Appendix B: Travel Demand Forecasts July 2017
Table of Contents 1 Introduction... 2 2 Model Review and Updates... 2 2.1 Overview of Smart Moves Model ( City of London Model )... 2 2.1.1 Network and Zone System... 3 2.1.2 Trip Generation... 6 2.1.3 Trip Distribution... 7 2.1.4 Mode Choice... 7 2.2 Model Refinements and Updates... 7 3 Future Demand Forecasts... 9 3.3 System-Wide Transit Forecasts... 9 3.4 Screenline Travel Demand... 10 3.5 Station Ridership Forecasts... 10 July 2017 i
1 Introduction IBI Group and WSP were retained in August 2014 by the City of London to undertake an Environmental Assessment (EA) for the City s Rapid Transit (RT) initiative. The EA follows from recommendations of the recently completed City of London Smart Moves 2030: Transportation Master Plan (TMP) by AECOM in 2013. The Rapid Transit EA will determine the recommended plan for implementation of rapid transit, including key decision items such as the preferred operational form and corridors for the rapid transit lines. The development of travel demand forecasts and transit ridership forecasts are key inputs into the Environmental Assessment. Future travel demand forecasts are derived using the City of London Travel Demand Model. The purpose of this report is to provide an overview of the London Travel Demand Model and updates made to the model to refine its suitability for evaluating rapid transit alternatives and key performance measures. Further details on the application of the model and the evaluation of alternatives is provided in the Rapid Transit Master Plan and the Rapid Transit Business Case. It should be noted that the review and refinement of the London Travel Demand Model for the purpose of the Rapid Transit Environmental Assessment was substantially carried out in 2015, as part of the assessment of alternatives. As a result, existing conditions described herein are reflective of that time. Minor updates were made to the ridership forecasts over the period from 2015 to late 2016 as new data became available. 2 Model Review and Updates 2.1 Overview of Smart Moves Model ( City of London Model ) The City of London Transportation Master Plan Smart Moves Model ( City of London Model ) is a multimodal travel demand forecast model based on the traditional four-stage modelling approach. The model was updated by AECOM in 2012-2013 for the TMP. It was made available to IBI Group by the City of London for the appropriate modelling purposes required to complete the EA. The TMP included an assessment of initial rapid transit networks, and these were used as a starting point for the development and testing of options for the current EA. The City of London Model s base year for the TMP was 2009, with a total resident population of 362,138 and a post-secondary non-resident student population of 41,635. The model is in TransCAD for auto and transit peak hour assignments. The model future year for the Smart Moves TMP was 2030. The 2030 horizon model for the TMP included two scenarios. Scenario 2 represents business-as-usual in terms of transit service levels and TDM measures (no major increase in downtown parking costs, etc.). Scenario 3 model output was provided, which represents a more aggressive TDM plan in an attempt to achieve the TMP targeted transit mode share of 20% within the rapid transit corridors. Details on the development of the Transportation Model are described in Appendix G of the Smart Moves TMP. The following sections provide a detailed review of the City s model processes and attributes. July 2017 2
2.1.1 Network and Zone System Road Network The base (existing) road network from the TMP model is shown in Exhibit 2-1. The network was reviewed in terms of number of lanes, freeflow speed, and capacity within the proposed rapid transit corridors and observed to match existing conditions. A number of road improvements were identified as part of the Smart Moves TMP. The list of improvements was subsequently refined as part of the 2014 Development Charges Background Study. Exhibit 2-2 indicates the improvements assumed for the future horizon year. This future network was used as the starting point for developing and modelling rapid transit alternatives. Exhibit 2-1: Base Road Network July 2017 3
