Appendix 3 CUUATS Transportation Model Report

Transcription

1 Appendix 3 CUUATS Transportation Model Report

2

3 TRANSPORTATION MODEL LONG RANGE TRANSPORTATION PLAN 2025 Champaign-Urbana Urbanized Area Transportation Study (CUUATS)

4 TABLE OF CONTENTS I. INTRODUCTION...1 II. DATA DEVELOPMENT TRAFFIC ANALYSIS ZONE (TAZ), CENTROIDS, AND EXTERNAL STATIONS NETWORK CODING...2 1) Highway Network...7 2) Transit Network SOCIOECONOMIC DATA III. DEFINITIONS FOR TRIPS TYPES OF TRIPS TRIP PURPOSES IV. TRIP GENERATION OVERVIEW TRIP PRODUCTION AND ATTRACTION ANALYSIS Trip Production Trip Attraction EXTERNAL TRIPS ANALYSIS ) Roadside origin-destination survey ) External Trips BALANCING V. TRIP DISTRIBUTION OVERVIEW ESTIMATION OF IMPEDANCES: TRAVEL TIMES ESTIMATION OF FRICTION FACTOR USE OF GRAVITY MODEL VI. MODE CHOICE MODE SPLIT PROCEDURES CREATION OF AN ORIGIN-DESTINATION TRIP TABLE VII. TRIP ASSIGNMENT HIGHWAY TRIP ASSIGNMENT ) Methodology i

5 2) Parameters TRANSIT TRIP ASSIGNMENT VIII. TIME-OF-DAY ANALYSIS IX. MODEL VALIDATION PLOT CHECK VOLUME CHECK ) % Root Mean Square Error ) Average Volume Ratio VEHICLE MILES TRAVELED CHECK X. REFERENCE XI. APPENDICES APPENDIX A: HOUSEHOLD SIZE BY TAZ APPENDIX B: EMPLOYMENT BY TAZ APPENDIX C: UNBALANCED TRIP PRODUCTIONS AND ATTRACTIONS APPENDIX D: BALANCED TRIP PRODUCTIONS AND ATTRACTIONS ii

6 LIST OF FIGURES Figure 1. The Structure of the Four-Step Travel Demand Model...1 Figure 2. Procedures to build transit network...9 Figure 3. Procedures for building transit skim file Figure 4. Trip Purposes Figure 5. Trip Generation Process Figure 6. Trip Distribution Process Figure 7. Process to Estimate Friction Factors from the Survey List of Tables Table 1. Node Attributes...7 Table 2. Link Attributes...7 Table 3. Link Speed/Capacity by functional classification and area type...8 Table 4. Link data (Type 1 Data Card)...9 Table 5. Speed Curve Use Data (Type 2 Data Card) Table 6. Speed Curve Data (Type 3 Data Card) Table 7. Speed curve data used by Champaign-Urbana Urbanized Area Table 8. Line numbers used in TROUTE.TEM Table 9. Employment Category Table 10. Summary of the Roadside Origin-Destination Survey Table 11. Percent Residency and Trip Purposes for CUUATS Table 12. External Trip Productions and Attractions (Person Trips) Table 13. Trip Generation Summary Table 13. Typical Terminal Times for Different Area Types Table 14. Calibrated Friction Factors Table 15. Production/Attraction code and Area types used in mode choice model Table 16. Volumes by Split Curves Table 17. BPR curve coefficients for Link Group Table 18. Percentage of peak periods by actual traffic counts done in 2002 (1/09/04) Table 19. Peak hour traffic volumes for Champaign-Urbana Urbanized Area Table 20. VMT per household and per capita Table 21. VMT by Road Classification LIST OF MAPS Map 1. Study Area Boundary for LRTP Map 2. Traffic Analysis Zones for LRTP Map 3. Highway Network for LRTP Map 4. Transit Network for LRTP Map 5. Cordon Count Locations Map 6. Highway Network iii

7 I. Introduction The purpose of this document is to outline the main processes of creating a transportation model (or travel demand model ). The model will be used as an analysis tool to evaluate existing and future network conditions as part of the Long Range Transportation Plan This document is divided into four chapters describing data requirements, definitions that CUUATS staff has developed for model inputs, trip generation process and the trip distribution process. The model uses a four-step travel demand model to estimate the traffic volumes, which is shown in Figure 1. Currently, the CUUATS travel demand model utilizes TRANPLAN version 9.0 and VIPER version 3.0 by CITILAB. The four-step travel demand model: Trip Generation: estimate how many trips are made by each household for each of the trip purposes (work, shopping, etc.) and how many trips are attracted to each location (work places, shopping centers, etc.). Trip Distribution: estimate how many trips go from one location to all other locations. Mode Choice: given that someone will travel from one location to another, compare the mode options and choose which mode the traveler would likely use. Trip Assignment: route the travel between zones onto public transportation services and roadways. Socioeconomic Data Trip Generation Trip Rates Highway Network Trip Distribution Friction Factor Transit Network Mode Choice Utility Coefficient Trip Assignment BPR Coefficients Time-Of-Day Analysis Peak-time Percentage Traffic Volume Figure 1. The Structure of the Four-Step Travel Demand Model 1

