Data Driven Approach - 15th TRB National Transportation Planning

Report
Simplified Data Driven Methods for FTA
Small Starts Project Evaluation
Jeffrey Roux (AECOM)
Prasad Pulaguntla (AECOM)
May 8, 2013
Agenda
• Richmond Broad Street BRT Project Overview
• The Challenge
• Approach
• Bump on the Road
• Project extensions
• Lessons Learned
• Acknowledge MANY contributions to the study
SERPM 6.7
February 17,
Page 2
Broad Street Corridor
• Broad Street is THE bus
corridor in Richmond, VA with
17,000 daily transit trips.
• Broad Street serves as main
GRTC artery in Downtown
• Today: bus bunching, substandard lanes and accidents
result in a relatively slow and
unreliable journey.
• Full BRT saves up to 3 min.
on local bus trips and 6 min.
on BRT/express bus trips
Page 3
Problem Statement
• Existing conditions
– 48 buses/direction/peak hour within downtown RIC
– Current service is “local” stopping every block
– Narrow Parking lane (8’) dedicated to buses in peakperiod (with poor enforcement)

Problems
– High bus volumes with multiple routes entering the
trunk line lead to bus bunching and low reliability
– Existing narrow downtown bus lanes cause conflicts
and accidents
– Conga-line conditions leads to very slow travel times
Page 4
Goals and Objectives
•Provide more attractive transit service
– Improve travel times
– Reduce accidents
– Improve customer experience/amenities
– Improve system efficiency
– Improve reliability
– Improve convenience, efficiency of transfers
– Expand market
Page 5
Project Tactics
• BRT trunk line to provide a higher quality transit service:
– Higher quality, low floor, branded vehicles
– Formal BRT transit stops
• Formal branded shelters
• Off-board fare collection
– Dedicated bus lanes for half of corridor to speed trips
– Traffic signal priority
– Trunk line BRT is 6 min faster to Downtown (vs. express bus without
priority treatments) and 14 min faster compared to local bus.
• Consolidated bus stops in Downtown for ALL buses
– Downtown operations consolidate to three stops
– Multi-platform boarding
– Local buses are 3 min. faster through Downtown
Page 6
The Challenge
• Prepare patronage forecast & FTA Project
Justification materials for Small Starts application
• Off the shelf tools (in 2010):
– VDOT Richmond/Tri-Cities Forecasting Model (RTFM):
• Highway focused forecasting model
• Very good representations of highway travel speeds
• Basic transit network and limited mode choice
capabilities
– Fall 2009 GRTC On-Board Survey
• Approach: Develop data-driven approach from
RTFM networks and on-board surveys
Page 7
Data Driven Approach – Two Elements
• Transit Networks:
– Recoded RTFM Transit Networks (CUBE
TP+/TRNBUILD)
– Calibrated bus travel times from VDOT peak/off-peak
highway travel times (equations of motion)
• On-Board Transit Survey – Trip Tables:
– VDRPT surveyed GRTC routes in Fall ’09
– Comprehensive picture of GRTC existing markets
• Approach: Measure how existing riders (surveyed)
impacted by project and estimate new riders through
representation of transit network changes
Page 8
Transit Network Development
Recoded RTFM Transit Networks to Reflect Nov. ‘09
Service (survey):
– Separate peak/off-peak transit networks:
• Peak (7-9 AM)
• Off-Peak (10-2 PM)
– Correct route routing (including loops through D’town)
– Zone splits in project corridor
– Properly reflect stop locations
– Developed bus speed relationships, using equations of
motion based approach:
• AM Peak – Congested highway travel times
• Off-Peak – Free flow highway travel times
Page 9
Bus Travel Time Validation
Approach: Accurately represent bus travel speeds by
pivoting off of RTFM highway speeds
Process built “equation of motion” approach to replicate
