QUIMPER PENINSULA TRAVEL DEMAND MODEL DOCUMENTATION

Transcription

1 Jefferson County QUIMPER PENINSULA TRAVEL DEMAND MODEL DOCUMENTATION Prepared for: Jefferson County October 2008 Prepared by: th Avenue NE, Suite 600 Kirkland, WA Phone: Fax: Transpo Group

2 Table of Contents INTRODUCTION...1 MODEL NETWORK...3 Transportation Analysis Zones... 3 Links... 3 Nodes... 6 Turn Penalties... 7 EXISTING AND FORECAST LAND USE...8 Residential Forecasts... 9 Employment Forecasts External Growth TRIP GENERATION...12 Trip Rates Trip Types Trip Balancing External TAZ Trip Generation TRIP DISTRIBUTION AND ASSIGNMENT...15 Trip Distribution Trip Assignment VALIDATION AND CALIBRATION RESULTS...16 Screen Line Analysis Link Volume Analysis BASELINE MODEL...19 Baseline Model Results APPENDIX A: MODEL PLOTS APPENDIX B: TURN PENALTIES APPENDIX C: LAND USE PROCESS APPENDIX D: EXISTING AND FORECAST LAND USE i

3 List of Exhibits Exhibit 1 Quimper Peninsula Model Study Area... 2 Exhibit 2 Quimper Peninsula Model TAZs... 4 Exhibit 3 Quimper Peninsula Model Links... 5 Exhibit 4 Model Link Types and Standards... 6 Exhibit 5 Node Classifications... 7 Exhibit Quimper Peninsula Model Land Use Categories... 8 Exhibit and 2031 Employment and Household Land Use Data... 9 Exhibit 8 Model Sub-Areas Exhibit 9 Weekday PM Peak Hour Rates Exhibit 10 Weekday Daily Trip Rates by Trip Type Exhibit and 2026 External TAZ Trip Generation Exhibit 12 Trip Distribution Equation and Parameters Exhibit Quimper Peninsula Model Screen Lines Exhibit 14 Quimper Peninsula Model Screen Lines Exhibit 15 Statistical Analysis of Calibrated Data Exhibit Baseline Model Network Changes ii

4 Introduction A Travel Demand Model was created for Jefferson County to assist in the analysis of future transportation system needs and alternatives on the Quimper Peninsula. The Travel Demand Model was constructed based on the 2006 Port Townsend Model using the software platform VISUM. The Quimper Peninsula Model is a PM Peak Hour model designed to model traffic volumes during the single hour of the afternoon that has the highest hourly volume and which usually represents the highest traffic volumes of any hour during the day. The Quimper Peninsula Model was calibrated to conditions for the existing year (2007) and then used to forecast volumes based on 2031 land use projections roadway network assumptions. The study area of the Quimper Peninsula Model encompasses all of the Quimper Peninsula north of SR 104 and east of US 101, including Port Townsend, the Irondale and Port Hadlock Urban Growth Area (UGA), and the Port Ludlow Master Planned Resort (MPR). Marrowstone and Indian Islands are considered external to the model. Exhibit 1 illustrates the Quimper Peninsula Model study area. The model consists of the following elements which will be documented in this report: 1. Model Network The model network consists of Transportation Analysis Zones (TAZs) to facilitate the gathering and organization of land use data, links that represent roadways, and nodes that represent intersections. 2. Existing and Future Land Use The 2007 existing and 2031 future baseline residential and commercial land uses were summarized based on data provided by the City of Port Townsend and by Jefferson County and sorted into the appropriate TAZs. 3. Trip Generation Trip generation is the process by which the total number of trips is estimated for travel coming to and going from each TAZ in the model during the PM Peak Hour. Trip Generation rates are based on the 2006 Port Townsend Model and are similar to rates suggested by the Institute of Transportation Engineers for each land use type. The rates are adjusted during calibration to reflect the travel characteristics of residential, commercial, and leisure trips that occur in the Quimper Peninsula. 4. Trip Distribution and Assignment Each trip generated in the model originates in one TAZ and has a destination in another TAZ during the PM Peak Hour. Trip Distribution is the process by which the interchanges between TAZs are determined. Assignment is the process by which trips between TAZs are assignment to specific links in the model. The process of determining the routes that each trip will take between TAZs is the main purpose and the ultimate objective of a Travel Demand Model. 5. Validation and Calibration Validation and calibration is the process by which the model is tested against existing conditions. The purpose of calibration and validation is to provide a reasonable assurance that the model accurately represents the travel patterns, behaviors, and decisions that influence travel in the model study area. The outputs of the model are tested against current field data and travel surveys Baseline Model The 2031 baseline model is the model against which future alternatives are tested. The baseline model contains land use forecasts for the year 2031 and all of the roadway and intersection capacity improvements that are most likely to be completed by Page 1

