Learning Center
Plans & pricing Sign in
Sign Out

Slide 1 - The University of Texas at Austin


									System-Level Analysis of Austin Arterials

  Presented by the University of Texas at Austin Center for
  Transportation Research for TxDOT
  Dr. S. Travis Waller
  Dec 15th, 2008
 Traditional Transportation Planning

§ Trip Generation
  § Based on demographic data and land use,
    develop the number of trips originating or ending
    in a region
§ Trip Distribution
  § Assigns specific origins and destinations to trips
§ Mode Split
  § Determines mode of travel for each trip
§ Traffic Assignment
  § Assigns trips onto specific roadway segments
  Traffic in Traditional Planning

§ Traditional planning only addresses vehicular traffic
  through Static Network Assignment
§ Major benefit of static assignment
   § Static assignment is relatively easy to solve making regional
     models possible in previous decades
§ Major detriment of static assignment
   § Static assignment treats traffic as a single steady state value
     (time insensitive) and neglects critical traffic behavior such as
     queue formation and traffic spillover from one street onto

§ Dynamic traffic assignment has been developed to
  maintain regional network modeling with correct
  traffic behavior
   § Including other operational considerations such as transit
     service, pricing, ITS, etc.
Expanding Traditional Planning
     Comparison: Static vs. Dynamic
•   Volume/ Capacity can be          • Volume/Capacity is always less
    greater than 1                     than 1
     – Inconsistency with real          – Consistent with real world
        world traffic counts               counts
                                     • Inflow >= Outflow
•   Inflow must = Outflow
                                        – Models queue formation,
     – Does not adequately                 shockwave propagation,
        model congestion where             congestion formation and
        inflow > outflow such as           dissipation
        queue formation              • Ideal for transit operations,
•   Cannot be used for signal          signal optimization, ITS
    optimization, ITS devices,         devices, ramp metering, transit
    precise transit modeling, ramp     operations, dynamic pricing
    metering                           and any time-varying traffic
     Impact of Traffic Dynamics
     One example: queue formation and dissipation due to signals

                                              DTA captures precise
                                              vehicle movements
                                              throughout the network

Static assignment can not capture traffic
signalization (due to the omission of time
Signals cause specific non-uniform impact
on traffic however (as shown)
Queues form and dissipate
    Scope and Objective
§ Implement cutting-edge Dynamic Traffic Assignment (DTA)
    techniques in regional traffic network modeling
§   Develop methods for incorporating the new modeling
    approach into the conventional 4-step travel demand
    modeling process
§   Develop a reliable, flexible, well-calibrated network traffic
    model for both transportation planning and operations
§   Conduct multi-modal, multi-level network-wide analysis and
    scenario evaluations

§ Outcomes:
     § Calibrated model for system-wide evaluation in central Texas
     § Integrated modeling process applicable to other regions
   Example Analysis Options: Traffic Humps (hypothetical)

Setting: Models reduction in speed by
reducing the free flow speed and
Saturation Capacity
Metrics: Changes in travel times, level of
service in the affected street and
surrounding neighborhoods
Capabilities: DTA can track individual
vehicle trajectory, and observe the path
change due to the introduction of speed
humps & alternate routes selected by
drivers plus the change in link flows
  Traffic Humps (Cont’d)

•About 1,200 vehicles during the peak hours     An example trip            Before
choose alternative routes, which accounts for    to Red River              After
56% reduction of the traffic volume
•Examples of changes in vehicle trajectories
     •To Red River (5%)
     •To I-35 (4%)
     •To Guadalupe (13%)
     •To Speedway (30%)

                                                          An example trip to I-35
Change Two-Way Street to One-Way (hypothetical)

Location: Set Guadalupe St (between W 29th St and FM 969 (MLK Blvd))
a southbound-only street
Setting: Modeled as a direction elimination and capacity reallocation of the
corresponding link
Metrics: Changes in travel times,
level of service & link flows in the
target street and its vicinity;
changes in vehicle trajectories;
changes in O-D travel time
  Change Two-Way Street to One-Way (Cont’d)

•About 1200 vehicles
detour during the peak
•Examples of changes
in vehicle trajectories
                                         An example trip
    •To Lamar (30%)                       to Lamar Blvd

    •To Rio Grande
    •To I-35 (5%)
Change One-Way Street to Two-Way (hypothetical)

Location: Set W 6th St from I-35 to MO-Pac Express to be a two-way street
Capabilities: Can track individual vehicle trajectory which can be aggregated
to obtain changes in link flows
                                            Metrics: Changes in travel times,
                                            level of service and link flows in
                                            the target street and its vicinity
                                            and system-wide traffic; changes
                                            in vehicle trajectories
   Change One-Way Street to Two-Way (Cont’d)

•The total link
flow is reduced
by 1,388
vehicles, which
accounts for 53%
of the traffic flow
•The largest part
of the detouring
vehicles (37%)
switch to a
parallel route, 5th
    Street Widening (hypothetical)

Dynamic assignment allows for system-wide
routing decisions to be explained as well as
precise time evolving traffic representation.

