September 2007Travel Forecasting for New Starts3-1 10. Uncertainty Analysis FTA requirements for New Starts FTA requirements for New Starts Implementation

September 2007September 2007 Travel Forecasting for New StartsTravel Forecasting for New Starts 3-3-11

10. Uncertainty 10. Uncertainty AnalysisAnalysis

FTA requirements for New StartsFTA requirements for New Starts ImplementationImplementation


FTA RequirementsFTA Requirements

Specific SAFETEA-LU provisionsSpecific SAFETEA-LU provisions– Project ratings and reliability of forecastsProject ratings and reliability of forecasts– Before-After studies of predicted/actualBefore-After studies of predicted/actual– FFGA bonus awards for “good” forecastsFFGA bonus awards for “good” forecasts– Tracking of contractor performanceTracking of contractor performance

Travel forecasting measureTravel forecasting measure– Guideway ridersGuideway riders– Measurable; most visible elementMeasurable; most visible element


Early Final Design

FFGA

Uncertainty Analysis

PE B-A Installment

Alternatives Analysis

Preliminary Engineering

Rating to Enter PE

Rating to Enter FD

Rating for FFGA

Bonus Award

Cumulative Records of Contractor and

Sponsor Performance

Interim and Final B-A Findings from Other

Projects by Contractor and Sponsor

FD B-A Installment

Bonus Claim

Construction; Opening

2 Years of Service

Opening B-A Installment

Final B-A Assessment

Uncertainty in and Accuracy of Cost and Ridership Forecasts for New Starts Projects

- Uncertainty Analysis- FTA Ratings- Before-After Studies- FFGA Bonus Awards- Performance Tracking


ImplementationImplementation

Prior to PE application (in AA, post Prior to PE application (in AA, post AA)AA)– Build-up of preferred-alternative Build-up of preferred-alternative

forecastforecast– Scrutiny of large contributorsScrutiny of large contributors– Range of forecastsRange of forecasts– Formal documentationFormal documentation

Updated at entry to Final DesignUpdated at entry to Final Design


Build-up of LPA Build-up of LPA ForecastForecast Series of forecasts for:Series of forecasts for:

– TodayToday– Plus future transit networkPlus future transit network– Plus new transit behaviorsPlus new transit behaviors– Plus future trip tablesPlus future trip tables– Plus future highway congestionPlus future highway congestion– Plus future parking costsPlus future parking costs– Plus alternative land use (?)Plus alternative land use (?)

Choice ridersPark/ride etc.Guideway effects


Scrutiny of “Drivers”Scrutiny of “Drivers”

PerspectivesPerspectives– Reliability of technical methodsReliability of technical methods– Consistency with current behavior, Consistency with current behavior,

trendstrends– Consistency with peersConsistency with peers

Alternative outcomesAlternative outcomes


Range of ForecastsRange of Forecasts

Complete forecasts (not factored)Complete forecasts (not factored)– Most likelyMost likely

Adjustments to key contributors?Adjustments to key contributors? For project evaluation, so with user For project evaluation, so with user

benefitsbenefits Template for opening-year forecastTemplate for opening-year forecast

– Lower bound: P(lower outcome) < 20% Lower bound: P(lower outcome) < 20% – Upper bound: P(higher outcome) < Upper bound: P(higher outcome) <

20%20%


DocumentationDocumentation

Range of forecasts (low, likely, Range of forecasts (low, likely, high)high)– Ridership patternsRidership patterns– Guideway ridershipGuideway ridership

DiscussionDiscussion– Key drivers of most-likely forecastKey drivers of most-likely forecast– Significant downside uncertaintiesSignificant downside uncertainties– Significant upside uncertaintiesSignificant upside uncertainties


11. Before-After 11. Before-After StudiesStudies

FTA requirements for New StartsFTA requirements for New Starts ImplementationImplementation Thoughts on good practiceThoughts on good practice



New/Small Starts New/Small Starts Before-After study Before-After study Element of project scopeElement of project scope

– Pre-approved work-plan requiredPre-approved work-plan required– Eligible for FTA New Starts fundsEligible for FTA New Starts funds

Dual purposesDual purposes– Impacts of the project: “before vs. after”Impacts of the project: “before vs. after”– Accuracy of forecasts: “predicted vs. Accuracy of forecasts: “predicted vs.

actual”actual” Annual report to Congress on findingsAnnual report to Congress on findings



Before versus afterBefore versus after– Conditions prior to project implementationConditions prior to project implementation– Conditions 2 years after project openingConditions 2 years after project opening– Understanding of project impactsUnderstanding of project impacts

Predicted versus actualPredicted versus actual– Accuracy of forecastsAccuracy of forecasts– Causes of differencesCauses of differences– Implications for methods, QC, Implications for methods, QC,

managementmanagement



MilestoneMilestone ActivitiesActivities

Post AAPost AA Uncertainty analysis; forecast Uncertainty analysis; forecast preservationpreservation

Post PEPost PE Analysis of revisions; forecast Analysis of revisions; forecast preservationpreservation

Pre-projectPre-project Collection of “before” dataCollection of “before” data

After opening (+2 After opening (+2 yrs)yrs)

Collection of “after” dataCollection of “after” data

Analysis of project impactsAnalysis of project impacts

Assessment of forecast accuracyAssessment of forecast accuracy



Uncertainties analysisUncertainties analysis Analysis of interim changesAnalysis of interim changes Preservation of forecastsPreservation of forecasts Collection of data (“before,” Collection of data (“before,”

“after”)“after”) Completion of the studyCompletion of the study



Analysis of interim changesAnalysis of interim changes– Identification of causesIdentification of causes

Changes in project scopeChanges in project scope Changes in demographic forecastsChanges in demographic forecasts OthersOthers

– Quantification of impact (no hand-Quantification of impact (no hand-waving)waving) Separate contributionsSeparate contributions Full travel forecastsFull travel forecasts



Preservation of forecastsPreservation of forecasts– DocumentationDocumentation– Networks, demographics, modelsNetworks, demographics, models– Preservation of Preservation of ability to replicate ability to replicate

forecastsforecasts Computer(s) in the closetComputer(s) in the closet Migration to new software, hardware, Migration to new software, hardware,

modelsmodels

– FTA oversight contractor FTA oversight contractor archives archives



Collection of data (“before,” “after”)Collection of data (“before,” “after”)– Conceptual design/budget in approved Conceptual design/budget in approved

planplan– Detailed designDetailed design

Sampling plan, methods, data itemsSampling plan, methods, data items Opportunity for FTA comment (approval?)Opportunity for FTA comment (approval?)

– Preservation of dataPreservation of data– FTA oversight contractorFTA oversight contractor



Completion of the studyCompletion of the study– Impact of the projectImpact of the project

Changes in services, ridershipChanges in services, ridership Meaningful differences in “before,” “after” Meaningful differences in “before,” “after”

datadata

– Accuracy of forecastsAccuracy of forecasts Ridership forecast versus “after” dataRidership forecast versus “after” data Analysis of differencesAnalysis of differences

– Full forecasts demonstrating impacts of changesFull forecasts demonstrating impacts of changes– No handwavingNo handwaving



Completion of the study Completion of the study (continued)(continued)

– DocumentationDocumentation– FTA contractorFTA contractor– FTA acceptance of completed studyFTA acceptance of completed study

Experience to dateExperience to date– Salt Lake City and DallasSalt Lake City and Dallas– ““After” with no “before;” limited After” with no “before;” limited

“predicted”“predicted”– Demonstrate importance of Demonstrate importance of preservationpreservation


12. Performance 12. Performance TrackingTracking FTA requirements for New StartsFTA requirements for New Starts ImplicationsImplications Implementation ideasImplementation ideas

– Outlines of a proposalOutlines of a proposal– Formal draft, comments, and final Formal draft, comments, and final

policy guidance in 2008policy guidance in 2008



Annual report to CongressAnnual report to Congress– Projects: FFGA New Starts; PCA Small Projects: FFGA New Starts; PCA Small

