Presented by Andrew Velasquez URS Corporation/Florida s Turnpike Enterprise

Transcription

1 Presented by Andrew Velasquez URS Corporation/Florida s Turnpike Enterprise

2 Traffic and Revenue Study Assessment of the viability of tolling a facility and expected toll revenue over a long term time frame. Evaluates congestion relief potential of tolling options. Allows agencies to evaluate financing the options. Cash Financing Debt Financing Public Private Pi Partnerships 2

3 Traffic and Revenue Study Levels Top Down Comparative analysis of similar facilities Sketch Level Involves analysis tools with generic inputassumptions Planning Level Involves some data collection and use of various modeling tools, including time of day diversion models. Investment Grade Extensive data collection (O/D, value of time, socioeconomic) and refined modeling tools. 3

4 Study Area District 4 Portion 4 Managed Lanes on I 75 from NW 170 th Street to I 595 Reversible lane direct connection to I 595 District 6 Portion 4 Managed Lanes on I 75 from SR 826 to NW 170 th Street Direct ramp connection from SR 826 to I 75 (two managed lanes) 4 Managed Lanes on SR 826 between SR 836 and SR 826 4

5 Traffic and Revenue Models Land Use Review Bluetooth OD SP/RP Survey Travel Demand Model (3 Period) Time of Day Model (Hourly) Probability Model Toll Revenue Model 5

6 OD Estimation Bluetooth Detection Technology Bluetooth Detector (Traffax) has 100 meter range and consecutive 5 second scan windows. Discoverable bluetooth devices are detected and timestamped. Bluetooth devices have a unique MAC ID. Chronological MAC IDs are linked together to form trips. Dataanalysis analysis provides studyareaspeed speed profiles and origin destination estimates. 6

7 Bluetooth Data Collection Effort 46 detectors deployed in the study area Deployment designed primarily to sequester interchanges Data collection occurred October 12 20, 2011 Penetration rate ranges from 3% 9%, 5% mean O D table by period (AM, PM, OP) 7

8 Bluetooth Data Collection Effort 5 detectors deployed in the I 95 Data collection occurred February 23 March 1, 2012 Purpose to understand vehicle movements eligible to use 95 Express 63% or 155,00 vpd O D critical for accurate model calibration 8

9 Stated/Revealed Preference Surveys Stated Preference Survey: Hypothetical choice scenarios to understand response to future conditions What would you do if...? Revealed Preference Survey: Observed choice bh behavior given existing iti conditions What did you do...? I 75/SR 826 Stated Preference Survey Estimate Values of Time (VTTS) for drivers in the I 75 and SR 826 corridors Input for traffic and revenue forecast for Express Lanes expansion I 95 Stated/Revealed d Preference Survey Estimate VOT for drivers in the I 95 corridor Understand why drivers choose to use the I 95 Express Lanes without a large travel time benefit Validate/calibrate travel demand model 9

10 I 95 Trip Characteristics Ranking of Reasons for Using the I 95 Express Lanes Reasons for using I 95 Express Lanes Faster travel time cited most frequently Faster travel time ranked highest Less congestion andtravel time reliability ranked 2 nd and 3 rd Faster travel time (N = 213) Less congestion (N = 193) More reliable travel time (N = 139) Safe road conditions (N = 44) No large trucks (N = 74) 33% 26% 18% 16% Only 1 respondent reported fuel savings as a reason Other (N = 8) 63% 13% 25% 18% 73% 29% 40% 66% 78% 22% 45% 27% Ranked 1st Ranked 2nd Ranked 3rd or lower 5% 10

11 Discrete Choice Modeling Overview Stated and Revealed Preference data used to estimate coefficients of a discrete choice model Coefficients provide information about sensitivity to: Travel time Toll cost Travel Time uncertainty Used to feed the ELTOD model 11

12 SP/RP Model Calibration Time and Cost Savings Reliability Other Variables Utility for alternative () (i): U(i) () = βtime * traveltime(i) () + βcost * tollcost(i) () + βentropylate * entropy () (i) + CalibratedTollConstant Entropy for alternative (i): Entropy(i) = 2.15*EXP( VC(i)* )/(1+EXP( VC(i)* )) Where: βtime = travel time coefficient βcost = toll cost coefficient βentropylate = entropy coefficient for travelers arriving late CalibratedTollConstant = Calibrated toll constant for appropriate travel direction and time period traveltime(i) = Travel time for alternative(i) tollcost(i) () = Toll cost for alternative(i) () VC(i) = Volume/Capacity ratio for alternative(i) Coefficient Before Calibration After Calibration Time Cost VTTS = $11.03 Entropy Toll Constant (Peak) 1.50 (Off Peak) 12

