An Integrated Urban Model of Transportation, Land-Use, Energy, and Environment

Kouros Mohammadian Department of Civil and Materials Engineering

University of Illinois at Chicago

Why an Integrated Urban Model?

More than 50% of the world’s population live in urban areas

By 2050, 70% of people are likely to be city dwellers – compared with less than 30% in 1950

Cities consume over 2/3 of the world’s energy and account for more than 70% of global CO2 emissions

This trend brings with it challenges and threats in urban areas

Need to consider a comprehensive, multi‐faceted, approach for an enhanced , efficient, resilience, and sustainable urban built infrastructure

To date, a comprehensive and holistic approach to study large-scale and inter-dependent urban infrastructures has not been developed thoroughly

An Integrated and Holistic Approach to Urban Infrastructure Development

Source: sklinternational.se

Transportation and urban form are fundamentally linked. How we build our city directly determines travel needs, viability of alternative travel modes, etc.

Transportation, in turn, influences land development and location choices of people & firms.

PhD Dissertation Proposal 5/10/2013

Overall Integrated Urban Model Framework

Population Synthesis

Vehicle Ownership Home/Work Location choice

Work/Home Change and Choice Vehicle Transaction

Household Composition

Transportation System

Regional Economy Land/Job Markets

Household Long-Term Context

Long-term Decision Making

Short-term Simulation

Activity/Travel Model

Network Simulation ADAPTS Model

Freight Model

Environment Energy Water Waste Health

FAME Model

6

Criticisms of Current Travel Demand Models

Unrealistic behavioral assumptions – Utility maximization

Artificially restrict activity scheduling to predefined choices – Can not represent full range of

“scheduling behavior” – No consideration of “dynamics” (full

day selected at one time)

Limitations on time-scale of analysis

Research Gaps in ABM

Simplification of scheduling process – Priority rules, fixed order of choices – Only out of home activities – Interactions between in-home and out-of-home activities

are ignored

No representation of planning dynamics – Impact of planning time on choices made

Short-term, diary data used – Most rule-based models implemented using 1-2 day

diary data

Limited integration with traffic simulation – Static assignment for fixed time-periods

Limited policy/scenario analysis capability

7/60

Scheduling Order Example A) Impulsive Shop - Preplan Eat Out Before Change

After Change

B) Preplan Shop - Impulsive Eat out Before Change

After Change

8

Motivation Hypothesis: When and how activity planning decisions are

made can impact final daily activity pattern

ADAPTS: – Next generation ABM paradigm accounting for “behavioral processes” – Retains link b/w activities and travel at the individual level – Adds element of activity planning, to activity generation and activity

scheduling, rather than relying on sequential series of models

Account for planning dynamics – when is each decision made in relation to other decisions,

activities, schedule, etc. – Planning and scheduling occurring in time-dependent manner,

also impacted by time-dependent traffic network – Within day and en-route re-planning

Represent macro-level changes from impacts of policies on planning dynamics at individual level

9

ADAPTS Simulation Framework

Household Planning

Individual Planning

Household Schedule

Household Memory

Social Network

Individual Schedules

Individual Memory

Land Use

Historical Network LOS

Institutional Constraints

Initialize Simulation •Initialize World •Synthesize Population •Generate routines

For each timestep

Trip Vector

Least Cost Routes

Information Flow Simulation Flow

Vehicles Movements

New Travel Times

Add to Historical Travel Times for Next Iteration

For each timestep

Prevailing Network LOS

Network Assignment

Household Socioeconomic

10

ADAPTS Planner/Scheduler ADAPTS planning and

scheduling framework

Handles at each timestep:

– Generation – Planning – Scheduling

Generation, planning and scheduling can occur at different times for same activity

Core of the framework is the Attribute Plan Order Model

Operationalized using multiple scheduling process surveys

At timestep t

Generate new activity

Update existing activity(s)

Execute activity

Attribute Planning Order model

Planned Activity Schedule

Time-of-Day

t = Ttime

Party

t = Twith

Mode Choice

t = Tmod

Destination choice

t = Tloc

Executed Schedule

Resolve Conflicts

Conflict Resolution Model

Set Plan Flags: (Ttime,Tloc, etc.)

Yes

Decision Logical test Model Simulated events

Yes

No

Yes

No

No

Activity G

eneration Activity

Planning Activity

Scheduling

11

ADAPTS Components

Activity Generation Activity generation with joint hazard-duration equations

– Significant socio-economic variables – Impact of hazard rate from other activities

Failure probability (generation) calculated each timestep – Based on time-since-last activity – Calculated using observed UTRACS data (14-day survey) – Fit to Chicago 1-day survey through updating

13

Activity Planning Order Framework

Assign plan horizon to each attribute

– After activity generated

Plan order model process – Assigns attribute flexibility – Get activity plan horizon – Attribute plan horizons

Plan horizons for each attribute based on:

– Attribute flexibilities – Activity plan horizon – General activity attributes – Socio-demographics, etc.

