ActivityActivity--Based Based Travel Demand ModelsTravel Demand Models

Understanding Travel Behavior

Derivednatureoftraveldemand(Jones,1979) Travelgenerallyundertakentofulfillactivityneedsanddesires Activitiesdistributedintimeandspacenecessitatingtravel

Desiretraveldemandmodelsthatreflectthisfundamentalnotion

Roots of Activity Based Paradigm

Microsimulation ofindividualactivitytravelpatterns

Bringtogethertwoschoolsofthought Hgerstrands ConstraintsSchool(1969) ChapinsActivitypull(NeedsandDesires)School(1971and1974)

The Perfect Storm

Whyisvisionbeingrealizednow? Disaggregatemodelingoftravelbehavior Advancesinstatisticalandeconometricestimationmethods

Quantumleapsincomputationalpower Policyquestionswithcomplexbehavioralimplications Availabilityofdatabases(landuse,travelbehavior,GPS,networks)

An Evolutionary Stream of Research Chapin(1971and1974)studiedhumanactivitypatternsinurbancontexts

Hgerstrand (1969)examinedactivitypatternsinthetimespacedomain(continuum)andidentifiedseriesofconstraints

Activitybasedanalysistracesitsrootstotheseschoolsofthought

Hgerstrands Constraints

Threeprimarytypesofconstraintsidentified Authoritative:Timespaceconstraints Capability:Biologicalneedsandresources(sleep,income)

Coupling:Interagentinteractions(children)

Activity-Based Paradigm in Transportation

Jones(1979)explicitlyidentifiedrelationshipsamongactivities,travel,time,andspace Traveldemandisaderiveddemand

Followedwithaconferencein1981onTravelDemandAnalysis:ActivityBasedandOtherNewApproaches

Impetusgrewinlate1980sandinto1990s KeylegislativeactssuchasISTEA(1991),CAAA(1990),andTEA21(1998)

MajorlawsuitfiledbySierraClubagainstSanFranciscoBayAreaMTC

FederalTravelModelImprovementProgram(TMIP)

Activity-Based Paradigm in Transportation

EarlyresearcheffortsunderTMIP DevelopmentofTRANSIMSatLosAlamosNationalLaboratory Activitybasedmodeldevelopmentresearch(AMOS)

FourkeypaperspublishedinTransportation(1996)offeredframeworksforactivitymodels

Activity-Based Paradigm in Transportation

FourpapersinTransportation LateKitamura/Pas BenAkiva/Bowman Stopher/Hartgen Slavin

Otherkeypublications Kitamura(1988);Axhausen andGrling (1992);Jonesetal(1990);Bhat andKoppelman (1999)

Activity-Based Paradigm in Transportation

Otherkeyevents 1996ActivityBasedModelingConference(NewOrleans)organizedbyTMIP everybodyfromMissouri?

1995and2004NetherlandsConferencesinProgressinActivityBasedAnalysis

TRBTaskForceonMovingActivityBasedApproachestoPractice(20032008) chairedbyProf.KostasGoulias

Activity-Based Paradigm in Transportation

A Daily Activity Itinerary

Home

WorkRestaurant

Shop

Kids School7:15 am

7:30 am

8:00 am

7:35 am

12:30 pm12:35 pm

1:00 pm

1:05 pm5:00 pm

5:30 pm6:00 pm

6:30 pm

drivedrive

drive drive

walk

walk

Interactions and Constraints Howdoesafoursteptripbasedmodelviewthisitinerary?

Drive

Drive

Drive

Drive

Walk

Walk

O1 D1

O2

O3

O4

O5

O6

D2

D3

D4

D5

D6

Two Home-based tripsFour Non home-based trips

Peak

Peak

Peak

Peak

Off Peak

Off Peak

Influence of Trip Chaining Consideratransitenhancement

Work

Shopping

Drive alone

Drive alone Drive alone

Enhanced Transit Service

Home

Influence of Trip Chaining Howcanaswitchtotransitbeaccommodated?

