O Centro de Excelência em BRT Across Latitudes and Cultures (ALC-BRT CoE) promoveu o Bus Rapid Transit (BRT) Workshop: Experiences and Challenges (Workshop BRT: Experiências e Desafios) dia 12/07/2013, no Rio de Janeiro. O curso foi organizado pela EMBARQ Brasil, com patrocínio da Fetranspor e da VREF (Volvo Research and Education Foundations).
Uneak White's Personal Brand Exploration Presentation
BRT Workshop - Fare Collection in the Broader Payments Environment
1. Ricardo Giesen, BRT Workshop, Rio 2013
Fare Collection in the Broader
Payments Environment
Ricardo Giesen
Pontificia Universidad Católica de Chile
BRT Workshop: Experiences and Challenges
Rio de Janeiro, July 2013
2. Ricardo Giesen, BRT Workshop, Rio 2013
Motivation
• OD matrices reflect demand’s behavior for a particular
time period
– Obtaining OD matrices is a long and expensive process
• Mobility Surveys
• Traffic Counts
• In-vehicle Passenger counts (Passenger per vehicle counts)
• New technologies allow to compile cheaper and higher
quality information
– Automated Fare Collection Systems (AFC)
3. Ricardo Giesen, BRT Workshop, Rio 2013
Transantiago AFC System
• bip! Transactions (card id and type, fare, vehicle id, time)
~ 35M transactions per week
> 3M bip! cards observed
10.000 stops
• AVL GPS (vehicle id, time, position)
~ 80 a 100 M observations
> 6.000 buses
4. Ricardo Giesen, BRT Workshop, Rio 2013
Outline
• Automated Fare Collection Systems (AFC) & Data
Collection Systems (ADCS)
• ADCS Relationship to key agency functions
• Role in Decision Support
• Examples of applications and services
• Passenger Flow and System Capacity
• OD Matrix Estimation
• Performance Measurement (PM)
• Real-time demand estimation and control (reliability)
• Traveler (Customer) information
5. Ricardo Giesen, BRT Workshop, Rio 2013
Automated Data Collection Systems (Buses)
• Automatic Vehicle Location Systems (AVL)
• bus location based on GPS
• available in real time
• Automatic Passenger Counting Systems (APC)
• bus systems based on sensors in doors with channelized
passenger movements
• passenger boarding (alighting) counts for stops/stations with
fare barriers
• traditionally not available in real-time
6. Ricardo Giesen, BRT Workshop, Rio 2013
Automated Data Collection Systems (Buses)
• Automatic Fare Collection Systems (AFC)
• increasingly based on contactless smart cards with unique ID
• provides entry (exit) information (spatially and temporally) for
individual passengers
• traditionally not available in real-time
• XFCD (extended floating car data)
• Maintenance
• Monitoring
7. Ricardo Giesen, BRT Workshop, Rio 2013 7
Manual
• low capital cost
• high marginal cost
• small sample sizes
• aggregate
• unreliable
• limited spatially and temporally
• not immediately available
Automatic
• high capital cost
• low marginal cost
• large sample sizes
• more detailed, disaggregate
• errors and biases can be estimated and
corrected
• ubiquitous
• available in real-time or quasi real-time
Transit Agencies are at a Critical Transition
in Data Collection Technology
We are in the era of BIG DATA!
8. Ricardo Giesen, BRT Workshop, Rio 2013
Opportunities
• ADCS
– monitoring status at various levels of resolution
– measuring reliability
– understanding customer behavior
• Data + Computing
– simulation-based performance models
– robust scheduling
– dynamic scheduling
• Communications
– real time information (demand)
– Dynamic response (supply)
• Systematic approaches for planning, operations, real time control
• Maintenance
9. Ricardo Giesen, BRT Workshop, Rio 2013
ADCS - Potential
• Integrated ADCS database
• Models and software to support many agency decisions
using ADCS database
• Monitoring and insight into normal operations, special
events, unusual weather, etc.
