Travel Demand Methodology Report. Durham Orange Light Rail Transit Project

Transcription

1 Durham Orange Light Rail Transit Project September 2012

2 TABLE OF CONTENTS 1. Introduction STUDY BACKGROUND RECENT PROJECT AND STUDIES CURRENT WORK PROGRAM Preliminary Review of the Triangle Regional Model REGIONAL MODELING TOOLS GEOGRAPHIC AREA COVERAGE TRAVEL MARKETS Review of the Mode Choice Model and Transit Elements STRUCTURE COEFFICIENTS Coefficient of In Vehicle Time (Civtt) Coefficient of Out of Vehicle Time (Covtt) Implicit Value of Time (VOT=Civtt/Ccost) Alternative Specific Constants TRANSIT SYSTEM Networks Technology Representation Model Enhancements, Calibration, and Validation Alternative Modeling Approaches AGGREGATE FORECASTING USING ARRF II CHOICE SET APPROACH Uncertainty Analysis References Durham-Orange Light Rail Transit Project September 2012 i

3 LIST OF TABLES Table 2.1 Travel Patterns by County (2006 Household Travel Survey) Table 2.2 Transit Travel Patterns by County (2006 Transit On Board Survey) Table 2.3 Trip Purpose by Transit Operator Table 3.1 Coefficients of Explanatory Variables in the Model Table 4.1 Pre Paid Ridership by Route in Durham Orange Light Rail Corridor Table 4.2 Chapel Hill Transit Ridership Before and After Implementing Free Fare Table 4.3 Triangle Transit Ridership Increases Following Fare and Service Policy Changes Table 4.4 Unmeasured Attributes of Guideway Transit Table 4.5 Proposed In Vehicle Travel Time Related Alternative Specific Effects Table 4.6 Assigned versus Observed Transit Ridership Table 4.7 Transit Trips by Trip Purposes (2005) Table 4.8 Observed (2006) and Estimated (2005) Mode Shares of Motorized Travel Table 4.9 Observed (2006) and Estimated (2005) Access Mode Shares of Transit Trips Table 5.1 ARRF II Assumptions and Ridership Estimates Table 5.2 Choice Set Approach: Revised Calibration Framework Table 5.3 Percentages of Trips Now Considering Local Bus Transit as an Option Table 6.1 Potential Forecasting Uncertainties with Triangle Regional Model Table 6.2 Potential Forecasting Uncertainties with ARRF II Model Table 6.3 Mode Choice Model Constants (HBW, HBSh, HBU) Table 6.4 Mode Choice Model Constants (HBSc, HBO, NHB) LIST OF FIGURES Figure 1.1 The Triangle Region Figure 2.1 TRM4E.1 Structure Figure 2.2 Regional Analysis Districts Figure 3.1 Mode Choice Model Structure Durham-Orange Light Rail Transit Project September 2012 ii

4 1. Introduction This report documents the assumptions and methodology for performing travel demand forecasting for the Triangle Regional Transit Program Alternatives Analyses (AA) and the Durham Chapel Hill Corridor Light Rail Transit (LRT) New Starts application. The report was prepared by Triangle Transit in coordination with the Capital Area Metropolitan Planning Organization (CAMPO), the Durham Chapel Hill Carrboro (DCHC) MPO, and the Triangle J Council of Governments. The AA process covered multiple corridors in the Triangle Region to help prepare one or more projects for entry into the Preliminary Engineering phase as part of the Federal Transit Administration (FTA) New Starts process. As recommended by the FTA, this methodology report documents the methodology and assumptions used to develop the AA ridership forecasts and the refinements incorporated into the analysis for supporting the New Starts forecasting for the relevant locally preferred alternative (LPA). This report is intended to present a clear record of the process used and to ensure that all parties are in agreement. Chapter 2 presents an overview of the Triangle Regional Transit Program (TRTP) and the regional travel demand forecasting model. Chapter 3 discusses the mode choice portion of the model in greater detail, focusing on how the model differs from FTA recommendations and standards. Chapter 4 covers model enhancements, calibration, and validation. Chapter 5 describes alternative approaches to modeling ridership that were investigated in consultation with the FTA Study Background The Triangle Region has a polycentric urban pattern that includes several sizeable downtowns, four major universities, three major medical centers, and many satellite communities, with travel and economic patterns that link them to the Region s core counties of Durham, Orange (D O), and Wake (see Figure 1.1 The Triangle Region ). Raleigh Durham International Airport (RDU) and the major economic engine, the Research Triangle Park (RTP), draw traffic to the center of the region. The TRTP is a collaborative framework for developing an efficient and sustainable regional transportation system that addresses the Triangle s critical need for improved connectivity and mobility choices while promoting its economic prosperity, job growth, and an enhanced quality of life. The following agencies currently provide transit service in the Triangle Region: Capital Area Transit (CAT) Chapel Hill Transit (CHT) C Tran (Cary Transit) Durham Area Transit Authority (DATA) Duke University Transit North Carolina State University (NCSU) Transportation (WolfLine) Triangle Transit In 2006, a transit on board survey found that approximately 80,000 transit trips occurred per day in the Triangle Region (slightly less than 1.5 percent of total travel). Durham-Orange Light Rail Transit Project September