5 D OAK BAY RD Strait of Juan De Fuca RIDGE DR MCMINN RD CA PE GEORGE RD HASTINGS AV W LOFTUS RD SAND RD S DISCOVERY RD SR 20 OTTO ST IVY ST COOK AV M HASTINGS AV DISCOVERY RD ILL RD PARKRID GE DR ENGEL RD 35TH ST 4 9T H ST 14TH ST 9TH ST SIMS W AY 57TH ST 53RD ST 25TH ST KALA POINT DR SAN JUAN AV W ST FIR ST F ST BLAINE ST CLAY ST 20 WATER ST Port Townsend Admiralty Inlet SCHWARTZ RD OLD GARDINER RD Discovery CUB RD 20 ANDERS O N LAKE RD PROSPECT AV KENNEDY RD E MOORE ST IRONDALE RD D ST CHIMACUM RD US Naval Reservation 116 WASHINGTO N LN Oak SR 116 EAST MARROWSTONE RD ROBBIN S RD W UNCAS R D VA N TROJ E N RD CENTER RD 19 N BAY WAY Puget Sound SNOW C REEK R D 101 LELAND VALLEY RD E Legend LI ND R D TARBOO LAKE RD EAGLEMOUNT RD PETERS O N RD Model Study Area Port Townsend OLD T ARBOO RD GIB BS LA KE RD 104 EGG & I RD WEST VALLEY RD SANDY SHORE RD DENA LN L ARSON LAKE RD SHINE RIDGE RD SWANSONVILLE RD TODD RD OLELE POINT RD WALKER WAY OLYMPUS BLVD LUDLO W BA Y RD HIGHLAND DR ANDY COO PER R D T EAL LAKE RD PARADISE BAY RD ALPINE CT DABO B R SHINE RD Exhibit 1 Model Study Area Quimper Peninsula Model Documentation

6 Model Network The network for the Quimper Peninsula Model was built starting with the 2006 Port Townsend Model. TAZs were created with input from Jefferson County that encompassed the remainder of the Quimper Peninsula north of SR 104 and were sufficiently refined to account for the Port Hadlock UGA and the Port Ludlow MPR. The remainder of the Quimper Peninsula network was constructed using NAVTEQ data. NAVTEQ data contain features such as water, parks, streets, street names, permitted and restricted turns, one-way streets, intersections, and posted speeds. NAVTEQ data are commonly used in personal navigation devices and are constantly updated to reflect existing conditions. After the NAVTEQ data were added to the model the TAZs, links, and nodes were checked for accuracy and systematically classified and adjusted for calibration. Transportation Analysis Zones Exhibit 2 illustrates the Quimper Peninsula Model TAZs. The TAZs in the study area were created based on the type and intensity of existing and future land use, zoning, natural features, and transportation system assumptions. Where possible, similar land uses were grouped together. TAZs inside of Port Townsend retain the same structure and shape as the 2006 Port Townsend Model. The Quimper Peninsula Model TAZs are organized accordingly: Zones numbered between 1 and 200 are Port Townsend TAZs. Zones numbered between 201 and 699 are Jefferson County TAZs including the Port Hadlock UGA and the Port Ludlow MPR. Zones 700 through 705 represent external inputs, such as the Port Townsend-Keystone Ferry, the Hood Canal Bridge, Center Road, and US 101. A map of the model TAZs and the sub-area boundaries for Port Townsend, the Port Hadlock UGA, the Port Ludlow MPR, etc., can be found in Appendix C. Links The links in the model network represent the state highways, principal arterials, minor arterials, collectors, and some business and local access roads as illustrated in Exhibit 3. Some of the links that represent business and local access roads appear as small stubs off of major roadways and are important because they represent locations where traffic is being loaded onto the roadway network to and from a particular TAZ. Some of the stubs in the model represent the location where future connections will be added and ensure that the proper level of detail is available in that portion of the roadway network to provide an accurate comparison between existing and future models. Each link in the model is coded with a capacity and speed based on the functional class, number of lanes, and relative number of access points and level of side street traffic. Page 3

7 D OAK BAY RD Strait of Juan De Fuca RIDGE DR MCMINN RD CA PE GEORGE RD HASTINGS AV W LOFTUS RD SAND RD S DISCOVERY RD SR 20 OTTO ST IVY ST COOK AV M HASTINGS AV DISCOVERY RD ILL RD PARKRID GE DR ENGEL RD 35TH ST 4 9T H ST 14TH ST 9TH ST SIMS W AY 57TH ST 53RD ST 25TH ST KALA POINT DR SAN JUAN AV W ST FIR ST F ST BLAINE ST CLAY ST 20 WATER ST Port Townsend Admiralty Inlet SCHWARTZ RD OLD GARDINER RD Discovery CUB RD 20 ANDERS O N LAKE RD PROSPECT AV KENNEDY RD E MOORE ST IRONDALE RD D ST CHIMACUM RD US Naval Reservation 116 WASHINGTO N LN Oak SR 116 EAST MARROWSTONE RD ROBBIN S RD W UNCAS R D VA N TROJ E N RD CENTER RD 19 N BAY WAY Puget Sound SNOW C REEK R 101 LELAND VALLEY RD E D Legend Model TAZs Externals LI ND R D TARBOO LAKE RD EAGLEMOUNT RD PETERS O N RD OLD T ARBOO RD GIB BS LA KE RD 104 EGG & I RD WEST VALLEY RD SANDY SHORE RD DENA LN L ARSON LAKE RD SHINE RIDGE RD SWANSONVILLE RD TODD RD OLELE POINT RD WALKER WAY OLYMPUS BLVD LUDLO W BA Y RD HIGHLAND DR ANDY COO PER R D T EAL LAKE RD PARADISE BAY RD ALPINE CT DABO B R SHINE RD Exhibit 2 Model TAZs Quimper Peninsula Model Documentation