StartsStarts Summary of forecastsSummary of forecasts Identification of forecasting “contractors”Identification of forecasting “contractors”

– New Starts projects opened to serviceNew Starts projects opened to service Summary of first-year ridershipSummary of first-year ridership Assessment of forecasts, Assessment of forecasts, causes of errorscauses of errors


ImplicationsImplications

Some areas of tensionSome areas of tension– Risk | controlRisk | control

Contractors assume the risk of “bad” performance Contractors assume the risk of “bad” performance gradesgrades

Sponsors and others control key resources for forecastingSponsors and others control key resources for forecasting Budget, schedule, data, existing models …Budget, schedule, data, existing models …

– Conditions | fundingConditions | funding Contractors’ risk minimized by identifying uncertaintiesContractors’ risk minimized by identifying uncertainties Sponsors’ funding put at risk by identified uncertaintiesSponsors’ funding put at risk by identified uncertainties

– AA contractor | PE contractorAA contractor | PE contractor May not be the same (good or bad for analytical rigor??)May not be the same (good or bad for analytical rigor??)


ImplicationsImplications

Some more areas of tensionSome more areas of tension– Evaluation measure | measurable impactsEvaluation measure | measurable impacts

Projects are evaluated on mobility benefitsProjects are evaluated on mobility benefits Project ridership is more measurable and visibleProject ridership is more measurable and visible

– Forecast year | performance yearForecast year | performance year Projects are evaluated with 2030 benefitsProjects are evaluated with 2030 benefits Contractor performance is based on opening yearContractor performance is based on opening year


Implementation IdeasImplementation Ideas

Scope: assessment of all Scope: assessment of all contributionscontributions

ContractorContractor Project sponsor, MPO, othersProject sponsor, MPO, others

Uncertainties analysisUncertainties analysis– Required prior to PE applicationRequired prior to PE application– Forecasts and methods preserved for Forecasts and methods preserved for

use in later analysesuse in later analyses



Principal measuresPrincipal measures– Guideway ridershipGuideway ridership– System ridershipSystem ridership

ConsistencyConsistency– 2030 & 12030 & 1stst year: same methods year: same methods– Allowance for initial maturation Allowance for initial maturation

effectseffects FTA oversight contractorFTA oversight contractor


Performance ScoringPerformance Scoring

Parallels project ratingsParallels project ratings– Five rating categories (High … Low)Five rating categories (High … Low)– Multiple measuresMultiple measures– Weighted average with judgmentWeighted average with judgment

CriteriaCriteria– Proximity of actual ridership to forecastProximity of actual ridership to forecast– Sources of error controlled by Sources of error controlled by

contractorcontractor– Sources of error controlled by othersSources of error controlled by others


CriteriaCriteria

Proximity: actual vs. most likely Proximity: actual vs. most likely forecastforecast– Within Within ±20 percent±20 percent– Below floorBelow floor– Above ceilingAbove ceiling

Sources of error (contractor)Sources of error (contractor)– Source identified/exploredSource identified/explored– Implications quantified Implications quantified – Adjustments high/low forecastsAdjustments high/low forecasts

In uncertainties analysis


CriteriaCriteria

Sources of error (others)Sources of error (others)– Source identified/exploredSource identified/explored– Implications quantified Implications quantified – Adjustments high/low forecastsAdjustments high/low forecasts

In uncertainties analysis


Ideas and CommentsIdeas and Comments

NowNow E-mail, soonE-mail, soon Formally in early 2008Formally in early 2008


13. Transit Path 13. Transit Path ChoicesChoices

No FTA requirements on this topicNo FTA requirements on this topic Some observationsSome observations Three presentationsThree presentations DiscussionDiscussion


Some ObservationsSome Observations

Transit choicesTransit choices– Access mode (walk, bus, PnR, KnR, Access mode (walk, bus, PnR, KnR,

etc.)etc.)– Line-haul mode (bus, rail, etc.)Line-haul mode (bus, rail, etc.)– Path (first boarding, last alighting)Path (first boarding, last alighting)

Central issue for model designCentral issue for model design– Choices handled by the pathbuilder?Choices handled by the pathbuilder?– Choices handled in “mode” choice?Choices handled in “mode” choice?


An ExampleAn Example

Setting: HonoluluSetting: Honolulu– Dense existing bus networkDense existing bus network– Corridor defined by geographic constraints Corridor defined by geographic constraints – Rail options imply lots of bus changesRail options imply lots of bus changes

PathbuilderPathbuilder– All-or-nothing (with combined headways)All-or-nothing (with combined headways)

Question: pathbuilder-alone adequate?Question: pathbuilder-alone adequate?


Some AnalysisSome Analysis

Transit tripsTransit trips– Build alternativeBuild alternative– HBW/peak = 81,200 tripsHBW/peak = 81,200 trips

Transit pathsTransit paths– Best bus-onlyBest bus-only– Rail, bus IVT weight = 1.2Rail, bus IVT weight = 1.2

ObservationsObservations– 34,500 trips have choice34,500 trips have choice– 9,200 trips, |9,200 trips, |ΔΔ| < 5 min.| < 5 min.– 16,100 trips, |16,100 trips, |ΔΔ| < 15 min.| < 15 min.– Best path Best path more rail trips more rail trips– Path choice Path choice more UBs more UBs

Transit Trips with Rail and Bus Paths

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

-145

-135

-125

-115

-105 -95

-85

-75

-65

-55

-45

-35

-25

-15 -5 5 15 25

Rail Impedance minus Bus Impedance

Per

cen

t o

f H

BW

Tra

nsi

t T

rip

s

freq

cuml

difference = 0

difference= 0

rail better

bus

better


A Modified ApproachA Modified Approach

Pathbuilder: four best pathsPathbuilder: four best paths Best bus/walkBest bus/walk Best bus/driveBest bus/drive Best rail/walkBest rail/walk Best rail/driveBest rail/drive

Mode choice modelMode choice model– ““Transit” mode = four discrete Transit” mode = four discrete

choiceschoices– Probably with some nested structureProbably with some nested structure


Two Design OptionsTwo Design Options

auto transit

bus/w bus/drail/w rail/d

walk drive

choice

auto transit

walk drive

choice

“MODE”

CHOICE

TRANSIT PATH BUILDE

R

best trn/w path

“best” bus/w path

best trn/d path

“best” bus/d path

“best” rail/w path

“best” rail/d path

Option A Option B


Actually,Actually,ManyMany Design Options Design Options Other transit choicesOther transit choices

– Ferry Ferry – Local bus, limited-stop bus, express Local bus, limited-stop bus, express

busbus– Walk-rail versus walk-bus-railWalk-rail versus walk-bus-rail

Other influencesOther influences– Transit pathbuilding algorithmTransit pathbuilding algorithm– Zone sizeZone size– Computational intensityComputational intensity


PresentationsPresentations

Path Choice with Substantial Reliance Path Choice with Substantial Reliance On Discrete-Choice Models On Discrete-Choice Models

Bill Davidson, Parsons BrinckerhoffBill Davidson, Parsons Brinckerhoff

Path Choice with Principal Reliance Path Choice with Principal Reliance On Networks and Path-builders On Networks and Path-builders

Bill Woodford, AECOM ConsultBill Woodford, AECOM Consult

““To Multipath or Not to Multipath” To Multipath or Not to Multipath” – The Denver Experience – The Denver Experience

David Kurth, Cambridge Systematics, Inc.David Kurth, Cambridge Systematics, Inc.