• Large, long-time series disaggregate panel data for
better understanding of travel behavior
10. Ricardo Giesen, BRT Workshop, Rio 2013
ADCS - Reality
• Most ADCS systems are implemented independently
• Data collection is ancillary to primary ADC function
• AVL - emergency notification, stop announcements
• AFC - fare collection and revenue protection
• Many problems to overcome:
• not easy to integrate data
• requires substantial resources
• lack of expertise
11. Ricardo Giesen, BRT Workshop, Rio 2013
Key Transit Agency/Operator Functions
A.Off-Line Functions
• Service and Operations Planning (SOP)
• Performance Measurement (PM)
B.Real-Time Functions
• Service and Operations Control and Management (SOCM)
• Customer Information (CI)
12. Ricardo Giesen, BRT Workshop, Rio 2013
Key Operator Functions: Off-Line Functions
A.1) Service and Operations Planning (SOP)
• Network and route design
• Frequency setting and timetable development
• Vehicle and crew scheduling
• ADCS Impacts on SOP
• AVL: Provide detailed characterization of route segment running times
• APC: Provide detailed characterization of stop activity
(boardings, alightings, and dwell time at each stop)
• AFC: Give detailed characterization of fare transactions for individuals over
time, supports better characterization of traveler behavior
13. Ricardo Giesen, BRT Workshop, Rio 2013
A.2) Performance Measurement (PM)
• Measures of operator performance against SOP
• Measures of service from customer viewpoint
• ADCS Impacts on PM:
• AVL: Supports on-time performance assessment
• AFC: Supports passenger-oriented measures of travel time and reliability
Key Operator Functions: Off-Line Functions
14. Ricardo Giesen, BRT Workshop, Rio 2013
B1) Operations Control and Management
• Dealing with deviations from SOP, both minor and major
• Dealing with unexpected changes in demand
• ADCS Impacts on management and control
• AVL: Identifies current position of all vehicles, deviations from
SOP or desired operational strategy
• AFC: Provide real-time information about demand
Key Operator Functions: Real-Time Functions
15. Ricardo Giesen, BRT Workshop, Rio 2013
B2) Customer Information (CI)
• Information on routes, trip times, vehicle arrival times, etc.
• Both static (based on SOP) and dynamic (based on SOP and
SOCM)
• ADCS Impacts on Customer Information
• AVL: Supports dynamic CI
• AFC: Permits characterization of normal trip-making at the
individual level, supports active dynamic CI function
Key Operator Functions: Real-Time Functions
16. Ricardo Giesen, BRT Workshop, Rio 2013
Key Functions
Off-line Functions
Real-time Functions
Supply Demand
Customer
Information (CI)
Service Management
(SOCM)
Service and Operations
Planning (SOP)
ADCSADCS
Performance
Measurement (PM)
System
Monitoring, Analysis, and
Prediction
17. Ricardo Giesen, BRT Workshop, Rio 2013
Real-Time Functions
Demand
CONTROL CENTER
Prediction
Estimation of current
conditionsSupply
ADCS
Incidents/Events
Vehicle Locations Loads
Monitoring
Dynamic
rescheduling
Information
• travel times
• paths
18. Ricardo Giesen, BRT Workshop, Rio 2013
Managing for uncertainty
Timing
Strategy
Operations Planning Real time
Preventive
Run/cycle times
Robust schedules
Deployment of recovery
resources (spare crews)
temporarily and spatially
Real time (minor)
adjustments
Supervision and dispatching
Corrective
Real time operations control
Dynamic service plan
adjustments and
rescheduling
Dynamic crew rescheduling
Use of spare resources
19. Ricardo Giesen, BRT Workshop, Rio 2013
Examples of ADCS in Decision Support
• Passenger Flow and System Capacity
• Public Transport OD Matrix Estimation
• Performance Measurement (PM)
• Real time demand estimation and control
• Customer information
20. Ricardo Giesen, BRT Workshop, Rio 2013
