comparing aggregate trip- based and disaggregate tour-based travel demand models: columbus highway...
TRANSCRIPT
Comparing Aggregate Trip-Based and
Disaggregate Tour-Based Travel Demand Models:
Columbus Highway Results
Research Project Overview• Research led by University of Texas-
Austin with AECOM, John Bowman, Mark Bradley & Ram Pendyala
• Main objective: examine the performance of the trip-based and tour-based frameworks for Columbus in the context of a before-and-after project analysis
Major Tasks
• Update trip-based model to resolve differences in estimation datasets, TOD, geographic coverage and other areas
• Develop 1990, 2000 and 2005 scenarios (including socio-economic data, networks, traffic counts, etc.)
• Compare models’ regional-level results to Census (1990, 2000), Household Interview Survey (1999), ACS (2005)
• Decide on study projects• Compare models’ project level-results
Brief Model Comparison
Trip-Based Model Tour-Based Model
Unit of Travel Trip Tour
Travel Aggregation
Zonal-level Person-level
Estimation Datasets
GenerationDistributionMode Choice
1999 HIS1999 HIS1993 COTA On-Board
1999 HIS1999 HIS1993 COTA On-Board + 1999 HIS
Geographic Coverage
1877 zones across 7 counties
Time-of-Day FidelityGenerationDistributionMode ChoiceAssignment
DailyDailyPeak/off-peak4 time periods
1-hour blocks4 time periods4 time periods4 time periods
Base Year(s) 2000, 2005 2000, 2005
Validation - VMTTOUR MODEL Observed Traffic Modeled Traffic Percent Difference
Facility Type # Links CountCount VMT
VolumeModel VMT
Volume VMTMax.
% VMT%
RMSE
1 Interstate 155 7,557,083 7,716,241 7,554,692 7,404,023 0% -4% 7% 17%
2 Expressway 96 2,247,915 1,610,448 2,205,784 1,597,832 -2% -1% 10% 18%
3 Arterial 2,521 22,159,792 6,385,080 22,471,650 6,321,376 1% -1% 10% 32%
4 Collector 1,531 3,962,091 1,654,805 3,732,848 1,553,047 -6% -6% 15% 56%
5 Local 932 1,012,435 413,312 982,722 389,390 -3% -6% 15% 92%
Total 5,235 36,939,316 17,779,886 36,947,696 17,265,668 0.0% -2.9% 3% 37%
TRIP MODEL Observed Traffic Modeled Traffic Percent Difference
Facility Type # Links CountCount VMT
VolumeModel VMT
Volume VMTMax.
% VMT%
RMSE
1 Interstate 155 7,557,083 7,716,241 7,859,723 7,934,125 4% 3% 7% 13%
2 Expressway 96 2,247,915 1,610,448 2,313,518 1,676,435 3% 4% 10% 20%
3 Arterial 2,521 22,159,792 6,385,080 20,872,011 5,938,085 -6% -7% 10% 34%
4 Collector 1,531 3,962,091 1,654,805 3,673,546 1,537,975 -7% -7% 15% 57%
5 Local 932 1,012,435 413,312 948,324 390,552 -6% -6% 15% 92%
Total 5,235 36,939,316 17,779,886 35,667,122 17,477,172 -3.4% -1.7% 3% 37%
Validation - % RMSE
Model-to-Model Comparisons
• Trip distance by trip purpose
• Delta volume bandwidth plots– ∆ volume = Tour model – Trip
model
Comparison of Trip Length by Purpose
8
Delta Volume Bandwidth Plot1990
9
Delta Volume Bandwidth Plot2000
10
Delta Volume Bandwidth Plot2005
11
Regional-Level Results
• Vehicle ownership– Tour model performs better in Franklin
County– Trip model performs better in other counties
• Work flows– Both models generally perform the same,
except for inter-county movements where the tour model generally performs better
• Average work travel time– Both models generally perform the same
Findings from the Report
• Need to investigate why tour-based model systematically under-performs in vehicle ownership outside Franklin County
• Overall there are few major differences between the two models (slight overall edge to tour-based model?)