Defines the meta-attributes of the activity attributes

Generate New Activity (Auld and Mohammadian 2009)

Attribute Flexibility Model

Attribute Plan Horizon Model

Demographic Characteristics

Activity/Travel History

Flex Person

Flex location

Flex Start

Flex Duration

Plan Person

Plan location

Plan Start

Plan Duration

Plan Mode

Activity Plan Horizon Model

Plan Activity

Data source

Model

Model result

Flex Mode

14

Scheduling Heuristics

Set of rules for scheduling activities – Attempts to resolve conflicts by modifying each activity

Includes results of conflict resolution model: – TASHA – conflict resolution based on heuristic rules – New rules – heuristic rules determine how conflict resolution

strategy is implemented New rules allow for the consideration of more complicated

conflict types and deletion operations A new optimization model is developed

15

Traffic Assignment and Simulation

Zone

System

Econ

omic

Activ

ity

Logis

tics D

ecisi

ons

Zone System Socio-

Economic Factors

Economic Activity

Data

Freight Generation

Model

Commodity Production

Consumption

Supplier Selection Model

Network Assignment

Establishment Survey

Annual Commodity Flow

(firm-to-firm)

GPS-based data

Supplier Evaluation

Model

IO accounts/ Industry-commodity

crosswalk

Establishment Freight Survey

Transportation Performance

Measures

Interview Survey (specialists)

IO Accounts

Firm Synthesis

List of Firms with Their

Characteristi

Empty Trucks / Backhauling

Shipping Chain Configuration (direct/non-direct shipping chains)

Shipment Size / Frequency Choice

Main Mode Choice Model

Vehicle Choice

Number of Stops per

Stop Type

Access/Egress Mode Choice

Simulated Individual Shipments

Netw

ork

Analy

sis

CBP Data

Industry- Commodity Crosswalk

Freight Model (FAME) Framework

FAME’s zoning system Zone System Township level zones in the Chicago area (118 zone) County level zones in rest of Illinois (95 zone) FAF zones in the rest of US (120 zone)

Integration with Network Simulation

Advanced Demand models (Activity Based Models)

Advanced Supply Models (Dynamic Traffic Assignment)

– MATSim, DynusT, TRANSIMS, AIMSUN, ...

Interfacing, Interacting, Integrating!?! – Feedbacks – Tight Integration

Issues: – Behavioral Realism – Dynamic Activity Based Models – Dynamic pricing scenarios – Congestion related unreliability

Demand Model

Supply

List of Trips

Travel Times

19

Integration with TRANSIMS TRANSIMS Version 5.0

Router – Label constrained time dependent shortest path

Microsimulator: executes the trip plans – explicitly simulate:

Multiple travelers per vehicle Multiple trips per traveler Vehicles with different operating characteristics Intermodal travel plans

– does not simulate: Walking legs Interaction between cars and pedestrians Rail road crossings

Issues: – Behavioral Realism – Dynamic Activity Based Models

Network: Chicago Zones: 1,961 Signals: 9,822 Nodes: 19,038 Act Location: 98,204 Link: 31,339

Population: Households: 2,910,510 Person: 7,755,490

20

Integration Issues Variable Demand

Vehicle Rerouting

Performance – Running two different software, reduces the performance

Issues with Traffic Assignment Software Packages – Documentation – Programming Language Differences – Zoning Differences – Availability of high-fidelity network – Open Source, Cost

21

PhD Dissertation Proposal 5/10/2013

Integration with Traffic Simulation

Characteristics

Disaggregate & Mesocopic level

Agents could respond to dynamics in terms of activity planning, scheduling, and trip characteristics (mode, route)

Updating network characteristic second by second

Considering users’ response to network real time attributes in their route choice process

Able to track each user during the simulation and access to its real time characteristics

Demand

Supply

22

Embedded Traffic Simulator

Integrated model has been implemented directly in ADAPTS simulation

Consists of individual person-agents who: – Perform activity planning and scheduling (ADAPTS model)

– Select routes based on prevailing traffic & information provision (A* router)

Travel through the network (Newell’s model)

Can interact with ITS elements to modify or update plans

Two Modules: Router and Simulator – Router Calculates Shortest Path for P1 – Simulator Moves P1 along shortest path

Send details back to activity planner

1

8

5

9 11 10

7

4 3 2

6

5

4

2

6

2

4

6

5

7

4

5

6 3

23

Computational Issues Slow performance discourages practitioners

who need many scenario analysis and calibration

We may have the resources and patience, but others may not, so it is important to optimize the program and use minimum hardware

Advance models are very time-consuming and computationally intensive (DTA, ABMs, in-home activities, complex and dynamic policies, high level visualization, etc.)