Exhibit 2-2: Future Road Improvements by 2034 Note: This exhibit does not represent the preferred Rapid Transit alternative. Road improvements on rapid transit corridors represent assumptions as of 2014. See Rapid Transit Master Plan for description of proposed improvements on RT corridors. Transit Network For the future planning horizon model, the Smart Moves TMP developed the following transit routes and networks: Business-as-usual (BAU): 82 bus routes; and 2030 TMP (with RT): 86 transit routes, 82 of which are bus routes and four of which are RT routes. Since the TMP, several changes have been made to the transit network. In particular, LTC has added new express routes and restructured some local routes. Future transit networks were modified to reflect changes to the existing network since the Smart Modes TMP. LTC also completed a Transit Network Rapid Transit Integration Study and Financial Plan in August 2016, which took place in parallel with the Rapid Transit Master Plan process. Exhibit 2-3 presents a map of proposed RT routes overlaid on the base case transit network. Zone System July 2017 4
There are 634 zones in the City of London Model, of which: 532 are City of London traffic analysis zones (TAZs); 96 are Middlesex County and nearby outlying zones (excluding City of London); and 6 are external zones. The primary study area for model review strictly consists of the 532 City of London TAZs, but all 634 model zones are properly accounted for in base and future horizon auto and transit assignments. See Exhibit 2-4 for the model zone system within the City. Exhibit 2-3: Base Case Transit Network with BRT Routes Highlighted July 2017 5
Exhibit 2-4: City of London Traffic Zones 2.1.2 Trip Generation Trip generation in the City of London Model is managed in a spreadsheet. Land use inputs for each TAZ (see Section 0) are used to calculate trip productions and attractions by purpose. The TMP model has a 2009 base-year, based on 2006 census data. The TMP utilized 2030 population and employment inputs developed by the City. Subsequent to the TMP, the City updated land use forecasts as part of the 2014 Development Charges Background Study. This included the development of population and employment by TAZ for 2011 (based on the 2011 Census) and extending the future year forecasts to 2034. July 2017 6
2034 forecasts are based off Scenario 2 from the TMP and reflect an intensification of land use along rapid transit corridors. Exhibit 2-5 illustrates the 2011 and 2034 population and employment forecasts for the City of London overall and for the rapid transit corridors. For the purpose of this summary, Rapid Transit corridors are defined as including traffic zones where the majority of the population is within 1 kilometre of the potential corridors. Population is expected to increase within the rapid transit corridors, from 152,200 to 170,500, which represents a 12% increase. Employment is expected to increase from 114,700 to 189,700, which represents a 14% increase. Overall, the City s population and employment are projected to increase by 24% and 22% respectively. Note that these figures differ slightly from The London Plan forecasts, which are based on a horizon year of 2035 and base year of 2015. Exhibit 2-5: 2011 and 2034 Land Use Model Inputs RT Corridor 2011 2034 2011-2034 Increase Citywide RT Corridor City-wide RT Corridor City-wide Population 152,200 366,200 170,500 454,700 12% 24% Employment 114,700 189,700 131,000 231,800 14% 22% 2.1.3 Trip Distribution As described in more detail in the TMP Model Development Report (Smart Moves Appendix G), trip distribution is determined through a doubly-constrained Gravity process. Each of the eight trip purposes has its own calibrated trip distribution function. The friction factors used in the function are based on distance between zones, and do not include travel time by auto, transit or other modes. As a result, the TMP model may underestimate the upside potential for rapid transit to shape travel patterns and trip distribution, and thus the ridership forecasts could be considered conservative. 2.1.4 Mode Choice The City of London Model employs a multinomial logit model to predict the choice of travel mode. Generalized auto and transit costs are calculated using assumed monetary value of distance based on fuel consumption/price and travel time, plus a flat fare for transit. For walking/cycling, the mode choice utility is calculated based on distance only. Refer to Appendix G of the City of London TMP for detailed parameters and utility function definitions. As discussed below, in order to add an additional level of sensitivity to the mode choice component to reflect the characteristics of rapid transit additional processes were executed. Following industry practices, a travel time elasticity model was developed to operate with the existing transportation model, where the change in transit travel time with BRT relative to BAU conditions was estimated on an origin-destination basis by the model based on coded transit networks, and elasticity applied to estimate the change in transit ridership. 2.2 Model Refinements and Updates Overall, the structure of London s existing Transportation Model, developed in 2012 for the TMP, is appropriate and provides a base for developing ridership forecasts and performance metrics for the Rapid Transit EA. However, given the time that has passed since the model development, several refinements and updates are appropriate including: July 2017 7
