Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sustainable road safety improvement through the promotion of an alternative mode : Okanagan Valley electric.. Boozarjomehri, Elham 2009-12-31

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
ubc_2009_spring_boozarjomehri_elham.pdf [ 12.53MB ]
Metadata
JSON: 1.0063128.json
JSON-LD: 1.0063128+ld.json
RDF/XML (Pretty): 1.0063128.xml
RDF/JSON: 1.0063128+rdf.json
Turtle: 1.0063128+rdf-turtle.txt
N-Triples: 1.0063128+rdf-ntriples.txt
Original Record: 1.0063128 +original-record.json
Full Text
1.0063128.txt
Citation
1.0063128.ris

Full Text

SUSTAINABLE ROAD SAFETY IMPROVEMENT THROUGH THE PROMOTION OF AN ALTERNATIVE MODE:  OKANAGAN VALLEY ELECTRIC RAIL byElham Boozarjomehri B.Sc. in Aerospace Engineering, Sharif University of Technology, 2007 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE inThe College of Graduate Studies (Civil Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Okanagan)March 2009 © Elham Boozarjomehri, 2009- ii - ABSTRACT Across North America there is an increasing demand for safer, faster, less energy intensive and less costly modes of transportation. There are enormous social and economic costs associated with road collisions, which have been recognized world-wide as a serious problem. One local alternative to reduce auto use and resultant road collisions is the placement of a railway line through the Okanagan Valley in British Columbia, Canada, connecting Osoyoos and US railways in the south with Vernon and cross-Canada railways in the north. This line would not only be able to service Okanagan Valley commuters and tourists, currently served by Highway 97, but also provide an additional freight link between the United States and Canada.  The objectives of this research were threefold: (1) To develop a set of macro-level collision prediction models for Highway 97, for use in calculating the Safety benefits of the proposed railway, (2) To conduct a comprehensive literature review of the design issues and costs of railway freight and passenger systems, and, (3) To conduct a social cost-benefit analysis of electric railways in Canada for a case study of the Okanagan Valley.To this end, a conceptual design, including route alignment, traffic forecast as well as a social benefit/cost analysis, was conducted. The paucity of data and studies for the Okanagan Valley regarding railway planning and engineering required several extrapolations from data reported in the literature.  These estimates were made to gain insight into the data needed to conduct full scale analyses where gaps in knowledge exist, for the purpose of further research. The results of this research were in line with expectations. Regarding sustainable road safety, the development of macro-level collision prediction models for Highway 97 was successful. Regarding knowledge gaps, several were identified on traffic and cost data, and in freight forecast models. However, what was available allowed an order-of-magnitude construction cost estimate of 990 million 2007 dollars for 176 kilometres of track. Moreover, if an electric railway began operating in 2050, the benefit/cost ratio of the project was estimated at 1.15, suggesting an Okanagan Valley railway warrants further research.- iii - TABLE OF CONTENTS ABSTRACT........................................................................................................................ ii TABLE OF CONTENTS...................................................................................................iii LIST OF TABLES............................................................................................................. vi LIST OF FIGURES ........................................................................................................... ix ACKNOWLEDGEMENTS............................................................................................... xi DEDICATION.................................................................................................................. xii1 INTRODUCTION ...................................................................................................... 1 1.1 Study area............................................................................................................ 1 1.2 Transportation network in the study area............................................................ 4 1.3 History of rail transportation in the study area ................................................... 4 1.4 Revitalization of railways in the Okanagan Valley ............................................ 5 1.4.1 Sustainable road safety (SRS):.................................................................... 6 1.4.2 Sustainable economic prosperity: ............................................................... 7 1.4.3 Sustainable ecological footprint.................................................................. 9 1.4.4 Wider relevance: ....................................................................................... 10 1.5 Objectives of the research................................................................................. 12 1.6 Structure of the thesis........................................................................................ 13 2 PASSENGER DEMAND FORECASTS ................................................................. 15 2.1 Introduction....................................................................................................... 15 2.2 Literature review............................................................................................... 15 2.2.1 Statewide models ...................................................................................... 16 2.2.2 Regional models........................................................................................ 20 2.3 Methodology for demand forecasting for the Okanagan Passenger Railway... 25 2.3.1 Trip generation & distribution .................................................................. 26 2.3.2 Mode choice.............................................................................................. 26 2.3.2.1 Model selection..................................................................................... 27 2.3.2.2 Mode choice model development ......................................................... 29 2.3.2.3 Calibration............................................................................................. 36 2.3.2.4 Validation.............................................................................................. 37 2.4 Demand forecast for the Okanagan Passenger Railway ................................... 41 2.4.1 Trip generation & distribution .................................................................. 41 2.4.2 Mode choice.............................................................................................. 42 2.4.3 Forecasts ................................................................................................... 48 2.5 Summary........................................................................................................... 49 3 FREIGHT DEMAND FORECASTS ....................................................................... 51 3.1 Introduction....................................................................................................... 51 3.2 Literature review............................................................................................... 51 3.2.1 Freight model classes and the components............................................... 52 - iv - 3.2.2 Rail freight demand modeling components .............................................. 54 3.2.2.1 “Trip generation & trip distribution” or “direct factoring”................... 54 3.2.2.2 Mode split ............................................................................................. 55 3.2.2.3 Rail assignment..................................................................................... 55 3.3 Methodology for demand forecasting for the Okanagan Freight Railway ....... 56 3.3.1 Direct factoring O-D table ........................................................................ 56 3.3.1.1 Likely origins and destinations ............................................................. 57 3.3.1.2 Data sources.......................................................................................... 64 3.3.2 Mode split ................................................................................................. 65 3.3.2.1 NB/SB mode shares .............................................................................. 66 3.3.3 Assignments.............................................................................................. 69 3.4 Demand forecast ............................................................................................... 70 3.4.1 O-D table development............................................................................. 70 3.4.2 Future growth............................................................................................ 77 3.4.3 Rail share calculation................................................................................ 80 3.4.4 Rail assignment......................................................................................... 85 3.4.4.1 Method # 1: shortest path...................................................................... 85 3.4.4.2 Method # 2: Lansdowne rule-base method........................................... 87 3.4.4.3 Method # 3: spatial separation method ................................................. 89 3.4.5 Demand forecasts for the Okanagan Freight Railway .............................. 90 3.5 Summary............................................................................................................93 4 COST ESTIMATION............................................................................................... 94 4.1 Introduction....................................................................................................... 94 4.2 Alignment selection .......................................................................................... 94 4.2.1 Literature review and methodology.......................................................... 94 4.2.2 Application of the method and results ...................................................... 97 4.2.2.1 Generation of the cost surfaces............................................................. 97 4.2.2.2 Proposed alignments ........................................................................... 109 4.3 Capital and O&M costs................................................................................... 112 4.3.1 Literature review and methodology........................................................ 112 4.3.1.1 Track cost............................................................................................ 113 4.3.1.2 Electrification...................................................................................... 114 4.3.1.3 Other capital costs............................................................................... 116 4.3.1.4 Operating and maintenance costs........................................................ 116 4.3.1.5 Construction management and contingencies..................................... 117 4.3.2 Application of the method and results .................................................... 117 4.4 Summary......................................................................................................... 117 5 BENEFITS.............................................................................................................. 119 5.1 Introduction..................................................................................................... 119 5.2 Rail benefits except safety benefits................................................................. 119 5.2.1 Literature review and methodology........................................................ 119 5.2.1.1 Rail revenue ........................................................................................ 119 5.2.1.2 Okanagan Passenger Rail externality benefits.................................... 120 5.2.1.3 Okanagan Freight Rail externality benefits ........................................ 121 - v - 5.2.1.4 Unquantifiable benefits....................................................................... 122 5.2.1.5 Travel time differences ....................................................................... 123 5.2.2 Application of the method and results .................................................... 124 5.3 Safety benefits................................................................................................. 124 5.3.1 Literature review..................................................................................... 124 5.3.2 Methodology for safety benefit evaluation............................................. 126 5.3.2.1 Collision Prediction Model (CPM) development ............................... 126 5.3.2.2 Safety benefits calculation .................................................................. 129 5.3.3 Model development ................................................................................ 130 5.3.4 Safety benefits calculation ...................................................................... 135 5.3.4.1 Application of developed CPMs......................................................... 135 5.3.4.2 Safety benefits for the city of Kelowna .............................................. 136 5.3.4.3 Results................................................................................................. 138 5.4 Summary......................................................................................................... 139 6 EVALUATION OF PROPOSED RAILWAY ....................................................... 141 6.1 Introduction..................................................................................................... 141 6.2 Literature review and methodology................................................................ 141 6.3 Social cost benefit analysis ............................................................................. 142 6.4 Sensitivity analysis.......................................................................................... 145 6.5 Summary......................................................................................................... 147 7 CONCLUSIONS AND RECOMMENDATIONS ................................................. 148 7.1 Introduction..................................................................................................... 148 7.2 Summary & conclusions................................................................................. 148 7.3 Research contributions.................................................................................... 149 7.4 Recommendations for future research ............................................................ 151 REFERENCES ............................................................................................................... 156 APPENDICESA. UTILITY FUNCTIONS ......................................................................................... 164 B. THE RELATIONSHIP BETWEEN AUTO AND TRANSIT CONSTANTS....... 167 C. DATA SOURCES FOR FREIGHT DEMAND ESTIMATES .............................. 169 D. COST BENEFIT ANALYSIS CALCULATIONS ................................................ 172 E. MONTE CARLO SIMULATION RESULTS........................................................ 178 - vi - LIST OF TABLES Table  1-1 Transportation network in the Okanagan Valley................................................ 5 Table  1-2 Impacts of rail station proximity on property values (Hass-Klau, 2003)........... 8 Table  2-1 FSUTMS 's utility parameters .......................................................................... 17 Table  2-2 Utility parameters for the Oregon’s long distance travel model ...................... 19 Table  2-3 CSX's utility parameters................................................................................... 22 Table  2-4 SMART’s mode split factors............................................................................ 23 Table  2-5 Mode share for all regional travels at the two Svealand line stations .............. 24 Table  2-6 Mode share for all regional travels between Eskilstuna/Strängnäs and Södertälje/Stockholm........................................................................................................ 24 Table  2-7 Intercity trips by OVTP model in 1994............................................................ 26 Table  2-8 Intercity trips by OVTP model in 2020............................................................ 26 Table  2-9 Attributes of reviewed railways........................................................................ 27 Table  2-10 Trips purposes in "2007 Okanagan Travel Survey" ....................................... 30 Table  2-11 Data on transit and auto trips from the Okanagan survey .............................. 31 Table  2-12 Mode choice model coefficients for FSUTMS and OKI models................... 33 Table  2-13 Values for the transit and auto constants for FSUTMS and OKI models ...... 34 Table  2-14 Mode choice coefficients values used in calibration as the Okanagan Values........................................................................................................................................... 35 Table  2-15 Observed and modeled transit share between zones 6 and 7.......................... 35 Table  2-16 Okanagan mode choice model........................................................................ 37 Table  2-17 Mode choice coefficients’ values used in the validation process................... 38 Table  2-18 Validation results............................................................................................ 38 Table  2-19 Home Based Work (HBW) share................................................................... 41 Table  2-20 Home Based Other (HBO) share.................................................................... 42 Table  2-21 Non Home Based (NHB) share ...................................................................... 42 Table  2-22 Car occupancy for different trips.................................................................... 42 Table  2-23 Default values for mode choice model coefficients ....................................... 43 Table  2-24 Oil price forecasts between 2008 and 2030.................................................... 46 Table  2-25 Forecasted demand for the Okanagan Passenger Railway............................. 50 Table  3-1 Freight model classes by component................................................................ 52 Table  3-2 Likely O-D pairs............................................................................................... 64 Table  3-3 O-D data sources .............................................................................................. 65 Table  3-4 SB rail and truck freight movements (in thousands of short tons) and rail share........................................................................................................................................... 67 Table  3-5 NB rail and truck freight movements (in thousands of short tons) and rail share........................................................................................................................................... 68 Table  3-6 2007 trades by rail at eastern ports in thousands of tons.................................. 73 Table  3-7 2007 trades by rail at western ports in thousands of tons................................. 73 Table  3-8  2007 KPR trades in thousands of tons............................................................. 74 Table  3-9 2007 freight movements by trucks to/from the Okanagan Valley in thousands of tons................................................................................................................................ 74 - vii - Table  3-10 2007 trades by truck from Oroville-Osoyoos in thousands of tons................ 76 Table  3-11 2007 trades by truck from Sumas-Huntingdon in thousands of tons ............. 76 Table  3-12 2007 trades by truck from Blaine-Douglas in thousands of tons ................... 76 Table  3-13 2007 trades by truck from Eastport-Kingsgate in thousands of tons.............. 77 Table  3-14 Compound annual growth rate for railways ................................................... 77 Table  3-15 Compound annual growth rate for trucks at customs ports............................ 78 Table  3-16 Commodity composition between Okanagan and Greater Vancouver Regional District............................................................................................................................... 80 Table  3-17 Commodity composition for truck crossings at Osoyoos customs port......... 81 Table  3-18 Commodity composition between Okanagan and the rest of Canada............ 82 Table  3-19 Possible rail share of the Okanagan freight movements ................................ 82 Table  3-20 Potential rail traffic in the Okanagan area...................................................... 82 Table  3-21 Divertible commodities from all border crossing (except Osoyoos) traffic... 84 Table  3-22 Potential rail traffic for Okanagan Railway.................................................... 85 Table  3-23 Centroid for origin/destination zones ............................................................. 86 Table  3-24 Actual distance between the centroids of O-D pairs ...................................... 86 Table  3-25 Assigned traffic to Okanagan Railway using shortest path method............... 87 Table  3-26 Assigned traffic to Okanagan Railway using Lansdowne rule-base method. 88 Table  3-27 Index of spatial separation for the centroids of O-D pairs ............................. 89 Table  3-28 Assigned traffic to Okanagan Railway using spatial separation method ....... 90 Table  3-29 Okanagan Railway freight demand forecast................................................... 91 Table  3-30 Payload factors by two-digit for border crossing ........................................... 92 Table  3-31 Payload factors for the valley......................................................................... 93 Table  4-1 Evaluation criteria in the literature................................................................... 95 Table  4-2 Scale for pairwise comparison.......................................................................... 96 Table  4-3 Relative criterion weights................................................................................. 97 Table  4-4 Cost values for slope’s cost surface.................................................................. 99 Table  4-5 Cost value for geology's cost surface ............................................................. 101 Table  4-6 Cost value for the soil's cost surface .............................................................. 102 Table  4-7 Discount factors for the ROW of KPR and KVR........................................... 107 Table  4-8 Primary attributes of the selected alignment for Okanagan Railway............. 112 Table  4-9 Unit cost of track construction ....................................................................... 114 Table  4-10 Other capital cost estimates.......................................................................... 116 Table  4-11 O & M costs.................................................................................................. 117 Table  4-12 Total estimated construction costs for Okanagan Railway .......................... 118 Table  5-1 Current inter-city transit fare in the Okanagan Valley ................................... 120 Table  5-2 Passenger rail externalities benefits................................................................ 121 Table  5-3 Freight rail externalities benefits.................................................................... 122 Table  5-4 Attributes of the twelve segments .................................................................. 131 Table  5-5 Other attributes of the twelve segments ......................................................... 132 Table  5-6 Developed macro CPMs................................................................................. 133 Table  5-7 Attributes of the validation segments............................................................. 134 Table  5-8 Predicted collisions by different models ........................................................ 135 Table  5-9 Road Collision Costs in Highway 97, British Columbia................................ 136 Table  5-10 Safety benefits of Okanagan Railway (Except savings in Kelowna)........... 136 Table  5-11 Road collision costs in Highway 97, Kelowna............................................. 137 - viii - Table  5-12 Safety benefits of Okanagan Railway in Kelowna....................................... 138 Table  5-13 Break down of Okanagan Railway benefits in 2007 $................................. 140 Table  6-1 Costs and benefits of the project in nominal 2007 dollars ............................. 143 Table  6-2 Net benefits, net costs and present worth of the project in real 2007 dollars. 143 Table  6-3 NPW and B/C for all scenarios ...................................................................... 144 Table  6-4 Random variables and outcomes .................................................................... 145 Table  6-5 Monte Carlo Simulation Results .................................................................... 146 Table  C-1 Data sources................................................................................................... 169 Table  D-1 Costs and benefits of the project in nominal 2007 dollars ............................ 172 Table  D-2 Net benefits, net costs and present worth of the project in real 2007 dollars 175 - ix - LIST OF FIGURES Figure  1-1 Study area..........................................................................................................2 Figure  1-2 Railway network at southern British Columbia................................................ 6 Figure  1-3 LOS in 2035 without any improvements ........................................................ 10 Figure  1-4 Peak-period congestion on the national highway system in 2035 .................. 11 Figure  1-5 Freight rail network at West Coast.................................................................. 13 Figure  2-1 Florida four-level nested model ...................................................................... 16 Figure  2-2 Person Transport Model Flow Diagram.......................................................... 18 Figure  2-3 Oregon mode choice model nesting structure................................................. 19 Figure  2-4 OKI/MVRPC Model Components.................................................................. 21 Figure  2-5 OKI / MVRPC mode choice model structure ................................................. 21 Figure  2-6 The Mälaren valley and surroundings Railways in 2002 (Svealand line thick black)................................................................................................................................. 23 Figure  2-7 Svealand line passengers’ previous mode of travel ........................................ 25 Figure  2-8 Distribution of population around the reviewed railways............................... 28 Figure  2-9 Okanagan mode choice model nesting structure............................................. 30 Figure  2-10 Commuter rail share Vs travel distance ........................................................ 39 Figure  2-11 Speed vs. HSR market share ......................................................................... 40 Figure  2-12 Commuter rail share vs. travel distance for different operating speeds........ 40 Figure  2-13 Number of passenger vs. headway time........................................................ 45 Figure  2-14 Oil price scenarios generated by NZ Transport Agency Research ............... 47 Figure  2-15 Oil price after 2030 by SAUNER & LOPEX ............................................... 48 Figure  2-16 Daily ridership of the southbound movements in 2030 ................................ 49 Figure  3-1 Highway network in Western North America ................................................ 62 Figure  3-2 Rail network in Western North America ........................................................ 63 Figure  3-3 Rail and Highway network at study area ........................................................ 66 Figure  3-4 Alternative rail routes in study area ................................................................ 69 Figure  3-5 USA/BC Rail Border Crossings...................................................................... 72 Figure  3-6 Export via BC-USA Customs ports by Rail in 2007\ ..................................... 72 Figure  3-7 Highway Border Crossings ............................................................................. 75 Figure  3-8 Export via BC-USA Customs ports by Truck in 2007.................................... 75 Figure  4-1 Land use’s cost surface for the Okanagan Valley........................................... 98 Figure  4-2 Slope’s cost surface for the Okanagan Valley .............................................. 100 Figure  4-3 Geology's cost surface in the Okanagan Valley............................................ 102 Figure  4-4 Soil's cost surface in the Okanagan Valley ................................................... 103 Figure  4-5 Road’s cost surface in the Okanagan Valley................................................. 104 Figure  4-6 Hydrology’s cost function in the Okanagan Valley...................................... 105 Figure  4-7 Provincial park's cost surface in the Okanagan Valley ................................. 106 Figure  4-8 Suitability cost surface .................................................................................. 108 Figure  4-9 Proposed alignment for Okanagan Railway Kelowna-Osoyoos................... 109 Figure  4-10 Proposed alignment for Okanagan Railway Vernon-Kelowna................... 110 Figure  4-11 Selected alignment for Okanagan Railway................................................. 111 - x - Figure  4-12 Unit cost of a new track on the existing ROW............................................ 114 Figure  5-1 Benefits of the railway .................................................................................. 124 Figure  5-2 Collisions vs VKT......................................................................................... 125 Figure  5-3 Population density (per square kilometre) in Okanagan Valley ................... 128 Figure  5-4 Permanent and temporary traffic counting locations in Highway 97 ........... 130 Figure  5-5 Number of collisions in Highway 97 ............................................................ 132 Figure  5-6 Validation segments...................................................................................... 134 Figure  5-7 6 Yrs collisions prediction for V1 and V2 .................................................... 135 Figure  5-8 Safety savings of Okanagan Railway (Thousands of 2005$) ....................... 139 Figure  6-1 Break even chart............................................................................................ 147 Figure  C-1 1999 NRS Canadian analysis zones ............................................................. 171 Figure  C-2 FAF analysis areas........................................................................................ 171 Figure  E-1 Probability distribution for Scenario "a" ...................................................... 178 Figure  E-2 Probability distribution for Scenario "b" ...................................................... 178 Figure  E-3 Probability distribution for Scenario "c" ...................................................... 179 Figure  E-4 Probability distribution for Scenario "d" ...................................................... 179 Figure  E-5 Probability distribution for Scenario "e" ...................................................... 180 Figure  E-6 Probability distribution for Scenario "f"....................................................... 180 Figure  E-7 Probability distribution for Scenario "g" ...................................................... 181 - xi - ACKNOWLEDGEMENTS I offer my enduring gratitude to my supervisor, Dr. Gordon Lovegrove, for providing me with the opportunity to work in his research group, introducing me to the sustainable transportation world and advising me during the course of this project.The contributions of Kevin Woods, General Manager of KPR, as well as those of Andrew Albiston, Kean Lew, Ken Tseng, Kathryn Weicker and Ron Westlake are deeply appreciated.My very sincere thanks go to my friend and colleague, James Sun, for his help, support, and friendship. My deepest thanks go to my beloved Soheyl, and my lovely Mahboobe. Without their love and endless support, I couldn’t have come so far.   And above all, my thanks to my God, the beneficent, the merciful. - xii - DEDICATION To1 the1 memory1 of1 Cheng1 Ging1 Butt1 and1 all1 the1 brave1 people1 who1 endured1 harsh1 working1and1 living1 conditions1 to1 not1 only1 help1 build1 the1 Canada1 Pacific1 Railway,1 but1 unite1 a1nation,1as1well. 1- 1 - 1 INTRODUCTIONThe topic of this thesis is placement of a railway line through the Okanagan Valley in British Columbia, Canada, connecting Osoyoos and US railways in the south with Vernon and cross-Canada railways in the north.  This topic is motivated by an increasing demand for safer, faster, less energy intensive and less costly modes of transportation than the current dominating road-based, rubber-tire transport modes (i.e. autos and trucks) used for commuting, tourism and freight transport across North America. There are also high levels of congestion in rail and road-based freight shipping lines across the continent. Moreover, there are enormous social and economic costs associated with this congestion, and with the high number of road collisions.  Injuries due to road collisions have been recognized world-wide as a serious problem for decades.  Consequently, in this introductory chapter, sections 1.1 through sections 1.3 provide brief overviews of the Okanagan Valley from geographic, social, transportation and economic perspectives. Section 1.4 presents background information on the significance of the research problem and the motivations for this study. Section 1.5 states the research objectives. Section 1.6 describes the structure of this thesis.1.1 Study area The Okanagan Valley, the study area for this research, is located in Southern British Columbia, Canada, as seen in Figure  1-1. The Okanagan is comprised of three regional government districts, including: the North Okanagan (NORD), Central Okanagan (CORD), and Okanagan-Similkameen Regional Districts (RDOS). More specifically, the transportation corridor between Osoyoos and Vernon is the main geographic area of focus for this research. Throughout this corridor, there are approximately fifteen different- 2 - Figure  1-1 Study area  - 3 - communities and First Nations’ Bands. The following paragraphs describe a few of the major communities in terms of location, population and employment. Additional information regarding transportation is provided where available (Urban Systems Inc, 2004).OsoyoosOsoyoos is located at the junction of Highway 3 and Highway 97, just north of the Canada / United States’ border, in the RDSO. The town is located approximately 60 km south of Penticton and 400 km east of Vancouver.  