may, 07, 2013 1 the corradino group, inc., 14th trb planning applications conference 14 th trb...

May, 07, 2013 1The Corradino Group, Inc., 14th TRB

Planning Applications Conference

1414thth TRB Planning Applications Conference TRB Planning Applications Conference

Columbus, OhioColumbus, Ohio

1

Developing an Alternate Approach to the use of Manual Post-Processing Techniques in Traffic Forecasting- A Case Study on Managed Lanes Forecasting on I-95 Corridor in Southeast Florida

Presented by Srin Varanasi



OutlineOutline

Design Traffic Forecasting NeedsDesign Traffic Forecasting Needs Traditional ApproachTraditional Approach Manual Post-Processing IssuesManual Post-Processing Issues Journey to an Alternate ApproachJourney to an Alternate Approach Case-Study ExampleCase-Study Example



Design Traffic Forecasting Design Traffic Forecasting NeedsNeeds

AADTAADT DDHVDDHV Traffic Factors, K and DTraffic Factors, K and D



Traditional ApproachTraditional Approach

Select a Travel Demand ModelSelect a Travel Demand Model Perform Subarea ValidationPerform Subarea Validation Develop Forecasts Develop Forecasts Apply Traffic FactorsApply Traffic Factors Ad hoc, NCHRP 255-style Post-Ad hoc, NCHRP 255-style Post-

ProcessingProcessing Develop Design TrafficDevelop Design Traffic



Manual Post-Processing Manual Post-Processing IssuesIssues

Manual Errors in Large-Scale projectsManual Errors in Large-Scale projects Forecasting Consistency IssuesForecasting Consistency Issues Heavy Relying on Traffic Factors- K, DHeavy Relying on Traffic Factors- K, D Issues with Managed Lanes Modeling Issues with Managed Lanes Modeling

ProjectsProjects Daily-to-Peak Correspondence is Daily-to-Peak Correspondence is

Different (K)Different (K)



Journey to an Alternate Journey to an Alternate ApproachApproach

Use of Peak-Period Model EstimatesUse of Peak-Period Model Estimates Use of Diurnal Factors Use of Diurnal Factors Use of CUBE Analyst Trip Table Use of CUBE Analyst Trip Table

Estimation ProcessEstimation Process Increased Confidence in Turning Increased Confidence in Turning

Movement Forecasts Movement Forecasts



Case StudyCase StudyI-95 PD&E Study in Southeast FloridaI-95 PD&E Study in Southeast Florida



Study Modeling Study Modeling ApproachApproach



Study GoalsStudy Goals

Produce traffic estimates for Produce traffic estimates for morning and afternoon peaks.morning and afternoon peaks.

Multiple target years: 2010, 2020, Multiple target years: 2010, 2020, 2030, 2040.2030, 2040.

Make systematic adjustments.Make systematic adjustments. Minimize “post-processing”.Minimize “post-processing”.



Subarea Modeling Subarea Modeling ApproachApproachRegional Model

2005 Base Year

Subarea Model

2005 Base Year

“Tight Subarea” Model

2010 Base Year



““Tight Subarea” Trip Table Tight Subarea” Trip Table AdjustmentsAdjustments

Adjustments to Base Trip Tables Using CUBE Adjustments to Base Trip Tables Using CUBE AnalystAnalyst

Inputs:Inputs: Extracted Tight Subarea Trip Tables for the Study Extracted Tight Subarea Trip Tables for the Study

CorridorCorridor Balanced AM and PM Traffic Counts for Freeways Balanced AM and PM Traffic Counts for Freeways

and Rampsand Ramps Trip ends Trip ends

Used Iterative Matrix Adjustment ProcessUsed Iterative Matrix Adjustment Process Trip ends NTrip ends Notot Altered. Altered. Output: Adjusted Base Trip TableOutput: Adjusted Base Trip Table



““Tight Subarea” Tight Subarea” Forecasting ApproachForecasting Approach

First, Compute Unadjusted GrowthFirst, Compute Unadjusted Growth Unadjusted Future-BaseUnadjusted Future-Base

Estimate Future Adjusted Trip TableEstimate Future Adjusted Trip Table Adjusted Base + Growth Adjusted Base + Growth

Difference Method Judged to be Difference Method Judged to be Better Performing than FactorsBetter Performing than Factors



Results Discussion-Results Discussion-Improved Validation Improved Validation

ExampleExampleVol Grp Count Range Model RMSE(%)

Allowable RMSE Range

Volume Count Volume/CountNumber of

Links1 1- 5,000 58% 45 - 55% 601,419 533,116 1.13 1802 5,000- 10,000 38% 35 - 45% 303,209 325,637 0.93 533 10,000- 20,000 30% 27 - 35% 545,149 444,128 1.23 284 20,000- 30,000 20% 24 - 27% 222,931 197,001 1.13 9

