Transit Modeling Update District One Implementation & Status Report presented to MTF Transit & Rail Committee presented by Dan Macmurphy, Traf-O-Data Corp. June 17, 2013 Purpose and Need Developed by FDOT Systems Planning, in conjunction with FTA, to provide transit travel demand forecasts consistent with federal expectations. Incorporates state of the practice techniques 1
Key Features New Trip Generation Rates and Purposes Peaking Factors split into peak and off-peak periods Mode Choice and Destination Choice (45 markets) Diurnal Factors split Highway Assignment into 4 periods A travel time Feedback Loop is included to account for the effects of congestion on Distribution and Mode Choice Network Generation (daily) Skims Feedback Loop Pre Load Peaking Factors Mode Choice Transit Assignment Build Transit Access Links Special Trips Model Flowchart Destination Choice Diurnal Factors Highway Assignments (am md pm nt) Reporting 2
Trip Generation New process, 2010 Census and 5-year ACS data Table HBW Trip Production Rates Household Income Household Autos Number of W orkers in H ousehold Wrks0 Wrks1 Wrks2 Wrks3+ $0 $24999 0.000 1.000 1.783 3.500 $25000 $49999 0.000 1.000 1.783 3.500 Veh0 $50000 $74999 Table 0.000 1 Trip Attraction 1.000 Rates 1.783 3.500 $75000+ 0.000 1.000 Industrial 1.783 Service 3.500 Commercial School $0 $24999 0.000 1.085 2.272 4.389 Households Employment Employment Employment Enroll. $25000 $49999 Trip 0.000 1.332 3.145 4.389 Veh1 Purpose Coeff. t Stat Coeff. t Stat Coeff. t Stat Coeff. t Stat R 2 $50000 $74999 0.000 1.145 2.172 4.389 HBW Inc1 0.0663 1.3 0.0910 7.4 0.2218 7.1 0.07 $75000+ 0.000 0.954 2.094 4.389 HBW Inc2 0.4764 6.9 0.1943 11.5 0.3642 8.5 0.17 $0 $24999 0.000 1.581 2.974 4.102 HBW Inc3 0.1457 2.6 0.1932 14.3 0.2560 7.4 0.15 $25000 $49999 0.000 1.604 3.171 4.102 Veh2 HBW Inc4 0.5202 7.9 0.4225 26.5 0.3132 7.7 0.35 $50000 $74999 $74999 HBSh 0.000000 1.572 2.863 4.102 3.2805 56.0 0.37 $75000+ HBSR 0.000 0.1199 1.330 1.0 3.111 0.23594.1028.0 0.6814 8.6 0.7368 35.5 0.42 $0 $24999 HBO 0.000 1.595 3.130 0.49105.07512.1 1.3213 12.9 1.1578 39.8 0.51 $25000 $49999 HBSc 0.000 2.297 3.109 5.075 1.326 Veh3+ $50000 $74999 HBU 0.000 1.967 3.247 5.799 0.549 $75000+ NHBW 0.000 Table 0.6346 1.551 HB College/University 7.6 3.119 0.45345.799 and 22.1 HB School 0.7118 Trip Production 13.0 0.1773 Rates 12.3 0.45 NHBO 0.2443 7.5 1.947 23.6 0.7514 32.1 0.51 Household Household Size Trip Purpose Autos /Income Size1 Size2 Size3 Size4+ HB College/University 0.016 0.036 0.266 0.363 All HB School 0.000 0.042 0.491 1.623 Peak Factors Use Peak / Off-Peak processes throughout the Model Polk TMU - PK Polk TMU - OP CFRPM - PK CFRPM - OP TBRPM -PK TBRPM - OP Time Period (AM, MD) 6:00 AM-9:00 AM 9:00 AM-3:00 PM 6:30 AM-9:00 AM 9:00 AM-3:30 PM 6:30 AM-9:00 AM 9:00 AM-3:30 PM Time Period (PM, NT) 3:00 PM-7:00 PM 