Validating the model and implementing a process to enhance corridor-level forecasting; Implementing an elasticity-based mode split model to better capture the change in transit use brought about by the implementation of rapid transit. To reflect existing (2015) conditions, model auto and transit volumes at rapid transit corridor screenlines were compared to observed traffic and transit counts for the p.m. peak hour. While the model for the TMP was calibrated to match system-wide counts, the comparison of observed and model-predicted counts indicated some discrepancies at rapid transit screenlines, requiring a modified process to allow the model to better reflect existing and projected future conditions in rapid transit corridors. The resulting process entailed a pivot-point approach where the 2009 O-D survey auto and transit matrices were used to reflect base year 2009 conditions, with the transportation model used to project changes on an origin-destination basis by mode to the horizon year of 2034. As such, the level-of-service sensitivities and behaviour elements of the model in forecasting 2009 to 2034 travel were captured, with the model-predicted growth change applied to or pivoted off of the 2009 Survey O-D matrix. The straight application of the above process resulted in a decrease of overall transit mode share for the City between 2009 and 2034, reflecting that much of the population growth was occurring in the outer areas of the City where transit mode split levels are lower. Since the process did not reflect the natural increases in transit use brought about by attitudinal, land use/urban design changes and on-going transit service investments over the next 30 years, it is likely that the future transit demand is underestimated. To reflect these effects, the 2034 transportation model-generated mode split was increased by 15% on an origin-destination basis. Based on the above, the model was validated for use in rapid transit corridors. As a second-stage refinement to the transportation model, an elasticity-based mode split model was incorporated as model tests and a review of the mode split model formulation confirmed that the transportation model was not sensitive to level-of-service improvements that would accompany the implementation of rapid transit. Following industry practices, a travel time elasticity model was developed to operate with the existing transportation model, where the change in transit travel time with BRT relative to BAU conditions was estimated on an origindestination basis by the model based on coded transit networks, and elasticity based approach applied to estimate the change in transit ridership. The introduction of rapid transit in London would result in both reduced transit travel times and improved service levels (i.e. increased frequency). Rather than introduce separate elasticities for travel time and service, a single travel time elasticity of -1.0 was used to capture both effects. Travel time elasticities observed in the industry are typically around the -0.6 range, but do not capture the uplift of introducing a new higher-quality mode and service improvements. Hence, an elasticity of -1.0 was considered reasonable. To estimate travel time savings, the BRT operating speed was assumed to be 25 km/h and the customer perceived benefit of the improved reliability and convenience of the mode was felt to be the equivalent of a 15 km/h speed bonus. Hence, the speed used for BRT for elasticity purposes was 40 km/h. For the Hybrid and LRT scenarios, further adjustments were made to the ridership forecasts to account for the perceived benefits of rail (e.g. comfort and image), which cannot be captured in a traditional-four stage model. July 2017 8
3 Future Demand Forecasts 3.3 System-Wide Transit Forecasts System-wide ridership projections to 2035 were developed based on the 2034 Travel Demand Model results and extrapolating back to 2015. Exhibit 3-1 shows the projected ridership for the preferred BRT network in comparison to historic trends. (Note that 2035 is selected as the horizon year for presentation of forecasts in the Business Case, but detailed model results are based on the 2034 model.) As shown on Exhibit 3-1, London s transit ridership has increased steadily every year, with the exception of 2015, where a change to the Ontario Works transit pass program affected transit ridership, and 2009, where a transit strike was experienced. Overall, the difference between the BAU and preferred rapid transit scenario is approximately 10%, which is attributable to the improved speed and reliability provided by the introduction of rapid transit: Annual ridership was 22.4 million in 2015 Under a Business-as-Usual (BAU) scenario, without rapid transit, ridership is expected to increase to 28.4 million riders in 2035, or an increase of 26.7%. This compares to a population growth of 20.2% over the same period. Under the preferred BRT scenario, overall system ridership is projected to reach 31.3 million in 2035, representing an increase of 39.7%. Exhibit 3-1: Past and Projected Future Transit Ridership (1998 to 2035) July 2017 9