8 II. Data Development 1. Traffic Analysis Zone (TAZ), Centroids, and External Stations The study area for the travel demand model encompasses the existing Champaign-Urbana Urbanized Area (not including Bondville). Traffic analysis zones (TAZs), centroids, and external stations were defined based on the study area. Map 1 shows the study area, and Map 2 shows traffic analysis zones, centroids, and external stations. The study area has 147 TAZs, 147 centroids, and 20 external stations. Traffic analysis zones are the geographical units for the travel demand model. Major land uses are defined for each TAZ. It is assumed that all travel activities and characteristics are homogeneous within each TAZ. When defining TAZs, several factors should be considered 1 : Geometric shape of zones Geographic, physical, and political boundaries Census boundaries Arterial roadways should not bisect a TAZ Relatively similar land use in a zone Centroids are the center of activities within a traffic analysis zone; the centroid is not necessarily physically centered in the TAZ. Centroids represent the origins and destinations of travel activity within each zone 2. These are determined based on aerial photos and knowledge of the local situations. Centroid connectors connect centroids to the nearby road network. These connectors represent all local residential streets that are not included in the highway network. Ideally, a centroid should be connected by at least four centroid connectors. External stations are the points that represent outside traffic entering, exiting, or passing through the study area. The study area has 20 external stations on the edge of the urbanized area. Roadside interview surveys were done at the stations in April 2003 to collect data about trip behaviors. 2. Network Coding The most important aspect of the transportation model is to build an accurate network for each mode representing the transportation system for the Champaign-Urbana urbanized area. CUUATS has developed a highway network for automobile users and a transit network for transit users. All the street characteristics were coded in geographic information system (GIS) format and StreetMap USA was used as a base map. Then, the base map was updated based on the survey file and aerial photos taken in Map 3 and Map 4 show each of output for the highway network and the transit network displayed in VIPER 1 Alvin Whyte (2000). Travel Demand Forecasting Manual 2: Highway Network Coding Procedures. Ohio DOT. 2 Martin, A. William, and Nancy A. Mcguckin (1998). Report 365: Travel Estimation Techniques for Urban Planning. National Academy Press. Washington D.C. 2

9 Map 1. Study Area Boundary for LRTP

10 Map 2. Traffic Analysis Zones for LRTP

11 Map 3. Highway Network For LRTP Miles 5

12 Miles Map 4. Transit Network For LRTP 2025 Legend Miles Transit Route Link Centroid Connector Interstate Highway Highway Ramp 6

13 1) Highway Network Highway network consists of nodes and links. Following sections describe each of highway network component. Once database were prepared, highway network can be built by the Tranplan function Build Highway Network. For highway network impedance coding, refer the first section of Trip Distribution. Highway node Nodes are the representation of street intersections and any changes in the transportation system. For example, if the street reduces from two lanes to one lane, the point where the number of lanes changes is the location of a node. The Champaign-Urbana Urbanized Area Highway Network has total 1,009 nodes uses State Plane Coordinate System (Illinois East). Node data is made in ASCII format and attributes for the node database are shown at Table 1. Table 1. Node Attributes Field Type Columns Field Name Description C 1 ID Record Identifier, always as N N 2-6 (5) N Node Number, Centroids, External Stations, Highway nodes, Node only for transit (2) - Not Used N 9-17 (9) X X-Coordinate, two digit decimal places (2) - N (9) Y Y-Coordinate, two digit decimal places C Comments Comments Highway link Highway links are connections between any two nodes in the transportation system. Links contain the important characteristics of a roadway such as functional classifications, distances, capacities, area types, and speed. Champaign-Urbana Urbanized Area has 2,886 links, 7 functional classifications, and 3 link group data. Data format for link is shown at Table 2 and detailed information for coding capacity is followed in Table 3. In Table 3 the capacity for the connector is assumed to be a , this is because the connector is a representation of the local streets inside the TAZ and may not represent an actual roadway. As a result, the model will not show any congestion in the connectors. Table 2. Link Attributes Field Type Columns Field Name Description N 1-5 (5) ANODE A node number which identifies the from node of the link N 6-10 (5) BNODE B node number which identifies the to node of the link N 11 (1) ASSIGNMENT Functional Classification of roadways GROUP 1 = Major Arterial, 2 = Minor Arterial, 3 = Collector 7

14 4 = Local, 5 = Connector, 6 = Interstate Highway, 7 = Ramp N (4) DIST Actual node to node distance (Hundredths of miles) C 16 (1) OPTCODE Time or speed, always as S N (4) SPEED1 Operation speed (Miles per hour, decimal point between columns 18 and 19) N (4) SPEED2 Posted speed (Miles per hour, decimal point between columns 18 and 19) N (2) DIRCODE Direction Code, here coded as 1 N (2) AREATYPE N (2) LANES Number of lanes (1 to 3) 1 = CBD, 2 = Fringe, 3 = Residential, 4 = OBD (Other Business Area such as Market Place Mall, North Prospect, U of I, and Parkland College), 5 = Rural N (2) Facility Type 1-17 by functional classification for BPR parameters. Refer the BPR parameters section of the trip assignment. N (6) CAPACITY 24-hour capacity. Refer the Table 3. N (6) VOLUME Average Annual Daily Traffic from traffic count N 45 (1) OPTION 1 = ignore all data in columns Table 3. Link Speed/Capacity by functional classification and area type Area Type Facility Type (ASSIGN. GROUP) CBD Fringe Residential OBD Rural 1 Major Divided Arterial* Undivided Minor Arterial* (25 mph) (30-40 mph) 3 Collector* Local* Connector Interstate Highway Ramp * Per lane capacity for 24hours 2) Transit Network Transit network uses a highway network as a base. Nodes and links of highway system are shared with a transit network and speed for transit is computed as a function of auto travel speed on highway network. The databases that represent transit network were made in the INET program. The coding procedures using INET program are shown on the Figure 2. First step is to generate input file into Tranplan function and then build the transit network using Tranplan. INET requires three input databases: TSYSIN.TEM, TROUTE.TEM, and HNET.TEM and Tranplan needs two inputs: HUDNET.TEM and additional nodes if needed. Following sections explain about transit travel impedance. 8

