Using Archived Stop-Level Transit Geo-Location Data for Improved Operations and Performance Monitoring

Robert L. Bertini Portland State University September 26, 2003

Objectives
Trends in transit technology AVL and APC BDS system and data archiving Preliminary route level analysis Other examples Conclusions and future research

Automatic vehicle location (AVL)

Determination of vehicle location
Global positioning systems (GPS) Signposts Ground-based radio Dead-reckoning

U.S. transit agencies

128 operational 172 planned

Automatic passenger counters (APC)

Count boarding and alighting passengers
Infrared beams

U.S. transit agencies

60 operational 124 planned

AVL/APC in the Bay Area

AVL AC Transit Central Contra Costa Livermore/Amador Napa County Transit Santa Clara VTA Santa Cruz Sonoma County Transit Western Contra Costa San Francisco MUNI SamTrans TOTAL 800 in 2002 112 [131 by 2005] 71 19 6 [600] 80 [86] 54 [62] 0 [37] 350 312 [362] 1804 [2518] APC 125 0 71 0 0 [100] 0 0 0 0 [135] 0 [36] 196 [467]

Source: APTS Deployment in the United States Year 2002 Update, USDOT, 2003.

Archiving AVL/APC Data

AVL not designed with archived data in mind. Most AVL systems do not deliver data for offline planning analysis. Transit agencies didnt insist on it. APCs designed with archiving in mind. APCs often only in 10-15% of fleet. TriMet system driven by APC pays for itself in saved reporting costs

Source: TCRP H-28, Uses of Archived AVL-APC Data to Improve Transit Performance and Management.

About TriMet
Serves 1.2 M population 575 mi2 62.8 M annual bus trips 206,600 daily bus boardings 95 bus routes 655 buses 8100 bus stops Also LRT, Paratransit

Performance measurement

Measuring system performance is the first step toward efficient and proactive management. Increasing attention to transit performance Transit Capacity and Quality of Service Manual
Quantitative/qualitative Passenger point of view Linked to agency operating decisions

NCHRP Performance Based Planning Manual

Accessibility Mobility Economic Development

Improve reliability
Reduce variability of system performance
Delay Travel time

Attract more riders Reduce operations costs Increase productivity Link to service standards

In the past

Data collection more difficult Low temporal and spatial resolution Many people to collect little data Focus on limited, general, aggregate measures for external reporting Natural air conditioning

Today
Unlimited coverage and continuous duration Design, extract and test specific measures Actual system performance Data management/processing challenges Need for generating relevant measures

TriMets Bus Dispatch System

Navstar GPS Satellites
Radio System

GPS Antenna
Doors Lift APC (Automatic Passenger Counter) Overhead Signs Odometer Signal Priority Emitters

Radio Antenna OnBoard Computer

Control Head Memory Card

Radio

Garage PCs

TriMets Bus Dispatch System

PCMIA Card Control Head

Schedule deviation

Real Time Elements

Navstar GPS Satellites
Radio System

GPS Antenna
Doors Lift APC (Automatic Passenger Counter) Overhead Signs Odometer Signal Priority Emitters

Radio Antenna OnBoard Computer

Control Head Memory Card

Radio

Garage PCs

Real Time Elements

Schedule Data

On-board Computer

GPS Location

Radio/Cellular Communications

90 sec Updates

Dispatch and Control

Arrival Prediction

Archived Elements
Navstar GPS Satellites
Radio System

GPS Antenna
Doors Lift APC (Automatic Passenger Counter) Overhead Signs Odometer Signal Priority Emitters

Radio Antenna OnBoard Computer

Control Head Memory Card

Radio

Garage PCs

Archived Elements

Schedule Data

On-board Computer

GPS Location

APC/ Lift

PCMIA Card

Event Data [Operator] Pass up Overload Traffic Delay Train/Bridge Delay Fare Evasion Graffiti/Vandalism

Stop Data [Automatic]

Event Data: Fare Evasion

Archived Elements

Schedule Data

On-board Computer

GPS Location

APC/ Lift

PCMIA Card

Event Data [Operator]

Stop Data [Automatic] Scheduled Unscheduled

Stop Data

REWRITTEN ARRIVE TIME (IF DOOR OPENS)

RS ETE 15 M S TER ME 30

ARRIVE TIME

TIME LINE

LEAVE TIME DOOR CLOSE DOOR OPEN STOP LOCATION

DWELL TIME

Stop Data
Arrive Time Location ID Leave Time Max Speed Stop Time Route No.

Est. Load

Direction

Pattern Distance

Trip No.

X Coor.

14 14 14

01NOV2001 8:53:32 8:49:15 8:53:28 01NOV2001 8:55:00 8:51:41 8:54:46 01NOV2001 8:56:22 8:52:00 8:55:08

285 285 285

0 0 0

1120 4964 1120 4701 1120 4537

0 4 36

0 0 3

0 0 0

0 0 6

0 1 0

21 20 26

41 50 34

10558.58 7644468 676005 15215.05 7649112 676328 15792.35 7649674 676220

Route Number Vehicle Number Service Date Actual Leave Time Scheduled Stop Time Actual Arrive Time Operator ID Direction Trip Number Bus Stop Location

Dwell Time Door Opened Lift Usage Ons & Offs (APCs) Passenger Load Maximum Speed on Previous Link Distance Longitude Latitude

Y Coor.