Shopping

TransitDrive alone

Drive alone

Home Work

Transit

Activity-Travel Inter-relationships

Noteseriesofinterrelatedbehavioralresponses Shiftinmodechoicetowork Shiftindestinationchoiceforshopping Shiftintimingofshoppingactivity Impactontripgeneration(4tripsinsteadof3trips)

Limits of Four-Step Trip-Based Models

Considerachangeinsystemconditions Increaseincapacity Increaseincongestion(traveltime) Changeinfuelprice

Whatdopeopledo? AccordingtoUSAToday/GallopPollinJune2008,>80%consolidateerrands Directimpactontripgeneration

However,activityparticipationremainedlargelyunaltered

Tripgenerationmodelsinpractice Largelyinsensitivetosystemconditions

Limits of Four-Step Trip-Based Models

Insensitivityoftripgenerationtosystemconditions Inabilitytomodelsuppresseddemandorinduceddemand

Couldincorporateaccessibilitymeasuresintripgenerationmodels,but Stillverylimitedinabilitytoreflectinteractionsandconstraints

Limits of Four-Step Trip-Based Models

Social Equity and Quality of Life Issues

Qualityoflifetightlyconnectedtohumanactivitypatternsandhowpeoplespendtime

Activitybasedparadigmoffersabilitytoconstructutilitymeasuresthatdirectlyaddresstheseissues

Planning Issues Traditional/novelmultimodalcapacityadditions/subtractions Transit/PedestrianOrientedDevelopment Bicyclefacilityenhancements HOV/HOTlanes,CongestionPricing,VariablePricing,Parking

Pricing

Telecommuting(Telecommunications),FlexibleWorkSchedules ITSdeployments Equity,SocialExclusion,EnvironmentalJustice Energy(FuelPrices)andEnvironment(AirQuality) Homelandsecurityanddisastermanagement

Basis for Model Design Policy issuesandquestionsofinterest Realisticbehavioral paradigm/representation Computationallyfeasibleandtractable

Estimation Implementation

Data availability(presentandfuture)

A Focus on Behavioral Considerations

Multitudeofchoicesdefineactivitytravelbehavior Activitytype/purpose Activitytiming(timeofday) Travelmodeanddestination Activityduration Activitylinkage(tripchaining) Accompanyingpersons Networklevelchoices

Behavioral Decision Processes Multitudeofdecisionhierarchiespossible

Whatisthesequenceinwhichchoicesaremade? Virtuallyallmodelsystemsimplyacertaindecisionhierarchy

Towhatextentarechoicesmadesequentiallyversussimultaneously/jointly?

Decision Hierarchies Largevarietyofdecisionhierarchiespossible

Heterogeneityinthepopulation Carefulmarketsegmentationbasedondecisionprocesses

Growingevidenceofsimultaneityinchoicedecisions Peoplechooseanactivitytravel(lifestyle)package

Ifchoiceprocessissequential,moreconstrainedchoiceprecedeslessconstrainedchoice Inhouseholdwithvehicleownershipconstraints,wouldmodechoice

precededestinationchoice?

The Role of Time

Thenotionoftimeandtimeuseiscentraltotheactivitybasedmodelingparadigm Timeisnotjustacost tobeminimized Rather,itisafiniteresourcewhoseuse peoplestrivetooptimize

Timeisanallencompassingentityinactivitybasedmodels

Timeappearsinactivitytravelagendasinnumerousways Dailytimeallocationtoactivitiesandtravel Thedurationofsingleactivityandtravelepisodes Thetiming(timeofday)ofactivities/trips Multiday(weekly)activityscheduling

The Role of Time

Considerationofrelationshipsbetweeninhomeandoutofhomeactivitytimeuse

Evidenceofincreasedavailabilityofleisuretime Evidenceofincreaseintraveltimeexpenditures

Productivityefficienciesbroughtaboutbyspecializedservicesandtechnologydeployment

The Role of Time

Dopeopletreattimeasacontinuousentityoradiscreteentity? Discretetimeofdaychoicemodels(breakthedayintodiscreteperiods)

Continuousdurationmodelswhereactivitytimingismodeledalongthecontinuoustimeaxis

Schedulingmaybediscretewhiletimeallocationmaybecontinuous

The Role of Time

Agent Interactions Taskallocationandjointactivitytravelengagement

Withwhomandforwhom? Activitydependency(children) Householdvehicleallocation Residentialandworkplacelocationchoices Realtimeactivityscheduling

Influenceofmobiletechnologies Generateactivitiesontheflyinmodel?

