CTS Workshop on: Access to Destinations Twin Cities Metrowide Traffic Micro Simulation: Feasibility

1
CTS Workshop on Access to DestinationsTwin
Cities Metro-wide Traffic Micro Simulation
Feasibility Investigation

• Panos Michalopoulos
• John Hourdos
• University of Minnesota Civil Engineering
  Department
• February 28th, 2007

2
Access to Destinations

• Project objective Measure accessibility for
  periodic evaluation the TC transportation system
  (LOS)
• Needs/requirements
• Travel Times
• Historical
• Future estimates
• Freeways
• From Any ramp to Any ramp
• 5-15 minute intervals
• Arterials
• Link travel times

3
Issues

• Freeway sections not all instrumented
• No queue detectors before 2003
• No historical data for most sections especially
  prior to 1994.
• Gaps and/or errors in loop data
• Arterials
• Instrumentation very sparse no records
• Only reliable way that meets needs is simulation
• Detailed future Travel Time forecasting
• Can be combined with Planning tools to
  incorporate land use other planning scenario.

4
Issues (cont.)

• Simulation needs to be on a Metro-wide basis to
  meet Accessibility project needs
• Approach should satisfy other needs requiring
  substantial time and effort
• Examining effects of HOT lanes on a
  corridor/metro-wide basis.
• Geometric improvements planning
• ATMS/ATIS
• Real time incident management
• Travel Time predictions
• Impacts of Transit Improvements/Expansion

5
Project Objectives

• Explore feasibility of Metro-wide simulation in
  Twin Cities
• World wide experience
• Local stakeholder needs
• Data requirements availability
• Existing tools suitability
• Implementation plans

6
World Wide Experience

• Considered best proven micro-simulators used in
  practice.
• Suitable for both Traffic Planning
• AIMSUN, CUBE, PARAMICS, TRANSMODELER, VISSIM
• 25 related projects and growing.
• Most important presented here

7
Projects Found

• AIMSUN NG (AIMSUN Planner AIMSUN)
• Hessen, Germany
• Singapore
• Barcelona, Spain
• Lausanne, Switzerland
• Helsinki, Finland
• Madrid, Spain
• Stockholm, Sweden

• PTV Vision (VISUM VISSIM)
• London, UK
• Madrid, Spain
• Birmingham, UK
• Munich, Germany
• Berlin, Germany
• Dubai, UAE
• Columbus, OH
• Kansas City
• Reno, NV
• Seattle, WA
• Houston, TX
• Denver, CO
• San Diego, CA

• TRANSMODELER (very new, Transcad)
• Orange County I-5 freeway
• Louisville, Kentucky
• Lexington, Kentucky

• PARAMICS
• Sydney, AU
• Xanadu, NY
• Manhattan, NY
• IH-94, WisDOT
• Toronto, CA

• CORSIM, CUBE
• No Data

8
Issues in Acquiring Info

• New technology information hard to get.
• Communication delays/responsiveness.
• Reluctance to respond to questionnaire (time
  consuming, confidentiality issues, other)
• No responses to emails or phone calls
• Used personal communications
• Frustrating Experience (started in May 2006)

9
Summary of Findings

• Wide range of applications including
• Real time large scale traffic management
• Advanced urban planning
• Most mature large implementations are overseas.
• Most projects had specific objectives.
• With out plans for on going use.
• Found few failures due to multiple model
  incompatibilities.
• Reluctance to disclose cost effort.

10
Island of Singapore

• Total length1650 Km
• Total lane length 4483 Km
• Intersections 10,580
• Zones 3153

11
Island of Singapore

• Objective
• Road network control
• Off-line management scenario analysis
• Real-time simulation scenarios for Incident
  Congestion Management.
• Provides short term forecasts.
• Traffic control center operations application.
• In operation since April 2006
• Was based on O/D data from original regional
  model.
• Problems occurred but the 3 year length of the
  project allowed for successful resolution.
• Current O/D data are maintained inside the new
  system and adjusted periodically.
• Calibrated with volume, travel time ,and
  classification data on key locations of the
  network.
• Cost not disclosed.

