Home

Developing a Mode Choice Model for Small and

1. cceeccecesccecesececeeeeeeeteceesaeeeesaeeeenaeeeens 59 8 3 4 Explanatory Vari bl S cs ese oes edie eects cen eaceteat a cloak eects E aE tances 60 8 33 Datarea a E titrinearbaatiubinnied EE TOEN 60 Chapter 9 COmCHISiOMs ausccicistevccaswesesatiscs Geaaespectesvcstoaswausovetanedecessisoussoiscocecasonsecubiviasseuncsbecdonieuss 63 Referentes iid iseiiscecudssscduvscdesiceds shvedesiedacsaadessdecboadsdcessssasbuvisdescesdsseeisdesdosdedessssiavsadessevdesoniscisssiarsadecdees 65 Appendix A Guide to Model Skim Generation Development in TransCAD and ArcMap 10 L cccisaasacscataseuticartnsscadaaenedsatisesemlaadeshiuetaesealdiaanosauaisasacaeestocdelusestcanustinerspdasncanestocsaSeseteess 69 Appendix B Forecasting Tool User Manual esessoessscsssccssecesocesoosesseessccssocescoossoosesseessocesooseo 107 Appendix C The Multinomial Logit MNL Model sesssesesooesoocssscsssccssocesoosssosesscessocesooseo 117 Appendix D The Nested Logit NL Model oesoessesooesosssesoossocssesooesocssessossocssessossoossessssssse 119 Appendix E Travel Demand Models of MPOs Outside of Texas e sseessooesooessocessecssocesooseo 121 viii List of Figures Figure 1 1 Four step trip based approach cccascedecscccenssccnsecerecetaveveaaveae socaaassesatasesaccsnseceendessaers 1 Figure 1 2 Texas MPOs with a travel mode choice component in their TDM 3 Figure 2 1 NL model structure of GCMPC s mode choice model ce
2. KY Validate Features Snapping gt More Editing Tools gt Editing Windows Options Figure A 46 Save edits and stop editing options e Now click on the Final_bus_stops Shapefile in the Table of Contents window and select Open Attribute Table You should see the table similar to Figure A 47 Figure A 47 Attribute table for bus stops e At this point we have the points and we do not know the coordinates of the points Let s add XY coordinate to the points Click at the top of the table and select Add Fields as shown in Figure A 48 96 BA Find amp Replace BS Select By Attributes Key Switch Selection Select All Turn All Fields On Show Field Aliases Arrange Tables Restore Default Column Widths Restore Default Field Order Joins and Relates Related Tables Create Graph Add Table to Layout Reload Cache Print Reports Export Appearance Figure A 48 Add field option to table e Provide a name e g X_Cord and under Type select Double see Figure A 49 Repeat this step to add the Y_Cord field m oa Field Properties Alias Allow NULL Values Yes Default Value Figure A 49 Defining field name and type e Now select the X_Cord field then right click and select Calculate Geometry Next select the X Coordinate of Point in the Property drop down list and click OK The X_Cord field gets populated with corresponding X coordinates
3. fege f 3J Ces Figure A 4 Geographic file selection window e To label the zones click on the Automatic Labels button a on the toolbar TransCAD opens a window as shown in Figure A 5 71 e Chose the field TAZ from the Field drop down list Adjust the font size and color as convenient and click OK This will display the TAZ number at the center of each TAZ Automatic Labels Layer LUBO6 TAZ RMC6767 m A toed I Smart Alignment Stretch T Allow rotation I Allow Duplicates Spacing 0 Inches J Limit Lines to Characters Size fi2 M Boldi Italic l Shows what Label Text w Figure A 5 TAZ visual setting window Add the Network Layer and Node Layer e Chose Map gt Layers or click the 2 button on the toolbar TransCAD opens a window as shown in Figure A 6 e Click on the Add Layer button Select the Network file and click Open This will add the Network file to the list of layers Labels _Autoscale Rename Metadata Geographic File CA tazskimsandnetworksfor6767 Lubbock_Trial Network DBD Figure A 6 Layer addition window 72 e Follow the same procedure and add the Node file to the layer list Select Node layer and click on Hide Layer button This will add the Node file to the layer list but hide the nodes on the map Figure A 7 If you wish to see the details of the Node file at any point in the modeling process simply make the layer visible
4. row_stop col_stop size Stop XY TAZ XY sortrows TAZ XY 1 Number of TAZs to consider based on the distance For example if the user sets a value of 5 the first five TAZs based on distance in ascending order are mapped for each of the stops Thus the first TAZ is closest to the stop followed by second third fourth and fifth User can change this value as per requirement Num_TAZ Reg 5 Outputl Output2 zeros row_stop Num_TAZ Req 1 zeros row_stop Num_TAZ Req 1 for i 1 row_stop if Lat_Lng 1 105 Temp Dist zeros row_TAZ 2 for j 1 row_TAZ delta_lat Stop XY i 2 TAZ XY j 2 delta_lon Stop XY i 3 TAZ XY j 3 a sin delta_lat 2 2 cos TAZ XY j 2 cos Stop XY i 2 sin delta_lon 2 2 c 2 atan2 sqrt a sqrt l a Temp Dist j 1 TAZ XY j 1 Temp Dist j 2 R c end Temp Dist sortrows Temp Dist 2 Outputl i 1 Stop XY i 1 Output2 i 1 Stop XY i 1 Output1 i 2 Num_TAZ_ Req 1 Output2 i 2 Num_TAZ_ Req 1 clear Temp Dist else Temp Dist zeros row_TAZ 2 for j 1 row_TAZ I Temp_Dist 1 Num_TAZ_Req 1 Temp_Dist 1 Num_TAZ_Req 2 X Diff Stop XY i 2 TAZ XY j 2 Y Diff Stop XY i 3 TAZ XY j 3 a sqrt X_Diff 2 Y Diff 2 Temp Dist j 1 TAZ XY j 1 Temp Dist j 2 a end Temp Dist sortrows Temp Dist 2 Output1i i 1 Stop XY i 1 Output2 i 1 Stop XY i 1 I Output1 i 2 Num_TAZ_ Req 1 O
5. 2000 2010 Population Ee Population Population growth Small MPOs Texarkana MPO 89 306 92 565 3 65 San Angelo MPO 105 781 111 823 5 71 Victoria MPO 111 663 115 384 3 33 Sherman Denison MPO 110 595 120 877 9 30 Wichita Falls MPO 151 524 151 306 0 14 Abilene MPO 160 245 165 252 3 12 Harlingen San Benito MPO 144 658 173 278 19 78 Medium small sized MPOs Tyler Area MPO 174 706 209 714 20 04 Longview MPO 194 042 214 369 10 48 Bryan College Station MPO 184 885 228 660 23 68 Waco MPO 213 517 234 906 10 02 Brownsville MPO 190 569 241 831 26 90 Amarillo MPO 226 522 249 881 10 31 Laredo MPO 193 117 250 304 29 61 Midland Odessa Transportation Organization 237 132 274 002 15 55 Lubbock MPO 249 700 284 890 14 09 South East Texas Regional Planning Commission 385 090 388 745 0 95 Killeen Temple MPO 330 714 405 300 22 55 Corpus Christi MPO 403 280 428 185 6 18 Medium large sized MPOs Hidalgo County MPO 569 463 774 769 36 05 El Paso MPO 679 622 800 647 17 81 Source Texas State Data Center 2011 3 1 2 Mode Choice Shares To obtain a sense of modal shares in the Texas small and medium sized urban areas the CTR team extracted information from the 2009 American Community Survey and obtained modal shares for work trips for the 21 MPOs that currently have not implemented mode choice models within their travel demand modeling framework see Table 3 2 Not surprisingly the vast majority of work trips in each urban area are pursued by driving alone
6. 80 1 AD Enable Disable Links Option Disable Links z Using By Expression 7 r Link Costs J7 Update costs on all links Network Field In Line Laye Figure A 21 Network update window e Click OK enter the condition in the Expression box and then click OK twice Here we identify the freeway and highway network as functional classification 4 as shown in Figure A 22 FUNCL 4 Name Figure A 22 Condition window e To ensure that the links have been disabled click on Info as shown in Figure A 23 and TransCAD shows the information on disable links You can click on the Info button at any time during the modeling process to check the network status 81 Figure A 23 Network info window e Run the Multiple Shortest Path on this network and minimize the distance This will create the distance matrix for bike and walk mode with constrained network Calculate the travel time matrix for the bike and walk by assuming average biking 11 mph and walking 3 mph speeds e TransCAD does not automatically enable the disabled links the next time you start a new session on the same network In order to work with full network select Network Paths gt Settings gt Updates and select Enable Links and All Features as shown in Figure A 24 Update node costs Network Node Field In Node Layer Figure A 24 Re enable all th
7. 82 42 on average across the 21 urban areas or carpooling 12 62 on average In several urban areas Denison Texarkana Midland Temple and Port Arthur the percentage of commuters that rely on the automobile by driving alone or carpooling to reach their workplace exceeds 97 Only in three urban areas College Station El Paso and Laredo does the public transportation share exceed 1 5 College Station registers the highest share of non motorized mode share 6 42 attributable to special generator trips from the Texas A amp M College campus 20 Table 3 2 Mode shares for HB work trips Modal Share ee ae me Carpool Transit Bike Walk Other Small MPOs Texarkana MPO Texarkana 85 43 11 81 0 23 0 00 1 55 0 97 San Angelo MPO San Angelo 81 46 11 01 0 61 0 06 5 00 1 86 Victoria MPO Victoria 80 16 15 53 0 78 0 34 1 95 1 24 Sherman Denison MPO Denison 88 15 10 37 0 32 0 46 0 70 0 00 Sherman 83 91 12 90 0 22 0 00 1 10 1 86 Wichita Falls MPO Wichita Falls 80 86 10 71 0 48 0 00 6 79 1 15 Abilene MPO Abilene 84 38 11 41 0 39 040 2 19 1 23 Harlingen San Benito MPO Harlingen 86 02 10 84 0 06 0 42 1 25 1 40 Medium small sized MPOs Tyler Area MPO Tyler 84 65 11 01 0 57 0 18 0 97 2 62 Longview MPO Longview 84 93 11 27 0 21 0 13 140 2 06 Bryan College Station Bryan 83 62 11 53 1 31 0 34 1 02 2 19 MPO College Station 78 25 10 42 3 84 2 52 3 90 1 08 Waco MPO Waco 82 79 13 13 0 30 0 21 2 80 0 78 Brownsville MPO Brownsville 78 50 14 52 1 26 0 08 2 71
8. Chapter 8 53 54 Chapter 8 Model Development 8 1 Introduction and Overview This chapter describes the mode choice model for two medium small sized MPOs Lubbock and Longview The need for the addition of a mode choice model to the four step planning process for medium small sized MPOs is evident all 21 small and medium MPOs in Texas plan to improve their public transportation systems incentivize non motorized modes reduce emissions and provide more multi model transportation options by year 2035 as a part of their strategic planning goals To evaluate the impact of such policies all the MPOs will require a mode choice model but currently none of the medium to small MPOs has a mode choice model in their four step planning process The mode choice model developed here is for HB work trips exclusively the data for the model development is derived from a 2005 survey 8 2 Lubbock MPO 8 2 1 Traffic Analysis Zones The Lubbock area has 723 internal TAZs Each TAZ is classified into one of four area types CBD CBD fringe urban and suburban 8 2 2 Modes In the current model development we considered five modes drive alone shared ride transit walk and bike 8 2 3 Network and Level of Service Preparation The level of service variables IVTT and OVTT travel cost and travel distance were generated for all five modes considered in the analysis The network file obtained from TxDOT forms the basis for development of lev
9. Fort Bend Waller and Harris HGAC develops its TDM in collaboration with TxDOT and the Metropolitan Transit Authority of Harris County METRO In the trip generation step trips are categorized into 14 purposes The trip household production models use cross classification trip production rates developed from the HGAC 1995 Household Travel Survey data while the trip attraction rates are stratified by area type and employment category An atomistic model is used for the trip distribution step HGAC 2012 The development of the HGAC mode choice model was based on the 1995 household survey data and on board transit rider survey data HGAC updated its previous mode choice model an NL model with a new NL model that encompasses more alternatives than the previous model and a different nesting structure Additionally separate NL models were developed for five income groups and three trip purposes HB work HB non work and NHB The model includes 15 alternatives as follows e Drive alone non toll e Transit walk access commuter bus e Drive alone toll e Transit walk access local bus e Two person auto non toll e Transit walk access express bus e Two person auto toll e Transit walk access urban rail e Three person auto non toll e Transit walk access commuter rail e Three person auto toll e Transit drive access PNR e Four plus person auto non toll e Transit drive access KNR e Four plus person auto toll All mode choice models estimated had the sam
10. Source AMBAG 2011 Figure 2 2 NL model structure of AMBAG s mode choice model e Metro MPO Metro 2008 An MNL model was used for the mode choice step Metro s model was applied to three trip purposes HB work HB other and NHB and two time periods peak 07 00 08 59AM and off peak 14 00 14 59 The mode choice alternatives considered include the nine choice alternatives shown in Table 2 3 Household demographic variables and income specific cost coefficients were used for the model specification Accessibility measures include household employment and intersection density Mode characteristics considered in the analysis include in vehicle time walk time first wait time modeled at 50 of headway transfer wait time and number of boardings Bike and walk travel times are calculated based on assumed speeds 2 3 Mode Choice Models in Texas As mentioned in Section 2 1 each MPO in Texas is responsible for the transportation planning and programming coordination within their urban area In this section we review the mode choice models developed by the four large MPOs in Texas Capital Metro MPO CAMPO San Antonio Bexar County MPO SABCMPO Houston Galveston Area Council 11 HGAC and North Central Texas Council of Governments NCTCOG All urban areas use a four step trip based approach to model and forecast travel demand 2 3 1 Capital Metro MPO CAMPO CAMPO is the MPO for the Austin area which includes the Bastrop Ca
11. The two urban areas that were employed in developing the model were a subset of the four considered in this chapter 46 Chapter 7 Procedure to Prepare Data The objective of this chapter is to provide the steps procedures adopted in order to prepare the data for mode choice model estimation Following the framework of Chapters 2 through 4 we focused exclusively on HB work trips The research project required the CTR research team to develop a mode choice model for two Texas MPOs We selected the Longview and Lubbock MPOs The selection is based on conversations with the PMC regarding survey data availability and the recommendations made in Chapters 2 and 3 The MPOs locations are presented in Figure 7 1 ma Lubbock MPO mm Longview MPO Figure 7 1 Selected study area For each of the MPOs we generated skims for five travel modes e Drive alone e Carpool e Transit bus e Walk e Bicycle The remainder of the chapter is organized as follows Section 7 1 provides information about the steps adopted in extracting necessary demographic data and trip characteristics from the survey data Section 7 2 presents the steps and assumptions made in the development of skims for the two selected MPOs Section 7 3 provides conclusions and recommendations where appropriate 47 7 1 Procedure for Survey Data Extraction The steps involved in extraction of demographic data and trip characteristics are identical for both the
12. dass raster Create a new coordinate system Edit the properties of the currently selected coordinate system Set the coordinate system to Unknown Save the coordinate system to a file C Figure A 39 Coordinate system window 90 e Again in ArcCatalog click on the working folder and select New gt Personal GeoDatabase Rename it appropriately Do not change the mbd extension e Click on the GeoDatabase and select New gt Feature Dataset e Provide a name for the dataset followed by coordinate system Click Next and Finish see Figure A 40 91 Choose the coordinate system that will be used for XY coordinates in this data The XY tolerance is the minimum distance between coordinates before they are Geographic coordinate systems use latitude and longitude coordinates on a spherical model considered equal The XY tolerance is used when evaluating relationships between of the earth s surface Projected coordinate systems use a mathematical conversion to features Name lt Unknown gt transform latitude and longitude coordinates to a two dimensional linear system poo Unknown Units m Geographic Coordinate Systems import H E Projected Coordinate Systems a i 0 001 M Tolerance 0 001 Reset To Default About Setting Tolerance 7 Accept default resolution and domain extent recommended sek Cae Cet Figure A 40 Creating feature dataset 92 e Click on t
13. seniors 65 yrs 1 75 and persons with disabilities or older and persons with disabilities Weekly pass for Medicare seniors 65 yrs 7 25 or older and persons with disabilities f Monthly pass for Medicare seniors 65 yrs 50 00 or older and persons with disabilities Source http www citibus com page services Citibus regular routes and number of stops are the following e 1 Dunbar Area 5 stops e 2 East Broadway 4 stops e 5 Boston S Quaker South Plains Mall 6 stops e 6 Buddy Holly 50 St Crosstown 5 stops Longview Transit route maps can be found at http www longviewtransit com routes php Citibus route maps are available at http www citibus com page routes 44 e 9 Ave Q S University S Quaker 5 stops e 12 Arnett Benson 4 St 8 stops to 9 stops back e 14 Cherry Point 5 stops e 19 Wayland Plaza South Plains Mall 7 stops e 34 34 St South Plains Mall 6 stops In addition the following routes operate to around Texas Tech e Texas Tech 10 stops e Red Raider 10 stops e Masked Rider 7 Stops e Overton Park North 4 stops e Overton Park South 5 Stops e Northwest 6 stops e North 4th 8 stops e North Indiana 8 stops e Tech Terrace 10 Stops e West 4th Express 2 stops e TTU s Bus Safe Ride 7 stops 6 3 Transit Skim Generation The guidelines to generate the transit skims were discussed in Chapter 5 The first step towards developing the transit skims is to generate the transit n
14. set between 0 15 to 0 25 miles However this maximum distance can be different in some areas For example SABCMPO uses a maximum distance of 0 6 miles in the CBD to acknowledge good connectivity and 0 15 in other areas 39 e Wait time is commonly computed as one half of the headway with the assumption that transit users arrive at bus stops at random that is as if they didn t know the bus schedule However for habitual users this assumption is not valid as these travelers tend to arrive at bus stops within 15 minutes of the bus departure as assumed by NCTCOG This case is particularly relevant when the bus service has low frequency and should be considered when defining the OVTT Cost The cost for transit corresponds to the bus fare In areas with large populations of elderly or students a weighted measure of the fare can be used Number of transfers Several MPOs penalize transit transfers in their mode choice models This penalization can be done in two ways e Based on the bus route system the number of transfers can be determined for each pair of TAZs Then a TAZ to TAZ matrix containing this information can be used as a skim e Another way to incorporate transfers into the modeling is to develop a matrix of time penalties Again the pair of TAZs in which individuals transfer from one bus to another has to be identified The time penalty varies from 3 minutes in TAZs with high transit density such as the CBD to 20 m
15. urban and suburban 8 3 2 Modes In the current model development we considered five modes Drive Alone Shared Ride Transit Walk and Bike 8 3 3 Network and Level of Service Preparation The level of service variables were prepared in exactly as for Lubbock The transit fare is 1 25 in the Longview area 59 8 3 4 Explanatory Variables The same configuration of explanatory variables was used for Longview as for Lubbock 8 3 5 Data The survey data for the model development was obtained from TxDOT The survey data corresponds to the year 2005 After careful examination and refinement of survey data a sample size of 1189 individual HB work trips was prepared The data preparation steps are same as discussed for Lubbock area Tables 8 8 and 8 9 provide the sample distribution based on household size and income Table 8 8 Distribution of sample based on household size Household Size persons 1 2o0r3 4or more Frequency 67 692 430 Percentage 5 63 58 20 36 16 Table 8 9 Distribution of sample based on household income Income Frequency Percentage Less than 25K 124 10 43 Between 25K amp 50K 417 35 07 Greater than 50K 648 54 50 Table 8 8 clearly indicates that a majority of the population is has a household size of two or three We may infer from Table 8 8 that most of the households are married couples with one or more children Similarly Table 8 9 appear
16. 2 93 Amarillo MPO Amarillo 84 76 11 80 0 39 0 14 163 1 27 Laredo MPO Laredo 77 86 15 96 1 97 0 06 242 1 73 Midland Odessa MPO Midland 85 21 12 51 0 32 0 02 0 95 0 99 Odessa 82 22 14 02 0 23 0 14 1 57 1 82 Lubbock MPO Lubbock 85 57 10 72 0 67 0 29 212 0 63 South East Texas RPC Beaumont 85 42 10 81 0 97 0 17 1 79 0 83 Port Arthur 84 60 12 60 0 37 0 17 154 0 71 Fort Hood CDP 80 26 11 77 0 14 0 02 609 1 72 Killeen Temple MPO Killeen 82 21 14 33 0 38 0 10 1 56 1 41 Temple 85 22 11 91 0 28 0 04 0 96 1 59 Corpus Christi MPO Corpus Christi 79 64 13 80 1 34 0 34 2 05 2 84 Medium large sized MPOs Edinburg 77 00 15 48 0 19 0 61 1 66 5 05 Hidalgo County MPO McAllen 84 00 11 73 0 71 0 05 1 22 2 30 Mission 74 56 16 17 0 00 0 00 0 71 8 56 El Paso MPO El Paso 82 06 11 37 1 85 0 16 2 27 2 29 Average 82 42 12 62 0 71 0 28 2 00 1 97 Std Deviation 3 21 1 83 0 80 0 47 145 1 62 Minimum 74 56 10 37 0 00 0 00 0 70 0 00 Maximum 88 15 16 17 3 84 2 52 6 79 8 56 21 Source U S Census Bureau American Community Survey 2009 3 1 3 Strategic Planning Goals Table 3 3 summarizes the strategic planning goals of the 21 small and medium sized MPOs in Texas The information presented in the table was obtained from the MPOs MTPs In order to assess if the MPOs have goals that align with multimodality the CTR research team classified the goals into the following categories e Improve and or expand public transportation system e Incentivize non motorized modes e Improv