Path Choice with Substantial Path Choice with Substantial RelianceReliance

On Discrete-Choice ModelsOn Discrete-Choice Models

Bill DavidsonBill Davidson

Parsons BrinckerhoffParsons Brinckerhoff

September 2007September 2007 Travel Forecasting for New StartsTravel Forecasting for New Starts 3-3-3838September 2007September 2007 Travel Forecasting for New StartsTravel Forecasting for New Starts 3838

Why Rely Heavily on Why Rely Heavily on Discrete Choice Discrete Choice Models?Models? Many shades of gray?Many shades of gray? What might be a decision What might be a decision

framework?framework? Considering the full range of Considering the full range of

choiceschoices Behavioral implicationsBehavioral implications


Los Angeles Nested Los Angeles Nested ModelModel


Miami Mode Choice Miami Mode Choice ModelModel


Thoughts about a Thoughts about a Decision FrameworkDecision Framework What are the choices to be What are the choices to be

considered?considered?– ExistingExisting– AndAnd future future

Understanding marketsUnderstanding markets– Context specific (“one size does not fit Context specific (“one size does not fit

all”)all”)– Survey data requirements – and Survey data requirements – and qualityquality

September 2007September 2007 Travel Forecasting for New StartsTravel Forecasting for New Starts 4141


Some Possible CriteriaSome Possible Criteria

Non-included AttributesNon-included Attributes– Facility relatedFacility related– Span of serviceSpan of service– Passenger amenitiesPassenger amenities– Trip characteristicsTrip characteristics

Vehicle, reliability, seat availability……Vehicle, reliability, seat availability……

CompetitionCompetition



(More) Possible (More) Possible CriteriaCriteria Market segmentationMarket segmentation

– Traveler, access/egress….Traveler, access/egress…. ElasticitiesElasticities

– TradeoffsTradeoffs Mobility influencesMobility influences

– More choices available to the More choices available to the travelertraveler



Choice DimensionsChoice Dimensions

Physical & operational characteristicsPhysical & operational characteristics Access/egressAccess/egress

– Market segmentationMarket segmentation San Diego and small area geographySan Diego and small area geography Differences in walk access options (bus v. Differences in walk access options (bus v.

rail)rail)

– Boarding location choiceBoarding location choice StationStation Bus stopBus stop



From the MSP From the MSP WorkshopWorkshop

Zone I: 1 mile square Walk-rail: 25% Walk-transit: 100%

Zone J: 1 mile square Walk-rail: 12.5% Walk-transit: 100%

RAIL LINE

LOCAL BUS

LOCAL BUS

STATION STATION

What transit options are available to whom?

LOCAL BUS

Maximum walk distance = 0.5 mi.


Access RepresentationAccess Representation

Paths from I to JPaths from I to J– DetailedDetailed

walk-rail-walkwalk-rail-walk walk-bus-rail-walkwalk-bus-rail-walk walk-rail-bus-walkwalk-rail-bus-walk walk-bus-walkwalk-bus-walk drive-rail-walkdrive-rail-walk drive-rail-bus-walkdrive-rail-bus-walk

– TypicalTypical walk-local-walkwalk-local-walk walk-premium-walkwalk-premium-walk drive-transit-walkdrive-transit-walk

Markets from I to JMarkets from I to J– DetailedDetailed

25 x 12.5 25 x 12.5 = 3.125%= 3.125% 100 x 12.5 100 x 12.5 = 12.5%= 12.5% 25 x 10025 x 100 = 25%= 25% 100 x 100100 x 100 = 100%= 100% 100 x 12.5100 x 12.5 = 12.5%= 12.5% 100 x 100100 x 100 = 100%= 100%

– TypicalTypical 100 x 100100 x 100 = 100%= 100% 100 x 100100 x 100 = 100%= 100% 100 x 100100 x 100 = 100%= 100%

!


More Choice More Choice DimensionsDimensions CompetitionCompetition

– Access (WMATA)Access (WMATA) Walk to bus to railWalk to bus to rail Direct walk to railDirect walk to rail

– Primary mode (Los Angeles)Primary mode (Los Angeles) Metrolink v. Urban Rail v. Transitway Metrolink v. Urban Rail v. Transitway

– long distance travellong distance travel Urban Rail v. Rapid Bus v. Local Bus Urban Rail v. Rapid Bus v. Local Bus

– Intra corridor travelIntra corridor travel



Even More Choice Even More Choice DimensionsDimensions Modal InteractionsModal Interactions

– Metrolink & Red LineMetrolink & Red Line– Orange Line (BRT) & Red LineOrange Line (BRT) & Red Line

60% of Orange Line riders transfer to 60% of Orange Line riders transfer to Red LineRed Line

– Implicit Hierarchy in Nested ModelsImplicit Hierarchy in Nested Models Where is that Red Line rider?Where is that Red Line rider?

– Metrolink, Urban Rail, BRT, Rapid Bus ??? Metrolink, Urban Rail, BRT, Rapid Bus ???



Behavioral Behavioral ImplicationsImplications Consideration of non-included Consideration of non-included

attributesattributes– Fixed v. variableFixed v. variable

Value of time differencesValue of time differences– Fare contribution to path choiceFare contribution to path choice– Express bus, urban rail, commuter railExpress bus, urban rail, commuter rail

ElasticitiesElasticities– 500 new spaces at Lot A500 new spaces at Lot A




Why Rely Heavily on Why Rely Heavily on Discrete Choice Discrete Choice Models?Models? Choice complexitiesChoice complexities

– Access/egress (market segmentation)Access/egress (market segmentation)– CompetitionCompetition– InteractionsInteractions

Behavioral considerationsBehavioral considerations– Non-included attributesNon-included attributes– Value of timeValue of time– Elasticities/mobility influencesElasticities/mobility influences



Path Choice with Principal Path Choice with Principal RelianceReliance

On Networks and Path-buildersOn Networks and Path-builders

Bill Woodford, AECOM ConsultBill Woodford, AECOM Consult


Range of Options;Range of Options;Not an Either/Or ChoiceNot an Either/Or Choice Discrete choice models depend on Discrete choice models depend on

network path builders for each choice network path builders for each choice (or component of a choice)(or component of a choice)

Most models that rely on transit path Most models that rely on transit path builders still have separate choices for builders still have separate choices for access mode (walk vs. drive access)access mode (walk vs. drive access)

Key question:Key question:What is a path-building decision and What is a path-building decision and what is a mode-choice decision?what is a mode-choice decision?


The Range of OptionsThe Range of Options

Person Trips

TransitAuto

Walk-Transit

Drive-Transit

Person Trips

TransitAuto

Local Bus

Express Bus

BRT LRT/ HRT

CommRail

Walk Access

Drive Access

KNR PNR

Walk Access

Drive Access

KNR PNR

Similar to Local Bus

Similar to LRT/HRT

Similar to LRT/HRT

Station Pair 1

Station Pair 2

Station Pair 1

Station Pair 2

Station Pair 1

Station Pair 2

Mostly Path Driven Mostly Discrete

Choice Driven


Philosophy Behind Philosophy Behind RelianceRelianceon Network/Path-builderson Network/Path-builders All other things being equal, a simple model is All other things being equal, a simple model is

preferable to a complex model since it is:preferable to a complex model since it is:– Faster to developFaster to develop– Easier to understand and explainEasier to understand and explain– Less likely to have unknown/undesirable interrelationshipsLess likely to have unknown/undesirable interrelationships

Complexity is needed when a simpler model doesn’t:Complexity is needed when a simpler model doesn’t:– Depict how travelers behave (mode and submode level)Depict how travelers behave (mode and submode level)– Provide important information on the operation of a projectProvide important information on the operation of a project– Tell the story of a projectTell the story of a project

Bottom Line:Bottom Line:– Start simple, add complexity as neededStart simple, add complexity as needed– Begin by building the best paths possible…good paths are Begin by building the best paths possible…good paths are

essential for choice based models also.essential for choice based models also.


