percentile of historicalbottlenecks

8:15 2-vehicles collide8:19 Crash reported8:27 VMS message: CENTER LANES CLSD8:40 COMET requests tow9:10 Tow arrives9:27 Lanes clear 9:30 Traffic starts to clear9:45 Traffic half clear10:00 Traffic all clear


Metropolitan Mobility the Smart WayMetropolitan Mobility the Smart Way

Deena – Are any of these the RTP slides you

wanted?

Deena – Are any of these the RTP slides you

wanted?


Portal In Action: Metropolitan Congestion Over TimePortal In Action: Metropolitan Congestion Over Time

2005

2006

2004

Winter Spring Summer

Fall


Cross Section StudyCross Section Study

Speed-Volume Analysis (2005)

2004-05 Speed Comparison

‘04

‘05

Volume

SpeedSpeed


Cross Section ComparisonCross Section Comparison

Geographic Bottlenecks

Mega-project!

Design Flaws

Good Free-flow performance

Looming Danger


Ramp Flow ML Flow ML Speed

Metering activated earlier under SWARM

Vehicle-Hours of Delay

Station 28-Sep (P) 21-Sep (S)

Sunnyside 272 161

Johnson Creek 1054 818

Foster 1075 711

Corridor Total 3775 2358

Despite a slightly higher metering rate, SWARM’s earlier activation appeared to delay the onset of congested speeds and allowed for higher and more stable mainline flows.

Note: SWARM Metering Activation Data not collected at Foster

9/21 (SWARM) & 9/28 Pre-Timed9/21 (SWARM) & 9/28 Pre-Timed

Metering Activation

Speeds dropped prior to activation

Pre-Timed

SWARM


Extras – PORTAL DetailExtras – PORTAL Detail


What’s Behind the Scenes?What’s Behind the Scenes?What’s Behind the Scenes?What’s Behind the Scenes?

Database ServerDatabase ServerPostgreSQL Relational Database PostgreSQL Relational Database Management System (RDBMS)Management System (RDBMS)

StorageStorage2 Terabyte Redundant Array 2 Terabyte Redundant Array of Independent Disks (RAID)of Independent Disks (RAID)

Web InterfaceWeb Interface

Development ServerDevelopment ServerCentOS Linux distributionCentOS Linux distribution


Performance Measures UsedPerformance Measures UsedPerformance Measures UsedPerformance Measures Used

VolumeVolume SpeedSpeed

OccupancyOccupancy Vehicle Miles TraveledVehicle Miles Traveled Vehicle Hours TraveledVehicle Hours Traveled

Travel TimeTravel Time DelayDelay

In near future will add: Fuel In near future will add: Fuel Consumption, Emissions, Consumption, Emissions,

Carbon MeasuresCarbon Measures


Grouped Data – Travel TimeGrouped Data – Travel TimeGrouped Data – Travel TimeGrouped Data – Travel Time


Performance Report - ReliabilityPerformance Report - ReliabilityPerformance Report - ReliabilityPerformance Report - Reliability


Monthly ReportMonthly ReportMonthly ReportMonthly Report


Incident ReportsIncident ReportsIncident ReportsIncident Reports

Incident on NB I-205, Incident on NB I-205, log truck rear-ended a log truck rear-ended a

nursery truck, two nursery truck, two cars also involved, cars also involved,

duration over 4 hours.duration over 4 hours.

11/15/2005 Northbound I-20511/15/2005 Northbound I-205

20,000 20,000 reported reported

incidents/yearincidents/year

92 Database 92 Database fieldsfields

4 Entries per 4 Entries per incidentincident


Mapping – Speed SubtractionMapping – Speed SubtractionMapping – Speed SubtractionMapping – Speed Subtraction

Average Evening Peak Speed (5-6 pm)Average Evening Peak Speed (5-6 pm)

Difference Difference July-December 2005July-December 2005


Google TrafficGoogle TrafficGoogle TrafficGoogle Traffic


Incident ReportsIncident Reports

Incident on NB I-205, log truck rear-ended a nursery truck, two cars also involved, duration over 4 hours.

11/15/2005 Northbound I-205

Incident on SB I-205, NB effects visible


Extras – Other ProjectsExtras – Other Projects


Ramp Flow ML Flow ML Speed

Vehicle-Hours of Delay

Station1-Oct

(P)

17-Sep (S)

Sunnyside 5 8

Johnson Creek 189 205

Foster 23 61

Corridor Total 262 491

(3) Slightly higher metering rates under SWARM than Pre-Timed

(2) But metering at Sunnyside (and likely Foster) activated later under SWARM than Pre-Timed

(4) SWARM appears to implement a lower metering rate, responding to lower speeds.

10/1 (Pre-Timed)& 9/17 (SWARM)10/1 (Pre-Timed)& 9/17 (SWARM)

(1) SWARM activation matches drop in speed


SWARM Summary SWARM Summary

• SWARM allows more vehicles onto the freeway at each on-ramp.