8 D OAK BAY RD Strait of Juan De Fuca RIDGE DR MCMINN RD CA PE GEORGE RD HASTINGS AV W LOFTUS RD SAND RD S DISCOVERY RD SR 20 OTTO ST IVY ST COOK AV M HASTINGS AV DISCOVERY RD ILL RD PARKRID GE DR ENGEL RD 35TH ST 4 9T H ST 14TH ST 9TH ST SIMS W AY 57TH ST 53RD ST 25TH ST KALA POINT DR SAN JUAN AV W ST FIR ST F ST BLAINE ST CLAY ST WATER ST Port Townsend Admiralty Inlet SCHWARTZ RD OLD GARDINER RD Discovery CUB RD ANDERS O N LAKE RD PROSPECT AV KENNEDY RD E MOORE ST IRONDALE RD D ST CHIMACUM RD US Naval Reservation WASHINGTO N LN Oak SR 116 EAST MARROWSTONE RD ROBBIN S RD W UNCAS R D VA N TROJ E N RD CENTER RD N BAY WAY Puget Sound LI ND R D EAGLEMOUNT RD PETERS O N RD GIB BS LA KE RD EGG & I RD WEST VALLEY RD DENA LN SWANSONVILLE RD OLELE POINT RD OLYMPUS BLVD SNOW C REEK R LELAND VALLEY RD E D Legend Model Links Local Roads TARBOO LAKE RD OLD T ARBOO RD SANDY SHORE RD L ARSON LAKE RD SHINE RIDGE RD TODD RD WALKER WAY LUDLO W BA Y RD HIGHLAND DR ANDY COO PER R D T EAL LAKE RD PARADISE BAY RD ALPINE CT DABO B R SHINE RD Exhibit 3 Model Links Quimper Peninsula Model Documentation

9 During calibration link attributes such as speeds, number of lanes, and capacities, were refined and adjusted to represent existing conditions. The roadway functional class, capacity, number of lanes, and speed values listed in Exhibit 4 were used as guidelines during calibration of link attributes. Appendix A contains model plots of links attributes. The speeds (or other parameters) can be adjusted in the model setup, as appropriate, to support the model calibration process. Link capacities adjusted for calibration purposes were kept to within 200 veh/hour of the standard link capacity. Link speeds were kept to within 10 mph of the standard link speed or were changed to match actual posted speed limits or real world driving conditions during the PM Peak Hour. Exhibit 4 Model Link Types and Standards 1 Type # Type Description Model Link Capacity Standards (vph) Model Link Speed Standards (mph) 21 Rural Principal (2 Lane) Urban Principal (2 Lane) 1, Rural Principal (3 Lane) 1, Urban Principal (3 Lane) 1, Rural Minor (2 Lane) Urban Minor (2 Lane) Rural Minor (3 Lane) Urban Minor (3 Lane) Rural Collector (2 Lane) Urban Collector (2 Lane) Rural Collector (3 Lane) Urban Collector (3 Lane) Urban Local Rural Local Link types, capacities, and speeds serve as guidelines to calibrate the model. During calibration the attributes on some links and nodes were adjusted from the model standard in an attempt to model actual conditions. For example, the average speeds of motorists on some roadways may not match the posted speed limits. Link attributes were developed based on attributes in the 2006 Port Townsend Model. The VISUM software adjusts travel speeds during the assignment process based on defined volume-delay functions. As congestion (measured by the volume-to-capacity ratio, or V/C ratio) increases, travel speeds on the link decrease. This results in traffic shifting to other corridors to minimize travel times between two TAZs. Centroid connectors are used to connect the trips generated within each TAZ to the model network. Within the study area, the multi-point assignment process within VISUM was used to assign a defined percentage of vehicle trips to/from a TAZ to a particular centroid connector based on observed conditions. Multi-point assignment allows the model to better reflect existing travel patterns by directing the model where to load traffic from a TAZ to the street system. Nodes Two nodes are used to define each link in the model network. A node is simply the location where two or more links come together to form the network. Nodes were classified according to intersection control type and roadway functional classification. The node classifications are listed in Exhibit 5. Page 6

