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
Bus Stops Bus Facilities
Area

Bus Time
Clearance Time Failure Rate Dwell Time
Variability

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 UFS – Universal Farebox System
Counter – TAP/Cash fare payments
– Alighting/boarding – Bicycle, wheelchair tallies, etc.
– Load factor – New data points for each fare
– Dwell time paid/tally recorded.
– New data points for each stop.
Methodology: Route Selection

Metro Rapid 720

Infrequent stops, frequent headways. Downtown LA
Avg. weekday ridership: 41,000
Avg. Saturday ridership: 29,000
Avg. Sunday ridership: 22,000

Serves many employment centers with

connections to rail transit.
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 • TAP Fare
Record
payments
1

APC UFS
Record
• Non-TAP Fare
payments
Record 2

UFS • Bicycles,
Record wheelchairs,
3 etc.
Constraints:
• Operator-dependent tallies may not be accurate.
• UFS and APC clocks may not be synchronized.
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. Dwell Time ≥ Total
180s
720 6 3,361 14,472 2,477 20,316
120 2 454 4,838 104 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
Mean Median Std. Deviation Min. Max.
Dwell Time (sec.) 26.7 17.0 25.6 1.0 180
Passenger Service Time (sec.) 7.3 5.0 11.1 0.0 180
Ons (#) 3.4 2.0 4.8 0.0 52
Offs (#) 3.6 2.0 4.8 0.0 65
Ons (no UFS) (#) 0.7 0.0 1.6 0.0 44
Offs (Offs > Ons) (#) 2.1 0.0 3.8 0.0 65
Dwell Load (#) 22.9 18.0 19.0 0.0 107
TAP Fare (#) 2.2 1.0 3.6 0.0 59
Non-TAP Fare (#) 0.7 0.0 1.6 0.0 31
TAP (Sale of Value or Pass) (#) 0.0 0.0 0.2 0.0 5
Wheelchairs (#) 0.0 0.0 0.1 0.0 3
Bikes (#) 0.0 0.0 0.1 0.0 3
N = 99,453
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)
 Variable B Std. Error Beta T Sig (p)
(Constant) 11.5* 2.8 4.1 0.0
Ons (no UFS) 3.8* 0.0 0.2 100.2 0.0
Findings Offs (Offs > Ons)
TAP Fare
0.8*
2.7*
0.0
0.0
0.1
0.4
46.4
130.1
0.0
0.0
Non-TAP Fare 4.6* 0.0 0.3 100.7 0.0
TAP (Sale of SV or Pass) 9.0* 0.3 0.1 29.0 0.0
• People paying with TAP Fares in Grace Period -2.6* 0.1 -0.1 -41.3 0.0
Cards take less time to Wheelchairs 36.9* 0.6 0.2 65.7 0.0
board. Bikes 4.5* 0.7 0.0 6.9 0.0
Dwell Load -0.01 0.0 0.0 -1.9 0.06
• Articulated buses Peak Hour (1=Yes/0=No) -1.0* 0.1 0.0 -7.9 0.0
experience shorter Night Time (1=Yes/0=No) -2.1* 0.1 0.0 -15.6 0.0
dwells than non- Articulated Bus (1=Yes/0=No) -3.3* 1.2 -0.1 -2.7 0.0
articulated buses.
Service Type
6.1* 1.2 0.1 5.0 0.0
• Rapid routes had (1=Rapid/0=Local)
longer dwell time than Wide Doors (1=Yes/0=No) 0.4 0.8 0.0 0.5 0.6
Low Floor (1=Yes/0=No) 1.0 2.9 0.0 0.3 0.7
local routes. Abnormal Boarding
24.0* 0.5 0.1 50.2 0.0
(1=Yes/0=No)
* Significant at the .001 Confidence Level
Adjusted R-Square: .45
N = 99,453
 Variable B Std. Error Beta T Sig (p)
(Constant) 9.0* 3.0 3.0 0.0
Ons (no UFS) 3.1* 0.1 0.3 32.2 0.0
Findings Offs (Offs > Ons)
TAP Fare
1.0*
3.0*
0.1
0.1
0.1
0.5
15.3
48.2
0.0
0.0
Passenger Congestion Non-TAP Fare 4.0* 0.2 0.3 26.3 0.0
Filtering the sample to records TAP (Sale of SV or Pass) 5.9* 1.3 0.0 4.5 0.0
with a load factor of 1 or higher. Fares in Grace Period -1.7* 0.2 -0.1 -6.9 0.0
Wheelchairs 42.5* 2.2 0.2 19.2 0.0
• TAP fare payments take longer,
Bikes 1.8 2.1 0.0 0.9 0.4
however are still less than Non- Dwell Load 0.04 0.0 0.0 1.6 0.1
TAP. Peak Hour (1=Yes/0=No) 0.3 0.4 0.0 0.6 0.6
• Articulated busses reduce Night Time (1=Yes/0=No) -2.1* 0.5 0.0 -4.4 0.0
dwell time more than in prior
Articulated Bus (1=Yes/0=No) -11.9* 5.0 -0.1 -2.4 0.0
model.
Service Type
14.1* 4.3 0.1 3.3 0.0
(1=Rapid/0=Local)
Wide Doors (1=Yes/0=No) 0.4 3.8 0.0 0.1 0.9
Irregular Passenger
15.8* 2.7 0.0 5.8 0.0
(1=Yes/0=No)
Low Floor - - - - -
* 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