15 TSYSIN.TEM TROUTE.TEM HNET.TEM INET HUDNET.TEM Build Transit Network Additional Nodes MTD.NET Figure 2. Procedures to build transit network Database o TSYSIN.TEM TSYSIN consists of header in the first record, control file, and three data cards. Data card 1 contains links not in the highway network, Data card 2 includes speed curve use data and Data card 3 gives the speed curve to be used for each mode and roadway combinations. Nodes in transit network share the same node system in the highway network. Additional nodes needed for the transit routes are added as a separate database in the process of running Tranplan. Following tables show the format of each data card. All the data were made with two-digit format. Table 4. Link data (Type 1 Data Card) Field Data Column Field name Length Type Description 1 1 CARD TYPE I1 Card Type = ANODE I5 ANODE BNODE I5 BNODE 1 12 Unused MODE I2 Walk = 1, MTD = Add. Mode 4I2 Leave the columns blank 9

16 DIST F5.0 Distance From A to B in miles Eg.0.5 mile => 0.5, 0.02 mile => coded as SPEED F5.0 Speed from A to B in miles per hour, coded 5 mile/hr less than highway speeds TIME F5.0 Time from A to B in minutes Unused Unused OPTION I1 Type 2 since speed, distance, and time are identical between Node A and Node B Blank 1 66 Fare Code I1 Blank (1 7) Unused Area Type I Facility Type I1 1-7 Table 5. Speed Curve Use Data (Type 2 Data Card) Field Data Column Field name Length Type Description 1 1 Card Type I1 Card Type = Low Mode I2 Low Mode High Mode I2 High Mode 1 6 Unused Low Area Type I1 Low area type from High Area Type I1 High area type from Unused Low Facility Type I1 Low facility type from High Facility Type I1 High facility type from Unused Curve Number I Unused Repeat Repeat Repeat Repeat Repeat Repeat 2-16 Note: For Mode 4, curve number 11 was assigned for facility type 8-9. Curve number from speed curve data (1-15) Table 6. Speed Curve Data (Type 3 Data Card) Field Length Column Field name Data Type Description 1 1 Card Type I1 Card Type = Unused Curve Number I2 Curve Number from 1 to Unused Low Hwy F5.0 Low auto speed value (X1) 1 11 Unused Low Transit F5.0 Low transit speed value (Y1) 1 17 Unused High Hwy F5.0 High auto speed value (X2) 10

17 1 23 Unused High Transit F5.0 High transit speed value (Y2) Champaign-Urbana Urbanized Area has three different area types and 7 facility types and grouped these by similar speed range. Table 7 shows the 15 speed curve data by area type and facility type. Transit speed is fewer by 5 miles per hour than that of highway speeds. Table 7. Speed curve data used by Champaign-Urbana Urbanized Area CBD Fringe OBD Residential Rural Major Arterial Curve No 1 (20, 15) Curve No 4 (20, 15) Curve No 7 (15, 13) Curve No 10 (20, 18) Curve No 13 (25, 23) Minor Arterial Collector Local Connector (30, 20) Curve No 2 (15, 10) (25, 20) Curve No 3 (15, 10) (20, 15) (45, 40) Curve No 5 (15, 10) (30, 25) Curve No 6 (15, 13) (35, 30) (45, 40) Curve No 8 (15, 11) (35, 30) Curve No 9 (15, 10) (35, 30) *Curve number 14 is for minor arterial, collector, local, and ramp. ** Curve number 15 is for highway. (45, 40) Curve No 11 (15, 10) (40, 35) Curve No 12 (20, 15) (35, 30) (45, 40) Curve No 14* (20, 16) (45, 40) Curve No 15** (45, 40) (55, 50) o TROUTE.TEM TROUTE contains route records in INET format and includes mode, line number, specific line name, headway, and stops of each transit route. One line with different directions was considered as separate lines and had two separate line numbers. Here are the samples of route file and the following table indicates the line numbers used in this TROUTE.TEM. &ROUTE M=4, L=1, ID='1YELLOW NB', RG=1, H=30.0, N=-1468,-1771,-2000,-1467,-2001, -17,-1016, 16,-1017,-1415,-1798, -1414, 2002,-2003,-2004,-1789,-1747,-1404,-1768,-1405,-1767,-1371, -1375,-2005,-1863,-1369,-1372,-1838,-1351,-1839,-1333,-2006, C=1 &END Table 8. Line numbers used in TROUTE.TEM Line Name Line ID Line Line Line Name Line ID Number Number 1 YELLOW North Bound 1 9A BROWN 9A BROWN 9 South Bound 19 9B BROWN 9B BROWN 99 2 RED North Bound 2 West Bound GOLD South Bound 29 East Bound North Bound 3 North Bound SILVER LAVENDER South Bound 39 South Bound BLUE West Bound 4 West Bound SCAMP East Bound 49 East Bound GREEN West Bound 5 East Bound LOOP East Bound 59 West Bound ORANGE West Bound 6 21 QUAD 21 QUAD 21 11