Other Questions Other Questions Influencing Model DesignInfluencing Model Design Does the software permit realistic mode-Does the software permit realistic mode-

specific paths?specific paths? Can I afford the time/storage associated with Can I afford the time/storage associated with

a separate set of skims for each choice?a separate set of skims for each choice? Can I define a transit sub-mode hierarchy the Can I define a transit sub-mode hierarchy the

properly represents the relationships among properly represents the relationships among the options?the options?

Will this mode hierarchy continue into future Will this mode hierarchy continue into future with the introduction of new projects?with the introduction of new projects?

Does added complexity help or hinder telling Does added complexity help or hinder telling the story of the project?the story of the project?


ExampleExample

What happens with rail replaces What happens with rail replaces bus in a simple network?bus in a simple network?

Calibration case (and baseline):Calibration case (and baseline):– Bus only systemBus only system– 5% transit share5% transit share

Modeling questionsModeling questions– UTPS or multipath?UTPS or multipath?– Path-based or choice-based?Path-based or choice-based?


Example – BaselineExample – Baseline

A

D C

B

Red Local Bus –20 min Headway




Run

tim

e: 1

5 m

in

Run

tim

e: 1

5 m

in

Run time: 20 min

Run time: 20 min

Example 1 - Baseline


Example – BuildExample – Build

A

D C

B

LRT – 20 min Headway




Run

tim

e: 1

5 m

in

Run

tim

e: 1

5 m

in

Run time: 20 min

Run time: 20 min

Example 1 - Build


Pathbuilder (UTPS Paths) Pathbuilder (multi paths)Nest Coef 1 Nest Coef 1

Baseline HWY Transit Vacant HWY Transit VacantImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -3.32750E+00 0.00000E+00 -3.75000E-01 -3.31875E+00 0.00000E+00e(UTIL/nest coef) 3.58827E-02 0.00000E+00 3.61981E-02 0.00000E+00

Sub Mode Share 100.0% 0.0% 100.0% 0.0%logsum -3.32750E+00 -3.31875E+00Eutil 6.87289E-01 3.58827E-02 6.87289E-01 3.61981E-02Main Mode Share 95.0% 5.0% 95.0% 5.0%

Share 95.0% 5.0% 0.0% 95.0% 5.0% 0.0%Trips (per 100) 95.0 5.0 - 95.0 5.0 - Build HWY Transit Vacant HWY Transit VacantImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -3.32750E+00 0.00000E+00 -3.75000E-01 -3.31875E+00 0.00000E+00e(UTIL/nest coef) 3.58827E-02 0.00000E+00 3.61981E-02 0.00000E+00


Share 95.0% 5.0% 0.0% 95.0% 5.0% 0.0%Trips (per 100) 95.0 5.0 - 95.0 5.0 - Evaluation MeasuresLRT Trips 0 or 5 2.5 Incremetal Transit Trips - - UB Min - -

Pathbuilder-Based – As Defined (No Time Savings)


Pathbuilder (UTPS Paths)- 1 min LRT saved Pathbuilder (multi paths)- 1 minute LRT savedNest Coef 1 Nest Coef 1

Baseline HWY Transit Vacant HWY Transit VacantImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -3.32750E+00 0.00000E+00 -3.75000E-01 -3.31875E+00 0.00000E+00e(UTIL/nest coef) 3.58827E-02 0.00000E+00 3.61981E-02 0.00000E+00


Share 95.0% 5.0% 0.0% 95.0% 5.0% 0.0%Trips (per 100) 95.0 5.0 - 95.0 5.0 - Build HWY Transit Vacant HWY Transit VacantImpedance TermsIVTT -0.025 15 34 15 34.5WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -3.30250E+00 0.00000E+00 -3.75000E-01 -3.30625E+00 0.00000E+00e(UTIL/nest coef) 3.67911E-02 0.00000E+00 3.66534E-02 0.00000E+00


Share 94.9% 5.1% 0.0% 94.9% 5.1% 0.0%Trips (per 100) 94.9 5.1 - 94.9 5.1 - Evaluation MeasuresLRT Trips 5.1 2.6 Incremetal Transit Trips 0.1192 0.0597 UB Min 5.0212 2.5165

Pathbuilder-Based – Adjusted (1 minute LRT Time Savings)


Choice-Based – As Defined (No Time Savings)Choice/UTPS Paths Choice/MultipathsNest Coef 0.3 Nest Coef 0.3

Baseline HWY Bus LRT HWY Bus LRTImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -1.10917E+01 0.00000E+00 -3.75000E-01 -1.10625E+01 0.00000E+00e(UTIL/nest coef) 1.52388E-05 0.00000E+00 1.56898E-05 0.00000E+00


Share 95.0% 5.0% 0.0% 95.0% 5.0% 0.0%Trips (per 100) 95.0 5.0 - 95.0 5.0 - Build HWY Bus LRT HWY Bus LRTImpedance TermsIVTT -0.025 15 35 35 15 35 35WAIT -0.0625 25 25 25 25WALK -0.0625 10 10 10 10XFER -0.125 1 1 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -1.10917E+01 -1.10917E+01 -3.75000E-01 -1.26250E+01 -1.26250E+01e(UTIL/nest coef) 1.52388E-05 1.52388E-05 3.28876E-06 3.28876E-06


Share 94.0% 3.0% 3.0% 96.1% 1.9% 1.9%Trips (per 100) 94.0 3.0 3.0 96.1 1.9 1.9 Evaluation MeasuresLRT Trips 3.0 1.9 Incremetal Transit Trips 1.0776 (1.1038) UB Min 45.6151 (46.2099)


Choice-Based – Adjusted (1 minute LRT Time Savings)Choice/UTPS Paths- 1 min LRT saved Choice/Multipaths - 1 min LRT savedNest Coef 0.3 Nest Coef 0.3

Baseline HWY Bus LRT HWY Bus LRTImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -1.10917E+01 0.00000E+00 -3.75000E-01 -1.10625E+01 0.00000E+00e(UTIL/nest coef) 1.52388E-05 0.00000E+00 1.56898E-05 0.00000E+00






Deep Nested Choice-Based – Adjusted (1 minute LRT Time Savings) Choice/UTPS Paths- 1 min LRT saved Choice/Multipaths - 1 min LRT saved

Nest Coef 0.01 Nest Coef 0.01Baseline HWY Bus LRT HWY Bus LRTImpedance TermsIVTT -0.025 15 35 15 35WAIT -0.0625 25 17.5WALK -0.0625 10 10XFER -0.125 1 1Constant 0 -0.14 -0.14 0 -0.6 -0.6Mode Choice ComputationsUTIL/nest coef -3.75000E-01 -3.32750E+02 0.00000E+00 -3.75000E-01 -3.31875E+02 0.00000E+00e(UTIL/nest coef) 3.07972E-145 0.00000E+00 7.38786E-145 0.00000E+00






Example: SummaryExample: Summary

UTPS Paths Multipaths Choice Choice Deep Nested Deep NestedUTPS Multipath UTPS Multipath

No Time Savings--Base Transit Trips 5.0 5.0 5.0 5.0 5.0 5.0 --LRT Trips 2.5 2.5 3.0 1.9 4.7 3.0 --Delta Transit Trips - - 1.1 (1.1) 0.0 (1.8) --UB Minutes - - 45.6 (46.2) 1.4 (74.7) --UB Min./Base Transt Trip - - 9.1932 (9.2359) 0.2782 (14.9331) 1 min LRT TimeSavings--Base Transit Trips 5.0 5.0 5.0 5.0 5.0 5.0 --LRT Trips 5.1 2.6 3.0 1.9 4.7 3.0 --Delta Transit Trips 0.1 0.1 1.2 (1.1) 0.1 (1.7) --UB Minutes 5.0 2.5 48.7 (44.2) 5.2 (72.3) --UB Min./Base Transit Trip 1.0 0.5 9.8182 (8.8357) 1.0443 (14.4444)

Note for UTPS, no time savings, equal impedance assumed to be evenly distributed among paths


QuestionsQuestions

Should multi-path credit be assigned Should multi-path credit be assigned to multiple bus paths also?to multiple bus paths also?