current scheduled travel times of all buses:
– Used typical bus acceleration/deceleration rates
– Calibrated an average “dwell” times per stop
– Process worked very well as virtually all buses are within 15% or (3-4
min.) of scheduled end-to-end running times
Page 10
2009 GRTC On-Board Survey
4,500 completed surveys representing 32,900
GRTC boardings:
– Included VCU and non-VCU bus routes
– For approach we processed:
• Trip Ends (Productions and Attractions)
• Mode of Access (PNR, drop-off and walk)
• Time of Day (peak & off-peak)
• Converted boardings to linked trips
– Conducted on all routes, except inter-city routes:
• Petersburg-Downtown Richmond Express
• Fredericksburg-Ashland-Downtown Richmond Express
• Not material to Project
Page 11
2009 GRTC On-Board Survey Trip Productions
Page 12
2009 GRTC On-Board Survey Trip Attractions
Page 13
“Traditional” Survey Tabulations
Table 1: GRTC Regional Bus Service Year 2009 Average Weekday Linked Bus Trips
Table 2: VCU Bus Service Year 2009 Average Weekday Linked Bus Trips
Home Based Work (HBW) Trips
Home Based Work (HBW) Trips
Period
Peak Period
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
Vehicles/Household
0 Car HH 1+ Car HH
2,236
1,972
71
1,449
2,307
3,421
2,541
1,714
72
146
2,613
1,860
4,920
5,281
Total
4,208
1,520
5,728
4,255
218
4,473
10,201
Period
Peak Period
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
Home Based Other (HBO) Trips
Period
Peak Period
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
Grand Total
Total
138
218
356
68
73
141
497
Vehicles/Household
0 Car HH 1+ Car HH
85
346
0
71
85
417
253
629
0
91
253
720
338
1,137
Total
431
71
502
882
91
973
1,475
Vehicles/Household
0 Car HH 1+ Car HH
151
780
13
726
164
1,506
245
924
21
632
266
1,556
430
3,062
Total
931
739
1,670
1,169
653
1,822
3,492
Home Based Other (HBO) Trips
Vehicles/Household
0 Car HH 1+ Car HH
1,359
913
35
145
1,394
1,058
2,696
2,021
71
101
2,767
2,122
4,161
3,180
Total
2,272
180
2,452
4,717
172
4,889
7,341
Period
Peak Period
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
Non Home Based (NHB) Trips
Period
Peak Period
Vehicles/Household
0 Car HH 1+ Car HH
10
128
4
214
14
342
36
32
0
73
36
105
50
447
Non Home Based (NHB) Trips
Vehicles/Household
0 Car HH 1+ Car HH
873
406
28
245
901
651
1,467
1,153
72
103
1,539
1,256
2,440
1,907
11,521
10,368
Total
1,279
273
1,552
2,620
175
2,795
4,347
Period
Peak Period
Access Mode
Walk
Drive
SubTotal
Off-Peak Period Walk
Drive
SubTotal
Total
21,889
Grand Total
Page 14
818
4,646
5,464
Development of a Base Year 2009 Transit Trip Table
• Developed CUBE/TP+ Survey-Based Trip Table (Fall ‘09)
– Stratified by:
• Time of Day (Peak and Off-Peak)
• Mode of Access:
– PNR & Walk
– KNR treated like walk (with drop-off location treated as PROD
end)
• VCU and non-VCU trips stratified
• VCU Trips Stratified Separately
– Operates under contract to GRTC
– Free w/valid student/staff ID
– Main to Downtown Campus shuttles & fringe parking
– Policy of VCU students on BRT unknown
Page 15
Calibration of Transit Path Weights
• AECOM calibrated TRNBUILD path-building and
assignment weights through iterative assignment of transit
trip table
• Started with FTA “national experience” weights and tuned
• Final weights used were:
–
–
–
–
–
–
IVTT Weight = 1.0
Waiting/Transfer Waiting Time Weight = 1.5
Walk Time Weight = 2.0
Drive Time Weight = 2.0
Transfer penalty = 6.0 min per transfer
Walk speed = 3 mph
Page 16
2009 Weekday Modeled vs. Observed by GRTC Route
Group
Local Routes
Route Number
1/2
3/4
5
6
7
8
10
11
13
16
18
19
20
22
24
32
34
53
37
61
62/63
67
68
70/71
72/73
74
91
92
93
100
999 - Express/BRT
Subtotal
2009 Surveyed
Boardings
2,343
1,658
3625
1,010
1,856
