Modeling in the “Real Modeling in the “Real
World”World” John BrittingJohn Britting
Wasatch Front Regional Wasatch Front Regional CouncilCouncil
April 19, 2005April 19, 2005
IntroductionIntroduction• Forecasting manager for Salt Lake City Forecasting manager for Salt Lake City
metropolitan planning organizationmetropolitan planning organization• MPOs maintain region’s short and long-term MPOs maintain region’s short and long-term
transportation planstransportation plans• The “3 C’s”The “3 C’s”
• Responsible for developing and using models Responsible for developing and using models to forecast future travel patternsto forecast future travel patterns• Mathematical models representing current travel Mathematical models representing current travel
behavior are used to forecast future travel behavior are used to forecast future travel behaviorbehavior
• Analyze future alternatives, quantify benefits and Analyze future alternatives, quantify benefits and costscosts
Quick Facts
• 2 MPOs
• 4 Counties
• 1300 Square Miles
•1.8 million people today
•2.7 million people by 2030
Typical AnalysesTypical Analyses
1) Air Quality Conformity -NAAQS
2) System Performance (aggregate)
-VMT, VHT, Mode Share, etc.
3) Corridor-level Analyses
-Identify and compare options
4) Facility Performance
-V/C, Ridership, speed
The other 3 C’sThe other 3 C’s
• ComplexityComplexity• Challenges (legal)Challenges (legal)• CreativityCreativity
Advancing the modeling practice is Advancing the modeling practice is not easy.not easy.
What is a Travel Model?What is a Travel Model?
A systematic tool to forecast future A systematic tool to forecast future travel. travel.
One of many tools used in decision-One of many tools used in decision-making process.making process.
The 5 steps of modeling process The 5 steps of modeling process (typically) are:(typically) are: 1. Land Use Forecasting 1. Land Use Forecasting 2. Trip Generation2. Trip Generation 3. Trip Distribution3. Trip Distribution 4. Mode Split4. Mode Split 5. Trip Assignment5. Trip Assignment
Model InputsModel Inputs
Network of zones Network of zones
and linksand links• 1300 zones 1300 zones
contain contain demographic demographic data data (people/jobs)(people/jobs)
• 20,000 links 20,000 links describe describe road/transit road/transit infrastructure infrastructure (lanes, speed, (lanes, speed, capacity, capacity, headway etc.)headway etc.)
NetworksNetworks
Trip GenerationTrip Generation
Trip GenerationTrip Generation Trip DistributionTrip Distribution Mode ChoiceMode Choice Trip AssignmentTrip Assignment
Each zone produces and attracts Each zone produces and attracts trips, based on the amount and trips, based on the amount and types of activities within the TAZ.types of activities within the TAZ.
Modeling Steps
TAZ Population Jobs
393
679
176
LANDUSE DATA
1000
500
0
0
300
800
Trip DistributionTrip Distribution
Trip Distribution Trip Distribution estimates the estimates the number of trips number of trips between zonesbetween zones
Trip GenerationTrip Generation Trip DistributionTrip Distribution Mode ChoiceMode Choice Trip AssignmentTrip Assignment
Modeling Steps
Mode ChoiceMode ChoiceMode Choice considers travel time, auto availability, and costs in estimating the likelihood of making trips by car, train, bus, etc.
Trip GenerationTrip Generation Trip DistributionTrip Distribution Mode ChoiceMode Choice Trip AssignmentTrip Assignment
Modeling Steps
Trip AssignmentTrip Assignment
Trip assignment estimates which road or route should be taken. Considers travel time, congestion, speed, distance, transit transfers, etc.