13 SP/RP Model Calibration V/C Entropy 0% % % % % % % % % % % % % % % % % % % % % 2.14 Entropy Entropy VC Relationship 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Volume/Capacity 13

14 95 Express Calibration Results 2,500 NB Observed vs Model Traffic 3,000 SB Observed vs Model Traffic 2,000 1,500 1, Observed NB Modeled NB 2,500 2,000 1,500 1, Observed SB Modeled SB Hour of Day Hour of Day NB Observed Toll vs Model Toll SB Observed Toll vs Model Toll $4.00 $4.00 $3.50 $3.50 $3.00 $3.00 $2.50 $2.50 $2.00 $1.50 $1.00 Model NB Observed NB $2.00 $1.50 $1.00 Model SB Observed SB $0.50 $0.50 $ $ Hour of Day Hour of Day

15 95 Express Calibration Results Calibration Results Average Weekday Observed Model % Difference Transactions 47,404 50, % Revenue $67,314 $67, % 0.5% For the period November 1, 2011 to December 16,

16 Model Interactions Land Use Review Bluetooth OD SP/RP Survey Travel Demand Model SERPM CORSIM Micro Simulation Step 1 Step 2 Time of Day Model ELTOD Toll Revenue Model Step 3 Step 4 Probability Model 16

17 Traffic and Revenue Process Step 1 Step 2 Initial feed dfrom the SERPM into the ELTOD model Run ELTOD for full experimental design into probability model. Probability model provides target revenue values. Step 1 Step 2 SERPM ELTOD Probability Step 4 Step 3 Toll Revenue Model 17

18 Southeast Regional Planning Model (SERPM) 3 Counties Miami Dade, Broward and Palm Beach County Project Level Validation Updated Highway Network to 2010 Benchmarked Population to 2010 Census Updated 3,700 links with counts by time of day 18

19 Subarea Feed from SERPM 19

20 TOD Model Flowchart Travel Demand Corridor Daily O D Matrix Model Parameters Akcelik parameters Value of time Model scale Equilibration settings ELTOD Equilibrium Assignment Supply Toll Policy Minimum toll Maximum toll Toll escalation Highway Network Geometry GP links ML links Access links Assumed tolls Equilibrium Flows by Period Toll levels, revenues Speeds, travel times V/C ratios Output Report Notes: Blue boxes represent elements from SERPM Green boxes are other ELTOD inputs Orange boxes are ELTOD calculations 20

21 ELTOD Express Lanes Time of Day Assignment Model Traffic allocation analysis for managed toll lanes projects Produces traffic forecast estimates in a pre defined limited access corridor for general use and managed toll lanes Utilizes output from the SERPM Traffic estimates are by: Direction, Project Section, Hour, Lane Type (GUL or ML) 21

22 ELTOD: Input Data VTTS & Choice Model Coefficients: $11.03 Maximum Service Volume (Vehicles/Lane/Hour): vph Free Flow Speed: mph Geometry (GP & ML) Number of Lanes, Link Length (Distance) Sub area O D matrix (AM, PM, Off Peak) Hourly traffic distribution Volume Delay Measure (Akcelik Function) 22

23 ELTOD: Volume Delay Curve 23

24 Input Data: ELTOD: Pricing Policy Minimum/Maximum Toll Rates Min = $0.15/mile, Max = $2.00/mile Pricing Curve Level (frequency of toll rate change) Exponential curve (not a step function) Higher exponential values represent more traffic serviced 24

25 ELTOD: Pricing Policy Curve Comparison p$1.00 To oll Rate er Mile ($ $) $2.00 $1.80 $1.60 $1.40 $1.20 $1.00 $0.80 $0.60 $0.40 $0.20 $0.00 Revenue Maximizing I-75/S.R. 826 IG I-75/S.R. 826 PL 95 Express Traffic Maximizing Revenue Potential Increases Curves are defined by the EXP value. Traffic Volume Increases LOS A LOS B LOS C LOS D LOS E Volume/Capacity 25

26 ELTOD: Output Traffic volumes Preliminary revenue Toll rates Speeds Volume to capacity ratios Application: UsedtheELTOD to generate hourly traffic and toll rates to feed the revenue model 26

27 CORSIM Micro Simulation Models developed for I 75/S.R. 826 Project Development and Environment (PD&E) Studies Consolidated two separate models into one model Developed interface to read ELTOD volumes into CORSIM Re calibrated to ELTOD volumes and Bluetooth th Speeds 27

28 Existing Speed Comparison Bluetooth ELTOD BT vs CORSIM Avg. Avg. ELTOD Avg. Direction Period Speed Speed Diff Speed Diff I 75 SB AM % % PM % % NB AM % % PM % % 28% SR 826 SB AM % % PM % % NB AM % % PM % 4.1% % 28