18 East Bound ILLINI 22 ILLINI FARTOISR 22 West Bound 7 22 ILLINI ISRTOFAR GREY East Bound SHUTTLE WEST 23 SHUTTLE WEST 23 8 OCHARD West Bound 8 23 SHUTTLE EAST 23 SHUTTLE EAST 123 DOWN East Bound PACK 26 PACK 26 o HUDNET.TEM This file is an output file generated from the INET Program and one of the inputs for building transit network. This contains link, line, and node data. Transit Travel Impedances In Tranplan, procedures to generate zone-to-zone travel impedances have two steps. First step is to create a minimum path file of transit travel that builds shortest paths from one zone to all other zones. Then the next step is to build a transit skim file that contains travel impedances from a minimum path file. Input to these functions are a transit network and several parameters and the output is a matrix with zone-to-zone travel impedances. The figure 3 shows the process of creating transit skim file. Transit Network Build Transit Paths Transit Paths Transit Selected Summation Transit Skims Figure 3. Procedures for building transit skim file Regarding the first step, minimum path is the summation of transit travel time on the links and wait times such as waiting time and transferring time penalties between non-transit modes and transit modes or between different transit modes. Transfer time penalties were one-half of headways and set as a default by the Tranplan. However, minimum and maximum of transfer time penalties can replace the calculated time when the last is more than maximum time set in the parameter section of control file or less than the minimum transfer time. From non-transit to 12

19 transit, maximum wait penalty is set as 10 minutes and minimum is 2 minutes. For the transfer between transit modes, maximum wait penalty is 20 minutes and minimum is 2 minutes. A transit skim file can be built using a minimum path file created in the transit path building function. Final output is a zone-to-zone matrices with 5 tables selected in the data section in the control file: walk access, total wait time, transfers, mode 4 time, and total time. 3. Socioeconomic Data The socioeconomic data is one of the major inputs for the travel demand model. The data include household and employment information aggregated by each traffic analysis zone. Census block data is a source for household data. After identifying the household data by each TAZ, the data is disaggregated by household size in groups of one person, two persons, three persons, and four or more persons. See Appendix A and B for household and employment data by each TAZ. Employment data is one of the most difficult inputs to collect and modify since each employer in the study area must be identified by their location, number of employees, and industrial classification. The Champaign County Chamber of Commerce, Illinois Department of Employment Security (IDES), the University of Illinois, and Urbana School District provided the data. CUUATS staff geocoded all employer locations in GIS. The data were then aggregated by TAZ and employment category (See Table 9). Table 9. Employment Category Category Description Industrial Two digit Standard Industry Classification Code Retail Two digit Standard Industry Classification Code Service Two digit Standard Industry Classification Code 60-81, Education Two digit Standard Industry Classification Code 82 Government Two digit Standard Industry Classification Code 91 13

20 III. Definitions for Trips 1. Types Of Trips Trips may be defined as a non-stop itinerary that starts in one place and ends in another place 3. There is a single origin and a single destination. Trips can be further divided by trip makers and origin locations. Person trip: a movement from one address to another by one person by any mode Vehicle trip: a movement by a private vehicle from one address to another for a purpose regardless of the number of the people in the vehicle Internal-internal Trips: trips within the study area Internal-External / External-Internal Trips: trips entering or exiting the study area. These trips are referred to as Internal-External Trips External-External Trips: trips passing through the study area 2. Trip Purposes A trip purpose is the main reason that motivates a trip 4. CUUATS defined five different trip purposes for the travel demand model: Home-based Work, Home-based School, Home-based Shopping, Home-based Other, and Non-home-based. Figure 4 shows the simplified trip definitions. Non-home trips Other Work Home to Work Non-home trips Home Home to Work Work Home to Shopping Shopping Non-home trips Figure 4. Trip Purposes 3 Thurston County Regional Planning Commission. 4 Martin, A. William, and Nancy A. Mcguckin (1998). Report 365: Travel Estimation Techniques for Urban Planning. National Academy Press. Washington D.C. 14

21 Home-Based Work (HBW): This category includes To/From Work or Work-Related Business trips. Home-Based School (HBShc) - This category includes trips to school, college or university for classes, or to school-related meetings. Home-Based Shopping (HBSho) One end of trip is shopping activities. Home-Based Other (HBO): This category includes family and personal business trips such as banking, haircuts, visiting friends and relatives, other Social or Recreational trips taken for entertainment and recreation, and for trips that do not fit any of the other categories. Non-home Based trips This category includes trips that do not start or end at home. 15

22 IV. Trip Generation 1. Overview Trip generation is a process that estimates the amount of trips made to and from each TAZ on a daily basis. CUUATS trip generation model is based on the cross-classification method because this is currently considered to be better than regression model in its ability to handle non-linear trip variables. This method was also used because it can integrate local trip rates from the household travel survey conducted in This model makes use of disaggregate socioeconomic data such as the household by family size classification to determine the amount of travel generated in the region 5. Figure 5 shows the input and output of this process. The final output through this process would be a set of balanced trip production and attraction rates for each TAZ and external station. Major input data sources are Census 2000, Champaign County Chamber of Commerce, and the 2002 CUUATS Household Travel Survey. As a result, the Champaign-Urbana urbanized area generated 566,700 person trips per day. Household Data Trip Production Production Rate Employment Data Trip Attraction Attraction Rate ADT External Trip Through trip Rate Balancing Balancing Factor Balanced Trip P & A Figure 5. Trip Generation Process 5 Martin, A. William, and Nancy A. Mcguckin (1998). Report 365: Travel Estimation Techniques for Urban Planning. National Academy Press. Washington D.C. 16

