Justifying Rail Bias Factor for Houston METROs Transit Model

1
Justifying Rail Bias Factor for Houston METROs
Transit Model

• Presentation by
• Vijay Mahal, HDR Inc
• Vincent Sanders, Houston METRO
• May 18, 2009
• TRB Applications Conference
• Houston, Texas

2
Outline of Presentation

• Houstons Long range Transit Plan- MetroSolutions
  
• Houstons first Light Rail Line- Phase I
• Description
• Ridership profile
• Houston METROs travel model earlier version
• Need for Rail Bias factor
• Incorporation of Rail Bias
• Model application / Conclusions

3
Houston METROs Long Range
Plan

• Development of METROSolutions
• Voter referendum
• Different Phases
• The Redline (Phase 1)

4
Houstons first LRT opens January
2004
5
Houstons first Light Rail

• Starter line is 7.5 miles long
• 16 stations
• Serves very important activity centers (CBD, TMC,
  UH, HCC, Rice)
• High frequency
• Uses German-made Siemens cars
• Top speeds achievable 66 MPH

6

• Stations a few blocks apart in CBD and about one
  mile apart outside of CBD
• Uses Clean electric power
• Moves large volumes of people
• Operates faster than local buses

7
LRT Ridership Profile
8
Forecasts from Earlier Models

• LRT ridership forecasts made during the 1990s
• Opening year (2004) forecast
  22,000
• 2020
  forecast 40,000
• Actual ridership in September 2004
  33,000
• Actual current ridership
  44,000
• The Red Line is considered the most successful
  line in the country.
• Nov. 2007 Breaks 40 Million boardings
• Sept. 2008 Breaks 50 Million boardings
• So, what factors make this line so successful?

9
Travel Surveys

• 1st Survey in Sept 2004 (elaborate OD survey)
• 2nd Survey in November 2004 (Supplemental survey
  to collect data on new riders)
• 3rd Survey in 2007 (OD survey to update model
  parameters)

10
The 2004 O/D survey format

• 19 Question survey
• Basic demographics question were asked
• age, gender, income, ethnicity, available
  vehicles
• Access-Egress patterns, type of trips, transfer,
  types of fare media, how long often have you
  used transit
• One question not asked Were you a METRO
• user before METRORail?

11
Supplemental Survey Instrument
12
Highlights from Survey
Survey conducted from November 8th 2004  to
November 17th 2004 Total surveys handed out
13,000 Valid surveys returned 5,940  (very high
response) Percent of new riders (Question 1 of
survey) was 37 Were you a METRO bus/shuttle
rider before METRORail opened?
13
Houstonians Offer Positive Comments

• Survey response
• 90 comments were positive
• 3 neutral, 7 negative
• 36 wanted immediate system expansion
• Most cited reasons for positive response
• reliability
• comfort / smoothness of ride
• faster than bus

14
Houston METROs original Mode Choice
Model

• Light Rail data was never a part of Houston
  METROs mode
• choice model estimation
• Primary Data Sources
• 1985 Home Interview Survey
• 1990 Journey to Work Census data
• 1995 and 2001 On-board OD surveys
• Periodic ride check data
• LRT came into existence in January 2004
• None of the data set contained the rail
  attributes

15
Earlier Planning Studies

• Major Planning Studies for this region began in
  the mid 1990s
• For the Red line,
• Alternative Analysis Environmental documents in
  progress
• Travel forecasts were needed urgently for a mode
  that didnt exist
• Modifications made to the mode choice model to
  accommodate LRT (quick fix)

16
Structure of the original Mode
Choice Model
17
Structure of the revised Mode
Choice model
18
Mode-Specific Constants

• Utility equations
• Original Model
• U LB C1 x Inveh TT C2 x Out-of-veh
  TT C3 x Fare
• U CB K 1 C1 x Inveh TT C2 x Out-of-veh TT
  C3 x Fare
• Revised Model
• U LB C1 x Inveh TT C2 x Out-of-veh
  TT C3 x Fare
• U CB K 1 C1 x Inveh TT C2 x Out-of-veh TT
  C3 x Fare
• U EX K 2 C1 x Inveh TT C2 x Out-of-veh TT
  C3 x Fare
• U LRT C1 x Inveh TT C2 x Out-of-veh
  TT C3 x Fare

19
Purpose of Mode Specific constant

• Most models consider travel time, travel costs
  and some measurable service characteristics ( of
  Xfers)
• Visibility, comfort of ride, reliability,
  passenger amenities etc. are non-quantifiable
  and not considered
• In theory, mode specific constant is supposed to
  represent the non-quantifiable factors
• Misuse of Mode Specific constant correction
  factors (for calibration purposes)

20
Underestimation of LRT forecasts

• LRT forecasts were significantly underestimated
  when its mode bias constant was zero
• Opening year (2004) forecasts 22,000 daily
  boardings
• Horizon year (2020) forecasts 40,000 daily
  boardings
• Current (2009) actual ridership 44,000 daily
  boardings
• What was the most likely reason for the
  underestimation?
• Survey results clearly demonstrated Houstonians
  perceive LRT as a Premium mode. So, a mode
  specific constant was introduced to represent the
  Rail Bias.

21
Incorporating Rail Bias Factor

• Methodology
• Started with a rail bias factor of 0.33 used in
  the old Minneapolis and San Diego models. In
  terms of IVTT, this equals 6.73 min in our
  models.
• Run model with and without LRT network
• Analyze ridership and new trips and compare them
  to survey results
• Increase bias factor if necessary and repeat the
  process
• Final factor was 0.50 (equal to 10.2 minutes of
  IVTT)

22
Calibration Results (Yr
2004)
23
Model Application

• Application results

24
Rail Bias Impacts on New Starts
projectsThe North Corridor Extension
25
Rail Bias Impacts

• Ridership projections increased by about 20 to 25
  percent
• User Benefits increased by about 15 percent
• Distribution of user benefits improved
• Most diagnostic statistics from SUMMIT reasonable

26
Conclusions

• Travel surveys reveal Houstonians perceive LRT as
  a premium mode. Its ride quality, reliability,
  station amenities are as important to them as its
  travel times and frequency. They have a positive
  Bias towards rail.
• The rail bias factor in Houston is worth
  approximately 10 min of IVTT in comparison to
  local bus mode
• Rail Bias can improve User Benefits and
  ridership- in our case about 15 to 20 percent