Dynamic Origin-Destination Trip Table Estimation for Transportation Planning Ramachandran Balakrishna Caliper Corporation 11 th TRB National Transportation.

Dynamic Origin-Destination Trip Table Estimation for

Transportation Planning

Ramachandran BalakrishnaCaliper Corporation

11th TRB National Transportation Planning Applications Conference, Daytona Beach,

Florida9th May, 2007

Outline

• Introduction• Within-day traffic dynamics• Limitations of static methods• Short-term planning methods• Obtaining dynamic OD flows• Case studies

Introduction

• Long-term planning– Land use, residential location choice– Infrastructure development

• Short-term planning– Congestion and incident

management– Work zone scheduling– Special events preparation– Evacuation planning

Within-Day Traffic Dynamics

• I-405, Orange County, CA

0

2000

4000

6000

8000

10000

12000

14000

0 50 100 150 200 250 300 350

Time of Day

Flo

w (

veh

/ho

ur)

Hourly Flows

5-Min Flows

• Temporal dynamics– Complex interactions of network demand– Aggregation error

[Source: PeMS on-line database]

Limitations of Static Methods

• Temporal patterns “averaged out”

– Average trip rates over long periods– Daily, Peak (AM, PM), Off-Peak (MD, NT)

• Boundary conditions inconsistent

– Trips assumed to finish within single period– Departure time effects ignored

• Capacity, dynamic traffic evolution ignored

– Volume/capacity ratios can exceed unity– No queue formation and dissipation,

spillbacks

Short-Term Planning Methods

• Growing popularity of dynamic models

– Microscopic simulation– Dynamic Traffic Assignment (DTA)

• Key input: origin-destination (OD) matrices

– OD departure rates by time interval– Interval width: 5 min, 15 min, 1 hour

Obtaining Dynamic OD Flows

• OD surveys

– OD information collected directly– Costly, difficult to repeat / update

• Profiling of static matrices

– Not based on real measurements– Can be counter-intuitive (e.g. negative

flows)

• OD Estimation

– Match actual traffic data (e.g. detector counts)

– Data are up-to-date, easy to collect– OD information is indirect (requires

modeling)

Dynamic OD Estimation Steps

• Start with initial OD flow estimates−e.g. Derived from static matrix

• Assign them to the network−Dynamic network loading model

• Compare assigned output to data−Goodness of fit statistics

• Adjust OD flows, iterate to convergence−Optimization algorithms

Challenges

• OD departures appear in future intervals

• Data collection:– Loop detector counts are widespread– Richer data are becoming available

• Easy to match counts– Harder to match speeds, travel times,

queue lengths

• Most methods are tailored for counts– Recent methods include other data

Case Studies

• Irvine, CA• South Park, Los Angeles, CA• Lower Westchester County, NY

Irvine, CA1

1Balakrishna, R., H.N. Koutsopoulos and M. Ben-Akiva (2005) Calibration and Validation of Dynamic Traffic Assignment Systems. Mahmassani, H.S. (ed.) Proc. 16 th International Symposium on Transportation and Traffic Theory, pp. 407-426.

South Park, Los Angeles, CA1

1Balakrishna, R., M. Ben-Akiva and H.N. Koutsopoulos (2007) Off-line Calibration of Dynamic Traffic Assignment: Simultaneous Demand-Supply Estimation. Transportation Research Record (forthcoming).

Lower Westchester County, NY1

1Balakrishna, R., C. Antoniou, M. Ben-Akiva, H.N. Koutsopoulos and Y. Wen (2007) Calibration of Microscopic Traffic Simulation Models: Methods and Application. Transportation Research Record (forthcoming).

Conclusion

• Time-dependent OD flows– Critical for short-term planning,

simulation

• Dynamic OD estimation– Practical for real networks and data– Several approaches using counts– Recent advances allow general traffic

data

• Thrust areas– Collecting richer data for large networks