7:00 PM-6:00 AM 3:3030 PM-6:30 PM 6:30 PM-6:30 AM 3:3030 PM-6:30 PM 6:30 PM-6:30 AM HBW 0.764 0.236 0.546 0.455 0.510 0.490 HBSH 0.409 0.591 0.282 0.718 0.263 0.737 HBSR 0.401 0.599 0.277 0.724 0.279 0.721 HBSC 0.816 0.184 - - 0.582 0.418 HBO 0.565 0.435 0.452 0.548 0.408 0.592 NHBW 0.436 0.564 - - 0.342 0.658 NHBO 0.435 0.565 0.305 0.695 0.286 0.714 Percent of Trips HBCU 0.581 0.419 - - 0.250 0.750 0.900 0.800 0.700 0.600 0.500 0.400 0.300 0.200 0.100 0.000 HBW Polk TMU - PK Polk TMU - OP CFRPM - PK CFRPM - OP TBRPM -PK TBRPM - OP 3
Transit Very similar to the Tier A/B/C methodology currently in use, but includes additional features: Use of REWALK for access-links Transit speeds are controlled by dwell at stops Change wait-time from ½ headway Use (new) recommended mode coefficients Mode Choice and Distribution (destination choice) have been revised to reflect detailed market segmentation Calculate probabilities for: HBW -- 7 markets HBNW -- all trips HBSH -- 7 markets HBSR -- 7 markets HBO -- 7 markets NHBO -- all trips HBW -- 7 markets HBSC -- 4 markets HBCU -- 4 markets Time-Of-Day Assignment Diurnal Factors assign trips to time period buckets TMU MODEL HBW_PA HBW_AP NHBW_PA NHBW_AP NHBO_PA NHBO_AP 0.524 0.005 0.015 0.258 0.119 0.119 6:00AM-8:59AM 0.474 0.169 0.472 0.466 0.451 0.451 9:00AM-2:59PM 0.039 0.432 0.678 0.049 0.381 0.381 3:00PM-6:59PM 0.221 0.136 0.062 0.000 0.049 0.049 7:00PM-5:59AM TIME TBRTM HBW_PA HBW_AP NHBW_PA NHBW_AP NHBO_PA NHBO_AP 0.491 0.028 0.336 0.048 0.177 0.177 6:00AM-8:59AM 0.261 0.327 0.297 0.328 0.421 0.421 9:00AM-2:59PM 0.030 0.451 0.262 0.354 0.323 0.323 3:00PM-6:59PM 0.184 0.228 0.096 0.279 0.080 0.080 7:00PM-5:59AM TIME 4
Feedback Loop Feedback Loop Number = 1 START TRANSITMODEL Thu 05/02/2013 17:06:20.97 START MODE CHOICE Thu 05/02/2013 17:07:15.45 START DISTRIBUTION Thu 05/02/2013 17:19:44.19 START HASSIGN Thu 05/02/2013 17:49:34.07 AM percent time change 0.69542 AM percent volume change 0.5959 MD percent time change 0.3764 MD percent volume change 0.28835 Feedback Loop Number = 2 START TRANSITMODEL Thu 05/02/2013 17:53:40.97 START MODE CHOICE Thu 05/02/2013 17:54:34.31 START DISTRIBUTION Thu 05/02/2013 18:07:39.58 START HASSIGN Thu 05/02/2013 18:37:18.43 AM percent time change 0.05759 AM percent volume change 0.12637 MD percent time change 0.0310803108 MD percent volume change 0.07725 Feedback Loop Number = 3 START TRANSITMODEL Thu 05/02/2013 18:41:03.19 START MODE CHOICE Thu 05/02/2013 18:41:57.27 START DISTRIBUTION Thu 05/02/2013 18:54:53.13 START HASSIGN Thu 05/02/2013 19:24:38.26... CONVERGED Thu 05/02/2013 21:52:12.48 START TASSIGN Thu 05/02/2013 21:52:20.85 START REPORTS Thu 05/02/2013 21:52:35.42 End Run Thu 05/02/2013 21:52:51.83 The model ran for a total of: 290 Minutes ( 4 Hr 50 Min) Skims Mode Choice Destination Choice Diurnal Factors Highway Assignments (am md pm nt) Polk Implementation CRTPA POLK 5