In 2003, the population of Osoyoos was estimated to be 4,565; historically, its population growth rate has been less than 1% per annum, with year 2000 population estimated at 4,440 residents. PentictonThe City of Penticton is the largest community in the RDSO.  In 2003, its population was 32,519, corresponding to a 0.4% increase over the 2002 estimate. Including the surrounding area, the City’s 2001 population was 41,574. Penticton is approximately 395 km east of Vancouver and 70 km south of Kelowna. Kelowna  Kelowna is located approximately 400 km east of Vancouver, 70 km north of Penticton and 600 km west of Calgary, and is the regional economic centre of the CORD.  The 2003 population estimate for Kelowna was 103,421, a 1.7% increase over the 2002 population estimate, and averaging just over 1% each year for the past five years.VernonVernon is located approximately 46 km north of Kelowna, 60 km south of Salmon Arm and 117 km south of Kamloops and is the primary service centre of the NORD. Vernon is the second largest city in the Okanagan with an estimated population of 35,073 (2003). This corresponds to a 0.3% increase in population compared to the 2002 estimate. For comparison, the 2000 population estimate for the town was 34,657.  - 4 - 1.2 Transportation network in the study area The transportation network in the Okanagan Valley consists of highway corridors, airports, transit systems, and railway corridors, as outlined in Table  1-1.   1.3 History of rail transportation in the study area Rail was introduced to the region by the Kettle Valley Railway (KVR) in 1916. The KVR was operated in the Okanagan region, connecting the Kootenays (Southeastern portion of British Columbia) to the west coast of British Columbia. From Hope, the KVR made its way through the Coquihalla River Gorge. It continued to Princeton, West Summerland, Penticton, along the south side of Okanagan Lake and finally south to Midway, for a total of 525 km. The line was completed by the construction of branch lines to Merritt and Osoyoos, which connected the KVR to Canadian and American railway national networks (see Figure  1-2). Okanagan Lake itself was once used to transport rail cars from Kelowna to Penticton, but that rail ferry service was discontinued in the 1960’s, around the time that the KVR ceased operations in 1964.  Currently, the Kelowna Pacific Railway (KPR) is the only railway in the study area (see Table  1-1), and operates on a rail line leased from CN Rail, between Vernon and Kelowna.  There is no rail link from Kelowna to Penticton.  Car ferry service using the Lake was discontinued shortly after construction of the Okanagan Lake Bridge in 1958.The KVR downturn began in 1949, when Highway 3 (the Hope Princeton) was opened. During the next few years, operation of the KVR became uneconomical. In 1962, the Coquihalla section of the tracks was removed. In January of 1964, the last passenger train ran from Merritt to Penticton and onto Midway and finally Nelson. Figure  1-2 shows the railway network in the study area.- 5 - Table  1-1 Transportation network in the Okanagan Valley 1. Road Transportation: Highway  Highway 97, Canada/US Border to Vernon – 190 km of national (primary) highway  A four-lane highway (except for two lane between Summerland and Peachland) with acceptable level of service B and C 2. Road Transportation: Bus Lines Greyhound Lines (throughout the valley) Penticton Transit System (City of Penticton)  Summerland Transit System (Summerland to Penticton)  Okanagan-Similkameen Transit System (Penticton to Naramata) Princeton and Area Transit System (Penticton to Keremeos / Princeton) South Okanagan Transit Society (Osoyoos/Oliver/Okanagan Falls/Penticton/ Kelowna) Kelowna Regional Transit System (Peachland to Lake Country) Vernon Regional Transit System (Enderby to UBC-Okanagan, North of Kelowna) 3. Air Transportation: Penticton Regional – 1,829 meter runway, gross revenues of $24.7 million, 247 jobs Kelowna International Airport – 2,225 meter runway, 11th busiest airport in Canada, 2003 passenger volume of 863,645 passengers, estimated $265 million economic impact to the local regions, 1,834 full-time jobs  Vernon Regional Airport – 1,036 meter runway servicing local commercial operators, approximately 100 businesses based at the airport industrial park 4. Rail Transportation: Kelowna Pacific Railway (KPR) – freight only, 167 km of mainline track, 10.2 km of associated sidings and spurs, and approximately 40.9 km of running rights on Canadian Pacific Railway. The KPR connects with Canadian National Railway’s (CN) Kamloops Yard. The KPR connects to the Canadian Pacific Railway (CPR) through Okanagan Valley (OKAN) Railway1.4 Revitalization of railways in the Okanagan Valley North America is switching back to the railway as the solution to many issues of road transportation.  For the first time in nearly a century; railroads are making large investments in their networks (Machalaba, February 13, 2008). Therefore, it is reasonable - 6 - to re-consider the merits of freight and passenger railway in the Okanagan Valley. Motivations for re-consideration of a valley-wide railway are briefly discussed below. Figure  1-2 Railway network at southern British Columbia  1.4.1 Sustainable road safety (SRS): In 2006 alone on Highway 97 between Vernon and Osoyoos, there were 1,040 road collisions, including 4 fatal and 440 injury collisions. The directly resulting auto insurance claims were estimated at $14 million by the provincially-mandated Insurance Corporation of British Columbia (ICBC). This excludes the social costs of these collisions, such as incident delays to other road users, lost productivity, and lost wages. The usual mitigations to reduce collisions and congestion include adding road capacity (e.g. new road lanes) and changing road and/or interchange geometries. This study investigates another possible solution to the safety problems on Highway 97, namely, launching an auto-alternative mode of transportation in the valley to reduce Highway 97 traffic volume and congestion.  Past research has shown that reduced traffic volumes are directly associated with reduced collisions (Lovegrove, 2007). Initial investigations - 7 - suggested that the distance between the cities in the Okanagan Valley was too short for commuters, tourists, and freight shippers to use air transportation in a cost competitive manner. Alternatively, resurrection of a previously operated regional railway is a reasonable research consideration. 1.4.2 Sustainable economic prosperity: Local economic development Rail transit can provide a variety of economic development benefits (Litman, 2004). These benefits are summarized below.1. Shifting Consumer Expenditures Although no Okanagan data was found, studies have found that expenditures on automobiles, fuel and roadway facilities provide relatively little regional economic activity, because they are capital intensive and largely imported from other areas (Miller et al., 1999). They found that each 1% of regional travel diverted from automobile to public transit increases the regional income by about 2.9 million dollars, resulting in 226 additional regional jobs. Rail transit availability has been shown to be a significant motivating factor for the residents for using public transport. Studies reveal that residents of cities where rail transit is available and convenient annually spend an average of 448 dollars per capita less on average on transportation than the residents of cities where bus is the only form of transit, due to the higher use of public transport (Litman, 2004). In brief, rail transit is considered an efficient way of creating additional jobs and additional regional income. 2. Increased Property Values Eppli showed that rail transit systems help increase property value due to improved accessibility and liveability (Eppli et al., 2000). Table  1-2 reviews the property value increases measured with increasing proximity to rail transit stations in different European - 8 - and North American cities (Hass-Klau, 2003). This increased value is due to the fact that businesses and individuals prefer to be located near rail stations, in order to improve access for employees, customers and residents. 3. Community Redevelopment New development patterns tend to occur in new communities in suburban areas, abandoning older neighbourhoods, due to the overwhelming congestions in older urban neighbourhoods. Rail transit can facilitate urban redevelopment thus avoid sprawl by reducing automobile traffic volumes through urban areas. A unique transit service can popularize tourist activities and help create community identity which stimulates economic development (Litman, 2004). Table  1-2 Impacts of rail station proximity on property values (Hass-Klau, 2003)City Factor Difference Newcastle upon Tyne, UK House prices 20% Greater Manchester, UK Not stated 10% Portland, Oregon House prices 10% Portland, Oregon Residential rent >5% Strasbourg, France Residential rent 7% Strasbourg, France Office rent 9.85% Rouen, France Rent and houses 10% Hannover, New Hampshire Residential rent 5% Freiburg, Germany Residential rent 3% Freiburg, Germany Office rent 14.8% Montpellier, France Property values Positive, no figure given Orléans, France Apartment rents None-initially negative due to noise Nantes, France Not stated Small increase Nantes, France Commercial property Higher values Saarbrûcken, Germany Not stated None-initially negative due to noise Bremen, Germany Office rents +50% in most cases - 9 - Develop tourism: a more convenient mode of transport The tourism industry is a vital part of the economic structure employing 11.8% of the Okanagan labour force (Okanagan Partnership Society, 2008). A tourism consultant study on a train tour in the region, conducted by IBI Group, concluded market potential exists for an additional high quality, moderately priced tour train excursion in the Okanagan Valley (IBI GROUP, 2007). Moreover, a railway can improve tourism by facilitating transportation between the cities of the valley. A transit planning consultant study revealed that a passenger rail, mainly designed for commuters, could be utilized to support tourism and other recreational activities, thereby adding to the trade within a region (Urban Systems, 2004). Local demand Local businesses first recognized the need for new infrastructure in the valley, and they are willing to support and contribute to this project. The initial idea of building a railway in the valley came from the Okanagan Partnership, a community-led, non-profit society dedicated to the vision of a prosperous and sustainable Okanagan region. It recognized that it is not possible to rely on only one transportation corridor (Highway 97) and expect that the level of prosperity of the Okanagan Valley to be sustained. 1.4.3 Sustainable ecological footprint Rail transit can provide considerable energy conservation and emission reduction. Rail travel consumes about a fifth of the energy-per-passenger-km as automobile travel (Shapiro et al., 2002). Moreover, electric powered rail produces minimal pollution and noise emission.  International studies indicate that per capita energy consumption declines with increased transit use (Kenworthy et al., 2000). Freight rails also can be very effective in providing a sustainable footprint. This fact has motivated the British Columbia Provincial Climate Action Plan team to issue a policy - 10 - and recommend enhancing the role of rail in moving freight in B.C. The plan proposes a greater use of rail transportation, which offers the potential for significant reductions in greenhouse gas emissions (B.C. Climate Action Team, 2008). 1.4.4 Wider relevance: As an exporting nation, Canada needs an integrated and efficient national transportation system, and so does British Columbia because it owns the most active ports of Canada. However, the principal freight railway system between British Columbia and its United States neighbours, Washington and Idaho states, is experiencing increasing traffic.  American studies have shown that railways at the border (between British Columbia and Washington and between British Columbia and Idaho) would be highly congested and might reach a Level of Service (LOS) of “E” or “F” (high congestion) in 2035, if no further improvement takes place (see Figure  1-3) (Association of American Railroads, 2007).Figure  1-3 LOS in 2035 without any improvements Source: (Association of American Railroads, 2007)- 11 - Other American studies have shown highways adjacent to railways would experience high traffic congestion volume in 2035 as well (see Figure  1-4) (Federal Highway Administration, 2007).  Figure  1-4 Peak-period congestion on the national highway system in 2035 Source: (Federal Highway Administration, 2007)While no similar data or forecasts were found on Canadian railway and highway segments near to and crossing the border, it can be reasonably assumed that congestion on Canadian segments will be similar to American segments.  The Canadian Government has already started to plan to solve them (Government of Canada, 2006). Long combination vehicles (LCV), trucks with two trailers are currently being demonstrated on highways in British Columbia, Alberta and Nova Scotia as a possible mitigation (Nova Scotia Department of Transportation and Infrastructure Renewal, 2008; Williams, 2007).  These vehicles allow truck drivers to carry additional payload, increasing their profits per trip, while reducing traffic and Greenhouse Gas emissions on our roads (Live Smart BC, 2008). However, they impose significant costs on the highway infrastructure, Therefore, the seven-year “Building Canada” plan suggests that investing in short-line railways may mitigate increased congestion. The plan contends renewed and upgraded short-line - 12 - railway infrastructure will provide better connections to mainline corridors, improve market access for rural shippers, strengthen rural economies and employment opportunities, and reduce stress on the regional road infrastructure caused by long haul freight truck transportation (Government of Canada, 2007). As Figure 1-5 shows there is a gap of roughly 100 kilometres between American railways ending in the town of Oroville at the US/Canada border, and Canadian railways ending in Kelowna.  A new ‘short-line’ railway between Oroville and Kelowna could connect US and Canadian railways, and thereby address rail and truck traffic congestion and transportation infrastructure concerns cited in the “Building Canada” plan. 1.5 Objectives of the research  Possible motivations for a passenger railway in the Okanagan Valley as an alternative to Highway 97 freight, commuter, and tourism traffic have been discussed. A railway connecting Canadian and American railways is one possible solution to existing and forecast congestion and infrastructure problems, and might also help local industry to flourish and sustain its prosperity. This research has sought to provide a science-based examination of the practicality of an Okanagan Valley railway.  Specifically, the objectives that this research set out to achieve were as follows: 1. Develop a set of macro-level collision prediction models that provide empirical, science-based evidence of the road safety benefits of railways, using data from the Okanagan Valley and Highway 97. 2. Conduct a comprehensive literature review on the design issues, costs, and benefits of railway freight and passenger systems, with particular focus on relevance to ‘short-line’ railways and the Okanagan Valley. - 13 - 3. Conduct a social cost-benefit analysis of an electric railway implemented to serve the Okanagan Valley. Figure  1-5 Freight rail network at West Coast 1.6 Structure of the thesis The results of this research have been presented in five chapters. Chapter 1 provides a brief introduction of the parameters of this research, including the study area, motivations, research objectives and organizational structure of this report.  In Chapter 2, passenger demand for the Okanagan Railway is estimated. The first regional mode choice model for the Okanagan Valley is developed. It is then applied to the trip estimates of the Okanagan Valley Transportation Plan -- which predicts the number of trips between the cities in the valley in 1994 and 2020 -- service is planned for the commuter railway using the demand forecast results.  - 14 - Chapter 3 focuses on the freight demand forecast for the Okanagan Railway, estimating the Origin-Destination table, mode choice analysis, and traffic assignment. In addition, this chapter reviews available data sources for the freight demand forecast in the British Columbia interior and the literature on the different methods of freight demand forecast. In Chapter 4, the cost of the Okanagan Railway is estimated. First, a comprehensive literature review explores the design criteria for the freight and passenger railroads. Possible criteria, such as, geology, hydrology, slope and socio-economics, are studied and their relative costs are determined. The least cost path is defined using the Analytical Hierarchy Process (AHP) in Geographic Information System (GIS). Capital costs for the selected path are calculated after reviewing the cost estimates of similar projects in the literature.In Chapter 5, all railway benefits, including safety benefits, are estimated. In order to quantify the safety benefits of the Okanagan Railway with reasonable accuracy, a macro-level collision prediction model (CPM) exclusive for Highway 97 is developed. The safety benefits of the Okanagan Railway, then, are estimated using the forecasted demand and the developed CPMs.In Chapter 6, a social cost-benefit analysis is conducted using the estimated costs and benefits of the railway and employing the process suggested by the Canadian Treasury Board for analysis of public projects. Finally, a Monte Carlo simulation of the results is performed to determine the sensitivity of the results to the assumptions. Chapter 7 contains a summary of the research, together with conclusions, contributions, and future research topics.- 15 - 2 PASSENGER DEMAND FORECASTS 2.1 Introduction Demand forecasting involves estimating the travel demand for a specific transportation facility or infrastructure. Passenger demand for the Okanagan Railway is estimated in this chapter, in three sections. In Section 2.2, five successful demand forecast models which include a commuter rail in the mode choice step, are reviewed. In Section 2.3, the methodology for the passenger demand forecasting for the Okanagan Railway is discussed. In section 2.4, a model for the Okanagan region is provided, and then passenger demand for the Okanagan Railway is estimated. 2.2 Literature review The traditional four-step travel demand forecast modeling process has been used for this research, including Trip Generation, Trip Distribution, Mode Choice, and Trip Assignment. Five existing travel demand forecast models were reviewed.  A commuter rail component was included in the mode choice step for each model. Two of these models are statewide models that have been used for predicting passenger demands for existing transportation infrastructure including regional/commuter railways. The others involve ridership forecasts for three commuter railways selected because they had the most similar attributes to the proposed Okanagan railway. Using Okanagan Valley data, the models were used in case studies to facilitate choosing the right model for forecasting the passenger demand for the Okanagan Railway.  - 16 - 2.2.1 Statewide models Commuter/regional railways in Florida and Oregon use statewide transportation models to forecast the ridership.  Commuter/regional railways have been considered as viable alternative modes of transport in these states for some time, and hence have been included in their models.  Florida Standard Urban Transportation Modeling Structure (FSUTMS) FSUTMS uses the conventional four-step model, with some adjustments, to predict the travel demand in South Florida (Zhao, 2003).  To estimate the share of trips in the mode choice step, FSUTMS has recently developed three mode choice models based on the purpose of trips: home based work trips (HBW), home based other (HBO), and non-home based (NHB). The mode choice models use a three-level nested logit model as shown in Figure  2-1. Figure  2-1 Florida four-level nested model Source: (Abdel-Aty, 2001)- 17 - The mode choice models use three sets of utility parameters including nesting coefficients, mode-specific constants, and level-of-service coefficients. Table  2-1 shows the list of the FSUTMS’ utility parameters (Abdel-Aty, 2001). Table  2-1 FSUTMS 's utility parameters Level-of-service Coefficients Nesting Coefficients Mode Specific Constants Transit Walk Time (TWT)   Transit Auto Access Time (TAAT)Transit mode Auto Driver Transit Run Time (TRT) Highway auto mode Auto Passenger Transit First Wait (TFW) Shared ride mode Walk to local bus (LB) Transit Transfer (2nd wait) Time (TTT)Other nests Walk to express bus (EP,WK) Transit Number of Transfers (TNT) Walk to metro rail (MR,WK) Transit Fare (TF)  Walk to tri rail (TR,WK) Highway Terminal Time (HTT)  Drive to express bus (EP,DV) Highway Run Time (HRT)  Drive to metro rail (MR,DV) Highway Auto Operating Costs (HAOC)Highway Parking Costs (HPC)   HOV Time Difference (HTD)Transit and auto utility functions for this model have been shown in Appendix A. Oregon Transitional Model The Oregon Transitional model is a part of a comprehensive “land use/transport” model in Oregon State called the Transportation and Land Use Model Integration Program (TLUMIP) (PB Consult Inc., 2006). The model’s Passenger Transport (PT) module includes the passenger transportation model, consisting of two sub-components: Short - 18 - Distance Transport (SDT), which predicts all regular work commutes and travel patterns less than or equal to 80 km, and Long Distance Transport (LDT), which predicts non-commute travel patterns greater than 80 km. Figure  2-2  provides an overview of the SDT and LDT models.Figure  2-2 Person Transport Model Flow Diagram Source: (PB Consult Inc., 2006)In this section, the mode choice model for LDT in the PT module has been explained in further detail, as it can be used as a reference to establish a mode choice model for the - 19 - Okanagan Railway. The LDT mode choice model is a two-level nested logit model which includes existing intercity transits (Greyhound and Amtrak services) and two high speed rail alternatives which are available if they are useful in future model applications as shown in Figure 2-3.Figure  2-3 Oregon mode choice model nesting structure  Source: (PB Consult Inc., 2006)Based on the type of passengers and purpose of the trips, three sets of models have been produced: 1) HH: household (travel in which the entire household participates), 2) WR: work-related (individual business travel), and 3) O: other (individual travel for non-work purposes). The mode choice models use three sets of utility parameters: nesting coefficients, mode-specific constants, and level-of service coefficients. Table  2-2 shows the full list of the utility parameters in LDT (PB Consult Inc., 2006). Table  2-2 Utility parameters for the Oregon’s long distance travel model Level-of-service Coefficients Nesting Coefficients Mode Specific Constants In-Vehicle Time (IVT) Transit-High Speed Rail Air  Walk-Access Time (WAT) Transit Walk-Drive  Transit-Walk Access  Drive-Access Time (DAT) High Speed Rail Walk-Drive  Transit-Drive Access  Wait Time (WT)  High Speed Rail-Walk Access Cost (C)  High Speed Rail-Drive Access  - 20 - Transit and auto utility functions of the Oregon model are shown in Appendix A. 2.2.2 Regional modelsIn this section, the ridership forecast models of three commuter/regional railways, available in the literature, are studied including; CSX, a proposed commuter rail in Ohio; Sonoma Marin Area Rail Transit, a proposed passenger rail service in California; and the Svealand line, a high speed regional railway in Sweden. CSX Commuter Rail in Ohio A Commuter Rail on the tracks of the CSX freight railway has been proposed by the North-South Transportation Initiative as an alternative transit mode in peak hours. CSX, 140 km in length, would operate between Piqua, Ohio, and downtown Cincinnati, Ohio, with 2007 estimates indicating a total resident population of 3.61 million. The commuter railway would serve 14 stations with two stations located at major cities of Dayton and Cincinnati. The suggested maximum speed for CSX is 125 km per hour. Ohio-Kentucky-Indiana Regional Council of Governments (OKI) and Miami Valley Regional Planning Commission (MVRPC) have developed a joint model, called the OKI/MVRPC regional travel demand model. In the CSX commuter railway proposal, this model has been employed to forecast the ridership. The OKI / MVRPC demand model is based on the conventional four-step model, complemented with a sub-model to forecast the generated trips at two important stations as shown Figure  2-4.  The OKI / MVRPC mode choice model is a three-level nested logit model as shown in Figure  2-5. The CSX Commuter Rail is studied in the Commuter Rail nest. Four mode choice models are developed in the OKI / MVRPC model, based on the purpose of the trips, including home-based work (HBW), home-based university (HBU), home-based other (HBO), and non-home based (NHB).  - 21 - Figure  2-4 OKI/MVRPC Model Components Source: (Parsons Brinckerhoff Ohio, 2004)Figure  2-5 OKI / MVRPC mode choice model structure Source: (Parsons Brinckerhoff Ohio, 2004)- 22 - The mode choice models use three sets of utility parameters, including level-of-service coefficients, nesting coefficients, and mode-specific constants. Table  2-3 shows the full list of the parameters that are used for calculating CSX mode share (Parsons Brinckerhoff Ohio, 2004). Table  2-3 CSX's utility parameters Level-of-service Coefficients Nesting Coefficients Mode Specific Constants In-Vehicle Time (IVT) Shared Ride P&R Access, Suburb-to-Central Business District (CBD) First Wait Time (FWT) Transit Submodes K&R Access, CBD Transfer Wait (TW) Transit Access Walk Access, CBD Walk Time (WT)  P&R Access, other Drive Access Time (DAT)  K&R Access, other Fare (F)  Walk Access, other Terminal Time (TT)   Auto Operating Costs (AOC)   Parking Costs (PC)   Transit and auto utility functions of the OKI model are shown in Appendix A. Sonoma Marin Area Rail TransitSonoma Marin Area Rail Transit (SMART) is a proposed passenger rail as an alternative to Highway 101 in Northern California. SMART, 110 km in length, would serve 14 stations, which are in the relatively small towns between Cloverdale, California and Larkspur, California. The suggested maximum speed for SMART is 125 km per hour. While SMART connects to the city of San Francisco via the existing Larkspur Ferry, it is projected that only a few dozen SMART passengers each day would continue into San Francisco (SMART, 2008). SMART employs Marin County’s 2020 daily home based work trip table from the county’s EMME2 highway and transit model to create the Origin-Destination (O-D) table. To be included into the O-D table, jobs of the commuter rail riders should be within an approximate 3 to 4 km radius or a 15-minute shuttle time connection of a station. - 23 - Instead of the traditional EMME2 mode choice model components, due to the nature of commuter rail services, SMART has used a set of mode split factors as shown in Table  2-4. It assigns a mode share factor to the trips based on their distances (Schiermeyer Consulting Services, 2002). Table  2-4 SMART’s mode split factors Origin Station to Destination Station (km) Mode Split Percentage Applied 16 – 24 2.5% 25 – 32 5% 32 – 48 10% 48 – 56 15% 56 – 64 20% 64 + 25% Data source: (Schiermeyer Consulting Services, 2002)Svealand line Svealand line (Svealandsbanan in Swedish) is a regional high speed railway, 114 km in length which was opened in 1997. It runs between Södertälje and Valskog, via Eskilstuna which are small towns in Mälaren Valley, Sweden, with 2008 estimates indicating a total resident population of 2.15 million. At Södertälje’s end, the line connects to Stockholm, Sweden capital, via Stockholm commuter rail (see Figure  2-6).  Figure  2-6 The Mälaren valley and surroundings Railways in 2002 (Svealand line thick black) Source: (Fröidh, 2003) - 24 - The maximum speed for the Svealand line is 200 kilometres per hour. This means that regional high-speed trains can cover the distance between Eskilstuna and Stockholm (115 km) in just an hour. Unfortunately, there is not data available about the travel demand modeling of this regional high speed railway. However, the actual ridership of the railway at the two stations almost in the middle of the line, Strängnäs and Läggesta is provided as shown in Table  2-5. The transit shares before and after the opening of the railway are also shown in the table.  Table  2-5 Mode share for all regional travels at the two Svealand line stations Station Name Svealand line share at 2000 Public transport share at 2000 Public transport at 1996 Strängnäs 28-30% 30-32% 18% Läggesta 29-31% 30-32% 14% Source: (Fröidh, 2003) Mode shares of rails from all the travel between the largest cities on the line, Eskilstuna / Strängnäs and Södertälje / Stockholm, are shown in Table  2-6. Table  2-6 Mode share for all regional travels between Eskilstuna/Strängnäs and Södertälje/Stockholm Mode 1996 2000 Car  85% 62-67% Public transport - buses 15% 0-5% Public transport - high speed trains 0 33-38% Regional journeys per year, in millions 3.0 3.7 – 4.2 Source: (Fröidh, 2003)Table  2-5 and Table  2-6 show about a 30% to 35% share for the Svealand line. Figure  2-7 shows the previous mode of the travel by current passengers of the Svealand line.  Note that “new journeys” refers to the journeys in the corridor that were made after the introduction of the Svealand line due to transport enhancements (Fröidh, 2003) - 25 - BusCarNew journeys generatedFigure  2-7 Svealand line passengers’ previous mode of travel  Source: (Fröidh, 2003)2.3 Methodology for demand forecasting for the Okanagan Passenger Railway To forecast the passenger demand for the Okanagan Railway, a travel forecast model similar to the models described in the previous section is needed. As mentioned in the previous section, regional travel demand forecast models are usually used to predict the demand for the proposed infrastructure in the area or at least to produce the O-D table of potential demand. The Okanagan Valley Transportation Plan (OVTP) model, which is the only regional travel model currently available in the Okanagan Valley, produces an O-D table for the passenger demand between the cities of the Valley, but it does not include a mode choice model. Considering the model was published in 1997, this study suggests a comprehensive travel forecast model needs to be developed including all of the four steps for future reference in the region. Nevertheless, for the purpose of this study, passenger demand for the Okanagan Railway was forecast using: 1) the 1997 OVTP model; 2) the most recent survey in the Okanagan Valley; and, 3) the reviewed mode choice models.  - 26 - 2.3.1 Trip generation & distribution The first step, trip generation, was needed to produce a matrix of inter-zonal trips between each origin and destination zone. Because the demand for the Okanagan Railway would come from intercity travels, each city in the valley was considered as one zone. Thus the number of trips between each pair of cities in the Okanagan Valley was required. The OVTP divided the Okanagan region into zones, and forecasts inter-city trips for 1994 and 2020 as shown in Table  2-7 and Table  2-8. Table  2-7 Intercity trips by OVTP model in 1994 From/To  Osoyoos  Penticton & Summerland  Peachland & Westbank Kelowna  Vernon  Osoyoos  1,427 2,387 81 289 10 Penticton & Summerland  2,686 1,046 1,112 3,119 58 Peachland & Westbank  75 1,631 1,227 14,230 477 Kelowna  171 3,076 14,283 6,431 4,268 Vernon  17 92 892 3,338 1,352 Table  2-8 Intercity trips by OVTP model in 2020 From/To Osoyoos  Penticton & Summerland  Peachland & Westbank Kelowna  Vernon Osoyoos  2,012 3,882 158 500 11 Penticton & Summerland  4,368 1,696 2,309 6,164 70 Peachland & Westbank 133 3,713 2,911 29,482 923 Kelowna  292 6,370 29,631 15,133 8,670 Vernon  20 108 1,762 6,364 3,363 2.3.2 Mode choice There was no mode choice model produced for the Okanagan Railway, so one was ‘borrowed’ from those found in the literature. Rossi showed that transferring mode choice model parameters in the absence of the local data is practical and, on average, transferred parameters seem to fit in the new models. However, it should be mentioned - 27 - that the complete model should be transferred, since variables may be correlated with one another (Rossi et al., 2002). That is, mode choice model should be adapted from the most similar region. Moreover, the other attributes of the selected railway – e.g. overall length, speed, frequency, and distance between stations – should be completely considered. 2.3.2.1 Model selection Table  2-9 presents the attributes of the reviewed railways. The length of all the reviewed railways is relatively similar to the distance between Vernon and Osoyoos. However, only the population of areas around the SMART line is comparable to the Okanagan Valley population.Table  2-9 Attributes of reviewed railways Name of rail Length (km)Max speed  (km/h) Population of cities around the rail (millions) Number of stationsShare tracks with freight Tri rail 116 110 5.50 18 YES CSX Line 143 110 3.61 14 NO SMART 110 110 0.39 12 YES Svealand line 114 200 2.15 9 YES As shown below in Figure  2-8, all of the railways have a mega city, with a metropolitan population above two million people, as a station or close to a station, which increases the values in the fourth column of Table  2-9 drastically (note that there are small towns at both ends of the SMART line with populations of less than 50,000 that are not shown). Studies show that a large portion of the Svealand line market comes from Stockholm using the connecting light rail between Södertälje and Stockholm. Therefore, the population of Stockholm has been added to the population of the cities around Svealand line, despite the fact that the Svealand line is not directly connected to those mega cities. However, it is projected that only a few dozen SMART passengers each day would continue their trip onto San Francisco. Therefore, the population of San Francisco has not been added to the population of cities around SMART. - 28 - Figure  2-8 Distribution of population around the reviewed railwaysFor comparison, the market population for the Okanagan Railway originates in three regional districts, with 2004 populations of 76,678 (NORD), 159,333 (CORD), and 77,904 (OSRD).  