10 Exhibit 5 Node Classifications Type # Type Description 20 Rural One-/Two-way stop, high flow 21 Urban One-/Two-way stop, high flow 30 Rural One-/Two-way stop, low flow 31 Urban One-/Two-way stop, low flow 40 Rural All-way stop 41 Urban All-way stop 51 Signal, Minor/Collector junction 52 Signal, Principal/Collector junction 53 Signal, Minor/Minor or Principal/Minor junction 54 Signal, Principal/Principal junction 81 Roundabouts Turn Penalties Turn penalties are used along with link capacities and speeds in the VISUM assignment to define the shortest travel time path between all TAZs. The travel times and distances are then used in assigning traffic to specific routes. Nodes can also be placed along corridors during the calibration process in order to introduce delay on routes where VISUM is overassigning volumes. A table containing the turn penalties used in the Quimper Peninsula Model can be found in Appendix B. Page 7

11 Existing and Forecast Land Use Land use data were used in the modeling process to estimate the quantity of travel activity associated with each TAZ. The land use process was a coordinated effort between Transpo, Jefferson County, and the City of Port Townsend. Jefferson County was primarily responsible for providing the 2007 and 2031 land use data outside of Port Townsend. The process by which the land use was calculated and refined for the Quimper Peninsula Model is documented in Appendix C. Existing and future land use data provided by Jefferson County were quantified according to the categories shown in Exhibit 6. Exhibit Quimper Peninsula Model Land Use Categories Land Use Type Units Land Use Description SFDU Dwelling Units Single-family dwelling units in Port Townsend CntySFDU Dwelling Units Single-family dwelling units outside Port Townsend MFDU Dwelling Units Multi-family dwelling units, including duplexes RetireDU Dwelling Units Retirement dwelling units/second homes NRC Employees Natural resource (Agriculture, Forestry, Fishing, Mining) and construction employment Manuf Employees Manufacturing and Industrial employment CTU Employees Communication, Transportation, Utilities employment Whole Employees Wholesale employment Retail-High Employees Retail employment along Upper Sims Way Retail-Low Employees Retail employment along Lower Sims Way/Water Street CntyRetail Employees Retail employment outside of Port Townsend FIRE Employees Finance, Insurance, and Real Estate employment Edu Employees Education employees (Elementary and Secondary) Med Employees Medical employment Office/Other Employees Office or other services employment CntyOffice Employees Office or other services employment outside of Port Townsend GOV Employees Government employment Motel Rooms Motels, hotels, and resort destinations Exhibit 7 summarizes the existing and future land use data for households and employment organized by sub-area including Port Townsend, North Peninsula, Mid-Peninsula, the Port Hadlock UGA, South Peninsula, and the Port Ludlow MPR. Page 8

12 Exhibit and 2031 Employment and Household Land Use Data 1 Model Sub-Area Number of Household Units Number of Employment Units Percent Increase Percent Increase Port Townsend 4,643 7, % 5,451 9, % North Peninsula 1,270 1, % 649 1, % Mid-Peninsula 1,692 2, % 927 1, % Port Hadlock UGA 1,230 2, % 1,007 2, % South Peninsula 1,644 2, % % Port Ludlow MPR 1,328 2, % % Total 11,807 18, % 8,593 15, % 1. Data provided by the City of Port Townsend and Jefferson County. 2. The boundaries of the model sub-areas as represented in the model documentation may or may not exactly match the boundaries as defined and adopted by local agencies because the model documentation boundaries are based on the model TAZ structure which is not as refined. However, the boundaries should be fairly close. Household land use data account for single-family and multi-family dwelling units as well as homes of person who are retired, and second homes or vacation homes that may or may not be occupied year round. The employment land use data comprise several categories ranging from office jobs to retail, manufacturing, and wholesale jobs as well as education related employment as outlined in Exhibit 6. Exhibit 8 illustrates the model sub-areas that are listed in Exhibit 7 while a detailed map of each sub-area can be found in Appendix C. Residential Forecasts Residential land use growth can be characterized by the following: A majority of the SFDU and CntySFDU growth will most likely occur in the City of Port Townsend, the Port Hadlock UGA, and the Port Ludlow MPR. A majority of the RetireDU, or second home/vacation home growth will most likely occur in the Port Ludlow MPR and along Paradise Road from SR 104 to the Port Hadlock UGA. The Port Ludlow MPR is a master planned resort community and the total number of dwelling units at build-out is capped by the programmatic Environmental Impact Study that was completed in Most of the 134 percent growth in MFDU units will most likely occur in the City of Port Townsend west of Discovery Road and between Sims Way and Hastings Avenue as well as along lower Sims Way. Significant MFDU growth outside of Port Townsend will most likely occur in the Port Hadlock UGA when sewer service is available. Page 9