Passenger Flow and System Capacity
• Estimation of passenger flows at the route level
• Peak of the peak period and peak segment
• Route choice in complex transit systems
• Route choice in corridors (parallel routes)
shipProfileofthePiccadillyLine25-March-2012
Ravichandran/MIT(harshavr@mit.edu)8
010002000300040005000600070008000900010000
ckfosters-Oakwood
akwood-Southgate
thgate-AmosGrove
rove-BoundsGreen
Green-WoodGreen
reen-TurnpikeLane
Lane-ManorHouse
ouse-FinsburyPark
nsburyPark-Arsenal
enal-HollowayRoad
ad-CaledonianRoad
nRoad-King'sCross
ross-RussellSquare
sellSquare-Holborn
orn-CoventGarden
en-LeicesterSquare
are-PiccadillyCircus
lyCircus-GreenPark
rk-HydeParkCorner
orner-Knightsbridge
e-SouthKensington
on-GloucesterRoad
erRoad-Earl'sCourt
Court-BaronsCourt
ourt-Hammersmith
ith-TurnhamGreen
Green-ActonTown
wn-EalingCommon
mmon-NorthEaling
thEaling-ParkRoyal
ParkRoyal-Alperton
rton-SudburyTown
yTown-SudburyHill
ryHill-SouthHarrow
arrow-RaynersLane
ynersLane-Eastcote
tcote-RuislipManor
uislipManor-Ruislip
Ruislip-Ickenham
ckenham-Hillingdon
Hillingdon-Uxbridge
Town-SouthEaling
hEaling-Northfields
fields-BostonManor
tonManor-Osterley
rley-HounslowEast
st-HounslowCentral
tral-HounslowWest
West-HattonCross
nCross-Heathrow4
ross-Heathrow123
ow123-Heathrow5
Figure5:WBRidership--Peak30minutes
CapacityRidership
12 trains
11 trains
22. Ricardo Giesen, BRT Workshop, Rio 2013
OD Matrix Estimation
Objective:
• Estimate passenger OD matrix at the network level
using AFC and AVL data
• Multimodal passenger flows
• AFC characteristics
• Open (entry fare control only)
• Closed (entry+exit fare control)
• Hybrid
Source:
"Intermodal Passenger Flows on London’s Public Transport Network: Automated Inference of Full Passenger Journeys Using Fare-
Transaction and Vehicle-Location Data. Jason Gordon, MST Thesis, MIT (September 2012).
24. Ricardo Giesen, BRT Workshop, Rio 2013
Source: Munizaga
and Palma 2012
Transactions localized spatially
25. Ricardo Giesen, BRT Workshop, Rio 2013
Vanishing routeBoarding point
First route
Second route
Third route
User i
B
iks1
2
1iks
1
1iks
4
1iks
5
1iks
j
iks1
ikV1
j
iks2
j
iks3
B
iks2
B
iks3
ikV2
ikV3
ikikikik VVVJ 321 ,,
d(a,b)
d(a,b) < M
ikd1
ikd2
ikd3
Estimated
alighting stop
Alighting Stop Estimation: Open AFC
Source: Chapleau et al
2008
26. Ricardo Giesen, BRT Workshop, Rio 2013
• Three types of transactions are distinguish:
• Bus
• Metro
• Multi-service stop
Alighting Stop Estimation: Open AFC
Source: Munizaga and Palma 2012
27. Ricardo Giesen, BRT Workshop, Rio 2013
• Identify service
• Position of the next transaction
• Closest stop to the next transaction
Alighting Stop Estimation: Bus
Source: Munizaga and Palma 2012
28. Ricardo Giesen, BRT Workshop, Rio 2013
i
min ti
+
di->xpost ypost
vcam
×(qtcam /qtvia)
s.a. dpost
£ d
boarding
Next tap
Closest
point ?
Alighting stop
estimation ?
Look for the
point that
minimizes
generalized
travel time
i
d
. . . . . .
. . . . . . .
Alighting Stop Estimation: Bus
Source: Munizaga and Palma 2012
29. Ricardo Giesen, BRT Workshop, Rio 2013
• Position of the next transaction
• Metro Station: minimum generalized time to the
position to the next transaction
• Route Estimation: minimum time
Alighting Stop Estimation: Metro
Source: Munizaga and Palma 2012
30. Ricardo Giesen, BRT Workshop, Rio 2013
• Identify services stopping at multi-service stop
• Position of the next transaction
• Identify common lines: minimum expected
generalized time to the position of the next
transaction
• Assign service: the first of the common lines
that passes at that stop
Alighting Estimation: Multi-service Stop
Source: Munizaga and Palma 2012
31. Ricardo Giesen, BRT Workshop, Rio 2013
Estimation of alighting stop
• Compute travel time
• Time until the next transaction: Transshipment or activity
(destination)?