5 Figure 1.1 The Triangle Region 1.2. Recent Project and Studies Planning for fixed guideway transit in the Triangle Region began 20 years ago, and a number of transit studies have been conducted to advance major transit investments in the area. The most advanced planning to date was the Regional Rail Project. In 1998, FTA authorized Preliminary Engineering and preparation of an Environmental Impact Statement (EIS) for the Regional Rail Project. The Final EIS and Preliminary Engineering were concluded in 2003, with the FTA issuing the environmental Record of Decision, and subsequently authorizing Triangle Transit to enter the Final Design phase in February Midway through the Final Design phase, construction cost estimates dramatically increased and ridership forecasts did not meet the FTA criteria for funding at this new cost. As a result, the project was scaled back in an attempt to maintain the desired cost effectiveness level. After numerous attempts to right size the project, it became apparent that the project would not qualify for a Federal Full Funding Grant Agreement (FFGA), and that sufficient local funding was not available to supplement the project. Based on the project s inability to meet FTA s revised project cost effectiveness and issues with the overall financial plan, Triangle Transit withdrew the Regional Rail Project from the FTA New Starts program in August Durham-Orange Light Rail Transit Project September

6 Since the withdrawal of the Regional Rail Project from the FTA New Starts program, the region has undertaken several initiatives to restructure its Regional Transit Plan, resulting in the following key changes: Between 2007 and 2009, system wide planning for future fixed guideway transit corridors was conducted through a cooperative regional planning effort. Recommendations from this planning process were adopted as the transit element of the Triangle Region s 2035 Long Range Transportation Plan (LRTP). In 2009, House Bill 148 was passed to enable local funding for transit projects through a 1/2 percent sales tax contribution, subject to referenda. Currently, all fixed guideway transit corridors in the 2035 Long Range Transportation Plan are being analyzed to determine implementation priorities. Also at this time, Triangle Transit is coordinating with other agencies pursuing commuter rail and high speed rail initiatives within the region. These changes and developments were described in more detail in the Project Introduction document circulated and presented to the Federal Transit Administration on July 19, Current Work Program Additional planning work to implement the 2035 LRTP was conducted in two phases: Phase 1 Transitional Analysis (Complete) Triangle Transit conducted a system level Transitional Analysis (TA) to define and prioritize transit corridors in the LRTP to carry forward into the AA process. Given the geography of the region, including three counties and two MPOs, one corridor within each of the MPO areas was carried forward into the AA process. A third commuter rail corridor spanning the Region is possible as an early implementation project, with potential collaboration with the North Carolina Railroad. The TA identified the priority corridors to be included in the AAs based on criteria that included mobility, socioeconomics, land use, and financial measures. The TA acted as the bridge between the system plan in the LRTP and the more focused and detailed analysis performed in each AA. Phase 2 Alternatives Analysis (Complete) Triangle Transit conducted AAs for the three priority corridors in the TA to evaluate and screen alternative alignments, station locations, and technologies. Evaluation criteria, including ridership forecasts, were developed and used to select the LPAs to be adopted by the appropriate MPO. The study design permitted one or more LPAs to advance to the next implementation phase of preliminary design and the required environmental impact analysis. At this time, a single LPA in the Durham Orange Corridor is being advanced for submission through the New Starts process. Durham-Orange Light Rail Transit Project September

7 2. Preliminary Review of the Triangle Regional Model The Triangle Regional Model Service Bureau (TRMSB), in cooperation with regional stakeholders (CAMPO, DCHC MPO, North Carolina Department of Transportation, and the Triangle Transit), performs travel modeling for the Region. TRMSB is housed at the NCSU Institute for Transportation Research and Education Regional Modeling Tools The adopted official Triangle Regional Model (TRM) at the start of this effort, Version 4, operates in TransCAD 4.8 Build 540, with a base year of 2005 and horizon year of This model is an enhanced and refined version of the regional model developed for the Triangle Transit s New Starts project in 2006 (TRM2006). With a base year of 2002, the TRM2006 model was estimated and calibrated using the 1995 Home Interview Survey and 2000 Census Transportation Planning Package data. During the evaluation of the New Starts forecasts, FTA identified several issues, including the overestimation of park and ride transit trips to central suburban destinations, such as RTP. TRMSB is in the process of developing a new version of the TRM, Version 5, which also includes expanded geographic coverage. This model development process has been performed in close coordination with the Non Motorized Model Development project sponsored by DCHC MPO and conducted by Cambridge Systematics (CS). The version of the model resulting from the DCHC MPO project (hereinafter referred to as TRM Version 4 Enhanced or TRM4E.1 ) included re estimated and calibrated trip generation and destination choice models using the latest household travel survey and enhanced treatment for non motorized trips. Figure 2.1 shows the TRM4E.1 structure. Figure 2.1 TRM4E.1 Structure Trip Generation Trip Distribution Motorized/ Non Motorized Split Binary Log Sum Mode Choice Assignment Durham-Orange Light Rail Transit Project September

8 TRM Version 5 incorporates the enhancements that were developed as part of TRM4E.1. The model development and validation work for TRM5 is continuing. There was some uncertainty as to the schedule of delivery, given the ambitious level of work that was being performed in this revision and the need for assembly of base and future year socioeconomic data for the expanded geography. Based on the schedule for the AA and New Starts application(s), the timeliness of data used, the status of Version 5, and the resources available, the study team elected to use TRM4E.1 as the base model for the AA and New Starts submittal. Any further project specific enhancements that have been made to the model were then made to this version, the final version of which is called TRM4E.2. The Version 4 Enhanced models (TRM4E.1 and TRM4E.2) include the following major features: A multinomial logit based trip generation model with six trip purposes (home based work [HBW], home based shopping [HBSh], home based school [HBSc], home based other[hbo], home based university [HBU], and non home based [NHB]) A multinomial logit based destination choice model A nested logit based mode choice model A multimodal multiclass user equilibrium highway assignment and pathfinder transit assignment, with an iterative feedback mechanism Three time periods (morning peak, evening peak, and off peak) for highway assignment (for transit assignment, peak and off peak levels of service and assignments are used) Model estimation and calibration using the 2006 household travel survey and 2006 transit onboard survey data Generation of total person trip productions and attractions carried through trip distribution (prior to the mode choice step, a binary choice model (developed using the 2006 household travel survey and the 2006 transit on board survey) is used to split non motorized and motorized trips) Incorporation into the destination choice model of the Logsum estimated from the binary nonmotorized split model 2.2. Geographic Area Coverage The TRM Version 4 Enhanced models cover the Triangle Region, which includes CAMPO, DCHC MPO, and portions of several surrounding counties including Chatham, Harnett, Johnston, Franklin, and Granville. Only Durham, Orange, and Wake counties are included completely within the model region; only a portion of the other five counties are within the model region. The major geographic divisions include Transportation Analysis Zones (TAZ), model analysis districts, and counties. Durham-Orange Light Rail Transit Project September

