Toward a Consistent and Robust Integrated Multi-Resolution Modeling Approach for Traffic Analysis Jeff Shelton, TTI Yi-Chang Chiu, Univ. of Arizona TRB – Transportation Planning Conference Houston, TX May 17-21, 2009 Transportation Operations Group Outline Introduction – Mesoscopic – Microscopic Multi-Resolution Modeling – Concept – Conversion Process – Modeling Issues Case Study Applications Transportation Operations Group Outline Introduction – Mesoscopic – Microscopic Multi-Resolution Modeling – Concept – Conversion Process – Modeling Issues Case Study Applications Transportation Operations Group Introduction Integrating mesoscopic dynamic traffic assignment (DTA) and microscopic traffic simulation and assignment models can be advantageous for regionwide operational planning projects – DTA – region-wide estimation of traffic redistribution – Microscopic – local operational analysis The integration synergizes the strengths of both models. Challenges remain in model translation and interface Modeling issues to be addressed – Consistency – Situation in which feedback is needed Transportation Operations Group Simulation-Based Dynamic Traffic Assignment (SBDTA) Address issues that may fall beyond the reach of both: – Microscopic models: (dynamic but small-scale) typically used by traffic engineers for project traffic studies – Macroscopic models: (large-scale but static) typically used by transportation planners for long-range planning – SBDTA – dynamic and large-scale The scenarios of interest may result in shifts of network or corridor-wide traffic flow patterns. – Significant change to roadway configuration – Certain corridor management strategies Transportation Operations Group Mesoscopic Dynamic Traffic Assignment DynusT v2.0 – Free version available for DYNASMART-P users Dynamic simulation and assignment tool for regional operational planning analysis Equilibrium-based Dynamic Traffic Assignment – Assigned paths are based on experienced (actual) travel time Applications – Assess impacts of ITS technologies – Work zone planning and traffic management – Evaluate HOV/HOT lanes – Congestion pricing – Special event/emergency evacuation Transportation Operations Group Microscopic VISSIM 5.1 A driver-behavior-based simulation tool capable of performing multiple applications including – Analyzing complex intersections – Border crossings inspection booths – Managed lanes – University campus settings Fined-grained analysis – Vehicle interactions – Individual lane analysis Simulate multiple modes of transportation simultaneously 3-D graphics Transportation Operations Group Outline Introduction – Mesoscopic – Microscopic Multi-Resolution Modeling – Concept – Conversion Process – Modeling Issues Case Study Applications Transportation Operations Group Concept What is multi-resolution modeling? – Integrating mesoscopic and microscopic models for the purpose of achieving a specific goal » Analyze network at both the system-wide and localized levels Why is multi-resolution modeling so important? – Mesoscopic & microscopic models are not mutually exclusive – They are complimentary to one another and can accomplish optimal modeling capabilities. – Retain the best characteristics of both » Realistic representation of regional traffic » Detailed interactions Transportation Operations Group Concept Sub-area Cut Mesoscopic Model Model Conversion Process Integration Tool Microscopic Model Transportation Operations Group Concept DynusT SubArea VISUM VISSIM Transportation Operations Group Multi-Resolution Modeling Framework Field Data Regional Travel Demand Model Initial Network Conversion (DynusT) Calibration •Speed Profile •OD •Traffic Model Sub-Area Cut Yes Detailed Analysis No DVC VISSIM Calibration Rerun DTA Network Modification Transportation Operations Group DTA Model Preparation Convert the GIS layer of the Travel Demand Model to Mesoscopic format. Disaggregate 24-hour matrix based upon car & truck – – – – – – – Home to work Work to home Home to private Private to home Thru External Local Non-home based external local Multiply each matrix by corresponding hourly factor Transportation Operations Group DTA Model Preparation Multiply each matrix by hourly factor H-W W-H H-P P-H THRU EXLO NHBEXLO Summation of matrices gives you directional 1-hour matrix Transportation Operations Group DTA Model Preparation 24 - one hour matrices Transportation Operations Group Calibration Traffic flow model – Traffic simulation in DynusT is based upon the Anisotropic Mesoscopic Simulation (AMS) model – Moves vehicles based upon speed-density (v-k) relationship – v-k relationship is derived from Greenshields equation Transportation Operations Group Calibration – Minimize the deviation between simulated and actual screen line counts & speed profile – Iterative process – Program solves linearized quadratic minimization problem – Results in updated OD matrices Estimated TimeDependent OD Matrices Traffic Assignment/ Simulation Assignment Results Linear Optimization