Model Task Force Meeting November 29, 2007 Activity-based Modeling from an Academic Perspective Transportation Research Center (TRC) Dept. of Civil & Coastal.

Model Task Force Meeting November 29, 2007

Activity-based Modeling

from an Academic Perspective

Transportation Research Center (TRC)

Dept. of Civil & Coastal Engg.

University of Florida

Siva Srinivasan

Overview

1. A critique of the trip-based method

2. What is the activity-based approach?

3. Feasibility of adopting activity-based methods

4. Concluding thoughts



Home

Work

Lunch

Shop

Drop-off Kid at School

7:15 am

7:30 am

8:00 am

7:35 am

12:30 pm 12:35

pm

1:00 pm

1:05 pm5:00 pm

5:30 pm6:00 pm

6:30 pm

Car-pool

Drive-alone

Drive-alone

Drive-alone

walk

walk

Consider the travel pattern of a person………

Critique of the Trip-based Approach

Drive

Drive

Drive

Drive

Walk

Walk

H A1

H

P3

P4

P5

P6

A2

A3

A4

A5

A6

One Home-based Other trip

Four Non home-based trips

Peak

Peak

Peak

Peak

Off Peak

Off Peak


The trip-based method views this person’s travel as:


One Home-based Shop trip

Trip Generation

(frequency)

Trip Distribution

(destination)

Mode Split

(mode)

Network Assignment

(route)

Link flows, speeds, travel times, transit-ridership, etc.

• Socio-economic characteristics

• Land-use patterns

• Transportation system characteristics


The analytical modeling procedure is the “four-step” process


The fundamental role of travel-demand models is to enable us perform quantitative assessments of the impacts of policy actions

The policy actions being considered these days may invoke complex behavioral responses from travelers which cannot be adequately captured within a trip-based framework

Some Illustrative Examples…..



Example 1: Effect of Transit Improvements


Work

Shopping

Drive alone

Drive alone Drive alone

Very Good Transit Service

Transit

Drive AloneDrive Alone


Home

HomeWork Shop

Example 2: Effect of Transit Improvements (alternate responses)

Drive alone

Drive alone

Drive alone

HomeShop

Drive alone

Drive alone

Home

HomeWork

Transit Transit



5:00 pm

5:15 pm

5:40 pm

6:00 pm

4:00 pm

4:15 pm

4:40 pm

5:00 pm

4:00 pm

4:30 pm

5:40 pm

6:00 pm

Example 3: Employer-based Demand-Management Actions



Home

Home

Work Pick-up child

ShopDrop-off child

Person 1

Person 2

Example 4: Inter-personal Dependencies and Indirect Effects



Home Hom

e

Work

Person 1

Pick-up child

ShopDrop-off childPerson 2



Example 4: Inter-personal Dependencies and Indirect Effects

What is the Activity-based Approach?


Home

Work

Lunch

Shop


7:15 am

7:30 am

8:00 am

7:35 am

12:30 pm 12:35

pm

1:00 pm

1:05 pm5:00 pm

5:30 pm6:00 pm

6:30 pm

Car-pool

Drive-alone

Drive-alone

Drive-alone

walk

walk

The activity-based method views the travel pattern as………


The conceptual modeling procedure is………

(1) Activity Generation:

Choose what out-of-home activities have to be undertaken during the day incorporating relevant constraints

Drop-off Kid at School Work

LunchShop

Space constraints:

Location of home, school, and work are fixed

Time constraints:

School and work timings

Total time = 24 hours & that includes time for sleeping, eating etc.

Inter-personal constraints:

Household task allocations

Joint travel of household members



The conceptual modeling procedure is………

(2) Activity Scheduling:

Sequence the activities into a “chain” or “pattern”.