9 Traa vel Demand Method ology Report Figure 2.2 Regional Analysis Districts Based on the 2009 estimates from the U.S. Census Bureau, the major centers in the Triangle Region are: the City of Raleigh, with a population of approximately 400,,000; the City of Durham, with a population of approximately 230, 000; the Town of Cary, with an estimated population of approximately 140,000; and the Town of Chapel Hill, with an estimated population off over 50,000. As of 2005, the model region covered a population of 1.3 million and approximately 680,000 jobs. By 2035, both the number of residents and the number of jobss in the region are expected to double, with 50 percent of the population growth and 70 percent of thee employment growth occurring in the rail corridors a currently being studied Travel Markets Three projects were considered through the AAs. The first project represented a potential transit corridor within Wake County. The second project represented a transit corridor between Durham and Orange counties. The third project was a connecting project that connects the two previous projects between Raleigh and Durham. a Broadly defined to include any modeling district containing the proposed alignments. Durham-Orange Light Rail Transit Project September

10 The regional travel patterns can be characterized in the following terms: Three primary counties Wake, Durham, and Orange Three primary urban districts Chapel Hill, Durham, and Raleigh Three primary trip purposes HBU, HBW, and NHB Regional travel patterns are dominated by the three central counties, with Wake County comprising more than half of regional travel. Most travel in the region occurs within a single county (intracounty travel). Transit trips echo these overall patterns, and the largest intercounty transit market is between Durham and Orange counties. Table 2.1 breaks down HBW travel at the County level based on the 2006 Household Travel Survey, while Table 2.2 uses the 2006 Transit On Board survey to analyze regional transit travel patterns for HBW. Table 2.1 Travel Patterns by County (2006 Household Travel Survey) Chatham Durham Franklin Granville Harnett Johnston Orange Wake Total Chatham 3,648 3, ,792 2,629 15,026 Durham ,571-1, ,735 22, ,700 Franklin - 2,931 2,628 1, ,128 20,839 Granville 182 4, , ,745 14,712 Harnett , ,636 8,227 Johnston - 1, ,245-27,612 44,656 Orange , ,656 8,261 98,417 Wake ,206 1, ,136 10, , ,052 Total 6, ,501 5,092 8,574 3,572 17, , , ,629 Table 2.2 Transit Travel Patterns by County (2006 Transit On Board Survey) Chatham Durham Franklin Granville Harnett Johnston Orange Wake Total Chatham Durham 6 5, , ,288 Franklin Granville Harnett Johnston Orange , ,253 Wake ,524 5,792 Total 6 6, ,021 4,771 18,511 *Data source: 2006 Transit On Board Survey. At the district level, three locations contribute most significantly to the regional travel market: North Central Wake County (district 12), Southwest Wake County (district 15), and Raleigh (district 18). These three districts account for 38 percent of regional trip productions and 42 percent of regional trip attractions. For HBW trips, other major attraction areas in the region include Chapel Hill (district 16), Durham Center (district 17), and Triangle Regional Park (district 19). The three urban districts of Raleigh, Durham-Orange Light Rail Transit Project September

11 Chapel Hill, and Durham Center are major centers for transit travel in the Triangle region, with 64 percent of transit productions and 87 percent of transit attractions. The projects being studied in the TRTP are designed to serve these locations. Transit trips in the region serve three major purposes, which account for almost 90 percent of transit travel: commuter trips (HBW), university trips (HBU), and NHB (primarily university related travel). For all three purposes, intracounty travel is dominant. Although HBU and NHB have almost 90 percent of their transit trips as intracounty trips, HBW is somewhat lower at 74 percent. Overall, university related trips are the most significant element of the regional transit market, accounting for almost 60 percent of total transit trips in the region. As shown in Table 2.3, the NCSU Wolfline and CHT are dominated by HBU trips, whereas C Tran, Triangle Transit, and CAT primarily serve work trips. NHB trips are particularly important for Duke University Transit, which may be related to the number of transit trips taken between the two Duke campuses. Table 2.3 Trip Purpose by Transit Operator Purpose TTA CAT CHT DATA Wolfline Duke Ctran Total HBW 62% 46% 28% 37% 4% 22% 63% 28% HBSh 1% 9% 2% 10% % 3% HBSc - 1% - 3% 1% - 4% 1% HBO 6% 21% 4% 21% - 3% 8% 7% NHB 12% 17% 14% 21% 27% 42% 8% 23% HBU 20% 6% 51% 8% 67% 33% - 38% A significant portion of low income and transit dependent riders use the various transit systems in the TRTP region. Regionally, 39 percent of transit trips are attributable to riders with household incomes below $15,000. This holds true for the university bus services as well as the more traditional transit operators. The exceptions are C Tran and Triangle Transit, which each have only 20 percent of riders with incomes below $15,000. Transit dependency is also a driving force for transit usage in the Triangle Region: 27 percent of riders report no household vehicles, and an additional 13 percent report living in households with fewer vehicles than workers. Durham-Orange Light Rail Transit Project September

