Modeling for Managed Lane Traffic and Revenue Forecasting

CDM Smith Modeling for Managed Lane Traffic and Revenue Forecasting

North Carolina Model Users Group 19 November 2015 Agenda

• General concepts of modeling for managed lanes – Managed lanes vs. general tolls • Managed Lane Model Development – Modifications to the Travel Demand Model – Replicating future year bottleneck conditions with microsimulation • Forecasting revenue – Static approximation of dynamic tolling • Special considerations in managed lane modeling

Managed Lane Modeling for Traffic and Revenue

CDM Smith’s Managed Lane Forecasting Model

• Adaptation of a standard regional Travel Demand Model • Focused heavily on a single highway corridor (or network of corridors) • Surrounding region is included to allow for alternative route diversion

Model Functionality

• Built to forecast ‘typical’ traffic and revenue conditions • Estimates typical tolls for model periods • Hourly during peaks • Period-wise during off peaks Levels of Managed Lane Study

Level 1 (Sketch Level)

• High-level traffic and revenue estimates • May use spreadsheet model as an alternative to a traditionsal TDM Covered in Presentation Level 2 (Planning Level)

• Employs travel demand model with volume and speed calibration

Level 3 (Investment Grade)

• Requires independent economist to validate model growth • Employs SP surveys to estimate model parameters (e.g. VOT, toll decision equation, eligible managed lane users) • May require risk analysis or sensitivity testing General Toll Decision Model

P = exp(b0 + b1x + b2x2 + … )

Toll Choice: • Cost A • Time A • Distance A

Free Choice: • Time B • Distance B Managed Lane Choice Model

Managed Lane Path: • Toll Cost • Time (freeflow)

Free Path: • Time (congestion)

Alternate Free Path • Time (congestion) General Toll Modeling Process

Land Use and Corridor Model Development Socioeconomic Input from MPO Control Data • Corridor (or alt Socioeconomic Data Model route) Counts • Adjustments by Calibration Screenline Counts • Independent Process Speed Data Economists Network Checks and Modifications Trip Generation Value of Time and Calibrated Base Year Vehicle Operating Cost Model Trip Distribution Calculation Transit Trip Tables

Mode Choice Future Year Model Future Year Network Highway Trip Tables Adjustments Improvements 4-Step Model

Future Year Traffic and Revenue Model Runs

Traffic and Revenue Forecasts Managed Lane Modeling Process

Mode Choice Future Year Model Future Year Network Highway Trip Tables Adjustments Improvements 4-Step Model

Skeleton Trip Table Base Year VISSIM Future Year Speed Flow Calibration Curves Base Year Corridor Future Year Traffic and

Design Revenue Model Runs Operational Network Adjustments

Future Network Design Future Year VISSIM Simulation Traffic and Revenue

Forecasts Operations Analysis (Microsimulation) Managed Lane Sensitivity

• Well-Calibrated Travel times are critical on corridor Corridor Speed • Simulating and replicating bottleneck conditions is necessary

• Necessary to model a peaking condition Peak Hour Condition • Replicate queue building across periods

• Dynamic pricing on managed lanes to maintain Traffic Management freeflow conditions Modifications for the Managed Lane Model: Refining Model Periods

AM1 AM2 AM3 AM4 AM5 Split MD1 Periods PM1 PM2 PM3

New NT New AM New AM Shifted Shoulder Periods New MD New PM

Original OP Original Original AM Periods Original Sum of Mainline PM Counts

Time of Day Modifications for the Managed Lane Model: Sub-Area Extraction

FROM 3,774 ZONES TO 1,621 ZONES Modifications for the Managed Lane Model: Sub-Area Extraction

Base Year Sub- Base Year Trip Base Year Trip Area Trip Table, Base year Sub- Regional Sub area Extracted from Sub Area Area Trip Table, Table, Not Calibration Table, Calibrated extraction Regional TT (Not Calibration Fully Calibrated Calibrated to Regional Level Calibrated)

Apply Growth to Calibrated Model Apply Growth to Calibrated Model

Future Year Sub- Future Year Trip Future Year Trip Area Trip Table, Future year Sub- Sub area Extracted from Area Trip Table, Table, Not Table, Calibrated extraction Regional TT (Not Fully Calibrated Calibrated to Regional Level Calibrated)

Modifications for the Managed Lane Model: Replicating Bottleneck Conditions

Handled by TDM Calibrating Future Year Speeds

Naïve method (consistent with base year VDF)

• Only accurate if travel patterns remain consistent across corridor and additional capacity does not create new bottlenecks

HCM analysis

• Load future year demand, calculate bottlenecks across time periods based on demand and capacity of new network • Use several loading combinations to determine shape of speed flow curves

VISSIM analysis (most accurate method)

• Build and calibrate base-year VISSIM network • Create future year network and loading conditions on calibrated sim • Load microsimulation for peak periods to determine new VDF curves