12
Real-time Traffic Management
Hessen, Germany
Total lenght1,512 Km Zones 245 Intersections
462 Controlled 200
Madrid, Spain (TSS)
First Phase (current) Intersections
2,109 Zones 207 Second Phase Intersections
8,000 Zones 2,000
13
Barcelona, Spain

• Total Length 1,438 Km
• Intersections 7,901
• Zones 344
• Completed 2002

14
Barcelona, Spain

• Collaboration between urban planning, transit ,
  and operation authorities.
• Objective
• Serve as a Master Model to support operational
  planning projects in various districts of the
  city.
• Projects since completion
• Studies in urban reshaping.
• Development of traffic management schemes in four
  city districts (2 completed).
• Feasibility study of Light Rail line through the
  CBD (ongoing).
• Initial O/D information from regional model
  (EMME/2).
• Only the 24h matrix was available.
• Traffic data for master model matrix adjustment
  and calibration was a problem.
• Complementary study conducted to determine best
  detection layout and identify number and location
  of additional detectors.
• Detector installation is ongoing.

15
Madrid, Spain

• Total Length 225 Km
• Links VISUM 8,345, VISSIM 926 (45 controlled)
• Zones VISUM 536, VISSIM 126
• Completed

16
Madrid, Spain

• Preceded earlier project (not real-time
  capability)
• Objective
• Renovation of the urban ring road M30 around the
  old city center of Madrid.
• Estimate hourly traffic flow for different
  projection years (2007, 2012, 2037).
• Capacity analysis at bottlenecks (weaving
  sections, on/off-ramps, spill-backs from adjacent
  intersections onto M30).
• Finding remedial measures to prevent bottlenecks
  at reasonable construction cost.
• Original data from regional model (EMME/2)
• Only peak periods available (two 4hour matrixes)
• Calibration
• Original EMME/2 demand model was calibrated
  against 55 sites counted VISUM model was
  recalibrated and fine-tuned against 20 detector
  counts.
• Effort spend
• 24 person months for converting the regional
  model to VISUM, minor recalibration of VISUM and
  building the future M30 cases in VISUM and VISSIM

17
Columbus, Ohio

• Area 30 sq miles
• Intersections 400
• 330 signalized
• 25 freeway interchanges
• Zones 295 VISUM 65 VISSIM
• Completed 2005

18
Columbus, Ohio

• Objective
• Investigate access to downtown area during AM
  peak period in anticipation of 377 population
  growth.
• Converted regional model (CUBE) to VISUM model.
• Adjusted O/D and assignment based on measured
  link flows.
• VISSIM model calibrated with data collected from
  detectors and from extensive manual counting
  efforts and floating car studies.
• New calibration methodology was necessary.
• No planned use of the model beyond current
  project.

19
London, UK

• Total Length 220 miles (VISUM)
• Intersections 2000
• 250 signalized (SCOOT)
• Started in 2006

20
Lausanne, Switzerland

• Total length 671 Km (area 100Km2 )
• Intersections 1,476, Controlled 148
• Zones 292
• Completed in 2004

21
Sydney, Australia

• Area 12 sq. Km
• Intersections 600, 200 signalized
• Completed 2002

22
Berlin, Germany

• Total Length VISUM 6,019 Km, VISSIM 272Km
• Intersections VISUM 1602, VISSIM 217 (98
  controlled)
• Zones VISUM 1118, VISSIM 152
• Completed

23
Munich, Germany

• Total Length 1,113 km
• Intersections 98 signalized, 6 Freeway
  interchanges
• Completed 2004

24
Birmingham, UK

• Total Length VISUM 21,200
• Intersections VISUM 5,895
• Zones VISUM 898
• Completed 2006

25
Local Needs

• Interviewing local stakeholders to explore needs
  and potential desire for metro wide
  micro-simulation.
• Metropolitan Council
• City of Minneapolis
• FHWA
• Regional Traffic Management Center
• Hennepin Co.
• Mn/DOT (on going)
• City of Eden Prairie
• City of Maple Grove