Simple rule: t > 45min Activity
Alighting Estimation: Multi-service Stop
Source: Munizaga and Palma 2012
32. Ricardo Giesen, BRT Workshop, Rio 2013
Route 1
Route 2
Route 3
Boarding stop
Alighting stop
Bus Route OD Inference: Closed system
33. Ricardo Giesen, BRT Workshop, Rio 2013
Journey 1
1. Enter East Croydon NR station, 7:46
2 & 3. Out-of-station interchange to Central Line at
Shepherds Bush, 8:30
4. Exit LU at White City, 8:35
5. Board 72 bus at Westway, 8:36
6. Alight 72 bus at Hammersmith Hospital, 8:42
Journey 2
7. Board bus 7 at Hammersmith Hospital, 16:17
8. Alight bus 7 at Latymer Upper School, 16:19
9. Board bus 220 at Cavell House, 16:21
10. Alight bus 220 at White City Station, 16:24
11. Enter LU at Wood Lane, 16:25
12 & 13. Out-of-station interchange from Circle or
Hammersmith & City to District or
Piccadilly, 16:40
14. Exit LU at Parsons Green, 16:56
34. Ricardo Giesen, BRT Workshop, Rio 2013
Example Transport for London
• Oyster fare transactions/day:
• Rail (Underground, Overground, National Rail): 6 million (entry & exit)
• Bus: 6 million (entry only)
• For bus:
• Origin inference rate: 96%
• Destination inference rate: 77%
• For full public transport network:
• 76% of all fare transactions are included in the seed matrix
• Computation time for full London OD Matrix (including both
seed matrix and scaling):
• 30 mins on 2.8 GHz Intel 7 machine with 8 GB of RAM
35. Ricardo Giesen, BRT Workshop, Rio 2013
Station Specific Analysis
36MIT, Transit Leaders
36. Ricardo Giesen, BRT Workshop, Rio 2013
Preliminary Results
• Sample size of 63.221 observations from the week first week of
September 2008.
• 80% of the cases were estimated
• 77% of the bip! Cards have complete information
37. Ricardo Giesen, BRT Workshop, Rio 2013
Preliminary Results: Location of trips
destinations
Histogram of trips per day for the subsample
38. Ricardo Giesen, BRT Workshop, Rio 2013
Histograma de Etapas para la submuestra
Preliminary Results:
Histogram of trips per day per bip!
39. Ricardo Giesen, BRT Workshop, Rio 2013
Histograma de Etapas para la submuestra
Preliminary Results:
Histogram of Stages per Trip
40. Ricardo Giesen, BRT Workshop, Rio 2013
O/D North West East Center South South-East Oi
North 552 145 195 410 115 125 1542
West 122 1093 660 983 125 196 3179
East 208 562 1557 1126 404 912 4769
Center 374 824 961 889 509 748 4305
South 124 150 428 612 476 264 2054
South-East 117 177 972 754 217 1261 3498
Dj 1497 2951 4773 4774 1846 3506 19347
Matrix
Preliminary Results
Origen-Destination Trip Matrix
41. Ricardo Giesen, BRT Workshop, Rio 2013
What can be obtained?
• Level of service for each “detected” user
- In Vehicle Travel Time
- Transshipment Time including wait
- Estimation of Waiting Time for the Initial Trip
• Desegregated by
– Residential zone
– Destination Zone (work, study)
– Operator
42. Ricardo Giesen, BRT Workshop, Rio 2013
What can be achieved?
– Load Profiles per service per period
– Public Transport O/D Trip Matrix
– Passenger Flows at each Stop
– Passenger Arrival Pattern at each Stop
43. Ricardo Giesen, BRT Workshop, Rio 2013
Conclusions
• ADCS provide information with a high level
of resolution never seen before.
• Big data can change the way we do public
transport planning and management.
• Analysis possibilities are endless …
45. Ricardo Giesen, BRT Workshop, Rio 2013
• Filtering data errors
Data Pre-processing
Source: Cortés et al 2011
46. Ricardo Giesen, BRT Workshop, Rio 2013
Data Pre-processing
Source: Cortés et al 2011
47. Ricardo Giesen, BRT Workshop, Rio 2013
Projection of GPS point to the route
Data Pre-processing
Source: Cortés et al 2011
48. Ricardo Giesen, BRT Workshop, Rio 2013
Can we use GPS data to monitor speed?
• We have the position of each bus every 30 secs.