By 2031 the population of these three districts has been predicted to increase by 38% growth (Population Section, BC Stats., 2005).  Kelowna, as the major city in the Valley, had a metropolitan population of 114,731 in 2007 and it is expected to add approximately 85,000 to its population by 2031 (Neale, 2005). Kelowna would not be a market comparable with mega cities such as Miami with a population of 2,387,170 people in 2007, Cincinnati with a population of 2,133,678 people in 2007, or Stockholm with a population of 1,949,516 people in 2007. - 29 - It was concluded that the SMART line’s attributes were the most similar to those of the Okanagan Railway’s, and the SMART mode choice factors, presented in Table  2-4, could be used for the purpose of mode choice in this study. However, no information on how SMART had developed these factors was found in the literature, or by contacting the Rail Planning Manager of the project. Therefore, instead of selecting a model, it was decided to develop a model for the Okanagan area by studying the three available mode choice models, and modifying and calibrating them. As stated earlier, Rossi showed transferring mode choice model parameters in the absence of local data is acceptable (Rossi et al., 2002).2.3.2.2 Mode choice model development As mentioned, the Florida and OKI models split the trips based on their trip purpose. Florida divides them into three groups of home-based work, home-base non-work, and non-home-based trips. OKI includes a group of home-based university trips in addition to Florida groups. However, the Oregon model groups the trips based on traveler characteristics, including household (travel in which the entire household participates), work-related (individual business travel), and other (individual travel for non-work purposes).The OVTP model was not as sophisticated, and combined all the trip purposes into one group. In order to apply a mode choice model to the O-D table, a second data source – the 2007 North and Central Okanagan Household Travel Survey provided by the city of Vernon – was used. The survey identified the purpose of the inter-city trips in the Okanagan Valley and divided the trips into eleven groups as shown in Table  2-10. Since the trip classification in the survey was based on the purpose of the trips (like Florida and OKI models) rather than characteristics of the traveller (like Oregon model), it was not possible to apply the Oregon model to the O-D table to estimate the rail share. Therefore, it was decided to remove Oregon from the potential mode choice models and develop an Okanagan mode choice model based on the OKI or FSUTMS models.  - 30 - Table  2-10 Trips purposes in "2007 Okanagan Travel Survey" Code Description 1  To Work - From Home 2  To Work - From Other 3  To School - From Home 4  To School - From Other 5  From Work - To Home 6  From Work - To Other 7  From School - To Home 8  From School - To Other 9  To Home - From Other 10  From Home - To Other 11  All Others (To Other - From Other) As Figure  2-1 and Figure  2-5 show, the transit nests in FSUTMS and OKI models have been divided into several sub-nests, which are the available transit modes in the area. However, bus transit is currently unavailable or infrequent between the cities in the Okanagan Valley. Therefore, it was decided to keep the nesting structure very simple and not divide it into any sub-nests. Figure  2-9 shows the proposed nesting structure for the Okanagan Valley. Although it is simple, it provides this study with the necessary information and it has the potential to be developed further in future research. Figure  2-9 Okanagan mode choice model nesting structure Table  2-12 shows the values for the coefficients of the Florida and OKI mode choice models. It was observed that most of the variables were analogous between the two models. The variables also shared similar values. This added evidence in support of the assumption that transferability was valid.  Table  2-13 shows the values for the transit and - 31 - auto constants. The Florida constants are grouped based on the number of cars per household, while all modal constants are normalized with respect to the drive alone mode. OKI constants are divided into four groups based on the number of workers and cars in the households. In this model, constants are used only for the purpose of model calibration.In the 2007 North and Central Okanagan Household Travel Survey, the origins and destinations are grouped into 8 zones, including the City of Vernon (1), Coldstream & Lumby (2), Spallmucheen & Armstrong, & Enderby (3), Lake Country (4), Westside & Westbank & Peachland (5), Central Kelowna (6), Suburban Kelowna (7) and outside the region (8).  An active transit system is only working between Central Kelowna (6) and Suburban Kelowna (7), where survey shows that there is a substantial transit share (comparing to other zones) for HBW, HBO and NHB trips. Table  2-11 shows the data from the survey on the trips between zones 6 and 7 and the transit share at 2007.Table  2-11 Data on transit and auto trips from the Okanagan survey HBW HBO NHB Average transit share 2.06% 1.37% 1.47% Average run time for auto (min) 16.36 15.06 14.45 Average run time for transit (min) 35.92 33.46 25.14 Average auto occupancy 1.18 1.79 1.71 Data source: (Winram, 2007) To understand how the Florida and OKI mode choice models would predict the trip mode shares in the Okanagan Valley, both models were employed to predict the trip mode for HBW, HBO and NHB trips between zones 6 and 7. As there was no survey showing regional values for several coefficients of the Florida and OKI mode choice models such as transit waiting time, transit walk access time, transit transfer wait time, or transit fare, the average value for Okanagan transit was used as shown in Table 2-14. Auto operating costs were provided by the Canadian Automobile Association’s Driving Costs (2008), which was based on the national average gas price (as of December 2007) of 110.1¢ per litre, and an annual driving distance of 18,000 kilometres. The fuel costs reflect the - 32 - purchase of unleaded, regular-grade gasoline, based solely on self-service gasoline prices (Canadian Automobile Association, 2008). The 2007 Okanagan survey showed that 25% of automobiles were SUVs or Minivans, 16% were pickup trucks and the remaining 59% were Sedans. Since there was no data on the operating cost of trucks, it is assumed to be similar to Sport Utility Vehicle (SUV) and Minivan operating costs. Unlike the mega cities, in all the cities of the Okanagan, even Kelowna, terminal parking times and parking costs are not determining factors (Groundworks, 2000). Therefore, for all the auto trips, one dollar has been considered as the parking cost, which is enough for two hours of on-street parking and one hour in a metered lot. All of the parameters are shown in Table  2-14..- 33 - Table  2-12 Mode choice model coefficients for FSUTMS and OKI models Home Based Work trips Home Based Other trips Non-Home Based trips  Mode Choice Model Coefficients FL OKI FL OKI FL OKI Transit In-Vehicle Time  -0.02 -0.0248 -0.015 -0.0085 -0.018 -0.0265 Transit Walk-Access Time  -0.045 -0.0876 -0.035 -0.0169 -0.045 -0.0623 Transit Drive-Access Time  -0.02 -0.0248 -0.015 -0.0085 -0.018 -0.0588 < 7 min           -0.045 < 15 min      -0.0409 -0.035 < 15 min      -0.0169 -0.045 < 15 min          -0.0405 Transit Wait > 7min          -0.023 > 15 min       -0.0248   > 15 min       -0.0085   > 15 min         -0.0265 Transit Fare  -0.0032 -0.0021 -0.0048 -0.0017 -0.0048 -0.003 Transit Transfer Wait -0.045 -0.0461 -0.035 -0.0169 -0.045 -0.0301 Transit Number of Transfers  -0.045   -0.035   -0.045   Drive Access Time / In Vehicle Time    -1.25   -1.25   -1.25               Auto Run Time -0.02 -0.0248 -0.015 -0.0085 -0.018 -0.0265 Auto Parking Time -0.045 -0.0248 -0.035 -0.0085 -0.045 -0.0265 Auto Operating Cost -0.0025 -0.0021 -0.0048 -0.0017 -0.0048 -0.003 Auto Parking Cost -0.0032 -0.0021 -0.0048 -0.0017 -0.0048 -0.003 -33-- 34 - Table  2-13 Values for the transit and auto constants for FSUTMS and OKI models FL   OKI HBW HBO NHB  HBW HBO NHB Auto Constant for auto passenger     0.5043       -0.697   Zero cars per household 1.2626 0.7173   zero cars 0.187 1.07     One car per household -1.1834 0.7564   worker<car 0.29 2.12     Two+ car per household -1.4036 0.7443   worker>car 0.599 2.673             worker=car 1.125 3.064   Transit Constant       -1.91435       1.398   Zero cars per household -7.2301 -15.758   zero cars -1.492 -0.788     One car per household -1.1613 -1.6495   worker<car -0.532 -1.082     Two+ car per household -1.5579 -2.0545   worker>car -0.634 -1.118             worker=car -0.527 0.125   -34- - 35 - Table  2-14 Mode choice coefficients values used in calibration as the Okanagan Values Transit Value Transit Walk-Access Time  5 min Transit Drive-Access Time  0 min Transit Wait (average) 15 min Transit Fare  200 ¢ Transit Transfer wait  0 min  Transit Number of Transfers  0   Auto Auto Terminal Parking Time 5 min Auto Parking Cost 100 ¢ Auto Average Speed in City 30 km/h Auto Operating Cost (per km) Cobalt LT Grand CaravanFuel 9.95¢ 12.97¢ Maintenance  2.36¢ 2.82¢ Tires  1.49¢ 1.91¢ Total 13.80¢ 17.70¢ Data source for auto operating cost: (Canadian Automobile Association, 2008)Table  2-15 shows the predicted transit share between zones 6 and 7 by the FSUTMS and OKI models, and compares them with the observed share. The FSUTMS model predicts the shares, especially HBO and NHB trips share, with less error. It is concluded that the FSUTMS model fits better in the Okanagan area. Therefore, the FSUTMS model has been chosen as the base model for developing the Okanagan mode choice model. Table  2-15 Observed and modeled transit share between zones 6 and 7  HBW HBO NHB Observed transit share  2.06% 1.37% 1.47% Modeled transit share  by OKI model 7.67% 2.16% 4.09% Error percentage 73% 37% 64% Modeled transit share  by FSUTMS model 4.62% 1.21% 1.37% Error percentage 55% 13% 7%  - 36 - 2.3.2.3 Calibration Since the data from the 2007 survey was not enough to calibrate all the mode choice coefficients and constants, it was decided to keep all the coefficient values constant, and only change the auto and transit constants to calibrate the model. Similarity of the coefficient values between OKI and FSUTMS models also supported this decision.Through the process of calibration, it was discovered that the auto constant and the transit constant were not independent, but that there was a linear relationship between their exponentials (between TrCe  and AuCe ) as seen in Equation 2-1. The equation has been explained in further detail in Appendix B.)...(...1......1221122112211AuCaaaabbTreeCeeCLnC DDDDEE     ( 2-1) Where,1C  is the transit mode share at 2007 survey; TrU is the transit utility function; AuUis the auto utility function; iD is ith parameter in transit utility function; iais the coefficient for iD ;TrCis the constant in transit utility function; iE is ith parameter in auto utility function; ibis the coefficient for iE ; and AuCis the constant in auto utility function.  - 37 - Therefore, it was only necessary to satisfy Equation 2-1, as the values of the transit and auto constants were not individually important. For ease of possible future research and later comparisons, it was decided to keep the highway coefficient the same as the FSUTMS model and only change the transit coefficient. Table  2-16 shows the calibrated model. This model would later be used to determine the share for the Okanagan Railway. Table  2-16 Okanagan mode choice model    HBW HBO NHB Transit In-Vehicle Time  -0.02 -0.015 -0.018 Transit Walk-Access Time  -0.045 -0.035 -0.045 Transit Drive-Access Time  -0.02 -0.015 -0.018 < 7 min       -0.045 Transit Wait > 7min        -0.023 -0.035 -0.045 Transit Fare  -0.0032 -0.0048 -0.0048 Transit Transfer Wait -0.045 -0.035 -0.045 Transit Number of Transfers  -0.045 -0.035 -0.045 Transit Constant -1.98 -1.74 -1.55 Auto Run Time -0.02 -0.015 -0.018 Auto Terminal Parking Time -0.045 -0.035 -0.045 Auto Operating Cost -0.0025 -0.0048 -0.0048 Auto Parking Cost -0.0032 -0.0048 -0.0048 Auto Constant 0.83 0.94 0.9 2.3.2.4 Validation2007 survey data showed that transit was an active mode for HBW trips between zones 5 and 6 (it was not the case for HBO and NHB trips). These data, which were not used in the calibration process, were used to validate the calibrated FSUTMS HBW model, as shown in Table  2-17.Table  2-18 shows the validation results. The calibrated FSUTMS model or Okanagan model predicted the transit share with 13% error. Overall, this process validated the  - 38 - accuracy of estimates with the Calibrated FSUTMS model (Okanagan model), while it showed there is plenty of room for improvement.   Table  2-17 Mode choice coefficients’ values used in the validation process Transit Value Transit Run Time 47 min Transit Walk-Access Time  5 min Transit Drive-Access Time  10 min Transit Wait 15 min Transit Fare  200 ¢  Auto Auto Run Time 26 min Auto Terminal Parking Time 5 min Auto Parking Cost 100 ¢ Auto Operating Costs 219.5 ¢ Table  2-18 Validation results Observed transit share  2.17% Modeled transit share by Okanagan model 1.91% Error percentage 13% Travel distance and the commuter rail mode share It is worth mentioning that an interesting correlation between travel distance and the commuter rail mode share has been observed in the literature review (Parsons Brinckerhoff, 2004), (Transystems, 2004), (Wilbur Smith Associates, 2005), (Schiermeyer Consulting Services, 2002) for this section. Figure  2-10 is created from the collected data, and shows the mode share for different commuter rails with operating speeds of 50 to 100 km-per-hour for the traveled distance below 70 km, a strong linear correlation between commuter share and the distance can be observed. While using graphs like Figure  2-10 is not a very reliable method to calculate the potential share for a new commuter rail, these graphs can be used to validate the results of the mode choice  - 39 - models or predict the possible growth in share. Note that such a relationship has previously been reported between distance and the air/high speed rail share (Segal, 2006).0%10%20%30%40%50%60%70%80%90%100%0 10203040506070809010Travel Distance (Km)Commuter Rail ShareFigure  2-10 Commuter rail share Vs travel distance Data sources: (Parsons Brinckerhoff, 2004), (Transystems, 2004), (Wilbur Smith Associates, 2005),(Schiermeyer Consulting Services, 2002)The role of rail speed in determining rail share has been discussed in the literature for High Speed Railways (HSRs) (de Rus, 2008). As stated in Figure 2-11, the HSR market share is correlated with the rail speed, with the exception of Madrid-Barcelona connection (recently launched). . Figure  2-12 shows the rail share versus the travel distance for three different commuter rails with different operating speeds. It implies that for rails running well below the speed of high speed rails (maximum speed of less than 130 km-per-hour) rail share is independent of rail speed. - 40 - Figure  2-11 Speed vs. HSR market shareData source: (de Rus, 2008)0%10%20%30%40%50%60%0 10203040506070Travel distance kmrail share 51 kph80-96 kph70 kphFigure  2-12 Commuter rail share vs. travel distance for different operating speeds Data sources: (Parsons Brinckerhoff, 2004), (Transystems, 2004), (Wilbur Smith Associates, 2005)  - 41 - 2.4 Demand forecast for the Okanagan Passenger Railway  In this section, the demand for the passenger railway has been determined using the methods and procedures described in the last section. 2.4.1 Trip generation & distribution As discussed previously, the Okanagan Valley Transportation Plan model (OVTP) model provided the O-D table of passenger demand for 1994 and 2020 (Table  2-7 and Table  2-8). For the years after 2020, a linear extrapolation of the data was used due to the absence of any more accurate data.     To estimate the mode share, the developed Okanagan mode choice model (Table  2-16)was used. Recall, the mode choice model needed more refined data (i.e. trip purposes) than what was provided by the OVTP model (all in one group). Therefore, four steps were taken to create it. In the first step, using the 2007 Okanagan survey, the percentages of each trip purpose between each O-D pair were determined as shown in Table  2-19,Table  2-20 and Table  2-21. In the second step, as the Okanagan survey only included the North and Central Okanagan districts, for the other O-D pairs, an average of the O-D pairs in the North and Central Okanagan has been used.Table  2-19 Home Based Work (HBW) share From/To Osoyoos Penticton Summerland PeachlandWestbank Kelowna Vernon Coldstream Osoyoos 17.06% 36.48% 37.15% 37.15% 37.15% Penticton Summerland 36.48% 17.06% 36.48% 37.15% 37.15% Peachland Westbank 37.15% 36.48% 13.69% 43.11% 36.36% Kelowna 37.15% 37.15% 34.41% 19.59% 32.06% Vernon Coldstream 37.15% 37.15% 37.93% 36.36% 17.91%  - 42 - Table  2-20 Home Based Other (HBO) share From/To Osoyoos Penticton Summerland PeachlandWestbank Kelowna Vernon Coldstream Osoyoos  57.07% 37.72% 37.36% 37.36% 37.36% Penticton Summerland  37.72% 57.07% 37.72% 37.36% 37.36% Peachland Westbank 37.36% 37.72% 68.05% 40.83% 33.33% Kelowna  37.36% 37.36% 46.01% 50.44% 31.76% Vernon Coldstream  37.36% 37.36% 41.38% 32.26% 52.73% Table  2-21 Non Home Based (NHB) share From/To Osoyoos Penticton Summerland PeachlandWestbank Kelowna Vernon Coldstream Osoyoos  25.86% 25.80% 25.50% 25.50% 25.50% Penticton Summerland  25.80% 25.86% 25.80% 25.50% 25.50% Peachland Westbank 25.50% 25.80% 18.26% 16.06% 30.30% Kelowna  25.50% 25.50% 19.59% 29.97% 36.18% Vernon Coldstream  25.50% 25.50% 20.69% 31.38% 29.36% Third, the number of trips in each purpose group between each O-D pair was estimated by using the total trips from the OVTP model and above tables. Last, the number of trips needed to be converted to the number of passengers before advancing to the mode choice step. The conversion was done using the Occupancy estimates in the Okanagan Valley provided by 2007 Okanagan Survey as shown in Table  2-22.Table  2-22 Car occupancy for different trips HBW HBO NHB 1.20 1.94 1.64 2.4.2 Mode choice Following these four steps, the Okanagan mode choice model (Table  2-16) was used to develop the share of the rail from the passengers moving between each O-D pair. Note that these mode choice model results assumed that the Okanagan Railway would be the  - 43 - only regional transit system in the valley, which was considered reasonable for this research.  Future research would be needed to test this assumption. Through the application of the mode choice model, it has been found that the mode share results are very sensitive to only three coefficients, which are Transit In-Vehicle Time, Transit Waiting Time, and Auto Operating Cost. The value of the other coefficients cannot change the outcome of the model drastically as long as they are in the reasonable ranges. Therefore, for the other coefficients, a reasonable value is selected as shown in Table 2-23, while the three sensitive coefficients are discussed in more detail below.    Table  2-23 Default values for mode choice model coefficients Transit Walk-Access Time  2 min Transit Drive-Access Time  15 min Transit Fare  the same as bus transit Transit Transfer Wait 5 min Transit Number of Transfers  1 Auto Run Time distance between O-D times the average speed in the area Auto Terminal Parking Time 2 min Auto Parking Cost 100 ¢ Transit In-Vehicle Time Train speed and station dwell time for boarding/detraining passengers would determine the Transit In-Vehicle Time. Operating Speed: usually the operating speed1 of a regional railway would be between 40 and 80 km per hour, while the maximum speed2 ranges from 80 to 130 km per hour (Vuchic, 2005). For the primary calculations, the operating speed of 60 km per hour and maximum speed of 100 km per hour, the averages of the existing regional railways, are chosen. Note that while the higher speed would decrease the transit run time and would increase the favorability of the rail and thus the rail share, it would increase the cost of track construction.  Sensitivity analysis in Section 6.4 would 1 Operating speed is the speed of travel on the line which passengers experience. 2Maximum speed is defined as the maximum speed a vehicle and rider can reach at steady state, with no grade.  - 44 - examine the sensitivity of the capital and maintenance cost and passenger demand to the speed. Sensitivity analysis would determine the impacts of different operating speed on the benefit cost ratio of the project.Dwell Time: TRB report 13: Rail transit capacity suggested different methods for estimating the dwell time while admitting that none of their methods are proven to be entirely satisfactory (Parkinson, T. Fisher, I., 1996). In this study, the first method, “Assigning a Value”, i.e. assigning a fixed dwell time value to each station, is applied. In the major stations, dwell time is considered to be one minute, while in the small stations, it would be 30 seconds. Note that for primary calculation eight stations for the Okanagan Railway are considered: Vernon, Lake County, Kelowna Airport, downtown Kelowna, Westbank, Peachland, Summerland, Penticton, and Osoyoos. Transit In-Vehicle Time (in seconds) would be calculated as shown in Equation 2-2: ��u MNVDTt3060155.02     ( 2-) where:V = operating speed;D =  distance between the origin and destination;  M = number of small stations between the origin and destination; andN = number of major stations between the origin and destination. Note that the acceleration and deceleration rate for an average regional rail is about 1 2sm.This means that it takes about 17 seconds (155 meters) for a train to stop in a station or reach its operating speed from standstill.  Transit Waiting TimeStatistically speaking, the average waiting time for a train would be half of the headway time. Determining the optimum headway time is difficult. Longer headways lead to fewer  - 45 - trains, thus lower operating costs. However, shorter headways increase the demand, and thus the revenue. Therefore, the optimum headway should be long enough to allow the train to reach its capacity, and short enough to remain attractive for the passengers.  Figure  2-13 shows the total southbound demand and the number of passengers on each train for different headways, if a train was running between Kelowna and Vernon in 2007. The number of total passengers (demand) decreases from 6800 persons to 1100 persons by increasing the headway time from 10 minutes to 2 hours. However, the number of passengers per train goes up by increasing the headway time from 10 minutes to one hour, and then drops down with further increases in headway from one hour to two hours. It seems that a headway time between 40 to 60 minutes can lead to the lowest operating cost (minimum number of trains) while maintaining the demand at an acceptable level. Sensitivity analysis would examine the effects of this choice. 0510152025300 20 40 60 80 100 120 140Headway in minNumber of passengers per train010002000300040005000600070008000Total number of passengersNumber of passengers per train Total number of passengersFigure  2-13 Number of passenger vs. headway timeAuto Operating Cost As shown in Table  2-14, auto operating costs include fuel, maintenance, and tire prices  (exclude ownership cost such as insurance, depreciation, finance expense as defined in FSUTMS mode choice model). The future tire and maintenance prices are expected to  - 46 - increase with inflation. However, prediction of the fuel price is an ongoing discussion in the economic literature. Oil price prediction for the Okanagan Railway is more challenging considering the fact that the Okanagan Railway is being planned for the distant future, for which very few studies about predicting the price of oil exist .Table 2-24 summarizes the literature on oil price prediction between now and 2030.Table  2-24 Oil price forecasts between 2008 and 2030 Forecast Name  CSIRO MED Goldman Sachs NYMEX IEA  Denominated USD  2008 2005 2008 2008 2008 Global CPI  1 1.08 1 1 1 Future Discount  0 0 0 15%2 02008 $100 $90 $108 $126 $84 2009 $95 $105 $110 $127 $772010 $200 $120 $120 $125 $74 2011 $210 $120 $120 $124 $712012 $220 $120 $75 $123 $68 2013 $215 $120 $75 $123 $662014 $210 $120 $75 $123 $63 2015 $175 $120 $75 $123 $602016 $165 $114 $75 $123 $60 2017 $155 $108 $75 $123 $602018 $145 $102 $75 $123 $60 2019 $175 $96 $75 $123 $60 2020 $180 $90 $75 $123 $60 2021 $140 $90 $75 $123 $61 2022 $100 $90 $75 $123 $62 2023 $95 $90 $75 $123 $63 2024 $96 $90 $75 $123 $64 2025 $97 $90 $75 $123 $64 2026 $98 $90 $75 $123 $66 2027 $99 $90 $75 $123 $67 2028 $100 $90 $75 $123 $68 2029 $101 $90 $75 $123 $69 2030 $102 $90 $75 $123 $70 Source: (Donovan et al., 2008) - 47 - New Zealand Transport Agency Research has summarized the above predictions using Monte Carlo simulation, and concluded three scenarios of High, Average and Low oil price as shown in Figure  2-14.  Figure  2-14 Oil price scenarios generated by NZ Transport Agency Research  Sources: (Donovan et al., 2008)   Only two models are found in the literature on predicting the oil price for years beyond 2030. The first is the SAUNER (Sustainability And the Use of Non-rEnewable Resource) model, which was developed by the Department of Economics and International Development at the University of Bath, United Kingdom. The project aims to generate a comprehensive database of regional and world hydrocarbon resources, and develop a long-range supply-cost model through 2100 (University of Bath et al., 2000). The second is the LOPEX (Long-term Oil Price and EXtraction) model, which was developed by researchers at the University of Stuttgart, Germany. This model generates long-term scenarios about the future world oil supply, and corresponding price paths up to the year 2100 (Rehrl et al., 2006).The results of the most optimistic scenarios for both models are shown in Figure  2-15. The SAUNER predictions are under-predicting the oil price considering its prediction for  - 48 - 2030 is the same as the prediction by the Low scenario of the NZ report. On the other hand, the LOPEX prediction for 2030 is a little lower than the oil price in the Average scenario of NZ report. While it seems reasonable to select the LOPEX model due to the compatibility with NZ report, in order to be conservative, another prediction for years between 2030 and 2100 is created by averaging the results of the SAUNER and the LOPEX models. In this newly developed model, the oil price would be between the oil price in the Low and Average scenarios of NZ report. From this point forward, oil price would be calculated using this developed model.  01002003004005006002030 2040 2050 2060 2070 2080 2090 2100Price [2008USD/barrel]AverageSAUNER: Impressive Technological Improvements and High Economic Growth  .LOPEX:Increased Recovery scenarioFigure  2-15 Oil price after 2030 by SAUNER & LOPEX2.4.3 ForecastsUsing the values assigned to the mode choice model, passenger demand for the Okanagan Railway is calculated for every segment of the railway for years between 2009 to 2079, as shown in Table 2-25.  It should be noted that the Okanagan Railway would run between 6:00 AM to 7:00 PM until 2060. During these years, there is not enough passenger demand for the rail before 6:00 AM and after 7:00 PM.  However, after 2060 the Okanagan Railway would run from 5:00 AM to 12:00 AM. Further studies are  - 49 - recommended to analyze the impact of this decision on attractiveness of the railway for tourism and late night socializing up and down the valley. Figure 2-16 summarizes passenger ridership for southbound movements for a typical weekday in 2030.Daily Ridership of the Okanagan Railway (Southbound) in 2030219233296415857694603633882515321109612882287306393474988611171132142922104538513359Vernon toWinfield,lakecountyWienfield, lakecounty to airportKelownaairport-downtownkelownaQeensway,Kelowna toWestBankExchangeWestBankExchange toBeach & Hwy 97,PeachlandBeach & Hwy 97,Peachland toSummerlandSummerland toPentictonPenticton toOsoyoos6:00 - 9:009:00 - 15:0015:00 - 18:0018:00 - 19:0019:00 - 24:00Figure  2-16 Daily ridership of the southbound movements in 2030 To calculate some of the benefits of the railway, it was necessary to estimate the number of trips that the forecasted passenger would make by auto or bus in Highway 97 if no railway is built in the valley. Those benefits include savings that are related to the number of cars that are removed from Highway 97. Last column of Table 2-25 shows the number of transferred trips from autos in Highway 97 to the Okanagan Railway.  2.5 SummaryIn this chapter, the passenger demand models for the similar commuter rails were reviewed. Afterward, the O-D table for the inter-city trips in the Okanagan Valley was produced using the existing Okanagan Valley transportation plan. In the next step, a mode choice model for the Okanagan Valley was developed using the reviewed models. Finally, as shown in Table  2-25, the demand for the Okanagan passenger railway was estimated using the produced O-D table and the developed mode choice model.   - 50 - Table  2-25 Forecasted demand for the Okanagan Passenger Railway 2009 2019 2029 2039 2049 2059 2069 2079 Vernon – Lake County 1,369 1,697 2,131 3,123 4,933 10,328 21,589 41,772 Lake County - Kelowna Airport 1,414 1,772 2,226 3,172 4,840 9,700 19,663 38,777 Kelowna Airport - Downtown Kelowna 1,760 2,207 2,771 3,949 6,026 12,077 24,482 48,280 Downtown Kelowna - Westbank 2,336 2,918 3,633 5,096 7,661 15,855 35,357 77,462 Westbank- Peachland 484 605 773 1,239 2,144 5,571 15,777 39,939 Peachland - Summerland 437 545 698 1,145 2,023 5,386 15,505 39,487 Summerland - Penticton 559 688 866 1,351 2,288 5,804 16,162 40,681 Penticton - Osoyoos 392 463 570 911 1,595 4,257 10,226 19,992 Total (Transferred Passenger) 8,751 10,896 13,669 19,985 31,511 68,978 158,760 346,390 Total (Transferred Trip) 5,368 6,684 8,385 12,259 19,328 42,310 97,380 212,469 - 51 - 3 FREIGHT DEMAND FORECASTS 3.1 Introduction  This chapter consists of three sections. In Section 3.2, the literature on freight demand modeling including freight model classes and the components is reviewed. In Section 3.3, methodology for producing O-D tables, calculating the rail shares, and assigning the rail freight to the rail network is discussed. Finally in Section 3.4, freight demand for the Okanagan Railway is estimated.  3.2 Literature review  Freight forecasting models use the same principles as passenger demand forecasting models.  However, freight demand modeling is more complex than passenger modeling and much less research has been done on modeling freight than passenger movement (Ortuzar et al., 2001). As mentioned in the motivations section, the demand for the Okanagan Freight Railway would be international as well as long haul domestic. Therefore, the focus of the literature review was on the long distance freight movements, and thus statewide/province-wide models. The most comprehensive literature on this group of freight forecasting models was found among the publications of the National Cooperative Highway Research Program (NCHRP) including; Report 606: Freight Toolkit; Report 358: Statewide Travel Forecasting Models; and Report 280: demand forecasting techniques (Memmott, 1983, Horowitz, 2006, Cohen et al., 2008). - 52 - 3.2.1 Freight model classes and the components TRB Report 606: Freight Toolkit suggests grouping the freight models into five basic model classes with modeling components as shown in Table  3-1. A brief explanation of each class is included below. Table  3-1 Freight model classes by component           Component ClassDirectFactoring TripGeneration TripDistribution Mode Split TrafficAssignment Economic/ Landuse  Modeling Direct Facility Flow Factoring Method Of  facilityflows      Truck Model  exogenously supplied zonal activity Included Not Applicable Included  Four-Step Commodity Model  exogenously supplied zonal activity Included Included Included  Economic Activity Model  outputs of economic model Included Included Included Included O-D Factoring Method Of O-D  tables   Included Included  Source: (Cohen et al., 2008) The direct facility flow factoring class estimates freight volumes on existing transportation links. It includes two steps: one that estimates the flow on the link based on factors that cause diversion from other links or modes and one that estimates the future demand based on appropriate growth factor. While this class of models is frequently used in planning, it cannot provide long-term forecast nor consider all important factors in freight transport (Cohen et al., 2008).   The truck model class uses trip generation and trip distribution models to produce an O-D table for truck trips and then applies assignment techniques to assign the trips to the network. While - 53 - this class is useful dealing with freight transportation on highways, it was not suitable for our purpose because it could not be adapted to rail transportation (Cohen et al., 2008). The four-step commodity model class is similar to the four-step urban travel demand model for passengers.  It also includes trip generation, trip distribution, mode split, and trip assignment model components. In the trip generation step, a set of trip generation rates or equations sorted by the commodity type provides demand as a function of population or employment data. In the trip distribution step, models apply gravity models. In the mode split step, usually existing mode share or qualitative adjustments of existing mode share is used due to the fact that developing mode split models involves considering too many relevant factors. Usually each mode employs a specific technique in the assignment step. The rail, water, and air assignments typically follow the rules-based assignment process. The assignment of truck freight has typically been done by a freight truck only or multiclass assignment techniques (Cohen et al., 2008).  The economic activity model class uses trip generation, trip distribution, mode split, and assignment model components to produce freight forecasts like the four-step commodity model. However, unlike the four-step model, the economic activity model is dynamic both with respect to land use and transportation (Cohen et al., 2008).The O-D factoring method class uses an existing O-D table instead of trip generation and distribution and applies growth factors to the O-D to forecast the future demand. After that, it follows the conventional mode split and trip assignment steps as described in the four-step commodity model. This method is particularly well suited to studies with too many large zones where trip generation rates are not available. For example, in the United States, where Reebie Associates’ TRANSEARCH database is available as a source of O-D table, many state models and the country-wide Freight Analysis Framework (FAF) apply this method to estimate freight demand (Cohen et al., 2008).  