Travel is a consequence of this sequencing – the need to move from one location to another at certain times of the day to participate in the different activities

Home

Work

Lunch

Shop


7:15 am

7:30 am

8:00 am

7:35 am

12:30 pm

12:35 pm

1:00 pm1:05 pm5:00 pm

5:30 pm6:00 pm

6:30 pm

Car-pool

Drive-alone

Drive-alone

Drive-alone

walk

walk

Home

School

7:15 am

7:30 am

Car-pool

4:00 pm

4:15 pm

Car-pool


Contrasting Trip-based and Activity-based Methods:

Summary of Conceptual Differences


Treatment of travel as though demanded on their own right

Individual trip is the unit of analysis

Internal consistency of the travel pattern NOT guaranteed

Impacts of personal and household constraints not captured

Represents time as simply a cost of making a trip and time-of-day of travel is not captured adequately

Treatment of travel as a derived demand

Activity-travel pattern is the unit of analysis

Ensures internal consistency of the activity-travel pattern

Accommodates the impacts of various constraints on activity-travel decision making

Models travel within the context of overall daily time-use (both durations and time-of-day)

Trip-based Activity-based

Contrasting Trip-based and Activity-based Methods:

Summary of Modeling Differences


Trip-based Activity-based

Number of HB and NHB trips Generation and sequencing of activities

Zonal-level trip attractions & gravity model for trip-end locations

Location of activity participation

Mode for each trip Mode for linked trips (tours)

Time of day using peak and off-peak factors

Duration and timing of activities and travel

Feasibility of Adopting Activity-based Methods


(1) Activity-based models are “complicated”

Yes…

There are several (definitely more than four) components in the model-system

Could employ advanced econometric methods

However…

The conceptual modeling approach is more intuitive and easier to explain

The overall complexity depends on what dimensions of the activity pattern we are modeling

Statistical software to estimate the advanced econometric models available


(2) Activity-based models are data intensive

Yes…

Household travel surveys are needed for model estimation

Disaggregate forecasts of future population characteristics needed for model application

However….

Activity-based models can be developed using data collected from currently-used household-travel-survey methods

In fact, activity-based methods make better use of the data we are currently collecting

Methods (Iterative Proportional Fitting) for synthesizing disaggregate population characteristics are available



(3) Activity-based models are computationally demanding

Yes….

More model components / disaggregate approach / probabilistic models requiring micro-simulation to determine choice outcomes

Requires significant run times and very large data processing capabilities

However…

Computers are only becoming faster !

Parallel processing, multi-threading, efficient database management methods are available

Can use clever sampling techniques



(4) There are no standardized procedures or software

Yes…

Unlike the trip-based approach, there is no one single well-established procedure for modeling activity-patterns

There are no rigorous comparative assessments of the different modeling methods

There are no generic software platforms for implementing activity-based methods

However…

There are common underlying structures across the different activity-based models that can provide guidelines on developing one for any region

There is considerable flexibility in customizing the modeling approach for the activity-travel dimensions of interest for any region



(5) Other Issues

Activity-based models have predominantly focused on internal, person travel – external travel and freight movements are not yet within this framework

The activity-patterns have to be aggregated to OD trip tables by time-of-day if dynamic traffic assignment techniques are to be used -- these may not necessarily be straight forward

Still need to make the models “empirically” useful

Incorporation of appropriate traveler characteristics

Choice of resolution for space and time

Inclusion of appropriate land-use descriptors

Inclusion of appropriate transportation system descriptors


Concluding Thoughts


• Activity-based methods are theoretically and intuitively better descriptors of the travel-generation process

• The fundamental advantage of activity-based approach is in terms of more providing more realistic behavioral responses to land-use and transportation system changes

• The practical necessity of activity-based methods (and the structure of this model) for any region depends on what kinds of questions we want the travel-demand model to answer for this region

• It does appear that the “questions” are rather different for different regions as are data and resource availability

• Customization of the modeling methodology to different regions is more appropriate ??

Model Task Force Meeting November 29, 2007 Activity-based Modeling from an Academic Perspective Transportation Research Center (TRC) Dept. of Civil & Coastal.

Documents

tripbased approach slide

pm example

tripbased approach example

tripbased method

shop trip slide

tripbased framework

activitybased approach

pm carpool drive