• We need to assign buses to services and
distinguish between stopping and moving time
Monitor the speed of each route
50. Ricardo Giesen, BRT Workshop, Rio 2013
Time-Space Diagram
D2
r
D1
r
Source: Cortés et al 2011
51. Ricardo Giesen, BRT Workshop, Rio 2013
Computing commercial speed
Source: Cortés et al 2011
52. Ricardo Giesen, BRT Workshop, Rio 2013
Average Speed per Service
GoodVery bad Bad Acceptable Excelent
Km/h
53. Ricardo Giesen, BRT Workshop, Rio 2013
Average Speed per service-segment
(Spatial desegregation)
Level of service
54. Ricardo Giesen, BRT Workshop, Rio 2013
Average Speed per service-segment
(Temporal desagregation)
Morning
Peak
Off-Peak
55. Ricardo Giesen, BRT Workshop, Rio 2013
• We can obtain a matrix sij per service
(i: segment; j: period)
• Global Indicator aggregated per segment
sR = reference speed
Commercial Speed Computation
I jk
=
sR
× 1
sijk
iå
N jk
for sijk
¹ 0
Source: Cortés et al 2011
56. Ricardo Giesen, BRT Workshop, Rio 2013
Definition of Speed Ranges sR=20[km/hr]
Condition Sijk [Km/h] Ijk Color
Very bad ≤ 15 ≥ 1.333 Red
Bad >15 a ≤19 < 1.333 to ≥
1.053
Orange
Barely
Acceptable
>19 a ≤20 < 1.053 to ≥
1.0
Yellow
Fair >20 to ≤25 < 1.0 to ≥ 0.80 Light Green
Good >25 to ≤30 < 0.80 to ≥
0.667
Dark Green
Excellent >30 < 0.667 Blue
Source: Cortés et al 2011
57. Ricardo Giesen, BRT Workshop, Rio 2013
Global Results (All Services)
Septembre 2008
March 2009
April 2009
Source: Cortés et al 2011
59. Ricardo Giesen, BRT Workshop, Rio 2013
Results for a Particular Service
Time-space
desegregated
60. Ricardo Giesen, BRT Workshop, Rio 2013
Results for a Particular Service (in the map)
61. Ricardo Giesen, BRT Workshop, Rio 2013
Allow
detecting
problems,
propose
solutions,
improve
management.
Results for a Particular Service (in the map)
62. Ricardo Giesen, BRT Workshop, Rio 2013
Real Time Control
• ADCS enabler measure reliability and its impact on individual pax
• Reliability metrics
• Contractor performance
• Performance from passenger’s point of view
• High frequency services
• Extensive vehicle interactions
• Most customers do not time their arrival to schedules
• On-time performance may not be as critical
• Schedules can be revised in real time
• Fleet management
• Communications
63. Ricardo Giesen, BRT Workshop, Rio 2013
Traveler Information
• The role of information
• Real-time information
▫ Location
▫ Comprehensiveness
▫ Type
64. Ricardo Giesen, BRT Workshop, Rio 2013
Customer Information
• Traditional
– Static
– Customer service call centers
– Pathfinding at stops / stations/ pedestrian access
• Initial ITS
– Displays at bus stops (scheduled arrivals / real-time ETA)
– Monitors at terminals
– Next stop information on-board vehicles (AVA)
Source: B. Hemily & A. Rizos
65. Ricardo Giesen, BRT Workshop, Rio 2013
Customer Information
• State of practice
– Web-based applications
• trip planning systems
• Google Transit trip planning
– Smart Phone Applications
• Static and dynamic information
• State of the art
– Social Media
• Facebook, Twitter
– Real-Time Information
– Open-Source Traveler Information Software Development
• Forthcoming
– Special Mobile Applications (e.g. customers with special needs)
– “Augmented Reality” and implementation in apps
• Combine compass, visual recognition, other tools
– Crowdsourcing (Waze for Buses?) Source: B. Hemily & A. Rizos
66. Ricardo Giesen, BRT Workshop, Rio 2013
Conclusion
• New automated data sources enable a range of
applications and services for improved level of
service and more efficient utilization of resources
• Lack of integration
– Databases
• Legacy systems
• Challenge going from data to information
• Level of “know – how”
Source: B. Hemily & A. Rizos
67. Ricardo Giesen, BRT Workshop, Rio 2013
Fare Collection in the Broader
Payments Environment
Ricardo Giesen
Pontificia Universidad Católica de Chile
BRT Workshop: Experiences and Challenges
Rio de Janeiro, July 2013
Hinweis der Redaktion
Passenger Flow and System CapacityPublic Transport OD Matrix EstimationPerformance Measurement (PM)Real time demand estimation and control Customer information