What does define an independent What does define an independent choice as distinct from a typical bus choice as distinct from a typical bus path choice?path choice?

Does it matter since a deeply nested Does it matter since a deeply nested outcome begins to mirror path-based outcome begins to mirror path-based models?models?

Can multi-path path-builders co-exist Can multi-path path-builders co-exist with nested choice models?with nested choice models?


Conclusion: Depends on Conclusion: Depends on Having a Meaningful Having a Meaningful ChoiceChoice Significantly different level of service / Significantly different level of service /

comfortcomfort– Guaranteed seatGuaranteed seat– Fare differentFare different– Substantial time improvementSubstantial time improvement– Independent marketing identityIndependent marketing identity

Evidence that presence of multiple Evidence that presence of multiple choices increases mode share choices increases mode share independent of time and costindependent of time and cost


Transit Path-Building:Transit Path-Building:““To Multipath or Not to Multipath”To Multipath or Not to Multipath”– – The Denver ExperienceThe Denver Experience

David Kurth, Cambridge Systematics, Inc.David Kurth, Cambridge Systematics, Inc.Based on work performed with:Based on work performed with: Suzanne Childress (Parsons)Suzanne Childress (Parsons) Erik Sabina & Sreekanth Ande (DRCOG)Erik Sabina & Sreekanth Ande (DRCOG) Lee Cryer (Denver RTD)Lee Cryer (Denver RTD)


Investigation ContextInvestigation Context

DRCOG Integrated Regional Model (IRM) DRCOG Integrated Regional Model (IRM) developmentdevelopment– Activity / tour-based modelActivity / tour-based model– Better representation of transit possibleBetter representation of transit possible– Correct options in estimation dataset Correct options in estimation dataset

required for proper estimationrequired for proper estimation Detailed Travel Behavior Inventory (TBI) Detailed Travel Behavior Inventory (TBI)

datadata– Provided for detailed path-checkingProvided for detailed path-checking


Simple Path-Builder + Simple Path-Builder + Simple Mode ChoiceSimple Mode Choice

Mode Choice

Auto Transit

Transit All-or-Nothing


Complex Path-Builder Complex Path-Builder + + Simple Mode ChoiceSimple Mode Choice

Mode Choice

Auto Transit

Transit Multi-Path

Local Bus

Premium Bus

Rail

Local & Premium

Local Bus & Rail

Premium Bus & Rail

All Modes


Simple Path-Builder + Simple Path-Builder + Complex Mode ChoiceComplex Mode Choice

Mode Choice

Local Only

Premium Only

Rail Only

Local & Premium

Local & Rail

Premium & Rail

Local, Premium,

& Rail

AON Assign

AON Assign

AON Assign

AON

Assign

AON Assign

AON

Assign

AON Assign


Complex Path-Builder + Complex Path-Builder + Complex Mode ChoiceComplex Mode Choice

Mode

Choice

Local Bus Trips

Premium Bus Trips

Rail Trips

Multi-path Assignment

(Local, Premium & Rail Networks)

Local Boardings

Premium Boardings

Rail Boardings


(Local & Premium

Networks)


(Local Only Network)


IRM Design OptionsIRM Design Options

Path-Path-BuilderBuilder

Mode Mode ChoiceChoice

SimpleSimple SimpleSimple Shown to not Shown to not workwork

ComplexComplex SimpleSimple

SimpleSimple ComplexComplex

ComplexComplex ComplexComplex Possible Possible “confusion”“confusion”


$

$

$

$

$

$

$

$

$$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$$

$$

$

$$

$$

$$

$

$

$ $

$

$$

$

$

$$

$$

$$

$

$$

$$

$$

$

$$

$

$$

$$

$$

$$

$$

$$ $ $ $

$

$

$

$

$

$

$

$

$

$

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$$ $

$

$

$$

$

$$

$

$

$

$

$

$

$ $

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

Example RTD Path Example RTD Path OptionsOptions

3 Reasonable Paths3 Reasonable Paths– Path 1:Path 1: 2 Local 2 Local

BusesBuses– Path 2:Path 2: 2 Local 2 Local

BusesBuses– Path 3: Local Bus, Path 3: Local Bus,

Rail, Rail, Mall Mall ShuttleShuttle

Travel Behavior Travel Behavior Inventory (TBI) had Inventory (TBI) had observations for all observations for all three!three!


Access Distance Access Distance ImpactImpacton Route Choiceon Route Choice

Possible true Possible true trip originstrip origins

Zone Zone centroid for centroid for path-path-buildingbuilding

$

$

$

$

$

$

$

$

$$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$$

$$

$

$$

$$

$$

$

$

$ $

$

$$

$

$

$$

$$

$$

$

$$

$$

$$

$

$$

$

$$

$$

$$

$$

$$

$$ $ $ $

$

$

$

$

$

$

$

$

$

$

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$$ $

$

$

$$

$

$$

$

$

$

$

$

$

$ $

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

¼ Mile


$

$

$

$

$

$

$

$

$$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$$

$$

$

$$

$$

$$

$

$

$ $

$

$$

$

$

$$

$$

$$

$

$$

$$

$$

$

$$

$

$$

$$

$$

$$

$$

$$ $ $ $

$

$

$

$

$

$

$

$

$

$

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$$ $

$

$

$$

$

$$

$

$

$

$

$

$

$ $

$

$

$$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$

$ $

$

$

$

$

$

$

$

$

$

$

$

$

$

$

Example RTD Path Example RTD Path OptionsOptions

3 Reasonable Paths3 Reasonable Paths– Path 1:Path 1: 2 Local 2 Local

BusesBuses– Path 2:Path 2: 2 Local 2 Local

BusesBuses– Path 3: Local Bus, Rail, Path 3: Local Bus, Rail,

Mall Shuttle, Mall Shuttle, Local Local BusBus

Travel Behavior Travel Behavior Inventory (TBI) Inventory (TBI) observations for all observations for all three!three!

I-25 / Broadway StationI-25 / Broadway Station


I-25/Broadway I-25/Broadway TransfersTransfers


Transit Network Transit Network Testing: TypicalTesting: Typical

Route specific travel timesRoute specific travel times– Modeled versus observedModeled versus observed

Selected transit pathsSelected transit paths– Logical? (“Yep, that makes sense…”)Logical? (“Yep, that makes sense…”)

Boardings per linked tripBoardings per linked trip– Assignment of observed on-board survey tripsAssignment of observed on-board survey trips– Comparison of assigned to observed Comparison of assigned to observed

boardingsboardings By routeBy route By service typeBy service type By access mode (walk versus drive)By access mode (walk versus drive)


TBI Data –TBI Data –– Access and egress modeAccess and egress mode– Individual routes usedIndividual routes used

RTD systemRTD system– Reasonable options for pathsReasonable options for paths– Reasonable options for modesReasonable options for modes

Transit Network Transit Network Testing: OpportunitiesTesting: Opportunities


TBI Path-Matching TBI Path-Matching Experiments Experiments Reviewed selected individual Reviewed selected individual

reported pathsreported paths– Some logical paths not selectedSome logical paths not selected– Some multiple path optionsSome multiple path options– Some poor reporting by respondentsSome poor reporting by respondents


TBI Path-Matching TBI Path-Matching Experiments Experiments

Review of selected individual Review of selected individual reported pathsreported paths– Some logical paths not selectedSome logical paths not selected– Some multiple path optionsSome multiple path options– Some poor reporting by respondentsSome poor reporting by respondents

……IS VERY LABOR INTENSIVE!IS VERY LABOR INTENSIVE! Automated procedureAutomated procedure

– Prediction success tablesPrediction success tables


Transit NetworksTransit Networksfor Path-Buildingfor Path-Building 7 Networks:7 Networks:

Local Bus OnlyLocal Bus Only Local & Premium BusLocal & Premium BusPremium Bus OnlyPremium Bus Only Local Bus & RailLocal Bus & RailRail OnlyRail Only Premium Bus & Rail Premium Bus & Rail All ModesAll Modes

4 Times-of-Day:4 Times-of-Day:AM PeakAM Peak PM PeakPM Peak Off-PeakOff-Peak

Early/LateEarly/Late 2 Access Modes:2 Access Modes:

Walk AccessWalk Access Drive AccessDrive Access 56 Sets of Paths56 Sets of Paths


How Good Is the How Good Is the Complex MC-Simple Complex MC-Simple Path Approach?Path Approach? Prediction success testsPrediction success tests

– Built paths for observed Built paths for observed interchangesinterchanges Based on observed mode combinationBased on observed mode combination

– Local only, premium only, rail only…Local only, premium only, rail only…

– Compared:Compared: Modeled to observed boardingsModeled to observed boardings

– Interchange-by-interchange basisInterchange-by-interchange basis


Prediction Success Prediction Success Results Results PM Work Trip – Walk to Rail OnlyPM Work Trip – Walk to Rail Only

Skimmed Boardings

No Path 1 2 3+

No Path 0 0 0 0 1 7 3 4 0 2 1 0 0 0 TB

I R

epor

ted

Boa

rdin

gs

3+ 0 0 0 0


Prediction Success–Prediction Success–Complex MC-Simple Path Complex MC-Simple Path ApproachApproach

67 percent “correct”67 percent “correct” Unaffected by access modeUnaffected by access mode

Percent with Skimmed Boardings: Aggregation Level

Number of Linked

Trips = Reported Boardings

> Reported Boardings

< Reported Boardings

All Trips 1,278 67% 24% 9% Walk Access 854 67% 23% 9% Drive Access 424 67% 25% 7%


Prediction Success Prediction Success Results Results

AM Walk Access TripsAM Walk Access Trips

Complex MC-Simple Path vs. Complex MC-Simple Path vs. Simple MC-Complex PathSimple MC-Complex Path

Percent with Skimmed Boardings: Path-Building Approach

Number of Paths Found

Boardings / Linked

Trip = Reported Boardings

> Reported Boardings

< Reported Boardings

Observed 308 1.5 – – – Simple 290 1.6 66% 22% 12% Complex 302 1.9 52% 38% 10%

Complex ApproachComplex Approach– Observed trips assigned Observed trips assigned

to “All modes” pathsto “All modes” paths

Simple ApproachSimple Approach– As beforeAs before


Some Observations…Some Observations…

Transit usersTransit users– Pick individual pathsPick individual paths– Do not necessarily:Do not necessarily:

pick the same pathspick the same paths pick logical pathspick logical paths accurately report pathsaccurately report paths

Transit multi-path buildersTransit multi-path builders– Representation of discrete choiceRepresentation of discrete choice– Do not capture choice behaviorDo not capture choice behavior


Conclusions – For Conclusions – For DenverDenver Transit pathsTransit paths

– Are choice behaviorAre choice behavior– Should be represented as discrete Should be represented as discrete

choiceschoices– Require substantial resources to Require substantial resources to

model and estimatemodel and estimate


Conclusions – In Conclusions – In GeneralGeneral Common network validation measures Common network validation measures

that may that may notnot be sufficientbe sufficient– Ability to assign all observed tripsAbility to assign all observed trips– Matching observed boardings / linked tripMatching observed boardings / linked trip

More detailed validation is feasible More detailed validation is feasible (prediction success tables)(prediction success tables)– Well designed on-board survey is neededWell designed on-board survey is needed

Good origin and destination reportingGood origin and destination reporting Access and egress modeAccess and egress mode Boardings Boardings by modeby mode for reported trip for reported trip


Some FTA Some FTA ObservationsObservations

- People choose different paths I-J- People choose different paths I-J- Pathbuilders do fares badly- Pathbuilders do fares badly- Need 1- Need 1stst-board-location choice-board-location choice- Different choices, different - Different choices, different εε’s’s- Others ……- Others ……

- Nesting - Nesting ββ’s always asserted’s always asserted- Pathbuilder–MC consistency- Pathbuilder–MC consistency- Favoring paths - Favoring paths distortions distortions- Path choices defy discrete - Path choices defy discrete

labelslabels- Others ……- Others ……

Path-Types / Discrete Choice

Network/Pathbuilder

Response:Response: DATA DATA ANALYSIS ANALYSIS SPECIFICATIONS SPECIFICATIONS

And kudos to DRCOGAnd kudos to DRCOG


14. Telling a Good 14. Telling a Good StoryStory

FTA requirements for New StartsFTA requirements for New Starts Useful “Make the Case” Useful “Make the Case”

documentsdocuments Thoughts on good practiceThoughts on good practice Participant experiencesParticipant experiences An exampleAn example



Make-the-Case documentMake-the-Case document– GuideGuide to project benefits and to project benefits and

“justification”“justification” For FTA staffFor FTA staff For FTA briefing papers, talking pointsFor FTA briefing papers, talking points For the Annual Report on New StartsFor the Annual Report on New Starts

– Element of project “justification” ratingElement of project “justification” rating


A Useful DocumentA Useful Document

No more than “five pages”No more than “five pages”– Project identificationProject identification– SettingSetting– PurposePurpose– Current conditions in the corridorCurrent conditions in the corridor– Anticipated conditions in 2030Anticipated conditions in 2030– The case for the proposed projectThe case for the proposed project– RiskRisk– SummarySummary


Some Not-Useful Some Not-Useful Elements Elements Topics relevant elsewhere (not Topics relevant elsewhere (not

here)here)– History of project developmentHistory of project development– Detailed project descriptionDetailed project description– Financial feasibilityFinancial feasibility– Public support; other supportPublic support; other support– ““Importance”Importance”– PicturesPictures


Project IdentificationProject Identification

One or two sentencesOne or two sentences– Transit modeTransit mode– Starter line, expansion, or extensionStarter line, expansion, or extension– Length of projectLength of project– LocationLocation


SettingSetting

MapMap Key jurisdictions, activity centersKey jurisdictions, activity centers Any key geographical featuresAny key geographical features Major transportation facilitiesMajor transportation facilities


Purpose of the ProjectPurpose of the Project

TransportationTransportation– Whom is it intended to serve?Whom is it intended to serve?– From where to where?From where to where?

Economic development (if Economic development (if applicable)applicable)– Development locationsDevelopment locations– Role of the project – specific Role of the project – specific

mechanisms mechanisms


Current ConditionsCurrent Conditions

CurrentCurrent ≈ ≈ today (usually, today today (usually, today ≠ ≠ 2000)2000) ConditionsConditions relevant to project benefitsrelevant to project benefits

– Key travel markets (and recent growth?)Key travel markets (and recent growth?)– Congestion & highway travel timesCongestion & highway travel times– Transit services & transit travel timesTransit services & transit travel times– Transit ridership, emphasis on key marketsTransit ridership, emphasis on key markets


Conditions in 2030Conditions in 2030

Key Key changeschanges: today to 2030 (No : today to 2030 (No Build)Build)– Travel marketsTravel markets– Highway systemHighway system– Transit facilities, services, and travel Transit facilities, services, and travel

timestimes– Transit ridershipTransit ridership

Well linked to current conditionsWell linked to current conditions


Case for the ProjectCase for the Project

Low-cost approach (TSM)Low-cost approach (TSM)– Brief description of key TSM elementsBrief description of key TSM elements– Impact on transit service qualityImpact on transit service quality– Impact on transit ridershipImpact on transit ridership– Mobility benefits (time savings)Mobility benefits (time savings)– Cost-effectiveness versus No-BuildCost-effectiveness versus No-Build– Success in addressing the purpose(s)Success in addressing the purpose(s)