Time-Space Interactions

Gainrealismbyincorporatingtimespaceprismconstraints

Constraintsonmodaltransition,publictransitavailability,anddestinationchoices

Generatework/schoolschedulesandtoursfirst(defineanchorpoints) Discretionaryactivitiessimulatedalongthetimeaxisrecognizingconstraintsimposedbyworkandschool

PrismConstrainedActivityTravelSimulator(PCATS)ofKitamura nowembeddedinOpenAMOS

Dividesadayintoopenperiodsandblockedperiods DefinesaHgerstrandsprismforeachopenperiodandsimulatesactivitiesandtravelwithinit

Time-Space Interactions

Home Work

Activity 1 (Fixed)

Activity 2 (Fixed)

T

i

m

e

Urban Space

1

vHome Activity

A

Activity atLocation A

Activity 1

Activity 2

Time-Space Interactions

Sequentialstructure Theattributesofanactivityandtriptoitaresimulatedactivitybyactivity,conditionalonpastactivityengagement

Operationalhierarchy:activitytypemodedestinationactivityduration

Representationofprismconstraints Activitytypechoice/generation remainingtimeinprism Modedestinationchoice constrainedchoiceset Durationchoice remainingtimeinprism

Time-Space Interactions

Decision Time Points for Discretionary Activities

Decision Point 1

Time

Open Period

Decision Point 2 Decision Point 3

Travel

Travel

Travel

Fixed ActivityActivity 1 Activity 2

Blocked Period

PCATS Model Components

Prismvertexmodels Stochasticfrontiermodelstodetermineunobservedprismvertices

Activitytypechoicemodels Multinomiallogitmodelsthatdetermineactivityengagementineachopenperiod

DestinationModechoicemodels Nestedlogitmodelsthatassignadestinationmodepairtoeachactivitywithinaprism

Activitydurationmodels Splitpopulationsurvivalmodelsthatdeterminelengthofeachactivitywhileconsideringtheprismsize

PCATS Model Components

Representing Prism Constraints

Home Work Urban Space

In-home Activity

T

i

m

e

1v

PM Work Activity

In-home Activity

AM Work Activity

Prism vertices generated by stochastic frontier models

A prism configured assuming the fastest travel mode in the choice set

Travel mode availability by time of day and mode continuity checked within and across prisms

Activity Generation Process

Time Left for Activities?

Activity Type Choice

Destination-Mode Choice

Activity Duration Choice

More Activities?

Mode Choice (to Fixed Activity)

Adjust Activity Duration

Mode Choice toNext Fixed Activity

YesNo

YesNo

Open Period Begins

Open Period Ends

Prism ConstraintsPrism Constraints

Prism ConstraintsPrism Constraints

Prism ConstraintsPrism Constraints

Prism ConstraintsPrism Constraints

Prism ConstraintsPrism Constraints

Integrate with Dynamic Traffic Simulator

Maximizeuseofinformationfromactivitybasedmodelsystem Activitiesandtripsgeneratedalongthecontinuoustimeaxis Loadtripsonthenetworkastheyaregenerated(atoneminuteresolution)

Dynamicinterfaceandconcurrentexecution,alongthetimeaxis,oftheactivitysimulatorandanetworksimulator

Nopostprocessingofmodeloutputs

Integration with Traffic SimulatorDecision Processor

Traffic Simulator

Event Manager

Time Axis

Decision to engage in some activity

Decision to engage in some activity

Determine destination and

mode

Determine destination and

mode

Travel

Scanning Interval (1)

Given arrival time, determine

activity duration

Given arrival time, determine

activity duration

Activity duration

Agent on Process Waiting

List

Agent on Traveler List

Agent on Traveler List

Agent on Actor List

Decision to engage in some activity

Decision to engage in some activity

Enhancing Behavioral Realism

Exacttripdurationsnotknownuntiltripsarecompleted

Needtoconsiderissuesofunmetmobility Anagentmaybelateforwork,cannotfinisherrand,cannotreturnhome,etc.

