Making Headways
Smart Card Fare Payment and Bus Dwell Time in Los Angeles
Daniel Shockley
Fehr & Peers
Julia Salinas
Los Angeles Metropolitan Transportation Authority
Brian D. Taylor
UCLA Institute of Transportation Studies
Transportation Research Board
2016 Annual Meeting
Washington, D.C.
Agenda
• Overview
• Hypothesis
• Methodology
– Data Sources
– Route Selection
– Exclusions
• Analysis
– Variables
– OLS Regression
• Findings & Interpretation
• Conclusions
Overview: Los Angeles Metro
Metro Rail
350,000 average weekday boardings
Six lines (four light rail and two heavy
rail)
80 Stations (26 under construction)
87 miles of track
Five extensions currently under
construction
Overview: Los Angeles Metro
Metro Bus
Approx. one million average weekday
boardings.
Local Service: Frequent stops and
infrequent headways.
Rapid Service: Infrequent stops and
frequent headways.
Bus Rapid Transit: Two lines operating in
exclusive right-of-way
Overview: Transit Access Pass (TAP)
Smart Card Fare Payment System
Stored cash value or pass.
Accepted at 24 transit systems in Los
Angeles County.
Required for Metro Rail.
Dwell Time
“the amount of time a transit vehicle spends at stops and stations serving passenger movements”
Transit Capacity and Quality of Service Manual (TCQM)
Dwell Time
Bus Transit Route
Capacity
Bus Loading
Area
Clearance Time
Bus Time
Variability
Bus Stops
Failure Rate
Bus Facilities
Dwell Time
Passenger Demand and Loading
Bus Stop Spacing
Fare Payment Procedures
In-vehicle Circulation
Research Question & Hypothesis
Question: All other factors held constant,
what is the influence of the TAP card on
transit bus dwell times?
Hypothesis: TAP card usage can help to
reduce bus transit dwell time by reducing the
amount of time to board per person.
Method: Ordinary Least Squares (OLS)
regression analysis with Dwell Time as the
dependent variable, while controlling for as
many other determinants of dwell time as
possible using the data at hand.
Why is this important?
… lowers operating cost per route.
… lowers headways per route.
… reduces passenger waiting.
… attracts more riders to faster service.
Time saved per stop…
Methodology: Sources
APC - Automatic Passenger
Counter
– Alighting/boarding
– Load factor
– Dwell time
– New data points for each stop.
UFS – Universal Farebox System
– TAP/Cash fare payments
– Bicycle, wheelchair tallies, etc.
– New data points for each fare
paid/tally recorded.
Methodology: Route Selection
Metro Rapid 720
Infrequent stops, frequent headways.
Avg. weekday ridership: 41,000
Avg. Saturday ridership: 29,000
Avg. Sunday ridership: 22,000
Serves many employment centers with
connections to rail transit.
Downtown LA
Methodology: Route Selection
Metro Local 120
Frequent stops, infrequent headways.
Avg. weekday ridership: 4,000
Avg. Saturday ridership: 2,000
Avg. Sunday ridership: 2,000
Serves mostly residential and major physical
rehabilitation center. Connection to Metro Rail.
Methodology: Constructing the Data
UFS
Record
1
• TAP Fare
payments
APC
Record
UFS
Record
2
UFS
Record
3
• Bicycles,
wheelchairs,
etc.
Constraints:
• Operator-dependent tallies may not be accurate.
• UFS and APC clocks may not be synchronized.
• Non-TAP Fare
payments
Methodology: Exclusions
•
•
•
•
Minimum Passenger Service Time (PST) < .5 second
Dwell time is zero
Stops at layovers, terminus, and time points.
Abnormally long dwell time >= 180 seconds
Route
PST< .5s
Dwell Time = 0s
Layovers, etc.
720
120
6
2
3,361
454
14,472
4,838
Dwell Time ≥
180s
2,477
104
Total
20,316
5,489
Grand Total 25,805
Methodology: Summary of Data
342 operators
187 vehicles
540,407 farebox records
99,453 APC records (N)
Analysis: Descriptive Statistics
Dwell Time (sec.)
Passenger Service Time (sec.)
Ons (#)
Offs (#)
Ons (no UFS) (#)
Offs (Offs > Ons) (#)
Dwell Load (#)
TAP Fare (#)
Non-TAP Fare (#)
TAP (Sale of Value or Pass) (#)
Wheelchairs (#)
Bikes (#)
Mean
26.7
7.3
3.4
3.6
0.7
2.1
22.9
2.2
0.7
0.0
0.0
0.0
Median
17.0
5.0
2.0
2.0
0.0
0.0
18.0
1.0
0.0
0.0
0.0
0.0
N = 99,453
Std. Deviation
25.6
11.1
4.8
4.8
1.6
3.8
19.0
3.6
1.6
0.2
0.1
0.1
Min.
