1 The Aggregate Rail Ridership Forecasting Model: Overview Dave Schmitt, AICP Southeast Florida Users Group November 14 th 2008.

1

The Aggregate The Aggregate Rail Ridership Rail Ridership

Forecasting Forecasting Model: OverviewModel: Overview

Dave Schmitt, AICPDave Schmitt, AICP

Southeast Florida Users GroupSoutheast Florida Users GroupNovember 14November 14thth 2008 2008

2

What is It?What is It?

A sketch-planning tool consisting of CTPP A sketch-planning tool consisting of CTPP 2000 data, GIS info, programs, control 2000 data, GIS info, programs, control files, and a spreadsheet collectively used to files, and a spreadsheet collectively used to develop an estimate of the ridership develop an estimate of the ridership potential for a new rail systempotential for a new rail system

Based on 20 recently-built light and Based on 20 recently-built light and commuter rail projectscommuter rail projects

Two spreadsheets: light rail and commuter Two spreadsheets: light rail and commuter rail; all other materials are identicalrail; all other materials are identical

Sponsored by FTA; developed by AECOMSponsored by FTA; developed by AECOM

3

CTPP 2000 DataCTPP 2000 DataPart 1 – Workers at home-endPart 1 – Workers at home-end

Part 2 – Workers at work-endPart 2 – Workers at work-end

Part 3 – Flows Part 3 – Flows

CTPP1INC_TZ.exe,CTPP1INC_TZ.exe,

CTPP1INC_BG.exe, CTPP1INC_BG.exe, andand

CTPP1INC_TR.exe CTPP1INC_TR.exe programsprograms

Calculates proportion of households in Calculates proportion of households in low, medium and high income low, medium and high income categories by geographic unitcategories by geographic unit

CTPP2EMP_TZ.exe,CTPP2EMP_TZ.exe,

CTPP2EMP_BG.exe, CTPP2EMP_BG.exe, andand

CTPP2EMP_TR.exe CTPP2EMP_TR.exe programsprograms

Calculates workers in each geographic Calculates workers in each geographic unit and estimates employment densityunit and estimates employment density

CTPP3.exeCTPP3.exe program program Helps to extract tract-level data from Helps to extract tract-level data from region- or state-wide filesregion- or state-wide files

GIS infoGIS infoRail station points;Rail station points;

Proportion of tracts/zones within range Proportion of tracts/zones within range of stationsof stations

RailMarket.exeRailMarket.exe program program

Calculates the number of workers who Calculates the number of workers who both live and work within particular both live and work within particular distances of a rail station by income distances of a rail station by income group and employment density categorygroup and employment density category

SpreadsheetSpreadsheetRecords service variables and Records service variables and RailMarketRailMarket results; produces ridership results; produces ridership potential estimatepotential estimate

4

LRT Model EquationLRT Model Equation

Total Weekday Unlinked Rail Trips =Total Weekday Unlinked Rail Trips =

Weekday Unlinked Drive Access to Work Rail Trips + Weekday Unlinked Drive Access to Work Rail Trips +

Weekday Unlinked Other Rail TripsWeekday Unlinked Other Rail Trips

Weekday Unlinked Drive Access to Work Rail Trips =Weekday Unlinked Drive Access to Work Rail Trips =

0.030 * CTPP PNR 6 -to-1 Mile JTW Flows (<50K Den) +0.030 * CTPP PNR 6 -to-1 Mile JTW Flows (<50K Den) +

0.202 * CTPP PNR 6 -to-1 Mile JTW Flows (>50K Den)0.202 * CTPP PNR 6 -to-1 Mile JTW Flows (>50K Den)

Weekday Unlinked Other (Non-Drive Access to Work) Rail Trips =Weekday Unlinked Other (Non-Drive Access to Work) Rail Trips =0.395 * CTPP 2 -to-1 Mile JTW Flows (<50K Den) +0.395 * CTPP 2 -to-1 Mile JTW Flows (<50K Den) +