Prismconstraintsmaynotalwaysbesatisfied Prismconstraintsincreasinglyfuzzy?(technologyeffects)

Activity-Based Model Systems

Numerousactivitybasedmodelsystemsdevelopedinresearcharena

Modelshavematuredtovaryingdegrees Attempttoincorporateaspectsofbehaviorhighlightedinpresentation

Activity-Mobility Simulator (AMOS, FAMOS)

HouseholdAttributesGenerationSystemHouseholdAttributesGenerationSystem

SyntheticPopulation(HouseholdsandPersons)

SyntheticPopulation(HouseholdsandPersons)

HouseholdTravelSurvey

Data

HouseholdTravelSurvey

Data

PrismConstrainedActivityTravelSimulator

PrismConstrainedActivityTravelSimulator

NetworkLevelofServiceDataNetworkLevelofServiceData

ActivityTravelRecordsforEachPerson

ActivityTravelRecordsforEachPerson

OutputProcessorOutput

Processor

OutputReportsOutputReports

GISVisualizationGISVisualization

ODFlowsbyPurposeandTimeofDay

ODFlowsbyPurposeandTimeofDay

CensusSocioEconomic

Data

CensusSocioEconomic

Data

DynamicEventBasedNetwork

Simulator

DynamicEventBasedNetwork

Simulator

Model Design: SimTRAVEL

Model Design

Model Design

Model Design

SimTRAVEL: Simulator of Transport, Routes, Activities, Vehicles, Emissions, and Land

Funded by FHWA through Exploratory Advanced Research ProgramSee: http://simtravel.wikispaces.asu.edu

Integrated Model: Supply and Demand

OpenAMOS

DynusT/MALTA

t =1 mint =0

Origin,Destination,

VehicleInfoforVehicleTrip1

ArrivalTime

VehicleisloadedandthetripisSimulated

Person(s)reachdestinationandpursueactivity

Origin,Destination,

VehicleInfoforVehicleTrip2

24 hr duration

Update Set of Time-Dependent Shortest Paths 1440 paths per O-D Pair

ODTravelTimesforDestinationandModeChoiceModeling

6sec.interval

CEMDAP (Bhat)

Input Data Coordinator

SimulationCoordinator

Model Modules

Household

Person

Pattern

Tour

Stop

Internal Data Entities

Decision to work

Work duration

Work start time

HH activity generation

Activity stop location

LOS & Zonal data queries

Comprehensive Econometric Microsimulator of Daily Activity-travel Patterns

Comprehensive Econometric Microsimulator of Daily Activity-travel Patterns

CEMDAPApplication of the Generation-Allocation Model System

Work and school activity participation and timing decisions

Childrens travel needs and allocation of escort responsibilities to parents

Independent activity participation decisions

Application of the Scheduling Model SystemWork-to-home and home-to-work commute characteristics

Drop-off tour of the nonworker escorting children to school

Pick-up tour of the nonworker escorting children from school

School-to-home and home-to-school commutes

Joint tour of the adult pursuing discretionary activity with children

Independent home-based and work-based tours for each worker

Independent home-based tours for each non-worker

Independent discretionary activity tour for each child

C

E

M

D

A

P

:

M

o

d

e

l

C

o

m

p

o

n

e

n

t

s

C

E

M

D

A

P

:

M

o

d

e

l

C

o

m

p

o

n

e

n

t

s

CEMDAP: Parent-Child Interaction

Children'sPattern

Parent(s)pattern

Go to School?

School Timing

Go to Work?

Work Timing

Mode to/from School

Drive by parent

Allocate to Parent(s)

Adjust Work Timingof Parent(s)

Integrated Activity-Travel Demand and Dynamic Traffic Assignment Model System (CEMDAP-VISTA: Bhat/Waller)

Activitytravelsimulator(CEMDAP)

Individualactivitytravelpatterns

Aggregatesociodemographics

Interface:ConvertPersonTourstoODTripTablesbyTimeofDay

Syntheticpopulation

generator(SPG)andInputgeneration(CEMSELTS)

Activitytravelenvironment(LOS)

Interface:LinkVolumesand

Speeds

Update LOS

ODTripTablesbyTimeofDay

LinkVolumesandSpeeds

Traffic SimulationFind 6 sec. traffic flowsAccounts for ramp metering, information provision, traffic incidents, etcOutput: Travel times per interval and road segment

Optimal routingFinds optimal rout for all OD pairs and departure timesSolve time-dependent shortest pathAccounts for non-travel time costs (tolls, stochasticity)Output: Optimal route per OD pair and departure time

Path AssignmentAssign paths to each individual vehicle on the networkOutput: vehicle path