17. 58 Line option on Editor Toolbar e To digitize the lines move the cursor to the start of the road you wish to digitize and click Move along the center of the road and double click to end the line Use small 102 segments for higher accuracy Each time you double click on the image a line will be added to the Final_bus_links shapefile see Figure A 59 ni 22x ERE ML hy O oe x oBJecTiD SHAPE SHAPE Length i fewine sar Figure A 59 Final_bus_links shapefile e To view the digitized lines uncheck the image in the Table of Contents and you should see the lines you just digitized Once the digitization of transit links is completed add the attributes such as travel time name etc using the Add Fields option discussed earlier e Save both the shapefiles Final_bus_stops and Final_bus_links when you are done editing To save the files as separate shapefiles outside the database right click on the respective shapefiles Final_bus_stops and Final_bus_links and select Data gt Export Data You should see a window similar to Figure A 60 Click on the yellow output button circled in black and select Shapefile option in Save as type see Figure A 61 and provide a name Click Save and you are done ames S Export Use the same coordinate system as this layer s source data the data frame the feature dataset you export the data into only applies if you export to a feature dataset in a geodatabase Out
18. County area with a population of 425 790 comparable to medium small sized MPOs in Texas e The study area was divided into 639 TAZs The main source of data was the Michigan Travel Counts Survey Trip productions and attraction rates were estimated for eight trip purposes HB work low income HB work high income HB shopping HB other HB school HB university NHB other and NHB work A cross classification model was used to model trip productions based on number of workers vehicles per household household size and household income data Trip attraction rates were calculated using a linear regression model calibrated by the Michigan Travel Counts database specifically employment variables total employment and total number of households data Association of Monterey Bay Area Governments AMBAG California e The AMBAG MPO is located in California and serves the Santa Cruz Monterey and San Benito counties with a combined population of 733 667 comparable to medium large sized MPOs in Texas 121 e The study area was divided into 1 884 TAZs The main sources of data were the 2000 2001 California Statewide Household Travel Survey and the 2002 Monterey San Benito Household Travel Survey Trip production and attraction rates are estimated for seven trip purposes HB work HB maintenance HB discretionary work based HB school other and visitor from private residence or hotel rooms A cross classification method was used to mode
19. Georeferencing 7 Layer Transit_Map png Zi E Figure A 29 Link table button Restore From Dataset Figure A 30 Link Table 85 e Click on the Add Control Points button shown in the circled area in Figure A 31 Georeferencing 7 Layer Transit_Map png j he Oa EA Figure A 31 Add control point button e Use the control point tool to draw connections between the corresponding points on the unregistered and registered images Create at least four points covering the top bottom right and left portions of the image as shown in Figure A 32 i D nh ame Z LTT 43 es Figure A 32 Adding control points e Now click on the Auto Adjust button to see the root mean square error RMSE value Figure A 33 X Source Y Source X Map Y Map Residual 939309 374798 7291796 777568 951178 748000 7288523 980000 1781 57225 859076 062526 7244763 456581 920046 850000 7275169 620000 1056 29661 948004 610611 7192196 803713 949251 162000 7255589 150000 868 87815 997211 740552 7268675 355066 966397 515000 7281331 190000 1594 15379 ma J 1st Order Polynomial Affine bd Restore From Dataset Figure A 33 RMSE value 86 e Select the points with high Residual value and delete them using the button highlighted with a circle in Figure A 33 Add new points by unchecking the Auto Adjust button until the RMSE value is within a reasonable limit Whe
20. LJTAZshp 1 10 2013 231 PM SHP File Fie nme I ls fee SE FF Open asgeadony 7 Open for excusive acos Figure A 1 File selection menu e Select the file and click Open It opens a window as shown in Figure A 2 C Users caee 1p226874a Desktop MPC 1000 Cancel 867984 565167 7196943 172258H1044439 55246 737 Type Polygon Coordinates Layer Name TAZ Import Layer IV Eliminate duplicate boundary lines Figure A 2 Import shapefile window e Check the Import Layer box and click OK e Upon clicking the OK button TransCAD opens a window as shown in Figure A 3 and allows the user to save the file as a dbd file Provide a name and save the file into the dbd format 70 leskin Ji TAZ J Oe Date modified Type Size WAZdbs 12 1 2013 926 PM Caliper Standard Figure A 3 File save menu Creating the Overview Map The first step in the skim development process is creation of the overview map by combining the network and TAZ files To create the overview map follow the steps in this section Add the TAZ layer e Choose File gt Open and select the TAZ Geographic File dbd format and click Open as shown in Figure A 4 Lookin J Lubbock Taal J o 8 Name 4 Date mosties Type Sze Pe aoe D tee Tene LAUDI PM Fe folder W ieaoo ses 12 2013 10ST PM Caliper Standard sa 12 1 2013 1058 PM Caliper Standard 4a Dones 12 2S1OS4PM Caliper Standard se
21. Model development report 2005 base year model Bhat C R amp Sardesai R 2006 The impact of stop making and travel time reliability on commute mode choice Transportation Research Part B Methodological 40 9 709 730 Also available directly from Dr Chandra Bhat Website http www ce utexas edu prof bhat ABSTRACTS Bhat_Sardesai_TRptB_rev pdf Bureau of Transportation Statistics BTS 2012 Table 3 17 Average Cost of Owning and Operating an Automobile Available at http www rita dot gov bts sites rita dot gov bts files publications national_transportation_ statistics html table_03_17 html Cambridge Systematics 2013 NCTCOG mode choice model documentation Available at http www nctcog org trans modeling documentation index asp Capital Area Metropolitan Planning Organization CAMPO Mode Choice Model Calibration Validation Report prepared for CAMPO Sept 2012 http www projectconnect com connect sites default files CAMPOModelCalibration Valid ationReport_2012 pdf Capital Area Metropolitan Planning Organization 2013 Travel Demand Model prepared for CAMPO May 2013 http www carson org Modules ShowDocument aspx documentid 36649 Capital Area Metropolitan Planning Organization Travel Demand Model Documentation Available at http www jeffcitymo org campo documents 1 1Jan2013CAMPOTDMDocumentation pdf Champaign Urbana Urbanized Area Transportation Study CUUATS 2009 Long range transportation plan 2025 CU
22. This suggests that people put twice the weight on OVTT as compared to IVTT The model also includes a different coefficient value on the cost variable for three different income categories this coefficient value decreases as income increases suggesting a decrease in sensitivity to cost as income increases Table 8 6 provides the implied money value of IVTT and OVTT for three income categories The implied money value for the income group greater than 50 000 is 22 11 hour which is close to the value obtained from the existing CAMPO model of 21 06 hour CAMPO combines the other two categories less than 25 000 and between 25 000 and 50 000 into one category for which the implied value of time is 5 02 hour see CAMPO Mode Choice Model Calibration Validation Report 2012 Table 8 6 Implied money value of travel time Income Category IVTT Value dollars hour OVTT Value dollars hour Less than 25K 3 50 7 00 Between 25K and 50K 8 08 16 16 Greater than 50K 22 11 44 21 Table 8 7 provides the implied mode share for the Lubbock area Table 8 7 Implied mode share for Lubbock Area based on estimated model Mode Frequency Percentage Walk 1 0 05 Drive Alone 1873 94 84 Shared Ride 99 4 99 Transit 1 0 07 Bike 1 0 05 8 3 Longview MPO 8 3 1 Traffic Analysis Zone The Longview area has 336 internal TAZs As with Lubbock each TAZ is classified into one of the four area types CBD CBD fringe
23. and 50K 0 0026 0 0026 0 0026 lt cents Travel Cost Income greater than SOK cents 0 00095 0 00095 0 00095 Finally Table 8 13 provides the implied mode share for the Longview area Table 8 13 Implied mode share for Longview area based on estimated model Mode Frequency Percentage Walk 4 0 33 Drive Alone 1145 96 31 Shared Ride 36 3 06 Transit 1 0 05 Bike 3 0 25 62 Chapter 9 Conclusions This project focuses on developing a process and a framework for 1 generating the inputs needed for estimating a travel mode choice model that includes the transit mode and 2 developing a framework for implementing the results of an estimated travel mode choice model to project mode shares in response to demographic changes and to improvements in transit service In terms of generating the inputs for estimating a mode choice model an important component is the generation of the necessary network skims travel times and costs by alternative modes Most metropolitan planning organizations MPOs have good geographic information systems GIS based representations of the highway network which can be used to generate drive alone and shared ride skims based on certain assumptions as discussed in detail in Chapter 8 However this is not the case with transit skims because of the lack of a good GIS based representation of the transit networ
24. by selecting the Node layer and clicking on the Show Layer button as shown in Figure A 8 Metadata Geographic File CA retazskimsandnetworksfor6767 Lubbock_Trial Zone DBD Figure A 7 Layer visualization window LUBO6 Nodes RMC6767 Hidden Style Labels Rename Metadata Geographic File C retazskimsandnetworksfor6767 Lubbock_Trial Nodes DBD Figure A 8 Layer visualization window e Click the Close button At this moment the TAZ boundaries and Network lines are not clearly visible Change the Zone and Network Style e From the drop down list of layers on the toolbar shown in Figure A 9 choose the layer on which you want to modify the visual settings 73 Cy el aeons rece Fl Di 3 3 ot 3 8 ea A 4 JB06 TAZ RMC676 NET RMC6767 Figure A 9 Toolbar e Click on the toolbar TransCAD opens a window as shown in Figure A 10 Change the setting as desired and click OK Style Layer LUBO6 TAZ RMC6767 oe x Fill Color P Option F Transparent Figure A 10 Visual adjustment window for layers e To add a title to the map click T in the Tools list Figure A 11 e Draw a wide rectangle on the top of the map and type the title e To change the style click R in the tools list Figure A 11 Click on the text then right click to select Properties and modify the settings 74 lt loll YIR lele iF e He fi
25. choosing two alternatives is independent of the presence or attributes of any other alternative The premise is that other alternatives are irrelevant to the decision making process when choosing between the two alternatives in the pair The IIA property has some important ramifications in the formulation estimation and use of MNL models in particular the IIA property allows the addition or removal of an alternative from the choice set without affecting the structure or parameters of the model MNL model development consists of formulating model specifications and estimating numerical values of the parameters for the various attributes specified in each utility function by fitting the models to the observed choice data The critical elements of this process become the selection of a preferred specification based on statistical measures and judgment The model estimation is conducted using the maximum likelihood technique see Koppelman and Bhat 2006 for details which is included in most statistical software 118 Appendix D The Nested Logit NL Model The MNL model structure has been widely used for both urban and intercity mode choice models primarily due to its simple mathematical form ease of estimation and interpretation and the ability to add or remove choice alternatives However the MNL model has been widely criticized for its independence of irrelevant alternatives IIA property discussed in Appendix A The IIA property may
26. estimates two files named INDIVIDUAL RECORD_EST and INDIVIDUAL_MODE_SHARE will be generated reporting mode choice probability and total mode share The INDIVIDUAL_RECORD_EST sheet contains all the individual records and appends the skims for various modes along with the probability of choosing a particular mode for each of the records The INDIVIDUAL_MODE_SHARE sheet will provide the summary of mode share as shown in Table B 4 113 Table B 4 Individual Level Mode Summary MODE FREQUENCY PERCENTAGE WALK 0 0 DRIVE ALONE 2074 99 71 SHARED RIDE 6 0 29 TRANSIT 0 0 BIKE 0 0 Total 2080 100 Now if the user chooses the option TAZ level estimates instead of Individual level estimates three sheets named TAZ HH INCOME DATA EST TAZ MODE SHARE and TAZ MODE SHARE FINAL will be generated The sheet named TAZ HH_INCOME_DATA_EST contains information on TAZ pair skims and mode availability The next sheet titled TAZ MODE_SHARE contains mode share for each TAZ pair for all possible combinations of household size and income For example if the model has three household categories and three income categories nine combinations will be formed and mode share for each combination will be reported in the sheet TAZ MODE_SHARE Finally the sheet TAZ MODE SHARE _ FINAL reports the weighted mode share To obtain the actual mode share simp
27. for the net file TransCAD creates the net file and makes it the active network as shown in the status bar in Figure A 15 76 Create links from Entire line layer o 5 Figure A 14 Binary network creation window Map scale 1 Inch 1 33269 Miles 1 84 439 a x 4 602365 32 578447 Network c uqviewmpodata combine netf net Figure A 15 Status bar Generating the Skims e Select Network Paths gt Multiple Shortest Path from the toolbar Make sure that Network file is selected in the drop down list on the toolbar otherwise the Multiple Shortest Path option will not be available e Select Time under the Minimize option and select Centroid for both the From and To fields This step will ensure that all the calculation starts and ends at the centroid node see Figure A 16 77 Line Layer LONG_NETWORK Network CAU odata Combine netf net m Settings Minimize From Centroid X To Centroid z r Store Results Route System Matrix File Figure A 16 Multiple shortest path menu e If the centroid set is not shown in the From and To option boxes select All Features This action will generate a shortest path for each Node pair The disadvantage with this step is that once the matrix is generated the user needs to remove the unnecessary node pair travel time e In order to generate the corresponding distance skim click
28. indicates that the majority of the population is distributed among household sizes of two or three and four or more We may infer from Table 8 1 that most of the households are married couples with one or more children Similarly Table 8 2 appears to indicate that the majority of the population is in the high income category greater than 50 000 Table 8 3 provides the descriptive statistics for level of service variables for all five modes Table 8 3 Descriptive statistics for level of service variable Variables Minimum Maximum Average Standard Frequency Deviation DY EE Oe ances 0 1 59 9 14 1 10 4 1975 Min ON EYAL ne 2 2 8 23 0 2 1975 Min Travel Distance for DA amp SR 2 Miles 0 1 47 1 10 1 8 2 1975 TENE COSitOE DS 0 01 11 8 2 5 2 1 1975 Dollars Travel Cost for SR 2 Dollars 0 01 5 9 1 3 1 1975 IVTT for Transit Min 4 0 25 0 9 2 4 54 OVTT for Transit Min 3 9 18 4 10 3 2 7 54 Travel Cost for Transit Dollars 1 8 1 8 1 8 0 54 Trip time for Walk Min 29 8 19 6 7 8 0 3 81 Trip Distance for Walk Miles 1 5 1 0 4 0 1 81 Trip time for Bike Min 0 3 40 23 1 9 8 997 Trip Distance for Bike Miles 0 1 7 3 4 2 1 8 997 DA drive alone SR shared ride For each level of service variable we report the minimum maximum average standard deviation and frequency The column frequency indicates the number of samples for which the cor
29. mode e TAZ TAZ distance The TAZ TAZ distance used by the carpool mode is same as the drive alone mode e Travel cost The travel cost used by the carpool mode is same as the drive alone mode cost divided by the number of passengers in the car In this study we assumed a two centroid link depending on the area type For example CAMPO uses an area based structure for travel time speed calculation for a centroid link 51 passenger carpool Hence the travel cost for the carpool is calculated by dividing the drive alone travel cost by two 7 2 3 Skim Generation for Transit Mode e In vehicle travel time In order to generate the transit IVTT skim we mapped the transit routes Longview and Lubbock Transit Agency Site Maps not dated and stops for both MPOs obtained from the websites of the respective city transit agencies onto a separate shapefile the bus route shapefile We also used the bus schedules to determine the travel time between stops To calculate the travel time between two stops we used the following procedure labeling two example stops S1 and S2 all the bus trips starting from S1 in the peak period 7 00 am to 10 00 pm and stopping at S2 were considered and the average of all trips was recorded as the travel time between S1 and S2 Further the bus route shapefile was joined with the respective city TAZ layer and the zones with transit access were identified In this instance we made the assumption that TAZ tha
30. not properly reflect the behavioral relationships among groups of alternatives That is other alternatives may not be irrelevant to the ratio of probabilities between a pair of alternatives In some cases this will result in erroneous predictions of choice probabilities This limitation of the MNL model results from the assumption of independent error terms in the utility of the alternatives Different models can be derived through the use of different assumptions concerning the structure of the error distributions of alternative utilities Among them the nested logit NL model is the simplest and most widely used The NL model represents important deviations from the IIA property but retains most of the computational advantages of the MNL model The NL model is characterized by grouping or nesting subsets of alternatives that are more similar to each other with respect to excluded characteristics than they are to other alternatives This characteristic is exemplified in Figure D 1 in which the modes Bike and Walk are grouped in one nest denoted non motorized modes Alternatives in a common nest exhibit a higher degree of similarity and competitiveness than alternatives in different nests Non motorized sy ee ieee om GD dd Drive Bus Bike Walk Figure D 1 Mode choice framework of NL models The derivation of the NL model is based on the assumption that some of the alternatives share common components in their random
31. of multiple alternatives individuals will choose the alternative that provides them the highest level of value or attractiveness or utility referred to as utility maximization The utility associated with an alternative has two components a deterministic or observable component that represents the portion of the utility observed by the analyst and is a function of the attributes of the alternatives and the characteristics of the decision maker and an unknown or unobserved component that can be the result of many sources imperfect information measurement errors omission of modal attributes and omission of the characteristics of the individual that influence his her choice Two of the most commonly used utility maximizing models are the multinomial logit MNL model and the nested logit NL model see Appendix A and B for details 2 2 Mode Choice Models Outside of Texas The TDMs of the following five MPOs outside of Texas with emphasis on the mode choice model component were reviewed e Champaign County Regional Planning Commission CCRPC Illinois e Lincoln MPO Nebraska e Genesee County Metropolitan Planning Commission GCMPC Michigan e Association of Monterey Bay Area Governments AMBAG California e Metro MPO Washington These MPOs were chosen to represent the four population based categories defined in Table 1 1 the medium small sized category had two representative MPOs All MPOs presented in this section use a t
32. on Skims button as shown in Figure A 16 and select Length under Field option and chose All Links under Skim Type Figure A 17 Click OK All fields Selected fields Lok cca x Figure A 17 Additional skim selection window 78 e Click OK one more time and TransCAD generates the shortest path travel time and corresponding distance matrix To select the travel time matrix select Time from the drop down list on the toolbar see Figure A 18 oaas m 3 ell Figure A 18 Toolbar e To save the matrix click a on the toolbar Matrices saved using this option are readable only by TransCAD e TransCAD also allows user to save the matrix in different formats To export the matrix select Matrix gt Export and click OK Figure A 19 P Matrix Export Export to a Table with One Record for Each Column in matrix TIME Matrices to Include Select All gth Skim Lok caca Figure A 19 Matrix export window e TransCAD opens a window and asks the user for the file type Select the desired file type under the option Files of type We recommend the txt or csv format as they are easy to view in Microsoft Excel Note that this option does not save the skim in the matrix form It will create a record per line for each non empty cell in the original matrix e With this step we finish the in vehicle travel time and the corresponding travel distance ma