13 D OAK BAY RD Strait of Juan De Fuca RIDGE DR MCMINN RD CA PE GEORGE RD HASTINGS AV W LOFTUS RD SAND RD S DISCOVERY RD SR 20 OTTO ST IVY ST COOK AV M HASTINGS AV DISCOVERY RD ILL RD PARKRID GE DR ENGEL RD 35TH ST 4 9T H ST 14TH ST 9TH ST SIMS W AY 57TH ST 53RD ST 25TH ST KALA POINT DR SAN JUAN AV W ST FIR ST F ST BLAINE ST CLAY ST 20 WATER ST Port Townsend Admiralty Inlet SCHWARTZ RD OLD GARDINER RD Discovery CUB RD 20 ANDERS O N LAKE RD PROSPECT AV KENNEDY RD E MOORE ST IRONDALE RD D ST CHIMACUM RD US Naval Reservation 116 WASHINGTO N LN Oak SR 116 EAST MARROWSTONE RD ROBBIN S RD W UNCAS R D VA N TROJ E N RD CENTER RD 19 N BAY WAY Puget Sound SNOW C REEK R 101 LELAND VALLEY RD E Legend D LI ND R D TARBOO LAKE RD EAGLEMOUNT RD PETERS O N RD Model Sub-Areas Port Townsend North Peninsula Mid-Peninsula South Peninsula Port Hadlock UGA Port Ludlow MPR OLD T ARBOO RD GIB BS LA KE RD 104 EGG & I RD WEST VALLEY RD SANDY SHORE RD DENA LN L ARSON LAKE RD SHINE RIDGE RD SWANSONVILLE RD TODD RD OLELE POINT RD WALKER WAY OLYMPUS BLVD LUDLO W BA Y RD HIGHLAND DR ANDY COO PER R D T EAL LAKE RD PARADISE BAY RD ALPINE CT DABO B R SHINE RD Exhibit 8 Model Sub-Areas Quimper Peninsula Model Documentation

14 Employment Forecasts Employment growth can be characterized by the following: Strong growth in Manufacturing, Wholesale, and Retail jobs, all above 100 percent total growth over 24 years with moderate growth in the other employment categories. The greatest increase in the density and number of new employees will likely occur in the southern portion of the City of Port Townsend along Sims Way (SR 20), Water Street, and along Discovery Road between Jacob Miller Road and Sheridan Street. The growth along Sims Way and Water Street will be concentrated mainly in the existing downtown core, the land around the southwestern portion of Discovery Road and Howard Street, and the southern end of Sheridan Street. A significant increase in Manufacturing employment is projected where the existing paper mill is located, on the vacant land between the Glen Cove light industrial/commercial area and the city boundary, and on the Jefferson County International Airport property in the Four Corners area. Only small amounts of employment growth are forecast for areas outside of Port Townsend, the Port Hadlock UGA, and the Port Ludlow MPR. External Growth Growth occurring external to the model was also accounted for by increased traffic on external connectors to the model. Growth on external connectors was determined with the help of historical growth data on state highways provided by the Washington State Department of Transportation (WSDOT). Ferry traffic was estimated to represent the peak off-loading capacity for the vessel most likely to be in service in the year 2031 as planned for by the Washington State Ferry System report for the Port Townsend-Keystone ferry route. Page 11

15 Trip Generation The trip generation process is used to convert the land use/socio-economic data into vehicle trips and estimate the number of trips per TAZ. The trip rate parameters can be adjusted to allow the model to better reflect existing conditions as part of the calibration/validation process. A simple trip generation methodology was applied to estimate the number of PM Peak Hour trips for each TAZ in the Quimper Peninsula Model. Trip generation rates were adjusted in an Excel spreadsheet for the 2007 base year and 2031 horizon year and were input directly into the VISUM model. Trip Rates Exhibit 9 contains the trip rates that were used in the 2007 and 2031 Quimper Peninsula Model. The trip rate calibration process began with the 2006 Port Townsend Model. Rates were created for SFDU, Retail, and Office land uses outside of Port Townsend to differentiate between travel patterns of people in a more urban area compared to people in a more rural area. Exhibit 9 Weekday PM Peak Hour Rates 1 Land Use Category Units Rates (Trips per unit) Percent Origins Percent Destinations SFDU Dwelling Units % 60 % CntySFDU Dwelling Units % 60 % MFDU Dwelling Units % 60 % RetireDU Dwelling Units % 50 % NRC Employees % 2 % Manuf Employees % 2 % CTU Employees % 2 % Whole Employees % 2 % Retail-High Employees % 79 % Retail-Low Employees % 79 % CntyRetail Employees % 70 % FIRE Employees % 2 % Edu Employees % 50 % Med Employees % 10 % Office/Other Employees % 2 % CntyOffice Employees % 2 % GOV Employees % 2 % Motel Rooms % 50 % 1. Trip rates based on Trip Generation, Seventh Edition, Institute of Transportation Engineers, Adjusted during model calibration to better replicate travel patterns and traffic counts within the study area. The trip generation process estimates origins and destinations within each TAZ. For a PM Peak Hour trip model, origins are normally associated with employment, since they correspond to where almost all PM Peak Hour trips begin. Destinations are normally associated with households because this is where PM Peak Hour trips are usually directed. The PM Peak Hour for the Quimper Peninsula Model falls between 4:15 pm and 5:15 pm which corresponds to the peak hour of a majority of the intersection counts used to calibrate the model. Most of the intersections that do not have a peak hour between 4:15 pm and 5:15 pm have a peak hour that starts within 15 minutes before or after 4:15 pm. Page 12