VISTA

ConvergenceCheck

After Convergence

SimAGENT for SCAG

Networks & Attributes

Accessibility (aggregate)

Land Use Design and Forecasting (including demographics)

Population Synthesis: PopGen

Population in Zones

(centroids)

Daily Scheduling

Parcels/Zones & Attributes

Long Term Choices

Daily AllocationAirports & Ports

External Trips

Passenger & Highway and TransitAll Other

(commercial)

REPORT

Origin Destination Trip Interchange Matrices

Network Assignment

Post Processor

Person Daily

Tours-stops & trips

EMFAC

ADAPTED CEMDAP MODEL

SimAGENTSimulator of Activities, Greenhouse Emissions, Networks, and Travel

Addressprovisions/mandatesofSB375 Requiresmetropolitanplanningorganizations(MPOs)toinclude

sustainablecommunitiesstrategies(SCS)forthepurposeofreducinggreenhousegasemissions

Addresswiderangeofpolicies,e.g.: Economicanalysis:locationbasedwelfare,wages,andexports Equityanalysis:changeinwelfarebyhouseholdincomeclass EvaluatetheenergyuseandGHGsproducedbyhouseholdsandworkers

inbuildingspace

Comprehensivelyevaluateeconomicdevelopmentimpacts Evaluatetimeofdayroadwaytolls

SimAGENT Phased Implementation Plan for SCAG

Phase Title Description

1 Model Development Plan and Strategy

Work closely with SCAG staff to finalize model development plan, model structure, model implementation path, and software and data requirements and specifications

2Development and Implementation of SimAGENT Version 1

Adapt CEMDAP to SCAG regionAdd Synthetic Population GeneratorCompare to 2003 Trip-Based ModelExtensive Validation and Sensitivity TestingConduct Hands-on Staff Training SessionsEstimate GHG using EMFAC

3Development and Implementation of SimAGENT Version 2

Enhanced CEMDAP Model SpecificationsMore Detailed Spatial/Network ResolutionFull Incorporation of Time-Space ConceptsExtensive Validation and Sensitivity Testing

4 Training and Reports Submission of Final Deliverables Conduct Hands-on Staff Training Sessions

3:00 AM-4:00AM

4:00 AM-5:00AM

5:00 AM-6:00AM

6:00 AM-7:00AM

7:00 AM-8:00AM

8:00 AM-9:00AM

9:00 AM-10:00AM

10:00 AM-11:00AM

11:00 AM- Noon

Noon-1:00PM

1:00 PM-2:00PM

2:00 PM-3:00PM

3:00PM-4:00PM

4:00PM-5:00PM

5:00PM-6:00PM

6:00PM-7:00PM

7:00PM-8:00PM

8:00PM-9:00PM

9:00PM-10:00PM

10:00PM-11:00PM

11:00PM-Midnight

Midnight-1:00 AM

1:00 AM-2:00AM

OPUS FRAMEWORK

URBANSIM

SpecialGenerators(eg, airport)

Trip Aggregator

Network traffic assignmentOD Matrices Network performance(skims)

External trips

HH/Personday-tour-trip list

Commercialmovements

AB HOUSEHOLD TRAVELDEMAND SIMULATOR

TRANSPORT MODEL SYSTEM

Person Day Simulator

Mobility Choice Simulator

Parcel Attributes(Land Development)

SyntheticPopulation Accessibility

TRANSPORT PLANNING

TransportNetworks

Parcel Attributes(Transport

Development)

REPORTINGAND QUERYSUBSYSTEM

P

S

R

C

M

o

d

e

l

D

e

s

i

g

n

(

B

o

w

m

a

n

,

B

r

a

d

l

e

y

,

C

a

s

t

i

g

l

i

o

n

e

)

Activity-Based Model System - DASH(Gliebe)MetroScopeLandUseDataMetroScopeLandUseData

PopulationSynthesisPopulationSynthesis

WorkplaceLocationChoiceWorkplaceLocationChoice School/College

LocationChoiceSchool/CollegeLocationChoice

AutoOwnershipChoice

AutoOwnershipChoice

LongTermChoices

DayActivityPattern/RoleChoice

DayActivityPattern/RoleChoice

StartingTimeChoiceStartingTimeChoice

InitialConditions(Day)