Metrics: Changes in travel times, level of
service and link flows in the affected street
and surrounding neighborhoods; changes in
vehicle trajectories and O-D travel times
     Network Improvement: (hypothetical)
         Street Widening
• About 1,400 vehicles during the peak hours
are attracted from other streets, which accounts
                                                   An example trip from S
for 30% increase of the traffic volume                    1St Ave
•7mph speed improvement
•Examples of changes in vehicles trajectories              After
    •From I-35 (1%)
    •From S 1st (10%)

                                                                    An example trip from I-35
Critical Link Analysis made possible
from Vehicle Path Analysis

 Traffic flow that uses a
southbound MO-Pac link

                            Traffic flow uses a northbound
                                       MO-Pac link
     Other Example Analysis Options

§ ITS Deployment
§ Dynamic pricing
§ More efficient corridor studies
§ Transit simulation
§ Improved mode choice
§ Etc.
    Overview of the Study Area

The Greater Austin Area, Central Texas
                                         •Approach boundaries are the CAMPO
                                          region: Travis, Williamson and Hays
                                         •Easily expandable to include other
                                          counties as well (Bastrop, Caldwell,

                                         •STILL NEED UPDATED FIGURE
Modeling Technique: Dynamic Traffic Assignment
                           §   Regional model scope
                           §   Captures time-varying
                               network & traffic movements
                           §   Correctly models street
                               capacities on traffic
                           §   Compatible with both
                               regional models and localized
                               traffic models
                           §   Capable of handling large-
                               scale networks while
                               modeling traffic control, ITS
                               and transit while tracking
                               individual vehicle movements
                           §   DTA output: time stamped
                               individual vehicle movements
                               through network
Model Development Procedure
     Network Profile (Updated)

§   Source -CAMPO
     § Nodes (9,622)
     § Links (19,519)
     § O-D demand pairs (234,729)
§   Source - City of Austin
     § Signals (580)
         • Pretimed: 456
         • Self-optimized: 124
§   Source - CAPMetro
     § Transit lines (316)
         • Covered road segments:
         • Bus stops: 4,202
     Highway Network Calibration
§ Calibration components
   § Supply: Road capacity and free-flow speed, signal phasing and
     timing settings, bus dwell times
   § Demand: O-D table and profile (from CAMPO: unmodified)
§ Calibration criteria
   § Road Network: traffic volume
   § Transit Network: GPS data on bus routes
§ Calibration data
   § Approximately 1,000 traffic counts of major highways and
     arterials (CAMPO/TxDOT)
   § Multiple days of GPS data for 14 bus routes (CAP Metro)
    Network Calibration Results

•RMSEs by the functional
 classification are:
     •18% (freeways)
     •22% (major arterials)
     •23% (minor arterials)

•Models substantially exceed
 typical planning standards for
 arterial street calibration
    The Calibrated Network Flow Pattern

•Traffic flow rate at
 the peak period
•The color scheme
 shows traffic
•Many bottlenecks in
 the vicinity of
 interchanges and
•Congested roadway
 segments on I-35,
 TX-1 (MO-Pac), TX-
 685, FM-620
Example Corridor Errors I-35
                IH 35 @ FM 969

            5% (211 veh.)


Example Corridor Errors - MoPac
      MoPac between US-183 and Cesar Chavez
  Transit Network Calibration

§ Calibration components
   § Transit route-link calibration
   § Transit schedule calibration
§ Calibration criteria
   § Schedule adherence
   § Headway variation
§ Calibration data
   § GPS-measured timepoint-based bus run time
   § Data collected January-June 2008
   § 20-50 whole-route run time samples available for each
     surveyed route during morning peak hours on weekdays
Transit Service Area
    Calibrated Bus Routes (GPS Data)

• Local service routes
   •   001 1l/1M, 003 Burnet & Manchaca
• Limited & Flyer Routes
   •   135 Dell Limited, 174 North Burnet
• Feeder Routes
   •   214 Lago Vista Feeder, 240 Parkfield
• Crosstown Bus Routes
   •   350 Airport Blvd, 383 Research
• ‘Dillo & Special Services
   •   412 Main Campus, 450 Congress ‘Dillo
• Express Routes
   •   982 Pavilion Express, 983 Noth 183
Local Service Route Example: Route 1 SB
Local Service Route Example: Route 1 SB
Cross Town Route Example: Route 350 NB
Cross Town Route Example: Route 350 NB
Express Route Example: Route 982 SB
Express Route Example: Route 982 SB
    Transit System Calibration Result

§   Schedule adherence compares the GPS-measured arrival times
    and the simulated arrival times against the scheduled arrival times
    over the transit system.

§   Arrival times are evaluated at the time points along each bus route,
    as defined by Capital Metro.

To top