3.4 Screenline Travel Demand In addition to system-wide forecasts, ridership was forecast for selected screenlines generally representing the peak point ridership on each corridor. For the purpose of this assessment, screenlines are defined as follows: 1. North Corridor: Richmond Street and Wharncliffe Road North of Oxford Street (Southbound) 2. East Corridor: Oxford Street, Dundas Street, King Street and York Street west of Adelaide Street (Westbound) 3. South Corridor: Wellington Street, Richmond Street and Ridout Street south of Horton Street (Southbound) 4. West Corridor: Oxford Street and Riverside Drive west of Thames River (Westbound) Exhibit 3-2 shows the projected transit demand for the four key screenlines, including a breakdown of the rapid transit component. It should be noted that the ultimate split between rapid transit and other transit will depend on the final LTC conventional service plans. The highest peak demands are projected for the north and east corridors, with an estimated 1,450 passengers/hr (peak point) north corridor and 1,360 passengers/hr (peak point) on the east corridor. Ridership on the south and west corridors is in the range of 600-650 passengers/hr at the two screenlines. Exhibit 3-2: Screenline Transit Demands Location Existing 2034 BAU 2034 with RT RT Volume 1. Richmond/Wharncliffe SB - - 1,460 2. Oxford/Dundas/ King/York WB - - 1,360 3. Wellington/Ridout SB - - 650 4. Oxford/ Riverside WB - - 590 Total Transit Volume 1. Richmond/Wharncliffe SB 1,357 1,670 1,830 2. Dundas/ King/ York WB 837 1,070 1,650 3. Wellington/Ridout SB 494 780 920 4. Oxford/ Riverside WB 477 720 850 3.5 Station Ridership Forecasts The passenger boardings and alightings at each station were developed through the travel demand model. The forecasted 2034 PM Peak Hour boardings and alightings, and resulting passenger load, is provided for one direction of travel for each corridor (Exhibit 3.3 and July 2017 10
Notes: Exhibit 3.4). The ridership forecasts will be considered in the next stage of design development to provide appropriate passenger amenities at stations with higher ridership. Exhibit 3.3: North + East: 2034 PM Passenger Boardings and Alightings by Station: From North to East North + East Corridor from North to East (2034 PM Peak Hour) Boarding (On) Alighting (Off) Passenger Load Masonville Place (Terminal) 200 N/A 200 Western Road at Richmond 50 10 240 Western Road at Windermere 80 10 310 Western Road at Elgin 450 10 750 WU Campus Centre**** 690 30 1410 Richmond Street at University Drive 70 30 1450 Richmond Street at Grosvenor Street 90 110 1430 Richmond Street at Oxford Street 60 150 1340 Clarence Street at Central Avenue 50 110 1280 Clarence Street at Queens Avenue 60 320 1020 Central Transit Hub** 150 530 640 King Street at Colborne 30 120 550 King Street at Adelaide Street 40 170 420 King Street at Ontario Street 10 20 410 Dundas Street at McCormick Blvd 30 40 400 Dundas Street at Highbury Avenue 30 220 210 Highbury Ave at LPH access 10 60 160 Highbury Ave at Oxford Street 30 80 110 Fanshawe College (Terminal) N/A 110 0 **The Central Transit Hub is proposed to be located at or near King Street and Wellington Street. **** RT station on Western University Campus to be determined in consultation with Western University. July 2017 11
Exhibit 3.4: South + West: 2034 PM Passenger Boardings and Alightings by Station: From West to South Notes: South + West Corridor from West to South (PM Peak Hour) Boarding (On) Alighting (Off) Passenger Load Oxford Street at Wonderland (Terminal) 70 N/A 70 Oxford Street at Beaverbrook 10 0 80 Oxford Street at Cherryhill 30 10 100 Oxford Street at Wharncliffe 50 10 140 Wharncliffe at Riverside 20 10 150 King Street at Talbot (EB) 50 20 180 Central Transit Hub** 140 50 270 Wellington Street at Horton 410 30 650 Wellington Street at South 40 60 630 Wellington Road at Bond 10 70 570 Wellington Road at Base Line 10 40 540 Wellington Road at Commissioners 20 50 510 Wellington Road at Wilkins 10 90 430 Wellington Road at Southdale 10 150 290 Wellington Road at White Oaks Mall (Terminal) N/A 290 0 **The Central Transit Hub is proposed to be located at or near King Street and Wellington Street. July 2017 12