TripListsTripListsAssignmentAssignment

TripTablesTripTables

PostProcessor

NextStopPurposeNextStopPurpose

NextStopLocationChoice

NextStopLocationChoice

DynamicActivityPatternGeneration

TourModeChoiceTourModeChoice

NextStopModeChoice

NextStopModeChoice

TourFirstStopPurposeChoice

TourFirstStopPurposeChoice

ADAPTS (Mohammadian)

Agent-based Dynamic Activity Planning and Travel Scheduling

Household Planning

Individual Planning

Household Schedule

Household Memory

Social Network

Individual Schedules

Individual Memory

Land Use

Network LOS

Institutional Constraints

Initialize SimulationInitialize WorldSynthesize PopulationGenerate routineactivities

For each timestep

Write Trip Vector

Traffic Assignment

Information Flow

Simulation Flow

ADAPTS (Mohammadian)

At timestep t

Plan new activities

Update existing activity(s)

Execute activity

Attribute Planning Order model

Planned Activity

Schedule

Time-of-Day model

t = Ttime

Party composition

model

t = Twith

Mode Choice model

t = Tmode

Destination choice model

t = Tloc

Executed Schedule

Resolve Conflicts Conflict Resolution Model

Set Plan Flags:(Ttime,Twith, Tloc,

Tmode)

Yes

Decision

Logical test

Model

Simulated events

Yes

No

Yes

No

No

Activity Generation

Activity Planning

Activity Scheduling

ALBATROSS (Timmermans/Arentze)

Alearningbasedactivitybasedtravelmodelsystem Employstheoriesofchoiceheuristicstorepresentbehavioralprocesses

Decisiontreeapproachesusedtoformalizeheuristicsandpredictchoicebehavior

Explicitconsiderationofmultitudeofconstraints

Albatrossassumesthatchoicebehaviorisbasedonrules thatareformedandcontinuouslyadapted throughlearning

Individualinteractswiththeenvironment(reinforcementlearning)orcommunicateswithothers(sociallearning)

Albatrossisbasedonalearningtheorywhichimpliesthatrulesgoverningchoicebehaviorare:

heuristic contextdependent adaptiveinnature

ALBATROSS (Timmermans/Arentze)

Components: amodelofthesequentialdecisionmakingprocess modelstocomputedynamicconstraintsonchoice

options

Asetofdecisiontreesrepresentingchoicebehaviorofindividualsrelatedtoeachstepintheprocessmodel

[a-priori definedderived from observed choice behavior

ALBATROSS (Timmermans/Arentze)

ALBATROSS (Timmermans/Arentze)

ALBATROSS (Timmermans/Arentze)

RuleRule--Based HeuristicsBased Heuristics

ALBATROSS (Timmermans/Arentze)

Classification of ActivitiesClassification of Activities

ALBATROSS (Timmermans/Arentze)

TRANSIMS (FHWA/LANL)

TransportationAnalysisandSimulationSystem Generatesandsimulatesactivity/travelpatternsforindividualsinaregionovera24hourperiod

Supportshighlydetailedroadandtransitnetworks Timedependentlinkdelaysareconsideredforroutingtripsthroughthenetwork

TRANSIMS: Framework

PopulationSynthesis

ActivityPatterns

ModePreference

Router Microsimulator

Stabilization

RefineModes

ChangeActivityTimesorPatterns

RouteAttributesAttractionBalancing

TRANSIMS: Activity Generator Generatesactivityengagementpatternsforeachmemberofahouseholdovera24hourperiod

Outofhomeactivitylocationsdeterminedusingadestinationchoicemodel

Activityengagementpatternsgeneratedbysamplingfromactivitypatternsofindividualsinatravelsurvey Inthecurrentimplementation,ClassificationandRegressionTreesare

used

TRANSIMS: Activity Generator (continued)

TRANSIMS: ApplicationHighwayNetwork

Conversion

TransitNetworkConversion

NetworkEditing

TripTableConversion

CensusDataConversion

PopulationSynthesisActivityGeneration

RouterandRouterFeedback

Microsimulator

ActivityGeneration

PopulationSynthesis

CensusDataConversionTripTableConversion

Tripbasedmodel Hybridmodel

Tourbasedmodel

NetworkPreparation

TRANSIMS: Feedback Processes

Router

PlanPrep(Merge)

PlanSum

PlanSelect

Done?