24

Number of Threads: 64 Machine: RAM 256 GB, 32 AMD Opteron(TM)

Processors 2.3 GHz Full Microsimulation with dynamic traffic

assignment Tracking vehicles and individual locations every 30s Activity generation and scheduling simulation every

15 minutes 28 days of out-of-home activity preplanning and

real-time in-home activity generation Average activity statistics are recorded for the

sample. Simulated results for one week

Simulation

Extensions of the ADAPTS Model

Travel Behavior Health

• Air Pollution • Physical Activity • Traffic • Comfort Travel • Noise Pollution

• Urban Form • Pop./Emp. Density • Trans. System • Transit Accessibility • Trip characteristics

like trip purpose • Transit Service • Socioeconomic Status

Travel Behavior and Health

Every percent increase in Transit-use: • Decreases Heart Attack by 0.07% • Decreases Obesity by 0.29% • Improves General Health by 0.09% • Increases chance of Asthma infection by 0.03%

Every percent decrease in Auto-use: • Reduce chance of High Blood Pressure by 0.26% • Reduce chance of High Blood Cholesterol by 0.18%

Pedestrian friendly neighborhoods motivate people to walk more and be healthier.

Incorporating In-Home Activities

What People do at home could affect: o Transportation

oNo Trips on the network o Electricity

o US Residential utility per day: 30 KWh o Water

o Every individual per day: 100 Gallons of Water

oNatural Gas o Economy

Incorporating In-Home Activities

o In-Home Activities that are being considered: – Sleep – Personal Maintenance

• Eating and Drinking • Personal Care

– Household Maintenance • Household Activities • Caring for & Helping Household Members • Caring For & Helping Non-HH Members

– Leisure/Social • Sports, Exercise, and Recreation (at Home) • Socializing, Relaxing, and Leisure

• Socializing and Communicating • Relax and Leisure

• Watching TV • Etc..

– Discretionary • Religious and Spiritual Activities • Telephone Calls

– Mandatory • Work & Work-Related Activities • Education Related

In-Home Personal Maintenance Time of Day Frequencies

In-Home Activity Duration Frequencies

Time step T

N

Y

Shift Planning Flag Times

Generate new Out-of-Home Activity

Update existing activity(s)

Attribute Planning Order model

Set Plan Flags: (Ttime,Tloc, etc.)

Y

N

t = Ttime t = Tmod t = Tloc t = Twith

Y

Planned Activity Schedule

Execute Schedule

Resolve Conflicts

Conflict Resolution Model 1 (OUT/OUT)

Is Any Activity in Progress?

Generate In-Home Activity

Conflict Resolution Model 2 (IN/OUT)

Time-of-Day Model

Party Composition

Destination Choice

Mode Choice

Y

N

Decision Logical test Model Simulated events

N

Y

Asleep?

N Y

N

o A combination of rule-based and econometric models that simulate in-home activities alongside with out-of-home ones

Assumptions:

o Out-of-home activities that are

preplanned have priority to in-home activities.

o Impulsive out-of-home activities compete with in-home activities.

ADPATS ABM Framework

ADAPTS Updated Framework

Simulation Time

Simulation Time Step T

Current time step interval under focus

Generated In-home Personal Maintenance

Travel to home

Work

Conflict of an In-home activity with a routine out-of-home activity

Simulation Time Simulation Time Step T

Resolved In-home Personal Maintenance Work

Rule-based conflict resolution by truncation :

Travel to work

Priority of out-of-home preplanned/routine activities to in-home activities

Conflict of an In-home activity with an impulsive out-of-home activity

Simulation Time Simulation Time Step T

Work

Priority conflict resolution by the reverse pairwise model: Travel to work

Generated impulsive shopping activity

Competition of out-of-home impulsive activities with in-home activities

shopping activity

Priority given to shopping activity

Simulation Time

Simulation Time Step T

Generated In-home Personal Maintenance

Travel to home

Work

0

1

2

3

4

5

Slee

p

Pers

onal

Mai

nten

ance

at h

ome

Hosu

ehol

d M

aint

enan

ce a

t hom

e

Leisu

re/S

ocia

l/Rec

reat

ion

at h

ome

Disc

retio

nary

-oth

er a

t hom

e

Man

dato

ry a

t hom

e

Average Frequency of Activities per Day

ATUS IL

0

60

120

180

240

300

Slee

p

Pers

onal

Mai

nten

ance

at h

ome

Hosu

ehol

dM

aint

enan

ce a

t hom

e

Leisu

re/S

ocia

l/Rec

reat

ion

at h

ome

Disc

retio

nary

-oth

er a

tho

me

Man

dato

ry a

t hom

e

Average Duration for Activity Episode (min)

ATUS IL

Validation of In-Home Activity Simulation

What’s next?

• Electricity • Water • Energy • Waste • Food • CAVs • ….

Total Electricity Consumption by Census Block

Per square foot Electricity Consumption by Census Block

Conclusion ADAPS is a framework for model development

ADAPTS’ unique framework incorporates planning dynamics – Enroute route-switching and activity re-planning – reactions to unexpected events

Potentially more useful for certain applications – Operational simulations (e.g., CAV fleets, …) – Agents responding to unexpected events – Emergency planning

Further Challenges – Still computationally intensive – Incorporating other urban systems – Incorporating social networks – Incorporating joint decision making processes – Incorporating agent history in decision making – Accounting for uncertainty – Data availability

38

Questions!

39