0.445 * CTPP 2 -to-1 Mile JTW Flows (>50K Den)0.445 * CTPP 2 -to-1 Mile JTW Flows (>50K Den)

5

CR Model EquationCR Model EquationCommuter Rail Weekday Unlinked Trips = Commuter Rail Weekday Unlinked Trips =

Nominal Ridership x Demand Adjustment FactorNominal Ridership x Demand Adjustment Factor

Nominal Ridership = Nominal Ridership =

0.069*High Income CTPP PNR 6-to-1 JTW flows + 0.069*High Income CTPP PNR 6-to-1 JTW flows +

0.041*Medium Income CTPP PNR 6-to-1 JTW flows + 0.041*Medium Income CTPP PNR 6-to-1 JTW flows +

0.151*Low Income CTPP 2-to-1 JTW flows0.151*Low Income CTPP 2-to-1 JTW flows

Demand Adjustment Factor=Demand Adjustment Factor= (1+0.3*Percent Deviation in (1+0.3*Percent Deviation in Average System Speed) xAverage System Speed) x

(1+0.3*Percent Deviation in Train Miles per Mile) (1+0.3*Percent Deviation in Train Miles per Mile) x x

Rail Connection IndexRail Connection Index

6

CR Model Equation (2)CR Model Equation (2)Percent Deviation in Average System Speed=Percent Deviation in Average System Speed=

System Average Speed-35.7 mph / [ System Average System Average Speed-35.7 mph / [ System Average Speed+35.7)/2]Speed+35.7)/2]

System Average Speed=System Average Speed=

Annual Revenue Vehicle Miles/Annual Revenue Vehicle Annual Revenue Vehicle Miles/Annual Revenue Vehicle HoursHours

Percent Deviation in Train Miles per Mile=Percent Deviation in Train Miles per Mile=

Weekday Train Miles per Directional Route Mile-10.3 / Weekday Train Miles per Directional Route Mile-10.3 /

[(Weekday Train Miles per Directional Route Mile+10.3)/2][(Weekday Train Miles per Directional Route Mile+10.3)/2]

Weekday Train Miles per Directional Route Mile=Weekday Train Miles per Directional Route Mile=

Annual Revenue Vehicle Miles/250/Average Train LengthAnnual Revenue Vehicle Miles/250/Average Train Length

7

ApplicationsApplications

8

Applications – City AApplications – City A New rail line between CBD and New rail line between CBD and

suburban activity centers; strong suburban activity centers; strong corridor bus ridership & servicecorridor bus ridership & service

Compared ARRF LRT model with travel Compared ARRF LRT model with travel demand model resultsdemand model results

ResultsResults ARRF LRT model results were 100% higher ARRF LRT model results were 100% higher

than travel demand model estimatesthan travel demand model estimates Stronger motivation to investigate transit Stronger motivation to investigate transit

model parameters; subsequently identified model parameters; subsequently identified issues with walk- and auto-access issues with walk- and auto-access connector methodologyconnector methodology

9

Applications – City A Applications – City A (cont’d)(cont’d)

ConclusionsConclusions ARRF model may partially explain ARRF model may partially explain

attractiveness of rail over existing attractiveness of rail over existing bus servicebus service

TDM path-builder probably better TDM path-builder probably better at evaluating bus/rail competition:at evaluating bus/rail competition: Equal service levels for bus & railEqual service levels for bus & rail Buses are just as close or closer to Buses are just as close or closer to

corridor activity centerscorridor activity centers

10

Applications – City BApplications – City B

New rail line between CBD and New rail line between CBD and suburban residential areassuburban residential areas

Used ARRF to develop rationale Used ARRF to develop rationale for alternative-specific constant for alternative-specific constant

Results on next slide…Results on next slide…

11

Ridership Forecasts – City BRidership Forecasts – City B

WalkWalk Drive/ Drive/ Drop-OffDrop-Off TotalTotal

ARRFARRF 14,79414,794 6,5486,548 21,34221,342

TDM ModelTDM Model

(no bias)(no bias)11,52011,520 4,5564,556 16,07616,076

TDM ModelTDM Model

(7.5 minute walk, 15 (7.5 minute walk, 15 minute drive)minute drive)