12 3. Review of the Mode Choice Model and Transit Elements This section details the elements of the model most directly related to forecasting transit ridership in the major travel markets: the mode choice model. The structure of the mode choice model elements is explained, with a focus on how the TRME4.2 differs from FTA recommendations and standards and why Structure The mode choice model of TRM has a nested logit model structure, with motorized trips split into auto and transit at the top level, as shown in Figure 3.1. As recommended by FTA, no nonlogit decision rules are used within the TRM. Figure 3.1 Mode Choice Model Structure Choice Automobile Transit SOV Shared Ride Auto Intercept Local Express Urban Rail HOV 2 Walk Access Walk Access Walk Access HOV 3+ Drive Access (Park) Drive Access (Park) Drive Access (Park) Drive Access (Drop Off) Drive Access (Drop Off) Drive Access (Drop Off) The auto mode is further split into single occupancy vehicles, shared ride (two person occupancy vehicles and three plus person occupancy vehicles), and auto intercept. Auto intercept is a mode in which, due to parking restraints at the University of North Carolina (UNC), drivers must park at satellite parking lots and take a shuttle bus to campus. The transit mode includes three submodes (local bus, express bus, and rail) and three modes of access (walk access, park and ride, and kiss and ride). FTA s modeling recommendations highlight the importance of maintaining consistency between the transit path builder and the mode choice model. Based on this recommendation, the time weights used for model estimation and model application need not be identical, but should be close in value. The TRM4E.2 weights have all been adjusted so that those used for the path builder are the same as those used in the mode choice model. Market segmentation for mode choice includes six trip purposes (home based work, home base shopping, home based school, home based university, home based other, and non home based) and Durham-Orange Light Rail Transit Project September

13 two time periods (peak and off peak). Peak trip tables are split into morning peak and evening peak after the mode choice process for assignment to the highway network Coefficients The mode choice model coefficients are asserted coefficients, as early household travel survey and onboard survey data were not adequate for model estimation. Table 3.1 summarizes the coefficients for the key explanatory variables by purpose. Table 3.1 Coefficients of Explanatory Variables in the Model Item HBW HBSh HBSc HBO HBU NHB In-Vehicle Time First Wait Transfer Wait Walk Time Drive Time Cost Coefficient of In Vehicle Time (Civtt) FTA recommends that the Civtt for HBW trips fall within the range 0.03 < Civtt < 0.02 and there be no variations by mode. In TRM4E.2, the Civtt of for HBW and HBU and 0.02 for NHB are within this recommended range. The Civtt of 0.01 for HBSh, HBSc, and HBO is half of Civt for HBW trips, which is within the FTA recommended range. The Civtt are the same across all of the various transit submodes as recommended by FTA Coefficient of Out of Vehicle Time (Covtt) FTA recommendations indicate that Covtt is dependent on Civtt such that: As shown in Table 3.1, Covtt/Civtt is equal to 2 for HBW and HBU and 2.5 for NHB, HBSh, HBSc, and HBO. These asserted values are within the FTA range Implicit Value of Time (VOT=Civtt/Ccost) FTA recommends that value of time (VOT), calculated as Civtt divided by the Coefficient of Cost (Ccost), should be within the range: Therefore, large cost related differences will not be observed in the utilities of different alternatives. Median household income in 2005 for the modeling area is about $51,911 per year. For HBW and HBU trips, the value of time is assumed to equal one third of the hourly wage. For all other purposes, the Durham-Orange Light Rail Transit Project September

14 value of time is assumed to equal one quarter of the hourly wage rate. Coefficients of cost were adjusted so that the implied values of time are consistent with the FTA recommendations Alternative Specific Constants The TRM4E.2 mode choice model has alternative specific constants specified by trip purpose, by time period (peak and off peak), by mode (Transit, Share Ride, Auto Intercept, HOV3+), and by five socioeconomic strata. The five socioeconomic strata are as follows: Strata 1 Households with no vehicles (of all income levels) Strata 2 Low income households with at least one vehicle Strata 3 Low medium and high medium income households with fewer vehicles than workers, but with some vehicles Strata 4 Low medium and high medium income households with vehicles equal to or more than workers Strata 5 High income households with at least one vehicle. Consistent with FTA recommendations, no geography based constants are used in the model. In addition, the constants for line haul modes (specifically express bus) remain constant across all strata. Strata based constants are applied to mode nests (i.e., auto, transit), auto nests (i.e., drive alone, shared ride, and auto intercept), and transit access mode nests (i.e., walk access, park and ride, and kiss and ride). Table 6.3 and Table 6.4 summarize the alternative specific constants used in the mode choice model. The transit constants are almost all negative. The few positive values are all associated with socioeconomic Strata 1 (households with no vehicles) and logically indicate a strong preference for transit. The highly negative transit constants indicate a strong preference for not using transit and are mostly associated with the highest income households (Strata 5) and households with access to many vehicles (Strata 4), as would be expected. Generally, the transit constants by strata show a plausible pattern for the majority of trip purpose/time period combinations, although there are a few anomalies. During the AA phase of this project, the use of the Aggregate Rail Ridership Forecasting (ARRF) model to assist in the development of new mode alternative specific constants were experimented with in consultation with the FTA. However, these were ultimately set aside for potential future use or discussion. Conventional travel demand modeling includes quantifiable variables that affect travelers mode choice behavior in terms of time and cost. A major factor that seems to impact the demand for transit is the preference for the trip maker to use a premium transit mode such as fixed guideway projects (LRT and Bus Rapid Transit (BRT)). However, the unmeasured attributes affecting the choice of mode by the trip maker cannot be directly addressed with level of service variables. The aggregate effects of those variables that cannot be quantified are typically reflected by constants in the mode choice model. These constants are usually calibrated using local transit survey data. As typical of a region without an operating fixed guideway, the Triangle region s travel demand model was developed and calibrated on transit surveys of the current bus users, which did not reflect the potential attractiveness of a fixedguideway service to riders if one were introduced to the market. Durham-Orange Light Rail Transit Project September