26
Summary of Findings (1)

• Current planning process is static.
• No real-time control, metering, ITS improvements
  can be assessed.
• Equilibrium Peak period O/D matrixes produced
• Implying constant Travel Times during the Peak
  Periods.
• Manual sort term partitioning of O/D for detailed
  analysis.
• Trial Error
• Calibration based on detector data also manual.
• Traversal O/D matrixes on sub-areas cannot be
  extracted with available SW (TP/CUBE)

27
Summary of Findings (2)

• Majority agrees on notion of at least Freeway
  Highway network micro-simulation.
• Application for assessing Ramp Metering effect
  including diversion in entire system.
• Model would assist in evaluating impact of major
  transportation projects (LRT, I-394 HOT lane)
• Model could be used for improving accuracy of
  Metro Planning model (traffic assignment).
• Impacts of
• Special events.
• Work zones their interactions

28
Summary of Findings (3)

• Improving Ramp Metering, optimization and/or new
  algorithm
• Planning for special events
• Real time management of special events
  (incidents)
• Post-incident management analysis and training.
• Access to Destinations Needs
• At minimum hourly Travel times (past projected)
• Land use interaction with transportation network
  development
• Dynamic effects of ATIS/ATMS for assessing
  effects on Accessibility

29
Special Software Requirements

• Transit operations modeling
• Priority/Exclusive lane modeling
• Dynamic Traffic Assignment capability
• API for interfacing with external systems
• Ability to Interact with planning package
• Ability to extract analyze a sub-area
• Partitioning of O/D matrices to short term
  periods.
• Similar needs recognized worldwide
• Trend is to integrate planning with
  micro-simulation
• Automation of a disjoined manual processes.

30
New Approach

• Next Generation Method Integrated Solution

• Traditional Method

1
Four Step Process
Manual Process
2
Analysis of a Specific Area O/D matrices
extracted from planning package and entered
separately into microsimulator

• 1 2 performed by different entities
• Static period O/D, manual partitioning

31
Micro Simulator Comparisons

• Several simulators commercially available
• Stand alone or integrated with the planning
  package
• Objective is to identify the top 2-3 most
  suitable for Twin Cities metro mrea based on SW
  requirements identified.
• Comparison in progress

32
Five Integrated Packages Selected for Review

• TransCAD/TransModeler
• Cube (Voyager Dynasim)
• Paramics
• PTV Vision (VISUM VISSIM)
• AIMSUN NG (AIMSUN PLANNER AIMSUN)
• Dynasmart-P and CORSIM do not meet
  micro-simulation or integration requirements.

33
All Packages Appear to Satisfy Most Criteria
Geometry Links and Nodes
34
Other Factors To Consider In Making A Provider
Decision

• Track record of the provider
• Actual implementation record in large scale
  applications.
• Proven methodology employed in the integrated
  macro and micro solution
• Fit of provider to local IT footprint
• Is it adaptable to the data bases / planning
  infrastructure / practices of the Metro Area?

35
Conclusions

• Implementation to Twin Cities is timely
  feasible subject to data availability.
• The most logical and feasible step is modeling of
  the entire Freeway Network
• Doable in lt2 years due to earlier work
• Accessible to State, Local agencies and
  consultants for saving time and in many
  applications.
• Project is still on going

36
Questions
37
Munich, Germany

• Objective
• Simulation test site for new Traveler information
  and management technologies
• En-route navigation with real-time travel time
  info.
• Dynamic speed control
• Shoulder lane usage
• VISEM activity based demand model.
• Socio-demographic data
• Household surveys
• Produced hourly O/D matrices per trip purpose.
• Calibrated based on average weekly traffic counts
• Microscopic model calibrated with 98 strategic
  detector sites
• 45 on freeways and 53 on urban streets
• Data collection effort 3 man months identify
  representative count profiles from a database of
  traffic counts plus re-calibration of the OD
  matrices.
• Project effort 18 man month.