33. passenser 3 trip purposes HB distance travel cost P 8 Metro MPO work HB other large MPO NHB 2 time periods accessibility measures household income number of workers per household number of vehicles per household household size 4 Bus only by walk access 5 LRT only by walk access 6 Bus LRT by walk access 7 Transit by PNR access 8 Bike 9 Walk Following is a summary of the modelling approaches of these MPOs e CCRPC CUUATS 2009 Until recently a fixed curve method was used for the mode choice step In 2011 CCRPC updated their mode choice model to the MNL using the five modes presented in Table 2 3 Three data sources were used to develop the MNL model transit on board survey data local transit district routes and ridership information data The resulting model was validated comparing the observed and estimated boardings and trough transit screen lines and cutline checks The CCRPC case study highlights that modeling can benefit a small MPO by identifying these elements o the uses and benefits of travel demand forecasting on a regional basis o the resources necessary to develop validate maintain and operate travel demand forecasting capabilities on a regional basis e Lincoln MPO Lincoln MPO 2011 In the mode choice step Lincoln MPO uses a mode split approach in which the percentage of non motorized trips and transit trips are identified with any remaining trips being classified as auto t
34. see Figure A 50 Do 97 the same for Y_Cord field but select Y Coordinate of Point in the Property drop down list and click OK see Figure A 51 NO a 70 Point 946809 016311 lt u gt 47 Point 947191742126 lt u gt Heen meee tu 13 Point 937385 480713 lt Null gt Property X Coordinate of Point X Coordinate System Use coordinate system of the data source Unknown Use coordinate system of the data frame Unknown Units Unknown Units Calculate selected records only Figure A 50 Calculate X coordinates 98 9 Point 946873 562007 7262816 70 Point 246809 016311 7268302 11 Point 947131742128 7273520 12 Point 939838182312 7273520 J Pom 937385 480713 7266367 Coordinate System Use coordinate system of the data source Unknown Use coordinate system of the data frame Unknown Units Unknown Units Calculate selected records only Figure A 51 Calculate Y coordinates This completes the digitization of bus stops from the image into a geographic file Next we turn our attention towards digitization of bus links First we must set the snapping environment ArcMap provides two ways to set the snapping environment the new snapping tool or the classic snapping tool Setting the Snapping Environment New snapping tool e To use the new snapping tool select Customize gt Toolbars gt Snapping To selec
35. sized MPOs population between 200 001 and 500 000 e Medium large sized MPOs population between 500 001 and 1 000 000 e Large MPOs population greater than 1 000 000 This categorization is based on the National Cooperative Highway Research Program NCHRP Report 716 NCHRP 2012 and allows us to compare Texas MPOs with other U S MPOs The categories and corresponding classification are presented in Table 1 1 Table 1 1 Population based classification of Texas MPOs Population Category 2010 MPO Name 92 565 Texarkana MPO 111 823 San Angelo MPO SAMPO Small 115 384 Victoria MPO MPOs population 120 877 Sherman Denison MPO between 50 000 and ae 200 000 151 306 Wichita Falls MPO 165 252 Abilene MPO 173 278 Harlingen San Benito MPO 209 714 Tyler Area MPO 214 369 Longview MPO 228 660 Bryan College Station MPO BCSMPO 234 906 Waco MPO Medium small sized 241 831 Brownsville MPO MPOs 249 881 Amarillo MPO population between 250 304 Laredo MPO 200 001 and 500 000 274 002 Midland Odessa Transportation Organization MOTOR 284 890 Lubbock MPO LMPO 388 745 South East Texas Regional Planning Commission SETRPC 405 300 Killeen Temple MPO KTMPO 428 185 Corpus Christi MPO Medium large sized 774 769 Hidalgo County MPO HCMPO MPOs 800 647 El Paso MPO population 500 001 to 1 000 000 1 716 289 Capital Area MPO CAMPO 2 142 508 San Antonio Bexar County MPO SABCMPO 5 946 800 Houston Galveston Area Council HGAC 6 371 773
36. the Customize menu select Toolbars and find Editor Fix the Editor box on the toolbar by placing the box in the empty region of the ribbon e Click on Start Editing under the Editor Toolbox Option and select Final_bus_stops see Figure A 43 and click OK i eR 2 x This map contains data from more than one database or folder Please choose the layer or workspace to edit O Pstreets2008 Source Type O c 6767_work Shapefiles dBase Files i C 6767_work Stops 1 mdb Personal Geodatabase About Editing and workspaces ox cae Figure A 43 Start editing option box 94 e This opens up a pane on the right side Select Final_bus_stops and the program will highlight the Construction Tools at the bottom Select the Point option as shown in Figure A 44 ax v Oi lt Search gt Q Final_Bus_Links EF Construction Tools Final_Bus_Links Point v Point at end of line Final_bus_stops Figure A 44 Create feature window e Select the Point option from the Editor Toolbox see the square blue box in Figure A 45 and start clicking on the image to digitize the points Editor gt PRIN hb x 9 Bp Figure A 45 Point option on Editor Toolbar e Click on all the points you want to digitize and then click on Save Edits followed by Stop Editing see Figure A 46 95 Editor 7 Stop Editing E Save Edits Move P Buffer
37. travel time is usually higher for HB work trips compared to NHB trips Table 4 1 Attributes to incorporate in mode choice models Attribute type Attributes to be incorporated by Texas MPOs Household size 1 person 2 persons 3 persons 4 persons 5 or more persons Demographics as dummy variables Income levels low medium and high Total travel time out of vehicle travel time in vehicle travel time Travel system attributes at the TAZ to TAZ level for each travel mode Total travel cost parking costs for auto modes only Presence and number of transfers for transit modes only Walk access time for transit and walk access distance to transit used to determine transit availability To exemplify this model specification consider an HB trip between two TAZs Assume that the only variables used are household size and total travel time Household size is categorized in three levels 1 person 2 persons 3 persons or more while total travel time is used as a continuous variable measured in minutes Only two modes are available transit and auto Then for each individual g and mode i the deterministic component of the utility function is given by Equation 4 3 Equation 4 3 Example of deterministic component of the utility function dummy variable equal to 1 if individual q belongs HHsize2 to a household with 2 persons 0 otherwise V Boa 6 HHsize2 a HHsize3 dummy variable
38. 05 TDM represents a cooperative effort among the SABCMPO Alamo Area Council of Governments VIA Metropolitan Transit Authority and TxDOT and its TPP Division Productions and attractions are estimated using TxDOT s TRIPCALS trip generation modeling software package and the trip distribution step is undertaken using TxDOT s Atom 2 gravity model distribution package SABCMPO 2011 For the San Antonio region a series of comprehensive travel surveys were conducted during 2005 2006 to update their TDM in particular household travel survey data was utilized for the development of the mode choice model A total of seven alternatives were considered e Drive alone e Shared ride two person carpool e Shared ride three person carpool e Bus walk access e Bus separate drive access or PNR e Bicycle e Walk SABCMPO s mode choice model estimates the person trips by travel mode at the zonal level by taking into consideration characteristics of the traveler and available highway and transit services Different mode choice models were for different time periods categorized as peak 6 30 9 00 AM and 3 00 6 00 PM and off peak all other time periods of the day and three trip purposes HB work HB other and NHB An NL model was used for the mode choice model see Figure 2 5 A wide range of explanatory variables were considered in the model including IVTT and OVTT income travel cost wait time number of transfers and parking cost Som
39. 08 and the transit network image is titled Transit_Map mealen Figure A 25 Input files in the workspace S This Network file is the same file obtained from the MPO 83 e Select the Georeferencing tool open the Customize menu select Toolbars and find Georeferencing Fix the Georeferencing box on the toolbar by placing the box appropriately in the empty region of the ribbon Figure A 26 shows the overview of the Georeferencing toolbar Georeferencing 7 Layer Transit_Map png AQ F E s Fit To Display Transformation p de Rotate Left Auto Adjust A Flip Horizontal Flip Vertical a Figure A 26 Georeferencing toolbar e To make the image visible within the workspace area select the transit image in the Georeferencing toolbar as shown in Figure A 25 and select Fit to Display This action might hide the registered image Select the shift tool Figure A 27 to drag the unregistered image to the side as shown in Figure A 28 Georeferencing Layer Transit_Map png LE CG Rotate 5 Shift KA vty Scale hg Figure A 27 Shift tool 84 EA gil WS i ji of Figure A 28 Images after using shift tool e Open the View Link Table note the circled area in the Figure A 29 ArcMap opens the Link table Figure A 30 Uncheck the Auto Adjust option at the bottom of the table Do not close the Link Table just drag it to the side
40. 1 Calculate Y cg tdinate S eaa A A a e ae aai 99 Figure A 527 Snapping toolbate niseni n a a a Bae ee 99 Figure A 53 Snapping toolbar Options sssssesssesesesessseessressetsseesseeessseesstessersseessseesseesseessesset 100 Figure A 54 New snapping tolerance setting WINdOW ee cesceesseeeseecnseceeeeeeneeeaeecnaecneenseee 100 Figure A 55 Classic snapping Option window ccccesssecessteceececesaeeceeaeeceeneeceeeeeceeeeecseeeeenaeees 101 Figure A 56 Classic snapping tolerance setting WiINdOW ccceesceceeececsseeeceteceeneeeeneeeeaees 102 Figure A 57 Create feature window line Option eeeesecssecsseceseeeeeecsaeceeeeseesseeesaeesaeenes 102 Figure A 58 Line option on Editor Toolbar s i2 5 acascscsachs ccc ceed jaczeds ds eens des eed es cdeed acai a eeeedds 102 Figure A259 Final b s links shapefile iiien Hah eaa E E net eG E a 103 Figure A 60 Data export window 252 205 4 6 2053 as ce ie Ace See ae 103 Figure A 61 Output feature class Window 2 1 4 4 2csceceisceiee tbs BAe ainda eeleeeee eee 104 Figure B 1 Forecasting Tool Input Sleetscctlei lt cceossercceagi sacceat sect ottemeiasisaccaterontetacceahascensis 108 igure Bo Empty Cell Messag inene a E A E E eres 112 Figure B 3 Empty Colored Cell nnrisinrsssisisisiniissisaesi iinis seiiet 113 Figure B 4 Delete the Old Sieets i255 iirinn eite t o a e alaei 115 Figure C 1 Mode choice framework of MNL models eeeceesseceeeeceeececeeeeeceeeeeceee
41. 1 Figure A 22 gt Condition window iisi ssc dascccsaesedeecneavacncegnndsvevenencses cbabavecendavasdisndeadedsenedtaasesysumnavents 81 Bisure A23 Network info Window ci esros eote yas er ee ei aioa s 82 Figure A 24 Re enable all the links sseesneeesesseseesesesessesserssesessresseserssresseserssresseserestesseseresresseeee 82 Figure A 25 Input files in the workspace sseseeeseseseesersseseessressrserssresseseresresseseresresseseresresseese 83 Figure A 26 Georeferencing toolba eessesesseseesresresressersesresstesttsersstesesetestesesetertessesetesresseese 84 1X Fig r A277 STUD carais ar i E a Sonne E TRES 84 Figure A 28 Images after using shift tool eeseseseeseseseessssessessressrserssresstseresressesererresseseresreeseese 85 Figur A 29 Link table Dutton 2sics 20 c22 2cedetversnntedesieneeee titer agra atten eantneesineerndeituneeds 85 Fis re A 307 Terme Table oozes hi iei osetia hes eta cen te ot et Res ieira sar e erates 85 Figure A 31 Add control point button 20 ceeescecceseccesseeceneecetseecensencoeteccenseceetneeconteecenteccenneese 86 Figure A 32 Adding control points sei ahes5 det oes Monch cde esol eons cage nl ese ies del ya ana 86 Preure A33 RMSE vale 3 jack cesies peas hcteexsleeez sede aelie e asees a a te eae ee esi iiaae ins 86 Figure A 34 Rectify option under georeferencing toolbar eee eee esseceeeceeeeesseecesecneeeseeeeaees 87 Figure A35 Savine themate sssi eels Beet eS i
42. 6 years Free 31 day monthly pass 40 00 Students ID required 0 65 Student semester 100 00 Medicare senior disabled 0 60 Source http www longviewtransit com ticket php 4 TRANSA Urban route map is available at http media gosanangelo com media static New_Bus_Schedule_for_web pdf 43 Longview Transit s routes and number of official stops are the following e 1 Mobberly LeTourneau Univ 5 stops e 2 Medical District Longview HS 4 stops e 3 Pine Tree Springhill 6 stops e 4 East Marshall Alpine 4 stops e 5 Loop 281 Silver Falls 6 stops e 6 MLK South Eastman 5 stops e 7 Hwy 80 West Gladewater Newest Route Monday Friday only 4 stops 6 2 4 Lubbock MPO The Lubbock MPO public transit system is called Citibus There are nine fixed routes throughout Lubbock as well as numerous campus routes to around Texas Tech Citibus operates from 5 25 AM to 7 45 PM Monday to Friday On Saturday the service begins at 6 45 AM and ends at 7 55 PM and there is no service on Sunday Citibus does not provide service on the following holidays New Year s Day Memorial Day Independence Day Labor Day Thanksgiving Day and Christmas Day Table 6 4 Table 6 4 Transit fares for Lubbock MPO Cash fares Tickets and passes Regular fare 1 75 One day pass 3 50 Children age 6 to 14 yrs 1 25 Weekly pass 14 50 Children under 6 yrs Free Monthly pass 50 00 Medicare seniors 65 yrs or older 0 85 One day pass for Medicare
43. College Station MPO San Angelo MPO Longview MPO and Lubbock MPO 6 1 Selected MPOs The selection of the four MPOs studied in this section was primarily based on data availability based on the suggestions of the PMC In addition these MPOs were identified in the research described in Chapters 2 and 3 as those that could benefit from the incorporation of a mode choice component into their TDMs A mode choice component was highly recommended for Bryan College Station given the high rates of population growth in the last 10 years the diversity in modal shares in the area considering the Texas context and the long range transportation goals defined by the MPO A mode choice model component was also suggested for the San Angelo Longview and Lubbock MPOs Among these San Angelo represents small MPOs in Texas while Longview MPO and Lubbock MPO represent medium sized MPOs Their locations are presented in Figure 6 1 Z _ Bryan College Station MPO San Angelo MPO Longview MPO Lubbock MPO wn Figure 6 1 The selected Texas MPOs 6 2 Transit Characteristics 6 2 1 Bryan College Station MPO Bryan College Station s transit provider is the Brazos Transit District referred to as The District The bus system operates Monday through Friday from 5 AM to 7 PM and is closed on the following holidays New Year s Day Martin Luther King Good Friday Memorial Day 4 Independence Day Labor Da
44. E Figure A 11 Tools ribbon Save Your Work e Chose File gt Save or click Hl on the toolbar Provide the file name and click Save At this point we have finished the overview map and are ready to begin the next step of skim development Creating the Centroid Set e Select the Node Layer from the drop down menu on the toolbar as shown in Figure A 12 If the Node layer is not shown in the drop down list follow the steps mentioned in the Add the Network Layer and Node Layer section e To view Node layer data click on the toolbar e Identify the field that indicates whether a node is a centroid Typically the node layer data table will have a field titled Centroid which indicates the type of node Sometimes the centroid nodes are given the same number as the TAZ to avoid extra work The user must identify the variable providing the information about node type In this example the centroid nodes are given the same number as the TAZ e To create the set of centroid nodes Chose Selection gt Selection by Condition or click gt z on the toolbar Type the condition in the Enter a Condition box followed by a name in the Set Name box and click OK as shown in Figure A 12 75 Select by Condition Dataview Nodes m Condition Builder estore Function Lst J Values of ID Me Previous Conditions Select from visible features only Figure A 12 Selection window TransCAD creates a se
45. E 37 5 3 1 Drive Alone and Car Sharing oesitsyiernetjs site iieis a e aS 37 Dado Transit eninin duh canis E a a e a AE E E OEE EE E AES E 39 5 3 3 Walk amG Bie yCle seccina e e E a a AE A ERSS 40 5 4 Summary and Next Stepsrra soen e E E T atts 40 Chapter 6 Transit Skim Generation for Texas MPOS sessoeesseessocssocesccsesocessecssccesooesoosssossose 41 Ol SSCIE CIEE MEP OB Ara eae ich AEE E E amend ayaa uate eae svg sas aaah Ce A aaleuan Nps 41 6 2 Transit Characteristics lt 3casssc avssessantadeysatianesedaviasj acta a a E a a a i 41 6 2 1 Bryan C Olle ge Station MPO nerien a a E a wits EE A S 41 0 2 2 Sat Angelo MPOn aii aae aa a a a A gee S 42 6 2 3 Longview MPO aaea E A E a A E E tenets 43 62 4 E bbock MPO Laineen aa a a e SS 44 6 3 Transit Skim Generation niseni nnsa n e E a a E a a ieia 45 6 4 Summary and Next Stepson nan a R e cide E a EASES 46 vii Chapter 7 Procedure to Prepare Data ccccccscscsssssssssssssssssssscssescssssscssssscsssssssssssesssssesseass 47 7 1 Procedure Tor Survey Data Extraction yicicasecsavesacyucedadssiiees scgan stead Does ast Seuedecesdcekademedseactaanten 48 TD Skim GEMEL AION seniii ar E E E A EAEE EEE A AET EEE A AAI 50 7 2 1 Skim Generation for Drive Alone Mod e cceccceceessececeessececeesescececcessaeeeesssseeeeeeeeaaes 50 7 2 2 Skim Generation for Carpool Mode 30 9 oy iaticienasnvscenieseuaaseass voeanastuuaceencianse ea aoe 51 7 2 3 Skim Generation for Transit Mode siscz
46. MPO MTP a P MPO Plan Fiscal Year 2010 2035 Medium large sized MPOs Hidalgo County MPO 2010 2035 MTP xX X X El Paso MPO Amended Mission r 2 x 2035 MTP 24 3 2 Recommendations Based on the information collected in Tables 3 1 to 3 3 Table 3 4 presents a recommendation on whether implementing a mode choice component has the prospect of adding substantial value given the population growth trends current modal splits and future planning priorities at each small and medium sized MPO in Texas e The first column classifies the population growth into three levels large population growth more than 20 medium population growth between 10 and 20 and small less than 10 e The second column non insignificant share of non auto modes evaluates whether the MPO non drive alone modal share is non insignificant or not A yes in the column means that the MPO has a non auto mode share considerably higher than the average across all MPOs e The third column evaluates whether the assessment of the MPO long range policies will benefit from a mode choice analysis or not A yes in the column conveys that the MPO has at least four strategic planning goals that are highly related to mode choices making a model choice model more important to incorporate e The last column provides a recommendation on incorporating a mode choice model in the TDM The assessment was defined in three levels Not recommended no diverse mode
47. MPOs This section discusses the steps involved for only one MPO Lubbock the same steps are assumed for the Longview MPO The mode choice survey data for both the MPOs was obtained from the Texas A amp M Transportation Institute with the permission of TxDOT The survey data consisted of following four files e Record Type 1 Household Information Data e Record Type 2 Personal Information Data e Record Type 3 Vehicle Information Data e Record Type 4 Activity Trip Data Each of the four survey files contains a unique household number for each of the households This unique household number is the key to finding information on households across different files The household information file contains information on household demographics which includes household size number of workers in the household number of vehicles owned by household household address etc The personal information file contains information about the individual such as gender age driver license status employment status etc The vehicle information file contains information about the type of vehicle car van motorcycle etc make model year of manufacture etc The activity trip file contains information on the purpose of the trip trip origin and destination locations mode of the trip trip arrival and departure time etc The survey recorded all the trips made by an individual in the household on the day of the survey To develop a mode choice model one ca
48. MPOs in Texas Champaign County Regional Planning Capital Metro MPO CAMPO Commission CCRPC Illinois San Antonio Bexar County MPO Lincoln MPO Nebraska SABCMPO Genesee County Metropolitan Planning Houston Galveston Area Council H GAC Commission GCMPC Michigan North Central Texas Council of Association of Monterey Bay Area Governments NCTCOG Governments AMBAG California Metro MPO Washington Based on a literature review of the MPOs listed in Table 5 1 we identified the components of the skims used in their mode choice models These components are presented in Table 5 2 and they represent a measure of impedance towards travel an increase in time or cost makes the mode less attractive Table 5 2 Skim components per mode Skims components Mode In vehicle Out vehicle Parking Number of travel time travel time Cost n feausters IVTT OVTT Drive alone v v v 4 Motorized Car sharing s V s modes Transit v v v 4 Non motorized Walk modes Bicycle v Table 5 2 shows that an in vehicle travel time IVTT matrix is used in every mode Some MPOs use distance instead of travel time for the non motorized modes However it is more appropriate to use the same measure of impedance for all modes and therefore travel time is preferred over distance In addition using travel time for all modes facilitates the comparison of coefficients across modes Out of vehicle travel time OVTT is only present f