ALL 1-500,000 44% 32 - 39% 1,672,708 1,499,881 1.12 270

AM SUBAREA MODEL ASSIGNMENT RMSE for year 2010, using 2011 counts

Vol Grp Count Range Model RMSE(%)Allowable

RMSE RangeVolume Count Volume/Count

Number of Links

1 1- 5,000 24% 45 - 55% 539,820 544,627 0.99 1842 5,000- 10,000 8% 35 - 45% 294,332 303,792 0.97 503 10,000- 20,000 3% 27 - 35% 456,930 459,168 1 294 20,000- 30,000 3% 24 - 27% 194,235 196,449 0.99 9

ALL 1-500,000 12% 32 - 39% 1,485,317 1,504,036 0.99 272

AM TIGHT SUBAREA MODEL ASSIGNMENT RMSE for year 2010, using 2011 counts



Results Discussion-Results Discussion-Improved Trip LengthsImproved Trip Lengths

AM PM OPTight Subarea 11.29 11.38 14.5I-95 Survey (unfactored) 11.27 11.23 14.22I-95 Survey Factored (Using different methods range) 10.57-10.95 10.38-11.23 9.7-10.8



Results Discussion- A look Results Discussion- A look at the AM DDHV Forecastsat the AM DDHV Forecasts



Results Discussion- A look Results Discussion- A look at the PM DDHV Forecastsat the PM DDHV Forecasts



Results Discussion- AADT Forecasts Results Discussion- AADT Forecasts Comparison to Historic Counts TrendsComparison to Historic Counts Trends

I-95 Segment North of 2010 Count StationI-95 PD&E 2040 NB AADT Projection

2040 Historic Trendline projections %Difference

Linton Blvd. 932193 290400 272200 7%Yamato Rd. 932192 269259 232000 16%Glades Rd./SR 808 932191 272613 300500 -9%Palmetto Park Rd. 932190 272000 218900 24%SR 810/Hillsboro Blvd 862507 268509 215500 25%SW 10th St. / SR 869 862506 262484 219300 20%Sample Rd. / SR 834 860163 263992 220100 20%Copans Rd. 862505 270250 245900 10%Atlantic Blvd. / SR 814 862504 291684 269600 8%Cypress Creek Rd. 862503 296733 311200 -5%Commercial Blvd. / SR-870 862502 296751 290800 2%Oakland Park. Blvd 862501 321798 284400 13%Sunrise Blvd. / SR 838 862500 362942 360900 1%Broward Blvd. / SR 842 862499 375070 304800 23%Davie Blvd. 862498 240347 295400 -19%I-595 862493 273854 297300 -8%Griffi n Rd. 862458 365468 318600 15%Stirling Rd. / SR-848 862456 379864 301600 26%



Conclusions and Conclusions and LimitationsLimitations

Effective Method to relate DDHV and OD Effective Method to relate DDHV and OD Matrices used in Operations AnalysisMatrices used in Operations Analysis

Use of Standard K-Based Projections Seem to Use of Standard K-Based Projections Seem to be Overestimated- “ignored” Peak Spreading??be Overestimated- “ignored” Peak Spreading??

The Projections Match well with Historic The Projections Match well with Historic Trends Trends

Matrix Estimation Software is Effective Tool if Matrix Estimation Software is Effective Tool if used with Careused with Care

Post-Processing is not Completely Eliminated. Post-Processing is not Completely Eliminated. Engineering Judgment Should be ExercisedEngineering Judgment Should be Exercised



AcknowledgementsAcknowledgements

Florida Department of Transportation, District Florida Department of Transportation, District 44 Shi-Chiang Li, AICPShi-Chiang Li, AICP Lisa Dykstra, PELisa Dykstra, PE

The Corradino Group PD&E DivisionThe Corradino Group PD&E Division Mike Ciscar, PEMike Ciscar, PE Ryan Solis, PERyan Solis, PE

The Corradino Group Systems Planning The Corradino Group Systems Planning DivisionDivision Ken Kaltenbach, PEKen Kaltenbach, PE Abhishek MishraAbhishek Mishra

May, 07, 2013The Corradino Group, Inc., 14th TRB

Planning Applications Conference 20

Thank You!Thank You!

Questions/DiscussionQuestions/Discussion

may, 07, 2013 1 the corradino group, inc., 14th trb planning applications conference 14 th trb...

Documents

dthe corradino group

traffic factorsad

base trip tables

pm traffic counts

tight subarea trip tables

managed lanes forecasting

pde study

study corridorbalanced