Model Update Demand Time-Dependent OD Traffic Network Traffic Flow Model Intersection Controls Results Optimized Affected, Time-Dependent OD Pairs Transportation Operations Group Conversion Process Sub-area cut – Remove unneeded sections of network – Renumbering of new zones, nodes and links – Retains paths and flows that travel through the sub-area Transportation Operations Group Conversion Process DynusT-VISSIM Converter – Developed by researchers from TTI and UA – Converts roadway network to VISUM network – Retains network geometry – Converts all timedependent paths and flows – Creates separate transportation systems (car, truck) Transportation Operations Group Conversion Process Microscopic model – Calibrate VISSIM model to reflect realistic roadway conditions – Perform detailed “finegrained” analyses » Speed profile for individual lanes » Lane-changing behaviors » Vehicle interactions at merge areas – Create 3-D graphics for presentations Transportation Operations Group Modeling Issues Consistency – Network » Lane configuration » Geometric design – Paths and flow » Verify same origin/destination paths » Verify number of vehicles generated – Speed profile » Perform field data collection to determine speed and vehicle counts » Obtain v-k curve from simulation output » Calibrate models with field data Transportation Operations Group Modeling Issues 40 35 30 Speed (mph) 25 Observed 20 Calibrated 15 10 5 0 0 20 40 60 80 100 120 140 160 Density (veh/mi) Transportation Operations Group Modeling Issues Field Data Regional Travel Demand Model Initial Network Conversion (DynusT) Calibration •Speed Profile •OD •Traffic Model Sub-Area Cut Yes Detailed Analysis No DVC VISSIM Calibration Rerun DTA Network Modification When Feedback is Necessary Transportation Operations Group Outline Introduction – Mesoscopic – Microscopic Multi-Resolution Modeling – Concept – Conversion Process – Modeling Issues Case Study Applications Transportation Operations Group Case Study City Council proposes ordinance to restrict trucks from using left lane on I-10 corridor How does the ordinance affect the freeway and surrounding arterials? Transportation Operations Group Case Study Which type of model do I use? Transportation Operations Group Case Study Truck restricted lanes – A case study to analyze the effectiveness of restricting trucks from left-most fast lane on freeway – 22-mile corridor of I-10 in El Paso, TX – Analyze a.m. peak, p.m. peak, & mid-day – Determine benefits » Speed on left-most lane » Acceleration/Deceleration patterns » Vehicle interactions at merge areas – DynusT estimates region-wide truck trajectories (route and flows) – VISSIM models detailed IH-10 truck lane operations given truck trajectories Transportation Operations Group Model Development 106 Origin/Destination links - 1895 Transportation Routes created Operations Group Model Development GPS unit was used to input freeway grading information Transportation Operations Group Model Development Field data collection-freeway speed profile (PM peak hour) Transportation Operations Group Model Development Data provided by TxDOT Automatic TrafficTransportation RecorderOperations Stations Group Model Development Transportation Operations Group Case Study I -10 EB @ Sunland Park (7-11 am) 65 64 63 62 61 60 59 58 57 56 1.5 Acceleration (ft/s2) Speed (mph) I-10 EB @ Sunland Park (7-11 am) 1 0.5 0 -0.5 -1 -1.5 0 3600 7200 10800 14400 0 3600 7200 Restricted Speed 14400 Time (sec) Time (sec) Base 10800 Base Restricted Accel/Decel Transportation Operations Group Case Study I-10 EB @ Paisano (3-7 pm) I-10 EB @ Paisano (3-7 pm) Acceleration (ft/s2) Speed (mph) 80 60 40 20 0 0 3600 7200 10800 14400 2 1.5 1 0.5 0 -0.5 -1 0 3600 7200 Time (sec) Base Restricted Speed 10800 14400 Time (sec) Base Restricted Accel/Decel Transportation Operations Group Case Study I-10 WB @ Paisano Left Lane (3-7 pm) I-10 WB @ Paisano Right Lane (3-7 pm) 80 Speed (mph) Speed (mph) 80 60 40 20 0 60 40 20 0 0 3600 7200 10800 14400 0 3600 7200 Time (sec) Base 10800 14400 Time (sec) Restricted Base Restricted Speed – Left vs. Right Lane Transportation Operations Group Outline Introduction – Mesoscopic – Microscopic Multi-Resolution Modeling – Concept – Conversion Process – Modeling Issues Case Study Applications Transportation Operations Group Applications Managed lanes – – – – Truck restricted lanes HOV lanes HOT lanes Time-dependent variable pricing Transportation Operations Group Applications Geometric design alternatives – Freeway direct connect » Various design configurations – Ramp reconfiguration » Braided ramps » “X” ramps Transportation Operations Group Applications Traffic impact studies – New retail shopping centers » Driveways » Pedestrian crossings – University campus planning » Integrating various modes of transportation (e.g. student, faculty, staff, pedestrians, transit) » New parking facilities » Campus core closure – Traffic calming Transportation Operations Group Questions ? Transportation Operations Group