49. North Central Texas Council of Governments NCTCOG Source Texas State Data Center 2011 Large MPOs population greater than 1 000 000 This report is divided into nine chapters and has five appendices including the guide with instructions for running the model The Forecasting Tool User Manual in Appendix B is also a stand alone document 0 6766 P1 Chapter 1 is an introduction to the project Chapter 2 provides a literature review of mode choice models Chapter 3 discusses how to incorporate a model choice model into a smaller medium sized MPO and Chapter 4 outlines how to develop a forecasting approach and model design Chapter 5 outlines the procedure to develop skims with Chapter 6 reviewing the procedures used to develop transit skims in four medium and small MPOs in Texas Chapter 7 outlines the procedure to prepare data for use in the model and Chapter 8 describes the model development and guide to utilizing the model Chapter 9 provides conclusions and recommendations for future work Chapter 2 Literature Review The initial task of the research study was to synthesize the available literature on mode choice models and develop an approach to assess the appropriateness of implementing a mode choice model in small and medium sized Texas MPOs This task also made recommendations regarding the incorporation of a mode choice step in Texas small and medium sized MPOs U S and Texas MPOs were reviewed to assess whether they have alre
50. O started with a highly restricted model specification in terms of alternatives considered and data used but are planning to improve the model for future TDMs For most MPOs the TDM validation process is quite helpful in discerning whether the mode choice model is correctly predicting mode shares and improving forecasting From an implementation perspective the small MPO reviewed CCRPC recognized that the development of TDMs including the mode choice component is a challenging task that requires identifying TDM s uses and benefits and the resources necessary to develop validate maintain and operate it Small and medium sized MPOs in Texas were analyzed to assess the appropriateness of developing a mode choice model in their areas Three variables were studied to make this assessment population growth modal shares and strategic planning goals As a result of this analysis we highly recommend incorporating a mode choice model step in the TDM of 3 MPOs 26 Bryan College Station MPO Laredo MPO and Killeen Temple MPO do not recommend implementing a mode choice model in 3 urban areas Sherman Denison MPO Amarillo MPO Midland Odessa MPO and recommend incorporating a mode choice model in the other 15 MPOs In conclusion the private automobile s dominance among travel modes used in Texas urban areas highlights the importance of developing the technical ability to evaluate multimodal projects that attempt to increase the shares of n
51. T is calculated in the following manner For a CBD area add 1 5 minutes on both ends of the trip For example if an OD pair happens to be a CBD CBD pair the total waiting time will be 3 minutes o Fora CBD fringe area add 1 25 minutes to both ends of the trip o For urban and suburban areas add 1 00 minute to both ends of the trip Using these assumptions we created OVTT skim for each TAZ pair depending on the type of area as mentioned in the IVTT skim generation description the TAZ attribute table provides information on the area type for each TAZ e TAZ TAZ distance The distance between each TAZ pair was calculated using the exact same approach as generation of IVTT In TransCAD one can specify additional attributes such as length or cost during IVTT estimation and TransCAD calculates the total length or cost for each TAZ pair along the shortest path e Travel cost From the literature we found that the range of travel cost mile varies from 12 8 to 21 2 mile with an average value of 17 7 mile CAMPO 2013 and Arizona Daily Star To generate the travel cost skim for the drive alone mode we multiplied the average cost miles 17 7 by the corresponding TAZ TAZ distance to get the total travel cost 7 2 2 Skim Generation for Carpool Mode e In vehicle travel time The IVTT used by the carpool mode is same as the drive alone mode o Out of vehicle travel time The OVTT used by the carpool mode is same as the drive alone
52. Technical Report Documentation Page 1 Report No 2 Government 3 Recipient s Catalog No FHWA TX 14 0 6766 1 Accession No 4 Title and Subtitle 5 Report Date Developing a Mode Choice Model for Small and Medium February 2014 Published September 2014 MPOs 6 Performing Organization Code 7 Author s 8 Performing Organization Report No Dubey S Deng J Hoklas M Castrol M Loftus Otway L 0 6766 1 and Bhat C 9 Performing Organization Name and Address 10 Work Unit No TRAIS Center for Transportation Research 11 Contract or Grant No The University of Texas at Austin 0 6766 1616 Guadalupe Street Suite 4 202 Austin TX 78701 Sponsoring Agency Name and Address 13 Type of Report and Period Covered Texas Department of Transportation Technical Report Research and Technology Implementation Office November 2012 December 2013 P O Box 5080 Austin TX 78763 5080 14 Sponsoring Agency Code Supplementary Notes Project performed in cooperation with the Texas Department of Transportation and the Federal Highway Administration Abstract This project developed a process and framework for generating the inputs needed for estimating a travel mode choice model that includes the transit mode and developing a framework for implementing the results of an estimated travel mode choice model to project mode shares in response to demographic changes and improvements in transit service In generating inputs for estimat
53. UATS Transportation Model Report Appendix 3 Commute Solutions Texas Not dated The True Cost of Driving Available at http commutesolutions org external calc html Federal Highway Association FHWA 2010 Status of Travel Model Improvement Program TMIP peer reviews Capital Area Metropolitan Planning Organization CAMPO Available at http www fhwa dot gov planning tmip resources peer_review_program campo index cfm 65 Genesee County Metropolitan Planning Commission GCMPC 2009 Model development and validation report Genesee County urban travel demand model improvements Houston Galveston Area Council HGAC 2012 2009 model validation and documentation Regional travel models Koppelman F S Bhat C R 2006 A self instructing course in mode choice modeling multinomial and nested logit models Prepared for U S Department of Transportation Federal Transit Administration Lincoln Metropolitan Planning Organization Lincoln MPO 2011 Travel demand model Model development and validation report Longview Transit Agency http www longviewtransit com routes php Longview MPO Traffic Operation through Public Transportation Plan Available at http mpo longviewtexas gov metropolitan transportation plan 2035 Lubbock Transit Agency http www citibus com page routes Lubbock Avalanche Journal http lubbockonline com stories 120104 edi_120104016 shtml Metro 2008 Metro travel forecasting 2008 trip based demand m
54. ady estimated calibrated and validated mode choice models The research team opted to focus on developing a framework only for home based HB trips to work This decision was reached mainly because in urban areas much emphasis has been placed on modeling mode choice for HB work trips primarily driven by the concentration of such trips during the morning and evening rush hours This decision was also taken because the primary audience for this research is TxDOT and Texas MPOs who are evaluating the need and therefore procedures for integrating a mode choice model into their TDM The synthesis however may also be useful to technical staff at other state Departments of Transportation DOTs MPOs transit agencies and planning agencies involved in travel demand modeling The objectives of this initial task were to 1 investigate the general methods and procedures adopted by MPOs across the U S that have incorporated a mode choice component into their TDMs Specific issues of interest include the alternative conceptual structures inputs outputs and the model formulation the steps taken to develop and implement mode choice models model estimation calibration and validation procedures and model application procedures 2 identify the challenges faced in the model development and application and document lessons learned and 3 develop an approach to assess the appropriateness of implementing a mode choice model for a specific urban area
55. an mode i for individual g gt a Xx Value of attribute k for mode i and individual q The previous utility specification is developed at the individual level However as discussed in Section 4 1 the Texas Package uses TAZs and not individuals or households as the unit of analysis Thus to incorporate a mode choice component into the Texas Package the model needs to be customized appropriately We recommend a forecasting approach to incorporate such models into the Texas Package 4 3 Model Specification To incorporate the disaggregate mode choice model of Equation 4 1 into the Texas Package the model should include demographic variables as dummy variables values of zero and one only and other travel system attributes at the TAZ level Table 4 1 lists some of the attributes that can be used by small and medium sized MPOs in Texas e For demographic variables the number of categories will depend on data availability and the particular characteristics of the study area These variables have to be included as dummy variables For example some MPOs may categorize household size in five groups 1 person 2 persons 3 persons 4 persons 5 persons or more while other MPOs may use only three groups 1 person 2 persons 3 persons or more 31 e Travel system attributes vary for different modes travel times tend to be higher for transit modes compared to auto These attributes may also vary by trip purpose For example total
56. and then taken through trip distribution followed by a conversion from production attractions to origin destinations The resulting traffic analysis zone TAZ to TAZ person trips by household size category and income level category can be taken as input by the estimated mode choice model to determine modal shares and thus zone to zone motorized vehicle trips The estimated models have been embedded into a software forecasting platform to predict modal share shifts between each pair of TAZs and the region as a whole arising in response to changes in income levels and or household size over time The models can also be used to assess the impacts of transit improvements in terms of in vehicle transit times as well as OVTTs such as increasing the frequency of service We should point out however that the model specifications embedded in the software platform need substantial improvement before actual implementation of the software very few transit riders appeared in the household survey data sets used in estimation This factor in addition to the usual difficulty in disentangling time and cost effects from observed revealed preference data led to a specification that has left very substantial room for improvement For a trip based mode choice model one possibility to improve the specification is to use additional data from on board transit surveys to increase the number of transit users in the mode choice estimation data set 63 Four inter
57. assification codes Please note that the area classification provided here is an example of the many classifications used by various metropolitan planning organizations MPOs Users can choose any classification based on their requirements Table B 2 Out of Vehicle Travel Time Based on Area Type Area Type Classification Code alae e oye Central Business District CBD 1 1 1 5 mins CBD Fringe 2 1 25 mins Urban and Suburban 3 amp 4 1 00 mins Travel times used by the Capital Area Metropolitan Planning Organization CAMPO INPUT Sheet Details Table B 3 provides the detail of the main sheet named INPUT where users can change the value of various inputs Table B 3 INPUT Sheet Detail Input Name Description Provide the total number of internal TAZs This number must not Number of TAZ exceed the size of skim sheet matrix Ber Mile Gas Costin Provide per mile gas cost for Drive Alone mode dollars Average Bike Speed Provide the value of average bike speed Generally a value of 11 mph mph is used by various MPOs Average Walk Speed Provide the value of average walk speed Generally a value of 3 mph mph is used by various MPOs Transit Fare dollars Provide the transit fare applicable to the area under analysis Provide the total number of area classification used in the analysis Number of Area This number should exactly equal the number of rows in the Area Clas
58. ban rural Income 3 categories less than 30K between 30K and 75K greater than 75K Household Number of autos in household variables Number of persons in household Auto Indicator 1 if fewer autos in household than person 0 otherwise NL models were developed for all HB work and non work purposes and an MNL was used for NHB trips However different nesting definitions and explanatory variables were used for each trip purpose Figure 2 6 presents the nesting structures used by NCTCOG Several constraints were imposed during the estimation process For HB work trips transit fare coefficient and auto fare coefficient were set to 0 550 and 0 770 to match the national averages For HB non work trips auto in vehicle time coefficient and transit fare coefficient were constrained to 0 016 and 0 008 respectively For NHB trips the auto and transit IVTT were constrained to 0 011 and 0 007 respectively Similarly the cost coefficient for two modes were set to 0 200 17 HB work trips Choice n SA Drive Shared alone drive Transit Shared 2 Shared 3 Walk Auto persons persons access Access HB work tri non work trips Giclee Non Auto Transit motornzed Dri Shared 2 Shared 3 rive ared ared i i alone persons persons Bus walk Bus drive Walk Bike Figure 2 6 NL model structure of NCTCOG s mode choice model 18 Chapter 3 Incorporating a Mode Choice Component for Small and Mediu
59. banized area which comprises a population of 144 361 comparable to small MPOs in Texas e As part of the 2025 long range transportation plan the study area was divided into 206 TAZs The main source of data was the Champaign Urbana Urbanized Area Transportation Study s 2002 2003 Household Travel Survey Trip productions and attraction rates are estimated for five trip purposes HB work HB school HB shopping HB other and NHB A cross classification method was used to model trip productions based on household size data Trip attraction rates were borrowed and modified from the NCHRP Report 365 Lincoln MPO Nebraska e Lincoln MPO is located in Nebraska and serves the Lancaster County area with a population of 285 407 comparable to medium small sized MPOs in Texas e The study area was divided into 502 TAZs The main source of data was the North Front Range Household Survey which was the dataset most similar to the Lincoln MPO s Trip productions and attractions rates are estimated for seven trip purposes HB work HB shopping HB recreational HB university HB other work based and other A cross classification method was used to model trip productions based on household size and income data Trip attraction rates were taken from the previous Lincoln MPO model and adjusted to balance trip production rates Genesee County Metropolitan Planning Commission GCMPC Michigan e GCMPC is located in Michigan and serves the Genesee
60. based on modal shares and the range of transportation planning needs policy questions project evaluations and travel demand forecasting exercises being considered The remainder of this chapter is organized as follows Sections 2 1 through 2 3 provide an overview of the current practices in mode choice modeling in the U S with a particular emphasis on Texas MPOs Section 2 4 develops an approach to assess the need to incorporate a mode choice component into the TDM of small and medium sized MPOs in Texas The final section of this chapter summarizes salient findings 2 1 Mode Choice Models 2 1 1 Overview Mode choice models provide the means to evaluate the ability of traffic congestion mitigation efforts to effect a change in travelers mode of travel from solo auto to high occupancy vehicles and non motorized modes of travel Koppelman and Bhat 2006 developed a self instructing manual on travel mode choice analysis that is now widely used by practitioners in the consulting arena as well as at MPOs As Koppelman and Bhat 2006 indicate some of the common types of independent or exogenous variables used to explain individual mode choice behavior include traveler characteristics trip characteristics and transportation system characteristics see Table 2 1 The models are separately estimated by trip purposes and sometimes further segmented based on statistical and intuitive considerations by traveler and trip characteristics or t
61. button Empty Cell Warning Figure B 2 Empty Cell Message 112 n hor E P Wsp Tot General E J noema tad Good Neant Gleulation E x Eaten oy A SYRURRRGHRS sees en 4 83 a6 7 81 Sears 3 3 s 1 18 549 s09 s an 19 255 sn s 62 23 3 5 sis 1638 435 764 s sat 5 63 n nn a 47 aw aes s ao osa 332 s sz 32 s en s osi w m su no 49 oin aas o ons ano nm 6s an ee ovens TT Eo eee OSU Oi ARATE OBTANCEDA INTTTRINSTT OVILIRANSIT OSTANCETRANS Figure B 3 Empty Colored Cell Upon identification of any type of unreasonable value a message will pop up asking the user to correct the value The program will not run until everything is corrected Now assuming that data in all the input sheets has passed the data consistency check a new sheet named OVTT_DA will be generated by the program containing the value of out of vehicle travel time for the Drive Alone mode for each of the TAZs depending on the TAZ area type Now depending on the data availability enter 1 in the Individual level estimates or TAZ level estimates option cell and click button CLICK FOR MODE SHARE 3 This button will generate the mode share output files Please note that all the output sheets which are program generated will have pink tabs Assuming that the user provided the individual level trip distribution data and chose the option Individual level
62. ce and number of transfers for transit Walk access and egress time for transit and walk access distance to transit used to determine transit availability In estimating mode choice models four elements are important to consider the decision maker the alternatives the attributes of alternatives and the decision rule 1 Decision maker The decision maker is a respondent in the survey who is observed to make a choice of mode for one or more trips of a specific purpose 2 Alternatives Individuals make a choice from a set of alternatives available to them The availability of an alternative for an individual in the context of travel mode choice may be determined by legal regulations a person cannot drive alone until the age of 16 or the non availability of a vehicle It also is typical to assume in mode choice models that transit e g bus is an available mode for an individual only if the transit stop is within 0 25 miles of the origin end and the destination end 3 Attributes of the alternatives The alternatives in a choice process are characterized by a set of attribute values as encountered by a specific individual Attributes include the transportation system characteristics such as travel times and costs 4 Decision rule A decision rule is a mechanism to process information and to evaluate alternatives Traditional mode choice models are based on utility maximization theory which assumes that when faced with a choice
63. ces Be E KERER E 87 Preure A30 Are atalos Window siisii e e E E TE A T 88 Figure A 37 Adding shapefile to the working folder eee eeeesscecsseceseeeeeeteneecsaeeneeeeeeeeneees 89 Fig re AAS Shape fle PrOperey OK sees So seen SAGS acs los cutoconss aan n ea aE Eaa ER Sa iaiT 90 Figure A 39 Coordinate system window s3 cce wean wae eM egies eaee 90 Figure A 40 Creating feature dataset iavcckcas so ccns co coals ce cael Socata cess pects tiuta ned Seated Weems asia 92 Figure A 41 Feature class enim Onss Gee s5 coe Mactagccceeld oct soc oe ocean teers elaine aes 93 Figure A 42 Shapefiles in table of contents window esceeseceseceseeeseecnaeceseeeseeesaeecseeeseeenneees 94 Figure A 43 Start editing option BOK aivescesccascuceseedonnsneicbpensavegsioesnde susesd ques tyesheavaasuatonegaouadecest nested 94 Figure A 44 Create feature windOWs ii2i 2isineaiiesddeislad niiehs delat hha 95 Figure A 45 Point option on Editor Toolbar c sctcnstriasopoesecrit sat cteetel elem ortati int cakes 95 Figure A 46 Save edits and stop editing Option 0 eeecceesseceeneceeseeceeneeceeeeeceeeeeceeeeecneeeeneeeees 96 Figure A 47 Attribute table for bus Stops vio oa eS i es 96 Figure A 48 Add field option to table 225 6 cede en Ruts halve eis ep Boe 97 Figure A 49 Defining field name and type 2 lt 0 23 9s scascsaceee ivendsai gosaes taspccassoremes iavedoeayemeeee ee 97 Figure A250 Calculate X co rdinaat S nr e a E S A E A E 98 Fig r A3
