toward a consistent and robust integrated multi-resolution modeling approach for traffic analysis

Transportation Operations Group

Toward a Consistent and Robust Integrated Multi-Resolution Modeling

Approach for Traffic Analysis

May 17-21, 2009

Jeff Shelton, TTIYi-Chang Chiu, Univ. of Arizona

TRB – Transportation Planning ConferenceHouston, TX


Outline

Introduction– Mesoscopic– Microscopic

Multi-Resolution Modeling– Concept– Conversion Process– Modeling Issues

Case Study

Applications


Introduction

Integrating mesoscopic dynamic traffic assignment (DTA) and microscopic traffic simulation and assignment models can be advantageous for region-wide 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


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


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


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


Outline



Case Study

Applications


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


Concept

Mesoscopic Model Model Conversion

Process Integration

Tool

MicroscopicModel


Concept

DynusT VISSIMVISUMSub-Area


Multi-Resolution Modeling Framework

Regional Travel

Demand Model

Initial Network

Conversion (DynusT)

Calibration•Speed Profile•OD•Traffic Model

Sub-Area Cut

DVC

VISSIMCalibration

Network Modificatio

n

Field Data

Rerun

DTA

Detailed Analysis

No

Yes


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



H-W W-H H-P P-HTHR

UEXLO

NHBEXLO

Multiply each matrix by

hourly factor

Summation of matrices gives you directional

1-hour matrix



24 - one hour matrices


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


Calibration

Time-Dependent OD – 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

Traffic Network Traffic Flow Model

Intersection Controls

Estimated Time-Dependent OD

Matrices

Traffic Assignment/Simulation

Linear Optimization

Model

Optimized Affected, Time-Dependent OD

Pairs

Results

Up

date

Dem

an

d

Assignment Results


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


Conversion Process

DynusT-VISSIM Converter– Developed by researchers

from TTI and UA– Converts roadway

network to VISUM network

– Retains network geometry

– Converts all time-dependent paths and flows

– Creates separate transportation systems (car, truck)


Conversion Process

Microscopic model– Calibrate VISSIM model

to reflect realistic roadway conditions

– Perform detailed “fine-grained” analyses» Speed profile for

individual lanes» Lane-changing

behaviors» Vehicle interactions

at merge areas– Create 3-D graphics for

presentations


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


Modeling Issues

Density (veh/mi)

Sp

eed

(m

ph

)


Modeling Issues

When Feedback is Necessary

Rerun

DTA

Regional Travel

Demand Model

Initial Network

Conversion (DynusT)

Calibration•Speed Profile•OD•Traffic Model

Sub-Area Cut

DVC

VISSIMCalibration

Network Modificatio

n

Field Data

Detailed Analysis

No

Yes


Outline



Case Study

Applications


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?


Case Study

Which type of model do I use?


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


Model Development

106 Origin/Destination links - 1895 Routes created


Model Development

GPS unit was used to input freeway grading information


Model Development

Field data collection-freeway speed

profile (PM peak hour)


Model Development

Data provided by TxDOT Automatic Traffic Recorder Stations


Model Development


Case Study

56575859606162636465

0 3600 7200 10800 14400

Spee

d (m

ph)

Time (sec)

I-10 EB @ Sunland Park(7-11 am)

Base Restricted

-1.5

-1

-0.5

0

0.5

1

1.5

0 3600 7200 10800 14400

Acce

lera

tion

(ft/s

2)

Time (sec)

I -10 EB @ Sunland Park(7-11 am)

Base Restricted

SpeedAccel/Decel


Case Study

0

20

40

60

80

0 3600 7200 10800 14400

Spee

d (m

ph)

Time (sec)

I-10 EB @ Paisano(3-7 pm)

Base Restricted

-1-0.5

00.51

1.52

0 3600 7200 10800 14400

Acce

lera

tion

(ft/s

2)

Time (sec)

I-10 EB @ Paisano(3-7 pm)

Base Restricted

SpeedAccel/Decel


Case Study

0

20

40

60

80

0 3600 7200 10800 14400

Spee

d (m

ph)

Time (sec)

I-10 WB @ PaisanoLeft Lane (3-7 pm)

Base Restricted

0

20

40

60

80

0 3600 7200 10800 14400

Spee

d (m

ph)

Time (sec)

I-10 WB @ PaisanoRight Lane (3-7 pm)

Base Restricted

Speed – Left vs. Right Lane


Outline



Case Study

Applications


Applications

Managed lanes– Truck restricted lanes– HOV lanes– HOT lanes– Time-dependent

variable pricing


Applications

Geometric design alternatives– Freeway direct connect

» Various design configurations

– Ramp reconfiguration» Braided ramps» “X” ramps


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


Questions ?

toward a consistent and robust integrated multi-resolution modeling approach for traffic analysis

Documents