23 2. Trip Production and Attraction Analysis Trip production analysis is used to estimate the amount of trips produced by each TAZ. It is assumed that households in each TAZ produce the trips. Trip production equations by trip purposes were developed from the CUUATS household survey as shown below. 1. Trip Production 6 HBWork = 0.16*6.1(X1) *9.8(X2) *11.4(X3) *12.7(X4) HBSchool = 0.15*6.1(X1) *9.8(X2) *11.4(X3) *12.7(X4) HBShopping = 0.08*6.1(X1) *9.8(X2) *11.4(X3) *12.7(X4) HBOther = 0.27*6.1(X1) *9.8(X2) *11.4(X3) *12.7(X4) NHB = 0.34*6.1(X1) *9.8(X2) *11.4(X3) *12.7(X4) X1 = Total number of 1-person households X2 = Total number of 2-person households X3 = Total number of 3-person households X4 = Total number of 4+person households Inversely, trip attraction analysis determines the trips attracted to each TAZ. Employers are the major source of attracting trips. The trip attraction rates for the model were borrowed and modified from the NCHRP Report The recommended NCHRP trip attraction rates are reasonable to use since the trip generation model is highly dependent on trip production analysis. 2. Trip Attraction 8 HBWork = 1.45 * (X1) OTHER = 9.0 * (X3) * (X4) * (X5) * (X6) HBSchool = 0.14 * [OTHER] HBShopping = 0.23 * [OTHER] HBOther = 0.64 * [OTHER] NHB = 4.1 * (X3) * (X4) * (X5) * (X6) X1 = Total employees in Total EMP for a TAZ X2 = Total employees in BASIC for a TAZ X3 = Total employees in RETAIL for a TAZ X4 = Total employees in SERVICE for a TAZ X5 = Total employees in OTHEREMP for a TAZ X6 = Total household for a TAZ 3. External Trips Analysis The purpose of external trip analysis is to find the number of external trip productions and attractions at external stations, which are the allocation of the observed traffic volumes crossing 6 Trip production rates are drawn from CUUATS Household Travel Survey Establishment Survey needs to be conducted to draw the trip attraction rates since Household Travel Survey is too small to provide them. 8 Trip attraction rates are based on the NCHRP 365 Travel Estimation Techniques for Urban Planning. 17

24 the regional study area using the estimated percentage of through trips, percentage of trip purposes and residency, and vehicle occupancy rates by each trip purpose. Two major data sources are the 24-hour average daily traffic (ADT) volume and the information on trip characteristics such as percent through trips, trip purpose, and residency. The total number of external stations in Champaign-Urbana-Savoy Area is 20 and the total ADT was 64,295. Cordon counts were done in Refer to the Map 5 to see the location and its ADT. 1) Roadside origin-destination survey CUUATS staff conducted a roadside origin-destination survey in April Five questions on the person s trip origin and purpose, destination and purpose, and place of residency were asked. Since a small sample of locations was surveyed among the 20 external stations, average numbers for percent through, percent trip purposes, and percent residency were used. The following table summarizes the roadside interview surveys. Table 10. Summary of the Roadside Origin-Destination Survey Item Percent Percent Through 9 % External trips (I-E or E-I) 91 % Percent Truck 4 % Percent Vans and Pickups 35 % Percent Residency 30 % Percent Non-Residency 70 % Percent Home-based work trips 57 % Percent Home-based school trips 6 % Percent Home-based shopping trips 9 % Percent Home-based other trips 21 % Percent Non-home-based trips 7 % Source: CUUATS Roadside Origin-Destination Survey, April ) External Trips Trips at external stations consist of through trips and external trips. Through trips are defined as those trips that begin and end outside the study area. External trips are those trips that either start or end outside the study area. External trips are calculated by subtracting through trips from the ADT. The percentage of through trips differs between areas depending on the functional classification of the roadway and population. The average percentage of through trips from the CUUATS roadside survey is 9%. Through trip = Percentage of through trips * ADT. External trip = ADT Through trips Application of trip purposes into external trips After finding the volume of external trips, these trips are allocated by trip purpose, based on the percentages of trip purposes. 18

25 Map 5. Cordon Count Locations 19

26 Table 11. Percent Residency and Trip Purposes from CUUATS Roadside Survey Trip Purpose Resident Non-resident 9 Total Home-based work 13 % 44 % 57 % Home-based school 4 % 2 % 6 % Home-based shopping 2 % 7 % 9 % Home-based other 7 % 14 % 21 % Non-home-based 4 % 3 % 7 % Total 30 % 70 % 100 % Source: CUUATS Roadside Origin-Destination Survey, Estimation of trip production and attraction In case of home based trips, trip production and trip attraction are estimated by residency. Persons who live outside of the study area produce the trip at the external station and persons who live inside the study area attract the trips at the external stations. 30% residency and 70% nonresidency percentages were applied to the allocated trip purposes based on the survey results. Estimated external trips are vehicle trips, whereas internal trips are person trips. Consistency must be achieved between the two in order to accurately illustrate production and attraction. A person trip was converted into vehicle trips by dividing the number of vehicle occupants by the number of vehicle trips. The average vehicle occupancy rate was 1.3 persons per vehicle. The table below shows the final results. A total of 67,844 trips were produced and 28,955 were attracted. Table 12. External Trip Productions and Attractions (Person Trips) Station Number Production Attraction HBW HBSch HBSho HBO NHB Total HBW HBSch HBSho HBO NHB Total , , , , , , , , , , , , , , , , , , , , , ,150 2, ,134 1, , , , , , , , , , , ,300 1, , , Total 29,700 1,505 6,267 13,904 3,840 55,216 8,966 2,006 1,393 7,361 3,840 23,566 9 Non-resident are the persons who live outside of the study area meaning the trip production at the external station. 20