The potential freight demand for the Okanagan Railway includes the Okanagan Valley’s originated or destined freight and through movements, which would use the Okanagan Rail as a connection. Considering the size of the study area for the freight demand, i.e. western United - 54 - States and Canada, direct factoring is the appropriate method for completing the O-D table.  Moreover, no trip generation rate or equation to predict commodity attraction and production, and no utility function to distribute the commodity has ever been developed for the Okanagan Valley or even for British Columbia. As shown in Table  3-1, direct factoring is the modeling component in two freight modeling classes: the direct facility flow factoring and the O-D factoring methods. The former, as discussed earlier, is only applicable to existing facilities. Thus, the O-D factoring method is selected as the appropriate method of freight demand forecasting for the Okanagan Valley. 3.2.2 Rail freight demand modeling components3.2.2.1 “Trip generation & trip distribution” or “direct factoring” Both direct factoring and the combination of trip generation and distribution produce volume estimates for the Traffic Analysis Zones (TAZs). The direct forecasting model component produces forecasts of link volumes using existing flows data. In practice, the O-D tables are produced from a trip table provided by x a public or commercial source; or  x by surveys of freight shippers, receivers, and/or carriers; or  x by observing freight flows.Cascade Gateway Rail has applied this methodology to produce the O-D tables for forecasting its future demand. They have used the TRANSEARCH database (Reebie Associates, 2002). This database is private and produced by Reebie Associates. It provides information on freight flows within the United States and between the United States and Canada. On the other hand, trip generation forecasts the production and attraction of freight movements that begin or end in a TAZ using trip generation rates or equations based on characteristics of the TAZ. The distribution models, which are usually gravity models, distribute trips between originating TAZs and destination TAZs based on the relative impedance. Some American states - 55 - like Oregon and Indiana have developed state-wide trip generation and distribution equations. These equations are then used for the regional rail demand forecast (Cohen et al., 2008). 3.2.2.2 Mode split The mode split models define future flows by specific modes. The mode split model may use modal shares from the base year data by origin, destination, and commodity group to forecast the mode split in the future year. If modal utility function like cost is available, that information can be used to modify the base year mode share forecasts. 3.2.2.3 Rail assignment As discussed earlier, rule-based assignment usually applies to rail networks due to the difficulties of including rail business practices in an assignment model. In the literature, two documents were found on rail assignment. First, Zachary F. Lansdowne (Lansdowne, 1981) proposed an algorithm for routing freight over a rail network whose tracks are controlled by several carriers while applying 1980s industrial practices:x Minimizing number of interline transfers,  x Finding the shortest route in each carrier’s subnetwork, x Maximizing revenue division for originating carriers, and x Dividing shipments among originating carriers based on shipping distance. The study proposed the following logit distribution function in the last step �� CxxCcLLP)exp()exp(DD          ( 3-1)Where  cP  The fraction of the shipment assigned to originating carrier c;  cL  Actual path distances; and, D = An input constant. - 56 - Although it is the most comprehensive study that can be found among the literature, the authors have not defined D in Equation 3-1.  Presumably it is defined by calibration with locally available data. Second, in Indiana Commodity Flow Modeling, William R. Black has suggested another algorithm to assign the rail traffic to the network (Black, 1999). The study assumed that there is some desire on the part of rail carriers to minimize the length of shipment but they have a tendency to use mainline trackage even though secondary lines may be more direct. A new “cost of movement” variable that addresses these issues has been proposed ,as follows: )11( DLI               ( 3-2)Where I = the index of spatial separation; L = the length of the line segment of the network; and, D = the traffic density of the line in millions of gross ton-miles per year. The variable would make the length of line segments with high traffic density shorter while the length of lines with low traffic density would not change. In the next step, Black proposes the same choice model as Lansdowne rule-based method(Equation 3-1) would be applied to the network. The only difference is that the new spatial separation index would be used instead of actual path distance at that equation.  3.3 Methodology for demand forecasting for the Okanagan Freight Railway 3.3.1 Direct factoring O-D table In the following sections, the O-D factoring method components – direct factoring of O-D tables, mode split, and assignment – are developed to forecast the Okanagan Railway freight demand. - 57 - 3.3.1.1 Likely origins and destinations The first step in developing the O-D table is defining the likely O-D pairs that produce future demand for the Okanagan Railway. A likely O-D pair is defined as: ijDDCurrentijOkanaganij�|  is a likely O-D pair where OkanaganijD Distance between origin i and destination j using the Okanagan Railway; and,  CurrentijD  Distance between origin i and destination j using the current path. In other words, when the Okanagan Valley lies along a desire-line between two cities, those cities make a likely O-D pair.  For instance, Prince George, BC, and Portland, OR, make a likely O-D pair, because a path along the Okanagan Valley can be a part of the path between the two cities. However, Prince George, BC, and Edmonton, AB, do not make a likely O-D pair, because the desire-line path between these two cities would not include the Okanagan Valley.Based on this definition, the likely O-D pairs are discussed below.  These likely O-D pairs may or may not produce demand for the Okanagan Valley.  Trip generation, trip distribution, mode choice, and trip assignment steps would be employed to determine the portion of the movements that would use the Okanagan Railway. Pair # 1: Okanagan Valley cities and Northern BC cities:  Currently most of the freight between the cities in the Okanagan Valley and the cities in Northern BC (including Kamloops, Prince George and Prince Rupert) is moved by truck. Trucks usually take one of the following paths (see Figure  3-1): x Highway 97 from origin to Vernon, then to Kamloops and finally to the destination and vice versa. x Highway 97 from origin to Westbank, Highway 97C to Merritt, Highway 5 to Kamloops and finally to the destination and vice versa. - 58 - As shown in Figure  3-2, in the future freights can be shipped between the cities in Okanagan Valley and the cities in Northern BC by rail taking the following path: x Okanagan Railway from origin to Vernon, KPR to Kamloops, and CN to the destination. Pair # 2: Okanagan Valley cities and the Lower MainlandCurrently most of the freight between the cities in the Okanagan Valley and the Lower Mainland is moved by truck. Trucks usually take the following path (see Figure  3-1): x Highway 97 from origin to Westbank, Highway 97C to Merritt, Highway 5 to Hope and finally to the destination and vice versa. As shown in Figure  3-2, in the future freights can be shipped between cities in the Okanagan Valley and cities or ports in Lower Mainland by rail taking the following path: x Okanagan Railway from origin to Vernon, KPR to Kamloops and CN or CPR to Lower Mainland.Pair # 3: Okanagan Valley cities and other Canadian provinces Currently most of the freight between the cities in the Okanagan Valley and cities in other Canadian provinces is moved by truck. Trucks usually take the following path (see Figure  3-1): x Highway 97 from origin to Vernon, Highway 97A to Sicamous, Trans-Canada Highway to Golden and finally to the destination in other provinces and vice versa. As shown in Figure  3-2, freights in future can be shipped between the cities in the Okanagan Valley and the cities in other Canadian provinces by rail taking the following path: x Okanagan Railway from origin to Vernon, KPR to Kamloops, and CN or CPR to the destination. Pair # 4: Okanagan Valley cities and American cities - 59 - Currently most of the freight between the cities in the Okanagan Valley and American cities is moved by truck. Trucks usually take one of the following paths (see Figure  3-1): x Highway 97 from origin to Osoyoos border, then to the destination in the United States and vice versa. x Highway 97 from origin to Westbank, Highway 97C to Merritt, Highway 5 to Hope to Blaine or Sumas customs port, and finally to the destination in the United States and vice versa.As shown in Figure  3-2, in the future freights can be shipped between the cities in Okanagan Valley and the cities in the United States by rail taking the following path: x Okanagan Railway from origin to Osoyoos, CSCD to Wenatchee, and BNSF to the destination.   Pair # 5: Northern BC cities and American cities Currently, the freight between the cities in the Okanagan Valley and Northern BC cities, including Kamloops, Prince George and Prince Rupert, is moved by truck and rail. Trucks usually take the following paths (see Figure  3-1): x Highway 97 from origin to Kamloops, then depending on the destination in the United States, Highway 5 to the customs ports at Eastern  BC (Sumas, Blaine) or Trans-Canada Highway to customs ports at Western  BC (Eastport), and finally from the customs port to the destination in the United States and vice versa.   The rail path for Kamloops and Prince George originating freight is almost the same as the road path (see Figure  3-2):  x CPR from origin to Kamloops, then depending on the United States destination, CPR to the customs ports in Eastern BC or CN to customs ports at Western BC, and finally from the customs port to the destination in the United States and vice versa (note that freight shipped by rail from Prince Rupert takes BC Rail to Vancouver, and then shipped from Vancouver to the destinations in the United States).   - 60 - As shown in Figure  3-2, the Okanagan Railway would add another route to the rail paths mentioned above. Freights in the future can be shipped between Northern BC cities and the cities in the United States by rail taking the following path: x CPR from origin to Kamloops, KPR from Kamloops to Vernon, the Okanagan Railway from Vernon to Osoyoos customs port, and finally from the customs port to the destination in the United States and vice versa Pair # 6: West Coast American cities and other Canadian provinces Currently, the freight is moved by truck and rail between the cities in the other Canadian provinces (excluding British Columbia as it was discussed in detail above) and the cities in California, Oregon, and West of Washington State (West Coast). Some of the freight use British Columbia roads and railways. The trucks usually take the following paths (see Figure  3-1): x Trans-Canada Highway via Golden or Highway 97 from origin to Kamloops, Highway 5 to the customs ports at Eastern  BC (mainly, Sumas, and Blaine) and finally from customs port to the destination in the United States and vice versa. x Trans-Canada Highway from origin to Golden, Highway 95 to the customs ports at Western BC (mainly Eastport) and finally from the customs port to the destination in the United States and vice versa. The freight can take two rail paths (see Figure  3-2):  x CPR or CN from origin to Kamloops, then to the customs ports in Eastern BC (Sumas, Blaine), and finally from the customs port to the destination at the United States and vice versa.As shown in Figure  3-2, Okanagan Railway would add another route to the rail paths mentioned above. Freight in the future can be shipped between the cities in the other Canadian provinces and the cities in West Coast states in the United States by rail taking the following path: x CPR from origin to Kamloops, KPR from Kamloops to Vernon, Okanagan Railway from Vernon to Osoyoos customs port, and finally from the customs port to the destination at US and vice versa.- 61 - Pair # 7: Middle West American cities and other Canadian provinces Currently, freight is moved by truck and rail between the cities in the other Canadian provinces (excluding British Columbia as it has been discussed in detail above) and the cities in Idaho, Arizona, Utah, Nevada and east of Washington State (Middle West). Some of the freight use British Columbia roads and railways. The trucks usually take the following path (see Figure  3-1): x Trans-Canada Highway via Golden or Highway 97 from origin to Kamloops, Highway 5 to the customs ports at Western  BC (mainly, Sumas, and Blaine) and finally from customs port to the destination at the United States and vice versa. x Trans-Canada Highway from origin to Golden, Highway 95 to the customs ports at Eastern BC (mainly Eastport) and finally from customs port to the destination at the United States and vice versa. Freight can take two rail paths (Figure  3-2):  x CPR or CN from origin to Kamloops, then to the customs ports at Western BC (mainly Sumas, and Blaine), and finally from the customs port to the destination in the United States and vice versa.x CN from origin to Golden, then to the customs ports in Eastern BC customs ports (mainly Eastport), and finally from the customs port to the destination in the United States and vice versa.As shown in Figure  3-2, the Okanagan Railway would add another route to the rail paths mentioned above. Freight in the future can be shipped between the cities in other Canadian provinces and the cities in the Middle West of the United States by rail taking the following path: x CPR from origin to Kamloops, KPR from Kamloops to Vernon, Okanagan Railway from Vernon to Osoyoos customs port, and finally from the customs port to the destination in the United States and vice versa.- 62 - Figure  3-1 Highway network in Western North America -62-- 63 - Figure  3-2 Rail network in Western North America-63-- 64 - Table  3-2 shows the likely O-D pairs as coloured cells. The cells’ contents show to which category each cell belongs. The next step would be to find data to fill the coloured cells. Table  3-2 Likely O-D pairs              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna Pair#1 Pair#2 Pair#3 Pair#4 Pair#4Penticton Pair#1 Pair#3 Pair#4 Pair#4Northern BC Pair#1 Pair#1    Pair#5 Pair#5Vancouver Pair#2       Rest of Canada Pair#3 Pair#3    Pair#6 Pair#7West USA Pair#4 Pair#4 Pair#5 Pair#6   Rest of USA Pair#4 Pair#4 Pair#5 Pair#7   3.3.1.2 Data sources There are three means to produce O-D tables (Cohen et al., 2008): x Obtain a filled table from a public or commercial source; x Develop a trip table from a survey of freight shippers, receivers, and/or carriers; or x Estimate a trip table from observed freight flows.  As discussed earlier, most studies use Reebie Associates’ TRANSEARCH database as the data source for producing O-D tables. However, the TRANSEARCH database is a private database and it is not available to academic researchers without significant cost. Therefore, in this study, the second and third method, i.e. surveys or observed trip tables, were used. Different databases are used for different analysis zones due to the size of the study area. Unfortunately there was no compatibility between the data sources and each source divided the study area into analysis zones differently. Table  3-3 reviews all of the databases and shows the zones of each database. Detailed information are included in Appendix C. - 65 - Table  3-3 O-D data sources  Database  Data 1 1999 NRS National Roadside Survey weekly truck movements (both domestic and transborder) in a representative week of the summer/fall of 1999 2 IMTC Manifest Survey International Mobility & Trade Corridor Project  - Cascade Gateway Almanac Survey annual truck movements through Blaine or Sumas in 2004 and 2006 3 BTS TransBorder Freight Data Bureau of Transportation Statistics  monthly/annual observed Transborder Truck and Rail movements from 1995 to 2007  4 FAF Version 2.2 Freight Analysis Framework annual Transborder Truck and Rail movements in 2002 and prediction for 2002 to 20355 Cascade Gateway Freight Demand Analysis Reebie Associates’annual Transborder Rail movements through Blaine or Sumas in 2002 and prediction for 2012  6 Okanagan traffic survey analysis local survey one day truck movement in Okanagan Valley via Highway 97 or 97C on June 6, 2007 3.3.2 Mode split As explained earlier, when the modal utility functions for different modes were available, the share for each mode was determined using a logit model driven from utility functions. However, in the absence of data to derive those utility functions, the mode share for each commodity in the base year was used to forecast the mode share in the future year. In this research, fixed share was employed due to the absence of utility functions. It meant that base year mode shares for each commodity in British Columbia were used as the future mode share.  - 66 - 3.3.2.1 NB/SB mode shares Although there were five border customs ports where trade happens by both railways and highways (Figure  3-3), only two sets of mode shares were available. Figure  3-3 Rail and Highway network at study area The first set was taken from Eastport where there was a railway (UP in ID and CN in BC) parallel to the highway (HWY 95), crossing the customs ports. The second set was at Blaine, where there was a railway (BNFS) parallel to the highway (I-5 in WA, HWY 99 in BC) crossing the customs ports. Because no major railway was running in the middle of BC, an Okanagan truck/rail mode split was developed by averaging the two available sets, as shown in Table  3-4 and Table  3-5.Different mode shares were created for northbound and southbound because the composition of the northbound commodities and the composition of the southbound commodities were different, even considering the commodities in the same commodity group.- 67 - Table  3-4 SB rail and truck freight movements (in thousands of short tons) and rail share  Customs port Blaine Eastport Average for BC SCTG Description3 truck rail rail share truck rail rail share truck rail rail share1 Live animals/fish 173.92 0.1 0% 43.36 0.02 0% 217.28 0.12 0% 2 Cereal grains 15.82 80.87 84% 5.42 57.34 91% 21.24 138.21 87% 3 Other ag. prods. 148.54 10.69 7% 34.34 6.96 17% 182.88 17.65 9% 4 Animal feed 94.62 121.83 56% 32.57 104.48 76% 127.19 226.31 64% 5 Meat/seafood 80.68  0% 12 0.02 0% 92.68 0.02 0% 6 Milled grain prods. 72.45 33.17 31% 11.12 17.55 61% 83.57 50.72 38% 7 Other foodstuffs 75.11 46.6 38% 14.47 29.03 67% 89.58 75.63 46% 8 Alcoholic beverages 85.1 10.91 11% 14 9.31 40% 99.1 20.22 17% 9 Tobacco prods. 0.27  0% 0.28  0% 0.55 0 0% 13 Nonmetallic minerals 57.4 309.58 84% 11.31 188.66 94% 68.71 498.24 88% 14 Metallic ores 1.83 26.91 94% 0.31 25.47 99% 2.14 52.38 96% 19 Coal-n.e.c. 141.46 646.77 82% 58.35 316.09 84% 199.81 962.86 83% 20 Basic chemicals 53.44 408.9 88% 16.21 289.17 95% 69.65 698.07 91% 21 Pharmaceuticals 0.9  0% 0.52 0.1 16% 1.42 0.1 7% 22 Fertilizers 80.8 884.49 92% 60.63 997.42 94% 141.43 1881.91 93% 23 Chemical prods. 59.64 14.06 19% 14.37 10.93 43% 74.01 24.99 25% 24 Plastics/rubber 147.54 140.44 49% 26.13 81.32 76% 173.67 221.76 56% 26 Wood prods. 1263.46 1657.59 57% 349.02 1335.24 79% 1612.48 2992.83 65% 27 Newsprint/paper 103.4 206.62 67% 21.02 128.69 86% 124.42 335.31 73% 28 Paper articles 489.03 469.44 49% 84.49 282.23 77% 573.52 751.67 57% 29 Printed prods. 22.55 0.27 1% 5.34 0.51 9% 27.89 0.78 3% 30 Textiles/leather 22.39 1.15 5% 3.97 0.75 16% 26.36 1.9 7% 31 Nonmetal min. prods. 134.87 8.98 6% 27.5 6.07 18% 162.37 15.05 8% 32 Base metals 107.74 149.46 58% 21.11 91.58 81% 128.85 241.04 65% 33 Articles-base metal 96.78 38.63 29% 17.39 49.12 74% 114.17 87.75 43% 34 Machinery 55.56 11.88 18% 9.01 5.92 40% 64.57 17.8 22% 35 Electronics 20.32 1.97 9% 2.8 1.7 38% 23.12 3.67 14% 36 Motorized vehicles 127.2 376.65 75% 18.65 286.07 94% 145.85 662.72 82% 37 Transport equip. 5.76 1.36 19% 1.54 1.03 40% 7.3 2.39 25% 38 Precision instruments 2.83  0% 1.46 0.09 6% 4.29 0.09 2% 39 Furniture 69.27 4.04 6% 10.14 4.44 30% 79.41 8.48 10% 40 Misc. mfg. prods. 8.95 0.38 4% 2.15 1.52 41% 11.1 1.9 15% 43 Mixed freight 106.34 14.29 12% 23.78 12.02 34% 130.12 26.31 17%                                                 3 For the complete descriptions of commodity groups, refer to Table  3-29 - 68 - Table  3-5 NB rail and truck freight movements (in thousands of short tons) and rail share Customs port Blaine Eastport Average for BC SCTG Description4 truck rail rail share truck rail rail share truck rail rail share1 Live animals/fish 75.28 2.11 3% 11.21 0.34 3% 86.49 2.45 3% 2 Cereal grains 77.07 63.62 45% 17.91 2.58 13% 94.98 66.2 41% 3 Other ag. prods. 837.56 9.74 1% 89.68 3.46 4% 927.24 13.2 1% 4 Animal feed 105.29 26.26 20% 7.59 0 0% 112.88 26.26 19% 5 Meat/seafood 30.96 0 0% 3.71 0 0% 34.67 0 0% 6 Milled grain prods. 60.33 2.51 4% 8.18 0.34 4% 68.51 2.85 4% 7 Other foodstuffs 152.98 16.36 10% 19.64 2.06 9% 172.62 18.42 10% 8 Alcoholic beverages 20 0.48 2% 1.36 0 0% 21.36 0.48 2% 9 Tobacco prods. 0.06 0 0% 0.12 0 0% 0.18 0 0% 13 Nonmetallic minerals 99.58 129.9 57% 17.71 20.51 54% 117.29 150.41 56% 14 Metallic ores 15.02 17.15 53% 14.44 40.2 74% 29.46 57.35 66% 19 Coal-n.e.c. 311.37 39.36 11% 51.42 14.13 22% 362.79 53.49 13% 20 Basic chemicals 101.08 36.24 26% 11.66 10.05 46% 112.74 46.29 29% 21 Pharmaceuticals 2.2 0 0% 0.76 0 0% 2.96 0 0% 22 Fertilizers 41.62 33.73 45% 18.98 4.56 19% 60.6 38.29 39% 23 Chemical prods. 77.52 4.78 6% 7.02 1.04 13% 84.54 5.82 6% 24 Plastics/rubber 106.3 6.01 5% 8.72 1.18 12% 115.02 7.19 6% 26 Wood prods. 285.57 13 4% 55.6 10.8 16% 341.17 23.8 7% 27 Newsprint/paper 4.01 2.08 34% 0.29 0.37 56% 4.3 2.45 36% 28 Paper articles 121.33 7.59 6% 17.46 3.52 17% 138.79 11.11 7% 29 Printed prods. 16.78 0.02 0% 1.48 0 0% 18.26 0.02 0% 30 Textiles/leather 14.17 0.04 0% 1.22 0.08 6% 15.39 0.12 1% 31 Nonmetal min. prods. 495.59 14.32 3% 56.55 12.11 18% 552.14 26.43 5% 32 Base metals 181.65 18.26 9% 19.95 0 0% 201.6 18.26 8% 33 Articles-base metal 78.42 1.48 2% 7.38 0 0% 85.8 1.48 2% 34 Machinery 118.26 1.06 1% 9.2 0.52 5% 127.46 1.58 1% 35 Electronics 28.15 0.12 0% 2.04 0.15 7% 30.19 0.27 1% 36 Motorized vehicles 107.17 8.99 8% 13.33 0.83 6% 120.5 9.82 8% 37 Transport equip. 8.71 0.77 8% 1.33 0.32 19% 10.04 1.09 10% 38 Precision instruments 12.81 0 0% 1.19 0 0% 14 0 0% 39 Furniture 16.21 0 0% 1.34 0.08 6% 17.55 0.08 0% 40 Misc. mfg. prods. 32.79 0.05 0% 3.8 0.02 1% 36.59 0.07 0% 43 Mixed freight 31.38 0.25 1% 3.94 0.06 2% 35.32 0.31 1%                                                 4 For the complete descriptions of commodity groups, please refer to Table  3-29 - 69 - Moreover, rail transportation is usually used for shipping large volumes of a commodity. Therefore, it is possible that rail transportation would be the favourable mode of transport where a large volume of a commodity is needed to be shipped, while it is unfavourable for the same commodity when the volume drops.   3.3.3 AssignmentsIn this section, the estimates in previous section were assigned to the rail network in the area including the following railways (see Figure  3-4) x Through Eastport, CN in Canada and UP in the United States; x Through Oroville, Ok-Railway in Canada and CSCD in the United States; and x Through Blaine, CN or CPR in Canada and BNFS in the United States. Where the Okanagan Railway was the only shipping railway, for example between Kelowna and Northern BC, no assignment was necessary and traffic estimates were directly considered as Okanagan Railway demands.  Figure  3-4 Alternative rail routes in study area - 70 - Where multiple rail corridors required an assignment process, the goal was to choose routes that minimized the shipping costs (time, distance or monetary costs). A shortest path method was developed for this purpose.  In this method, the cost function only considered path distance, and assigned traffic to the shortest available path using the logit distribution model as shown below: �� CxxCcLLP)exp()exp(         ( 3-3)Where  cP  The fraction of the shipment assigned to originating carrier c;  cL  Actual path distances.The method is expected to be supportive of local and short railways given the practical rules in the current rail industry. Therefore, it was used as an optimistic scenario.   As noted in section 3.2.2.3, the literature suggests two other methods for assigning traffic to the railways. Lansdowne rule-base method is the first method where cost function is made up of series of rules (Lansdowne, 1981). This method’s results are expected to be the closest to reality since it applies industry rules. Spatial separation method is the second method. In this method, the cost function is made up of time and spatial separation index. As explained in section 3.2.2.3, it is assumed to be highly in favour of major railways. 3.4 Demand forecast 3.4.1 O-D table development This section’s objective is to provide estimates for the likely i-j pair cells in Table 3-2, using the databases that were reviewed in the previous section. To achieve this goal and based on the data available, the potential freight was divided into two groups, including Group 1: Rail movements; and Group 2: Truck movements.  Group 1 consists of all the - 71 - rail freight that can divert from another rail path to the Okanagan Railway and Group 2 consists of the truck freight that can divert from truck to the Okanagan Railway.   Group 1: Rail movements The rail movements group included the freight that was shipped by rail between the likely i-j O-D pairs. It was assumed that the Okanagan Railway could act as a minor alternative path, and during peak or congested flows, take some freight from original carriers.   Bureau of Transportation Statistics (BTS) provides the annual observed trans-border truck and rail movements at all the customs ports shown in Figure  3-5. However, it does not provide the necessary data on trip distribution, i.e. the location of the traded freight’s origin and destination. As stated in Appendix C,  Freight Analysis Framework (FAF) Version 2.2 and Cascade Gateway Freight Demand Analysis provide some data on the trip distribution only for movements crossing Blaine, WA (Douglas, BC) and Eastport, ID (Kingsgate, BC), which are the major custom ports into British Columbia as shown in Figure  3-6. Therefore, it was assumed that the trip distribution for Boundary and Laurier custom ports were similar to the trip distribution in Eastport due to the close proximity. These three custom ports would be analyzed as one group, called “Eastern ports”. Moreover, the trip distribution for Sumas port would be similar to the trip distribution in Blaine due to the close proximity. Blaine’s distribution would be employed for Sumas’s distribution. These two custom ports were analyzed as one group called “Western ports”.- 72 - Figure  3-5 USA/BC Rail Border Crossings 015003000450060007500Blaine,WashingtonSumas,WashingtonBoundary,WashingtonLaurier,WashingtonEastport,IdahoExport in thousands of tonsFigure  3-6 Export via BC-USA Customs ports by Rail in 2007\ Table  3-6 and Table  3-7 summarize annual trade by railway at these major ports in 2007. For this thesis, the format and terminology in Table 3-2 has been followed. In each - 73 - section discussed below, only those cells relevant to the discussion group have been filled. For example, in Table  3-6 and Table  3-7 only coloured cells related to the rail movements through eastern ports were filled.  Table  3-6 2007 trades by rail at eastern ports in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     880.11 621.96 Vancouver        Rest of Canada     2,470.49 411.42 West USA 13.31  50.07   Rest of USA 21.72  81.69   Table  3-7 2007 trades by rail at western ports in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     1,198.15 1,062.51Vancouver        Rest of Canada     1,268.63 1,125.01West USA 227.16  122.32   Rest of USA 88.34  47.57   At the time of this study, there was only one active local freight railway between Kelowna and Kamloops, i.e. the Kelowna Pacific railway (KPR). The traffic on the KPR has been shown in Table  3-8.   It was assumed that the estimated movements in the above tables would remain captive to rail in the future. In other words, none of the traffic would divert from rail to truck, which - 74 - eliminated the need to perform a mode split calculation, and allowed for direct consideration of the route assignment step.  Table  3-8  2007 KPR trades in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 24.4PentictonNorthern BC 24.4     Vancouver        Rest of Canada     West USA    Rest of USA    Group 2: Truck movements Table  3-9 shows the estimates of freight that was shipped by truck between the Okanagan Valley cities and other Canadian cities. As an alternative mode along the same route, the Okanagan Railway would attract some of this freight from the truck mode.  Table  3-9 2007 freight movements by trucks to/from the Okanagan Valley in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 243.89 206.60 40.51 Penticton 34.52  23.16 Northern BC 280.08 35.65    Vancouver 439.90       Rest of Canada 181.44 80.82    West USA    Rest of USA    In addition to domestic movements, trans-border truck movements were considered.  Trans-border truck movements included freight that was shipped by truck between - 75 - Canadian and American cities through the customs ports, as shown in Figure  3-7. As shown in Figure  3-8, several of these customs ports are no longer active truck freight ports (assumed not to change in the future). The major customs ports assumed for this research therefore included: 1) Blaine, WA - Douglas, BC; 2) Sumas, WA - Huntingdon, BC; 3) Oroville, WA - Osoyoos, BC; and, 4) Eastport, ID - Kingsgate, BC.Figure  3-7 Highway Border Crossings 050010001500200025003000Point Roberts, WABlaine, WALynden, WASumas, WANighthawk, WAOroville, WAFerry,WADanville, WALaurier, WABoundary, WAMetaline Falls, WAPorthill, IDEastport, IDExport in thousands of tonsFigure  3-8 Export via BC-USA Customs ports by Truck in 2007 - 76 - Table  3-10 to Table  3-13 show the 2007 annual trade at these major customs ports. Table  3-10 2007 trades by truck from Oroville-Osoyoos in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 154.18 188.44 Penticton small small Northern BC     2.71 3.31Vancouver        Rest of Canada     6.10 7.45West USA 102.78 small 1.81 4.07   Rest of USA 241.54 small 4.25 9.56   Table  3-11 2007 trades by truck from Sumas-Huntingdon in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     51.39 2.34Vancouver        Rest of Canada     30.19 5.75West USA 20.49  16.39   Rest of USA 6.83 5.46   Table  3-12 2007 trades by truck from Blaine-Douglas in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     54.71 13.93 Vancouver        Rest of Canada     17.41 0.00West USA 275.76  0.00   Rest of USA 30.64  0.00   - 77 - Table  3-13 2007 trades by truck from Eastport-Kingsgate in thousands of tons              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     78.17 154.03 Vancouver        Rest of Canada     277.13 546.12 West USA 193.30  14.55   Rest of USA 307.30  23.13   It was assumed that a portion of freight in Group 2, truck movements, would have the potential of diverting from truck to rail, if the conditions were prepared (divertible commodities). To determine the portion diverted, a mode split analysis was conducted in section 3.3.2, before route assignment step.  3.4.2 Future growth Future rail traffic demand was estimated by applying growth factors to the traffic estimates of the previous section. A growth factor for each commodity between each O-D pair was pursued, but unsuccessful as data was not available.  Instead, for Group1, rail movements, growth factors provided by the customs ports, as shown in Table  3-14 and based on the carrier owner were used to forecast the volume of rail freight in the future years. In the eastern ports, the carrier owners were Union Pacific (UP) in the US, and Canadian National Railway (CN) in Canada. In the western ports, the carrier owner was Cascade railway both in Canada and US.Table  3-14 Compound annual growth rate for railways  Southbound NorthboundCascade 3.61% 2.97%UP/CN 2.41% 3.43%- 78 - For group 2, truck movements, growth rate of truck transportation in the Okanagan Valley was used to forecast the volume of truck freight in future years.  Local studies projected a 3.4% annual growth rate for the truck transportation in the Okanagan College Region (BC STATS, 2007).For the trans-border truck movements, three different sets of growth factors for each commodity based on the location of customs ports were available (see Table 3-15) including: one set for the freight crossing at western ports; one set for the freight crossing at eastern ports; and finally, one set for the freight crossing at customs ports in the middle of BC, such as Osoyoos. Different growth factors were created for northbound and southbound because the compositions of the northbound commodities and the southbound commodities, even in the same commodity group, were different. Table  3-15 Compound annual growth rate for trucks at customs ports Southbound Northbound Commodity Group West BC East BC Middle BC West BC East BC Middle BC Alcoholic beverages -0.0089242 -0.0104691 -0.0093606 0.0138174 -0.0176557 -0.0066807 Animal feed 0.0570856 0.0154733 0.0589497 -0.0091996 0.0297171 -0.0005958 Articles-base metal 0.0416934 0.0230148 0.0408039 0.0257781 0.0030174 0.0247586 Base metals 0.0282315 0.02257 0.0281561 0.0529626 0.0330468 0.0617225 Basic chemicals 0.0173324 0.0363869 0.0197904 0.0249549 -0.027427 0.0102004 Cereal grains 0.0286122 0.006233 0.0315822 0.0226031 -0.0137277 0.0430945 Chemical prods. 0.0390906 0.0390777 0.0403562 0.0622327 -0.0162707 0.0218419 Coal-n.e.c. 0.0102783 0.0097555 0.0114854 0.0314137 0.0958418 0.0334804 Electronics 0.0806137 0.0838682 0.0814517 0.08792 0.0449054 0.0827885 Fertilizers 0.0097404 0.0129827 0.0127064 0.0242149 -0.0096744 0.0011956 Furniture 0.1012065 0.0372797 0.101068 0.046406 0.026761 0.0525614 Live animals/fish 0.0560973 0.0305368 0.0570244 0.060687 0.0076306 0.0623073 Machinery 0.0674613 0.051452 0.0666453 0.0397008 0.0302856 0.0431454 Meat/seafood 0.0269768 0.0176874 0.0270238 0.076337 0.046872 0.0650334 Metallic ores 0.0315544 0.0060941 0.026319 0.0118266 0.0233607 0.0336457 - 79 - Table 3-15 Compound annual growth rate for trucks at customs ports “Continued” Southbound Northbound Commodity Group West BC East BC Middle BC West BC East BC Middle BC Milled grain prods. 