16 Trip Types Trips generated by each TAZ are also classified by a distinct trip type, or the general purpose of the trip. Within the Quimper Peninsula Model there are five primary trip types: Home to Work (HW): A vehicle trip that has an origin at a place of residence and a destination at a place of work. Home to Other (HO): A vehicle trip that has an origin at a place of residence and a destination at a non-work location such as a trip from a home to a restaurant or store for non-work purposes. Non Home-Based (NHB): A vehicle trip has neither an origin nor destination at a place of residence but may have an origin or destination at any other combination of land use types. Work to Home (WH): A vehicle trip that has an origin at a place of work and a destination at a place of residence. Other to Home (OH): A vehicle trip that has an origin at a non-work location and a destination at a place of residence. The trip rates used to estimate weekday PM Peak Hour vehicle trips by trip type are shown in Exhibit 10. Exhibit 10 Weekday Daily Trip Rates by Trip Type HW WH HO OH NHB Land Use Code 1 O 2 D 3 O D O D O D O D SFDU CntySFDU MFDU RetireDU NRC Manuf CTU Whole Retail-High Retail-Low CntyRetail FIRE Edu Med Office/Other CntyOffice GOV Motel See land use categories in Exhibit Rates for trip origins 3. Rates for trip destinations Page 13

17 Trip Balancing For each of the trip purposes, the total number of origins were balanced to the total number of destinations. The destinations within a PM Peak Hour model for the Home-Based Work and Home-Based Other trip purposes are generally linked to household information. Estimating the number of households is usually more accurate than estimating the number of jobs or employees. Subsequently, the trip generation values were balanced to the home end of the trip. External TAZ Trip Generation Trips for external TAZs (TAZs numbered ) were estimated from existing traffic counts and future forecasted traffic volumes. These trips were then separated into two groups: trips associated with other external TAZs, commonly classified as external-to-external (XX) trips; and trips associated with study area TAZs classified as internal-to-external (IX) and externalto-internal (XI) trips. Exhibit 11 lists the number of IX/XI and XX trips for the external TAZs in the Quimper Peninsula Model for the year 2007 and the year Exhibit and 2026 External TAZ Trip Generation External Internal Trips (IX/XI) External External Trips (XX) Zone Location X-I I-X From To Total Origins 1 Total Destinations 2 Average Annual Growth Rate External Trip Generation 700 Ferry Marrowstone Island Hood Canal Bridge Center Roadway Highway 101 South Entrance Highway 101 North Entrance External Trip Generation 700 Ferry Marrowstone Island % 702 Hood Canal Bridge , % 703 Center Roadway % 704 Highway 101 South Entrance % 705 Highway 101 North Entrance % 1. Origins = X-I + XX From 2. Destinations = I-X + XX To 3. The average annual growth rate represents the growth rate of the sum of the total origins and destinations between 2007 and No growth rate was applied to the ferry terminal as the vessel capacity anticipated for 2031 was used as the total origins and destinations allowed during the PM Peak Hour. All of the growth rates are based on WSDOT historical growth data. The Quimper Peninsula Model contains one XX trip type that accounts for external trips from the Port Townsend/Keystone ferry terminal to locations outside city limits. The total number of trips to and from the ferry was provided by WSDOT and accounts for the seasonal summer peak travel period. The percentage of origin and destination ferry trips that were XX in nature came from the Washington State Ferry 1999 Travel Survey Analysis and Results Report Table Page 14

18 Trip Distribution and Assignment The trip distribution and assignment of the Quimper Peninsula Model was performed by procedures inherent to VISUM and using parameters that were calibrated to represent conditions prevalent in Jefferson County. Trip Distribution The Quimper Peninsula Model applies the VISUM software s gravity model to distribute the PM Peak Hour trips between TAZs. The travel forecasting gravity model is built on Isaac Newton s theory that, all else being equal, the attraction between two masses will be proportional to the size of the masses and inversely proportional to the distance between the masses. In a travel forecasting model, the number of trips in a TAZ (for a trip purpose) is used to reflect the size of the mass, and a combination of travel time and distance is used to represent the distance factor in the gravity model. For trip distributions between external zones and stations an External to External trip table is applied during trip assignment. The trip distribution model contains parameters that adjust the relationship between travel time and distance based on trip purpose. The Quimper Peninsula Model uses the TModel Utility Function to calculate travel time using the coefficient values listed in Exhibit 12. During Trip Distribution, a separate trip table is constructed for each trip purpose. Exhibit 12 Trip Distribution Equation and Parameters TModel Parameters 1 Trip Type a b c Work-to-Home Home-to-Work Home-to-Other Other-to-Home Non-Home Based TModel Distribution Equation = 1/(U b + c U a ) Trip Assignment A trip assignment is performed using the trip matrices constructed as part of the trip generation and distribution phases. A multi-equilibrium assignment is used to assign the vehicle trips to the model roadway network. Three initial assignment iterations are used to load 33 percent, 33 percent, and 34 percent of the total vehicle trips onto the model roadway network in each incremental step, respectively. Termination of the iterations was also allowed if the maximum relative gap of 0.01 was met or a maximum of 20 network balancing iterations occured. Page 15

