Results of FY 08 COGTPB Travel Forecasting Research

1
Results of FY 08 COG/TPBTravel Forecasting
Research

• Presentation to Travel Forecasting Subcommittee
• September 19, 2008

Rich Roisman, AICP, Senior Transportation
Planner Hongtu Maggie Qi, P.E., Transportation
Engineer Jose Joe Ojeda, Transportation
Engineer Vanasse Hangen Brustlin,
Inc. rroisman_at_vhb.com Phil Shapiro, P.E.,
PTOE Shapiro Transportation Consulting,
LLC psshapiro_at_verizon.net
2
FY 08 TPB Travel ForecastingResearch Topics

• Expanded evaluation of peak spreading
• Estimating the impact of exurban commuters on
  travel demand

3
Expanded Peak spreading Analysis
4
FY 2007 Peak Spreading Analysis

• State of the art and state of the practice review
• Initial evaluation of traffic count data
• Determine data availability
• Evaluated peak spreading in TPB area
• Relationship of peak v/c to of lanes
• Proposed approaches for TPB model

5
Expanded 2008 Analysis

• Omitted use of V/C ratio
• Obtained MD SHA historical hourly traffic count
  data from ATR sites
• Comparable VDOT and DDOT data not available

6
Compared Ratio of Hourly to Peak Hour Volume by
Year
7
Compared Volume Through AM Peak Period by Year
8
Compared Volume Through PM Peak Period by Year
9
Volume per Lane Relationship

• Peak hour volume per lane divided byADT volume
  per lane
• Radial freeways
• I-270
• I-95
• US 50
• Combined
• Beltway in Prince Georges County

10
AM Peak Hour Regression All Radial Facilities
r2 .726
11
PM Peak Hour Regression All Radial Facilities
r2 .598
12
Potential for Radial Freeways

• When peak hour capacity reached trend emerges
• As ADT/Lane increases peak hour goes down
• Breaks down when ADT/lane 29,000
• Value to TPB and member agencies
• Estimate peak used to determine ADT capacities
  for model assignment
• Project Planning peak hour volumes

13
ADT / Lane and Peak Hour PercentRadial Freeways
14
Regression for Capital Beltway in Prince Georges
County

• Results not as promising as for radial freeways
• NB AM Peak Hour r20.30492
• NB PM Peak Hour r20.36988
• SB AM Peak Hour r20.02074
• SB PM Peak Hour r20.08723

15
Regression for Capital Beltway Prince Georges
County

• r2 - not very good fit
• Off-peak direction regression line slopes up
  instead of down
• Probably due to significant available off-peak
  capacity in Prince Georges County segment
• Further analysis needed to find useful
  relationship

16
Future steps for TPB

• Obtain more good hourly count data
• Peak Hour and ADT
• VA, DC Other MD Locations
• Freeway and major arterials
• Test Beltway locations with higher peak volumes
• Test arterial roadways
• Radial
• Circumferential

17
Estimating the Impact of Exurban Commuters on
Travel Demand
18
Estimating the Impact of Exurban Commuters on
Travel Demand

• Purpose to allow continual evolution of TPBs
  forecasting methods
• Identification of exurban travel patterns to the
  TPB region
• Literature review
• Data review
• Forecasting external trips using regression
  equations

19
Background

• FY 06 research on MPO methods of external
  forecasts
• Initial review of data indicates TPB region
  experiences high level of E-I travel
• E-I travel from workers outside the region
• Extreme commuters who live more than 100 miles
  away from the Capitol
• Impacts on travel forecasting and transportation
  planning are significant

20
Long Travel Times Experienced by Area Workers
Well-Documented by 2003 ACS

• Average county commute time rankings
• Prince William and Prince Georges counties
  exceeded only by four outer boroughs of NYC
• Fairfax County (21) and District of Columbia (45)
  also in top 50 nationally for average commute
  times
• Among state rankings
• Maryland (2) District (4) Virginia (9)

21
2003 ACS Also HighlightsExtreme Commuters

• Census defines as traveling 90 minutes or more
  (one-way) to work
• Nationally, only 2 of workers face extreme
  commutes
• Prince William 4.5 Prince Georges 3.8
  Montgomery 2.2
• Maryland 3.2 Virginia 2.3 District 2.2

22
Consider West Virginia

• Eastern panhandle part of new frontier for
  Washington-area workers
• Ranked 12th in commute times nationally
• Largest increase in average commute times between
  1990 and 2000
• Jefferson County part of TPB modeled area
• Berkeley County outside modeled area, growing
  source of E-I trips
• Many other nearby jurisdictions like Berkeley
  County

23
Results of Literature ReviewPopular Press

• Covered extreme commuting heavily following
  release of ACS data
• ACS data as jumping-off point to cover more
  extreme commuters
• Northeastern PA as commute shed for NYC
• Antelope Valley to Los Angeles
• Relationship between transportation and housing
  costs
• Even a recent documentary film on extreme
  commuting

24
Results of Literature ReviewProfessional /
Academic Press

• TTI Urban Mobility Report used as baseline data
  for popular press articles
• Transportation and housing costs as metric for
  regional affordability
• Can areas essential workers can afford to live
  there?
• When answer is no, likelihood of extreme
  commuting higher
• Pisarski, Commuting in America III

25
Results of Literature ReviewCommuting in
America III

• Several key findings
• Increases in the proportion of workers traveling
  60 mins and 90 mins to work
• Increases in the percentage of workers leaving
  before 6 AM
• Nationally, about 11 of work trips to the city
  center arrive from outside the metropolitan area
• CIA III also asks will long-distance commuting
  continue to expand?

26
Data Analysis

• Review of CTPP and BEA data
• Time series comparison
• Comparable regions (CMSA)
• NYC, Atlanta, San Francisco, Los Angeles
• Forecasting external trips

27
1970
TPB Modeled Region
28
1980
TPB Modeled Region
29
1990
TPB Modeled Region
30
2000
TPB Modeled Region
31
External Travel 1970-2000
32
Forecasting External Trips

• By type of jurisdiction
• Central, Inner Suburb, Outer Suburb, etc.
• Regression equations tested
• External trips vs. (employment minus workers) by
  jurisdiction
• Strong relationship for Central jurisdictions
• Less strong for outer areas
• Less strong for areas with more employment than
  workers

33
Results Central Jurisdictions
34
Results Fredericksburg area and Other
Jurisdictions
35
Results for Areas with More Employment than
Workers
36
Implications for TPBModeling Process

• Conclusions of literature based on cheap
  gasoline
• Impact of further price increases?
• Predictive equations can be used for forecasting
  external travel in some areas
• Further analysis using this data set should be
  performed

37
Implications for TPBModeling Process (2)

• Backcasting using predictive equations as
  additional test
• Challenge of data availability
• E-I market will continue to grow
• TPB model must do a reasonable job at capturing
  these trips

38
Questions?