• It is difficult to make disaggregate model comparisons when the models have different units of travel– Translating units leads to inconsistencies at a disaggregate
level, making the tour-based model’s full range of potential benefits difficult to compare to trip-based models
• More comparisons between trip- and tour-based models are needed to verify these findings
Project-Level Analysis
• Polaris – IR 71/ SR 750 Polaris Parkway– Large retail and employment growth– Interchange and other roadway improvements
• Rome-Hilliard – IR 70/IR 270– Large land-use development– No roadway improvements
• Spring-Sandusky – downtown Columbus– No major land-use changes– Major roadway improvements
• Control area – IR 71 in southwest Columbus– No major land-use changes or roadway
improvements
All Study Areas
Spring-Sandusky Study
Area
Polaris Study Area
Hilliard-Rome Study Area
Control Area
Polaris - 1988
Polaris - 2008
Spring-Sandusky Study Area
19
20
Hilliard-Rome Study Area
22
23
Control Area
Volume to CountsPolaris – 1990
Red – Overassigned
Blue - Underassigned
Trip Tour
Volume to CountsPolaris – 2000
Red – Overassigned
Blue - Underassigned
Trip
Tour
Volume to CountsPolaris – 2005
Red – Overassigned
Blue - Underassigned
Trip Tour
Volume to CountsSSI – 1990
Red – Overassigned
Blue - Underassigned
TripTour
Volume to CountsSSI – 2000
Red – Overassigned
Blue - Underassigned
TripTour
Volume to CountsSSI – 2005
Red – Overassigned
Blue - Underassigned
Trip
Tour
Design Forecasts
• Add 1 more “model”– Fratared a matrix of 1s to the Trip Ends
from the Tour model
Volume/Counts and Forecasts
• Polaris
IR 71 South of SR 750Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 40400 48773 42827 145805 8373 2427 1054052000 104500 115352 101922 201255 10852 -2578 967552005 122730 139013 123209 257432 16283 479 134702
Growth Rate 13.9% 12.5% 12.7% 4.9% -1.4% -1.2% -9.0%
Volume Difference
IR 71 South of SR 750Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 404002000 104500 98749 93461 99583 -5751 -11039 -49172005 122730 127923 119991 129175 5193 -2739 6445
Growth Rate 13.9% 14.4% 13.1% 14.6% 0.5% -0.8% 0.7%
NCHRP 255 Forecast Difference
Volume/Counts and Forecasts
• Spring-Sandusky Interchange
IR 670 West of US 23Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 42718 52873 62615 104115 10155 19897 613972005 117860 126611 138121 278010 8751 20261 160150
Growth Rate 11.7% 9.3% 8.0% 11.1% -2.4% -3.7% -0.6%
Volume Difference
IR 670 West of US 23Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 427182005 117860 113499 112786 197411 -4361 -5074 79551
Growth Rate 11.7% 11.0% 10.9% 24.1% -0.7% -0.8% 12.4%
NCHRP 255 Forecast Difference
Volume/Counts and Forecasts
• Rome-Hilliard
Rome-Hilliard Rd. Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 14504 9349 8794 26238 -5155 -5710 117342005 23986 27551 23467 37630 3565 -519 13644
Growth Rate 4.4% 13.0% 11.1% 2.9% 8.6% 6.8% -1.5%
Volume Difference
Rome-Hilliard Rd. Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 145042005 23986 34409 30962 24120 10423 6976 134
Growth Rate 4.4% 9.1% 7.6% 4.4% 4.8% 3.2% 0.1%
NCHRP 255 Forecast Difference
Volume/Counts and Forecasts
• Control Area
IR 71 South of Stringtown Rd.Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 38804 34829 36823 46870 -3975 -1981 80662000 53780 50668 49657 66288 -3112 -4123 125082005 62460 56232 52769 67826 -6228 -9691 5366
Growth Rate 4.0% 4.2% 3.0% 3.1% 0.1% -1.1% -0.9%
Volume Difference
IR 71 South of Stringtown Rd.Count Trip Tour Trip Ends Trip Tour Trip Ends
1990 388042000 53780 53577 49635 51944 -203 -4145 -18362005 62460 60964 55050 58514 -1496 -7410 -3946
Growth Rate 4.0% 3.8% 2.8% 3.4% -0.2% -1.2% -0.6%
NCHRP 255 Forecast Difference
Speed ComparisonIR 71 South of Morse Rd
Speed Trip Tour Trip Ends Trip Tour Trip Ends Trip-TourNB 2000 62.06 2000 55.86 55.89 28.65 -6.2 -6.17 -33.41 -0.03
2009 69.30 2005 55.81 55.8 24.07 -13.49 -13.5 -45.23 0.011.3% 0.0% 0.0% -3.2% -1.3% -1.3% -4.5%
SB 2000 56.39 2000 55.85 55.94 34.45 -0.54 -0.45 -21.94 -0.092009 70.82 2005 55.72 55.88 27.39 -15.1 -14.94 -43.43 -0.16
Growth Rate 2.8% 0.0% 0.0% -4.1% -2.9% -2.9% -6.9%
Volume Difference
IR 71 North of Stringtown Rd.Speed Trip Tour Trip Ends Trip Tour Trip Ends Trip-Tour
NB 2000 65.96 2000 57.97 57.98 56.93 -7.99 -7.98 -9.03 -0.012009 66.73 2005 64.88 64.94 61.33 -1.85 -1.79 -5.4 -0.06
0.1% 2.4% 2.4% 1.5% 2.3% 2.3% 1.4%SB 2000 61.22 2000 57.98 57.99 57.58 -3.24 -3.23 -3.64 -0.01
2009 60.75 2005 64.93 64.95 63.58 4.18 4.2 2.83 -0.02Growth Rate -0.1% 2.4% 2.4% 2.1% 2.5% 2.5% 2.2%
Volume Difference
IR 670 E of Grandview AveCount Trip Tour Trip Ends Trip Tour Trip Ends Trip-Tour
EB 2000 70.6 2000 54 54 54 -16.6 -16.6 -16.6 02009 67.44 2005 55.9 55.95 51.96 -11.54 -11.49 -15.48 -0.05
-0.5% 0.7% 0.7% -0.8% 1.2% 1.2% -0.3%WB 2000 62.97 2000 54 54 54 -8.97 -8.97 -8.97 0
2009 65.81 2005 55.99 56 54.75 -9.82 -9.81 -11.06 -0.01Growth Rate 0.5% 0.7% 0.7% 0.3% 0.2% 0.2% -0.2%
Volume Difference
Conclusions
• With an aggregate assignment, there isn’t much difference between the demand models for your run-of-the-mill project traffic forecasts.
• Biggest difference is in what questions your model can answer
• Develop a model that answers questions that are being asked in your region.– Use your crystal ball to determine what
questions are likely to be asked over the next 20 years.
Contacts
• Greg Giaimo – ODOT – [email protected]
• Rebekah Anderson – ODOT – [email protected]
• Zhuojun Jiang – MORPC – 614-233-4147• Chandra Bhat – UT at Austin
• Dave Schmitt – [email protected]