RouterStabilization

MicrosimulatorStabilization

UserEquilibrium

Router

PlanPrep(Merge/Sort)

PlanSelect/ProblemSelect

Microsimulator

PlanCompare

Done?

PlanPrep(Merge/Sort)

Microsimulator

Router

Done?

No

Yes Yes

Yes

NoNoStop

MATSIM-T (Axhausen/Nagel)

MultiAgentTransportSimulationToolkit Iterativeagentbasedtrafficsimulationframework Onlyautosaresimulated Involvestwomaincomponents

Agentgeneration(groupedashouseholds) ActivityScheduling

MATSim-T: Scheduling Tasks

TaskFrequencyper

runModeltype

Number,sequenceandtypeofactivities

Once Conditionalprobability

Startanddurationofactivities PeriterationBestresponsemodel(GAbasedoptimizer)

Compositionofthegroupundertakingtheactivity

Expenditureanditsallocationamongtheparticipants

Secondarylocationchoice OnceImputed(Proportionaltosizeanddistance)

Mode/vehiclechoice PeriterationImputed

(ChainbasedMNL)

Traveldemandqisgeneratedandmicrosimulated Resultinggeneralizedcostskareusedtoadjustschedules,

capacitiesandpricesoffacilities

Route(r)adaptationprocessalsoextendstowardstimechoice(t),modechoice,locationchoice(j),etc

MATSim-T: Framework

Competitionforslotsonnetworksandinfacilities

MentalMap

ActivityScheduling

Population

Parameters

Scenario k(t,r,j)

q(t,r,j)

ILUTE (Miller) IntegratedLandUse,Transportation,Environment

(ILUTE)modelsystem

ILUTEmainlytriestomodelthespatialmarketsandthepersons dailydecisionmakingwithinahouseholdbasedcontext

Simulatestheevolutionofagentsandobjectsovertime Agentsandobjectsincludeindividuals,transportationnetworks,

thebuiltenvironment,theeconomy,andthejobmarket

ILUTE: Framework

Demographics LandUse

LocationChoice

Activity/Travel&Goods

Movement

DynamicTrafficAssignment

VehicleOwnership

RegionalEconomics

GovernmentPolicies

ExternalImpactsFlows,Times,etc.

TransportSystem

ILUTE: Structure and Current ImplementationObservedBaseYear

AggregateDistributionsofAgentsandAttributes

AGENTSYNTHESIS

SyntheticAgentPopulationT=0

T=T+T

DemographicUpdate

LabourMarket

HousingMarket

AutoOwnership

ActivityBasedDailyTravel(TASHA)

RoadandTransitNetworkAssignments

TransportationEmissions&DispersedPollutionConcentrations GHGEmissions

Link&ODTravelTimes/CostsLink,

CongestionLevels,Etc.

CommercialVehicleMovements

Employment@timeT

Road&TransitNetworks@TimeT

PopulationExposuretoPollutantsbyLocation

andTimeofDay

ExogenousInputs@timeT:

InterestRatesEnergyRatesVehicletechnologyZoningIn/outmigrationrates

ILUTE: Evolutionary Engine

ExogenousInputs,TimeTInmigrationPolicyChanges

EMME/2TransportationNetworkModel(Computetraveltimes/costsbymode)

ILUTEEvolutionaryEngine

ForT=T0+1,T0+NTdo:DemographicUpdate

DemographicsFamily/householdcompositionupdateSchoolparticipationupdate

BuildingStockUpdateResidentialHousingCommercialFloorspace

Firm/JobLocationUpdateWorkParticipation&LocationUpdateResidentialLocationUpdateAutoOwnershipUpdateCommercialVehicleMovementUpdateActivity/TravelUpdate(TASHA)

SynthesizeBaseYearPopulation,Employment,

Dwellings,etc.