64. ch you will digitize fields If the coordinate system is undefined click on Edit to select from the various coordinate system options If you know the coordinate system of the registered image exactly assign the same to the Shapefile otherwise leave it undefined Note undefined coordinate system files do present a problem the user does not know the units of measurement Once the file is created with appropriate data trial and error is required to determine the unit of measurement For example once the bus stops are mapped into the shapefile we assign a certain coordinate system to the shapefile and determine the distance between them using shapefile units and compare the obtained value with the true value The procedure is repeated with different coordinate systems until a satisfactory result is obtained To assign a coordinate system at the time of the shapefile s creation click the Edit button to the define coordinate system given various 89 options If the shapefile is already created right click and select Properties Both of these options lead to the same window as shown in Figure A 39 4 E Show Details E Coordinates will contain M values Used to store route data E Coordinates will contain Z values Used to store 3D data select Select a predefined coordinate system Import a coordinate system and X Y Z and M domains from an existing geodataset e g feature dataset feature
65. choices medium to small population growth goals do not relate to mode choice Recommended one or two of the following diverse mode choices large population growth and goals highly related to mode choice analysis Highly recommended the MPO has diverse mode choices large population growth and goals highly related to mode choice analysis The literature review of mode choice models in MPOs outside of Texas shows several differences in approaches MPOs use a variety of models although the MNL and NL models are the most prevalent All MPOs developed different mode choice model for different trip purposes HB work HB non work and NHB were the most common categories used The number and type of choice alternatives considered varies among MPOs based mainly on region specific characteristics In Texas only four MPOs have developed a mode choice model Among them CAMPO and SABCMPO rely on TxDOT s Texas Package for travel demand modeling Therefore their modeling approach can be adapted by small and medium sized MPOs that also rely on the Texas Package for modeling purposes The four MPOs use an NL model but the number and type of alternatives vary among regions along with differences in the manner of the disaggregation of trips by purpose and time period The HGAC mode choice model is far more detailed in terms of the representation of modal alternatives However the NCTCOG mode choice model appears to use the most detailed data input
66. ciscssencs ieavcecaacasccived sais sasscaes eiecataienidersseedcaaasetiants 52 7 2 4 Skim Generation for Walk Mode ssccccssssssceceesesececsssseeecsesnecececseaeeecsssneeeceeesenaes 53 7 2 5 Skim Generation for Bike Mode vxieicsa escneiisa sed iden adec bate tisndiep cod eialusinadiaeoks 53 T 3 Summary and Next Stepsisccissacecccnssescsassjeaedsanewcseisngeaesasuaeadsasegeausvaveatevsnceceensbensaeesepaaoeeannces 53 Chapter 8 Model Development ssesssocssooesooesooesssecssccesocesooesoocesocessocesocescosesosesocessocssoossoosssosesse 55 8 1 Introduction and Overview essssssesessseeeessereessssstesssseresssrreessereesssstessssrressseeeesseseesssstesssse 55 Be Lubbock MPO erinan e i vat ded A E E E E A E 55 8 2 1 Traffic Analysis ZOneGS i scccsisisdenecvavnsetassidvedeaavessesaus EAEE S E EAEE EES aSa EA 55 P OTa AES EE E EET pss sues es se N ET 55 8 2 3 Network and Level of Service Preparation ccessccceseceseseceesseceeseceeseceesaeeeeaeeeeas 55 8 2 4 Explanatory V ariablesi iccciscacsastvesgicetincasbacsassiactvuigoaaaasuaeseas dey sadeandenda eeee joacteunasen ts leseade 56 Ci Dataren e eens an tess tude ore eau cuat E iA ies ae cock sah as coe A ies teas 56 8 3 Lons view MPO mre ing as heasn ceed cataah a eg wear Sean ons eRe See a AER 59 Oy Traffic Analysis Zoe 4026S Oh E Ae ale es Vee Rd ads is ee Se 59 8 3 2 MOIES hy r a a a ENE E seas A EO too ESSE EESE 59 8 3 3 Network and Level of Service Preparation
67. cision maker that has to choose one alternative among several available alternatives as depicted in Figure C 1 The multinomial logit MNL model is a discrete choice model that allows researchers to identify the factors influencing mode choice and forecast a future scenario to evaluate transportation policies Drive Figure C 1 Mode choice framework of MNL models The MNL model is based on the utility maximization theory The utility function has two components a deterministic or observable component that represents the portion of the utility observed by the analyst and an unknown or unobserved component Formally the utility is as shown in Equation C 1 Equation C 1 MNL utility function U true utility of the alternative i to the decision maker q deterministic or observable portion of the utility U i Vi HEj Vi estimated by the analyst for alternative i and decision maker q _ error or the portion of the utility unknown to the analyst T for alternative i and decision maker q The deterministic or observable portion of the utility of an alternative V is a mathematical function of the attributes of the alternative and the characteristics of the decision maker The systematic portion of utility can have any mathematical form but the function is most generally formulated as additive to simplify the estimation process as shown in Equation C 2 117 Equation C 2 Deterministic component of the MNL utility
68. d income and level of service variables travel time and cost Further the household size was divided into three categories one person household two and three person household and four or more person household Annual household income was also divided into three categories less than 25 000 between 25 000 and 50 000 and greater than 50 000 Both the household level attributes were used as indicator variables in the mode choice model One of the main reasons to include only two household attributes is to maintain consistency with the Texas Package The Texas Package uses household size and income as two independent variables in the trip generation and trip distribution step One main goal of this project is that MPOs should be able to use this mode choice model for policy evaluations such as change in mode share due to improvement in transit service Hence in the future when MPOs use this model they won t need to collect any additional data as household size and income variable is readily available from the Texas Package 8 2 5 Data The survey data for the model development was obtained from TxDOT The survey data corresponds to the year 2005 After careful examination and refinement of survey data a sample size of 1975 individual HB work trips was obtained These steps were followed in data preparation e We extracted the trip information for each household This information includes unique household number person number indicate
69. d their median access type divided undivided or continuous left turn These characteristics are obtained from the transportation network database maintained by the MPO TxDOT adds further information on road operational characteristics including the area type link capacity and speed and link length All of these road characteristics supplied by the MPO and TxDOT are then applied to create a matrix of network travel times from zone to zone These inputs with some additional information are taken into the software to produce individual trip tables These trip tables are then imported into TransCAD and altered into an O D matrix which is ultimately translated into 24 hour vehicle O D trip tables STEP 3 Traffic assignment assigns the trips to the network using TransCAD The traffic assignment models are based on a user equilibrium procedure and represent daily travel Besides the 24 hour O D trip table produced in the trip distribution step other inputs are the Level of Service E Capacities and the travel times No inputs in this step are provided by the MPO 4 2 Mode Choice Model Recommendations The literature review of mode choice models in MPOs in the U S in Chapter 2 showed that the MNL and NL models are the most prevalent The models are separately estimated by trip purposes and sometimes further segmented based on statistical and intuitive considerations To estimate such models urban household travel surveys of the type conducted b
70. day and therefore different transit skims have to be generated for each time period considered in the analysis In vehicle travel time Because the transit network shares nodes and links with the highway network IVTT for transit is computed as a function of auto travel IVTT Two approaches are used to compute IVTT e Travel time for transit lines is the congested time on roadway links This information is directly obtained from the network skims e The travel time for transit lines can be computed as a percentage of auto travel time For example IVVT for transit can be 25 higher than auto travel time Out of vehicle travel time The OVTT includes access egress time and wait time e Access and egress times can be computed from the transit network by measuring the distance between the transit network nodes bus stop locations and the zonal centroids at the origin TAZ for the access time and the destination TAZ for the egress time This distance is then multiplied by the walk speed 3 mph to obtain the time e In some cases the node centroid distance is very large and consequently the mode is not available for that trip This result is not necessarily because of great distance but due to the zonification To correct this problem MPOs define a maximum distance If the node centroid distance is larger than a certain value the distance is modified to this maximum value The maximum walking distance varies among MPOs but it usually
71. e bike speed 11 mph to obtain the bike travel time skim e TAZ TAZ distance Here we use the distance configuration matrix obtained during OVTT skim generation for the bike mode 7 3 Summary and Next Steps The estimation of mode choice models requires information on an individual s demographic and trip characteristics Demographic data and a certain set of trip characteristics such as the mode of travel purpose of travel location of travel etc are available from the survey data However data on other trip characteristics known as skims are generally not available from the survey and need to be constructed for each of the travel modes considered in the analysis In this chapter we extracted the required set of demographic variables and trip characteristics from the survey data for the Longview and Lubbock MPOs Further we also constructed the IVTT skim OVTT skim travel distance skim and travel cost skim for both MPOs using the guidelines developed in Chapters 4 through 6 We documented any assumptions made during construction of a skim All the assumptions were made after referring to other MPOs TDMs and skim generation guidelines assumptions to obtain consistent estimates In Chapter 8 we will estimate the MNL and NL model for both MPOs using the data prepared in Chapter 7 Further the estimated model coefficients will be used to develop an Excel based forecasting tool for mode choice at the TAZ level which will be discussed in
72. e ceeeeeseesseeeteeeeeeeeneees 10 Figure 2 2 NL model structure of AMBAG s mode choice model cee eeeeeseesteceseeeeeeeenees 11 Figure 2 3 NL model structure of CAMPO s mode choice model eeseseeeeeeereresserereerrese 14 Figure 2 4 NL model structure of H CAG s mode choice model ee eeeeeeeeseeseceeeeeeeeeeneees 14 Figure 2 5 NL model structure of SABCMPO s mode choice model ceeeeeeeeneeeseeeeeeeenees 16 Figure 2 6 NL model structure of NCTCOG s mode choice model eee ceeeeeseesseeeseeeeeeeenees 18 Fis re4 l Texas Ac Ame AUS aone E E E vin al Walesa ce sa te eles wuanteg EE ca waa es aaoh ou dene 29 Figure 6 1 The selected Texas MPOS ict jscccsscasadeacagsiccanusiaaetunssuatasnanad Soccescnca isie ii eiii niii 41 Fig t 7 1 Selected study ar dan a teal E E A E a 47 Fiure AT File selection Men s nscsi ies t e eet e aaea E E eE Oaa A EdE SOEST 70 Figure A 2 Import shapefile window isicscccccccssccsante cass cacesesdansteosducds vad cdassustacsdaegesaeaseuacss donddaeaneceass 70 Bist A 3 Bile save Menu siei iei Eaa ea aiene aS sar aaa RaT aoa 71 Figure A 4 Geographic file selection window ccsssccesesecesececeeeeeceseeecnsccecesceecssceecnseeeeeseeeees 71 Figure A 5 TAZ visual setting Window sssssesssesssssesssessessseeesseeesstesseesseessetessseessressesseeessessees 72 Figure Av Os Layer addition WTO Wy csigncts oie us sensncky ce Saved RE och tog ae asad a ae eee Ae 72 Fi
73. e component in their TDM 3 Strategic planning goals Transportation planning involves identifying broad regional problems and challenges that the region expects to face over the next years In long range transportation plans also referred as Metropolitan Transportation Plans MTPs MPOs usually define their long term goals and strategies Because transportation is interconnected with health quality of life social equity and the environment several of these goals are strictly related to promoting the use of alternative transportation modes To examine the potential impact of such policies developing a travel mode choice model is important The population growth modal shares and types of policies being considered in the urban areas will shape the need for and the structure of travel mode choice models In the following sections we describe each factor in the context of small and medium sized MPOs in Texas 3 1 1 Population Growth Table 3 1 presents small and medium sized MPOs area and population As the table illustrates Texas MPOs vary widely in both the spatial area and population they serve Many small and medium sized MPOs for example Hidalgo County MPO Laredo MPO and Bryan College Station MPO have experienced significant growth in the past decade growing at even higher rates than the state average Overall small MPOs tend to have a percentage growth smaller than medium sized MPOs 19 Table 3 1 MPO population growth
74. e for the Lubbock area is as follows o 1 75 for adults o 1 25 for children of age 6 12 o 0 85 for senior citizens persons 65 years of age or older 1l Access time is defined as the time spent in reaching the bus stop from the origin location 12 Egress time is defined as the time spent in reaching the destination from transit stop 52 7 2 4 Skim Generation for Walk Mode e Walk time To calculate the walk time between TAZ pairs we assume an average walking speed of 3 mph Then we calculated the shortest path between each TAZ pair in terms of distance by constraining the highway segments Here we assumed that people tend to avoid highway segments when using the walk trip mode for commuting Next we divided the zone pair distances by the average walking speed to obtain the TAZ walk travel time skim e TAZ TAZ distance Here we use the distance configuration matrix obtained during the walk time skim generation for the walk mode 7 2 5 Skim Generation for Bike Mode Out of vehicle travel time To generate the OVTT skim for the bike mode we assumed an average bike speed of 11 mph We found that CAMPO Lubbock Avalanche Journal not dated uses a speed range of 10 12 mph for the bike mode Similar to the walk mode calculation we assumed that people do not use highway segments while using a bike in a commute mode Hence we use the same TAZ distance configuration matrix obtained for the walk mode and divided the distances by averag
75. e links 82 Development of Skims for Transit The Network file available from the MPO does not contain information on bus routes Hence to develop the skims for transit the user needs to construct an entire or at least partial transit network One strategy is to map the bus stops on the network file available from the MPO and perform further analysis to obtain transit in vehicle travel time The second strategy is to construct the entire transit network file add the necessary centroid links to connect bus stops with TAZ centroid and run the Multiple Shortest Path For both of these strategies the starting point is a digital image of the transit network easily available from the city transit operator website With the help of ArcMap a GIS software we can convert the transit network image into a geographical file which can then be used for skim generation The first step is to georeference the image Geo referencing is the process of assigning a coordinate system to any given image Georeferencing an Image e Open ArcMap Click File gt Open and select the Network file the file should in shp format We will call this the registered image e Add the transit network image the unregistered image to the workspace Don t be concerned if you are unable to see the image at this time The image should be in either a PNG or TIFF format Figure A 25 shows the two files in the ArcMap workspace Here the network file is titled Streets 20
76. e parameter values were fixed in order to facilitate consistent model estimation including IVTT parameter fixed to 1 0 wait time transfer time walk access time walk egress time transfer penalty time all parameters fixed to 2 5 and cost parameter fixed to 0 06 The TDM validation process showed that the model replicates base year travel for both highway and transit modes 15 Choice ee Auto Transit Non motorized r m Drive Shared 2 Shared 3 Bus walk Bus drive Walk Bike alone persons persons Figure 2 5 NL model structure of SABCMPO s mode choice model 2 3 4 North Central Texas Council of Governments NCTCOG The study area for the Dallas Fort Worth mode choice model encompasses the whole Dallas Fort Worth area The trip generation step uses information on population households median household income basic employment retail employment and service employment for eight trip purposes The trip distribution step is formulated using a gravity model The 1996 NCTCOG household survey data set was used for the purpose of model estimation This dataset was further enriched by adding data from the 1998 DART and the 1996 FWTA transit on board surveys Overall the dataset presented a mix of both auto and transit trips with a total of 56 095 trip observations A series of checks were performed to remove trips with incomplete information origin or destination zone missing chosen mode not availab
77. e public health e Reduce emissions and or protect the environment e Reduce energy consumption e Provide multimodal transportation options e Enhance integration and connectivity Several observations may be drawn from the information in the table First all MPOs seek to improve and or expand the public transportation system train and bus in the future except for the Abilene Brownsville and Hidalgo County MPOs Second most MPOs encourage the use of non motorized modes by adding on street bike lanes off street multi use paths and signed bicycle routes for bike mode choice and designing a network of sidewalks and multi use paths to accommodate pedestrians mode choice However only few MPOs articulate intent to improve public health in the long range transportation plan Most of them are concentrating on protecting the environment and improving the air quality to meet the National Ambient Air Quality Standards established by the Environmental Protection Agency Finally since most MPOs target development of transportation modes other than drive alone the provision of multimodal transportation options is quite common 22 Table 3 3 Strategic planning goals related to mode choices Improve Incentivize Reduce Provide Enhance Improve ne Reduce expand public non emissions multimodal integration MPO Source i A public energy transportation motorized protect the transportation and health consumption n ae sys
78. e specification and nesting structure see Figure 2 4 The explanatory variables used are IVTT wait time two categories less than 4 5 minutes and greater than 4 5 minutes walk time transfer time number of transfers transit fare drive to transit time parking cost highway operating cost tolls and residential density factor 13 Twa oven a ert Die a eee so To E ee wa os eee oe Eon war Jae Jon Source CAMPO 2010 Figure 2 3 NL model structure of CAMPO s mode choice model Choice E3 Drive Shared Walk Drive alone rid access access e 4 Commuter Express a Commuter Park and Kiss and Toll No toll Toll No toll Toll No toll Figure 2 4 NL model structure of H CAG s mode choice model 14 The model validation process showed that the number of highway trips obtained from auto trips was lower than expected because of inconsistencies in number of occupants per vehicle Then the HB non work and NHB models were modified auto costs were no longer shared among vehicle occupants and an additional household size variable was added to the models Finally the models were applied at the TAZ level and the mode specific constants were adjusted to match observed 1995 control values This last step was required for forecasting purposes 2 3 3 San Antonio Bexar County MPO SABCMPO The study area for the San Antonio encompasses five counties Bexar Comal Guadalupe Kendall and Wilson The development of the 20
79. eal Eat Work Related School Shopping Personal such as laundry or banking Social Recreation and Pick up Drop off of Others e Household income The household income was categorized into the following fifteen categories less than 5 000 5 000 to 9 999 10 000 to 14 999 15 000 to 19 999 20 000 to 24 999 25 000 to 29 999 30 000 to 34 999 35 000 to 39 999 40 000 to 49 999 50 000 to 59 999 60 000 to 74 999 75 000 to 99 999 100 000 to 124 999 125 000 to 149 999 and 150 000 or more Table 7 2 shows the mode share for both of the MPOs for the alternatives drive alone carpool transit walk and bike Since we report only the trips corresponding to HB trip purpose and five modes the total sample size and frequency for each of the modes reported here will differ from the total survey sample size and corresponding mode frequencies 49 Table 7 2 Mode share for the HB work trip based on survey data Shared Ride 2 99 501 33 278 Table 7 2 clearly indicates that the majority of mode shares belong to the drive alone and shared ride mode for both the MPOs The average distance traveled to reach the workplace is in the order of 8 10 miles for both the areas as shown in Chapter 8 s Table 8 3 and 8 10 This average distance could well be the main reason for low walk and bike mode shares as people tend to go no further than 1 mile for walk trips and around 3 miles for bike trips Further