Proposed approach Proposed approach – Brief description of the projectBrief description of the project– Impact on transit service qualityImpact on transit service quality– Impact on transit ridership in key Impact on transit ridership in key

marketsmarkets– Mobility benefits (time savings)Mobility benefits (time savings)– Success in achieving the purpose(s)Success in achieving the purpose(s)– Cost-effectiveness versus TSMCost-effectiveness versus TSM


RiskRisk

Uncertainties in the costsUncertainties in the costs– Project scopeProject scope– Unit pricesUnit prices– Track recordTrack record

Uncertainties in the benefitsUncertainties in the benefits– Time savingsTime savings– Guideway ridershipGuideway ridership– Track recordTrack record


SummarySummary

One paragraph; one sentence per One paragraph; one sentence per topictopic

Essential elements of the caseEssential elements of the case– What is the purpose?What is the purpose?– How urgent is the problem?How urgent is the problem?– Why is a low-cost approach insufficient?Why is a low-cost approach insufficient?– How well does the project succeed?How well does the project succeed?– Are costs in scale with the benefits?Are costs in scale with the benefits?– How firm are the costs and benefits?How firm are the costs and benefits?


Thoughts on Good Thoughts on Good PracticePractice FocusFocus

– All discussion sections should help All discussion sections should help explain the benefits of the projectexplain the benefits of the project

– A strategyA strategy1.1. Figure out the principal benefits Figure out the principal benefits

(markets: geography, trip purposes, (markets: geography, trip purposes, etc.) that make the caseetc.) that make the case

2.2. Focus the introductory sections Focus the introductory sections (setting, current and future conditions) (setting, current and future conditions) on those marketson those markets


Thoughts on Good Thoughts on Good PracticePractice QuantificationQuantification

– Forecasts have numbers for Forecasts have numbers for ≈ ≈ everythingeverything

– Use them to avoid hand-waving.Use them to avoid hand-waving. ClarityClarity

– ““To write well is to think clearly. To write well is to think clearly. That’s why it’s so hard.” That’s why it’s so hard.” – David – David McCollough, 2003McCollough, 2003

– Assign someone who can do both.Assign someone who can do both.


Thoughts on Good Thoughts on Good PracticePractice ResourcesResources

– Basic summaries often not enoughBasic summaries often not enough– Subtask: extract information from Subtask: extract information from

forecastsforecasts– Preservation of resources for this workPreservation of resources for this work

FTA assistanceFTA assistance EthicsEthics

– Reliable numbers for decision-makingReliable numbers for decision-making– Bringing project benefits to the discussionBringing project benefits to the discussion


Participant Participant ExperiencesExperiences Attempts at Make-the-Case narrativesAttempts at Make-the-Case narratives Methods to find/correct errorsMethods to find/correct errors

– SummitSummit– Other tools/proceduresOther tools/procedures

Methods to better understand a Methods to better understand a projectproject– SummitSummit– Other tools/proceduresOther tools/procedures


Making the Case:Making the Case: An ExampleAn Example

Perris Valley Commuter Rail Perris Valley Commuter Rail ExtensionExtension– Riverside County Transportation Riverside County Transportation

Commission (California)Commission (California)


Perris ValleyPerris ValleyLineLineIdentificationIdentification

23-mile extension of 23-mile extension of the Metrolink the Metrolink commuter rail system commuter rail system from Riverside to from Riverside to communities in Perris communities in Perris Valley southeast of Valley southeast of RiversideRiverside


SettingSetting

City of RiversideCity of Riverside– 50 miles east of downtown LA50 miles east of downtown LA– 30 miles northeast of central Orange 30 miles northeast of central Orange

CountyCounty Perris Valley and I-215 to southeastPerris Valley and I-215 to southeast Moreno Valley and SR-60 to the eastMoreno Valley and SR-60 to the east Metrolink linesMetrolink lines

– Riverside Line to LA via PomonaRiverside Line to LA via Pomona– 91 Line to LA via Fullerton91 Line to LA via Fullerton– Inland Empire line to Orange County Inland Empire line to Orange County


Purpose of the ProjectPurpose of the Project

The Perris Valley extension will The Perris Valley extension will improve transit access to the improve transit access to the Metrolink system and the Metrolink system and the locations it serves for residents of locations it serves for residents of Perris and Moreno Valleys.Perris and Moreno Valleys.



DemographicsDemographics– 425,000 people and 123,000 jobs425,000 people and 123,000 jobs– One of the most rapidly growing counties One of the most rapidly growing counties

nationallynationally– Housing prices 25-35% less than in LA and OCHousing prices 25-35% less than in LA and OC

Long commutes and drive timesLong commutes and drive times– Riverside to LA CBD: 54 miles, 100 minutesRiverside to LA CBD: 54 miles, 100 minutes– Riverside to Orange: 35 miles in 76 minutes)Riverside to Orange: 35 miles in 76 minutes)



Key travel markets from Perris ValleyKey travel markets from Perris Valley– 18,000 workers to LA County18,000 workers to LA County– 30,000 workers to Orange County 30,000 workers to Orange County

Metrolink service from RiversideMetrolink service from Riverside– 37 trains per day on two lines to LA and one line to OC37 trains per day on two lines to LA and one line to OC– Focused on peak periods and commutersFocused on peak periods and commuters

Metrolink ridership: Riverside and adjacent stationsMetrolink ridership: Riverside and adjacent stations– 4,000 weekday trips total; 3,000 at Riverside station4,000 weekday trips total; 3,000 at Riverside station– 84% commuters; 65 % Perris Valley residents84% commuters; 65 % Perris Valley residents– 90 percent use auto access; 10 percent connector bus90 percent use auto access; 10 percent connector bus– Drive from South Perris to Riverside: 21 miles, 32 mins. Drive from South Perris to Riverside: 21 miles, 32 mins.



Rapid growth in Perris Valley Rapid growth in Perris Valley – +76% population to 600,000 people+76% population to 600,000 people– +115% employment to 210,000 jobs+115% employment to 210,000 jobs

Resulting growth in commuter marketsResulting growth in commuter markets– 24,000 workers to LA County (+33%)24,000 workers to LA County (+33%)– 46,000 workers to Orange Co. (+53%)46,000 workers to Orange Co. (+53%)

Consequent lengthening of peak Consequent lengthening of peak periods for auto travelperiods for auto travel



Large Metrolink changesLarge Metrolink changes– 126 trains per day (versus 37 per day currently)126 trains per day (versus 37 per day currently)– 16,300 trips per day using Riverside Co. 16,300 trips per day using Riverside Co.

stationsstations– 11,700 of these from Perris Valley11,700 of these from Perris Valley– Same commuter-oriented characteristicsSame commuter-oriented characteristics

More difficult drive-accessMore difficult drive-access– South Perris to Riverside, 21 milesSouth Perris to Riverside, 21 miles

32 minutes (39 mph) today32 minutes (39 mph) today 67 minutes (19 mph) in 203067 minutes (19 mph) in 2030



Low-cost alternativeLow-cost alternative– New express bus service to Riverside New express bus service to Riverside

stationstation– Additional park/ride facilitiesAdditional park/ride facilities– Mixed-traffic operationsMixed-traffic operations– An increase of 216 riders/day over No-An increase of 216 riders/day over No-

BuildBuild– Key limitation: long travel times because Key limitation: long travel times because

of congested highwaysof congested highways



Proposed projectProposed project– 23-mile commuter rail line23-mile commuter rail line– Six stations (5 park/ride with 1,800 Six stations (5 park/ride with 1,800

spaces)spaces)– Extension of the 91 line to downtown LAExtension of the 91 line to downtown LA

Travel times: Perris Valley to RiversideTravel times: Perris Valley to Riverside– 67 minutes by driving67 minutes by driving– 87 minutes by bus87 minutes by bus– 40 minutes by commuter rail40 minutes by commuter rail