27 4. Balancing Trip productions and attractions should be equal since each trip must have two trip ends: a production and an attraction. However, they are often different since trip productions and attractions are estimated separately. Through this last step in trip generation analysis, the productions and attractions can be balanced. As trip production is more reliable than the trip attraction, it was used to estimate the regional control total of trips, which totaled 566,700. External trips are a function of observed traffic volume so they cannot be changed. Trip production is the control total, which cannot be changed to balance production and attraction. Thus the only factor that can be used to make the total production and attraction the same is trip attraction in TAZs. The balancing factor is calculated from the equation below. The factor is then applied into the internal trip attractions by trip purposes in study area. Production Attraction TAZs Pz Az External Pe Ae Total P A Production = Attraction {Pz + Pe} = {(Az * x) + Ae} Balancing factor x = (Pz+Pe Ae) / Az The calculated balancing factors are listed below for each trip purpose. Based on the balancing factors, the Champaign-Urbana urbanized area generates a total of 566,700 trips per weekday as shown in Table 13. Home-based work trips = 0.96 Home-based school trips = 1.13 Home-based shopping trips = 1.29 Home-based other trips = 1.26 Non-home-based trips = 1.16 Total trips = 1.15 Table 13. Trip Generation Summary Purpose Trips Generation Internal TAZs External Stations Home-based Work 117,202 Production 87,502 29,700 Attraction 108,236 8,966 Home-based School 37,981 Production 36,476 1,505 Attraction 35,975 2,006 Home-based Shopping 68,670 Production 62,404 6,267 Attraction 67,278 1,393 Home-based Other 188,357 Production 174,453 13,904 Attraction 180,996 7,361 Non-home-based 154,489 Production 150,650 3,840 Attraction 150,650 3,840 Total 566,700 Production 511,484 55,216 Attraction 543,134 23,566 21

28 V. Trip Distribution 1. Overview Trip distribution is the estimation of how many trips go from one TAZ to all other TAZs. This process uses trip generation output data as an input and converts it into matrices of person trip tables that represent the movement between TAZs. The most common methodology for trip distribution is the gravity model, where the number of trip exchanges between two zones is directly related to land use patterns or activities and inversely related to separation between two locations. The trip distribution process has two major steps: 1) Estimation of friction factors based on travel times, and 2) Distribution of trip generation using gravity model. These processes can be implemented using a transportation planning software package such as TRANPLAN, EMME2, or TRANSCAD. The basic model structure is standard but the parameters and measures for separation such as time, cost, or combination of time and cost are different between regions. The next part of this documentation follows the procedural order that CUUATS underwent while implementing trip distribution. Highway Network Step 1 Build Travel Time Step 2 Estimate Friction Factors Step 3 Use of Gravity Model Production Attraction Person Trip Table Figure 6. Trip Distribution Process 22

29 2. Estimation of Impedances: Travel Times Travel impedance may be defined by the path of least resistance between each pair of zones 10. Travel time, distance, etc. are summed for the links between each zone pair and the results are stored in a zone-to-zone travel impedance matrix. TRANPLAN calculates travel impedance. The steps to estimate impedance are as follows: Estimation of free-flow interzonal travel times: Free-flow travel times are calculated by using link lengths and speeds. Link length is identified as a part of network data and link speed is estimated by a posted speed for each link. Once the length and speed were defined, shortest time path and travel impedance between zones can be calculated. Estimation of intrazonal travel times: Many of intrazonal travel take place on the local street network that is not coded; intrazonal travel times was estimated by the nearest neighborhood technique: Nearest neighborhood technique: Average interzonal travel times to the nearest adjacent zones Intrazonal travel time = 2 Terminal times: Terminal times is a representation of impedances at both ends of a trip required to walk to and from a transit mode, to park or access a parked car, and so forth. Terminal times are different for area types. Modified NCHRP terminal times were used in CUUATS model. Table 13. Typical Terminal Times for Different Area Types Area Type Terminal Time (minutes) CBD 3 CBD Fringe 2 Other Business District 2 Residential 1 Rural 1 3. Estimation of Friction Factor Friction factors or f-factors are inputs into the gravity model along with trip productions and attractions. The friction factor quantifies the impedance and is inversely related to the impedance (as the travel time increases, the friction factor decreases). There are several methods to calculate friction factors: 1) generated from survey data, 2) borrowed from another study, 3) generated from a power function, 4) generated from exponential functions, or 5) generated from a gamma function. Since survey data are available, it is possible to generate friction factors from the local data for the CUUATS model. The process to estimate friction factor is shown in Figure 7. Refer to Table 14 for the results. 10 Martin, A. William, and Nancy A. Mcguckin (1998). Report 365: Travel Estimation Techniques for Urban Planning. National Academy Press. Washington D.C. 23

30 Survey Data Build Trip Table Network Skim File Survey (Production/ Attraction Format) GMHFIL.IN File Build Gravity Model History File Gravity Model History File Calibrate Gravity Model Friction Factor Figure 7. Process to Estimate Friction Factors from the Survey 4. Use of Gravity Model Once the friction factor is prepared from the survey data, the trips are distributed using Gravity Model. This model can be simply described in that all trips produced in a zone are attracted to other zones depending on the distance and activity. The gravity model assumes that all trip attractions are directly proportional to the relative activity level, and inversely proportional to the distance between the zones. The following equation explains the gravity model: 24