80. eeenaeees 117 Figure D 1 Mode choice framework of NL models eeseseeeseeseesesereesessessresressrsrrssressrseresresse 119 List of Tables Table 1 1 Population based classification of Texas MPOS cessscessceceeneeceeeeeceeeeeceeeeeeseeeesaes 4 Table 2 1 Common variables used in travel mode choice MOdels eeeeeeeeeeeeeseeceeeeteeeeeeenees 6 Table 2 2 TDMs of MPOs outside of Texas wicccsedtecsie centecteed Ghee eesti eenns Gheiecneeeeeieds 8 Table 2 3 Mode choice models of MPOs outside Of Texas 0 ee ceeeeeeeescecsseceseeeeeeeeseecnaeenseeees 9 Table 2 5 List of explanatory variables in NCTCOG mode choice model eeeeeeeeeeeeeeees 17 Table 3 1 MPO population growth 3 is2sca 2scecetoas snacatesaeqecuaieg ec pcaareeeeatsbecatateteceadagacancasvertoat aseaaens 20 Table 3 2 Mode shares for HB Work 1psi4 05 Scacesaiactesondaa teen adanesemenleoo tah owcweas 21 Table 3 3 Strategic planning goals related to mode CHOICES ee eee eee eeeeeeeeeeeeeeceeeeeeeeeeeeeaeees 23 Table 3 4 Recommendation on incorporating a mode choice model in the TDM 26 Table 4 1 Attributes to incorporate in mode choice models 20 0 eee eeseceseceeeeeeseecsaeeeseenseeeeaees 32 Table 5 1 MPOs reviewed to develop the guidelines ssssnnnsnsnsssessssssseessesesssessseesseesseesseesseee 36 Table 5 2 Skim components per mode sssesssseesssesseesseeeseeesstetssttssersseessseessseessreesseesseesseeessete 36 Table 5 3 OVTT minute
81. el of service variables for all but transit mode The IVTT and travel distance for the drive alone and shared ride modes were determined by running multiple shortest paths in TransCAD To obtain the travel cost for the drive alone and shared ride modes a per mile gasoline cost of 0 25 was assumed A skim development guide is provided in Appendix A detailing all the steps involved in the development of level of service variables for the five modes in TransCAD and ArcMap In a similar fashion the travel distance and OVTT for walk and bike mode was developed given certain assumptions i e people tend to avoid freeways and highways when using the walk and bike mode to commute and the average walking and biking speeds are respectively 3 mph and 11 mph The level of service variables for the transit mode were developed using the transit network obtained from the transit operator s website The transit fare was also obtained from the transit operator s website For the Lubbock area the fare is 1 75 for the general public We would like to point out that the transit network obtained from transit operators website was not a geographical file but a simple digital image The process of creating a geographical file from a digital image is complex and thus it is not described in this chapter see the skim development guide in Appendix A 55 8 2 4 Explanatory Variables The mode choice model includes two household attributes household size and househol
82. en ArcCatalog by clicking on the toolbar This action will open a pane on the right side showing the working folder Figure A 36 Here the working folder is 6767_work Catalog ax e 34204 H a amp Location C 6767_work v Home Documents ArcGIS E Folder Connections E C 6767_work C Users Subodh Desktop Paper_da E C Users Subodh Desktop Paper_da E Toolboxes Database Servers E Database Connections Ka GIS Servers Figure A 36 ArcCatalog window e Click on the working folder in the ArcCatalog window and select New and then Shapefile see Figure A 37 88 e x e i B H h t Location C 6767_work v H GJ Home Documents ArcGIS ej Folder Connections E F C Us o E CAUse j amp Toolboxes Rename Database be Disconnect Folder E Database Gy GIS Serve Refresh Folder L File Geodatabase f Properties L Personal Geodatabase 3 Spatial Database Connection d ArcGIS Server Connection Layer Group Layer Toolbox dBASE Table Address Locator Composite Address Locator x XML Document Figure A 37 Adding shapefile to the working folder e This action opens a Shapefile box as shown in Figure A 38 Provide a name for the Shapefile and select the Feature Type Point to hold the stops and Polyline to hold the roads you will digitize also select a new polygon Shapefile into whi
83. eneration process Key Words 18 Distribution Statement Mode Choice Model Network Skims Metropolitan No restrictions This document is available to the Planning Organizations Urban Travel Demand Trip public through the National Technical Information Generation Trip Distribution Texas Package Suite Service Springfield Virginia 22161 www ntis gov Transportation Demand Model 19 Security Classif of report 20 Security Classif of this page 21 No of pages 22 Price Unclassified Unclassified 134 Form DOT F 1700 7 8 72 Reproduction of completed page authorized THE UNIVERSITY OF TEXAS AT AUSTIN CENTER FOR TRANSPORTATION RESEARCH Developing a Model Choice Model for Small and Medium MPOs Subodh Dubey Jun Deng Megan Marie Hoklas Marisol Castrol Lisa Loftus Otway Chandra Bhat CTR Technical Report 0 6766 1 Report Date February 2014 Project 0 6766 Project Title A Generic Mode Choice Model Applicable for Small and Medium Sized MPOs Sponsoring Agency Texas Department of Transportation Performing Agency Center for Transportation Research at The University of Texas at Austin Project performed in cooperation with the Texas Department of Transportation and the Federal Highway Administration Center for Transportation Research The University of Texas at Austin 1616 Guadalupe Suite 4 202 Austin TX 78701 http ctr utexas edu Disclaimers Author s Disclaimer The contents of this report re
84. equal to 1 if individual q belongs a HHsize3 B TT E to a household with 3 persons or more 0 otherwise total travel time between TAZs in minutes for mode i i transit or auto In Equation 4 3 all coefficients are individual specific sub index q The first coefficient is the alternative specific constant The second and third coefficients B and f are associated with the demographic dummy variables of individual g and the fourth coefficient P a is associated with the total TAZ to TAZ travel time independent of whether choosing transit of auto If there are K alternatives only K 1 alternative specific constants can be estimated see Koppelman and Bhat 2006 32 4 4 Forecasting Approach To forecast using the disaggregate mode choice model proposed in Section 4 1 one record for each demographic category needs to be created for each TAZ to TAZ pair For each of these records the disaggregate model would estimate a probability of choice by each mode Then these category specific mode probabilities would be applied to the TAZ to TAZ flows and then added up to determine total TAZ to TAZ by travel mode This method works because of the specification of dummy variables for the individual specific demographic variables In the previous example considering a pair TAZs three records are required household size of 1 person household size of 2 persons and household size of 3 persons or more The d
85. error terms Following the example of Figure D 1 the utilities for each mode are presented in Equation D 1 the subscript g is omitted for ease of presentation 119 Equation D 1 Utility expressions for NL model of Figure D 1 true utility of the alternative i i drive i bus i bike or i walk deterministic or observable portion of the utility estimated by the analyst for alternative i i drive i bus i bike or i walk U U drive V irive T arive V U pus Vpis Ebpus U sire Vike T E bike E E non motorized E E i non motorized U vaik Vai E walk error or the portion of the utility unknown to the analyst for alternative i i drive i bus i bike or i walk non motorized error or the portion of the utility unknown to the analyst for non motorized modes bike or walk It is convenient to interpret this structure as if there are two levels of choice even though the derivation of the model makes no assumptions about the structure of the choice process Figure D 1 depicts an upper level marginal choice among Drive Alone Bus and Non Motorized modes and a lower level conditional choice between Bike and Walk given that a non motorized mode is chosen 120 Appendix E Travel Demand Models of MPOs Outside of Texas Champaign County Regional Planning Commission CCRPC Illinois e CCRPC is located in Illinois and serves the Champaign Urbana Savoy Bondville ur
86. etwork Unfortunately the metropolitan area networks were not available at this point of the project and the only information the research team had are the network skims Therefore a different and simpler approach was used to generate the transit skims as follows e In vehicle travel time IVTT was computed as a function of IVTT for drive alone mode as obtained from the network skims provided by TxDOT In particular VTT for transit was assumed to be 25 higher than IVTT for auto for all TAZ pairs e Out of vehicle travel time The access egress time was defined as 3 minutes for all TAZ pairs and the wait time was computed as half of the headway based on the reported service frequencies Finally the OVVT is the sum of the access egress time plus the wait time e Cost A flat cost was used for all TAZs pairs equal to the regular fare e Number of transfers considered to be zero for all TAZ pairs 45 6 4 Summary and Next Steps The estimation of mode choice models requires an accurate representation of the transportation system serving the region In particular the model estimation requires that the different skims are generated for each mode In the next chapter we prepared data for mode choice estimation in one small sized urban area and one medium sized urban area following the guidelines proposed earlier In particular for the network skims we were able to obtain network data in order to then generate the transit network
87. ey data for a sample size of 1189 for the HB work trips Table 8 11 Mode share for Longview area Mode Frequency Percentage Walk 4 0 33 Drive Alone 1148 96 55 Shared Ride 33 2 78 Transit 1 0 08 Bike 3 0 25 Table 8 11 demonstrates that as in Lubbock almost all of the trips were made via the drive alone or shared ride mode The share for transit walk and bike is negligible A mode share distribution of this kind makes it infeasible to estimate a mode choice model with all five modes due to an insufficient number of observations for transit walk and bike modes Hence we borrow the level of service variable coefficient values from the Bhat and Sardesai 2006 study and adjust the alternative specific coefficients of the MNL model to represent the market share as obtained from survey data as was done for Lubbock area Table 8 12 provides the estimated 61 parameter values for the Longview area values that are highly similar to those in Table 8 5 for Longview Table 8 12 Mode choice model coefficients 3 Drive Shared Variables Aione Ride Transit Walk Bike Alternative Specific Caa i ees 3 20 1 85 3 00 5 65 4 75 5 36 6 05 10 51 IVTT min 0 035 0 035 0 035 lt OVTT min 0 070 0 070 0 070 lt Travel Cost Income less than 25K cents ORe 006 per I eee I ee Travel Cost Income between 25K
88. flect the views of the authors who are responsible for the facts and the accuracy of the data presented herein The contents do not necessarily reflect the official view or policies of the Federal Highway Administration or the Texas Department of Transportation TxDOT This report does not constitute a standard specification or regulation Patent Disclaimer There was no invention or discovery conceived or first actually reduced to practice in the course of or under this contract including any art method process machine manufacture design or composition of matter or any new useful improvement thereof or any variety of plant which is or may be patentable under the patent laws of the United States of America or any foreign country Engineering Disclaimer NOT INTENDED FOR CONSTRUCTION BIDDING OR PERMIT PURPOSES Project Engineer Dr Chandra Bhat Professional Engineer License State and Number Texas No 88971 P E Designation Research Supervisor Acknowledgments The authors wish to thank Wade Odell RTI Research Project Manager James Burnett Transportation Program and Planning TPP Project Advisor Gabriel Contreras TPP Project Advisor Greg Lancaster TPP Project Advisor George Petrek TPP Project Advisor Mike Schofield TPP Project Advisor and Janie Temple TPP Project Advisor Products Appendix B contains 0 6766 P1 Forecasting Tool User Manual also available as a stand alone document Accompanying CD The acc
89. function deterministic or observable portion of the utility Vi estimated by the analyst for alternative i and decision maker q K V gt XxX parameter which defines the direction and importance of a oe B qi the effect of attribute k on the utility of an alternative i for decision maker q Xjix Value of attribute k for alternative i and decision maker q The MNL model assumes that the error term components E are 1 extreme value or Gumbel distributed 2 identically and independently distributed across alternatives and 3 independently distributed across observations individuals The three assumptions taken together lead to the mathematical structure known as the MNL model which gives the choice probabilities of each alternative as a function of the systematic portion of the utility of all the alternatives The general expression for the probability of choosing an alternative i from a set of J alternatives is presented in Equation C 3 Equation C 3 Probability expression of the MNL model ev P qi probability of the decision maker q choosing alternative i P 4 7 a e deterministic or observable portion of the utility j l V estimated by the analyst for alternative i and decision maker q One of the most widely discussed aspects of the MNL model is its independence from the irrelevant alternatives IIA property The IIA property states that for any individual the ratio of the probabilities in
90. gure A 7 Layer visualization Window os 2c asscacqesadisnccastoceseioat snsiasesauqecaasapacamaouseeteaneate a aeaniee 73 Figure A 8 Layer visualization Window siti sescscsteracnisden ton Gotetiadatapieestedieniasem gat ecient asters 73 aitare NO LOOM Tei EE E EE EE E A E 74 Figure A 10 Visual adjustment window for layers eessccesssecescecesneeceeeeeceeeeeceeceeceeeeeeeeeeees 74 Fig r A 11s Tools tI DDO nonn ina sasa aa e ae A EE eTR a PREES 75 Fig re A12 Selection Window 4 acsue toed cect ncaeckosnacs eae n a aa aee a aesae Taas 76 Figure A 13 Viewing the centroid SObicisscsecessaccedesevenerdvosccedsvaieaeanavnracaien daeesasessdanelesaatesesineesoncaens 76 Figure A 14 Binary network creation window sssssssessessseeesssessseesseessersseessseesssressresseesseeesseee TT PTs A TSS ABUSE aves an da E pu EE E wa See E R AT ERN 77 Figure A 16 Multiple shortest path menu siis5 cc5 u52 cchia spccaseaavceiucsiaabead sbeeasadeaeeena ea ceaucasease slaseescenns 78 Figure A 17 Additional skim selection window ssssssesssesessseesseeessessresseresseeesseesseesseesseeesseee 78 Sire ATS OO DA a a ile ae 79 Fissure A PO NGAGE export WINdOW sosie ene e i ais i A E ea ane a ia 79 Figure A 20 Network setting WindOW ccccssssccsssccsessscssnsccsennncessccessssceenecsensteessacceensceesnecsees 80 Figure A 21 Network update Window lt sce caesdesads cecesshpebuseassinehetiel outed desta gubeded spaces natous soaden seen 8
91. he Feature Dataset and select New gt Feature Class Provide a name for the Feature class and select the type For bus stops provide a name for the final bus stops e g Final_Stops and select Point Features from the Type drop down menu For bus links provide a name for the final bus links e g Final_bus_links and select Line Features from the Type drop down menu see Figure A 41 Accept the default in the next window and click Finish Final_Stops Type of features stored in this feature class Type of features stored in this feature class ine Geometry Properties Geometry Properties Coordinates include M values Used to store route data Coordinates include M values Used to store route data Coordinates include Z values Used to store 3D data Coordinates include Z values Used to store 3D data Figure A 41 Feature class definition e Both the shapefiles should appear in the Table of Contents window on the left side see Figure A 42 If not add them manually 93 Gh bus_stops B Final_bus_stops B Final_Bus_Links E C 6767_work g Stopsl 3 O Streets2008 C 6767_work 3 O Transit_Map2 tif RGB MM Red Bandi Ml Green Band_2 W Blue Band 3 Figure A 42 Shapefiles in table of contents window e Select the Editor Toolbar To select the Georeferencing tool open
92. heir areas assign a fixed OVTT for drive alone mode depending on the location of the workplace as presented in Table 5 3 From this table small and mid sized MPOs can select the OVTT that better represents their region For car sharing the OVTT is assumed to be the same as for drive alone plus 5 minutes Table 5 3 OVTT minutes for drive alone mode Workplace location ae area type corre r aee Metro Central business district CBD 3 0 1 5 5 0 Other business districts 2 0 1 0 2 0 Residential 1 0 1 0 2 0 Rural 1 0 0 75 2 0 Cost Drive alone and car sharing mode costs include gas tires and maintenance related costs According to the Bureau of Transportation Statistics BTS 2012 the average automobile operating cost in 2011 was 19 64 per mile Car sharing costs can be computed as half the drive alone cost that is an average of 9 82 per mile Then the travel costs can be computed as the price per mile times the travel distance Parking cost Parking costs have been shown to have a significant effect on transit ridership levels and must be treated carefully MPOs have to collect information regarding the average parking cost in each TAZ For instance H GAC and NCTCOG estimated the actual out of pocket cost paid on a daily basis per vehicle If the information is not available parking costs can be defined by area type CBD other business districts residential and rural For example the parking cost for TAZs in
93. hicle travel time out of vehicle travel time travel distance and travel cost in order to calculate the mode share at both the traffic analysis zone TAZ level and individual level It also has the capability of obtaining the mode share given any change in the model explanatory variables such as in vehicle travel time out of vehicle travel time etc via the tool s scenario module Input Figure B 1 shows the main INPUT sheet of the tool the red tab the single input sheet tabs appear along the bottom To make the tool user friendly all the sheets in the tool are named according to their functionality The user needs to appropriately fill in all 11 sheets to run the tool In contrast to the main INPUT sheet the single input sheets have green tabs see Figure B 1 107 H S Ss FILE HOME INSERT PAGELAYOUT FORMULAS DATA REVIEW 2 m Record Macro Ya E Properties B Use Relative References S S View Code Vi Basic A Macro Security K sual Macros Add Ins COM Insert Design Add Ins Mode E Run Dialog Code dd ins Controls 36 E f VIEW ource Number of TAZ Per Mile Gas Cost in dollars Average Bike Speed mph Average Walk Speed mph 5 Transit Fare dollars Number of Area Classifications Number of Passengers in Car for Shared Ride Maximum Walk Time min Maximum Bike Time min Put 1 if you want to limit the availability of Walk mode ba
94. hlights calendar Year 2012 Available at http ftp dot state tx us pub txdot info trf crash_statistics 2012 01_2012 pdf The snapshot of travel modeling activities The state of Texas Available at http media tmiponline org clearinghouse FHW A E2 80 90HEP E2 80 9012 E2 80 90005 Snapshot_of_Modeling_in_Texas pdf TransCAD Transportation Planning Software User s Guide Caliper Corporation Texas State Data Center 2011 Descriptive tables population 2000 2010 total population by race ethnicity Available at http txsdc utsa edu Reports Subject Population aspx U S Department of Commerce United States Census Bureau State amp County Quick Facts Texas Available at http quickfacts census gov qfd states 48000 htm l 67 68 Appendix A Guide to Model Skim Generation Development in TransCAD and ArcMap 10 1 Development of Skims for Drive Alone and Shared Ride The development of skims refers to the process of developing the in vehicle travel time out of vehicle travel time travel distance and travel cost matrix for each traffic analysis zone TAZ pair for any given area Skims are the required model input in any mode choice model as they capture the traveler s sensitivity to time in vehicle and out of vehicle travel time cost and distance Required Input for Skim Development in TransCAD To develop the skims for any ground operated mode the user needs two files e Coded street layer network lines w
95. idance on how TxDOT can develop the skims for different travel modes The guidance uses the Texas Package as the basis for travel demand modeling and is therefore specific to the Texas context Appropriate procedures were also developed to determine when individuals consider or do not consider a travel mode as being available to them Following the approach described in Chapters 3 and 4 the framework was developed for HB work trips only The urban regions considered in this analysis were e Bryan College Station MPO e San Angelo MPO e Longview MPO e Lubbock MPO For each of these urban areas we obtained and assembled the following data is needed for generating skims for five travel modes e Drive alone e Car sharing e Transit bus e Walk e Bicycle The remainder of this chapter is organized as follows Section 5 1 outlines the skim components while Section 5 2 covers mode availability Section 5 3 presents the development of transit skims for the four selected MPOs 5 1 Skim Components The guidelines provided in this chapter are based on the review of five MPOs outside Texas and four MPOs in Texas Chapter 2 listed in Table 5 1 The MPOs outside of Texas were 35 chosen to represent different population levels and therefore different modeling needs The MPOs in Texas are those that have developed and implemented a mode choice model in their area Table 5 1 MPOs reviewed to develop the guidelines MPOs outside of Texas