15 FTA has issued guidance regarding appropriate levels of constants to consider to represent potential unmeasured attributes of fixed guideway modes. Three categories of attributes are recognized for credits, including guideway like characteristics, span of good service, and passenger amenities. Two types of adjustments can be made for unmeasured attributes of fixed guideway in the areas where a new fixed guideway would be introduced. First, FTA assigns a credit in terms of equivalent minutes of travel time savings to increase the attractiveness of the new guideway for guideway trips. Second, a discount on the weight applied to in vehicle travel time on the guideway is determined to increase the attractiveness of guideway travel. FTA assigns specific values for these two types of credits, based on the specific characteristics of a project in each of the three categories of unmeasured attributes. The maximum values are 15 minutes of time savings for each rider and a 20 percent discount on the travel time weight. In the travel demand forecasting for the AA, a 15 minute fixed credit was used for the proposed LRT projects in the base year model validation and forecasting. Addition of this credit, as implemented in terms of adjusted constants in the mode choice model, increased the forecast ridership and affected all tested alternatives in the same fashion. As the forecasting methodology was refined for the New Starts application, FTA suggested refining the credits that would be applied to the Locally Preferred Alternatives in the study corridors. These refined credits were implemented in the mode choice model to accommodate two different credits for two types of trips guideway only trips and guideway plus local bus trips. At the same time, a discount on the travel time weight for guideway trip travel time was incorporated. These credits/discounts are discussed further in Chapter Transit System Networks The transit networks represented in the TRM4E.2 are based on the long range planning transit networks used with the TRM4 model and have been updated based on the service plans proposed in the projects under study through the AA. Feeder bus networks have been considered for the guideway alternatives Technology Representation Across the three build projects considered in the AA, three premium transit mode services were considered. For Project 1 in Wake County, light rail transit was under consideration. For Project 2 in Durham and Orange counties, both light rail transit and bus rapid transit were considered. For Project 3, commuter rail was considered. As noted previously, the TRM4 considers express transit and rail transit separately from the local transit mode. There is no existing rail transit service in the Triangle Region. Use of ARRF or other FTA guidance to inform new mode alternative specific constant(s) is under consideration. Durham-Orange Light Rail Transit Project September

16 4. Model Enhancements, Calibration, and Validation The TRM4E.2 model development process included expanded calibration and validation efforts, especially including use of the 2006 household travel survey and the 2006 transit onboard survey. However, highway assignment results were not calibrated, and validation was simply reported. This is expected to have a somewhat limited effect on transit ridership forecasts developed by the model. Major refinements to the original mode choice model and associated components for the Alternatives Analysis forecasts included the following: The mode choice model constants were recalibrated using the 2006 household and transit on board travel surveys. Coefficients of cost were adjusted so that the implied values of time are consistent with the FTA recommendations. Weights used for the path builder were adjusted so that they are the same as or close to those used in the mode choice model. Drive access link coding was limited to 40 minutes for auto intercept lots and rail termini, 30 minutes for several strategic lots, and minutes for the rest. For a model run to include guideway effects of guideway facilities, alternative specific constants for guideways were, initially in the AA phase of this project, assumed to be an equivalent of 15 minutes to account for the effects of non included attributes. Additional refinements for the New Starts forecasting included: The mode choice model constants were recalibrated again using the 2006 household and transit onboard travel surveys. Drive access constants were calibrated for express bus and were assumed the same values for fixed guideways. Mode choice source code was modified to correct an auto intercept bug as identified by Triangle Regional Model Service Bureau (TRMSB). HBU trip tables were adjusted to obtain better fit between simulated and observed ridership for the Duke transit services, as similarly done by TRMSB. The TRM mode choice model was modified to comply with further FTA guidance on representing two types of adjustments that can be made for unmeasured attributes of fixed guideway in the areas where a new fixed guideway would be introduced. o First, a credit in terms of equivalent minutes of travel time savings is given to increase the attractiveness of the new guideway for all guideway trips. o Second, a discount on the weight applied to in vehicle travel time on the guideway is determined to increase the attractiveness of guideway travel, especially favoring those with a long guideway travel time. A distinction was made for guideway only and guideway and local bus trips; a full credit was given to the drive access guideway only trips, and a partial credit was given to guideway and local bus trips and non drive access guideway only trips. To account for expanded pre paid transit services and programs since 2005, the model validation year, appropriate average fare model inputs were estimated based on recent data. Durham-Orange Light Rail Transit Project September