Metrolink ridershipMetrolink ridership– 8,800 more weekday riders than in TSM8,800 more weekday riders than in TSM

User benefits: 3,100 hours/day savedUser benefits: 3,100 hours/day saved– 79% by commuters; 83% by PV residents79% by commuters; 83% by PV residents– Key markets – Perris Valley to:Key markets – Perris Valley to:

Orange County: Orange County: 1,000 hrs; 1,000 hrs; ≈≈18 min/trip18 min/trip Los Angeles: Los Angeles: 700 hrs; 700 hrs; ≈≈29 min/trip29 min/trip Riverside: Riverside: 400 hrs; 400 hrs; ≈22 min/trip≈22 min/trip



Cost effectivenessCost effectiveness– Capital: $180 million in 2007 dollarsCapital: $180 million in 2007 dollars– Added O&M cost: $1.5 million/yearAdded O&M cost: $1.5 million/year– Time savings: 850,000 hours/yearTime savings: 850,000 hours/year– $22.40 per hour of time savings$22.40 per hour of time savings– Competitive for federal fundingCompetitive for federal funding


Risks (Some Thoughts)Risks (Some Thoughts)

Ridership and transportation benefitsRidership and transportation benefits– Sources of risk?Sources of risk?

Very high growth projectionsVery high growth projections Very large congestion increasesVery large congestion increases Very large Metrolink service increases (NB)Very large Metrolink service increases (NB)

– Aspects that help contain riskAspects that help contain risk Existing Metrolink ridership from Perris Existing Metrolink ridership from Perris

ValleyValley Large Metrolink system, ridership, DATALarge Metrolink system, ridership, DATA

Costs: from formal risk analysisCosts: from formal risk analysis


SummarySummary

Rapid growthRapid growth Long-distance commutesLong-distance commutes Difficult access to Metrolink systemDifficult access to Metrolink system Large time savings (total and per Large time savings (total and per

rider)rider) Low capital costLow capital cost Costs in scale with the benefitsCosts in scale with the benefits


15. Economic 15. Economic DevelopmentDevelopment

SAFETEA-LU New Starts SAFETEA-LU New Starts requirementsrequirements

FTA thoughts, activitiesFTA thoughts, activities Discussion / ideasDiscussion / ideas


RequirementsRequirements

SAFETEA-LU: Evaluate projects SAFETEA-LU: Evaluate projects on:on:

“… “… a comprehensive review of its … a comprehensive review of its … economic development effects, and public economic development effects, and public transportation supportive land use policies transportation supportive land use policies and future patterns.”and future patterns.”


FTA ThoughtsFTA Thoughts

ChallengesChallenges– Land use versus economic developmentLand use versus economic development

Need clearly distinguished definitions, measuresNeed clearly distinguished definitions, measures So:So:

– Land use = attributes of the project settingLand use = attributes of the project setting– Econ-dev = changes because of the projectEcon-dev = changes because of the project



Challenges Challenges (continued)(continued)

– User benefits (UBs) versus User benefits (UBs) versus economic development benefits (EDBs)economic development benefits (EDBs)

Need to avoid double-counting Need to avoid double-counting mobility/accessibilitymobility/accessibility

So: looking for clear evidence that a measurable So: looking for clear evidence that a measurable portion of economic development impacts are portion of economic development impacts are separable and independent of user benefitsseparable and independent of user benefits




– Demonstrated impactsDemonstrated impacts Need to have analytical basis for EDBsNeed to have analytical basis for EDBs So:So:

– Literature reviewLiterature review– Apparently sparse evidence that transit station Apparently sparse evidence that transit station

proximity, by itself, has consistent impacts on land proximity, by itself, has consistent impacts on land prices (and by extension, development benefits)prices (and by extension, development benefits)

– Few existing studies distinguish the impacts of the Few existing studies distinguish the impacts of the project from the impacts of zoning changes, project from the impacts of zoning changes, development incentives, and other policies that development incentives, and other policies that affect developmentaffect development




– Useful measureUseful measure Need a measure of EDBs that provides a reasonable Need a measure of EDBs that provides a reasonable

accounting of benefits and disbenefitsaccounting of benefits and disbenefits So: concerns on “trip not taken” measurementSo: concerns on “trip not taken” measurement

– Location choice = f( travel costs, schools, amenities … )Location choice = f( travel costs, schools, amenities … )– So, different choices So, different choices different bundles of attributes different bundles of attributes– Relocation to location with lower travel costs cannot be Relocation to location with lower travel costs cannot be

evaluated solely on the basis of reduction in travel costsevaluated solely on the basis of reduction in travel costs– Direct parallel to evaluating mode-shift benefits using a Direct parallel to evaluating mode-shift benefits using a

strict accounting of “time savings”strict accounting of “time savings”




– Predictive toolsPredictive tools Need method for predicting development impacts Need method for predicting development impacts

and EDBs for individual projects in individual and EDBs for individual projects in individual contextscontexts

So: FTA will be evaluating existing predictive toolsSo: FTA will be evaluating existing predictive tools– Residential-location choice modelsResidential-location choice models– Workplace/employer-location choice modelsWorkplace/employer-location choice models– Others? Others?



NPRMNPRM– Evaluate presence of EDB-supportive conditionsEvaluate presence of EDB-supportive conditions

Opportunity: availability of land for (re)developmentOpportunity: availability of land for (re)development Market conditions: regional and corridor activityMarket conditions: regional and corridor activity Supporting policies: zoning, tax, & otherSupporting policies: zoning, tax, & other Accessibility impacts: consequence of the projectAccessibility impacts: consequence of the project Permanence: characteristics of the projectPermanence: characteristics of the project

– Premise: favorable conditions Premise: favorable conditions large EDBs large EDBs– Part of the measure of project Part of the measure of project effectivenesseffectiveness– Continued standard allowance in Continued standard allowance in cost-cost-

effectivenesseffectiveness



““Measures” documentMeasures” document– Rely on location choice models for Rely on location choice models for

predictions and measures of benefitspredictions and measures of benefits– Possible advantagesPossible advantages

Project-specific quantification of EDBsProject-specific quantification of EDBs Possible inclusion in cost-effectiveness Possible inclusion in cost-effectiveness

calculationscalculations Probability that some projects are “above Probability that some projects are “above

average” in that they have more EDBs than they average” in that they have more EDBs than they get from the standard allowance (implications for get from the standard allowance (implications for others?)others?)


FTA ActivitiesFTA Activities

NPRMNPRM– Receipt of formal comments; then … ?Receipt of formal comments; then … ?

FTA-sponsored applied researchFTA-sponsored applied research– Literature review (Literature review ( FTA website) FTA website)– Kick-off: meeting of expert panel Kick-off: meeting of expert panel

10/200710/2007– Development of predictive tool(s)Development of predictive tool(s)

Ideas from travel forecasters?Ideas from travel forecasters?


16. Wrap-Up16. Wrap-Up

Additional comments by Additional comments by participantsparticipants

FTA to-do listFTA to-do list FTA objectives for travel FTA objectives for travel

forecasting in support of New forecasting in support of New StartsStarts


Additional Comments Additional Comments


FTA To-do ListFTA To-do List

Research?Research? Written guidance?Written guidance? Training?Training? Future workshop?Future workshop? Other?Other?


FTA ObjectivesFTA Objectives

Travel forecasting for New StartsTravel forecasting for New Starts– Sufficient data to inform technical workSufficient data to inform technical work– Meaningful testing of travel modelsMeaningful testing of travel models– Adequate QC Adequate QC and analysisand analysis of forecasts of forecasts

– Understanding of project benefitsUnderstanding of project benefits

Documents

September 2007Travel Forecasting for New Starts3-1 10. Uncertainty Analysis FTA requirements for New Starts FTA requirements for New Starts Implementation