38
Berlin, Germany

• Objective
• Traffic Management study of the Berlin CBD
  (VISSIM part of project).
• The bigger VISUM model was generated to support
  trip generation for the area.
• Model build for specific project and not used
  anymore.
• O/D information was based on Regional model in
  VISEM.
• O/D adjusted with traffic counts
• 250 urban road detectors
• 200 freeway detectors.
• Calibration of Micro model with detector
  measurements
• Effort spend 9 man months for the VISSIM part
  only.
• Mainly coding of intersections.

39
Birmingham, UK

• Objective
• Develop regional model in VISUM and PRISM
  geo-database
• Introduce enough detail to allow VISSIM
  applications.
• Extension of existing LRT.
• Coventry ring road study.
• Supports assessment of various issues
• Local transport planning
• Metro and LRT extensions
• Park and Ride schemes
• Motorway widening
• General regional and local planning and policies
• Regional model developed and calibrated with
  census and household survey data.
• Roadside interviews and detector counts used to
  calibrate path costs and link flows.
• Cost several million USD

40
Manhattan, NY

• Objective
• Evaluate designs for the reconstruction of Route
  9A passing from the old Twin towers site.
• Model is currently used on other projects.
• Modeled 5 sq. miles of lower Manhattan.
• Many signals.
• Data from the regional model.
• Matrices adjusted based on volumes, speeds
  collected from detectors.
• 15 minute O/D matrices
• Model calibrated based on volume, speeds, and
  queues (manually collected).
• Cost Modeling part 200K-300K (50 was data
  collection). Overall project gt1M.
• Duration 1.5 years

41
Xanadu, NY

• Objective
• Evaluate from new Giants sports stadium
  construction.
• Project was for private entity so model was not
  used for anything else.
• Area 5 sq. miles.
• Very complicated, with parking isles.
• Standard and toll freeway interchanges
• Data from the regional model.
• Matrices adjusted based on volumes, speeds
  collected from detectors.
• 10 minute O/D matrices.
• Model calibrated based on volume, speeds, and
  queues (manually collected).
• Over 28,000 trips fro the stadium alone plus
  background traffic.
• 6-10 different time periods to represent lane and
  link closures by the sports authority.
• Cost 1M (50 on data collection) for
  calibration/validation.
• Duration 1 year

42
IH-94, WisDOT

• Objective
• Study the impacts of growth on the IH-94 corridor
  between the Illinois border and Milwaukee.
• Sub-areas are used for other projects.
• Size about 40 miles of freeway with two
  alternative routes.
• 165 zones
• Received from client only ramp counts. Generated
  O/D from that with Paramics Estimator because the
  caretaker of the planning model did not cooperate
  with the study team.
• Study team conducted O/D study fro one major
  interchange
• 15 minute O/D matrices
• Calibrated with automatically and manually
  collected volumes, speeds and queues.
• Cost 300K-400K
• Project is ongoing.

43
Sydney, Australia

• Area 12 sq. Km
• Intersections 600, 200 signalized
• Completed 2002

44
Sydney, Australia

• Objective
• To serve as Master model for metro area studies.
• O/D based on regional model.
• Manually increased zone density
• Used car park and link volume data to adjust.
• Effort 2000 man hours.

45
Vision

• Politicians/Management/Engineers often need quick
  answers to pressing questions/problems.
• Answers cannot always be readily produced.
• In most projects one has to start from scratch in
  terms of model building and data collection.
• Time consuming and expensive
• Deal breaker
• Simulation has matured for actual implementation
  in traffic and planning.
• Worldwide applications in
• Real time and off-line traffic management
• Metro/CBD access studies
• Transportation operations planning
• Alternative network design selection/effectiveness
  
• Specialized projects (special events, incident
  management, etc.)

46
Vision

• Trent is to integrate planning models with
  micro-simulation
• Vision
• Expand Metro area model capabilities to support
  micro-simulation.
• Begin to develop and maintain necessary
  databases.
• Simulate on demand for planned or emergency
  projects.
• Save time cost
• Obtain robust comprehensive solutions.
• Use for ongoing improvements (what if.)