19 Validation and Calibration Results Before the model is used to estimate future traffic volumes and travel patterns, the trip distribution results are validated and trip assignments are calibrated based on existing data. The validation process involves comparing the trip distribution results of the study area to what is generally known of the area. The relative distribution of trips between zones is reviewed to verify that the model is distributing trips appropriately and that the overall lengths of trips are reasonable. The calibration process consists of comparing model travel assignments for the base year to actual traffic volumes and data on travel patterns. The calibration process consists of three primary parts: trip generation (discussed previously), screen line analysis, and link volume analysis. Screen Line Analysis Screen lines were defined to facilitate comparison of model volumes versus actual counts across several roadways serving similar travel patterns. Exhibit 13 summarizes the comparison of traffic counts to model volumes at each screen line including the percent difference between counts and model volumes. Exhibit 14 illustrates the five major screen lines created to calibrate the Quimper Peninsula Model to the 2007 base year. The allowable percentage deviation from actual counts depends on the volume of the actual traffic counts. The higher the actual traffic volumes, the lower the percent deviation allowed, per Highway Traffic Data for Urbanized Area Project Planning and Design (NHCRP 255, TRB, Dec pg. 49). The maximum allowable deviation for the Quimper Peninsula Model screen lines is 50 percent. Exhibit Quimper Peninsula Model Screen Lines Map ID Screen Line Location Traffic Counts Model Volumes Percent Difference IN (EB/NB) OUT (WB/SB) IN (EB/NB) OUT (WB/SB) IN (EB/NB) OUT (WB/SB) 1 West Port Townsend 969 1,417 1,050 1,350-8 % 5 % 2 McPherson Street 823 1, ,045-2 % 5 % 3 Sheridan Street 1,167 1,110 1,055 1, % -6 % 4 Hastings Avenue % 1 % 5 San Juan Avenue 1,231 1,287 1,095 1, % 6 % 6 Port Hadlock UGA 2,202 2,032 2,410 2,415-9 % -16 % 7 Four Corners 777 1, ,100-9 % -7 % Page 16

20 D OAK BAY RD Strait of Juan De Fuca OLD GARDINER RD RIDGE DR MCMINN RD CA PE GEORGE RD HASTINGS AV W LOFTUS RD SAND RD Discovery S DISCOVERY RD SR 20 OTTO ST IVY ST COOK AV M HASTINGS AV DISCOVERY RD ILL RD PARKRID GE DR ENGEL RD CUB RD TH ST 1 4 9T H ST SIMS W AY 2 ANDERS O N LAKE RD 14TH ST 9TH ST 25TH ST KALA POINT DR SAN JUAN AV PROSPECT AV KENNEDY RD 57TH ST 53RD ST 3 W ST FIR ST F ST E MOORE ST Port Townsend IRONDALE RD 6 D ST 5 CLAY ST CHIMACUM RD 4 20 BLAINE ST WATER ST Admiralty Inlet US Naval Reservation 116 WASHINGTO N LN Oak SCHWARTZ RD SR 116 EAST MARROWSTONE RD ROBBIN S RD W UNCAS R D VA N TROJ E N RD CENTER RD 19 N BAY WAY Puget Sound SNOW C REEK R 101 Legend LELAND VALLEY RD E D LI ND R D TARBOO LAKE RD EAGLEMOUNT RD PETERS O N RD Model Screen Lines 1 - West Port Townsend 2 - McPherson Street 3 - Sheridan Street 4 - Hastings Avenue 5 - San Juan Avenue 6 - Port Hadlock UGA 7 - Four Corners OLD T ARBOO RD GIB BS LA KE RD 104 EGG & I RD WEST VALLEY RD SANDY SHORE RD DENA LN L ARSON LAKE RD SHINE RIDGE RD SWANSONVILLE RD TODD RD OLELE POINT RD WALKER WAY OLYMPUS BLVD LUDLO W BA Y RD HIGHLAND DR ANDY COO PER R D T EAL LAKE RD PARADISE BAY RD ALPINE CT DABO B R SHINE RD Exhibit 14 Model Screen Lines Quimper Peninsula Model Documentation