Trip GenerationTrip Generation Trip DistributionTrip Distribution Mode ChoiceMode Choice Trip AssignmentTrip Assignment
Modeling Steps
Trip-based ModelsTrip-based Models
Limitations of Traditional Limitations of Traditional ModelsModels• Aggregate and Trip-basedAggregate and Trip-based
• Poor accountingPoor accounting• Assume similarity within zonesAssume similarity within zones• Over-simplifies family dynamics and location Over-simplifies family dynamics and location
choicechoice• No feedback to land-use forecasting No feedback to land-use forecasting
processprocess• Land-use does not change with transportationLand-use does not change with transportation
• Simplistic response to land-useSimplistic response to land-use• No sensitivity to urban form (diversity, No sensitivity to urban form (diversity,
density, design)density, design)
Tour-based ModelsTour-based Models
Difficult Emerging Difficult Emerging QuestionsQuestions
• Land-use affects Land-use affects transportation transportation decisionsdecisions
• Transportation Transportation affects land-use affects land-use growthgrowth
• New technologies New technologies (e.g. ITS, rail)(e.g. ITS, rail)
• New policies (e.g. New policies (e.g. tolls, taxes)tolls, taxes)
Introduction to Introduction to UrbanSimUrbanSim
Forecasts future land-use (households, jobs)Forecasts future land-use (households, jobs) Effective means to incorporate city and Effective means to incorporate city and
county land-use plans into regional county land-use plans into regional transportation planstransportation plans
State-of-the-artState-of-the-art Defensible microeconomic theoryDefensible microeconomic theory
Incorporates transportation Incorporates transportation accessibilityaccessibility
Locally calibratedLocally calibrated Tremendous interest across the U.S.Tremendous interest across the U.S.
WFRC InterestWFRC Interest
Committed to exploring and discussing Committed to exploring and discussing linkages between land-use and linkages between land-use and transportation in LRTPtransportation in LRTP Wasatch Choices visioning effortWasatch Choices visioning effort
Extensive staff time fine-tuning UrbanSim Extensive staff time fine-tuning UrbanSim database and modeldatabase and model Major technical questions have been answeredMajor technical questions have been answered Testing about to begin anew in visioning effortTesting about to begin anew in visioning effort
UrbanSim – Travel Model UrbanSim – Travel Model InteractionsInteractions
UrbanSimTravel Models
Households by Income Age of head Size Workers ChildrenEmployment by sector
AccessibilityHighway Travel TimesVehicle Ownership Probabilities
Linked Urban MarketsLinked Urban Markets
Governments Infrastructure
Land
FloorspaceHousing
Households BusinessesLabor
Services
Developers
Flow of consumption from supplier to consumer
Regulation or Pricing
Overview of Modeling Overview of Modeling systemsystem
>30 models within local UrbanSim >30 models within local UrbanSim applicationapplication Land Value (by type of use)Land Value (by type of use) Real Estate Development (by type of Real Estate Development (by type of
use; intensity)use; intensity) Residential location (by type of Residential location (by type of
household)household) Employment location (by type of Employment location (by type of
industry)industry)
Key Variables in ModelsKey Variables in Models Land value Land value Vacant land (for developer models)Vacant land (for developer models) Accessibility measures (for example)Accessibility measures (for example)
Proximity to transportation facilitiesProximity to transportation facilities Jobs/households within 30 minutesJobs/households within 30 minutes
Neighborhood traits (for example)Neighborhood traits (for example) Housing and employment within walking distanceHousing and employment within walking distance Neighborhood mix (e.g. by income, by type of real Neighborhood mix (e.g. by income, by type of real
estate)estate) Decision-maker’s characteristics (e.g. Decision-maker’s characteristics (e.g.
income, HH size, sector)income, HH size, sector)
Model ConstraintsModel Constraints Environmental featuresEnvironmental features
Steep slopeSteep slope Wetlands/lakes/streamsWetlands/lakes/streams SuperfundSuperfund
Regional PoliciesRegional Policies Urban growth boundaryUrban growth boundary Open SpaceOpen Space
Local Land PoliciesLocal Land Policies Type of useType of use Allowable density of useAllowable density of use
Observed Predicted
Lan
d P
rice V
ali
dati
on
Lan
d P
rice V
ali
dati
on
Resi
den
tial
Locati
on
R
esi
den
tial
Locati
on
V
ali
dati
on
Vali
dati
on
Observed Total Observed % Modeled Utility
VisioningVisioning
Plans to test UrbanSim extensively Plans to test UrbanSim extensively over next 4-6 monthsover next 4-6 months
Plenty of opportunity for local Plenty of opportunity for local review and feedbackreview and feedback
Relatively safe opportunity to vary Relatively safe opportunity to vary land and transportation policies and land and transportation policies and see what the model sayssee what the model says
Political ChallengesPolitical Challenges
Political issues can be more Political issues can be more challenging than the technicalchallenging than the technical
Inherent resistance to changeInherent resistance to change Committing to a tool like UrbanSim Committing to a tool like UrbanSim
affects entire planning realm affects entire planning realm (local/regional/state)(local/regional/state)
Implications for project development Implications for project development must be well understoodmust be well understood