Urban Arterials: Linking Traffic, Transit, and Air Quality Data and Performance MeasuresMiguel Figliozzi, Associated Professor, Civil and Environmental Engineering

PSU Friday Seminar

Friday, October 19th, 2012

Special acknowledgements

Eric Albright Wei FengCourtney Slavin

Arterial traffic signal systems: adaptive or fixed‐cycle traffic lights?• Signal timing and cycle lengths can be constant throughout the day or change  

• Change:– As a function of time of the day (TOD)– As a function of detected traffic flows (adaptive)

• Adaptive: – Vehicle detection is key– $$$: for detection and adaptive control– $$$: staff time for maintenance/updating vs. software calibration  

3

Background I• The City of Portland implemented an adaptive traffic signal system (SCATS) along Powell Boulevard on October 8th, 2011

• SCATS: Sydney Coordinated Adaptive Traffic System  

• SCATS adjusts cycle length, phase splits, and offsets to optimize traffic

4

Background II• Buses running along this section of Powell Boulevard use TSP (Transit Signal Priority)

• To the best of our knowledge, this is the first time that SCATS and TSP have been integrated in Oregon (and in the USA)

5

Background III• Traditional GOALS

• Traditional Goals: a system more responsive to vehicle demand changes;  to coordinate traffic flows maintaining the highest acceptable degree of saturation and reducing overall delay … … there can be other goals

6

Adaptive vs. Time of Day Cycle Length

7

07:00 07:30 08:00 08:30 09:00

6080

100

120

140

Cycle Volume & Length

Vol

ume

per C

ycle

(veh

icle

s)

07:00 07:30 08:00 08:30 09:00

8010

012

014

016

0

October 26, 2011

Cyc

le L

engt

h (s

ecs)

Typical Constraints:- Maximum cycle length- Delays/queues for

secondary approaches

- Pedestrian requests- Transit priority

Objectives1. Are there significant improvements in traffic 

conditions after implementing SCATS?2. What about transit? Does SCATS affects transit 

operations?3. What about TSP (transit signal priority)? How 

effective is TSP,  where, how much?  4. Impacts on air quality and other livability 

related performance measures?

8

Data Sources• Traffic 

– Wavetronix radar detection units  – SCATS Loop detectors 

• Transit – TriMet AVL and APC (automated vehicle location and passenger counting)

– TSP logs • Air Quality 

– AQ measurements at some key intersections– In situ vehicle classification and presence (heavy vehicles)

9

BackgroundPowell Boulevard between 10th and 86th Streets 

10

• Urban arterial  • Connects 

Portland downtown and Gresham via the Ross Island Bridge

• Commuter corridor with peak period congestion

Study area: approx. 5.1 miles, 14 signalized intersections, 54 bus stops.

Traffic Volumes

11

• Peak Hour– Westbound:  up to 1,700 vehicles (7‐8 am)– Eastbound: up to 1,800 vehicles (4‐5 pm)

• Daily– Westbound: 19,000 – 22,000 vehicles– Eastbound: 18,000 – 20,000 vehicles– Both directions: approx. 41,000 vehicles

Volume‐speed vs. Time of Day

12

• Thibauds’ figure7am to 10am(red)

10am to 1pm(blue)

1pm to 4pm(green)

4pm to 7pm(orange)

7pm to 10pm(black)

Eastbound Movement (from Portland to Gresham)

Wavetronix unit at 35th @ PowellNext traffic light at 39th @ Powell

SP

EE

D (e

astb

ound

)

Volume (eastbound)

25

1,5001,000

30

35

0

2

4

6

8

10

12

7:00:00 AM 7:30:00 AM 8:00:00 AM 8:30:00 AM 9:00:00 AM

Speed Differen

ce (m

ph)

Powell @ 26th – Speed Difference –After‐BeforeAM Peak Period ‐Westbound

0

5

10

15

20

25

30

35

40

7:00:00 AM 7:30:00 AM 8:00:00 AM 8:30:00 AM 9:00:00 AM

Speed (m

ph)

Before SCATS AverageAfter SCATS Average

13

‐20

‐10

0

10

20

30

7:00:00 AM 7:30:00 AM 8:00:00 AM 8:30:00 AM 9:00:00 AM

% Differen

ce

Powell @ 26th ‐ Volume Difference – After‐Before AM Peak Period ‐Westbound

0

20

40

60

80

100

120

140

160

180

7:00:00 AM 7:30:00 AM 8:00:00 AM 8:30:00 AM 9:00:00 AM

Volume/5 minutes

Before SCATS Average

After SCATS Average

14

1. Traffic Evaluation

• In most cases, statistically significant improvements in travel speeds– With higher traffic volumes after SCATS  (statistically significant increases)

– Accounting for weather conditions– Accounting for day of the week  and other variables (comparing similar conditions one week before and after SCATS implementation)

15

2. Transit Performance Measures• Travel time (corridor) • Time point reliability (Powell @39th )• Controlling for…

– Passenger flows: no statistically significant differences in passenger boarding/hour and passenger load/hour between before and after SCATS

– TSP requests, time of day, direction of travel, bus driving speed…

16

17

Peak Period Off‐Peak PeriodWest East West East

With SCATS  221 88 0 ‐33Without SCATS   112 130 24 0

Regression Results• Peak periods: morning 7:30 to 8:30 am (westbound) and afternoon 4 to 6 pm (eastbound)