Service Date

Badge

Dwell

Door

Offs

Ons

Lift

TriMets Bus Dispatch System

Stop Data: Passenger Movement

Route 14 Case Study

Route 14 Case Study

7.9 miles long 105 scheduled trips per weekday 64 scheduled stops 40-45 min scheduled trip time (mean 43.3, SD 2.7 min) 3-55 min headways (mean 11.4 min) Focus on two weeks April 1-12, 2002 [>1000 runs] Morning inbound from SE 94th/Foster to NW 4th/Hoyt Crossing Hawthorne Bridge

April 1 A.M. Inbound Trips

7 6 Distance (miles) 5 4 3 2 1 0 5:00

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:0

April 1 A.M. Inbound Trips

7 6 Distance (miles) 5 4 3 2 1 0 5:00

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:0

April 1 A.M. Inbound Trips

7 6 Distance (miles) 5 4 3 2 1 0 5:00

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:0

April 1 A.M. Inbound Trips

7 6 Distance (miles) 5 4 3 2 1 0 5:00

6:00

7:00

8:00

9:00

10:00

11:00

12:00

13:0

April 1: Two Inbound Trips

7.0 6.0 Distance (miles) 5.0 4.0 3.0 2.0 1.0 Time 0.0 8:30 8:45 9:00 9:15 9:30

April 1: Two Inbound Trips

One late, one early arrival

7.0 6.0 Distance (miles) 5.0 4.0 3.0 2.0 1.0

Time

Trip 1290 Badge 996 (28 yrs exp) Median 1.2 min late

Trip 1295 Badge 2606 (4 yrs exp) Median 0.9 min late

Two late departures

0.0 8:30

8:45

9:00

9:15

9:30

April 1: Impact of Pax Load

50 45

7.0 6.0

40

35

Distance (miles)

5.0
30

4.0 3.0 2.0

Trip 1290 Load

25

20

15

10

1.0 Time 0.0 8:30

8:45

9:00

9:15

9:30

Passenger Load

On-time Performance
5,000 4,500 4,000 3,500 Frequency 3,000 2,500 2,000 1,500 1,000 500 0 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Late/Early (minutes) 100%

All "stops" n = 66,012 Median = 0.6 min Late

90% 80% 70% 60% 50% 40%

32% Early

30% 20% 10% 0%

On-time Performance
5,000 4,500 4,000 3,500 Frequency 3,000 2,500 2,000 1,500 1,000 500 0 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Late/Early (minutes) 100%

TriMet Service Standard [-1 min +5 min] 85% On-time 11% Early 4% Late

90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

OTP vs. Time of Day

30

20

Early/Late (minutes)

10

-10

-20

-30 5:00

7:00

9:00

11:00

13:00

15:00

17:00

19:00

21:00

23:00

Time

TriMet Operator Experience

600 n = 3,249 Mean = 9.8 years 100% 90% 80% 70% 60% 300 50% 40% 200 30% 20% 10% 0 0 4 8 12 16 20 24 28 32 36 40 0%

500

400 Frequency

100

TriMet Operator OTP

6 5 4 Median Late/Early 3 2 1 0 1 -1 Operator -2 -3
April 1-12, 2002 Route 14 Inbound 94 Operators 65,848 stops

TriMet Operator OTP

6 5 4 3 2 1 0 -1 -2 Years Experience -3 0 5 10 15 20 25 30 Late Early April 1-12, 2002 Route 14 Inbound 94 Operators 65,848 stops

Median Early/Late

Stop Level Performance

0:30

April 1, 2002 Stop No. 2606 Hawthorne/ 22nd Ave Mean Headway 11:02 min St. Dev. 7:10

0:25

Scheduled Headway

0:20

0:15

0:10

0:05

0:00 0:00

0:05

0:10 0:15 0:20 Actual Headway

0:25

0:30

Stop Level Performance

50 45 40 35 30 25 20 15 10 5 0 5:00 Scheduled Actual

Cumulative Bus Arrival Number

April 1, 2002 Stop No. 2606 Hawthorne/SE 22nd Ave Mean Headway 11:02 min St. Dev. 7:10
6:00 7:00 8:00 Time 9:00 10:00 11:00 12:00

Route 14: Trip Time Model

T = T0 + aNd + bNa + cNb T0 = average nonstop trip time Nd = number of dwells Na = passengers alighting Nb = passengers boarding

Dwell Time
7,000 100% 90% 6,000

5,000

Frequency

4,000

Nonzero Dwells n = 34,456 Mean = 13.3 s SD = 20.3 s Lift use = 232 times

80% 70% 60% 50%

3,000

40% 30% 20%

2,000

1,000 10% 0 0 10 20 30 40 50 60 70 80 90 100 0%

Dwell (sec)