0.0168244 0.0259394 0.0177047 0.0205316 0.0149535 0.0157402 Misc. mfg. prods. 0.071964 0.0548505 0.0700921 0.0097097 0.0548728 0.0609848 Mixed freight -0.0147112 0.063633 -0.0145848 0.0600986 0.0114631 0.0175886 Motorized vehicles 0.0270481 0.0062819 0.0248709 0.0261819 0.0180539 0.0232186 Newsprint/paper -0.0191158 -0.0243421 -0.015518 0.0081046 0.0124374 0.0221976 Nonmetal min. prods. 0.0147561 0.0144822 0.0154065 -0.0082355 -0.0091093 -0.0102494 Nonmetallic minerals 0.0445636 0.0362308 0.0471336 0.0764061 0.0371769 0.0543011 Other ag prods. 0.0280224 0.0289116 0.0282385 0.0272245 0.0183403 0.028238 Other foodstuffs -0.0048895 0.0281188 -0.0040302 0.0680432 0.0083314 0.0742141 Paper articles 0.025242 0.0218335 0.0268634 0.0605014 0.0398512 0.0611192 Pharmaceuticals 0.113523 0.0954383 0.1008688 0.0990976 0.0554604 0.0580853 Plastics/rubber 0.0622651 0.0338483 0.0607173 0.0527332 0.0142626 0.0506349 Precisioninstruments 0.1188837 0.0845263 0.1176885 0.0579498 0.0373618 0.0561369 Printed prods. 0.031262 0.0069198 0.0306324 0.0469007 0.026731 0.047166 Textiles/leather 0.0637081 0.0406869 0.0625512 0.0106775 -0.0074992 0.0114714 Tobacco prods. -0.1488962 -0.0556283 -0.1516052 -0.0479034 -0.0842763 -0.0556419 Transport equip. 0.1129797 0.0587389 0.1129772 0.0610148 0.0436198 0.0545743 Wood prods. 0.0493419 0.025709 0.0500816 0.0007535 -0.0069289 0.0049606 Note that long term compound annual growth might be a suspicious assumption considering de-industrialization and globalization of goods production. Further research in this subject is recommended.  - 80 - 3.4.3 Rail share calculation As mentioned earlier, freight that is currently shipped by rail (Table  3-6 to Table  3-8) was assumed to remain captive to rail in the future. Therefore, the mode split step only will be applied to freight demand that is currently shipped by truck (Table  3-9 to Table  3-13), which are divided into two groups:1) Demand from Okanagan Valley’s: Table  3-9 and Table  3-10  2) Demand from the eastern and Western BC ports: Table  3-11 to Table  3-13 The first group consists of the freight movements in the Okanagan Valley, i.e. trucks that use at least one segment of Highway 97 between Osoyoos and Vernon, which includes through truck traffic crossing from Osoyoos-Oroville customs port (Table  3-10), and truck traffic between the Okanagan Valley and the rest of Canada (Table  3-9). As there was no railway in the area, trucks did not have the option of switching to rail. To calculate what share of this traffic would be shipped by Okanagan Railway if it existed, several steps were needed.  First the composition of traffic between each O-D pair due to different rail shares for different types of commodities is estimated. The composition of commodities moving through the Okanagan Valley is shown in Table  3-16, and Table  3-17. Table  3-16 Commodity composition between Okanagan and Greater Vancouver Regional District Commodity Okanagan to GVRD GVRD to Okanagan Automotive 0.00% 6.82% Building materials 37.50% 20.45% Empty 0.00% 27.27% Food/beverage 25.00% 15.91% Forest products 12.50% 9.09% General freight 12.50% 4.55% Mattresses 0.00% 11.36% Alcoholic beverage 12.50% 4.55% - 81 - Table  3-17 Commodity composition for truck crossings at Osoyoos customs port Commodity  Canada to USA USA to Canada Live animals/fish 5% 0% Cereal grains 0% 5% Other ag. prods. 6% 4% Animal feed 2% 2% Other foodstuffs 1% 7% Alcoholic beverages 0% 3% Non-metallic minerals 0% 11% Metallic ores 0% 4% Plastics/rubber 2% 6% Logs 27% 6% Wood prods. 27% 6% Newsprint/paper 9% 5% Paper articles 1% 5% Printed prods. 1% 5% Non-metal min. prods. 15% 5% Base metals 0% 8% Machinery 1% 2% Motorized vehicles 2% 6% Furniture 1% 1% others 5% 11% Applying the appropriate rail share based on the type of commodity (last columns of Table  3-4 and Table  3-5) to the numbers in Table  3-16, and Table  3-17, the total share that could have been shipped by Okanagan Railway if it had existed in 2007 was estimated as shown in Table  3-19.  Applying the rail share provided in Table  3-19 to the estimates in Table  3-9, and Table  3-10, the volume of freight that could have been shipped by the Okanagan Railway if it had existed in 2007 was determined as shown in Table  3-20. - 82 - Table  3-18 Commodity composition between Okanagan and the rest of Canada Commodity Okanagan to northern BC and the rest northern BC and the rest to Okanagan Automotive 2.00% 5.00% Building materials 5.00% 14.00% Empty 46.00% 38.00% Food/beverage 12.00% 8.00% Forest products 16.00% 12.00% General freight 13.00% 17.00% Petrochemical 1.00% 8.00% N/A 4.00% 0.00% Table  3-19 Possible rail share of the Okanagan freight movements O/D  rail share Okanagan to northern BC and the rest 17.15% northern BC and the rest to Okanagan 18.25% Okanagan to GVRD 26.25% GVRD to Okanagan 19.20% Canada to USA through Osoyoos 57.00% USA to Canada through Osoyoos 17.75% Table  3-20 Potential rail traffic in the Okanagan area              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 41.83 54.23 6.95 87.89 113.28 Penticton 5.92 3.97 0.00 0.00Northern BC 51.11 6.51    1.55 1.89Vancouver 84.48       Rest of Canada 33.11 14.75    3.48 4.25West USA 18.25 0.00 0.45 0.72   Rest of USA 42.88 0.00 0.75 1.70   - 83 - The second group of studied freight consisted of the freight traffic through customs ports other than Osoyoos, which did not use any segments of Highway 97 between Osoyoos and Vernon. This group included through truck traffic crossing Sumas-Huntingdon customs port (Table  3-11); through truck traffic crossing Blaine-Douglas customs port ( Table  3-12); and through truck traffic crossing Eastport-Kingsgate customs port (Table  3-13). As there were already railways parallel to the highways at these ports, freight shippers have had the option of switching from trucks to rail for some time.  That portion diverted has been shown in Table  3-4 and Table  3-5.  While rail is the preferred mode for shipping large volumes of low-value bulk commodities long distances, handling characteristics of the cargo also impact the mode choice. Therefore, rail typically has not been used to handle high-value and perishable commodities.  However, more diversions from truck to rail could be reasonably expected as rail alternatives improved. (Columbia River Crossing, 2006). To calculate the amount of freight that could divert to rail (divertible commodities), the commodities were divided into three groups: Commodities captive to truck: commodities with average rail share of less than 15%; Commodities captive to rail: commodities with average rail share of more than 85%; and, 3) Divertible commodities: commodities with an average rail share of 15% to 85%.  It was assumed that none of the captive commodities would change their transportation mode, while divertible commodities would have some potential of changing modes, depending on conditions. Table  3-21 shows the volume of divertible commodities estimated from Table  3-11 to Table  3-13. The percentage of diversion depended on several factors. First, the rail industry had more focus on long-haul, single-commodity cargo trips between major hubs, because these movements require less handling and operational costs.  They also generated the highest profits for rail carriers. Second, while trains were more fuel-efficient (especially electrical) than trucks, both modes would be affected by a rise in fuel costs (Columbia River Crossing, 2006). - 84 - Table  3-21 Divertible commodities from all border crossing (except Osoyoos) traffic              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelownaPentictonNorthern BC     261.84 272.76 Vancouver      Rest of Canada     246.08 411.39 West USA 168.44  9.66  Rest of USA 182.51  10.48  Based on these factors, plus reflecting on truck/rail diversion data in Table  3-4 and Table  3-5, a conservative estimate of diverted commodities was assumed to be between 5 and 15 percent. A cross elasticity of demand calculation was done for comparison. As shown in Equation 3-4, cross elasticity of demand measures the responsiveness of the quantity demanded for rail to a change in the price of truck transportation.TRTRPQE'' %%,           3-4 Where TRE,is cross elasticity of demand for rail to truck; RQ'%  is % change in quantity demand for Rail; and TP'% is % change in truck cost.  Therefore, a cross elasticity of 0.52 (suggested by Association of American Railroads) means that a 1 percent increase in truck costs will result in 0.52 percent increase in rail demand. It reveals that a 5 to 15 percent diversion would be expected if truck costs rose by 19.6 to 28.8 percent (e.g. fuel costs due to peak oil), all other factors remaining constant (Reebie Associates, 2002, HNTB Team, 2007). Therefore, for this research, it was assumed that 10% of the divertible commodities would switch from truck to rail after introduction of an Okanagan Railway.  - 85 - The rail share for the all the railway in the region was calculated by summation of current rail traffic between the likely origin/destination pairs; potential rail traffic in the Okanagan area; and a 10% of potential diverted traffic from truck border crossings. A summary of the estimates is given in Table  3-22 wherein freight volumes total about 10.5 million tons. Table  3-22 Potential rail traffic for Okanagan Railway              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 41.83 54.23 6.95 87.89 113.28 Penticton 5.92 3.97 0.00 0.00Northern BC 51.11 6.51    2105.99 1713.63 Vancouver 84.48      Rest of Canada 33.11 14.75   3767.208 1581.81 West USA 18.25 0.00 257.76 174.07 Rest of USA 42.88 0.00 129.06 132.00 3.4.4 Rail assignmentRail traffic was assigned to the Okanagan Railway using all three methods described in Section 3.3.3, including shortest path, Lansdowne rule-base and spatial separation methods as follows:   3.4.4.1 Method # 1: shortest path  First, a centroid was assigned to each origin and destination. The centroid location was defined as the point at which shippers have at least two choices of railway. The centroids are shown in Table  3-23. At the second step, the shipping distances between centroids of origin and destination zones were calculated for three routes as shown in Table  3-24. - 86 - Table  3-23 Centroid for origin/destination zones zone centroids West USA Everett Rest of USA Sandpoint, Wenatchee, Spokane Northern BC Kamloops Rest of Canada Kamloops, Golden Table  3-24 Actual distance between the centroids of O-D pairs railroad1: Blaine railroad2: Oroville railroad3: Eastport Kelowna - Everett 6.68831 7.43491 18.88963 Kelowna - Wenatchee 13.00142 8.46663 12.57652 Kelowna - Spokane 6.68831 7.43491 18.88963 Kelowna - Sandpoint 14.31343 9.77864 11.26452 Penticton - Everett 7.59032 6.04444 19.79164 Penticton - Wenatchee 13.90343 7.07616 13.47853 Penticton - Spokane 7.59032 6.04444 19.79164 Penticton - Sandpoint 15.21543 8.38816 12.16652 Kamloops - Everett 4.70416 9.41906 16.90548 Kamloops - Wenatchee 7.34485 6.77837 14.26479 Kamloops - Spokane 11.01727 10.45078 10.59237 Kamloops - Sandpoint 16.46047 15.89398 5.14918 Golden - Everett 8.83535 13.55025 12.77429 Golden - Wenatchee 11.47605 10.90956 10.13360 Golden - Spokane 15.14846 14.58198 6.46118 Golden - Sandpoint 16.46047 15.89398 5.14918 Finally, using the logit distribution model as stated in Equation 3-3, the share for the Okanagan Railway was estimated. Table  3-25 shows the amounts of assigned traffic to the Okanagan Railway by the shortest path method. - 87 - Table  3-25 Assigned traffic to Okanagan Railway using shortest path method              Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 99.00 54.23 6.95 87.89 113.28 Penticton 5.92 3.97 0.00 0.00Northern BC 51.11 6.51    20.35 719.36 Vancouver 84.48       Rest of Canada 33.11 14.75    36.69 323.41 West USA 18.25 0.00 2.81 2.25   Rest of USA 42.88 0.00 60.46  22.65   3.4.4.2 Method # 2: Lansdowne rule-base method As explained earlier, the Lansdowne rule-base method is based on four principles: 1. Minimizing the number of interline transfers; 2. Maximizing the revenue division for the originating carrier;  3. Finding the shortest route in each carrier’s subnetwork; and  4. Dividing the shipments among the originating carriers based on the shipping distance. To apply the above rules, it was necessary to define the ownership of Okanagan Railway. The volume of assigned traffic to the Okanagan Railway, which is running under CN or CPR management and ownership would be different from an Okanagan Railway that is managed by a third company. Given the history of CN and some discussions with KPR management committee, it was assumed that the Okanagan Railway, like many other short lines in Canada, would be owned by a private company other than CN or CPR. It was also assumed that the Okanagan Railway, KPR, and CSCD would form a new carrier between Kamloops, BC to Wenatchee, WA.  The first rule indicated that the shippers tend to choose the routes with fewer transfers between couriers.  The second rule suggests that the originating carrier tends to select - 88 - routes that maximize its own revenue. A reasonable approximation of revenue for each carrier is the ratio of the distance on its tracks to the total shipping distance between the origin and the destination. Therefore, the originated carrier usually tends to keep freight on its tracks as long as possible. Applying these two rules to Table  3-22, some of the coloured cells would be crossed off. Based on the first rule, none of the movements between Northern BC and West of the United States would be assigned to the Okanagan Railway, mainly because the number of transfers would increase. Based on the second rule, none of the movements between the rest of the Canada and the United States (both west and the rest) would be assigned to the Okanagan Railway because the originated carriers would prefer to maximize the shipping distance on their tracks, and thus maximize their revenue.  The third and the fourth rules, finding the shortest paths and dividing the movements among them based on distance, basically followed the same procedure as described in the shortest path method. Lansdowne proposed the same cost function and choice model as the shortest path method. Therefore, the freight assigned to the Okanagan Railway with the Lansdowne rule-based method was the same as the assigned traffic to Okanagan Railway using the shortest path method, for all the coloured cells except those that are removed due to the first and second rules, as shown in Table  3-26.Table  3-26 Assigned traffic to Okanagan Railway using Lansdowne rule-base method             Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 99.00 54.23 6.95 87.89 113.28 Penticton 5.92 3.97 0.00 0.00Northern BC 51.11 6.51    0.00 719.36 Vancouver 84.48       Rest of Canada 33.11 14.75    0.00 0.00West USA 18.25 0.00 0.00 0.00   Rest of USA 42.88 0.00 60.46  0.00   - 89 - 3.4.4.3 Method # 3: spatial separation method As explained earlier, spatial separation method has also applied the industry practices. It combines all the practices and forms a new variable called “the spatial separation index”. It then uses the index instead of the cost function in the logit distribution model. As explained earlier, the spatial separation index for each railway would be defined using the actual distance and the initial traffic of the railway. Table  3-27 shows the index of spatial separation between centroids of origin and destination zones, calculated for three possible routes. The effect of traffic density can be observed by comparing Table  3-27 and Table  3-24. The indices for highly congested railways such as CN between Vancouver and Kamloops are much less than the actual length of those railways. However, the spatial separation indices are the same as the actual length for the railways with low traffic density. Table  3-27 Index of spatial separation for the centroids of O-D pairs rail1: Blaine rail2: Oroville rail3: Eastport Kelowna - Everett 3.62270 6.92384 13.92738 Kelowna - Wenatchee 8.74813 7.79002 8.80195 Kelowna - Spokane 3.62270 6.92384 13.92738 Kelowna - Sandpoint 9.95366 8.99555 7.59642 Penticton - Everett 4.52470 5.53337 14.82938 Penticton - Wenatchee 9.65013 6.39955 9.70395 Penticton - Spokane 4.52470 5.53337 14.82938 Penticton - Sandpoint 10.85567 7.60508 8.49842 Kamloops - Everett 1.63855 8.90799 11.94323 Kamloops - Wenatchee 3.76818 6.77837 9.81360 Kamloops - Spokane 6.76398 9.77417 6.81780 Kamloops - Sandpoint 7.96951 10.97970 5.61226 Golden - Everett 2.19197 9.46142 11.38980 Golden - Wenatchee 4.32160 7.33179 9.26018 Golden - Spokane 7.31740 10.32759 6.26438 Golden - Sandpoint 8.52293 11.53313 5.05884 - 90 - Using the logit distribution model as stated in Equation 3-3, the share of the Okanagan Railway is estimated. Table  3-28 shows the amounts of the assigned traffic to the Okanagan Railway by the shortest path method. Table  3-28 Assigned traffic to Okanagan Railway using spatial separation method             Origin DestinationKelowna PentictonNorthernBCVancouverRest of CanadaWestUSARest of USAKelowna 99.00 54.23 6.95 87.89 113.28 Penticton 5.92 3.97 0.00 0.00Northern BC 51.11 6.51    3.02 122.79 Vancouver 84.48       Rest of Canada 33.11 14.75    6.10 33.85 West USA 18.25 0.00 0.64 0.84   Rest of USA 42.88 0.00 4.35 3.64   3.4.5 Demand forecasts for the Okanagan Freight RailwayTraffic has been assigned to the Okanagan Railway using three methods of rail assignment. The shortest path method is an optimistic method, which predicts the demand of 1.8 million tons for the Okanagan railway. Lansdowne rule-base method is an academic method based on the industry practices. It predicts the demand of 1.4 million tons for the Okanagan Railway. Spatial separation method is a method in favour of major railways. It predicts the demand of 0.8 million tons for the Okanagan Railway. The results from Lansdowne’s method would be used as the base for the benefit cost analysis. However, in the sensitivity analysis section, the two other results would be used as the optimistic and pessimistic estimates, respectively.  All of the above numbers are demand forecast values for the year 2007. Using the developed growth factor, the demand for the years between 2009 and 2079 has been forecasted as shown in Table  3-29. Note that the results from the Lansdowne rule-based method assignment have been used as the base. - 91 - Table  3-29 Okanagan Railway freight demand forecast Osoyoos-VernonKelowna-VernonKelowna-Osoyoos Penticton-VernonTotal ‘000 metric tons Total Truck2009 841.99 434.03 192.71 32.84 1,501.57 24,714 2019 1,166.04 550.06 241.39 41.32 1,998.80 32,898 2029 1,490.10 666.08 290.07 49.79 2,496.03 41,081 2039 1,814.15 782.10 338.75 58.27 2,993.27 49,265 2049 2,138.20 898.12 387.43 66.74 3,490.50 57,449 2059 2,462.26 1,014.15 436.11 75.22 3,987.73 65,633 2069 2,786.31 1,130.17 484.79 83.69 4,484.96 73,816 2079 3,110.37 1,246.19 533.47 92.17 4,982.19 82,000 To calculate some of the railway benefits, it was necessary to estimate the number of trucks that the forecasted freight demand would have replaced, in case no railway was made in the valley. Those benefits include the savings that are related to the number of trucks that are removed from the highways (transferred trucks). The volumes of commodities were converted to the number of transferred trucks, using Table  3-30 for border crossing freights and Table  3-31 for local freights. Note that there is a percentage of empty trucks in each corridor. The number of empty trucks is added to the number of transferred trucks that was calculated in the last step.  To convert the number of annual trucks to daily trucks, it was assumed that there are 300 trucking days per year, as the Highway Capacity Manual suggests (an average truck works 5 workdays per week, plus at 44% capacity on the weekends, yielding 306 workdays per year, minus 6 federal holidays yields 300 days per year). The Last column of Table  3-29 shows the number of transferred trucks due to the diversion of freight from truck to the Okanagan Railway.- 92 - Table  3-30 Payload factors by two-digit for border crossing  Description Tons per Truck Live animals and live fish 11.11 Cereal grains 12.68 Agricultural products except live animals, cereal grains, and forage products 10.04 Animal feed and feed ingredients, cereal straw, and eggs and other products of animal origin n.e.c. 10.42 Meat, fish, seafood, and preparations 13.92 Milled grain products and preparations, and bakery products 5.37 Prepared foodstuffs n.e.c. and fats and oils 11.76 Alcoholic beverages 9.33 Tobacco products 11.42 Products of petroleum refining n.e.c. and coal products 19.90 Basic chemicals 13.33 Pharmaceutical products 4.65 Fertilizers and fertilizer materials 9.00 Chemical products and preparations n.e.c. 11.08 Plastics and rubber 8.77 Wood products 15.91 Pulp, newsprint, paper, and paperboard 8.39 Paper or paperboard articles 14.99 Printed products 11.92 Textiles, leather, and articles 5.00 Non-metallic mineral products 9.35 Base metal in primary or semi-finished forms and in finished basic shapes 14.08 Articles of base metal 11.09 Machinery 6.53 Electronic and other electrical equipment and components, and office equipment 11.83 Vehicles 6.27 Transportation equipment n.e.c. 7.12 Precision instruments and apparatus 15.55 Furniture, mattresses and mattress supports, lamps, lighting fittings, and illuminated signs 4.09 Miscellaneous manufactured products 6.40 Sources: (Alam et al., 2007)- 93 - Table  3-31 Payload factors for the valley Commodity Tons per truck Automotive 4.40 Building Materials 11.64 Empty 0.00 Food & Beverage 12.56 Forest Products 15.28 General Freight 3.88 Petrochemical 34.70 N/A 20.47 Data source: Okanagan traffic survey analysis local survey3.5 Summary In this chapter, first, freight model classes and the components were reviewed and the O-D factoring method was selected to be employed to in the Okanagan Valley. Then, an O-D table for the freight movements in the region was produced using observed and survey data. The rail share was estimated using the fixed shares based on the commodity type. Finally, the predicted share of freight rail was assigned to the rail network using three methods of assignment, which provided three freight demand forecasts. The mid-range forecast was selected as the freight demand for the Okanagan Valley.- 94 - 4 COST ESTIMATION  4.1 Introduction In this chapter, the costs of the Okanagan Railway are reviewed.  In section  4.2, the optimum alignment for the railway is selected. The process includes the evaluation of socio-economic, geological, hydrological and environmental parameters of the valley to minimize the environmental impacts of the railway. Section  4.3 includes a review of the capital, operating/ maintenance, construction management, and contingency costs of similar railways, and uses these to estimate the costs of an Okanagan Railway. 4.2 Alignment selection 4.2.1 Literature review and methodology The optimal alignment for the Okanagan Railway was determined using Geographic Information System (GIS) tools. A six-step methodology, employed by most of the literature, evaluated socio-economic, geological, hydrological and environmental parameters of the valley to minimize environmental impacts of the railway, including: 1) Definition of the set of evaluation criteria2) Determination of the suitability ranking for each criterion  3) Data collection and generation of the cost surface for each criterion based on each suitability ranking 4) Estimation of the weights for each criterion  5) Generation of the suitability cost surface  6) Determination of the shortest alignment based on the suitability cost surface. - 95 - GIS technology was selected because of its ability to integrate geo-referenced data from different sources and file formats - land-use, soil, water, infrastructure, settlements, and vegetation - to assist in the evaluation process. GIS also offers several Multi Criteria Evaluation (MCE) operators which are useful for planning problems. For this research, ArcGIS 9.1, provided by ESRI, was used. Definition of evaluation criteria and suitability rankings Most research found in the literature used similar evaluation criteria, with variations influenced mainly by author bias and study purpose.  Table  4-1 shows the list of the evaluation criteria in the most comprehensive studies with similar study purposes (Dane et al., 2007, Djenaliev, 2007, Roets, 2002). Table  4-1 Evaluation criteria in the literature Roets, 2002 Djenaliev, 2007 Dane, 2007 Gradient and Slope Slope/Gradient Slope Geology and Soil  Litology Hydrology Water bodies River Areas of Conservation   Vegetation  Land Cover  Land Use Land Use Land Use/ TopographyExisting Infrastructure Rail Stations /Roads/PowerLineExistingRoads/ Fault Line Built-up Areas Settlements Settlement Considering the scope of this study and the data availability, the following criteria were chosen to be considered for the Okanagan route evaluation: Land Use, Slope, Hydrology, Safety, Parks and geology.  Each criterion was classified, and each class ranked by a defined parameter, called suitability cost. Suitability costs’ values were between 1 and 10 where 1 was the most suitable class and 10 was the least suitable class. A suitability cost - 96 - value of 1000 was considered for the classes that were restricted to construction i.e. no railway could be built there.  Note that all GIS data was provided by DMTI spatial in CanMap® RouteLogistics unless otherwise stated. Estimation of the weight of each criterion  While all the defined criteria are influential in the decision making process, some are more important than the others. In this study, the Analytical Hierarchy Process (AHP) was used to determine the weights of relative importance of the criteria in rail route planning. The AHP method uses a scale with values from 1 to 9 to rate the relative preferences for two criteria as shown in Table  4-2.  Table  4-2 Scale for pairwise comparison. Intensity of Importance Definition Explanation 1 Of equal value Two requirements are of equal value 3 Slightly more value Experience slightly favours one requirement over another 5 Essential or strong value Experience strongly favours one requirement over another 7 Very strong value A requirement is strongly favoured and its dominance is demonstrated in practice 9 Extreme value The evidence favouring one over another is of the highest possible order of affirmation 2, 4, 6, 8 Intermediate values between two adjacent judgments When compromise is needed The comparison matrix is reciprocal; that is, if criterion A is twice as preferred as criterion B, we can conclude that criterion B is preferred only one-half as much as criterion A. Therefore if criterion A receives a score of 2 relative to criterion B, criterion B should receive a score of ½ when compared to criterion A. Djenaliev calculated the relative criterion weights of all the criteria by completing the matrix of pairwise comparison as shown in Table  4-3 (Djenaliev, 2007). - 97 - Table  4-3 Relative criterion weights LandUse(LU) Slope(SL)Park(PR)Road(RD) Hydrology(HY) Geology(GE)WeightLand Use (LU) 1 2 3 2 3 5 0.31Slope (SL) 0,5 1 2 4 3 2 0.24Park (PR) 0,5 0,25 2 1 2 3 0.17Road (RD) 0,33 0,33 2 0,5 1 0,33 0.1Hydrology (HY) 0,2 0,5 3 0,33 3 1 0.13Geology (GE) 0,2 0,25 0,5 0,17 0,25 0,5 0.04Sources: (Djenaliev, 2007)Proposed alignments Once the suitability cost surface is created, spatial analysis tools of ArcGIS 9.1 including Cost Distance tool and Cost Path tool were employed to determine the least-cost route, given a shortest optimum path for the proposed rail route.4.2.2 Application of the method and results 4.2.2.1 Generation of the cost surfaces Land Use Land use is recognized in all similar studies as an influential criterion that can impact both the capital cost of the railway and the passenger demand. The Land Use data for the study area is classified into four classes. A cost value of 10 is assigned to the commercial and the industrial areas, where the value of right of way acquisition is too high. A cost value of 6 is assigned to the residential areas, which means these areas are moderately suitable due to the noise and safety impacts of the railway. A cost value of 2 and 1 is - 98 - assigned to the open areas, and areas with no Land Use, respectively, which are the most suitable Land Uses for the railway alignment. It should be noted that the cost value assignment in this study is consistent with the assignments in the literature. Figure  4-1 shows the Land Use’s cost surface for the Okanagan Valley.  Figure  4-1 Land use’s cost surface for the Okanagan Valley - 99 - SlopeSlope and gradient (slope percentage) of the land can impact the construction costs and passenger amenity drastically. In mountainous areas like the Okanagan Valley, slope is the most important criterion for railway route selection. If only used for high speed transportation, railways can incorporate steeper gradients. Gradients of high speed rails in France and Germany can reach 4%. For freight railways where the gradient is a critical element, it could be 1% for loaded trains and 2% for empty trains (Beale, 2006). The slope of the Okanagan Valley is calculated using the Digital Elevation Modal (DEM) data that is provided by the Canadian Council on Geomatics (CCOG). The DEM is converted to a raster file and then the slope of the area is calculated using the ArcGIS Spatial Analysis tool. Slope data is categorized into eleven classes according to the standards for railway construction (Djenaliev 2007). Following the literature review, the cost values for the slope are assigned as shown in Table  4-4.   Table  4-4 Cost values for slope’s cost surface Percentage of Slope Cost value 0 % - 3 % 1 3 % - 5 % 2 5 % - 8 % 3 8 % - 10 % 4 10 % - 15 % 5 15 % - 18 % 6 18 % - 20 % 7 20 % - 25 % 8 25 % - 30 % 9 30 % - 35 % 10 35 % - 594 % 1000 The slope’s cost surface developed based on the cost values in Table  4-4 is shown in Figure  4-2. Note that while the values in Table 4-4 are for similar mountainous area, they might change in narrow valleys such as Okanagan. Further studies in this subject are suggested.- 100 - Figure  4-2 Slope’s cost surface for the Okanagan Valley - 101 - Geology and Soil According to the literature, the stability of the soil should also be investigated. Therefore, the Okanagan Valley’s earth is classified based on the type of the soil and geology. Geology data of the area is provided by the Digital Geology Map of British Columbia. Table  4-5 includes the information regarding the cost value of the simplified geology (Djenaliev, 2007, Roets, 2002).  Using these cost values, the geology’s cost surface for the Okanagan Valley were developed as shown in Figure  4-3. Table  4-5 Cost value for geology's cost surface Type of geology Cost value Quartzite, Quartzite with interbedded shale and conglomerate, and Diabase. 5 Sandstone 1 Acid lava, Andesite, Basaltic lava, Basic & ultrabasic rocks, Hornfels, Hybrid rocks, Hortonolite dunite, Hybrid, metasomatised & fenitised rocks, Pyroclasts, Lava, Granophyne, Diamictite and Volcanic rocks. 5Dolomite, Breccia, and Chert 10 Dolerite 1Granite, Gneiss, and Grey 5 Gabbro and Ferrogabbro 1 Shale 1Surface deposits; Alluvium 10 Digital Geology Map of British Columbia only includes the rock types, and does not provide any information about the soil. The soil analysis of the area was conducted from a geomorphologic perspective, meaning that the type of geology was related to the soil formations in terms of its erosion potential. Roets provides information regarding the cost value of the soil as shown in Table  4-6. Using these cost values, the soil’s cost surface of the Okanagan Valley is developed as shown in Figure  4-4. - 102 - Table  4-6 Cost value for the soil's cost surface Soil (Simplified)  Underlying Geomorphology  Cost Value SandAlkali-Feldspar, Arenite, Breccia, Chert, Conglomerate, Gneiss, Granite, Granodiorite, Granophyre, Metamorphic rocks, Migmatite, Pyroclastic Breccia, Rhyolite, Slate, and Tillite  1 ClayAmphibolite, Andesite, Basalt, Diorite, Dolerite, Dunite, Gabbro, Lava, Norite, Sedimentary rocks, Shale, and Syenite  5Deep Eroded Soil  Dolomite  10 Figure  4-3 Geology's cost surface in the Okanagan Valley - 103 - Figure  4-4 Soil's cost surface in the Okanagan Valley Road (Safety/Crossing) The literature on rail planning recommended avoidance of highway crossings, for several reasons: disturbance in road traffic, related safety issues, and the costs of grade crossings. A cost value of 5, as suggested in the studied literature, was assigned to the roads (including the 50 meter buffer) in the study area. Figure  4-5 shows the road’s cost surface.- 104 - Figure  4-5 Road’s cost surface in the Okanagan Valley HydrologyDue to the high cost of bridge construction, considerations were also given to keeping the rail alignment away from lakes, wetlands, and river crossings. In addition to cost, these impact aquatic ecosystems. Therefore, a detailed understanding of the hydrological cycle of the study area was conducted in order to predict any changes in the surface and groundwater flows as well as the aquatic ecosystems. Given the many mountain rivers in - 105 - the Okanagan Valley study area, and it was not possible to avoid them entirely. Therefore a cost value of 2 was assigned to mountain rivers (including the 50 meter buffer), and, a cost value of 6 was assigned to the Kelowna rivers due to their width. A cost value of 1000 was assigned to crossing Okanagan Lake. Figure  4-6 shows the hydrology’s cost function of the area.Figure  4-6 Hydrology’s cost function in the Okanagan Valley - 106 - Provincial Parks A cost value of 10 was assigned to provincial parks where construction was least desirable (see Figure  4-7).  Figure  4-7 Provincial park's cost surface in the Okanagan Valley - 107 - Final Suitability cost surface Using the weights in Table  4-3, the final suitability cost value would be defined as in Equation 4-1.GEHYRDPRSLLUCCCCCCyCostSuitabilit04.014.010.017.024.031.0     ( 4-1) The existing rail network in the area including the Kelowna Pacific railway (KPR) and the abandoned Kettle Valley railway (KWR) imposed some extenuating conditions on the cost surface. It was assumed that the suitability cost value of the zones in the Right of Way (ROW) of the railways would be different than the other zones because some infrastructures were already installed in the case of the KPR rail line, or, because a primary earthwork like Right of Way preparation had been done in the case of the KVR. Therefore discount factors were applied to the ROW of KPR and KVR as explained in Table  4-7. A discount factor of 0% meant that there would not be any change in the cost. A discount factor of 50% meant that the cost would be cut into half, and a discount factor of 100% means that the cost would be eliminated.  