13,14513,145 6,3416,341 19,48719,487

TDM ModelTDM Model

(10 minute walk, 15 (10 minute walk, 15 minute drive)minute drive)

14,77014,770 6,2776,277 21,04721,047

12

Applications – City CApplications – City C Streetcar in low density urban Streetcar in low density urban

activity center; existing service activity center; existing service is local & primarily captive is local & primarily captive marketmarket

ARRF LRT model compared with ARRF LRT model compared with travel demand model (2000 trip travel demand model (2000 trip tables, 2030 networks)tables, 2030 networks)

13

Applications – City C Applications – City C (cont’d)(cont’d)

ResultResult Aggregate model forecast 120% higher Aggregate model forecast 120% higher

than travel demand modelthan travel demand model

ConclusionConclusion ARRF model may partially explain ARRF model may partially explain

attractiveness of rail over existing service, attractiveness of rail over existing service, but does not well-represent benefits of but does not well-represent benefits of project since:project since:

The project mode is different than calibrated The project mode is different than calibrated modemode

Lack of choice market not consistent with LRT Lack of choice market not consistent with LRT sample citiessample cities

14

Applications – City DApplications – City D Commuter rail between two Commuter rail between two

adjacent metropolitan areas; adjacent metropolitan areas; some express bus service to each some express bus service to each CBD, but no service between CBD, but no service between CBD’sCBD’s

Commuter rail ARRF model Commuter rail ARRF model compared with travel demand compared with travel demand model (2000 trip tables, 2030 model (2000 trip tables, 2030 networks) applied to networks) applied to eacheach CBD CBD

15

Applications – City D Applications – City D (cont’d)(cont’d)

ResultResult Aggregate model forecast 130% higher Aggregate model forecast 130% higher

than travel demand modelthan travel demand model

ConclusionConclusion ARRF model may partially explain ARRF model may partially explain

attractiveness of rail over existing attractiveness of rail over existing commuter bus service, but does not commuter bus service, but does not well-represent benefits of project since well-represent benefits of project since lack of service between CBDs unlike CR lack of service between CBDs unlike CR sample citiessample cities

16

Applications – City EApplications – City E New commuter rail line to high New commuter rail line to high

mode share CBD with established mode share CBD with established “choice market” commuter bus “choice market” commuter bus service from large park and ride service from large park and ride facilitiesfacilities

Commuter rail ARRF model Commuter rail ARRF model compared with travel demand model compared with travel demand model (2000 trip tables, 2030 networks) (2000 trip tables, 2030 networks) appliedapplied

17

Applications – City E Applications – City E (cont’d)(cont’d)

ResultResult Aggregate model forecast 30% Aggregate model forecast 30%

lower than travel demand modellower than travel demand model

ConclusionConclusion Existing commuter (“choice”) Existing commuter (“choice”)

market in corridor stronger than market in corridor stronger than CR sample citiesCR sample cities

18

ProcessProcess

19

General ProcedureGeneral Procedure1.1. Obtain basic input filesObtain basic input files

2.2. Determine the socio-economic Determine the socio-economic characteristics of the geographycharacteristics of the geography

3.3. Prepare the CTPP Part 3 flow dataPrepare the CTPP Part 3 flow data

4.4. Determine the relationships between rail Determine the relationships between rail stations & geographystations & geography

5.5. Run Run RailMarketRailMarket program to determine the program to determine the number of work for both live & work number of work for both live & work nearby rail stationsnearby rail stations

6.6. Enter the information from Enter the information from RailMarketRailMarket into the model spreadsheetinto the model spreadsheet

20

Station Buffers Station Buffers

21

SpreadsheetsSpreadsheetsProject:Alternative:Date:

Input Data

1. System Operational Characteristics 1a. Annual Revenue Vehicle Miles 1b. Annual Revenue Vehicle Hours 1c. Average Consist Length 1d. Weighted Operating Days per Year 1e. Directional Route Miles 1f. Rail Connection Index (1 if connects to urban rail distributor to CBD, else .5)

2. CTPP Flows 2a. Home within 2 miles of any station and Work within 1 mile of any station 2.a.i Low Income

2b. Home within 6 miles of a PNR station and Work within 1 mile of any station 2.b.i Medium Income 2.b.ii High Income

Parameters

1. Calibration Average system speed 35.700 2. Demand elasticity with respect to speed 0.300 3. Calibration Average Weekday Train Miles per Directional Route Mile 10.300 4. Demand elasticity with respect to Train Miles per Directional Route Mile 0.300

5. Unlinked Other (non-Drive Access to Work) low income/CTPP Flow 0.151 6. Unlinked Drive Access to Work medium income rail trips/CTPP Flow 0.041 7. Unlinked Drive Access to Work high income rail trips/CTPP Flow 0.069

Demand Adjustment Factor

Computed Speed 0.0Adjustment for Speed 0.40 Computed Weekday Train Miles Per Dir. Rte Mile 0.0Adjustment for Train Miles Per Directional Route Mile 0.40 Adjustment for Rail Connection - Total Demand Adjustment Factor -

Commuter Rail Unlinked Trips

Nominal Daily Unlinked Other (non-Drive Access to Work) low density rail trips - Nominal Daily Unlinked Drive Access to Work low density rail trips - Nominal Daily Unlinked Drive Access to Work high density rail trips -

Subtotal Nominal Daily rail trips - Demand Adjustment Factor - Total Daily Unlinked Commuter Rail Trips -

Project:

Alternative:

Date:

Input Data

1. Rail Line Directional Route Miles 68.8

2. CTPP Flows 2a. Home within 2 miles of any station and Work within 1 mile of any station 2.a.i Destination has <50,000 work trip ends per square mile 33,523 2.a.ii Destination has >50,000 work trip ends per square mile 16,292

Total (check) 49,815 2b. Home within 6 miles of a PNR station and Work within 1 mile of any station 2.b.i Destination has <50,000 work trip ends per square mile 69,298 2.b.ii Destination has >50,000 work trip ends per square mile 40,057

Total(check) 109,355

Parameters

1. Unlinked trips per mile 772.000 2. Unlinked Other (non-Drive Access to Work) low density rail trips/CTPP Flow 0.395 3. Unlinked Other (non-Drive Access to Work) high density rail trips/CTPP Flow 0.449 4. Unlinked Drive Access to Work low density rail trips/CTPP Flow 0.030 5. Unlinked Drive Access to Work high density rail trips/CTPP Flow 0.202

Distance-Based Model

Daily Unlinked Rail Trips 53,114

CTPP-Based Model

Daily Unlinked Other (non-Drive Access to Work) low density rail trips 13,242 Daily Unlinked Other (non-Drive Access to Work) high density rail trips 7,315

Subtotal Daily Unlinked Other (non-Drive Access to Work) rail trips 20,557

Daily Unlinked Drive Access to Work low density rail trips 2,079 Daily Unlinked Drive Access to Work high density rail trips 8,092

Subtotal Daily Unlinked Drive Access to Work rail trips 10,170

Total Daily Unlinked rail trips 30,727

St Louis LRT

Year 2000 Calibration Run

1/25/2006

22

Materials Available Materials Available from FTAfrom FTA

Detailed documentationDetailed documentation Part-1: Model Application GuidePart-1: Model Application Guide Part-2: Input Data Development Part-2: Input Data Development

GuideGuide Part-3: Model Calibration ReportPart-3: Model Calibration Report

CTPP CTPP and and RailMarketRailMarket programs programs Spreadsheets: LRT and CRSpreadsheets: LRT and CR Contact: Contact: [email protected]@dot.gov

23

Thank you!Thank you!