17 4.1. Pre paid Fare Ridership Modeling Adjustment Several major employers in the Triangle Region have demonstrated a strong commitment to providing pre paid transit service to their employees and students. Pre paid transit fares occur one of two ways in the Triangle Region: 1. Fare Free service Since January 2002 the University of North Carolina at Chapel Hill (UNC) has had an agreement with the Town of Chapel Hill and the Town of Carrboro, under which UNC prepays approximately 60% of the fares for all ridership, the Town of Chapel Hill pre pays approximately 30% of the fares, and the Town of Carrboro pre pays approximately 10% of the fares on the system, creating a Fare Free environment for all transit riders. In 2010 Durham Area Transit Authority (DATA) implemented the Bull City Connector also as a Fare Free service in an attempt to boost economic development in downtown Durham. On any Chapel Hill Transit service or on the Bull City Connector, passengers simply walk onto vehicles and ride. No swiping or displaying of any fare media is required for any passenger. 2. GoPasses Employers and Institutions purchase passes for their students and employees allowing them to ride any transit service provided by Triangle Transit (Triangle Region), DATA (Durham), Capital Area Transit (CATs) (Raleigh), and C Tran (Cary) without paying a fare, although they do have to swipe a card. Both of these pre paid systems are fairly unique in that unlike most transit passengers the rider is not paying per use or per ride on the transit system. The fares are paid by property taxes, by students of the universities as part of their fees, or by employers and provided as a benefit to their employees. In each case the transit system appears to provide fare free service to the rider since they are personally not paying for their trip Pre paid Transit Service in the D O Light Rail Corridor The Durham Orange (D O) Light Rail Corridor is a heavily traveled corridor linking UNC and the UNC Hospitals in Chapel Hill with Duke University, Duke Medical Center, downtown Durham and East Durham. See Table 4.1 which shows the pre paid ridership by route by each of the transit agencies. Currently, there are six bus routes that operate directly in the D O Light Rail Corridor, each of these six routes are planned to be replaced with the D O Light Rail Project. In March of 2012, the average daily ridership on these six routes was approximately 6,700 of which approximately 6,000 trips (89 percent) were pre paid trips. Table 4.1 Pre Paid Ridership by Route in Durham Orange Light Rail Corridor Provider Route Average Daily Average Daily Pre % of Trips that are Trips Paid Trips Pre Paid Triangle Transit % Triangle Transit % Chapel Hill Transit FCX 1,909 1, % Chapel Hill Transit S 1,309 1, % Chapel Hill Transit HU % DATA BCC 1,715 1, % Total 6,733 5,983 89% Source: Triangle Transit Average Daily boardings for March 2012 Durham-Orange Light Rail Transit Project September

18 Impact of Pre paid Transit Service on Transit Ridership In January of 2002 Chapel Hill Transit began providing transit service in a manner that allowed anyone to ride Chapel Hill Transit without paying a fare for the trip. This resulted in a 76 percent increase in ridership from 2001 to 2003 as shown in Table 4.2 Table 4.2 Chapel Hill Transit Ridership Before and After Implementing Free Fare Year Average Weekday Unlinked Trips (Motor Bus) , ,078 Increase 8,649 % Increase 76% Source: National Transit Database Due to the success of the Chapel Hill Transit pre paid transit service, UNC decided to expand the 100% pre paid transit program beyond the transit service area of Chapel Hill Transit. UNC accomplished this by purchasing 5,000 annual transit passes from Triangle Transit in July 2006, and distributing them to faculty, staff, and students who requested a pass at no additional charge, although transit is a line item on UNC s student fees paid by all students whether they have a transit pass or not. The pre paid annual transit pass became what is now commonly referred to as GoPass. Since 2006, the GoPass program has expanded to other large regional employers and institutions. Most notably for the D O Light Rail Corridor, Duke University became a GoPass employer in August The GoPass program is one of several factors, which has led to an 85 percent increase in transit ridership on Triangle Transit buses in the D O Light Rail corridor, from FY 2006 to FY 2012, see Table 4.3. Other key factors include significant increases in gasoline prices and bus route restructuring within the corridor to better serve the GoPass users and also major destinations in the corridor, including UNC, Duke Medical Center, and downtown Durham (which will be served by D O Light Rail) b. It is important to note that during this time period, headways for Triangle Transit service between Durham and Chapel Hill have generally remained unchanged during both peak and off peak periods. b Due to the rerouting of Triangle Transit service, it is difficult to provide a one to one comparison of routes. Prior to the route restructuring in 2010, routes 402/3 and 412/13 operated in a loop manner providing a one seat ride between Durham, Chapel Hill, and Research Triangle Park. In 2010, these routes were divided into six routes that provide point to point service between Durham and Chapel Hill (routes 400 and 405), Chapel Hill and Research Triangle Park (routes 800 and 805), and Durham and Research Triangle Park (route 700). Durham-Orange Light Rail Transit Project September

19 Table 4.3 Triangle Transit Ridership Increases Following Fare and Service Policy Changes Year Triangle Transit Ridership in D O Routes Corridor , /403 and 412/ , , 405, 700, 705, 800, and 805 Increase 316,209 % Increase 85% Source: Triangle Transit Accounting for Pre paid Fare in Ridership Modeling It is clear that capturing the pre paid fare phenomenon is important to developing a reasonable forecast of transit usage in the corridor. Therefore, to account for the pre paid transit pass program, a weighted average fare input was developed for each service provider using available data on average fare paid. For the proposed Baseline and Build services, the estimated fare used was calculated based on the expected market penetration of pre paid fares. That is, it is assumed that the same level of pre paid trips in the corridor would occur by the Build or Baseline services as the available bus usage data for the corridor suggest Unmeasured Attributes of Guideway Transit (Alternative Specific Effects) Table 4.4and 4.5 show the set of unmeasured attributes of guideway transit that FTA considered to determine appropriate credits for a particular New New Start project and their representation in the D O corridor under this study. Durham-Orange Light Rail Transit Project September