21 All of the variances in volumes at city screen lines fall within the allowable deviation based on NHCRP 255. The largest percentage difference in the model screen lines is found in the Port Hadlock UGA screen line. Model volumes flowing out of the Port Hadlock UGA screen line are 16 percent lower than the actual traffic counts and the volumes flowing into the Port Hadlock UGA screen line are 9 percent lower than the traffic counts. The percent differences are very reasonable for Port Hadlock UGA where the primary trip generator is residential dwelling units. It is also important to note that because the traffic volumes in the model are so low each percent difference represents tens of vehicles out of thousands and the relative impact that would be noticed in real world conditions is not as large as it would be if the traffic volumes were in the high thousands or tens of thousands. Therefore, the degree of accuracy of the Quimper Peninsula Model is well within acceptable limits. However, for detailed intersection capacity analysis, additional adjustments will be made to account for these differences. Generally, the screen line analysis indicates that overall traffic generation and travel distribution patterns reasonably reflect existing conditions in the study area. Link Volume Analysis Another measure of model calibration is the analysis of roadway link volumes. This process compares roadway model volume outputs to actual traffic counts, by direction, for all locations where actual counts are provided. Exhibit 15 contains the output of key statistical analysis parameters for the calibrated model including the coefficient of determination, or R 2 value, and the percent root mean square error (%RMSE). Exhibit 15 Statistical Analysis of Calibrated Data R 2 %RMSE The R 2 value represents the goodness of fit, indicating how well the model output matches the actual traffic counts. The R 2 value also represents the likelihood that the correlation between the traffic counts and the model output could have occurred randomly or by chance. The closer the R 2 value is to 1.00 the less likely it is that the correlation between data sets occurred randomly and the more likely it is that relationship between the data is strong. An R 2 value of 0.88 or higher is considered acceptable in the industry according to Federal Highway Administration (FHWA) guidelines. The overall R 2 value for the Quimper Peninsula Model is The %RMSE measures the deviation between the model volumes and the actual traffic counts. The greater the %RMSE, the greater the deviation between the model volumes and the actual traffic counts. It is recommended that the %RMSE value for a model be 40 percent or less. The %RMSE of the Quimper Peninsula Model is approximately 23 percent. Page 18

22 2031 Baseline Model After the calibration process was completed and determined to be within acceptable parameters, the model was used to develop a baseline 2031 traffic forecast. The baseline model was used to determine where future operational and capacity deficiencies were likely to occur based on the assumed land use growth. Exhibit 16 lists the model network changes that represent the 2031 Baseline condition. All of the 2031 Baseline improvements were made in the City of Port Townsend based on Port Townsend s Draft Transportation Functional Plan. WSDOT or Jefferson County does not have significant capacity improvements planned outside of Port Townsend at this point. Relevant and available Hood Canal Bridge and Washington State Ferry studies were also reviewed for potential traffic impacts that would affect the Quimper Peninsula Model and will continue to be considered as future alternative analyses are conducted. Exhibit Baseline Model Network Changes 1 Project Limits Description Howard Street Sims Way to 35 th Street Extend Howard Street from Sims Way to 35 th Street. The new roadway will be classified as an Urban Minor street. Umatilla Avenue Howard Street / Sims Way McPherson Street / Sims Way Thomas Street / Sims Way Sheridan Street / Sims Way Howard Street / Discovery Road Sheridan Street / Discovery Road Kearney Street / Blaine Street Discovery Road Hastings Avenue 1. Based on input from City staff. Existing terminus to 35 th Street Intersection Intersection Intersection Intersection Intersection Intersection Intersection Sims Way to Sheridan Street Howard Street to Sheridan Street Construct Urban Collector between the existing western terminus of Umatilla Avenue and the eastern terminus of 35 th Street. Construct roundabout at the intersection of Howard Street and Sims Way as part of the extension of Howard Street. Limit access to right-in and right-out by prohibiting left-turns. Construct roundabout at the intersection of Thomas Street and Sims Way. Install traffic signal at the intersection of Sheridan Street and Sims Way. Construct roundabout at the intersection of Howard Street and Discovery Road as part of the extension of Howard Street. Install traffic signal at the intersection of Sheridan Street and Discovery Road. Install All-Way Stop at the intersection of Kearney Street and Blaine Street. Reclassify Discovery Road from Rural Minor to Urban Minor. Reclassify Hastings Avenue from Rural Minor to Urban Minor. Baseline Model Results The completion of the Howard Street corridor and the traffic control improvements along Upper Sims Way, at Howard, McPherson, and Thomas, will provide the necessary circulation for all of the forecast land use increases along Discovery Road. Appendix A contains model plots showing the volume-to-capacity (V/C) ratios and model volume bandwidths for both of the years 2007 and A comparison of the volume bandwidths and V/C ratios plots illustrates the increase in traffic that is expected on SR 20 and SR 19 that will require additional intersection and roadway improvements to be completed by The section of state highway that is anticipated to require access management and intersection mitigation is the section of SR 19/SR 20 from Four Corners Road to Sheridan Page 19

23 Street. The V/C ratio plot from the 2031 model shows that the volume demand will most likely exceed capacity if no improvements are made and all else, such as travel behavior, remains equal. The interchange of trips between the Port Hadlock UGA and Port Townsend is one of the primary trip interchanges in the model. The increase in volumes between Four Corners Road and Sheridan Street is primarily a result of the housing and employment growth both in the Port Hadlock UGA and in the southwestern portion of Port Townsend. Although most of the other roads in the model network will experience some increase in traffic volumes by the year 2031, few improvements will be required as most of the roadways are not nor will be near capacity in the future. Additional intersection improvements may be needed by 2031 and will be identified in a forthcoming traffic operations study. Page 20