96. ime periods This approach is used because the motivations preferences and modal choices for an HB work trip are very different from those for an HB shopping trip To estimate such models urban household travel surveys of the type conducted by TxDOT or TxDOT TPP are used to obtain information on trip mode choice traveler characteristics and trip purpose characteristics while supplementary land use and transportation system data are used to generate origin destination O D characteristics and transportation system characteristics these are typically developed at the level of the TAZ and appended to trips based on the origin and destination TAZs of each trip Table 2 1 Common variables used in travel mode choice models Factors influencing mode choice Examples Individual demographics age gender Traveler Household socio demographics income number of workers number characteristics of adults auto ownership level Household structure single adult nuclear family Trip purpose HB work HB non work NHB Trip O D characteristics area types of origins and destinations built characteristics environment measures at the origin and destination end distance between origin and destination Total travel time out of vehicle travel time OVTT in vehicle travel time IVTT for each travel mode Total travel cost for each travel mode and parking costs for auto Transportation system modes characteristics i Presen
97. ing a mode choice model an important component is network skims travel times and costs by alternative modes Most metropolitan planning organizations MPOs have good geographic information systems GIS based representations of the highway network which can be used to generate drive alone and shared ride skims However this is not the case with transit skims due to the lack of a good GIS based representation of the transit network especially for bus stops The project manually geo coded bus stop information onto the highway network and used assumptions to generate transit paths and corresponding zone to zone transit skims A guidebook provides a step by step procedure for developing skims The database for estimation was developed using household survey data 2004 on trip characteristics Two demographic variables were used in the mode choice model household size and income The models have been embedded into a software forecasting platform to predict modal share shifts between each pair of TAZs and the region as a whole due to changes in income levels and or household size over time The models can also be used to assess the impacts of transit improvements for in vehicle and out of vehicle transit times Further data collection from transit surveys is recommended to enhance the model s capacity to estimate the time and cost effects based on preferences A georeferenced coordinate system for bus stop locations would also improve the transit skim g
98. inutes in other areas although it can also be computed as half the headway 5 3 3 Walk and Bicycle In vehicle travel time MPOs compute the IVTT for walk and bicycle modes using a non motorized network based on the standard regional highway network excluding freeway facilities where bicycles and pedestrians are not allowed If this information is not available walk and bicycle distances are the same as auto distance Then using the distance among TAZs walk and bicycle travel times are calculated based on assumed speeds Most MPOs use uniform speeds of 3 mph for pedestrians and 10 12 mph for bicyclists to convert non motorized distance into travel time 5 4 Summary and Next Steps In this chapter we proposed guidelines to generate such skims based on a literature review within the specific context of small and medium sized Texas MPOs The guidelines proposed here should be evaluated by each MPO with the technical support of TxDOT to decide if the assumptions made here reflect the characteristics of their study area In addition data availability can make the process of generating the skims difficult particularly with the transit skims Then additional assumptions can be made to generate the skims as demonstrated at the end of the next chapter Section 6 3 40 Chapter 6 Transit Skim Generation for Texas MPOs In this chapter we review the procedures used to develop transit skims in four medium and small MPOs in Texas Bryan
99. isaggregate model Equation 4 3 estimates the probability of choosing transit and auto for each of these three records At this point six probabilities are computed e Probability of choosing transit by individuals that live in 1 person households Pansit szzsize e Probability of choosing auto by individuals that live in 1 person households Pi size e Probability of choosing transit by individuals that live in 2 persons households P transit HHsize 2 e Probability of choosing auto by individuals that live in 2 persons households Pig jtsizer e Probability of choosing transit by individuals that live in 3 persons households P transit HHsize3 e Probability of choosing auto by individuals that live in 3 persons households Pi Hrsize3 Then these household size category specific mode probabilities would be applied to the TAZ to TAZ flows by household size and then added up across the three household size levels to determine total TAZ to TAZ trips by transit and auto as shown in Equation 4 4 Equation 4 4 Example of computation of trips by mode TTri Total number of trips between the TAZs obtained l ponsi OP mika F E nU r ps from ATOM2 T E r Hiie x TTrips Tans Total number of trips by transit between the TAZs P auto HHsize1 Ean tied Pouto tise X T Trips Tuo Total number of trips by auto between the TAZs auto 4 5 Next Steps The purpose of this task was to provide model de
100. ith the following attributes length speed travel time one way capacity direction code link type and any other link attribute e TAZ layer with the following attributes TAZ type TAZ number TAZ centroid node number and any other information on the TAZ These two files should be available from the MPO in a TransCAD compatible format Possible File Formats To ensure compatibility with TransCAD the files need to be in one of these two formats gt cdf dbd Indicates a file format directly compatible with TransCAD gt shp Indicates a shapefile compatible with ArcGIS If the required input files are available in the cdf or dbd format the user is ready for the next modeling steps However if the input files are available in the shp format the user needs to first convert them into the format required by TransCAD by performing the following steps e Open a session of TransCAD and go to File gt Open and point to the required folder e If you do not see your shp file change the file type to ESRI shapefile shp at the bottom of the window under the title Files of type as shown in Figure A 1 8 Link type refers to the functional classification of the roadway facility such as state highway freeway arterial etc 14 TAZ type refers to the classification of an area into central business district urban area urban fringe rural area etc 69 Lookin Jp TAZ Sab z Date modified A Name d pai 4
101. k Even in cases where a GIS based transit bus network representation of stops is available our experience has been that these stops are not locationally consistent with the highway link networks In our project we had to manually geo code the bus stop information onto the highway network and then make assumptions to generate transit paths and corresponding zone to zone transit skims While we have provided a guidebook that provides a step by step procedure for doing so a good georeferenced coordinate system even just for stop locations will substantially help in the transit skim generation process In any case MPOs should be prepared to expend about 4 weeks of time to develop a good representation of the transit network from which to develop transit skims The estimation of the mode choice model given a set of trips and their characteristics can be achieved using any standard software package The database for this estimation may be developed using household survey data that provides information on trips trip characteristics origin and destination from which the network skims can be created and the characteristics of the individuals pursuing the trips In this project only two demographic variables were used in the mode choice model household size and income This approach was taken because the trip generation part of the Texas Package uses only these two demographic variables Thus person trips may be generated in the Texas Package
102. l trip productions based on income quartile age of head of household and auto ownership data Trip attraction rates were borrowed and modified from the NCHRP Report 365 when not available from survey data Metro MPO Oregon e Metro MPO is located in Oregon and serves Clackamas Multnomah and Washington counties comprising a population of 285 407 compared to large MPOs in Texas e The study area was divided into 2013 TAZs The main source of data was the 1985 Household Travel Survey Trip production and attraction rates are estimated for eight trip purposes HB work HB shopping HB recreation HB other NHB work NHB non work HB college and HB school A cross classification model was used to model trip productions that were based on household size number of workers age of household head number of children and household size by worker status Trip attraction rates are no longer computed except for HB work and HB college attractions which are calculated and then scaled to production rates 122
103. lable if the walking distance between TAZs is less than 5 miles or 30 minutes and the bicycle mode available when the distance between a TAZ pair is less than 10 miles 5 3 Skim Development 5 3 1 Drive Alone and Car Sharing In vehicle travel time The IVTT matrix for autos is produced in the Texas Package using TransCAD software and then exported as a binary file using a Texas Package utility This network skim represents 37 the daily travel times between all TAZ pairs derived from the minimum network travel time path for each TAZ pair This matrix can be used directly for both drive alone and car sharing modes Then no modification is needed with respect to the current procedures used by TxDOT Out of vehicle travel time If TxDOT or the MPO has detailed information about the parking lot location with respect to the workplace the walk time can be computed as the walk distance multiplied by the walking speed usually 3 mph Information regarding the parking lot location can be obtained potentially from workplace and special generator surveys For instance CAMPO computes the OVTT based on actual times skimmed from the highway network and walk times are based on a coded speed of 3 mph but capped at a maximum time of 10 minutes However small and medium sized MPOs generally don t have detailed information about parking location and therefore some assumptions have to be made The MPOs that have already implemented mode choice models in t
104. late out of vehicle travel time for the Drive Alone and Shared Ride modes This sheet accepts the travel distance matrix corresponding to the shortest path travel time matrix for Drive Alone mode i e TAZ TAZ travel distance This sheet accepts the in vehicle travel time matrix for Transit mode i e TAZ TAZ in vehicle travel time This sheet accepts the travel time to the nearest bus stop for a TAZ in a matrix format i e each cell represents the walking time to the nearest bus stop for the corresponding TAZ DISTANCE_TRANSIT This sheet accepts the travel distance matrix for Transit mode i e TAZ TAZ travel distance DISTANCE_BIKE DISTANCE_WALK This sheet accepts the travel distance matrix for Bike mode i e TAZ TAZ travel distance This sheet accepts the travel distance matrix for Walk mode i e TAZ TAZ travel distance 109 Please note that all the skims sheets IVTT_DA DISTANCE_DA IVTT_TRANSIT OVTT_TRANSIT DISTANCE_TRANSIT DISTANCE BIKE and DISTANCE WALK should be symmetrical the number of rows should equal the number of columns If the sizes for all the skims sheets are not same the program will return an error message In cases where no transit service is available between any TAZ pair simply provide a value of zero in the corresponding cell However for Drive Alone the user must provide a non negative in vehicle travel time and travel distance value Table B 2 provides the area cl
105. ldwell Hays Travis and Williamson counties CAMPO has a four step daily vehicle trip based model trip generation trip distribution mode choice and traffic assignment These are the basic steps of TxDOT s Texas Package described in Section 1 2 with the addition of the mode choice model step CAMPO compiles their data from many sources They conduct surveys in household travel workplace travel commercial vehicle external travel and on board transit They also obtain 24 hour traffic counts speed limit data and demographical data to help validate the models This data along with that provided by TxDOT allows CAMPO to run a successful model for their area CAMPO is one of the few MPOs in Texas that currently have a mode choice model implemented in their travel demand modeling system The data to estimate the mode choice model was obtained from home interviews and on board transit survey for year 2005 CAMPO currently uses an NL model Figure 2 3 shows the full flow chart of all 14 mode choice alternatives e Drive alone e Express bus PNR access e Shared ride 2 person e Express bus KNR access e Shared ride 3 person e UT Shuttle walk access e Local bus walk access e UT Shuttle PNR access e Local bus park and ride PNR e UT Shuttle KNR access access e Walk e Local bus kiss and ride KNR e Bicycle access e Express bus walk access Walk and drive times were obtained from highways skims A walking speed of 3 mph was used to es
106. le and auto ownership information missing which provided a final dataset with 35 377 observations The study considered the following five alternatives and three trip purposes HB work HB non work and NHB e Auto drive alone e Auto two occupants e Auto three or more occupants e Transit auto access e Transit walk access A wide range of explanatory variables were considered in the study Cambridge Systematics 2013 Similar to SABCMPO some parameter values were fixed in order to facilitate consistent model estimation including walk and wait time parameters fixed to 2 0 for HB work trips and 2 5 for NHB trips A detailed list of the variables included in the mode choice model is presented in Table 2 4 16 Table 2 5 List of explanatory variables in NCTCOG mode choice model Characteristics Variables considered Auto travel time Total transit travel time Auto out of vehicle time Transit walk access time Transit wait time Transit transfer time Level of service Auto access time transit auto access only variables Walk egress time transit mode only Transit out of vehicle time Fare Auto operating cost Parking cost Sum of operating and parking cost auto mode only Number of transfers transit mode only Population density at the production zone Employment density at the attraction zone Zonal variables ee l y t Type of attraction zone central business district other business district suburb ur
107. ly multiply the total trips going from Origin TAZ to Destination TAZ This completes the mode share calculation Next we discuss the scenario package of the tool which allows the user to obtain mode share given a specific set of changes such as change in in vehicle travel time etc Running the Scenario Module To run the scenario module enter 1 for Scenario Alternative Related Variables and click the button SCENARIO RUN 4 Before you click the button make sure that you have made the appropriate changes to the options Reduce Transit IVTT in Percentage Increase Drive Alone IVTT in Percentage and Increase Drive Alone OVTT in Percentage Please do not provide a negative positive sign before the numbers Once you click the button SCENARIO RUN 4 the same sets of sheets with extension _SCENARIO depending upon option Individual level estimates or TAZ level estimates will be generated reporting the mode share under the changed scenario All the files generated with an extension _SCENARIO have the same order of variables and meaning as their counterparts with no SCENARIO extension Some Useful Information At any point during calculation the user can shift from Individual level estimates to TAZ level estimates by placing 1 in the appropriate cell or vice versa However doing so invokes certain commands the next time when the user clicks the buttons the program will ask
108. ly one of the two options should be enabled at a time Number of Provide the total number of alternatives in the model It should be Alternatives equal to the number of alternatives provided in the model Number of Explanatory Variables Including Constants Provide the number as mentioned eae m Make this 1 if you want to create a TAZ TAZ skim configuration Configuration Household Category Provide the number of household category used in the model i e household classification based on household size Income Category Provide the number of income category used in the model i e household classification based on income range Scenario Alternative Related Variables Make this 1 if you want to obtain the mode shares under the scenario change option Reduce Transit IVTT in Percentage Provide a number to reduce transit in vehicle travel time by a certain percentage Enter only a numeral do not add a positive negative sign at the beginning or the sign at the end 111 Input Name Description Provide a number to increase Drive Alone in vehicle travel time by a certain percentage Enter only a numeral do not add a positive negative sign at the beginning or the sign at the end Increase Drive Alone IVTT in Percentage Provide a number to increase Drive Alone out of vehicle travel time by a certain percentage Enter only a numeral do not add a positive negative sig
109. m Sized MPOs in Texas To assess whether it is appropriate to implement a mode choice component in small and medium sized MPOs in Texas three factors were considered 1 Population growth According to the U S Census Bureau 2011 the state s population has increased by 20 6 between 2000 and 2010 This increase translates into 4 3 million people From a transportation planning perspective this rapid population growth is associated with more vehicles in the roadways and therefore increased travel times traffic congestion and greenhouse gas emissions The incorporation of a mode choice component in the TDM could help MPOs to understand and control the effects of a fast growing population 2 Mode choice shares Texas transportation systems are integral to the state s economic and functional viability and vibrancy providing accessibility for the daily travel needs of residents and tourists freight shipments and commuting trips While both roadways and public transportation systems are important to providing services for all residents more than 91 of Texas commuters use a personal automobile or carpool to get to work On the other hand less than 2 of commuters use the public transportation system and non motorized forms of transportation U S Census Bureau 2009 However in some urban areas in Texas the use of alternative transportation modes is more widespread therefore these areas may benefit from the inclusion of a mode choic
110. model is used for the mode choice step see Figure 2 2 estimated using data from the 2001 2002 Caltrans household survey The model structure was updated to comply with the Federal Transit Administration FTA guidance for New and Small Starts forecasting because some coefficients of the 10 previous mode choice model for year 2000 were outside the accepted FTA range and the model had county specific constants that are not allowed by the FTA The model is a three level structure for three trip purposes HB work HB other and NHB and divides the person trips into the eight alternative modes highlighted in Figure 2 2 The explanatory variables included in the model are IVTT OVTT wait time transfer wait time number of transfers operational cost and parking cost For transit related characteristics data was drawn from the transit network which consists of a description of bus lines that are superimposed on the road network Transit line characteristics include the locations of stops walk access links and peak and midday headways Transit speed was obtained by adjusting the average speed of all vehicles by link The TDM model was validated and a 40 75 root mean squared error was obtained for the predicted boardings Person Trips between Two TAZ s Motorized Trips Non Motorized Trips Walk Bicycle Automobile Shared Ride Walk Access 2 Person 3 Persons Premium Local Park and Ride Kiss and Ride Transit Transit Service Service
111. more the survey was conducted in 2004 and the bus connectivity may have been limited during that period contributing to the low transit ridership With better bus connectivity and better bus quality we can expect to see an increase in transit ridership in the future Therefore a generic mode choice model with a potential range of alternatives will provide enhanced variability for future mode share prediction Keeping this in mind we included five of the modes shown in Table 7 2 as alternatives in the mode choice model Section 7 2 shows the steps and assumptions involved in developing skims for all five modes 7 2 Skim Generation In this section we describe the steps and assumptions involved in development of skims for all the five modes Under the skim generation task we developed IVTT OVTT travel cost and distance skims for each of the five modes 7 2 1 Skim Generation for Drive Alone Mode e In vehicle travel time In order to generate the IVTT for each TAZ pair we used the TAZ and network file provided by TxDOT The TAZ and network file was available in the form of a shapefile A shapefile is a file that represents a geographic area in a visual form using the coordinate system A shapefile consists of a map shape of an area and an attribute table providing details about the area For example the TAZ shapefile shows the division of a geographic area into small blocks labeled as TAZs on a map and the corresponding attribute table pro