20 Table 4.4 Unmeasured Attributes of Guideway Transit Guideway Attributes Guideway like characteristics Max credit Existing Local Bus Baseline* Durham Orange LRT 8 1. Reliability of vehicle arrival 4 No exclusive guideway and not grade separated. No signal priority or pre emption at grade locations. Operate in mixed traffic, with frequent stops. Operate in mixed traffic, without exclusive guideway and gradeseparation. Plan to implement signal priority or pre emption at grade locations. Station spacing is similar to the proposed LRT. Exclusive guideway and grade separated, except in select locations. Plan to implement signal priority or preemption at grade locations. Long station spacing: 1 mile per station, higher than the median of existing LRT systems as of 2002 (0.83). Proposed Credit: 0.0 Proposed Credit: Branding/ visibility/ learnability 2 Four transit providers operate in the corridor, including Chapel Hill Transit, Durham Area Transit Authority, Triangle Transit, and Duke University. Limited service branding is planned. Stations are not as distinctive as those of proposed LRT. Limited targeted marketing campaigns planned. Proposed Credit: 0.5 Stations, vehicles, and fixed guideways are distinctive and the system is easy to identify and use. Transit referendum was carried out in Durham County in November 2011, increasing the awareness and visibility of the proposed fixed guideway projects in the Triangle region. Transit referendum is scheduled in Orange County for Nov 6, Targeted marketing campaigns planned. Proposed Credit: 2.0 Durham-Orange Light Rail Transit Project September

21 Guideway Attributes Max credit Existing Local Bus Baseline* Durham Orange LRT 3. Schedule free service 2 Headways: peak 15 min, off peak 60 min Two peak period routes with headways of 12 min (all stops) and 30 min (limited stops). One off peak route with headways of 20 min. Headways: peak 10 min, off peak 20 min. Proposed Credit: 1.5 Proposed Credit: 0.5 Span of good service 3 Operation hours vary by providers and routes, and some examples are as follows: Triangle Transit: 6:00 a.m. to 10:30 p.m.; DATA: 5:30 a.m. to 12:30 a.m; CHT: 5:00 a.m. to 1:15 a.m.; Duke: 7:20 a.m. to 3:55 a.m. Operation hours from 5:30 a.m. to 12:00 a.m., and headways in offpeak hours are 20 min. Proposed Credit: 0.5 Note: In part, a smaller credit is proposed for the Baseline than the Build for this attribute in recognition of differing perceptions of extended span of express bus service as compared with an extended span of LRT fixed guideway service. Operation hours from 5:30 a.m. to 12:00 a.m., and headways in off peak hours are 20 min. Service span would be vastly increased during off peak times and particularly weekends, which is far superior than is provided today in the corridor. Proposed Credit: 3.0 Passenger amenities at stations/stops 1. Physical characteristics 4 3 No station amenities. Limited passenger amenities at stations beyond upgrades to shelters and signage. Proposed Credit: 0.5 Station amenities include: Safety and security features such as good lighting Modern good looking shelters to protect from bad weather Good maintenance and cleanliness Durham-Orange Light Rail Transit Project September

22 Guideway Attributes Max credit Existing Local Bus Baseline* Durham Orange LRT Comfortable benches Spacious platform with good visibility and sightlines Provision of bicycle storage at PnR lots Information kiosks Proposed Credit: Dynamic schedule information 1 Triangle Transit has recently provided real time arrival predictions for six transit providers in the region. Provision of real time information on vehicle arrivals, the same as that provided today. Proposed Credit: 0.0 Provision of real time information on vehicle arrivals Signs show minutes until next arrival/ departure. Transit routes & schedules are clearly posted. Service change information is posted and announced. Triangle Transit has provided real time arrival predictions for six transit providers in the region. Proposed Credit: 1.0 TOTAL Lump Sum Credit 15 Proposed Credit 2.0 Proposed Credit 12.5 *Proposed Baseline credits are presented for discussion purposes. For this submission, due in part to technical challenges, the Baseline was coded in the same manner as existing express bus services. Durham-Orange Light Rail Transit Project September

23 Table 4.5 Proposed In Vehicle Travel Time Related Alternative Specific Effects Guideway Attributes Max credit Existing Local Bus Baseline Durham Orange LRT Passenger amenities on transit vehicles 0.75 guideway time weight Proposed Credit: none Proposed Discount: Reliability of travel time Local buses operate in mixed traffic, subject to the effects of traffic congestion and delays at signals. Local buses operate in mixed traffic, subject to the effects of traffic congestion and delays at signals. The alignment is mostly exclusive guideway, grade separated with signal priority and preemption at downtown locations, and it has long station spacing. 2. Availability of a seat Heavy loadings occur on some routes, affecting seat availability. Heavy loadings can occur on some routes, affecting seat availability. Heavy loadings are not expected in most stations. Seats are generally available. 3. Ride quality Transit ride is subject to stopand go traffic and may be bumpy. Transit ride is subject to stop and go traffic and may be bumpy. Transit ride is expected to be smooth and quiet, state of art design and construction. 4. Vehicle amenities Transit vehicles are standard buses. Transit vehicles are hybrid buses. Modern light rail vehicles are used, with larger capacity than standard buses. Seats are comfortable. Variable signs in vehicle. Durham-Orange Light Rail Transit Project September