Dwell Time vs. Passenger Movement

200 180 160 140 Dwell Time (sec) 120 100 80 60 40 20 0 0 5 10 15 20 25 30 35 40 45 Passenger Movement

On + Off 4%

None 7%

Off 28% On 61%

Boardings
12,000 100%

10,000

n = 37,441 Mean = 1.33 SD = 2.45 Max = 37

90% 80% 70% 60%

8,000 Frequency

6,000

50% 40%

4,000

30% 20% 10%

2,000

0 Boardings
10 11 12 13 14 15 16 17 18 19 20 0 1 2 3 4 5 6 7 8 9

0%

Alightings
16,000 14,000 12,000 10,000 Frequency 8,000 6,000 4,000 20% 2,000 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Alightings 20 10% 0% 100%

n = 36,978 Mean = 1.31 SD = 2.62 Max = 37

90% 80% 70% 60% 50% 40% 30%

Dwell Time Model

Dwell Time = 5.94 + 2.87Nb + 0.96Na n = 24,995 [non - zero, no lift, APC =" G" , no layover] Boarding Only : Dwell = 6.03 + 4.27Nb n = 15,357 Alighting Only : Dwell = 6.63 + 0.95Na n = 7,021

Nonstop Trip Time

2,000 1,800 1,600 1,400 Travel Time (sec) 1,200 1,000 800 600 400 200 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Link Distance (miles)

n = 30,036 links NSTT=15.3+191.2x (sec)

Trip Time Model

T = 1606 + 21.2Nd + 0.96Na + 2.87Nb Nd = number of dwells Na = passengers alighting Nb = passengers boarding

Run Times: April 1-12, 2002

70

60

50 Run Time (min)

40

30

20

10

n=856 trips Mean on=40/trip Mean off=40/trip Mean no. of dwells=35/trip

8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00

0 6:00

Arrive Time

Run Times: April 1-12, 2002

100 90 80 70 Frequency 60 50 40 30 20 10 0 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 Run Time (minutes) 60 100%

n=856 trips Mean=41.5 min SD=5.1 min

90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Run Time vs. Operator Experience

60

50

40 Mean Run Time

30

20

10

n=856 trips 83 operators Run time/experience relationship (95% confidence)

1976 1979 1982 1984 1987 1990 1993 1995 1998 2001

0 1973

Hire Date

Actual vs. Predicted Run Time

60 55 50 Predicted Run Time 45 40 35 30 25 20 20 30 40 Actual Run Time 50 60

n=856 Mean SD

Actual 41.5 min 5.1 min

Predicted 41.6 min 7.4 min 594

Veh-hrs 592

Stop Consolidation Analysis

Route 14 length 41710 feet, 64 stops Mean stop spacing 670 feet Consolidate stops to 1000 foot spacingeliminate 10 stops Reduce trip time 21.2 sec per stop, save 3.5 min per inbound run 105 inbound trips per day 6.1 hours savings ($60/hr +/-) Add ~9 trips using existing resources Improve mean headway 11.4 10.4 min Does it affect demand?

Boarding Area Improvement Analysis

Streamlining program to reduce dwell time Curb extensions Nearside farside conversion Smart cards Mean boarding time estimated 2.87 sec Reduce boarding time by 1 sec at top ten locations Total of 1800 passengers boarded at these locations Save 30 min/day

Other Applications

Other Applications

1 census tract 0.25-mi buffers 38% of population

Transit Signal Priority

Transit Signal Priority

Reduce run time and schedule variability 180 intersections complete, 100 more in 2003 Emitter turned on if bus is >90 sec late Remains on until <30 sec late Green extension/red truncation

Transit Signal Priority

Express Buses as Freeway Probes

Buses as Probes on Arterials

Buses as Probes on Arterials

Buses as Probes on Arterials

Buses as Probes on Arterials

Buses as Probes on Arterials

Buses as Probes on Arterials

Buses as Probes on Arterials

ttveh= 1.23 ttpseudo

Vveh= 0.84 Vpseudo

Buses as Probes on Arterials

Cleveland Ave.

Birdsdale Ave.

Pkwy.

Walters Rd.

Probe Vehicle Run

Legend Signalized Intersection Bus Stop

Hood Ave.

Ea stm

TriMet Route 9

Hogan Ave.

Towle Ave.

Main Ave.

an

Buses as Probes on Arterials

EASTBOUND WESTBOUND
-6% Savings 9% Savings

27% Savings 2% Savings 22% Savings

14% Savings

EB Probe Before (sec) After (sec) 172 169

EB Bus 200 182

EB Hypothetical 168 131

WB Probe 178 130

WB Bus 212 224

WB Hypothetical 142 122

Discussion
Other uses for the BDS data? Conclusions? Save all of your data! Ask for archiving capabilities

Acknowledgements
Steve Callas, TriMet Ahmed El-Geneidy, James Strathman, Thomas Kimpel, Kenneth Dueker, Center for Urban Studies Sutti Tantiyanugulchai, Civil & Environmental Engineering National Science Foundation Oregon Department of Transportation City of Portland Federal Transit Administration

More information
Email: bertini@pdx.edu Web: http://www.its.pdx.edu TriMet data readily available.