Calibrating Future Year Speeds

1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.5 1 1.5 2 V/C Ratio Calibrating Future Year Speeds

1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.5 1 1.5 2 V/C Ratio 1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.5 1 1.5 2 V/C Ratio Calibrating Future Year Speeds

1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 V/C Ratio Calibrating Future Year Speeds

1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 V/C Ratio Calibrating Future Year Speeds

1.2

0.8

0.6 T0/T

0.4

0.2

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 V/C Ratio Vissim Model Development and Calibration Process

1. Data Collection for VISSIM 2 Base Model 3. Error Checking Input Development • Review Inputs •Traffic Volumes • Review Animation •Base Maps/Inventory • Data input •Travel Time Surveys • Quality assurance of •Field Observations available VISSIM network

5. Future Model 4. Compare Model Development MOE’s to Field Data YES • Code Future Network Model Calibrated • Volumes and Speed Match? • Input Forecasted Demand • Congestion in the right places?

Final Results Adjust Global • Measure Model MOE’s parameters • Generate Speed-Flow Curves • Modify Global Parameters • Modify Link Parameter Sample Results from Vissim Load Sets Modified Speed-Flow Curves

1.2

0.8

0.6 T0/T

0.4

0.2

0 V/C Ratio Modifications for the Managed Lane Model: Modified Speed-Flow Curve Example

SR 112

Opa Locka Blvd/ NW 62nd St NW 69th St NW 81st St NW 95th St NW 103rd St NW 119th St NW 125th St NW 135th St

To Downtown Miami

Halllandale Beach Miami Gardens Dr NW 151st St Golden Ives Dairy Rd Blvd Pembroke Rd Hollywood Blvd Sheridan St Stirling Rd Glades PNR

I-595

SR 84 Davie Blvd Broward Blvd Griffin Rd Sunrise Blvd

Legend Curve 2 Curve 4 Curve 85 Curve 87 Curve 89 Curve 91 Curve 93 Curve 95 Curve 97 Static Approximation of Dynamic Tolling

• Given: – Calibrated Base Year Model – Modified future VDF curves for links – Accurate coding of proposed future year managed lane configuration(s) • We want to estimate how much revenue the managed lanes will generate • Considerations: – What optimization point will we use? • Maximize traffic flow • Optimize revenue – Specific toll agency requirements (minimum toll, trip reconstruction) Typical Traffic Conditions

• Dynamic traffic pricing – will adjust automatically (or manually depending on toll agency) to retain near free-flow speed in managed lanes • Model estimates are static (within standard TDM) • Model itself represents a typical weekday traffic condition – Interpretation of final traffic is that this would be the expected toll necessary to maintain traffic flow in lanes Toll Sensitivity and Optimization Points

2500 2000 1800

2000 1600 1400

1500 1200 1000 Revenue Traffic 1000 800 Optimal

600 Revenue(dollars) 500 400 200 0 0 0 50 100 150 200 250 0 50 100 150 200 Toll Rate (cents per mile) Toll Rate (cents per mile)

70 Revenue Optimization 60 • Maximize the revenue generated at a toll gantry or 50 for multiple toll movements Traffic 40 Optimal GP speed 30 ML Speed Traffic Optimization Speed(mph) 20 10 • Maximize the uncongested traffic in managed lanes • 50 miles per hour minimum speed 0 0 50 100 150 200 • Achieved in the model through a maximum traffic Toll Rate (cents per mile) threshold (typically about 1600 vplph) Post-processing and Revenue Calculations

Extract point-to-point Import into revenue Run assignments to test movements to pull spreadsheet: rates for traffic volumes for express • Point-to-point traffic optimization lanes by period, vehicle • Toll matrix type • Share of toll by segment

Apply leakage factors Interpolate between Sum by time and annualization analysis years to period/direction and factors to traffic and estimate annual segment, ETC vs. Video revenue revenues Special Considerations

Multiple Decision • Multinomial/Nested Decision Structure • Time Penalties Points • Depends heavily on configuration of decision points Fully Constrained • Limit minimum volume on roadways Sections (no alt • Trip suppression/time of day shifting in outer years route) Operations Analysis • Strict calibration of microsimulation model of Managed Lane • Reconfiguration of access point design Access

Sensitivity Testing • ‘Stress-testing’ model with specific input parameters & Risk Analysis • Statistical simulation Important Considerations of Managed Lanes

• Managed lanes are fully dependent on growth – Managed lanes to solve problems of reasonably congested GP lanes • Will make gp lanes freeflow (in many cases) for opening years – Managed lanes in extreme congestion must be carefully planned • If complete breakdown of lanes (demand > C*1.4), access points may exacerbate queues • Managed lanes should not be seen as a method for fully funding construction – ML construction rarely breaks even in a typical roadway lifespan • Most important model consideration is to benchmark forecasts against managed lanes that are currently operating