MATRIX ADJUSTMENT MACRO (DEMADJ.MAC AND DEMADJT.MAC) APPLICATIONS:

SEATTLE EXPERIENCE

Murli K. Adury

Youssef Dehghani

Sujay Davuluri

Parsons Brinckerhoff Quade & Douglas, Inc.

Seattle, WA, USA

Eric Chipps

Sound Transit

Seattle, WA, USA

Project(s) Context

Microsimulation on Two Major Highway Corridors :» I-405 Corridor Alternatives Analysis

– Highway Capacity Expansion Alternatives to Alleviate Congestion on I-405

» Alaskan Way Viaduct (AWV) Project– Project to Replace and Rebuild a highway

vulnerable to earthquake– Microsimulation to Evaluate Traffic Impacts Due

to Construction

Project Area Map- I-405 Corridor Study

- Alaskan Way Viaduct Project

Microsimulation for I-405 & AWV - 1

- Purpose and Need

I-405 Study» Be able to evaluate freeway capacity enhancement

elements of the preferred alternative» Be able to refine and/or modify capacity

enhancement elements of the preferred alternative » Be able to highlight potential deficiencies – e.g., lid

concept thru downtown Bellevue AWV Project

» Evaluate options for construction sequencing» Develop traffic control plans as part of overall

mobility plan

Microsimulation for I-405 & AWV- 2

INTEGRATION was selected as preferred microsimulation software program

INTEGRATION model to interface with regional EMME/2 Travel Demand Model» Obtain base highway network – freeway

and parallel arterials» Obtain PM peak hour O-D trip tables

Microsimulation for I-405 & AWV – 3

Model Development Process

Network preparation/calibration Data collection

» Details of network geometry» Signal timings » Traffic counts

Trip table development Model validation analysis Model application

Microsimulation for I-405 & AWV– 4

Trip Table Development

Regional model produced PM peak period vehicle trip tables for HOVs and Non-HOV» Apply factors to convert peak period trips to peak

hour trips Post regional modeling procedure:

» Directional peak-hour base year counts were seeded on designated links on I-405 and arterials

» Matrix adjustment macro DEMADJ.MAC developed Heinz Spiess was used to update original matrix

Microsimulation for I-405 & AWV– 5

Trip Table Development Continued

Post regional modeling procedure:» Screenline Analysis performed on assignment from

adjusted matrix» Utilized Traversal Matrix assignment feature of

EMME/2 to extract sub-area O-D trip tables» Allocated and proportioned O-D trip table from

EMME/2 model to correspond to detailed zone structure for microsimulation model

Microsimulation for I-405 & AWV - 6 Model Validation Analysis

Initial analysis indicated that study area networks needed to be streamlined:

Prepared link-level comparisons:» Total flow, Travel time, Speeds» Queue length

Comparative analysis of queue build-up against recurring bottleneck areas posted on the WSDOT real time traffic flow map

Accomplished reasonable base year validation results Model successfully utilized for future year applications

Model Validation Analysis- Continued -

Lessons Learned

Trips input into microsimulation from planning models need to be adjusted to counts

Adjustment of “seed” trip table from a planning model is better than synthesizing O-D trips for microsimulation

Useful features of EMME/2» Can use Demand Adjustment Macro» Can use matrix manipulation techniques to extract

sub-area matrices and to proportion trips to detailed zone structures

Demand Adjustment for Transit Trips Analysis Motivation

Sound Transit (ST) to build light rail (LRT) Requirement of agreement between Federal

Transit Adminstration and ST is : » To maintain model databases representing

conditions before and after the (LRT) project “Before” Conditions model circumstances:

» No current transit onboard survey data being available

» Link and route-level transit passenger counts by time of day being available

Transit Components Map

Sound Transit Model History

Sound Transit (ST) EMME/2 model:» Initially developed in 1992 to produce more

accurate route-level ridership forecasts» Incremental using comprehensive transit

onboard survey data» Relies on the regional 4-step model for

external factors such as change in demographics, congestion & travel costs

» Updated recently to reflect 1999 transit service levels and transit travel patterns

Matrix Adjustment Process-Initial Steps

Updated transit network reflecting service levels for fall of 2002 – over 400 routes

Performed transit network calibration analysis» Compared line times from EMME/2 model

against scheduled (actual) line times» Detailed verification of line headways,

operating hours and miles

Matrix Adjustment Process - Continued -

Relied on the methodology developed by Heinz Spiess (DEMADJT.MAC)» Minimization of difference between

estimated and observed link volumes on each route

» Travel patterns in seed matrix not changing more than necessary

Identified transit passenger counts locations

Matrix Adjustment Process - Continued -

Extracted and processed appropriate data from the Automated Passenger Count (APC) GIS database» Counts by time period» Counts on over 200 links representing over 300

directional transit route at heavy load segments» Used complementary vicinity count locations to

avoid under- or over-representation» Achieved convergence after 10 iterations

Validation Analysis Results

Validation Analysis Comparison» System-wide » Screenline» Route-level» Travel patterns – before & after matrix

adjustment Rigorous Peer Review and Verification

of Counts Used

Summary Conclusions

Matrix adjustment for transit is a viable and cost effective option when:» Seed matrix is not too old and initially developed

from rich survey data» Study area should have relatively good transit

coverage» Availability of a well-calibrated transit network» Availability of accurate and adequate counts data,

counts should be analyzed and understood» Availability of multi-path transit assignment

capability