Analysis of Time of Day Models from Various Urban Areas William G. Allen, Jr. Transportation Planning Consultant Windsor, SC TRB Transportation Planning Applications Conference Daytona Beach, FL 9 May 2007
Overview Comparisons among cities: can you borrow your neighbor’s TOD model? Peak spreading: a myth? NHB sub-purposes: worthwhile? Validation: it’s a good thing
Comparison Among Cities and Years Atlanta & 2002 Baltimore Charlotte New Orleans Reading, PA Washington, DC & 1994
TOD Analysis Data from home interview surveys, processed the same for each city 30 minute increments Separate by trip purpose and directionality Home to non-home and non-home to home Vehicle trips only Simple process for aggregate 4-step models
Survey Processing Summary of trips by “time in motion” Compute reported vehicle-minutes by trip Tabulate veh-minutes by 30 min. period Get fraction of VHT by period Group 30 min. periods as desired for assignment Apply fractions to daily vehicle trip table
Assignment Periods AMMDPMNTOP Atlanta45411 Baltimore36312 Charlotte36312 New Orleans37311 Reading37311 Washington3318
Atlanta – 1995 – Work
Atlanta – 2002 – Work
Baltimore – Work
Charlotte – Work
New Orleans – Work
Reading – Work
Washington – 1989 – Work
Washington – 1994 – Work
Observations on Work Trips Work trips are pretty regular AM peak usually higher than PM peak 7:30 – 8:00 AM is the highest half- hour everywhere Some pattern differences are logical: New Orleans: tourist-based economy Reading: shift workers Washington: regular pattern of government workers
Atlanta – 1995 – All Trips
Atlanta – 2002 – All Trips
Baltimore – All Trips
Charlotte – All Trips
New Orleans – All Trips
Reading – All Trips
Washington – 1989 – All Trips
Washington – 1994 – All Trips
Observations on All Trips Atlanta & Washington, over time Peaks get lower; midday/night higher People leave earlier, return later Comparisons among cities Washington & Baltimore similar PM high, AM less peaked Atlanta, Charlotte, Reading (?!) similar New Orleans unique
Peak Trip Share Declines AtlantaWashington Work6:30-9:30 a38%34%36%34% 3:30-6:30 p35%30%34%33% off-peak27%36%30%33% Other6:30-9:30 a14%9%12%15% 3:30-6:30 p24%23%39%25% off-peak62%68%49%60%
Peak Trip Share Declines – All Cities Work Trend Line Other Trend Line
Causes of Peak Spreading Increased traffic congestion Changing lifestyles More flex-time More part-time workers
Non-Home-Based Sub-purposes A fast-growing trip category NHB categories based on tour type JTW: home-other-work or work- other-home JAW: work-other-work NWK: home-other-other-other-home Trip generation and distribution are similar TOD is very different
Washington – 1994 – NHB JTW
Washington – 1994 – NHB JAW
Washington – 1994 – NHB NWK
Validation Compare link volumes to counts by assignment period This type of TOD model sometimes overestimates peak period volumes Reduce peak fractions, increase off- peak fractions until volumes ≈ counts A very necessary step Difficult to get counts Easy to adjust fractions
Conclusions Work trip patterns are generally consistent Some peak spreading over time Increased congestion is part of the reason Splitting NHB into sub-purposes is important for TOD This approach quickly produces a usable TOD model, but validation is important TOD models are not really transferable
Questions Please use Microphone