BaseYearCensusData,OtherAggregate

Data T0=BasetimepointT=CurrenttimepointbeingsimulatedNT=Numberofsimulationtimesteps

TASHA (Miller/Roorda) Travel/ActivitySchedulerforHouseholdAgents Simulatesoutofhomeactivityandtravelpatternsforindividualsrecognizinghouseholdlevelinteractionsandconstraints

TASHAusestheconceptofproject introducedbyAxhausen(1998)

TASHAcomprisesof: Anactivityepisodegenerator Anactivityscheduler Arandomutilitytourbasedmodechoicemodel

TASHA: Class Structure and Project Definitions

TASHA: Activity Generation, Scheduling and Mode Choice

Time Use Utility Measures

Timeuseallocationiscentraltotheactivitybasedmodelingparadigm

Offersstrongframeworkforanalyzingmeasuresofwelfarethatpeoplederivefromtheiractivitytravelpatterns

Addresssocialequityandqualityoflifeissues

Formulation of Time Use Utility Measure

Utilityformulation)1ln(])1ln([ qqqqq TSxU

Uq isutilityderivedfromactivityoftypeq

Tq iscumulativedailytimeexpenditureonactivityoftypeq

Sq iscumulativedailytimeexpenditureontravelforactivityoftypeq

xq isavectorofcovariatesaffectingutilityUq isascalarcoefficientassociatedwithln(Sq+1) isavectorofcoefficientsassociatedwithxqq isani.i.d.randomerrorterminUq.

Utilityformulation)1ln( ss TU

sq

q UUU

Us istheutilityderivedfromsleepU isthetotalutilityderivedfromthetimeusepatternTs iscumulativedailytimeexpenditureonsleepTf isthetotaltimeavailableinaday.

Maximize

Subjectto fsq

qq

q TTST

Formulation of Time Use Utility Measure

Baseline Activity Pattern

Activity Type Daily Duration (min)

Sleep 472In-home maintenance 202Out-of-home maintenance 53Travel for out-of-home maintenance 37In-home discretionary 166Out-of-home discretionary 76Travel for out-of-home discretionary 16Commute time (round trip) 60

Modified Activity Pattern: After Telecommuting

Activity Type Daily Duration (min)

Sleep 492 (+20)In-home maintenance 202Out-of-home maintenance 53Travel for out-of-home maintenance 37In-home discretionary 186 (+20)Out-of-home discretionary 90 (+14)Travel for out-of-home discretionary 22 (+6)Commute time (round trip) 0

Example Timeuseutilitymeasureformulatedasafunctionof: Socioeconomicanddemographiccharacteristics Traveldurationstoandfromactivities Activitydurationsfordifferentactivitytypes/episodes

TimeUseUtilitybeforecapacityenhancement=25.570 TimeUseUtilityaftercapacityenhancement=27.531 Couldtranslateintomonetarybenefits Alsoexamineequityacrossmarketsegments

Activity Type Utility Value Before

Telecommuting

Utility Value After

Telecommuting

Sleep 6.159 6.201In-home maintenance 2.939 2.939Out-of-home maintenance 0.777 0.777In-home discretionary 2.882 2.946Out-of-home discretionary 12.813 14.669Total 25.570 27.531

Example

Key Considerations

Representationoffuzzytimespaceprismconstraints,interagentinteractions,andtimeusebehavior

Greaterlevelofsimultaneityinchoiceprocessestoreflectchoiceoflifestylepackage

Recognitionofheterogeneityinpopulation behavioralstructure,decisionhierarchy,parameters/coefficients Carefulmarketsegmentation,trippurposedefinition,representationof

time,space,andnetworks

Centralroleoftimeandspace Disaggregaterepresentationoftimespacedomain Continuousrepresentationoftime Disaggregatespatialrepresentation

Maximizeuseofinformationfromactivitybasedtravelmodel

Key Considerations

Things to Think About Feedbackprocesses

Feedbackwithinactivitytravelsimulatorfromdestination/modechoicetotimeofdaychoicetoactivitytype/generation

Feedbackfromnetworkassignmenttoactivitytype/generation(tourstops),andmodeanddestinationchoice

Criteriaforconvergenceandequilibriumconditions

Stochasticsimulation Onerunrepresentsonerealizationofstochasticprocess Howmanyrunsarerequiredtoachievestableresults? Impactsoncomputationtimeandhardware/softwarerequirements

Datarequirements Travelsurveydata Multimodalnetworkdatabytimeofday Detailedlandusedata Greaterlevelofdisaggregationforactivitymicrosimulation

Things to Think About

Inhouseresources Stafftrainingandexpertise Computationalresources Phaseddevelopmentplan

Comprehensivemodeldesignupfrontwithstageddevelopmentandimplementationschedule

Things to Think About