31 T ij = P i zones k = 1 A A k j F F ik ij K K ik ij Where: T ij = the number of trips from zone i to zone j P i = the number of trip productions in zone i A j = the number of trip attractions in zone j F ij = the friction factor relating the spatial separation between zone i and zone j K ij = an optional trip-distribution adjustment factor for interchanges between zone i and zone j Input data for this process include: Trip Production Table, Trip Attraction Table, Friction Factor, and Highway Network Skim File. As a result, the final trip person distribution matrix is produced. The Output of trip distribution was five person-trip tables. Although the control totals of each trip purpose were the same as the trip generation results, the output from this process was twodimensional person-trip matrices showing the movement of trips between zones. Table 14. Calibrated Friction Factors Minutes HBWork HBSchool HBShopping HBOther NHB 1 150,680 54, , ,764 52, ,724 77, , , , ,836 98, , , , , , , , , , , , ,205 86, ,442 18,516 89,115 88,368 68, ,938 94,615 71,847 71,174 57, ,534 77,057 59,196 58,070 50, ,522 59,203 49,729 47,966 46, ,011 43,337 42,496 40,089 44, ,990 30,524 36,858 33,884 43, ,378 20,892 32,371 28,944 43, ,073 14,034 28,722 24,975 43, ,977 9,344 25,687 21,756 43, ,031 6,228 23,103 19,121 43, ,230 4,197 20,848 16,946 42, ,660 2,888 18,832 15,136 40, ,538 2,049 16,991 13,618 37, ,287 1,514 15,274 12,333 33,266 25

32 VI. Mode Choice 1. Mode Split Procedures Mode choice is the third step among the four major steps in travel demand model. It is a process to split all person trips into auto travelers and transit passengers according to the relationships between the two different modes: private automobile and public transit. Split is performed using diversion curves that are provided by the Tranplan program. The curves specify the percentages of transit travel as a function of the ratio or differences between one of the trip purpose, origin zone code, destination zone code, and trip impedance range. The model is using 5 trip purposes and 4 different area types (See the Table 15 below). Among 5 trip purposes, home-based shopping trips and home-based other trips share the same split curve since the percentages of transit users for those trips are similar. Thus, 4 split curves were used in this model. Table 15. Production/Attraction code and Area types used in mode choice model Production or Attraction Code Area Type Traffic Analysis Zone 1 CBD, Fringe Other Business Area U of I, Parkland College 44-66, , Residential Area , Rural Area , , Input in this step is total person trip, a transit skim file, and a highway skim file. Output is the transit passenger trip tables and the auto trip tables. As results, volumes by 4 split curves were shown in the Table 16. Table 16. Volumes by Split Curves Scurves Trip Purposes Transit Highway Total 1 Home-Based Work 2 Home-Based School 3 Home-Based Shopping & Home-Based Other 4 Non-Home-Based Total 15,632 (13%) 9,468 (25%) 7,342 (3%) 9,787 (6%) 42,231 (7%) 101,564 (87%) 28,514 (75%) 249,687 (97%) 144,701 (94%) 524,466 (93%) 117,196 (100%) 37,982 (100%) 257,029 (100%) 154,490 (100%) 566,697 (100%) The number of auto drivers can be determined through the auto occupancy curve, which specifies and splits the auto person trips into auto drivers and auto passengers. Auto occupancy curves were used to convert person trips to vehicle trips. Resulting volumes by those curves are 409,973 for auto drivers and 115,093 for auto passengers. The auto occupancy ratio was 1.3 persons per vehicle. 26

33 2. Creation of an origin-destination trip table After a mode choice and before a trip assignment, production-attraction format of trip table should be converted into origin-destination format to get actual directions of trips. Productionattraction format of trips expresses the directions going from home-end of the trip (production) to non-home end of the trip (attraction). That doesn t reflect the real directions from origin to destination. The method to convert production-attraction trip tables into origin-destination trip tables is to add one-half of the trip table to one-half of the transposed trip table. In Tranplan, this conversion is performed by a matrix transpose function and applied to home-based trips ranging from trip purpose 1 to trip purpose 4. Non-home-based trips have origin-destination direction by definition. These home-based and non-home-based trips in OD format were combined into one table by using the function of Matrix Manipulate. 27

34 VII. Trip Assignment Trip assignment loads the highway and transit person trips on networks. Theses traffic loadings are done on the 24-hour time period. Final output of this process is assigned highway and transit trips on each of networks. Inputs for trip assignments are coded networks and trip tables generated from the mode choice process. Following sections describe the highway and transit trip assignment, respectively. 1. Highway Trip Assignment 1) Methodology There are several methods for loading trips on the network: All-or-nothing assignment, capacity restraint assignment, incremental assignments, stochastic assignments, and equilibrium assignment. Equilibrium assignment is used with all newly developed models and mostly recommended. This technique is used in this model. By definition, equilibrium is reached when no travelers can reduce his or her travel time from origin to destination by switching to another path 11. This method is performed in the Tranplan function Equilibrium Highway Load. Modified Bureau of Public Roads (BPR) curves given in Highway Capacity Manual 2000 were used in this model. Free flow speeds and LOS E capacities are additional requirements. 2) Parameters The Bureau of Public Roads (BPR) Curves has been used to estimate link travel times as a function of the volume-to-capacity ratio. Basic formula is shown in the box below. Highway Capacity Manual 2000 provides recommended coefficients for BPR curves according to the various road classifications. Modified BPR curves 12 were used in this model. For 17 road classifications resulting from the combinations of functional class, speed, and area types, the BPR curve coefficients are shown in the Table 17 below and coded on Link Group 3 of Tranplan link database. v Tc = Tf 1+ α c β T c = congested link travel time T f = link free-flow travel time v = assigned link traffic volume (vehicles) c = link capacity α, β = volume/delay coefficients 11 ODOT, Traffic Assignment Procedures, The modified BPR curves and coefficients were based on the Traffic Assignment Procedures from Ohio Department of Transportation, which is originated from Highway Capacity Manual