New Data Requirements Household Income Household Size Household Workers Auto Ownership University / College Students Peak Hour Traffic Counts Diurnal Factors Peak Hour Factors Our Experience Added more error checking (out of range) Replaced hard-coded parameters with {keys} Revised some matrix calculations (autos > persons) Improved CLUSTER functionality Adjusted script to prevent out of memory errors Added reporting Timeframe for first run: 8 weeks 6
Out-of-the-Box Summary Reasonable Results and Run Times C:\FSUTMS\Polk\R3\Base \ \ \ VOLUME AND COUNT SUMMARY BY SCREENLINE Total VOL= 9,552,803 CNT= 9,719,061 VOL/CNT= 0.98 N=1,083 Overall Summary Total Number of Links: 8,132 Total Centerline Miles: 2,611.75 Total Lane Miles: 3,426.56 Total Directional Miles: 2,654.00 Total VMT using Volumes: 4,765,186 (Links With Counts) Total VMT using Counts: 4,576,271 (Links With Counts) Total VMT Volume over Counts: 1.04 (Links With Counts) Total VHT using Volumes: 131,757 (Links With Counts) Total VHT using Counts: 126,656656 (Links With Counts) Total VHT Volume over Counts: 1.04 (Links With Counts) Total Volumes All Links: 53,000,190 Total VMT All Links: 17,818,308 Total VHT All Links: 517,371 Original Speed (MPH): 35.49 Congested Speed (MPH): 33.85 ================================================================ End Run Tue 04/30/2013 17:02:51.03 The model ran for a total of: 354 Minutes ( 5 Hr 54 Min) Potential Model Enhancements External Trips Special Trips Identify by trip type and occupancy Split by trip purpose and trip length Airports Theme Parks Trucks Calibration Procedures Mode Choice Destination Choice 7
Managed Lanes Implementation Incorporating the Phase 1 congestion pricing Toll Policy Curve VC TOLL TOLL2 toll procedure in the Polk TMU 024 0.24 25 25 Preliminary results are encouraging 0.39 150 100 0.57 300 200 0.76 500 350 0.98 600 500 1.00 700 700 Willingness to Pay Curve COST PCTWTP 0.00 5.00 8.00 20.00 10.00 25.00 16.30 40.00 20.00 48.00 23.70 55.00 31.40 65.00 41.70 75.00 51.80 85.00 58.30 90.00 66.70 98.80 D1RPM Implementation CRTPA 2,240240 TAZs Scalability? Long Run Times (7 days) Optimization Procedures Cluster Utilization Hybrid Model Testing D1RPM 5,651 TAZs 8
TMU Project Documents Trip Distribution Review and Recommended Improvements Trip Generation Review and Recommended Improvements Time of Day Stratification Review Mode Choice Review and Recommended Improvements Travel Time Feedback Review Reporting Functionality Recommendations Principles of Model Calibration & Validation User Benefit Guidelines and Procedures Quality Control Guidelines Available for Download at FSUTMSONLINE.NET Q & A 9