Table  4-7 Discount factors for the ROW of KPR and KVR Criterion KPR KVR Land Use 0% 0% Slope 50% 50% Park 100% 50% Road 0% 0% Hydrology 50% 0% Geology 100% 50% Therefore, the suitability costs for the ROW of KPR and KVR are defined as shown in Equation 4-2 and 4-3 respectively. - 108 - 2/14.010.02/24.031.0HYRDSLLUKPRCCCCyCostSuitabilit       ( 4-2) 2/04.014.010.02/17.02/24.031.0GEHYRDPRSLLUKVRCCCCCCyCostSuitabilit         ( 4-3) Using the above suitability cost, the suitability cost surface is generated as shown in Figure  4-8.  Figure  4-8 Suitability cost surface - 109 - 4.2.2.2 Proposed alignments Kelowna Osoyoos segment Three possible alignments are found for this segment, as shown in Figure  4-9. The blue one is built using the cost function without applying any discount factors to the ROW of KPR and KVR (described above). The red one is built using the same cost function only this alignment is forced to pass through the Westbank and Penticton, where a high demand is forecasted. The black one, which lies over the red alignment for most of the way, is based on the modified cost function, which includes the discount factors for the ROW of KPR and KVR. It is forced to pass through the Westbank and Penticton as well. As Figure  4-9 shows, this alignment only uses the KVR’s ROW in a short part of its path. Figure  4-9 Proposed alignment for Okanagan Railway Kelowna-Osoyoos - 110 - Kelowna Vernon segment Three possible alignments are found for this segment as shown in Figure  4-10. The blue one is built using the cost function, without applying any discount factors to the ROW of KPR and KVR. The purple alignment is built using the same cost function; only this alignment is forced to pass through Lake County and Kelowna Airport where high demand is forecasted. The black one has used the modified cost function which includes discount factors for the ROW of KPR and KVR. It is forced to pass through the Lake County and the Kelowna Airport as well. As shown in Figure  4-10, the black alignment lays over the KPR between Vernon and Lake County, but enters Kelowna through a less urban path than KPR.Figure  4-10 Proposed alignment for Okanagan Railway Vernon-Kelowna - 111 - The entire alignment is shown in Figure  4-11. As it was explained in the passenger demand forecast section, there was not enough data accessible to accurately locate the stations. It was found reasonable to assume that stations would be located in Vernon, Lake County,  Kelowna Airport, Kelowna-downtown, Westbank, Peachland, Summerland, Penticton and Osoyoos (border). However, more research and analysis would be needed to provide the necessary information to decide about the other communities in the valley such as Oyama, and Oliver.  Figure  4-11 Selected alignment for Okanagan Railway - 112 - Table  4-8 summarizes the primary attributes of the selected alignment, which would be utilized as the basis for all the other estimations. Table  4-8 Primary attributes of the selected alignment for Okanagan Railway Distance between stations Length (km) Vernon - Lake County 30.30 Lake County – Kelowna Airport 8.27 Kelowna Airport - downtown, Kelowna 10.99 Downtown, Kelowna - Westbank 11.02 Westbank -  Peachland 14.96 Peachland - Summerland 25.57 Summerland - Penticton 16.41 Penticton - Osoyoos 58.71 Total length 173.69 Track condition Length (km) New track on existing ROW of KVR 29.44 New track including ROW preparation 105.69 Upgrading existing track of KPR 38.57 Number of crossings  River crossing 21 Lake crossing 1 Highway crossing 4 4.3 Capital and O&M costs 4.3.1 Literature review and methodology Capital costs are those outlays for building and equipping a line with the necessary tracks, yards, structures, signals, rolling stocks and appurtenances. In every feasibility study a group of engineers are responsible for estimating the local capital costs as the cost of some items might vary drastically by the location. Unfortunately, railway capital cost - 113 - estimates were not found in any literature for the Okanagan region. Therefore, it was decided to estimate the capital costs using the cost estimations of similar projects found elsewhere in North America in the literature. Given the uncertainty this would introduce when translated to the Okanagan Valley, to capture any impact of overestimation or underestimation, a complete sensitivity analysis was done. 4.3.1.1 Track cost Two similar projects were found in the literature, where comprehensive cost estimation was conducted (T.Y. Lin International Bascor, 1999, R.L. Banks & Associates, Inc., 2005). First, the Outer Circumferential (OC) commuter rail feasibility study, has thoroughly investigated the capital costs of a commuter rail with maximum speed of 130 km-per-hour in Illinois, US.  The entire route was 600 kilometres, where the distance between the stations varies from 70 km to 150 km.  Second, the Riverside County Transportation Commission (RCTC) commuter rail feasibility study investigated building a commuter rail with a maximum speed of 79 mph (130 kph) in Riverside, California. The route consists of two parts of 50 and 120 kilometres. In both studies, capital cost estimates were developed for several scenarios, including building a new track, building a new track while using the old track’s ROW; and upgrading the existing freight track. Capital costs included building or upgrading tracks, turnouts, at-grade, highway rail crossings, signals, drainage and earthwork. Figure  4-12 shows the cost of building a new track on the prepared ROW for the different segments of OC and RCTC commuter rails. It shows that the costs were independent of the length of the segment and while RCTC costs were slightly higher than OC estimates, they were in the same range.  - 114 - -0.501.001.502.002.500 50 100 150 200Length (km)Cost (million dollars per km)OC RCTCFigure  4-12 Unit cost of a new track on the existing ROW Table  4-9 shows the estimates for three different scenarios similar to the possible scenarios for the Okanagan Valley. Maximum and minimum estimates would be employed in the sensitivity analysis.  Table  4-9 Unit cost of track construction Item Average2004$ million Maximum 2004$ million Minimum 2004$ million New track on existing ROW                     1.82         2.02         1.61  New track including ROW preparation                     2.18         2.28         2.01  Upgrade existing track to 130 kph                     1.71         1.91         1.49  4.3.1.2 ElectrificationThe existing North American railway system employs mostly diesel propulsion including the two discussed projects. Therefore, the above costs did not include the electrification costs. However, considering the peak oil phenomena, environmental concerns about fossil fuel, and the last 30 years of European experience, it was expected that the North American railway system would eventually switch to electric propulsion. Therefore, electric propulsion was selected for the Okanagan Railway. While electrification may impose some higher initial costs, it was found that in the long run it was more - 115 - economically beneficial (Shonsey, 2006). In any case, the lifecycle cost difference was found to be minimal given emerging new types of energy storage technology, including ultra-capacitors. Ultra-capacitors store charge in an electric field similar to regular capacitors but instead used advanced materials in a sponge or matrix to increase surface area drastically. The increase in area greatly increases storage capacity and so ultra-capacitors have a higher energy density. Flaherty proposed using ultra-capacitors in conjunction with the Green Goat® design to improve its battery service life (Flaherty, 2005). A small bank of ultra-capacitors would act as a buffer between the batteries and the traction motors. The ultra-capacitors would do most of the quick charges and discharges, thus prolonging the functional life of the lead-acid batteries without a significant increase in cost.  Ultimately, Flaherty proposed that the entire power system of the locomotive could be replaced with ultra-capacitors. With current technologies, a fully ultra-capacitor locomotive could operate at speed for roughly 23 minutes with a 1.1 minute recharge time. Currently, this option is cost prohibitive, as a locomotive with only ultra-capacitors would cost $1.5 million while a bank of lead-acid batteries would be $0.2 million. With ongoing research in the field, Flaherty envisioned ultra-capacitor locomotives that could be made at a comparable cost to current locomotives. These future locomotives could operate for 2.4 hours with a recharge time of just 7 minutes. For long distances, this means that only 5% of the track would have to have an appropriate supply rail or, at a speed of 200 km-per-hour, only 23 km of supply rail would be needed for every 480 km traveled. Furthermore, for smaller distances the train could be recharged during its stops at stations. Without the need for electrified third rails or overhead power lines, significant construction and maintenance costs could be saved. Overall, the possibilities of ultra-capacitors in rail locomotives result in environmental benefits of electric rail, without the increased initial construction costs. However, to be conservative in this study, the cost of electrification has been estimated using the current overhead lines, at 0.64 million dollars - 116 - per kilometre, and added to the initial capital cost. Substations assumed to be spaced at 40-km intervals and 2% additional track mileage were assumed to be added for electrification of yards, and terminals (National Research Council, 1991). 4.3.1.3 Other capital costs The cost of the other infrastructure, including stations, bridges and rolling stock was estimated using the literature as shown in Table  4-10. The cost of a bridge over Okanagan Lake was estimated to be the same as the new William R. Bennett Bridge ($150 million).  Table  4-10 Other capital cost estimates Item Unit Cost in million 2007 US$ Bridge over minor river Each 0.85 Bridge over the Lake Each 150? Small station Each 4 Large station Each 8 rolling stock   Freight locomotive  Each 3.0 cars Each 0.1 passenger Electric Multiple Unit (EMU) Each 3.0 Passenger rolling stock was assumed to have a capacity of 100 passengers, while freight car capacity was set at 120 tons. Freight locomotive were assumed capable of pulling 30 cars.4.3.1.4 Operating and maintenance costs Operating and maintenance costs per passenger-kilometre and ton-kilometre are shown in Table  4-11. Passenger rail O&M costs were estimated using the existing commuter rail costs in the literature and freight O&M costs were estimated using existing class I railways (Bureau of Transportation Statistics, 2008).- 117 - Table  4-11 O & M costs Item Unit Cost  Passenger rail Passenger-km 0.20Freight rail Ton-km 0.024.3.1.5 Construction management and contingencies The cost estimates included a contingency level of 30% of estimated capital costs. This contingency level was considered appropriate as no Okanagan Valley rail facility had any in-depth design or engineering, even conceptually. This could be reduced, and the confidence in the capital cost estimates increased, pending further research and a comprehensive engineering design study. Moreover, a 15% allowance for project management - design, engineering, and construction – was also included (T.Y. Lin International Bascor, 1999, R.L. Banks & Associates, Inc., 2005).   4.3.2 Application of the method and results The engineering, capital, operating and maintenance costs of railway were calculated based on similar NA railways, using the methodology described in above section.. Finally, the costs of the Okanagan Railway were estimated, and summarized in Table  4-12, inflated to the 2007 US dollars using the historic inflation rate. 4.4 SummaryThis chapter documented the pursuit of a main objective of this research, to review the design and cost criteria of building an Okanagan railway. The process of selection of the optimum alignment included the evaluation of socio-economic, geological, hydrological and environmental parameters. The resulting recommended rail alignment should minimize the environmental impacts of the railway as well as the costs. Moreover, the costs of construction and operating the railway were estimated. - 118 - Table  4-12 Total estimated construction costs for Okanagan Railway Item Unit Unit price Quantity Cost in million 2007 US$ New track on existing ROW km 1.98 29.44 58.21 New track including ROW preparation km 2.36 105.69 249.23 Upgrade Freight track to 79 mph commuter track km 1.85 38.57 71.44 Electrification km 0.64 173.69 111.05 Bridge over minor roads Each 0.85 21.00 17.85 Bridge over lake Each 150.00 1.00 150.00 Large station Each 8.00 1.00 8.00 Small  station Each 4.00 8.00 32.00 Design & construction management  15%   Contingency  30% Total Estimated Construction Costs (Including Engineering, Construction Management and Contingencies) 993.78- 119 - 5 BENEFITS 5.1 Introduction Once the costs of the railway were considered, an estimate of the benefits of the Okanagan Railway was needed. In section 5.2, the revenue of the railway resulting from fares, and the user and externality benefits of the passenger and freight rail, including noise, pollution, and congestion reduction, are estimated using the regular rates in North America. In section 5.3, a new method for calculating the road safety benefits of transportation infrastructure, such as railways and roads, has been introduced to replace collision rates traditionally used in cost-benefit analyses, to estimate the safety benefits of the Okanagan Railway.5.2 Rail benefits except safety benefits 5.2.1 Literature review and methodology Railways, when compared to highways, increase both freight and passenger external benefits (Gleason et al., 2005, Forkenbrock, 2001, Transport Concepts, 1994, Forkenbrock, 1999). In addition to user benefits, non-market benefits, or externalities,were examined, including congestion, noise, and pollution. To compare the benefits, the costs of rail transportation were subtracted from the costs of highway transportation. This was done to illustrate the savings that could be generated if the proposed railway were used instead of current rubber tire methods. All values were changed to 2007 US Dollars using historic inflation rate.5.2.1.1 Rail revenue Sources of the revenues of the new railway are the fares paid by the passengers and the freight carriers. Passenger fares are assumed to be the same as the current bus transit in - 120 - the area, as shown in Table  5-1.  For freight rail, fares charged would be similar to the current fare of Class I rail, which is approximately $0.03 per ton-mile because the proposed railway would connect with Class I rail lines at either end of the route. All values were changed to 2007 US Dollars using exchange rate of 0.9 provided by U.S. Treasury, Financial Management Service (U.S. Treasury, 2008).Table  5-1 Current inter-city transit fare in the Okanagan Valley O-D Bus fare Penticton to Kelowna $6.25 Penticton to Summerland $3.75 Okanagan Falls to Penticton $2.50 Okanagan Falls to Summerland $5.00 Okanagan Falls to Kelowna $7.50 Summerland to Kelowna $6.25 Oliver to Penticton $5.00 Oliver to Kelowna $10.00 Oliver to Okanagan Falls $3.75 Oliver to Summerland $7.50 Osoyoos to Kelowna Airport $10.00 Osoyoos to Summerland $7.50 Osoyoos to Penticton $5.00 Osoyoos to Okanagan Falls $3.75 Osoyoos to Oliver $2.50 Vernon to Kelowna $2.00 Kelowna to Vernon  $2.50 Oyama to Vernon $2.50 Oyama to Kelowna  $2.00 5.2.1.2 Okanagan Passenger Rail externality benefits In general, passengers benefit from commuter rail through reduced collisions, congestion, and pollution. It also helps to reduce the highway infrastructure maintenance costs for governments and the ownership and operating costs for users. An Okanagan Valley collision prediction model was developed for Highway 97 to estimate the road safety - 121 - benefits of the railway more accurately than traditional methods. As local values for other benefits were not available, data was used from the literature on inter-modal comparisons of long-run average benefits (Kågeson, 1993, Levinson et al., 1996a). The values for the autos were generated using the cost values for different cities of North America and Europe, and as such, are different than the ones that may be observed in the Okanagan Valley. A summary of the associated costs and benefits can be seen in Table  5-2.   Table  5-2 Passenger rail externalities benefits Cost Category$ per passenger-kilometre Auto Electric-rail Okanagan Railway savings  External: Congestion   0.006348 0.00000 0.00635  External: Noise   0.006210 0.00138 0.00483  External: Pollution   0.002989 0.00028 0.00142 Highway Infrastructure: Construction and Maintenance  0.016560 0.00000 0.01656 User: Operating and ownership 0.105578 0.00000 0.105578 5.2.1.3 Okanagan Freight Rail externality benefits As was done for the passenger rail benefits, the difference between the rail benefits and the truck benefits is analyzed. In the absence of the local data, the valuation estimates for trucking and for rail are from the literature (Gleason et al., 2005, Forkenbrock, 2001, Transport Concepts, 1994, Forkenbrock, 1999). Freight rail and truck benefits were summarized in Table  5-3.  Noise costs imposed by electric rail were assigned a value of zero, as the literature suggested that a given level of noise produced by a freight train (especially electric) is usually perceived as less annoying than noise produced by vehicle traffic on a highway. Moreover, the ``Green Book'' of the Commission of the European Communities stated that the cost of road traffic noise is over six times greater than noise from freight rail (Forkenbrock, 2001). However, further studies on the subject are needed. - 122 - Table  5-3 Freight rail externalities benefits Cost Category$ per ton-kilometre Truck Electric-rail Okanagan Railway savings  External: Congestion   0.010240 0.00000 0.014033  External: Noise   0.000494 0.00000 0.000552  External: Pollution  and greenhouse gases 0.003174 0.00041 0.002760 Infrastructure: Construction and Maintenance  0.007952 0.00000 0.009156 User: Operating and ownership 0.116196 0.00000 0.116196 Regarding transfers from rail to road, some were expected as necessary to get passengers or freight from their origins to the rail stations or from the rail stations to their destinations. However, this study was not detailed enough to address these movements, and so this was acknowledged as a topic for future researchers to address.5.2.1.4 Unquantifiable benefits The major unquantifiable benefits for both the freight and the passengers were the value of reliability and scheduling. For passengers, the value of reliability was significant as it can simplify scheduling appointments and planning daily life (Lam et al., 2001). Companies shipping freight by railway highly value reliability for shipping goods to customers as all parties can more accurately schedule workers and docking space, potentially saving large sums of money throughout the year. However, an approximate value for reliability is not known and is therefore not accounted for when enumerating the benefits of the railway. Furthermore, both passenger and freight benefit from an appropriate schedule that coincides with work schedules and commuting rush hours. Again, this simplifies the scheduling of appointments, work hours, and events. This is also not considered when enumerating the benefits of the railway as no value could be estimated for the region. An additional source of benefits would be the increased attractiveness to tourists coming to the Okanagan Valley. With the construction of a railway through the Okanagan Valley, there is an opportunity to implement a system similar to the Euro Pass which would allow tourists to ride the railway for a discounted price. This would encourage tourism as it - 123 - would reduce the cost of traveling the valley significantly and would allow a much wider tour through the valley as well as ease of mobility. Similarly, it would offer a more affordable option to commuters than that of a personal vehicle and could open employment options for the public that cannot afford a personal vehicle. A final benefit not considered is the multiplier effect which would be felt throughout the Okanagan Valley during the construction of the railway. The workers that would be constructing the railway would be staying within the region and would open job opportunities for local residents in the form of construction and extra service providers. 5.2.1.5 Travel time differences  Since the Okanagan Railway was projected to travel at a lower overall speed than cars on Highway 97, some allowance was made to account for Okanagan Railway commuters’ longer travel time. To calculate the cost difference, first the value of travel time in the region was defined. Studies by the Economic Development Commission (EDC) of the Central Okanagan showed that the local time value is $12.12 per person hour (HDR et al., 2007). Assuming the average speed of the autos on Highway 97 at 80 km per hour, and the average speed of the Okanagan Railway to be 60 km per hour (calculated in section 2.4.2), an estimate of the added travel time cost were calculated as shown in Equation 5-1.$05765.06012.128012.12)(    kilometerpassengerTimeSaving    ( 5-1) Captive to rail commodities usually do not require an express delivery. Therefore, the value of time is not considered in the enumeration of the cost of the freight shipped by truck (Litman, 2008). - 124 - 5.2.2 Application of the method and results Unlike capital costs, the benefits of the railway are dependent on the demand, thus, vary from year to year. The benefits of the Okanagan Railway (except road safety) for a set of sample years are shown in Figure  5-1, inflated to the 2007 US dollars using the historic inflation rate. The benefits of railway were calculated using the methodology described in above section. -$1,500-$500$500$1,500$2,500$3,500$4,500$5,5002020 2025 2030 2035 2040 2045 2050 2055 2060Annual Worth of externalities  (in thousand 2007 USDollar)-$10,000$0$10,000$20,000$30,000$40,000Annual Worth of rail revenues and user benefits  (in thousand 2007 US Dollar)CongestionReductionNoiseReductionPollutionReductionHighwayInfrastructureSavingsUser TimeSavingsFareUser O&MCost SavingsFigure  5-1 Benefits of the railway 5.3 Safety benefits 5.3.1 Literature review To estimate the safety benefits of the Okanagan Railway, the drop in the number of collisions on Highway 97 due to the diversion of trips from the highway to the railway, were estimated. Traditionally, a safety evaluation was conducted using a linear rate between the number of collisions and vehicle-kilometres-travelled (or vehicle-miles-travelled), calculated as follows:  - 125 - nnVKTTCCR610u           ( 5-2) Where  CR is collision rate; TCnis total number of collisions in year n; and VKTn is vehicle-kilometres-travelled in year n. Subsequently, the drop in the number of collisions due to the drop in VKT (or VMT) was forecast using a developed rate given in the literature (Levinson et al., 1996b, Forkenbrock, 1999, Zhang et al., 2005, Highway Agency, 1998, Jack Faucett Associates, 1991)However, this methodology usually overestimates the number of collisions in high VKTs and underestimates the number of collisions in low VKTs. The literature provides ample research demonstrating that collisions are not linearly related to exposure (Lovegrove, 2007), with a typical association shown in Figure  5-2. Consequently, the collision rate cannot be assumed to be constant with respect to traffic volume (or VKT), so, to be conservative in this research, a linear-assumption rate was not be used to estimate number of reduced collisions and safety benefits (Qin et al., 2004). 0102030405060- 50,000 100,000 150,000 200,000 250,000 300,000 350,000Vehicle-Kilometres-TravelledCollisionsObserved collisions non-linear model Collision rateFigure  5-2 Collisions vs VKT Data sources: the Insurance Corporation of British Columbia (ICBC) - 126 - Instead, a macro-level collision prediction model (CPM) for the intercity part of Highway 97 was developed following the methodology described by Lovegrove (2007). It recommends using generalized linear regression modeling (GLIM) techniques, and assuming a negative binomial error structure (Lovegrove, 2007). Using this CPM, the number of collisions on Highway 97 for two different scenarios – (a) doing nothing, or (b) adding the Okanagan Railway to the transportation infrastructure of the valley – has been calculated. The savings from reducing the number of collisions (transferred commuter/tourist trips and freight truck trips) would be considered as a safety benefit of the Okanagan Railway. 5.3.2 Methodology for safety benefit evaluation 5.3.2.1 Collision Prediction Model (CPM) development Model form and fitness tests The general model form used is shown in Equation 5-4. Only exposure variables meeting the ‘zero-exposure = zero collision frequency’ principle were used as an external exposure parameter.   � /iiXbaeZaE1)($        ( 5-3) In this equation, E(/) is the predicted collision frequency (over 6 years); ao, a1, and bi are model parameters; Z is the external exposure variable (i.e. VKT); and  Xi is the explanatory variables (Lovegrove, 2007).Goodness of fit follows the methodology in Lovegrove, 2007. The 2FPearson statistic, the Scaled Deviance statistical measures, t statistic (equal to the parameter estimate divided by its standard error and equivalent to the Wald statistic) and N (shape parameter) were used to assess the model’s goodness-of-fit. The 2FPearson statistic is defined as: - 127 - � / niiiyEyPearson122)var())((F         ( 5-4) whereiy  is the observed number of collisions in zone i; )(iE /  is the predicted number of collisions for the zone as obtained from the collision prediction model; and )var(iy  is the variance of the observed collisions.  The Scaled Deviance (SD) is defined as the likelihood ratio test-statistic, which measures twice the difference between the maximized log-likelihoods of the studied model and the full or saturated model. The full model is the one with as many parameters as there are observations, which fits the data perfectly but becomes impractical as a forecasting model. Therefore, the full model, which possesses the maximum logarithm-likelihood achievable under the given data, provides a baseline for assessing the goodness-of-fit of an intermediate model with p parameters. In our case that error structure follows the negative binomial distribution, SD is as follows: })(ln)()()({21NNN � iiiniiiiyEyyyEyySD        ( 5-) The SD, and the 2FPearson  statistics should be less than the2F  distribution value with (n - p - 1) degrees of freedom at a 95% confidence level. The parameter estimate t statistic for all variables except the constant (suggesting that the constant term leading the CPM may not significantly differ from 1.0) should be significant at the 95% confidence level (>1.96). N (shape parameter) should exceed 1.0. VariablesPrevious studies on Collision Prediction Models (CPMs) for highways have demonstrated that highway collisions are less dependent on socio-demo-graphic, transportation demand management, and network variables, and are mainly dependant on exposure variables.  Exposure variables include: annual average daily traffic (AADT), segment length, speed, geometry, lane width, shoulder width, and pavement width (Qin et al., 2004).  However, - 128 - due to the size of the study area (176 km between Vernon and Osoyoos), and lack of data, macro (not micro) level collision prediction models were pursued in this research. Highway geometry variables were excluded in the absence of adequate quality data. Moreover, two new variables were added to the set of potential variables. The first variable was truck percentage (truck AADT/auto AADT), to capture any possible effect of trucks influencing the safety of the highway. The second variable was population density around the studied highway segments. This variable was included to reflect that some intercity parts of Highway 97 were significantly denser in population than others, as shown in Figure  5-3.  Figure  5-3 Population density (per square kilometre) in Okanagan Valley - 129 - As vehicle-kilometres-travelled (VKT) has been found to be better fitted than AADT in highway CPMs, whether macro or micro (Qin et al., 2004), VKT replaced average annual daily traffic (AADT) and segment length. To calculate VKT between points i and j in the highway, the following equation was used:ijjiijSLAADTAADTAvgVKT u ),( (Qin et al., 2004)     ( 5-6)whereiAADT is the annual average daily traffic at the point i, andijSL is the length of the highway segment between point i and j.5.3.2.2 Safety benefits calculation After developing collision prediction models for highway 97, the number of escaped collisions due to the reduction of trips in Highway 97 can be calculated. In next step, monetary value of the escaped collisions was calculated using the ICBC insurance method and the distribution of collisions (fatal, injury, property-damage-only) in Highway 97.While there are many fewer concerns about railway safety than highway safety, rail collisions are still inevitable and occasionally collisions at highway/railway crossings occur. Therefore, the value of safety for a railway was reviewed in the literature. The literature suggested that the collision cost for a passenger moved one kilometre by passenger rail was 6.3% of the cost of a passenger moved one kilometre by auto (Levinson et al., 1996a). Moreover, it was found that the collision cost for a ton of freight shipped one kilometre by freight rails is 15.5% of the cost of a ton of freight shipped one kilometre by autos (Gleason et al., 2005, Transport Concepts, 1994).In the last section, the safety concerns of the Okanagan Railway were taken into account and net safety benefits of the railway were calculated by subtracting railway safety costs (due to crossings) from railway safety benefits (due to trip reduction in Highway 97). - 130 - 5.3.3 Model developmentAlong Highway 97, there were several permanent and temporary traffic counting locations, as shown in Figure  5-2. Using these locations, Highway 97, between Vernon and the border (corridor of the railway), was divided into 12 segments.  Only highway segments outside the boundaries of towns were considered, as the Okanagan Railway would impact mainly the number of intercity trips, with the exception possibly of Kelowna, where two stations were expected - at Kelowna Airport, and in Downtown Kelowna. To calculate the number of reduced collisions due to the reduction of trips in Highway 97 inside the City of Kelowna, previously developed CPMs for Kelowna were used (Khondaker et al., 2009). Figure  5-4 Permanent and temporary traffic counting locations in Highway 97 - 131 - Table 5.4 and 5.5 show the attributes of the selected segments of Highway 97. Note that the segments in this study are longer (on average 9 km) than the segments in Qin (on average 1 km). Therefore, Equation 5-3 cannot be reliably applied, and future research is needed in this area.  Note that VKTs in Table  5-4 are driven from the observed, 24-hour AADT, unlike other CPM research (Lovegrove, 2007), which rely on morning peak period, modelled VKT. Zegeer (1982) recommended using an observation time period of between one and three years for collisions, to minimize the effects of random fluctuations but still remain sensitive to the changes over time. However, an observation period of 6 years (2000-2006) was chosen for this study, due to the low number of annual collisions on Highway 97. Figure 5-5 shows that number of collisions on Highway 97 in the years between 2000 and 2006, where no drastic change is observed.Table  5-4 Attributes of the twelve segments    Length (Km) AADT VKT Posted Speed (kph) 1 Border-Osoyoos 3 1,771 5,313 80 2 Osoyoos-Oliver 16 8,916 142,661 50 3 Oliver-Okanagan falls 19.5 8,123 158,402 70 4 Okanagan falls-3A junction 5 13,150 65,750 90 5 3Ajunction-Penticton airport 8.5 12,916 109,782 75 6 Penticton-Summerland 9.5 18,501 175,757 100 7 Summerland-Peachland 14.5 12,016 174,225 90 8 Peachland-97C 2.5 14,585 36,463 90 9 97C-Westbank 2.5 23,723 59,308 90 10 Kelowna airport-Winfield 5.5 20783 114,307 75 11 Winfield-Oyama 6 20,992 125,954 80 12 Oyama-Vernon 15 23,045 345,677 90  Min 2.5 1,771 5,313 50  Max 19.5 23,723 345,677 100  Average 9.0 14,877 126,133 82 - 132 - Table  5-5 Other attributes of the twelve segments   Truck Percentage (Trp) Population density (Popd) Total 6 Yrs collisions Severe 6 Yrs collision 1 Border-Osoyoos 48.00% 3.89 6 6 2 Osoyoos-Oliver 15.80% 8.63 48 48 3 Oliver-Okanagan falls 17.34% 8.13 46 46 4 Okanagan falls-3A junction 10.71% 6.22 1 1 5 3Ajunction-Penticton airport 10.91% 6.21 18 18 6 Penticton-Summerland 12.03% 23.29 19 19 7 Summerland-Peachland 17.33% 33.59 11 11 8 Peachland-97C 14.54% 290.99 2 2 9 97C-Westbank 8.94% 50.35 25 25 10 Kelowna airport-Winfield 12.72% 442.14 26 26 11 Winfield-Oyama 11.66% 75.86 44 44 12 Oyama-Vernon 10.62% 14.5 10 10  Min 8.94% 3.89 1 1  Max 48.00% 442.14 48 48  Average 15.88% 80.32 21 21 0200400600800100012002000 2001 2002 2003 2004 2005 2006 2007Observed collisions in HWY 97Figure  5-5 Number of collisions in Highway 97 - 133 - Using the above data, a set of macro CPMs were developed to predict the total number of collisions as well as a set to predict the number of severe collisions. Only three models met the goodness of fit criteria which are shown in Table  5-6  Table  5-6 Developed macro CPMs  2FPerason SD2Ft-StatisticsTotal Collisions149.0071.06/ VKTyrsCollisions  1.67 8.26 13.1 18.31 const  -0.8 VKT  2.02 2SpdeVKTyrsCollisions032.042.002.26/ 1.71 10.41 13.36 16.92 const -0.4 VKT  2.03 Spd -2.15Severe Collisions 347.0088.06/ VKTyrsCollisions  2.18 12.33 12.16 18.31 const  -2.15 VKT  2.71 To validate the accuracy of the developed models, they have been employed to predict the number of collisions for two segments of Highway 97 north of Vernon (see Figure  5-6, where V1 = validation segment 1, and V2 = validation segment 2). These segments are excluded from the model development step because they are outside of the study area. Table  5-8 presents the attributes of these two segments and the observed number of collisions there. Table  5-8 lists the number of collisions for these two segments predicted by developed Models, and also the regional Collision Rate. This table also shows the error of each model in predicting the collisions. It seems that the best model is Model 2, which predicts the number of collisions in Segment V1 with 27% error, and in Segment V2 with 7% error. Observing some error in the estimation by the models was predictable since they are developed with only 10 degrees of freedom. However, they are much more accurate than the linear models which traditionally were used in collision cost estimation. As Table  5-8 shows collision rate predicted number of V1’s collisions with 68% and number of V2’s collisions with 61% error (see Figure  5-7). - 134 - Figure  5-6 Validation segments Table  5-7 Attributes of the validation segments   Length  (km) AADT VKT Posted  Speed  Observed 6 Yrs  Collision Observed 6 Yrs  severe Collision V1 Falkland-Monte Creek 45 4,638 208,710 80 21 11 V2 Vernon-Falkland 37.5 8,881 333,059 75 35 17 - 135 - Table  5-8 Predicted collisions by different models Predicted  6 Yrs Collision by Model 1 Predicted 6 Yrs Collision by Model 2 Predicted 6 Yrs  Severe Collision by Model 9 Predicted 6 Yrs  Collision by Regional Collision Rate V1 Falkland-Monte Creek 29 (39.1%) 27 (27.5%) 13 (14.2%) 35 (68.1%) V2 Vernon-Falkland 37 (5.0%) 33 (6.6%) 19 (10.4%) 56 (60.9%) Therefore, the developed CPMs are selected to be utilized in the next section to estimate the cost of the safety benefits of the Okanagan Railway. To increase the accuracy of the models, it has been suggested that future studies divide each one of those 12 segments into several smaller segments by collecting AADT at more points in the highway. 