29 Probability Modeling Variable B Variable A Variable C Probability distributions for traffic and revenue 29

30 Improving the Reliability of Traffic and Revenue Forecasts There is broad recognition of issues with traffic forecasts and of ways of improving those forecasts. Improve the models that are used for forecasting Progress has been made along a number of fronts Model structure Model inputs Recognize and quantify inherent uncertainties and simplifications Identify key uncertainties in model structure and inputs Use quantified probability analysis

31 Sensitivity Analyses are Essential But Have Clear Limitations Typically run for each input separately, without accounting for interactions between inputs Generally need to evaluate effects of changes to more than one variable in combination; this cannot be accomplished without A new model run this can be time consuming given the number of possible combinations of possible uncertainties titi 31

32 How Are the Variables Assessed? 1. Estimate range of uncertainty associated with each significant variable 2. Outline scenarios involving combinations of factors 3. Determine sensitivityof traffic andrevenue by: Running travel demand model for those scenarios Developing statistical model to approximate the travel demand model 4. Use simulation method to determine full distribution of traffic and revenue accounting for uncertainties in input factors 32

33 Key Forecast Variables of Possible Significance Define key variables and their uncertainties, including: Vl Values of time distribution ib ti from survey Network changes FDOT work program Local road improvements Population, employment BEBR scenarios Employment projections 33

34 Experimental Design 2040 Model Population Roadway Network VTTS 1 6,263,300 Needs Plan $ ,263,300 CF Plan $ ,918,700 Needs Plan $ ,918,700 CF Plan $ ,918,700 E+C $ ,626,250 E+C $ ,263,300 Needs Plan $ Travel 8 6,263,300 Demand CF Plan $ Time of Day 9 6,263,300 Model E+C $ Model ,918,700 SERPM Needs Plan $ ELTOD ,918,700 CF Plan $ ,918,700 E+C $ ,626,250, Needs Plan $ ,626,250 CF Plan $ ,626,250 E+C $ ,918,700 Needs Plan $ ,918,700 CF Plan $ ,918,700 E+C $ ,626,250 CF Plan $ ,626,250 E+C $

35 Probability Model Risk Factors Revenue Risk Variable Case Factor Demographics Low 0.91 High 1.10 Value of Time Low 0.73 High Road Network Low 1.00 High General Uncertainty Low 0.75 High

36 Probability Analysis Process Input Distributions tions Input Distributions tions Travel Demand Model with Toll Module Regression Model Representative scenario runs Monte Carlo simulation runs Traffic & Revenue Estimations Traffic & Revenue Estimations * 20 runs for each year * 10, runs for each year 36

37 Application of Probability Model Monte Carlo simulations used to determine distributions of traffic and revenue for each model year Sample Output 37

38 Generic Probability Distribution P50 P75 P25 P90 P10 10% 25% 50% 75% 90% 38

39 The Final Steps Step 3 Step 4 Probability model supplies target revenue for P50 and P75 scenarios to ELTOD. Final ELTOD runs to produce hourly refinement for the Toll Revenue Model Step 1 Step 2 SERPM ELTOD Probability Step 4 Step 3 Toll Revenue Model 39

40 Revenue Forecasting Process ELTOD Gross Revenue Traffic and Revenue Factors Toll Collection Expenses Preliminary Net Revenue Other Operating Expenses Final Net Revenue Maintenance Expense 40

41 Traffic and Revenue Factors under FDOT Directive Weekend Traffic Weekend Toll Rates Toll Evasion No Trucks No Toll by Plate Tolls Indexed annually 41

42 Annual Gross Revenue ($000) $160,000 $140,000 $120,000 $100,000 $80,000 Final IG P50 Final IG P75 $60,000 $40,000 $20,000 $0 Note: Numbers are expressed in 2011 dollars. 42

43 Managed Lanes in Operation SR 167 Seattle (10 miles) I 15 Salt Lake City (40 miles) I 394 Minneapolis (9 miles) I 495 Washington DC (14 miles) I 85 Atlanta ( 7 miles) SR 91 Orange County (10 miles) I 15 San Diego (16 miles) I 25 Denver (7 miles) I 10 Houston (13 miles) I 95 Miami ( 7 miles) 43

44 Managed Lanes Traffic and Revenue Stats Facility Length General Use Lanes Express Lanes Annual Revenues (millions) Average Daily Rate/Mile Avg. Revenue/Lane Mile I 85 GA $ $0.07 $160 S.R. 167 WA $ $0.09 $28 I 394 MN $ $0.13 $73 I 25 CO $ $0.10 $179 I 10 TX $ $0.04 $146 I 95 FL $15.0 $0.12 $564 S.R. 91 CA $ $0.30 $1,000 I 595 FL Reversible $5.0 $0.06 $167 I 75/SR 826 IG $21.9 $0.24 $219 44

45 Questions & Answers 45