Transit Ridership ESTIMATE OF AVERAGE WEEKDAY RIDERSHP for 2007 BY ROUTE # Route Route Name Ridership TMU LAMTD 1 10 Shuttle 117 LAMTD 10 431.7 2 11 E. Main/Combee 256 LAMTD 11 462.9 3 12 Lakeland/WinterHaven 247 LAMTD 12 355.6 4 20 Grove Park/Crystal Lake 296 LAMTD 20 436 5 21 Edgewod 106 LAMTD 21 742.7 6 22X Bartow Express to Lakeland 285 COUNTY 22XL 424.6 7 30 Cleveland Heights 81 LAMTD 30 635.1 8 31 S Florida Ave 545 LAMTD 31 2205.7 9 32 Medulla Loop 16 LAMTD 32 155 10 37 South 27 LAMTD 37 60.1 11 40 Ariana/Beacon 68 LAMTD 40 581.8 12 41 Central Ave 220 LAMTD 41 352.7 13 42 W Memorial 414 LAMTD 42 1638.4 14 50 Kathleen/Providence 190 LAMTD 50 1160.4 15 51 N US98/Duff Rd 599 LAMTD 51 1930.9 16 52 N Florida Ave 513 LAMTD 52 771.8 17 53 Lakeside Village 32 LAMTD 53 475.3 18 56 Kathleen/Mall Hill Rd 170 LAMTD 56 334 19 57 Kidron/Flightline 88 LAMTD 57 223.1 Subtotal 4,270 Subtotal 13,378 WHAT 1 10 Northside 232 WHAT 10 88.6 2 12 Lakeland/WinterHaven 269 WHAT 12 355.6 3 15 Haines City 101 WHAT 15 59.3 4 20 PCC/Hospital 57 WHAT 20 233.2 5 22x Bartow Express to Winter Haven 170 COUNTY 22XWH 97.7 6 30 Eagle Ridge/Winter Haven 305 WHAT 30 895.8 7 40 Southside 179 WHAT 40 738.7 8 44 Southwest 167 WHAT 44 161.3 9 50 Westside 119 WHAT 50 286.8 Subtotal 1,599 Subtotal 2,917 Polk County 1 25 Bartow /Fort Meade 81 COUNTY 25a 33.5 2 35 Frostproot to Eagle Ridge Mall 111 COUNTY 25b 6.8 COUNTY 35 19.5 Subtotal 192 Subtotal 60 Total 6,061 Total 16,355 Run Optimization Distribute MULTI step HASSIGN example Distribute INTRA step HASSIGN example MODEL MODEL START START AM Zones 1-1866 PROCESS 1/3 AM Zones 1-5600 PROCESS 1 PROCESS 2 MD zones1-5600 AM Zones 1867-3732 PROCESS 2/3 END END AM Zones 3733-5600 PROCESS 3/3 Key Wait For Files MODEL CLUSTER MODEL CUBE Parallel Processing Manually Set up for # of Cores Must use when summary stats needed Disadvantage if one process is longer than the other Parallel Processing Automatically Scales for # of Cores 10% time savings Cannot be used for summary stats 10
Mode Choice 0.7 0.7 0.5 0.5 Tracking Trips GENERATION MODE DESTINATION CHOICE DAILY person-trips DAILY vehicle-trips DAILY vehicle-trips AUTO_DAY AO 399,614 HBW DA 275,185 327,021 1.22 DA 1,177,555 HBW SR2 39,098 SR2 410,374 SR3 12,738 SR3 157,077 339,072 HBSH DA 147,361 228,635 1.48 EE 77,731 HBSH SR2 62,403 LT 144,408 SR3 18,871 HT 47,938 276,433 HBSR DA 119,082 184,705 1.50 2,015,083 HBSR SR2 50,374 SR3 15,249 610,088 HBO DA 264,275 410,048 1.49 HBO SR2 111,846 ASSIGNMENT SR3 33,927 AM vehicle-trips MD vehicle-trips PM vehicle-trips NT vehicle-trips 607,309 NHB DA 294,454 417,330 1.46 NHB SR2 81,273 DA 208,446 439,406 380,318 149,373 SR3 41,603 SR2 72,667 156,602 132,515 48,585 77,732 EE DA 33,851 77,731 1.00 SR3 29,499 62,896 49,037 15,643 EE SR2 28,669 EE 10,610 36,456 28,255 2,409 SR3 15,211 LT 17,184 65,128 55,019 7,076 177,277 IE DA 77,198 177,267 1.00 HT 5,601 21,534 18,229 2,245 IE SR2 65,380 2,014,733 SR3 34,689 192,040 LT 144,408 192,346 1.00 TRK HT 47,938 2,679,565 2,015,083 11