» Counter to ODOT’s initial assumptions

• Pilot study on OR-217 SB demonstrated a tradeoff between decreased ramp delay and increased mainline delay

» Could not conclude that higher on-ramp volumes were the sole cause.

» SWARM’s earlier activation times reduce mainline delay under some conditions.

• Adjustment of metering rates and other SWARM parameters is needed to improve performance of the system

• Communications failures impact quality of SWARM operation

»Tradeoff between frequently updating ramp metering plans, and increased need for maintenance and tuning w/adaptive system

• Logging capabilities for SWARM/ATMS would make evaluation efforts easier

» Ramp queue loop detectors, meter activation times, and actual metering rates set by the SWARM system

• SWARM allows more vehicles onto the freeway at each on-ramp.

» Counter to ODOT’s initial assumptions

• Pilot study on OR-217 SB demonstrated a tradeoff between decreased ramp delay and increased mainline delay

» Could not conclude that higher on-ramp volumes were the sole cause.

» SWARM’s earlier activation times reduce mainline delay under some conditions.

• Adjustment of metering rates and other SWARM parameters is needed to improve performance of the system

• Communications failures impact quality of SWARM operation

»Tradeoff between frequently updating ramp metering plans, and increased need for maintenance and tuning w/adaptive system

• Logging capabilities for SWARM/ATMS would make evaluation efforts easier

» Ramp queue loop detectors, meter activation times, and actual metering rates set by the SWARM system


Travel Time Estimation ErrorTravel Time Estimation Error

• 85% of runs within error threshold of 20%• 85% of runs within error threshold of 20%2.9%

6.1%

14.7%

29.0%

10.3%

2.6% 2.9%

31.4%

0%

5%

10%

15%

20%

25%

30%

35%

< -30% -30% to -20%

-20% to -10%

-10% to0%

0% to10%

10% to20%

20% to30%

> 30%

Percent Error

Pe

rce

nt

of

Ru

ns


Real Time Travel Time EstimationReal Time Travel Time EstimationReal Time Travel Time EstimationReal Time Travel Time Estimation

• Goal: Assess accuracy of current travel time Goal: Assess accuracy of current travel time estimates and suggest improvementsestimates and suggest improvements

• AnalysisAnalysis

• 500 ground truth runs (GPS-enabled iQue)500 ground truth runs (GPS-enabled iQue)

• Compared ground truth with estimates using Compared ground truth with estimates using PORTAL dataPORTAL data

• ResultsResults

• Average error 11%Average error 11%

• Identified need for additional detectionIdentified need for additional detection

• Methods for evaluating benefits of additional Methods for evaluating benefits of additional detectiondetection


Sensor Data QualitySensor Data QualitySensor Data QualitySensor Data Quality

• ODOT products (speed map, ramp ODOT products (speed map, ramp metering) are only as good as the metering) are only as good as the input datainput data

• Use PORTAL to identify poorly Use PORTAL to identify poorly performing detectors; prioritize performing detectors; prioritize maintenance on those detectors maintenance on those detectors (improve efficiency)(improve efficiency)

• Key Question: How do data anomalies Key Question: How do data anomalies correlate with problems in the field?correlate with problems in the field?


Gap FillingGap FillingGap FillingGap Filling

Correlated information can help find mechanisms for filling the data gaps

By looking at available information from nearby stations, models fitted on historical data can provide an online estimate of the missing conditions. Different choices of estimation models exist, some more computationally intensive than others.

A B C

SB SCSA

Direction of flow

),(ˆCAB SSfS


Bottleneck IdentificationBottleneck IdentificationBottleneck IdentificationBottleneck Identification

A

B

C

Bottleneck

Estimated Propagation Speed

A – 25 mphB – 22 mphC – 21 mph

Activation

Deactivation

90% percentile of historicalbottlenecks


Incident Autopsy: 6/12/06Incident Autopsy: 6/12/06

8:15 2-vehicles collide8:19 Crash reported8:27 VMS message:

CENTER LANES CLSD8:40 COMET requests tow9:10 Tow arrives9:27 Lanes clear 9:30 Traffic starts to clear9:45 Traffic half clear10:00 Traffic all clear


Incident Autopsy: 6/12/06Incident Autopsy: 6/12/06

0

500

1000

1500

2000

2500

3000

3500

4000

4500

50000

8:0

0

08

:10

08

:20

08

:30

08

:40

08

:50

09

:00

09

:10

09

:20

09

:30

09

:40

09

:50

10

:00

10

:10

10

:20

10

:30

10

:40

10

:50

11

:00

veh

icle

s p

er

5 m

inu

tes

Crash

All Lanes Clear

All Traffic Clear

Tow Arrives

1 Intelligent Transportation Systems: Saving Lives, Time and Money PORTAL: Transportation Data Archive Intelligent Transportation Systems Laboratory Deena.

Documents