155
312
630
249
294
32
327
430
2,176
1,649
1,902
3,304
19
1,406
1,457
902
219
17
63
26,035
2009 Survey
Assignment
2,275
3,079
3,639
1,391
1,172
56
260
384
333
121
556
2,520
1,835
1,676
2,840
155
1,373
1,301
568
367
27
25,928
Group
Express Routes
VCU Routes
Route Number
26
27
28
29
64
65
66
81
82
Subtotal
84
86
87
95
99
Subtotal
Total
Page 17
2009 Surveyed
Boardings
196
228
26
464
214
176
308
1,612
3,162
987
804
2009 Survey
Assignment
480
178
84
275
242
169
82
371
1,881
4,953
3,248
862
374
577
5,061
32,600
32,870
Developing Project Forecast/FTA Justification Materials
• 2009 demand grown to represent 2015 no-build using MPO estimates
of POP growth
• Development of CUBE/TP+ process where calibrated path-weight
parameters are multiplied to create an equivalent IVTT “transit
impedance” score for each I-J pair
• Alternative specific effects for BRT built into the transit impedence
– 5% discount for BRT IVTT
– 5 min. “constant effect” for BRT only riders, 2 min for BRT & Local Bus Riders
• Compare the “transit impedance” between alternative and use an
elasticity range of -0.3 (low) to -0.7 (high) to estimate “new” riders
through alternative progression (No-Build – TSM Baseline – Build)
• Assign the resulting transit trip tables to the network for project ridership
forecasts
• Compare resulting transit impedance to estimate Transit System User
Benefits
Page 18
Bump in the Road – New PNR Lot
Park-and-Ride at Staples Mill (near end of line)
– Scale is small (about 100 spaces)
– No existing PNR behavior in corridor - PNR’s are in fringe suburban
areas feeding express buses to Downtown
– Since modest number of spaces, we assumed:
• 75% utilized (additional new riders, about 150)
• Calculate UB’s by treating these riders as “new” riders (i.e. half the benefit)
Not sophisticated, but how far could we be off?
Bigger transformation would require more sophisticated
approach.
Page 19
Project Extensions
• GRTC used on-board survey data to plan their 2015 bus network
• Refinements to transit networks/procedures were embedded in the
VDOT Richmond/Tri-Cities Forecasting Model update (2012)
– Bus speeds
– Network coding
– Path/assignment weights
• On-Board survey used for RTFM mode choice calibration targets
• Comprehensive on-board transit survey allowed us to develop
procedures to simulate “non traditional” trips:
– Drop-off transit trips to non-formal locations
– VCU fringe parking trips
Page 20
Summary/Wrap-Up
•
Small Starts project justification requires forecaster to tell
the story of the project “who will benefit and why”
•
Data driven approach is ideal for developing patronage
forecasts for incremental transit service improvements:
–
–
–
–
Routed in real traveler patterns, not models
Transparent & easy to articulate benefits of projects
Easy to describe cause and effect in the forecasts
Forecaster can articulate risk by segmenting:
•
•
•
What is “known” from the survey data (X riders in the corridor who will
save at least 5 min. from the project)
What is not “known” (penetration of choice market, given BRT
introduction)
FTA was very supportive of approach
Page 21
Acknowledgements
Virginia Department of Rail & Public Transportation – Amy Inman
GRTC – Larry Hagin and Scott Clark
VDOT – Rick Tambellini, Juyin Chen and Paul Agnello
FTA – Jim Ryan, Ken Cervenka, Nazrul Islam
AECOM – Jeff Bruggeman
Connectics Transportation Group – Jim Baker
Michael Baker – Lorna Parkins, Scudder Wagg
Parsons Transportation Group – Gibran Hadj-Chikh
VHB - Ram Jagannathan
Project “Graduates” – Chandra Khare, Abishek Komma, Jeremy Raw, Bill Woodford, Mike
Lambert, Frank Spielberg and Sharon Hollis
Page 22
Thank You!
[email protected]
[email protected]

similar documents