35 Table 17. BPR curve coefficients for Link Group3 Free Flow Roadway Link Functional Speed Classification Group3 Class (mph) Highway and Ramps Multi-lane or Rural Roadway Urban Street Area Type* Lanes a b Any Any Any Any Any Any Any Any , 4 Any Any Any ,4 Any Any Any , 4 Any Any Any , 4 Any Any Area type is the same coded on link file of the highway network. Original source: Exhibits C30-1 and C30-2 on page 30-39, HCM a and b are the BRP parameters for the equation. 2. Transit Trip Assignment Transit trip assignment is to load transit passengers on the transit network. Input in this function is transit volume, transit network, and a minimum transit path. Output is a loaded transit legs file sorted by lines. In case that more than one line operate on one link, the method of Frequency Split will be applied, which is a way of dividing transit volume by the relative frequencies of the transit lines. 29

36 VIII. Time-Of-Day Analysis Time of day analysis in this model is performed after trip assignment. The area transportation situation needs to figure out the specific time of day traffic volumes rather than 24-hour daily traffic volumes. Daily time period and percentages of traffic volume are split into three-time period shown at Table 18. PM-peak time which is from 4:40 PM to 5:30 PM has the most traffic volume reaching 12% of daily traffic volume. Table 18. Percentage of peak periods by actual traffic counts done in 2002 (1/09/04) Period Time Percentage Traffic Volume* AM Peak 7:30 8:30 AM 10.23% Off-Peak 12:00 1:00 PM 11.16% PM Peak 4:30 5:30 PM 11.73% * Source: ADT and turning movement traffic counts 2002, CUUATS. Method selected for doing time of day analysis after trip assignment may be sufficient for smaller MPOs where the duration and intensity of congestions are limited. This is mostly commonly used and simplest method. Data required are peak hour factors that reflect peak period link-level travel demand. Limitations are that this method does not consider peak travel times in assignment and does not account for localized effects of changes in demand. 24-hour assigned traffic volume is factored using the traffic percentages of each time period. The Table 19 is the traffic volume for Champaign-Urbana urbanized area resulting from the time-of-day analysis. Table 19. Peak hour traffic volumes for Champaign-Urbana Urbanized Area Period Estimated Volume Percentage AM Peak 929,086 10% Off-Peak 842,976 11% PM Peak 886,031 12% 24-hour 7,553, % 30

37 IX. Model Validation Once transportation model was set up, model validation should be done to verify the estimated model results comparing with the actual traffic count data. The root mean square error check, volume check, plot check and vehicle miles traveled (VMT) check are commonly performed for a base year trip assignment. 1. Plot Check This is the comprehensive check for the trip assignment. Plot shows the assigned traffic volumes on each link of the network. These volumes are compared with actual traffic count data considering whether the trip assignment results seem reasonable in general and whether there are links with overestimated or underestimated volumes when compared in detail. The map 6 shows the overall display of traffic volumes for year Volume Check 1) % Root Mean Square Error The percent root mean square error (%RMSE) is a measure of the relative error of the assignment compared to ground counts. The equation for this is given in the box below. The %RMSE is often used to compare the accuracy between estimated and measured traffic volumes. It is considered that acceptable range for the %RMSE is about 40% or less. The base year model for Champaign-Urbana Urbanized Area has 39% of %RMSE. Equation % RMSE = 100* ( Model j Count j ) j ( NumberofCounts 1) 2 Count j j ( ) NumberofCounts 2) Average Volume Ratio Another method for model validating using traffic volume is comparing the estimated model results with traffic count data results according to the road classifications. Average volume ratio by road classification is shown on the Table 20. Table 20. Average Volume Ratio by Road Classification Road Classification Observations (Number of links) Ratio (Model Estimation / Count) Major Arterials Minor Arterials Collectors Total

38 Map 6. Highway Assignment with Multi-band Width 32

39 3. Vehicle Miles Traveled Check Vehicle miles traveled (VMT) is obtained by multiplying the assigned volume and the distance of the link. The total VHT estimated was 1,581,367 miles and the study area has 125,264 population and 50,254 household. VMT per household/ capita and VMT by road classification are considered for checking the reasonableness of the model. Regarding VMT per household, acceptable range for this is 30 to 40 VMT per household and model result is 31.5 VMT per household as shown in the Table 21. Estimated VMT per capita is 12.6 while the acceptable range for this is 10 to 16 VMT per capita is acceptable. Table 20. VMT per household and per capita. ITEM Acceptable range* Model Results VMT per household VMT/HH 31.5 VMT/HH VMT per capita VMT/capita 12.6 VMT/Capita * Source: Model calibration and validation seminar by TMIP, FHWA Regarding VMT by road classification, arterials and collectors were considered for checking. Difference between actual VMT and estimated VMT were calculated and compared. VMT difference for arterials and collectors were 7.3% and 17.7%, respectively. If VMT difference for arterials is 10% or less and collectors is 20% or less, then they are acceptable. Table 21. VMT by Road Classification Roadway Type Acceptable range* Model Results Arterials Under 10% 7.3 % Collectors Under 20% 17.7 % * Source: Model calibration and validation seminar by TMIP, FHWA 33