1.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Max.
180
180
52
65
44
65
107
59
31
5
3
3
Analysis – Controlling for other factors
• Passenger Activity
–
–
–
–
–
Ons (no UFS)
Offs (Offs > Ons)
Dwell load
Bikes and wheelchairs loading and unloading
Abnormally long passenger boarding (>18s for one passenger)
• Service & Vehicle Characteristics
–
–
–
–
Peak hour service
Night-time service
Bus type (low/high floor/articulated/wide doors)
Service type (rapid/local)
Findings
• People paying with TAP
Cards take less time to
board.
• Articulated buses
experience shorter
dwells than nonarticulated buses.
• Rapid routes had
longer dwell time than
local routes.
Variable
(Constant)
Ons (no UFS)
Offs (Offs > Ons)
TAP Fare
Non-TAP Fare
TAP (Sale of SV or Pass)
Fares in Grace Period
Wheelchairs
Bikes
Dwell Load
Peak Hour (1=Yes/0=No)
Night Time (1=Yes/0=No)
B
11.5*
3.8*
0.8*
2.7*
4.6*
9.0*
-2.6*
36.9*
4.5*
-0.01
-1.0*
-2.1*
Std. Error
2.8
0.0
0.0
0.0
0.0
0.3
0.1
0.6
0.7
0.0
0.1
0.1
Beta
Articulated Bus (1=Yes/0=No)
-3.3*
Service Type
(1=Rapid/0=Local)
Wide Doors (1=Yes/0=No)
Low Floor (1=Yes/0=No)
Abnormal Boarding
(1=Yes/0=No)
0.2
0.1
0.4
0.3
0.1
-0.1
0.2
0.0
0.0
0.0
0.0
T
4.1
100.2
46.4
130.1
100.7
29.0
-41.3
65.7
6.9
-1.9
-7.9
-15.6
Sig (p)
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.06
0.0
0.0
1.2
-0.1
-2.7
0.0
6.1*
1.2
0.1
5.0
0.0
0.4
1.0
0.8
2.9
0.0
0.0
0.5
0.3
0.6
0.7
24.0*
0.5
0.1
50.2
0.0
* Significant at the .001 Confidence Level
Adjusted R-Square: .45
N = 99,453
Findings
Passenger Congestion
Variable
(Constant)
Ons (no UFS)
Offs (Offs > Ons)
TAP Fare
Non-TAP Fare
TAP (Sale of SV or Pass)
Fares in Grace Period
Wheelchairs
Bikes
Dwell Load
Peak Hour (1=Yes/0=No)
Night Time (1=Yes/0=No)
Filtering the sample to records
with a load factor of 1 or higher.
• TAP fare payments take longer,
however are still less than NonTAP.
• Articulated busses reduce
dwell time more than in prior
Articulated Bus (1=Yes/0=No)
model.
Service Type
(1=Rapid/0=Local)
Wide Doors (1=Yes/0=No)
Irregular Passenger
(1=Yes/0=No)
Low Floor
B
9.0*
3.1*
1.0*
3.0*
4.0*
5.9*
-1.7*
42.5*
1.8
0.04
0.3
-2.1*
Std. Error
3.0
0.1
0.1
0.1
0.2
1.3
0.2
2.2
2.1
0.0
0.4
0.5
Beta
0.3
0.1
0.5
0.3
0.0
-0.1
0.2
0.0
0.0
0.0
0.0
T
3.0
32.2
15.3
48.2
26.3
4.5
-6.9
19.2
0.9
1.6
0.6
-4.4
Sig (p)
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.4
0.1
0.6
0.0
-11.9*
5.0
-0.1
-2.4
0.0
14.1*
4.3
0.1
3.3
0.0
0.4
3.8
0.0
0.1
0.9
15.8*
2.7
0.0
5.8
0.0
-
* Significant at the .001 Confidence Level
Adjusted R-Square: .49
N = 7,327
Conclusion
1. People paying with TAP contribute fewer seconds to dwell
time, which can equate to large benefits later.
2. On a per-stop level, other factors seemed more important.
3. Technology can be improved to assist future analyses.
Thank You!
Contact:
Daniel Shockley - d.shockley@fehrandpeers.com
Photo Credits
Metro local bus 2 - Jonathan Riley
https://flic.kr/p/r4AEzy
720 - Oran Viriyincy
https://flic.kr/p/qcdZ71
Metro Rail – Steve and Julie
https://flic.kr/p/bDZRtC