• Interaction between SCATS, direction and time of day (seconds)

• Overall: reduction of 23 seconds on average

18

2. Transit Evaluation Conclusions• Mixed Positive Results• Travel time (corridor)

– During off‐peak improvements in both directions– During peak hour improvements only eastbound (not westbound) 

• On‐time reliability  (at 39th @ Powell)– During off peak improvements in schedule delay in the westbound direction

– During peak no improvement in headway delay

19

3. Transit Signal Priority (TSP) Evaluation

Source: Global Traffic Technologies

20

3. Effectiveness of Active Transit Signal Priority (TSP)

• TSP can significantly reduce travel time and does facilitate schedule recovery (statistically significant impacts)

• BUT both priority and driver behavior (bus speed) have significant and discernible effects on travel time

• Schedule recoveries are also affected by many factors and are intersection dependent – Cross traffic volumes– Bus Stop Bay characteristics (length, location)– Bus Stop Location (mid block, far side, near side)– Time of day– Direction of travel … 

21

3. Transit, the big picture• Approx. 5 miles• Typical bus travel time (not accounting for stops,

loading/unloading passengers) between 500 and 700 seconds

• SCATS decreases travel time 23 second on average

• TSP decreases travel time up to 30 seconds (2 seconds per intersection on average)

• Improvements of up to 10% of transit travel time• As a reference: peak periods add around 90

seconds on average

22

3. TSP Complexity• Can you have too much of a good thing?

• YES !  In some cases TSP can facilitate bus bunching 

• A bus travel time is affected by the number of stops, passenger boardings/alightings, TSP requests…etc… of the 3 previous buses (statistically significant impacts)

• Current system is myopic (one bus, not system‐wide)

23

4. Air Quality Insights I

• First research to quantify the impacts of traffic signals on vehicle emissions at the sidewalk level 

• Clear link between traffic/signal variables (vehicle volumes, vehicle type, queuing, and traffic signal phase durations) with exposure levels  

• Overall, SCATS tends to improve air quality per vehicle served by reducing queuing along Powell 

24

4. Air Quality Insights II• For particulate matter (PM), the impact of  heavy duty diesel engines (queuing/idling) at the intersection = hundred or more passenger vehicles

• Some heavy vehicle factors:  – Angled tailpipe (angled away reduces exposure) (‐)– Bus engine efficiency (EMP) (‐)– Idling time is a more significant factor than age/mileage of the bus  

• The impact of vehicle presence on particulate matter can take up to 2 minutes (lagged effect)

We are doing a good job measuring performance measures related to… 

25

• Capacity• Mobility

• Specific Technology Contributions/Understanding (e.g. TSP much more difficult !)

In general, the evaluation of  Accessibility & Livability measures have not received much attention: ‐ Pedestrian/traffic LOS tradeoffs?‐ Empirical modeling of air quality/exposure impacts? For 

example, what are the health impacts of bicycling along a busy arterial bike lane vs. bicycling along a quiet local street?

Lessons Learned• Support from the City of Portland 

and TriMet was essential • Value of data integration and 

availability • Challenges to integrate diverse data 

sources– Diverse temporal/spatial structures– Variety of devices/calibration– Data cleaning and validation… 

• Detailed research questions required sophisticated statistical analysis (time series, autocorrelation, lagged variables, time series, …) 

Drinking from the data, analysis & modeling fire hose

Acknowledgements• Oregon Transportation, Research and Education

Consortium (OTREC)• Peter Koonce & Willie Rotich (City of Portland

Transportation Bureau)• David Crout, Steve Callas, & Kurtis McCoy (TriMet)• Graduate Research Assistants

• Courtney Slavin, Eric Albright, Wei Feng, Adam Moore (PSU)

• Thibaud Barrel (exchange student from France)• Air Quality Data Collection team

– Pam Johnson, Brian Davis, and Katherine Bell• Chris Monsere• Linda George Christine Kendrick

27

Related Papers I Albright, E., Figliozzi, M., Leveraging AVL/APC Bus Data

to Determine Factors Influencing Transit Signal Priority Effectiveness, Forthcoming 2012, Transportation Research Record

Moore, A.; M. Figliozzi; C. Monsere. Bus Stop Air Quality: An Empirical Analysis of Exposure to Particulate Matter at Bus Stop Shelters. Forthcoming 2012, Transportation Research Record

Albright, E., Figliozzi, M., Analysis of the Impacts of Transit Signal Priority on Bus Bunching and Performance, Proceedings of the Conference on Advanced Systems for Public Transport, June 2012, Santiago, Chile.

28

Related Papers II • Slavin, C., Feng, W., Figliozzi, M., Koonce, P., A Statistical

Study of the Impacts of SCATS Adaptive Traffic Signal Control on Traffic and Transit Performance (TRB 2013 Conference Proceedings)

• Slavin, C., Figliozzi, M., The Impact of Traffic Signal Timing on Pedestrian Particulate Matter Exposure (TRB 2013 Conference Proceedings)

• Albright, E., Figliozzi, M., Schedule recovery for late buses: What are the individual and joint contributions of Transit Signal Priority and bus operator behavior? (submitted Public Transport Journal)

29

30

Questions ?

Prof. Miguel Figliozzifigliozzi@pdx.eduhttp://web.cecs.pdx.edu/~maf/Civil and Environmental EngineeringPortland State University

Source: Portland Afoot30

Source: Portland Afoot