24 Validation summaries for the TRM4E.2 model are shown in Table 4.6 through Table 4.9. As shown in Table 4.6 Assigned versus Observed Transit Ridership, the model modestly overestimates 2005 transit ridership, with a deviation of only 5.8 percent. Table 4.8 compares the observed mode shares from the 2006 household travel survey with the models estimated 2005 mode shares. On a regional basis, the percent splits are quite similar for all six modes and all five socioeconomic strata. Table 4.9 compares the observed and estimated values of transit access mode shares. Table 4.6 Assigned versus Observed Transit Ridership Transit Operator Observed Ridership Modeled Peak Ridership Modeled Off Peak Ridership Modeled All Ridership Modeled Percentage Deviation TTA 4,472 3, , % CAT 13,912 8,446 7,997 16, % CHT 26,201 15,294 14,980 30, % DATA 14,241 6,458 9,099 15, % NCSU 11,272 4,384 5,429 9, % Duke 13,985 6,269 6,213 12, % Total 84,083 44,793 44,139 88, % Table 4.7 Transit Trips by Trip Purposes (2005) Trip Purpose Peak Off Peak Daily Estimated Percentage Total Observed Percentage Total HBW 12,027 6,535 18, % 28.1% HBSh 1,264 1,731 2, % 3.5% HBSch % 0.8% HBO 2,563 3,346 5, % 8.1% NHB 6,630 9,543 16, % 22.0% HBU 12,366 14,343 26, % 37.5% Total 35,185 35,737 70, % 100% Durham-Orange Light Rail Transit Project September

25 Table 4.8 Observed (2006) and Estimated (2005) Mode Shares of Motorized Travel SOV HOV2 HOV3+ Auto Intercept Local Bus Express Bus Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est. Strata % 17.8% 42.4% 59.9% 1.9% 3.9% 0.8% 0.5% 24.5% 17.0% 1.3% 0.9% Strata % 54.1% 30.8% 32.0% 12.0% 12.2% 0.0% 0.0% 3.0% 1.7% 0.1% 0.0% Strata % 42.3% 30.6% 39.0% 26.1% 17.5% 0.0% 0.0% 3.3% 1.2% 0.1% 0.0% Strata % 63.6% 25.2% 23.8% 13.0% 12.2% 0.0% 0.0% 0.4% 0.4% 0.0% 0.0% Strata % 62.1% 25.6% 24.4% 14.6% 13.2% 0.0% 0.0% 0.2% 0.3% 0.0% 0.0% Total 58.5% 57.5% 26.5% 28.5% 13.5% 12.4% 0.0% 0.0% 1.4% 1.5% 0.1% 0.1% Table 4.9 Observed (2006) and Estimated (2005) Access Mode Shares of Transit Trips Local Bus Walk Local Bus Park and Ride Local Bus Kiss and Ride Express Bus Walk Express Bus Park and Ride Express Bus Kiss and Ride Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est. Obs. Est. Strata % 86.6% 12.3% 9.9% 3.3% 3.6% 56.5% 44.7% 34.4% 53.8% 9.1% 1.5% Strata % 78.5% 14.3% 13.8% 7.1% 7.7% 65.5% 66.5% 24.9% 31.1% 9.6% 2.4% Strata % 86.4% 6.5% 5.4% 7.4% 8.2% 62.8% 72.0% 28.0% 24.7% 9.2% 3.3% Strata % 58.7% 42.5% 38.5% 1.9% 2.9% 25.3% 15.6% 68.7% 84.1% 6.0% 0.3% Strata % 77.8% 18.9% 17.6% 4.1% 4.5% 44.1% 47.4% 48.3% 51.8% 7.7% 0.7% Total 79.3% 81.4% 16.7% 14.1% 4.0% 4.6% 48.0% 40.7% 43.8% 58.0% 8.2% 1.4% Durham-Orange Light Rail Transit Project September

26 5. Alternative Modeling Approaches 5.1. Aggregate Forecasting Using ARRF II The FTA ARRF Model II estimates total unlinked rail transit trips for light rail and commuter rail systems by applying a series of expected rail shares to the amount of total (all mode) travel to work occurring within the rail corridor as recorded in the Year 2000 Census Transportation Planning Package (CTPP) (AECOM 2009). ARRF II was developed to provide insights into the reasonableness of forecasts, to better understand potential markets, to offer potential targets for model calibration in so called New New Starts cases, and to provide a basis for quality control comparisons in the system expansion cases. Table 5.1 shows assumptions used in estimating rail ridership in the two proposed corridors in the Triangle region. ARRF II was applied to estimate the 2000 rail ridership for proposed D O and Wake LRT systems as part of the Alternatives Analysis. The 2005 ridership estimates were generated by growing the 2000 CTPP data to 2005, using the Fratar method. The population and employment growth rates at the TAZ level were initially used to create marginal control totals, with additional marginal controls at the county level. Similarly, the 2000 CTPP data were also grown to 2035 for estimating the 2035 ridership. Table 5.1 ARRF II Assumptions and Ridership Estimates Item Durham-Orange Wake Directional Route Miles Estimated Average Speed (mph) Average Station Spacing (mile) Estimated Trains per Day per Direction Estimated 2005 CTPP Flows (homes within 2 miles and work within 1 mile of any station) Unlinked Daily Rail Trips (2005) Work trips PNR access 26,700 36,900 15,100 6,300 3,300 23,500 11,000 7, Choice Set Approach Cambridge Systematics developed an adaptation of the enhanced TRM model (V4E.2) to employ a choice rider approach similar to an approach previously presented at an FTA Workshop on Travel Forecasting in March 2009, which was used to forecast the Charlotte South Corridor LRT ridership as part of a before after study. This approach is based on the idea that many choice riders will not consider using regular buses, but will consider using fixed guideway services. As in the Charlotte example, the modified choice set methodology was used for all premium services, including express bus. This analytical framework was presented as an alternative method to conducting ridership forecasting, illustrating uncertainties associated with the conventional modeling approach. Strata 4 and 5 as defined in the TRM model are households with higher income and were considered part of choice riders in this approach. Bus services are not considered a viable option to some people in these higher income groups, while premium services such as fixed guideway services are an option for everyone. Durham-Orange Light Rail Transit Project September