1020304050600 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000VKTCollision 6/yrs Observed Predicted by Model 1Predicted by Model 2 Predicted by Regional Collision RateFigure  5-7 6 Yrs collisions prediction for V1 and V2 5.3.4 Safety benefits calculation 5.3.4.1 Application of developed CPMs In order to estimate the number of reduced collisions on Highway 97, the transferred VKT was calculated.  It is the number of transferred trips - calculated in Chapters 2 and 3 - multiplied by the length of the trip in Highway 97 between Osoyoos and Vernon. The number of reduced collisions has been shown in Table  5-10. - 136 - In order to estimate the monetary value of the safety, the average cost of a collision was estimated. Table  5-9 lists two estimates of the cost per road collision type. Based on the number of collisions in the area, an average cost per collision was determined for each method. To be conservative, an average cost calculated by the ICBC Claims method was chosen. Safety savings are shown in Table  5-9. Table  5-9 Road Collision Costs in Highway 97, British Columbia  Distribution of collisions in Hwy 97 Insurance (ICBC) claims Human capital cost Willingness to Pay Fatal Collision 1.20% $241,000 $1,000,000 10,000,000 Injury Collision 40.60% $26,000 $35,000 100,000 Property Damage Only 58.20% $3,000 $5,000 6,000 Average cost per collision $15,000 $29,000 94,000 Table  5-10 Safety benefits of Okanagan Railway (Except savings in Kelowna)   Transferred truck VKT Transferred auto VKT Total transferred  VKTTransferred collisions  Savings (000$) 2009 7,414,136 31,648,683 39,062,819 3.19 45 2019 9,869,262 39,361,200 49,230,462 3.51 50 2029 12,324,388 49,368,611 61,692,999 3.86 55 2039 14,779,515 72,503,313 87,282,828 4.47 64 2049 17,234,641 114,946,472 132,181,113 5.32 75 2059 19,689,767 255,132,343 274,822,111 7.24 102 2069 22,144,894 597,656,203 619,801,097 10.19 142 2079 24,600,020 1,320,177,288 1,344,777,308 14.12 196 5.3.4.2 Safety benefits for the city of Kelowna As the railway would also help divert some of the traffic from Highway 97 between the airport and downtown Kelowna, the previously developed community-based, macro-level collision prediction models for the City of Kelowna were used.  This was done instead using the models developed in this study, because this part of Highway 97 is in the City - 137 - of Kelowna and it was predicted to have characteristics different than the intercity parts of the highway. Khondaker et al have developed 16 rural CPMs and 14 urban CPMs for the city of Kelowna, which relate different characteristics of the TAZ, including Exposure, Socio-Demographic, Transportation demand management, and Network Variables, to the number of collisions in each TAZ. Since most of Highway 97 lies in the rural zones, it was decided to use the rural CPMs. Moreover, the simplest rural model, which only includes VKT and Volume to Capacity ratio (VC) was selected (see Equation 5-7) as it helped the results of this part be consistent with the results for the inter-city parts of the highway.evcVKTyrsCollisionsTotal868.26478.043.03/       ( 5-7) The Volume to Capacity ratio and AADT change from point to point in Highway 97 between the airport and downtown as there are many arterial roads connecting to Highway 97. Therefore, Highway 97 is dived into three segments, including: 1) Airport-Sexsmith, 2) Sexsmith -HWY 33, and, 3) HWY 33-Okanagan Lake Bridge, as shown in Figure  5-4. Subsequently, transferred VKT, total VKT and Volume to Capacity were calculated for each segment. Finally, the number of reduced collisions in each segment was calculated by Khondaker’s CPM (Equation 5-7).  A new average cost of collision was calculated for the City of Kelowna using the same ICBC claims costs and collision severity distribution in Highway 97 in Kelowna as shown in Table  5-12. Table  5-11 Road collision costs in Highway 97, Kelowna  Distribution of collisions in Highway 97 Insurance (ICBC) claims Fatal Collision 0% $241,000 Injury Collision 45.20% $26,000 Property Damage Only 54.80% $3,000 Average cost per collision $15,000 - 138 - The number of reduced collisions and the safety benefits of the Okanagan Railway for Highway 97 in Kelowna have been shown in Table  5-12. Table  5-12 Safety benefits of Okanagan Railway in Kelowna  Airport-Sexsmith intersectionSexsmith -HWY 33 intersectionHWY 33-OK Lake BridgeTotal YrTransferred VKTTransferred collisions Transferred VKTTransferred collisions Transferred VKTTransferred collisions Transferred collisions Savings(000$) 2009 1,398 9 1,069 7 1,535 8 24 325 2019 1,740 11 1,331 8 1,911 9 28 375 2029 2,183 12 1,669 9 2,397 11 32 434 2039 3,192 16 2,441 12 3,505 14 41 555 2049 5,032 21 3,848 16 5,526 19 56 746 2069 11,661 37 8,917 27 12,804 32 96 1,285 2079 26,877 63 20,553 46 29,512 55 165 2,207 5.3.4.3 ResultsBased on analysis using CPMs, 94% of the safety benefits of the railway come from the diversion of autos from the highway, while 6% come from the diversion of trucks from the highway. Therefore, the net safety benefits of a railway were calculated as shown in Equation  5-8. SBSBNetSBuuuu )15.01(06.0)063.01(94.0      ( 5-8) In this equation, SBNet  is the Okanagan rail safety benefits minus Okanagan rail safety costs and SB is the safety saving for Highway 97 (a summation of the last columns of Table  5-10 and Table  5-12). Figure  5-8 shows the net safety benefits of the Okanagan Railway between 2007 and 2100 in 2007 dollars calculated based on three conventional methods mentioned in Table 5-9. However, Insurance claim results were used for cost benefit analysis.- 139 - -5,000,00010,000,00015,000,00020,000,00025,000,00030,000,00035,000,0002000 2020 2040 2060 2080 2100 2120Net safety benefits of Okanagan Railway in 2007$Human Capital Cost Willingness to pay Insurance ClaimsFigure  5-8 Safety savings of Okanagan Railway (Thousands of 2005$) Note that while trip numbers on highways may decrease, some trips would be generated between origins of the trips and railway stations. However, calculating those trips is not possible by the data available at this step, and would be the subject of more detailed studies in the future. Using the more conservative ICBC claims costs is expected to mitigate any collision over predictions due to these unaccounted trips. 5.4 SummaryIn this chapter, the benefits of the Okanagan Railway were estimated. These benefits included revenue of the railway as well as externalities and user benefits. Moreover, one of the stated objectives of this study – namely, developing collision prediction models for Highway 97 – was achieved. Three models were successfully developed following recommended methodology, with GLIM assuming a negative binomial error structure, and were used to calculate the safety benefits of the Okanagan Railway. In the next chapter, the economic feasibility of the proposed railway would be determined using the calculated costs and benefits. - 140 - Table  5-13 Break down of Okanagan Railway benefits in 2007 $ Yr Fare Congestion Noise Pollution Safety User Time User O&M Cost Hwy Infr. Total 2019 $12,866 $2,398 $303 $688 $268 -$2,381 $28,451 $2,343 $44,935 2029 $12,596 $2,452 $291 $699 $292 -$2,206 $28,977 $2,350 $45,451 2039 $15,554 $2,551 $382 $745 $406 -$3,295 $30,660 $2,645 $49,647 2049 $17,817 $2,565 $453 $764 $530 -$4,200 $31,296 $2,839 $52,064 2059 $24,747 $2,729 $668 $854 $787 -$6,829 $34,531 $3,497 $60,984 2069 $41,834 $3,276 $1,189 $1,112 $1,318 -$13,142 $44,031 $5,196 $84,815 - 141 - 6 EVALUATION OF PROPOSED RAILWAY 6.1 Introduction This chapter consists of two main sections. In section 6.2, the methodology for conducting a benefit cost analysis using the treasury benefit cost analysis guide is discussed. In section 6.2, the present net worth and benefit cost ratio for the Okanagan Railway are calculated. In section 6.3, a Monte Carlo simulation is conducted to analyze the sensitivity of the results of the last section.  6.2 Literature review and methodology After identifying and quantifying the benefits and the costs of the proposed railway, the economic feasibility of the project was evaluated. The comparison was done in the form of a present worth (PW) analysis and a benefit/cost (B/C) ratio. The present worth analysis uses the difference between the project costs and benefits to determine the possible economic gain that could be obtained within the region. Conversely, the benefit/cost ratio compares the project benefits to its costs, to determine a factor of return for the region.  The Benefit-Cost Analysis Guide published by the Treasury Board of Canada was used as the main guide throughout the analysis (Treasury Board of Canada Secretariat, 1998). The guide suggests that all the costs and benefits across all periods should be tabulated initially in nominal dollars. The Guide suggested that, all figures, or at least the net cash-flow line, should be converted to constant dollars before proceeding with the calculations of final present value or benefit-cost ratio. Treasury Board has required that benefit-cost analysts use a social discount rate of 10 per cent ‘real’ per annum, with 8% and 12% for the risk analysis (Treasury Board of Canada Secretariat, 1998). However, most Canadian - 142 - economists agree that a 5 percent discount rate is closer to society’s true rate of time-preference for the rail projects (HDR et al., 2005). Most of the recent Canadian rail projects such as Canada line and Evergreen Line applied a 6% social discount to calculate net present worth (Translink, 2008, Canada Line Rapid Transit Inc, 2006), while 3% to 10% social discount rate is used for cost benefit analysis of the rail projects around the world (Bourn, 2006, Transportation Economics & Management Systems, Inc., 2007, Acharya et al., 2006, Abelson, 1995, de Rus et al., 1997). In this study, a 6% real discount rate is employed. Benefit-Cost Analysis Guide suggests using simulation as the only practical approach to financial and economic risk analysis (Treasury Board of Canada Secretariat, 1998). Simulation predicts the possible results of the benefit-cost model, given the variables that influence those outcomes. This process is essential for this study, given the large number of assumptions. Most risk-analysis programs use the Monte Carlo method which is a simple random sampling according to a specified probability distribution. The advantage of the Monte Carlo over other methods of sensitivity analysis is that it makes it possible to analyze the combined impact of multiple sources of uncertainty, while other methods only allow analyzing one source of uncertainty at a time. Therefore, in the last step a Monte Carlo simulation was programmed to develop an overall picture by incorporating all the risks. 6.3 Social cost benefit analysis In this study, all the costs and benefits are calculated in the 2007 nominal US dollars. A sample of the calculations is shown in Table  6-1. Refer to Appendix D for the detailed calculations. In next step, the total cost and the total benefit and the net cash flow were converted to the real 2007 US$. A historic inflation rate of 2% was used for the conversion. A 6% real discount rate is applied to the real costs and benefits of the Okanagan railway while discount rates of 3% to 12% are used for the risk analysis. Table  6-2 shows a sample of the discounted costs, benefits and the net present value over time for years between 2009 - 143 - and 2079 in real 2007 dollars. The detailed calculations are presented in Appendix D. Note that benefit-cost analysis does not use accruals, depreciation allowances, or other ‘non-cash’ items.  Table  6-1 Costs and benefits of the project in nominal 2007 dollars    YrCapital  costRolling  stock O & M  costSafety  benefitsExternalbenefits Revenue Total costs Total benefits2009-198,756,877     -198,756,877 02019 -200,000 -15,436,868 361,584 39,736,449 16,317,089 -15,636,868 56,415,1222029 -200,000 -18,086,740 393,662 49,581,398 19,472,776 -18,286,740 69,447,8362039 -200,000 -29,091,735 548,603 62,627,333 29,312,562 -29,291,735 92,488,4972049 -200,000 -42,318,044 715,800 76,524,289 40,929,398 -42,518,044 118,169,4862059 -3,200,000 -76,485,825 1,062,579 98,444,266 69,299,413 -79,685,825 168,806,2582069 -200,000-167,072,083 1,779,934 141,979,088 142,804,209 -167,272,083 286,563,2312079 -3,200,000-348,490,636 2,917,168 220,313,252 288,974,841 -351,690,636 512,205,261Table  6-2 Net benefits, net costs and present worth of the project in real 2007 dollars Year Benefit Real 2007$ Cost Real 2007$ Present Worth  Real 2007 US$ 2009 0 -151,869,344 -151,869,3442019 11,227,504 -3,111,985 8,115,520 2029 3,599,865 -947,903 2,651,962 2039 1,248,691 -395,469 853,222 2049 415,540 -149,513 266,026 2059 153,983 -72,984 80,999 2069 68,361 -39,903 28,457 2079 31,825 -21,852 9,973 2089 12,119 -8,780 3,339 The net present worth (NPW) and benefit-cost ratio (B/C) of the railway are calculated for seven different scenarios, including: a: The Okanagan rail starts working in 2009;  b: The Okanagan rail starts working in 2019;- 144 - c: The Okanagan rail starts working in 2029;  d: The Okanagan rail starts working in 2039; e: The Okanagan rail starts working in 2049; f: Commuter rail between Kelowna and Vernon  starts working in 2019; and g: Commuter rail between Kelowna and Vernon  starts working in 2049. The results of the analyses are listed in Table  6-3. Note that a life cycle of 50 years and a salvage value of 600,000 dollars per mile (the current rate for Canadian Pacific purchases) are considered. Table  6-3 NPW and B/C for all scenarios  Scenario NPWreal million 2007$ B/Ca -369.87 0.58b -128.17 0.71c -33.09 0.86d 0.26 1.00e 10.80 1.13f -67.89 0.55g -3.99 0.95Values for the net present worth and the benefit cost ratio for scenarios a, b, c are below the acceptable range. It means that considering all the assumptions and calculations made in this study, the Okanagan railway is not economically feasible in the next 20 years. However, the benefits and the costs of scenario d are equal, suggesting that in 2040 the Okanagan railway may be an economically feasible project. The results for scenario econfirm this conclusion.  Note that the results for scenario f and g are below the acceptable range. It means a commuter rail between Kelowna and Vernon is not profitable for the next 40 years.  These results indicate that a commuter rail should be accompanied with a freight rail (connecting to the border) for a feasible project.  - 145 - 6.4 Sensitivity analysis Deterministic methods of benefit cost analysis (such as the above calculations) offer a single figure for NPW or B/C, but it is always unclear what the probability of this single outcome is. As mentioned in the methodology section, a Monte Carlo simulation was conducted to analyze the combined impact of multiple sources of uncertainty. The following steps were taken in developing a Monte Carlo simulation model for this study. First, all the uncertain variables that affect the net present worth of the project were determined. Then, a probability distribution is established for each variable. Table  6-4 shows the selected variables and the probability distribution for each variable.  Table  6-4 Random variables and outcomes                          Probability Variable25% 25% 50% Freight demand Assignment#1 Assignment#2 Assignment#3 Fuel price SAUNER scenario LOPEXscenarioAveragescenarioTrain operating speed (kph) 40 80 60 O&M cost of commuter rail 125% of the original value 75% of the original value original value O&M cost of freight rail 125% of the original value 75% of the original value original value Social discount rate 3% 10% 6% Track construction costs maximum Estimate minimum estimate averageContingency and design 28% of total cost 45% of total cost 50% of total cost External benefit 125% of the original value 75% of the original value original value In the next step, a code was written to generate a random number and then assign a new value to each variable based on the probability distribution of that variable. Finally, new - 146 - NPW and B/C were calculated using the new values for the above variables. The same procedure was going to be repeated several times. That is, a new fuel price and train speed were chosen, and a new passenger demand was forecasted. A new method of rail assignment was selected; and, a new freight demand was forecasted. A new track construction, O&M cost and contingency, and design percentage was chosen, resulting in new capital cost estimates. A new coefficient for external benefits was assigned; hence a new value for the benefits from externalities was calculated. Using all the new estimates, a new present worth was calculated.  Monte Carlo simulation results are shown in Table  6-5.   Table  6-5 Monte Carlo Simulation Results Scenario Avg NPW real million 2007$ Probability of B/C>1 a -265 15% b -49 26% c 28 43% d 50 60% e 53 72% f -60 4% g -3 37% The results confirmed that scenarios a, b, f and g are deemed to fail, scenario c meets minimum success. Scenario d is considered reasonably successful and scenario e is the most successful scenario with more than 70% chance of having benefits greater than costs. Refer to Appendix E, for detailed results.  In addition to Monte Carlo simulation, break-even analysis was also conducted to identify those variables that have the most impact on the present worth of the project. Figure  6-1 shows the break-even points for capital cost, O&M cost, revenue, fuel price, speed of the railway and social discount rate.  - 147 - $(20.00)$(10.00)$-$10.00$20.00-50% -30% -10% 10% 30% 50%Percantage variation from base valueNet Present Worth in Million 2007 US DollarsCapital CostDiscount RateO&M CostsExternalities & FaresSpeedFuel PriceFigure  6-1 Break even chart  6.5 SummaryThe main object in this research was to evaluate the feasibility of a new electric passenger and freight railway in the Okanagan Valley. In this chapter, net present worths and benefit cost ratios of the project were estimated for seven scenarios, including five scenarios for a passenger and freight railway running between Osoyoos and Vernon in different time periods and two scenarios for a commuter railway between Kelowna and Vernon in different time periods. While the benefit cost ratios for the passenger and freight Okanagan railway were < 1 for all scenarios that assumed launches before 2030, they were > 1 for all scenarios assuming launches after 2040. These results suggest that a passenger and freight rail between Osoyoos and Vernon running after 2040 would be an economically feasible project. It would appear that without the added revenues from freight rail, a Kelowna-Vernon commuter rail project would not be profitable even after 2050. A Monte Carlo simulation confirmed the validity of these results. - 148 - 7 CONCLUSIONS AND RECOMMENDATIONS  7.1 Introduction This chapter is comprised of three main sections. In Section 7.2, the thesis summary is presented along with the main research conclusions. Section 7.3 highlights the research contributions by justifying how they add to the current knowledge of the field of sustainable road safety and promotion of alternative mode of transportation. Finally, Section 7.4 describes some recommendations for future research topics, which can enhance and strengthen the methodologies and the results of this thesis. 7.2 Summary & conclusions Between 1900 and 1950, rail transportation was a widely used mode of transportation in the Okanagan Valley. However, after 1950, the construction of new highways and the rapid development of the auto industry made it impossible to keep the railways running. Today, the ever-growing traffic congestion on the major highways and the environmental issues associated with this congestion appear to signal the beginning of a new era for rail transportation. Moreover, the seven-year “Building Canada” Plan suggested investments in short-line railways are one possible way to mitigate increasing congestion in existing infrastructure. Furthermore, railways can also provide regions with sustainable road safety by diverting trips from highways.  This research examined the feasibility of laying down a new railway linking Vernon to Osoyoos in the Okanagan Valley by exploring the economic, environmental and safety benefits of such a railway. To address the existing issues, the Okanagan Railway would serve both as a freight railway and a commuter one. Local, domestic and international demands for the Okanagan freight rail were forecasted by an O-D factoring method. An - 149 - O-D table was produced using the observed and survey data, rail shares for each commodity was estimated using the existing shares, and the estimates were assigned to the rail network using three methods of assignment. Demand for the Okanagan passenger was forecasted following the traditional four-step models. Available data from the regional transportation model made it possible to escape the trip generation and distribution steps.  A new mode choice model was developed for the region due to the absence of a local mode choice model. The alignment for the railway was evaluated considering several socio-economic, environmental and geologic factors. In ArcGIS 9.1, these factors were combined by using the AHP extension and the optimum path was selected using GIS spatial tools. The costs of the project including capital, operating and maintenance and time loss, and the benefits of the project including revenue, environmental and safety benefits were evaluated. To estimate the safety benefits of the railway, a macro collision prediction model was developed for Highway 97 which estimates collision number in the highway as a function of Vehicle-kilometres and other highway attributes. Finally, an economic evaluation of benefits and costs of the railway was conducted by calculating the net present worth and the benefit cost ratio of the project for different scenarios. A Monte Carlo simulation was employed to analyze the sensitivity of the results.  The study concluded that the construction of a new commuter and freight railway is not economically feasible in the next 30 years. However, the project would be considered valuable for the years after 2039 or in the exceptional conditions such as peak oil when the price of oil is very high (400 US $ per barrel).  7.3 Research contributions The following six items represent the main contributions of this research: x Evaluation of feasibility of short line railways in Canada in the next 40 years: case study of the Okanagan Valley - 150 - Revitalizing old short line railways is an economically feasible solution for the congestion and environmental issues of road transportation in the less populated areas of Canada, like the Okanagan Valley after 2039. In this research study, a comprehensive review on the demand forecast models; planning and design issues; and costs and benefits of such short-line railways was conducted. The review was followed by a demand forecast, an alignment selection, and cost benefit estimation for the case study of the Okanagan Railway which enabled the study to evaluate the feasibility of the railway. It was determined that the Benefit-cost ratio of such a project would be greater than one after 2039.x Introducing a reliable tool for calculating safety benefits of transportation projectsIt is suggested that the collision rate method that is currently used to determine the safety costs or benefits of a project is not a reliable method. Collision rate tends to overestimate the number of collisions in high VKTs and underestimate the number of collisions in low VKTs. This research study introduced macro Collision Prediction Models as a dependable substitution. They can predict the number of collisions with less than half of the error of the collisions rates.   x Evaluation of impacts of a new mode of transportation in sustainable safety The safety benefits of the Okanagan Railway were then estimated using the developed CPMs as a successful case study. The study found that if the Okanagan Railway was running in the year 2008, it could avoid 27 collisions on Highway 97. Considering the enormous capital costs of commuter and freight rail, the safety benefits do not seem to be appealing enough. It is concluded that rail transportation cannot be assumed an alternative for road transportation to achieve Sustainable Road Safety. x Developing the first mode choice model for the Okanagan region Given the fact that autos (for passengers) and trucks (for goods) have been the only possible modes of transportation between the cities of the Okanagan Valley, it was not surprising that the current Okanagan Valley transportation model does not include a - 151 - mode choice step. In this study, the first mode choice model for the region was developed. This is due to the need in the passenger demand forecasting section. This model was adopted from the FSUTMS and OKI (Ohio-Kentucky-Indiana) models, while other accessible mode choice models such as Translink model, Oregon model, and California high speed rail model were reviewed. Finally, the mode choice model was calibrated using the recent trip survey in the North and Central Okanagan regional districts.x Developing a new procedure for rail assignment Rail assignment techniques are among the forgotten problems of freight demand forecast due to their complexity. They should consider practical rules of the rail industry in addition to the cost functions that are used in truck or auto assignments. In this study, a comprehensive review was conducted on the rail freight demand forecast including assignment techniques. A new assignment procedure was developed by combining the available techniques, mathematical choice models, and practical rules of the Canadian rail transportation industry.  x Implementing GIS as a useful tool in transportation planningGIS provided the opportunity of integrating data regarding different sources such as land-use, soil, water, and vegetation, to assist in the evaluation process of alignment. It offers several Multi Criteria Evaluation (MCE) operators that are suitable for the planning problems. In this study, GIS was proven to be a useful tool in mapping and ranking the planning criteria and selection of the alignment that minimizes the construction costs and environmental impacts.7.4 Recommendations for future research The results of this research study suggest that the Okanagan Railway warrants further research. This study was only a first step towards identifying knowledge gaps that must be addressed prior to any possible North American railway renaissance. Therefore, this section presents some of the recommendations for the future researches to enhance and - 152 - strengthen the methodologies described in this thesis, as well as to address various issues raised. These are as follows: x Accurate demand forecast is a basic need for an accurate feasibility study. Considering the growing population of the Okanagan Valley, it seems necessary to modify and update the existing travel demand forecast model: 1997 OVTP model. While this study includes some early modifications of the model trip generation and introducing a mode choice model, there is a need to conduct a comprehensive regional travel survey and update the traffic generation and distribution sections. Five case studies that were reviewed in Chapter 2 can be reliable references, since they included the most successful models in North America.x Developing an exclusive mode choice model for the Okanagan must be the subject of future studies. It is recommended that the developed Okanagan mode choice model be employed as a base. However, the model modification should include calibration of all the coefficients. Given the growth of the transit system in the region, this mode choice model can be used to evaluate the transit projects that may be proposed. x There is room for further modification of station selection. In this study, starting points and end points of each segment were selected as the stations due to the high volume of demand from/to them. While other stations in the locations like Oliver and Oyama may seem necessary, no finer data was available to determine the potential demand for these locations. Using the modified passenger demand forecast model, a detail study on the stations area can be the subject of future studies. x After reviewing the models developed for long distance freight demand forecast in North America, it was found that the studies on mode choice models are outdated and inadequate.  Most of the current demand forecast models have not developed any mode choice model and have employed fixed mode share based on the historical data. However, those historical shares can change given a high fuel price or environmental issues surrounding truck transportation. Such factors could work as game changers in - 153 - favour of railways. This study suggests that developing new sets of mode choice models be included among the priorities of the future studies. x The other gap in the freight modeling knowledge is in the rail assignment section. As it was discussed earlier, very few studies have addressed the rail assignment issues and proposed assignment techniques. This study suggests that future researchers need to focus on rail assignment techniques from a mathematical perspective as well as from a practical point of view.   x A comprehensive freight demand model is a requirement for any accurate demand forecast for existing and new infrastructure. While several country-wide and state-wide models are found among the American literature, there is no freight demand model for Canada or British Columbia. Canada should start developing a country-wide transport demand forecast model which includes both passenger and freight demands. This model can be used in any major transportation infrastructure project. However, on a smaller scale, such a model is totally necessary in the Okanagan Valley as freight export is one of the major businesses in the valley. Furthermore, improvement of the Osoyoos custom ports can bring new jobs and prosperity to the region. Hence, developing a comprehensive freight model for the Valley can be the subject of the future researches. x Another issue in freight modeling is the lack of data on freight movements. The latest and most comprehensive data on trucking is available from 1999 national roadside survey (NRS). While Ontario and Quebec have expressed interest in a future NRS in 2003 and conducted a new NRS in 2006, British Columbia has not participated in any national or provincial survey since 1999. As a result, new surveys in the area are highly recommended. The results can be used to develop the regional freight models that have been discussed above. x The lack of data on rail transportation is even more severe. No data sources can be found in the literature on the distribution of rail freight in the province. The most - 154 - detailed data is in the provincial scale which is not suitable for the regional planning purposes. However, unlike trucking, data on rail transportation does not need any survey due to the limited number of rail carrier in BC. Data availability depends on how much data CN and CPR want to share. It seems that for a better accessibility to rail transportation data, an agreement between the rail industry and the research institutes is needed.   x The other issue that intensifies the lack of data (both passenger and freight) is accessibility. American data and models can be found through on-line sources, where many web sites are devoted to provide the latest transportation data for the public. However, Canadian data is not often published publicly and on-line. Sometimes it is due to confidentiality issues. However, sometimes it is only due to the lack of data management on part of the Canadian sources. It is acceptable that a Canadian data provided by a Canadian source can be only found in the American web sites. Developing a comprehensive data source on the Canadian transportation data, like the American database “BTS”, is highly recommended. This database can improve and motivate research on the transportation in Canada.  x As it was discussed in the section above, a non-linear CPM was found to be a reliable empirical tool to be included in the cost benefit studies instead of conventional collision rates. However, in this study, the CPMs were developed with limited degrees of freedom which means models were fitted using a limited number of inputs. Developing better fitted CPMs for the provincial highways can be the subject of future studies. The CPMs can be used as a reliable tool to predict the safety costs and benefits in the provincial cost-benefit analyses or traffic impact studies. x Demand for the Okanagan Railway is highly dependent on the oil price as it is shown in the mode choice model. High oil price increases the operating cost of autos and trucks and make an electric railway more popular. Moreover, oil price can affect the capital costs of the railway as it affects the construction process. Therefore, an accurate estimation of the oil price is necessary for an accurate cost benefit analysis. - 155 - In this study, literature review on the oil price prediction reveals a gap of knowledge in this area where only two studies have tried to forecast the oil price for years beyond 2030. Moreover, there is no consistency between the literature’s estimations of oil price for years before 2030 where more publications exist. Therefore, developing a model for oil price prediction is proposed as a subject for the future research. x Capital cost estimates are expected to vary drastically from one region to another region. In this study, the attempt was to employ only the capital cost estimates for similar regions and similar projects. However, the estimates still embody a high level of uncertainties. As shown in break-even chart, capital costs have the most critical impact on the feasibility of this project. Therefore, future researches to develop more accurate capital costs estimates are highly encouraged. x In freight demand forecast section, it is assumed that the Okanagan Freight Railway would cooperate with the major railways in the region i.e. CN and CP railway. This cooperation would have the most significant impact on the success of the Okanagan Railway project. While some consultations with the local rail mangers were conducted, more consultations with the managements of CN and CPR are suggested for the future studies. - 156 - REFERENCES ABDEL-ATY, M., 2001. Calibration of Nested-Logit Mode-Choice Models for Florida. Center for Advanced Transportation Systems Simulation, University of Central Florida, Orlando, Florida.ABELSON, P., 1995. Cost benefit analysis of proposed major rail development in Lagos, Nigeria. Transport Reviews, 15(3), pp. 265-289.ACHARYA, J., BASNET, S., SHARMA, K., SUBEDI, P.S. and ADHIKARI, Y.R., 2006. Development of an East-West Electric Railway in Nepal. Nepal: Promotion of renewable energy, energy efficiency and greenhouse gas abatement (PREGA). Available from http://www.adb.org/REACH/NEP-PFS-Development-East-West-Electric-Railway.pdf. Accessed in March 2009. ALAM, M. and RAJAMANICKAM, G., 2007. Development of Truck Payload Equivalent Factor (TPEF). Office of Freight Management and Operations Federal Highway Administration, Washington, D.C. ASSOCIATION OF AMERICAN RAILROADS, 2007. National Rail Freight Infrastructure Capacity and Investment Study, Cambridge Systematics, Inc. Cambridge, Massachusetts. B.C. CLIMATE ACTION TEAM, 2008. Meeting British Columbia’s Targets. Availablefrom http://www.climateactionsecretariat.gov.bc.ca/cat/report.html. Accessed in March 2009.BC STATS, 2007. British Columbia: Regional Employment Projection Model INDUSTRY AND OCCUPATION PROJECTIONS: 2006 to 2011. Victoria, BC.BEALE, M., 2006. Route selection criteria for a new railway, Rail Achieving Growth, 2006, Halcrow Pacific Pty Ltd, Sydney.BLACK, W.R., 1999. Commodity Flow Modeling, Transportation Research Board, Dec,1998, Irvine, California.BOURN, J., 2006. The Modernisation of the West Coast Main Line. London,UK:National Audit Office.  - 157 - BUREAU OF TRANSPORTATION STATISTICS, 2008. National Transportation Statistics. US Department of Transportation, Washington, DC.  CANADA LINE RAPID TRANSIT INC, 2006. Canada Line final project report. Vancouver, BC. CANADIAN AUTOMOBILE ASSOCIATION, 2008. Driving Costs, CAA, Ottawa, ON.COHEN, H., HOROWITZ, A. and PENDYALA, R., 2008. Forecasting Statewide Freight Toolkit, NCHRP Report 606. Transportation Research Board, Washington, D.C.  COLUMBIA RIVER CROSSING, 2006. Feasibility of Diverting Truck Freight to Rail in the Columbia River Corridor, Vancouver, WA.DANE, G. and TECIM, V., 2007. GIS based route determination for light rail system: A Case Study in Izmir, Turkey, Joint Congress of the (ERSA) European Regional Science Association (47th Congress) and ASRDLF (Association de Science Régionale de Langue Française, 44th Congress), August 29th - September 2nd, 2007, Paris.DE RUS, G., 2008. The economic effects of high speed rail investment, August 2008, JOINT TRANSPORT RESEARCH CENTRE, International Transport Forum Available from www.internationaltransportforum.org/jtrc/DiscussionPapers/DP200816.pdf. Accessed in March 2009.DE RUS, G. and INGLADA, V., 1997. Cost-benefit analysis of the high-speed train in Spain. The Annals of Regional Science, 31, pp. 175-181.DJENALIEV, A., 2007. Multicriteria decision making and GIS for railroad planning in Kyrgyzstan, Royal Institute of Technology (KTH), Stockholm, Sweden.  DONOVAN, S., GENTER, J., PETRENAS, B. and MUMBY, N., 2008. Research Report 357: Managing transport challenges when oil prices rise. New Zealand Transport Agency, Wellington, New Zealand.  EPPLI, M. and TU, C.C., 2000. Valuing the New Urbanism; The Impact of New Urbanism on Prices of Single-Family Homes. Urban Land Institute, Washington, DC.  FEDERAL HIGHWAY ADMINISTRATION, 2007. Freight Facts and Figures. Available from http://www.trb.org/news/blurb_detail.asp?id=8617. Accessed in March 2009.- 158 - FLAHERTY, P., 2005. Multi-stage hybrid drives for traction applications, Joint Rail Conference, March 16-18, 2005, Pueblo, Colorado.FORKENBROCK, D., 1999. External costs of intercity truck freight transportation. Transportation Research Part A, 33.FORKENBROCK, D., 2001. Comparison of External Costs of Rail and Truck Freight Transport. Transportation Research A, 35.FRÖIDH, O., 2003. Introduction of regional high speed trains. A study of the effects of the Svealand line on the travel market, travel behavior and accessibility, KTH, Division of Transportation and Logistics, Stockholm, Sweden.  GLEASON, M., BRUNO, B., TANTRI, A. and TYREE, I., 2005. An Analysis of Externality Costs of Freight Transportation in Vermont. Available from http://www.lifecyclesproject.ca/initiatives/food_miles2/downloads/MiddleburyCollegeStudy2005.pdf. Accessed in March 2009. GOVERNMENT OF CANADA, 2007. Building Canada Modern Infrastructure for a Strong Canada. Available from http://www.buildingcanada-chantierscanada.gc.ca/plandocs/booklet-livret/booklet-livret02-eng.html. Accessed in March 2009. GOVERNMENT OF CANADA, 2006. Canada’s Asia-pacific gateway and corridor initiative. Transport Canada. Available from http://www.tc.gc.ca/CanadasGateways/APGCI/document/APGC-PCAP_en.pdf. Accessed in March 2009. GROUNDWORKS, 2000. Public Opinion Survey- Kelowna Transit System Report. BCTransit/Kelowna Transit System, Victoria, BC.  HASS-KLAU, C., 2003. Bus Or Light Rail: Making The Right Choice, Environment and Transport Planning, Brighton Press, Brighton, UK.  HDR and HLB DECISION ECONOMICS INC, 2007. HighwayDEC, A cost benefit model for highway infrastructure investment. Transport Canada, Ottawa, ON.HDR and HLB DECISION ECONOMICS INC, 2005. Economic Evaluation of Transit Investment Options in the Region of Waterloo. Available from http://www.tc.gc.ca/policy/acg/acgd/reports.htm#Rep1. Accessed in March 2009. - 159 - HIGHWAY AGENCY, 1998. Cost Benefit Analysis (COBA) Manual. London, UK.HNTB TEAM, 2007. Statewide Truck Lanes Needs Identification Study, Forecasting and Analysis, Georgia Department of Transportation, Atlanta, Georgia.HOROWITZ, A., 2006. Statewide Travel Forecasting Models, NCHRP Report 358. Transportation Research Board, Washington D.C.  IBI GROUP, 2007. Potential for a Central Okanagan Tour Train: Preliminary Analysis. Central Okanagan Transportation Corridor Project, City of Kelowna, BC.  JACK FAUCETT ASSOCIATES,, 1991. Highway Economic Requirements System Technical Report. Bethesda, MD: Federal Highway Administration, U.S. Department of Transportation, Washington, DC.  KÅGESON, P., 1993. Getting the Prices Right; A European Scheme for Making Transport Pay its True Costs. European Federation for Transport and Environment (T&E), Stockholm, Sweden.  KENWORTHY, J. and LAUBE, F., 2000. Millennium Cities Database For Sustainable Transport. Institute for Sustainability and Technology Policy, Distributed by the International Union of Public Transport, Brussels, Belgium.  KHONDAKER, B., SAYED, T. and LOVEGROVE, G., 2009. Transferability of Community-Based Macrolevel Collision Prediction Models Between Different Time-Space Regions for Road Safety Planning Applications, 2009, Transportation Research Board Washington, DC.  LAM, T.C. and SMALL, K.A., 2001. The Value of Time and Reliability: Measurement from a Value Pricing Experiment. Transportation Research E, 37.LANSDOWNE, Z.F., 1981. Rail Freight Traffic Assignment. Transportation Research Part A: General, 15(2) .  LEVINSON, D., GILLEN, D., KANAFANI, A. and MATHIEU, J.M., 1996a. The Full Cost of Intercity Transportation – A Comparison of High Speed Rail, Air and Highway Transportation in California. Institute of Transportation Studies, University of California at Berkeley. Berkeley, CA. LEVINSON, D., GILLEN, D., KANAFANI, A. and MATHIEU, J., 1996b. The Full Cost of Intercity Transportation – A Comparison of High Speed Rail, Air and Highway - 160 - Transportation in California. Insititute of Transportation Studies University of California at Berkeley. Berkeley, CA. LITMAN, T., 2008. Transportation Elasticities: How Prices and Other Factors Affect Transport Behaviour. Victoria Transport Policy Institute, Victoria, BC.  LITMAN, T.A., 2004. Rail Transit In America: Comprehensive Evaluation of Benefits. Victoria Transport Policy Institute, Victoria, BC.LIVE SMART BC, 2009. fuel savings for truckers. Available: Available from http://www.livesmartbc.ca/transportation/truckers.html. Accessed in March 2009. LOVEGROVE, G., 2007. Road Safety Planning: New Tools for Sustainable Road Safety and Community Development. VDM Verlag Dr. Mueller e.K Saarbrücken, Berlin, Germany.  MACHALABA, D., February 13, 2008. New Era Dawns for Rail Building. The Wall Street Journa, New York, NY.  MEMMOTT, F.W., 1983. Application of statewide freight demand forecasting techniques, NCHRP Report 260.Transportation Research Board, Washington D.C..  MILLER, J., ROBISON, H. and LAHR, M., 1999. Estimating Important Transportation-Related Regional Economic Relationships in Bexar County, Texas. VIA Transit, Texas.NATIONAL RESEARCH COUNCIL,, 1991. TRB Special Report 233: "In Pursuit of Speed: New Options for Intercity Passenger Transport". Transportation Research Board, Washington, DC.NEALE, T.L., 2005. Impacts of Climate Change and Population Growth on Residential Water Demand in the Okanagan Basin, British Columbia, University of British Columbia, Vancouver, BC. NOVA SCOTIA DEPARTMENT OF TRANSPORTATION AND INFRASTRUCTURE RENEWAL, 2008. Long Combination Vehicle Pilot Project-Application Guide, NovaScotia, Canada.OKANAGAN PARTNERSHIP SOCIETY, 2008. Fact sheet - tourism cluster HR study. Kelowna, BC.- 161 - ORTUZAR, J. and WILLUMSEN, L.G., 2001. Modelling transport. Third edn. Wiley, New York, NY.PARKINSON, T. FISHER, I., 1996. TCRP Report 13, "Rail Transit Capacity". Transportation Research Board, Washington, DC. PARSONS BRINCKERHOFF, 2004. VRE strategic plan, phase 2 report, ‘the VRE ridership market”, Virginia Railway Express, Washington, DC.  PARSONS BRINCKERHOFF OHIO, 2004. North South Transportation Initiative Technical Report & Appendices, Columbus, OH.  PB CONSULT INC., 2006. Oregon2 Transitional Model, Model Description at Finalization-Draft Report v1.2, Oregon Department of Transportation, Salem, OR.  POPULATION SECTION, BC STATS., 2005. BRITISH COLUMBIA LOCAL HEALTH AREA POPULATION PROJECTION: P.E.O.P.L.E. Province of British Columbia: Ministry of labour and citizens service. Victoria, BC. QIN, X., IVAN, J. and RAVISHANKER, N., 2004. Selecting exposure measures in crash rate prediction for two-lane highway segments. Accident Analysis and Prevention, (36),pp. 183-191.R.L. BANKS & ASSOCIATES, INC., 2005. RCTC Commuter Rail Feasibility Study Final Report. Riverside County Transportation Commission, Riverside, CA.  REEBIE ASSOCIATES, 2002. Cascade Gateway Freight Demand Analysis, Stamford, CT.REHRL, T. and FRIEDRICH, R., 2006. Modelling long-term oil price and extraction with a Hubbert approach: The LOPEX model. Energy Policy, 34, pp. 2413-2428.ROETS, W., 2002. Terrain suitability analysis for the proposed rapid-rail link between Pretoria, Johannesburg and Johannesburg international airport, RAND AFRIKAANS UNIVERSITY, Johannesburg, South Africa.ROSSI, T.F. and OUTWATER, M.L., 2002. How Valid Is It to Transfer Mode Choice Model Parameters? Seventh TRB Conference on the Application of Transportation Planning Methods, 1999 2002, Transportation Research Board, Washington, DC.  - 162 - SCHIERMEYER CONSULTING SERVICES, 2002. Cloverdale to San Rafael Commuter Rail Ridership and Revenue Forecast. Available from http://www.sonomamarintrain.org/documents/RidershipReport-LEHRev.PDF. Accessedin March 2009.SEGAL, J., 2006. High speed rail - The competitive environment, Sep 18, 2006, European Transport Conference, Strasbourg, France.  SHAPIRO, R.J., HASSETT, K. and ARNOLD, F., 2002. Conserving Energy and Preserving the Environment: The Role of Public Transportation. American Public Transportation Association, Washington, DC.  SHONSEY, J., 2006. Commuter Rail Vehicle Technology. RTD, Denver, CO.SMART, 2008. White Paper No. 1 "SMART Travel in the 21st Century". Sonoma-Marin Area Rail Transit Project. Available from http://www.sonomamarintrain.org/userfiles/file/1_whitepaper_21stcentury.pdf. Accessed in March 2009. T.Y. LIN INTERNATIONAL BASCOR, 1999. Outer Circumferential Commuter Rail Feasibility Study. Metra, Chicago, IL.  TRANSLINK, 2008. Evergreen Line Rapid Transit Project Business Case. Burnaby, BC.TRANSPORT CONCEPTS, 1994. External Costs of Truck and Train. Brotherhood of Maintenance of Way Employees, Ottawa, ON.  TRANSPORTATION ECONOMICS & MANAGEMENT SYSTEMS, INC., 2007. TheOhio & Lake Erie Regional Rail Ohio Hub Study, Frederick, MD.TRANSYSTEMS, 2004. Northern Illinois commuter rail initiative commuter rail feasibility study. City of Rockford, IL.TREASURY BOARD OF CANADA SECRETARIAT, 1998. Canadian Cost-Benefit Analysis Guide. Government of Canada, Ottawa, Canada.  U.S. TREASURY, 2008. Exchange Rates for 2007. Financial Management Service, Washington, DC.UNIVERSITY OF BATH, IER, U.S. and MONTANUNIVERSITÄT LEOBEN, 2000. SAUNER Sustainability And the Use of Non-rEnewable- Resources - Summary Final - 163 - Report. Contract ENV4-CT97-0692. The European Commission DGXII Environment and Climate Programme.  URBAN SYSTEMS Inc, 2004. OKANAGAN VALLEY TRANSPORTATION CORRIDOR An Assessment of Select Projects and Initiatives. Available from http://www.rdno.ca/publications/pdf/ovtc_report.pdf. Accessed in March 2009 VUCHIC, V.R., 2005. Urban Transit-Operations, Planning, and Economics. John Wiley & Sons, Inc, Hoboken, N.J.WILBUR SMITH ASSOCIATES, 2005. DRAFT FINAL REPORT - Santa Barbara Commuter Rail Stud. Santa Barbara County Association of Governments, Santa Barbara, CA.WILLIAMS, S., 2007. Long-Combination Vehicles to run Burnaby to Kamloops. NEWS RELEASE, BC Ministry of Transportation, Victoria, BC.  WINRAM, J., 2007. 2007 North and Central Okanagan Household Travel Survey. TheCity of Vernon & The City of Kelowna, BC.ZHANG, A., BOARDMAN, E., GILLEN, D. and WATERS, W., 2005. TowardsEstimating the Social and Environmental Costs of Transportation in Canada. TransportCanada, Ottawa, ON.ZHAO, F., 2003. Development of FSUTMS Life Cycle and Seasonal Resident Trip Production Models for Florida Urban Areas. Florida Department of Transportation, Tallahassee, Florida. - 164 - APPENDICESA. UTILITY FUNCTIONS Transit and auto utility functions of the FSUTMS model have been shown in Equation A-1 to A-6: 45.10032.0045.0045.0023.002.002.045.0, TFTNTTTTTFWTRTTAATTWTUHBWTr     A-1 10.2005.0035.0035.0035.0015.0015.035.0, TFTNTTTTTFWTRTTAATTWTUHBOTr    A-2 91.10048.0045.0045.0045.0018.0018.045.0, TFTNTTTTTFWTRTTAATTWTUNHBTr    A-3 22.1018.00032.00025.002.045.0, HTDHPCHAOCHRTHTTUHBWAu      A-4 75.0015.0005.0005.0015.035.0, HTDHPCHAOCHRTHTTUHBOAu      A-5 50.0018.00048.00048.0018.045.0, HTDHPCHAOCHRTHTTUNHBAu      A-6 Transit and auto utility functions of the Oregon model have been shown in Equation A-7 to A-12 775.30003.001.001.001.0005.0, CWTDATWATIVTUHHTr     A-7 - 165 - 436.10003.002.002.002.001.0, CWTDATWATIVTUWRTr     A-8 779.10003.001.001.001.0005.0, CWTDATWATIVTUOTr     A-9 314.00003.001.001.001.0005.0, CWTDATWATIVTUHHAu     A-10 278.20003.002.002.002.001.0, CWTDATWATIVTUWRAu     A-1 144.20003.001.001.001.0005.0, CWTDATWATIVTUOAu     A-12 Transit and auto utility functions of the OKI model have been shown in Equation A-13 to A-20:93.1)/(25.10021.00248.00876.00461.00248.00248.0, IVTDATFDATWTTWFWIVTUHBWTr    A-13 34.0)/(25.10021.00248.00876.00461.00248.00248.0, IVTDATFDATWTTWFWIVTUHBUTr    A-14 99.0)/(25.10017.00085.00169.00169.00085.00085.0, IVTDATFDATWTTWFWIVTUHBOTr    A-15 69.0)/(25.1003.00588.00623.00301.00265.00265.0, IVTDATFDATWTTWFWIVTUNHBTr    A-16 59.00021.00021.00248.00248.00248.00248.0, PCAOCDATWTTTIVTUHBWAu      A-17 - 166 - 95.00021.00021.00248.00248.00248.00248.0, PCAOCDATWTTTIVTUHBUAu      A-18 86.50017.00017.00085.00085.00085.00085.0, PCAOCDATWTTTIVTUHBOAu      A-19 39.1003.0003.00265.00265.00265.00265.0, PCAOCDATWTTTIVTUHBWAu      A-20 The Transit Share then can be calculated using Equation A-21: )()()(AuTrTrTrUExpUExpUExpP          A-21 :- 167 - B. THE RELATIONSHIP BETWEEN AUTO AND TRANSIT CONSTANTS Through the process of calibration, it has been discovered that the auto constant and the transit constant are not independent, but there is a linear relationship between their exponentials (between TrCe  andAuCe ). This observation can be proven as follows:AuTrAuTrAuTrTrAuTrTrAuTrTrAurTrTrCaaaabbCbbCaaaaCbbCaaCaaCbbCaaCaaCCbbCaaCaaUUUTreeCeeCeeeCeCeeeeCeeCeeCeeeeeeCeeeCeeeCP.....)..(...........1......1...1...1......1...1...1.........1.........11221122112211221122112211221122112211221122112211221122112211       DDDDEEEEDDDDEEDDDDEEDDDDEEDDDD)...(...1......1221122112211AuCaaaabbTreeCeeCLnC DDDDEEWhere,  1C  is the transit mode share at 2007 survey; TrU  is the transit utility function; AuU is the auto utility function; iD  is ith parameter in transit utility function; ia is the coefficient for iD ;TrC is the constant in transit utility function; iE  is ith parameter in auto utility function; ib is the coefficient for iE ; and AuC is the constant in auto utility function. - 168 - Therefore, it is only necessary to satisfy the above equation, and the values of transit constant and auto constant are not independently important- 169 - C.DATA SOURCES FOR FREIGHT DEMAND ESTIMATES Table  C-1 Data sources number Database  Data Analysis zones in BC Analysis zones in Rest of the Canada Analysis zones in the United states ref#1 1999 NRS National Roadside Survey weekly truck movements (both domestic and transborder) in a representative week of the summer/fall of 1999 4 analysis areas: British Columbia #1: southBritish Columbia #2: north British Columbia #3: lower mainland (Hope to Vancouver) British Columbia #4: west (islands and portion of mainland shore with ferry rather than road access) 22 analysis areas (Figure  C-1)51 analysis areas: Each state is an analysis area. ref#2 IMTC Manifest Survey International Mobility & Trade Corridor Project - Cascade Gateway Almanac survey annual truck movements through Blaine or Sumas in 2004 and 2006 3 analysis areas: West lower mainland East lower mainland Rest of the British Columbia 3 analysis areas: Alberta West of Canada East of Canada 6 analysis areas: Whatcom Puget Sound West Washington State East Washington State West of USA Rest of USA ref#3 BTS TransBorder Freight Data Bureau of Transportation Statisticsmonthly/annual observed Transborder Truck and Rail movements from 1995 to 2007  1 analysis area: Entire British Columbia 13 analysis areas: Each province/territory is an analysis area. 51 analysis areas: Each state is an analysis area. -169-- 170 - C-1 Data sources “continued” numberDatabase DataAnalysis zones in BC Analysis zones in Rest of the CanadaAnalysis zones in the United statesref#4 FAF Version 2.2 Freight Analysis Framework annual Transborder Truck and Rail movements in 2002 and prediction for 2002 to 2035-1 analysis area: Entire Canada 112 analysis areas: (Figure  C-2) Washington state is divided into three zones: Seattle-Tacoma- OlympiaBlaine Rest of Washington ref#5 Cascade Gateway Freight Demand Analysis Reebie Associates’annual Transborder Rail movements through Blaine or Sumas in 2002 and prediction for 2012  2 analysis area: Vancouver Non-CMA (rest of British Columbia)  4 analysis areas: Edmonton, AB Calgary, AB Rest of Alberta Rest of Canada 10 analysis areas: Spokane, WA Richland, WA Seattle, WA  Portland, OR Los Angeles, CA San Francisco, CA Phoenix AZ Salt Lake City UT Jonesboro AR Rest of USA ref#6 Okanagan traffic survey analysis local survey one day truck movement in Okanagan valley via Highway 97 or 97C in June 6, 2007 Individual cities in British Columbia   Individual cities in CanadaIndividual cities in United States -170-- 171 - Figure  C-1 1999 NRS Canadian analysis zones Figure  C-2 FAF analysis areas - 172 - D. COST BENEFIT ANALYSIS CALCULATIONS Table  D-1 Costs and benefits of the project in nominal 2007 dollars    YearCapital  costRolling  stock O & M  costSafety  benefitsExternalities benefits Revenue Total costs Total benefits2009 -198,756,877      -198,756,877 0 2010 -198,756,877      -198,756,877 0 2011 -198,756,877      -198,756,877 0 2012 -198,756,877      -198,756,877 0 2013 -198,756,877 -14,100,000     -212,856,877 0 2014  -200,000 -13,970,914 342,270 34,759,948 14,626,430 -14,170,914 49,728,648 2015  -200,000 -14,264,105 346,185 35,755,248 14,964,562 -14,464,105 51,065,995 2016  -200,000 -14,557,296 350,073 36,750,548 15,302,694 -14,757,296 52,403,315 2017  -200,000 -14,850,487 353,935 37,745,849 15,640,825 -15,050,487 53,740,609 2018  -200,000 -15,143,677 357,772 38,741,149 15,978,957 -15,343,677 55,077,878 2019  -200,000 -15,436,868 361,584 39,736,449 16,317,089 -15,636,868 56,415,122 2020  -200,000 -15,730,059 365,372 40,731,749 16,655,220 -15,930,059 57,752,342 2021  -200,000 -15,991,913 369,137 41,715,044 16,968,282 -16,191,913 59,052,462 2022  -200,000 -16,253,766 372,256 42,698,338 17,281,344 -16,453,766 60,351,937 2023  -3,200,000 -16,515,619 375,359 43,681,632 17,594,405 -19,715,619 61,651,397 2024  -200,000 -16,777,473 378,447 44,664,927 17,907,467 -16,977,473 62,950,840 2025  -200,000 -17,039,326 381,519 45,648,221 18,220,529 -17,239,326 64,250,269 2026  -200,000 -17,301,180 384,577 46,631,515 18,533,590 -17,501,180 65,549,682 2027  -200,000 -17,563,033 387,619 47,614,810 18,846,652 -17,763,033 66,849,081 2028  -200,000 -17,824,887 390,648 48,598,104 19,159,714 -18,024,887 68,148,466 2029  -200,000 -18,086,740 393,662 49,581,398 19,472,776 -18,286,740 69,447,836 2030  -200,000 -18,348,593 396,662 50,564,693 19,785,837 -18,548,593 70,747,193 2031  -200,000 -18,797,873 399,649 51,619,793 20,248,840 -18,997,873 72,268,282 2032  -200,000 -19,247,152 404,707 52,674,893 20,711,842 -19,447,152 73,791,443 2033  -200,000 -19,696,431 409,728 53,729,994 21,174,845 -19,896,431 75,314,566 2034  -200,000 -20,145,711 414,712 54,785,094 21,637,847 -20,345,711 76,837,653 2035  -200,000 -20,594,990 419,662 55,840,194 22,100,849 -20,794,990 78,360,705 2036  -200,000 -25,510,688 424,577 58,606,453 26,136,987 -25,710,688 85,168,016 2037  -200,000 -26,704,370 513,792 59,946,746 27,195,512 -26,904,370 87,656,050 2038  -3,200,000 -27,898,053 531,364 61,287,039 28,254,037 -31,098,053 90,072,440 - 173 - Table D-1 Costs and benefits of the project in nominal 2007 dollars “Continued” 2039  -200,000 -29,091,735 548,603 62,627,333 29,312,562 -29,291,735 92,488,497 2040  -200,000 -30,285,418 565,531 63,967,626 30,371,086 -30,485,418 94,904,244 2041  -200,000 -31,622,376 582,169 65,362,811 31,544,232 -31,822,376 97,489,212 2042  -200,000 -32,959,335 599,800 66,757,996 32,717,378 -33,159,335 100,075,174 2043  -200,000 -34,296,293 617,135 68,153,180 33,890,524 -34,496,293 102,660,839 2044  -3,200,000 -35,633,251 634,190 69,548,365 35,063,669 -38,833,251 105,246,225 2045  -200,000 -36,970,210 650,983 70,943,550 36,236,815 -37,170,210 107,831,348 2046  -200,000 -38,307,168 667,528 72,338,734 37,409,961 -38,507,168 110,416,223 2047  -200,000 -39,644,127 683,838 73,733,919 38,583,106 -39,844,127 113,000,863 2048  -200,000 -40,981,085 699,925 75,129,104 39,756,252 -41,181,085 115,585,281 2049  -200,000 -42,318,044 715,800 76,524,289 40,929,398 -42,518,044 118,169,486 2050  -200,000 -43,655,002 731,473 77,919,473 42,102,544 -43,855,002 120,753,490 2051  -200,000 -47,302,871 746,954 80,200,006 45,124,418 -47,502,871 126,071,378 2052  -200,000 -50,950,740 790,250 82,480,538 48,146,292 -51,150,740 131,417,080 2053  -3,200,000 -54,598,610 832,215 84,761,071 51,168,167 -57,798,610 136,761,452 2054  -200,000 -58,246,479 872,991 87,041,603 54,190,041 -58,446,479 142,104,636 2055  -200,000 -61,894,348 912,697 89,322,136 57,211,915 -62,094,348 147,446,748 2056  -200,000 -65,542,217 951,430 91,602,669 60,233,790 -65,742,217 152,787,888 2057  -200,000 -69,190,086 989,275 93,883,201 63,255,664 -69,390,086 158,128,140 2058  -200,000 -72,837,956 1,026,304 96,163,734 66,277,538 -73,037,956 163,467,576 2059  -3,200,000 -76,485,825 1,062,579 98,444,266 69,299,413 -79,685,825 168,806,258 2060  -200,000 -80,133,694 1,098,155 100,724,799 72,321,287 -80,333,694 174,144,241 2061  -200,000 -89,793,515 1,133,081 105,308,609 80,152,723 -89,993,515 186,594,412 2062  -200,000 -99,453,336 1,224,178 109,892,419 87,984,159 -99,653,336 199,100,755 2063  -3,200,000 -109,113,157 1,311,541 114,476,228 95,815,594 -112,313,157 211,603,363 2064  -200,000 -118,772,978 1,395,687 119,060,038 103,647,030 -118,972,978 224,102,755 2065  -200,000 -128,432,799 1,477,022 123,643,848 111,478,466 -128,632,799 236,599,336 2066  -6,200,000 -138,092,620 1,555,872 128,227,658 119,309,902 -144,292,620 249,093,431 2067  -200,000 -147,752,441 1,632,500 132,811,468 127,141,338 -147,952,441 261,585,305 2068  -3,200,000 -157,412,262 1,707,126 137,395,278 134,972,773 -160,612,262 274,075,177 2069  -200,000 -167,072,083 1,779,934 141,979,088 142,804,209 -167,272,083 286,563,231 2070  -3,200,000 -176,731,904 1,851,082 146,562,898 150,635,645 -179,931,904 299,049,624 2071  -200,000 -195,816,208 1,920,703 154,757,382 166,006,667 -196,016,208 322,684,751 2072  -3,200,000 -214,900,511 2,059,632 162,951,865 181,377,688 -218,100,511 346,389,186 2073  -3,200,000 -233,984,815 2,193,417 171,146,349 196,748,710 -237,184,815 370,088,477 - 174 - Table D-1 Costs and benefits of the project in nominal 2007 dollars “Continued” 2074  -3,200,000 -253,069,118 2,322,713 179,340,833 212,119,732 -256,269,118 393,783,278 2075  -200,000 -272,153,422 2,448,040 187,535,317 227,490,754 -272,353,422 417,474,111 2076  -3,200,000 -291,237,726 2,569,825 195,729,800 242,861,776 -294,437,726 441,161,401 2077  -3,200,000 -310,322,029 2,688,419 203,924,284 258,232,797 -313,522,029 464,845,500 2078  -3,200,000 -329,406,333 2,804,116 212,118,768 273,603,819 -268,713,572 488,526,703 2079  -3,200,000 -348,490,636 2,917,168 220,313,252 288,974,841 -351,690,636 512,205,261 2080  -3,200,000 -367,574,940 3,027,791 228,507,736 304,345,863 -370,774,940 535,881,389 2081  -3,200,000 -386,659,243 3,136,171 236,702,219 319,716,885 -389,859,243 559,555,274 2082  -3,200,000 -405,743,547 3,242,470 244,896,703 335,087,906 -408,943,547 583,227,080 2083  -3,200,000 -424,827,850 3,346,834 253,091,187 350,458,928 -428,027,850 606,896,949 2084  -200,000 -443,912,154 3,449,389 261,285,671 365,829,950 -444,112,154 630,565,009 2085  -3,200,000 -462,996,457 3,550,248 269,480,154 381,200,972 -466,196,457 654,231,374 2086  -3,200,000 -482,080,761 3,649,514 277,674,638 396,571,993 -485,280,761 677,896,145 2087  -3,200,000 -501,165,064 3,747,276 285,869,122 411,943,015 -504,365,064 701,559,414 2088  -6,200,000 -520,249,368 3,843,620 294,063,606 427,314,037 -526,449,368 725,221,263 2089  -3,200,000 -539,333,671 3,938,619 302,258,089 442,685,059 -542,533,671 748,881,767 2090  -3,200,000 -558,417,975 4,032,341 310,452,573 458,056,081 -561,617,975 772,540,995 2091  -3,200,000 -577,502,278 4,124,851 318,647,057 473,427,102 -580,702,278 796,199,010 2092  -3,200,000 -596,586,582 4,216,205 326,841,541 488,798,124 -599,786,582 819,855,870 2093  -200,000 -615,670,885 4,306,456 335,036,025 504,169,146 -615,870,885 843,511,626 2094  -3,200,000 -634,755,189 4,395,653 343,230,508 519,540,168 -637,955,189 867,166,329 2095  -3,200,000 -653,839,492 4,483,841 351,424,992 534,911,190 -657,039,492 890,820,023 2096  -3,200,000 -672,923,796 4,571,062 359,619,476 550,282,211 -676,123,796 914,472,750 2097  -3,200,000 -692,008,099 4,657,356 367,813,960 565,653,233 -695,208,099 938,124,549 2098  -3,200,000 -711,092,403 4,742,759 376,008,443 581,024,255 -714,292,403 961,775,457 - 175 - Table  D-2 Net benefits, net costs and present worth of the project in real 2007 dollars Year Benefit Real 2007$ Cost Real 2007$ Present Worth  Real 2007 US$ 2009 0 -151,869,344 -151,869,3442010 -132,752,923 -132,752,923 2011 0 -116,042,765 -116,042,7652012 -101,435,983 -101,435,983 2013 0 -94,957,996 -94,957,9962014 19,392,084 -5,526,061 13,866,023 2015 17,406,987 -4,930,414 12,476,5732016 15,614,373 -4,397,163 11,217,210 2017 13,997,238 -3,920,038 10,077,1992018 12,539,809 -3,493,359 9,046,450 2019 11,227,504 -3,111,985 8,115,5202020 10,046,882 -2,771,271 7,275,610 2021 8,979,945 -2,462,259 6,517,685 2022 8,022,336 -2,187,132 5,835,204 2023 7,163,521 -2,290,836 4,872,685 2024 6,393,801 -1,724,371 4,669,430 2025 5,704,355 -1,530,565 4,173,789 2026 5,087,169 -1,358,229 3,728,940 2027 4,534,976 -1,205,027 3,329,949 2028 4,041,193 -1,068,873 2,972,320 2029 3,599,865 -947,903 2,651,962 2030 3,205,610 -840,451 2,365,159 2031 2,862,353 -752,455 2,109,898 2032 2,554,791 -673,295 1,881,496 2033 2,279,304 -602,141 1,677,163 2034 2,032,691 -538,233 1,494,458 2035 1,812,048 -480,873 1,331,175 2036 1,721,559 -519,708 1,201,851 2037 1,548,821 -475,381 1,073,439 2038 1,391,186 -480,315 910,871 2039 1,248,691 -395,469 853,222 2040 1,120,023 -359,777 760,246 2041 1,005,708 -328,283 677,425 - 176 - Table D-2 Net benefits, net costs and present worth of the project in real 2007 dollars “Continued” 2042 902,434 -299,016 603,418 2043 809,222 -271,916 537,306 2044 725,176 -267,572 457,604 2045 649,466 -223,875 425,590 2046 581,324 -202,734 378,590 2047 520,045 -183,368 336,677 2048 464,981 -165,665 299,316 2049 415,540 -149,513 266,026 2050 371,177 -134,803 236,373 2051 338,744 -127,636 211,107 2052 308,660 -120,138 188,522 2053 280,780 -118,664 162,116 2054 255,026 -104,890 150,136 2055 231,306 -97,410 133,896 2056 209,514 -90,151 119,364 2057 189,543 -83,176 106,367 2058 170,600 -76,528 94,072 2059 153,983 -72,984 80,999 2060 138,845 -64,316 74,529 2061 130,057 -62,980 67,076 2062 121,281 -60,962 60,319 2063 112,654 -60,058 52,595 2064 104,277 -55,612 48,665 2065 96,224 -52,558 43,665 2066 88,546 -51,536 37,010 2067 81,276 -46,191 35,085 2068 74,797 -43,832 30,965 2069 68,361 -39,903 28,457 2070 62,360 -37,521 24,839 2071 58,818 -35,729 23,089 2072 55,192 -34,751 20,441 2073 51,545 -33,035 18,511 2074 47,942 -31,200 16,742 2075 44,428 -28,984 15,444 2076 41,039 -27,390 13,649 - 177 - Table D-2 Net benefits, net costs and present worth of the project in real 2007 dollars “Continued” 2077 37,800 -25,494 12,305 2078 34,725 -23,642 11,083 2079 31,825 -21,852 9,973 2080 29,105 -20,138 8,967 2081 26,565 -18,509 8,056 2082 24,204 -16,971 7,233 2083 22,016 -15,527 6,489 2084 19,995 -14,083 5,912 2085 18,134 -12,922 5,212 2086 16,425 -11,758 4,667 2087 14,859 -10,682 4,177 2088 13,426 -9,746 3,680 2089 12,119 -8,780 3,3392090 10,929 -7,945 2,984 2091 9,845 -7,181 2,665 2092 8,862 -6,483 2,379 2093 7,970 -5,819 2,151 2094 7,162 -5,269 1,893 2095 6,431 -4,744 1,688 2096 5,771 -4,267 1,504 2097 5,175 -3,835 1,340 2098 4,638 -3,444 1,193 - 178 - E. MONTE CARLO SIMULATION RESULTS  The results are shown in Figure  E-1 to Figure  E-7. Note that there are more than 120 Monte Carlo simulation trials included in each graph. Benefit-Cost Ratio for Okanagn rail running between 2009 and 20590%4%8%12%16%20%0.150.250.350.450.550.650.750.850.951.051.151.251.351.451.55B/CProbabilityFigure  E-1 Probability distribution for Scenario "a" Monte Carlo simulation results (see Figure  E-1) show that for the scenario that has the railway start operating in 2009, Scenario “a”, the probability of having a benefit cost ratio > 1 is 15%. The average net present worth of the project is -264.69 million dollars.  Thus, it is deemed to fail. Benefit-Cost Ratio for Okanagn rail running between 2019 and 20690%2%4%6%8%10%12%14%0.150.250.350.450.550.650.750.850.951.051.151.251.351.451.551.651.75B/CProbabilityFigure  E-2 Probability distribution for Scenario "b" - 179 - Monte Carlo simulation results (see Figure  E-2) show that for the railway to start operating in 2019, Scenario “b”, the probability of having a benefit cost ratio greater than one is 26%. The average net present worth of the project is -48.54 million dollars.  Thus, it is deemed to fail also. Benefit-Cost Ratio for Okanagn rail running between 2029 and 20790%4%8%12%16%0.150.250.350.450.550.650.750.850.951.051.151.251.351.451.551.651.751.85B/CProbabilityFigure  E-3 Probability distribution for Scenario "c" Monte Carlo simulation results (see Figure  E-3) show that for the railway to start operating in 2029, Scenario “c”, the probability of having a benefit cost ratio greater than one is 43%. The average net present worth of the project is 28.41 million dollars.  This meets minimum success. Benefit-Cost Ratio for Okanagn rail running between 2039 and 20890.00%4.00%8.00%12.00%16.00%20.00%0.250.350.450.550.650.750.850.951.051.151.251.351.451.551.651.751.851.95B/CProbabilityFigure  E-4 Probability distribution for Scenario "d" - 180 - Monte Carlo simulation results (see Figure  E-4) show that for the railway start operating in 2039, Scenario “d”, the probability of having a benefit cost ratio greater than one is 60%. The average net present worth of the project is 49.59 million dollars.  This is again considered reasonably successful.Benefit-Cost Ratio for Okanagn rail running between 2049 and 20990.00%4.00%8.00%12.00%16.00%20.00%0.350.450.550.650.750.850.951.051.151.251.351.451.551.651.751.851.952.05B/CProbabilityFigure  E-5 Probability distribution for Scenario "e" Monte Carlo simulation results (see Figure  E-5) show that for the railway start operating in 2049, Scenario “e”, the probability of having a benefit cost ratio greater than one is 72%. The average net present worth of the project is 52.88 million dollars.  This presents the most successful scenario.Benefit-Cost Ratio for Kelowna-Vernon Comuter Rail running between 2019 and 20690.00%5.00%10.00%15.00%20.00%25.00%0.050.150.250.350.450.550.650.750.850.951.051.151.251.35B/CProbabilityFigure  E-6 Probability distribution for Scenario "f" - 181 - Monte Carlo simulation results (see Figure  E-6) show that for the Commuter rail start operating between Kelowna and Vernon in 2019, Scenario “f”, the probability of having a benefit cost ratio greater than one is 4%. The average net present worth of the project is -60.22 million dollars. Thus, it was deemed as not feasible and a failed scenario. Benefit-Cost Ratio for Kelowna-Vernon Comuter Rail running between 2049 and 20990.00%5.00%10.00%15.00%20.00%25.00%0.050.150.250.350.450.550.650.750.850.951.051.151.251.351.451.55B/CProbabilityFigure  E-7 Probability distribution for Scenario "g" Monte Carlo simulation results (see Figure  E-7) show that for the Commuter rail start operating between Kelowna and Vernon in 2049, Scenario “g”, the probability of having a benefit cost ratio greater than one is 37%. The average net present worth of the project is -3.00 million dollars.  Thus, while closer to breaking even, it was deemed unsuccessful as well. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Country Views Downloads
United States 41 0
China 37 6
Canada 17 6
New Zealand 7 0
Turkey 5 0
France 5 0
Germany 2 5
Russia 2 0
Taiwan 2 0
Italy 2 0
Iran 1 1
Brazil 1 1
Vietnam 1 0
City Views Downloads
Unknown 29 9
Ashburn 18 0
Beijing 16 0
Hangzhou 12 0
Shenzhen 9 6
Washington 7 0
Vancouver 5 3
Mountain View 4 0
Toronto 4 0
Stuttgart 2 0
Wilmington 2 0
Milan 2 0
Los Angeles 2 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0063128/manifest

Comment

Related Items