112. n at the beginning or the sign at the end Increase Drive Alone OVTT in Percentage The user must also provide values for the Area Type Wait Time table located between the main input listing and the parameters table as shown in Figure B 1 Do not change the position of any input item as doing so may cause problems during calculation Mode Share Estimation With everything set proceed to the four buttons that appear underneath the parameters table on the main INPUT page The buttons need to be clicked sequentially from 1 to 4 First click the button REFRESH ALL 1 This button s function is to remove any sheets left from previous runs After this click the button CLICK TO CHECK DATA 2 It checks for lack of data consistency such as empty cells inappropriate value etc and warns the user if any are found prompting the user to address any inappropriate values For ease in handling a message will pop up describing the problem along with the sheet name and the corresponding cell in the sheet will be colored green for easy identification For example Suppose that the first cell in the sheet IVTT_DA is empty When the user clicks the button CLICK TO CHECK DATA 2 a message will pop up see Figure B 2 and the program will be terminated The user now can go to the corresponding sheet find the empty cell highlighted with green as shown in Figure B 3 change the value accordingly and re click the same
113. n potentially include the whole range of demographic and trip characteristics in the model However in practice we include only the variables that are available used in the Texas Package We maintain this consistency to avoid any additional future data collection effort Hence we include only household size and household income as demographic characteristics in the mode choice model development Further from the activity data file we obtain the purpose of the trip as the focus of the current work is HB work trips mode of trip and trip O D in terms of a TAZ number For ease in readability Table 7 1 provides a list of variables whose extraction from survey data is required for the mode choice model estimation 7 The mode choice model will be used to predict the mode share for the future year Hence including only the demographic variables available in the Texas Package ensures we will not need to collect additional data in upcoming years 8 The Texas Package uses household size and household income as explanatory variable in the trip generation step and hence this data is available for each of the TAZ 48 Table 7 1 List of demographic and trip variables Household Size Demographic Variables Household Income Purpose of Trip Trip Characteristics Mode of Trip O D of the Trip at TAZ Level Given the list of variables required for the mode choice model estimation we adopted the following steps to extract the required variables from
114. n you are happy with the RMSE value select the Rectify option from the Georeferencing toolbar as shown in Figure A 34 Georeferencing Layer Transit_Map png J Pal n Update Georeferencing 3 Rectify 4 Flip or Rotate gt Transformation v Auto Adjust Update Display Delete Control Points Reset Transformation Figure A 34 Rectify option under georeferencing toolbar e The Rectify option lets you save the unregistered image with the registered image coordinate system Select TIFF as the output file format provide a name with extension tif and save the image see Figure A 35 Make sure you save the image in the same folder where all other files are being stored Along with the image ArcMap generates some additional supporting files that are required for the image to work properly If you fail to store all the files in the same folder i e in the working folder the image will not work properly E E Cell Size NoData as Resample Type Nearest Neighbor for discrete data z Output Location C 6767_work Name Transit_Map 1 tif Format n Compression Quality Compression Type NONE Y 4 100 Figure A 35 Saving the image 87 e Add the image you just saved and remove the old unregistered image At this point we have assigned the coordinate system to the image and are ready for the next step digitizing the new image Digitizing the Registered Image e Op
115. nd Houston Galveston MPOs however TxDOT TPP does have a technical advisory or oversight role with the El Paso MPO Among the 25 Texas MPOs only four urban areas Austin Dallas Fort Worth Houston and San Antonio currently have developed a travel mode choice model see Figure 1 2 The remaining urban areas do not have a mode choice modeling step in their TDM systems At the same time many of the 21 small and medium urban areas that do not have a mode choice step have been experiencing significant demographic population growth in the past decade This growth motivates TxDOT TPP s efforts to develop a mode choice model that would be applicable for small and medium sized MPOs MPOs with a travel mode choice model Texas MPOs 1 Abilene 2 Amarillo 3 Austin 4 Beaumont 5 Brownsville 6 Bryan College Station 7 Corpus Christi 8 Dallas Ft Worth 9 El Paso 10 Harlingen San Benito 11 Houston 12 Killeen Temple 13 Laredo 14 Longview 15 Lubbock 16 McAllen 17 Midland Odessa 18 San Angelo 19 San Antonio 20 Sherman Denison 21 Texarkana 22 Tyler 23 Victoria 24 Waco 25 Wichita Falls Figure 1 2 Texas MPOs with a travel mode choice component in their TDM MPOs in Texas have been classified into four population based categories e Small MPOs population between 50 000 and 200 000 e Medium small
116. ndividuals trip purposes and times of the day For example older individuals are less likely to consider walking as an alternative Similarly transit may not be available at night Procedures to define mode availability are the following e The drive alone mode is always available given the high motorization rates in Texas If car ownership data is available from travel surveys zero vehicle households should not have the drive alone mode available e Car sharing is always available when the household does not have a car individuals from other households can pick up the traveler e Transit availability is more complex to define because it depends on the household location workplace location bus routes and bus stop location Following the procedures implemented by the reviewed MPOs transit is not available when the OVTT including access time egress time and wait time is longer than 30 minutes If this rule is too strict for certain areas particularly those where the bus frequency is low transit can be omitted from the choice set if at least one of the trip ends is within a 14 or 2 mile of a transit stop In theory walk and bicycle modes are always available because individuals can walk bike between any pair of TAZs However this assumption is not reasonable when the distance between TAZs is too long particularly when considering a trip undertaken on a daily basis such as a commute trip MPOs usually consider the walk mode avai
117. odel methodology report Planning Department Transportation Research and Modeling Services National Cooperative Highway Research Program NHCRP National Research Council U S American Association of State Highway and Transportation Officials United States amp Cambridge Systematics 2012 Travel demand forecasting parameters and techniques NHCRP Report 716 Washington D C Transportation Research Board Pinjari A R Pendyala R M Bhat C R Waddell P A 2011 Modeling the choice continuum an integrated model of residential location auto ownership bicycle ownership and commute tour mode choice decisions Transportation 38 6 933 958 San Antonio Bexar County MPO SABCMPO 2011 San Antonio Bexar County 5 county travel demand model documentation Sener I Ferdous N Bhat C R Reeder P 2009 Tour based model development for TxDOT Evaluation and transition steps Center of Transportation Research and Texas Transportation Institute Report 0 6210 2 prepared for the Texas Department of Transportation Texas Department of State health Services Texas Population 2015 Available at http www dshs state tx us chs popdat ST2015 shtm Texas Department of Transportation FTP server Total daily Statewide 2035 Forecast VMT by County Available at http ftp dot state tx us pub txdot info tpp plan_2035 boards vmt_map pdf 66 Texas Department of Transportation FTP server Texas Motor Vehicle Traffic Crash Hig
118. ompanying CD contains the Excel based forecasting tool and the Forecasting Tool User Manual Also included are a MATLAB script and input files for testing purposes This script is discussed at the end of Appendix A e Stop_TAZ_Code m MATLAB script file e LUBBOCK_TAZ_XY csv input file for TAZ number and coordinates e LUBBOCK_Stops csv input file for Stop number and coordinates vi Table of Contents Chapter 1 lntrod uct on gases ccscssvicesscoescosucsvsvssnvanenssenvvevenkeoyenesiesvensousaupochurovenesseunceosusvesseonsvyaussvavensss 1 Tigh ACK BEOUTG EEE ork EE E E uote ts sored Gott Welton hota ah oie Tacs Anak saan 1 1 2 Objective of Research Project cisisiccsenaviostessasaacianvcenadesogeaodssonecsuesegeadeaavccssevandeuetbencaaneravnnciast 2 Chapter 2 Literature Review e sseeesccesoocssoesssccssocesocesoosesoccsoccesocesocescosesoesssecesocesocessosesoessseessseee 5 2 1 Mode Choice MOdeI Ss s sssscccesscccecsseceaas uia n nn an k a a n a a 5 PAPIR ONTE AEE EEE A T E A F A E TES 5 2 2 Mode Choice Models Outside of Texas osnssoneessoseessssssesssereessereesssstesssseresssereesseseesssseesssse 7 2 3 Mode Choice Models in Texas lt ij cccsiscwiscatsvaccartiavavacaisaacateashavadtaleenaneessnseadasdeendeaduataatedaaceens 11 2 3 1 Capital Metro MPO CAMPO esesseesssesssessseseseseesseesserssersseeessseessresseesseeeseeesseresseesse 12 2 3 2 Houston Galveston Area Council HGAC sssssseseseessssssesoseesesssse
119. on and traffic assignment Figure 4 1 shows these steps and corresponding data inputs TxDOT leveldata MPO level data Trip production average household size medium household income Trip attraction population employment by category Travel survey data gt Workplace survey data Trip Generation External local auto and TripCALS truck production Productions and attractions by TAZ Special generators l BS ___ Roads physical Roads operational Trip Distribution characteristics characteristics ATOM2 On board public transit travel surveys optional O D trip table LOS E capacity estimated speed travel times Traffic Assignment TransCAD Figure 4 1 Texas Package inputs The three Texas MPOs that TxDOT TPP does not assist with model development are the Dallas Fort Worth El Paso and Houston Galveston MPOs however TxDOT TPP does have a technical advisory or oversight role with the El Paso MPO 29 STEP 1 Trip generation predicts the numbers of trips originated in and destined to each TAZ using the software TripCALS Trip attractions and productions are calculated at the TAZ level To calculate the trips produced originating and trips attracted destination to each TAZ the MPO and TxDOT have to provide the data to be inputted into the software The MPO provides zonal based estimates of household size household income and median household income to compute trip productions The MPO will also pr
120. on drive alone modes of travel such as public transportation and walking to mitigate congestion and air quality issues The implementation of a mode choice model in the TxDOT TPP s TDM can help in these efforts and also contributes to understanding unmet needs for local populations an issue that is of substantial importance for addressing policy concerns related to equity mobility accessibility and overall quality of life 27 28 Chapter 4 Develop a Forecasting Approach and Model Design This chapter describes the steps to develop a forecasting approach and overall model design recommendations to incorporate a mode choice component into the Texas Package Section 4 1 provides an overview of the Texas Package as it currently stands that is without a mode choice component Section 4 2 presents the model design recommendations to incorporate a mode choice component into the Texas Package the chapter concludes with summation of the most salient findings 4 1 The Texas Package Each Texas MPO is responsible for the transportation planning and programming coordination within their urban area Of those 25 MPOs 22 are led by TxDOT TPP for travel demand modeling As noted in Chapter 1 TxDOT has created a standardized approach for travel demand modeling in its Texas Package The Texas Package in conjunction with TransCAD is a three step daily vehicle trip based model The three steps of the Texas Package are trip generation trip distributi
121. or the auto modes drive alone and car sharing and transit For the auto modes OVTT is associated with the time spent walking from the parking lot to the workplace if the parking lot is located away from the final destination For transit OVTT reflects the time spent walking to from the bus stop and wait time at the bus stop This last time is related to the transit frequency Some MPOs differentiate 36 between access time egress time and wait time but we won t make that distinction OVTT can also include transferring time penalties between non transit modes for example park and ride and or transit modes transfer among buses Monetary cost is a skim component of motorized modes only For auto modes it represents the expenditure on gasoline and other maintenance costs associated with the vehicle such as registration and inspection costs For the transit mode the cost represents the fare Parking costs are only present for the drive alone mode while number of transfers is present for transit only Before presenting the guidelines to develop skims Section 5 3 it is important to discuss mode availability 5 2 Mode Availability One of the most important decisions to develop and implement a mode choice model is the mode availability that is whether a mode is available when making a choice The set that comprises the modes that are available for each individual is called the choice set Note that the choice set can vary among i
122. or the continual efforts to improve process the one that has perhaps received the most attention is travel mode choice Travel mode choice is arguably the single most important determinant of the number of vehicles on roadways and this dimension of travel may be influenced by policy actions that improve the level of bicycling relative to the drive alone modes of travel Such actions may include high occupancy preference lane provision park and ride PNR facilities provision transit oriented development mixed land use development improved pedestrian bikeway facilities toll pricing and travel is particularly important at a time when travel demand on roadways continues to rise in urban areas Doing so also leads to a more efficient use of the roadway infrastructure less traffic congestion lower mobile source emissions less energy dependence and improved mobility and However to simultaneously include all these choices in a single travel demand modeling states are based on either a trip based or an activity based approach In Texas currently a shown in Figure 1 1 trip generation trip distribution mode travel mode obtaining both the number of personal vehicle times and assigns the person trips to the transit network for these methods While there have been efforts to enhance each step of the trip based modeling service of non drive alone modes of travel such as carpooling using the bus walking and improvements in public tran
123. ovide zonal household and employment by category to produce the trip attractions The four basic employment categories used are basic retail service and education All of this information is obtained from conducting annual household surveys of the region Then this data is given to TxDOT who estimates a matrix of households by size and income for each urban area and used as a constraint in TripCALS The MPO is also required to identify any special generators of traffic Special generators are locations that have different travel characteristics than those found by trip generating models This includes hospitals colleges and airports TxDOT will acknowledge these special generators but attempts to limit their use in trip generation models TripCALS5 will then use all of these inputs to compute the productions and attractions by each TAZ for each trip purpose identified STEP 2 Trip distribution uses ATOM2 a spatially disaggregate trip distribution model where an origin destination O D table is created that specifies the number of trips leaving each origin and arriving at each TAZ To create individual trip tables ATOM2 utilizes the productions and attractions generated by TripCAL5 as well as road features provided by the MPO and TxDOT The MPO supplies information about the physical characteristics of the roads within each zone This includes the number of lanes posted speed limits direction one way or two way functional classification an
124. put feature class C 6767_work Copy Export_Output shp Figure A 60 Data export window 103 Saving Data ES Lookin 5 6767_work Copy 0 al ala e J bus_links shp Stops shp Stopsl shp Streets2008 shp a TAZ shp Name Eo Ot Save as type Shapefile Cancel Figure A 61 Output feature class window Now you have the node bus stops and network files bus links for transit Take them to TransCAD and perform the additional processing to obtain transit skims Some additional tasks may be necessary depending upon node and network file compatibility e g making sure that you have only one link between two stops In this manner we can treat the stops as nodes and get the travel time between each stop If TransCAD does not accept the shapefiles created using ArcMap you can use a simple script to determine which zones are transit accessible by calculating the straight line distance between a TAZ centroid and the bus stop and picking the TAZs that fall within a reasonable range say 500 meters or first N nearest TAZs A MATLAB script and input files are provided on the accompanying CD for testing purposes this script calculates the TAZ accessibility by using the stop and TAZ coordinates Basically the MATLAB script calculates the straight line distance between each stop and all the TAZs Then based on the user s criteria e g the first five TAZs based on straigh
125. r San Angelo MPO Cash Daily Weekly Monthly Passenger 8 fare fare fare fare Regular fare 2 00 2 00 10 00 30 00 Seniors 60 students military or disabled Children 5 and under must be accompanied by fare paying adult 1 00 1 00 5 00 15 00 Free Source http www cvcog org cvcog trans_urban html The six routes of TRANSA Urban along with the number of stops are the following e Route 1 12 stops e Route 2 13 stops e Route 3 11 stops e Route 4 13 stops e Route 5 20 stops e Route 6 Goodfellow Express 12 stops only runs on Friday from 18 00 to 01 00 and Saturday 12 00 to 01 00 6 2 3 Longview MPO The Longview Transit system consists of seven fixed routes with fixed stops However travelers can also wave down a bus at any point on the route to board The buses operate Monday through Friday from 6 15 AM to 7 15 PM and on Saturdays from 7 15 AM to 7 15 PM There is no service on Sundays In addition no service is provided on the following holidays New Year s Day Memorial Day Independence Day Labor Day Thanksgiving Day and Christmas Day although for these last two holidays service may end early the day before The transit fares are presented in Table 6 3 Table 6 3 Transit fares for Longview MPO Cash fares Tickets and passes Regular fare 1 25 Five tickets 6 25 Day pass 3 00 Ten tickets 12 50 Children age 6 to 14 years 0 65 Twenty tickets 25 00 Children under
126. related directions present themselves as potential avenues for implementation of this project s results The first is to examine transit skim development methods in more detail In the current project the skims were developed using several assumptions An implementation project can examine the correctness of these assumptions and propose alternative assumptions where appropriate It can also develop clear protocols and recommendations for procedures that MPOs can follow that will make the construction of these transit skims much easier The second is to integrate the mode choice framework developed in this project into the current trip based modeling system used by TxDOT s Transportation Planning and Programming Division The third is to pilot implement the proposed modeling framework and approach in a few MPOs in the state with the improved transit skim methods also developed as part of an implementation and better mode choice data that also uses on board transit survey data The fourth is to provide workshops on the actual implementation of the mode choice model integrated as part of the broader trip based travel model system 64 References Arizona Daily Star Find what your car costs to drive per mile Available at http azstarnet com business find what your car costs to drive per mile article_662b7ce9 d499 5f5c a95 1 5e65d3ff83dc html Association of Monterey Bay Area Governments AMBAG 2011 AMBAG regional travel demand model
127. responding mode is available In order to obtain the availability of walk and bike modes a maximum limit of 30 minutes on walking and 40 minutes on biking was imposed and then trips were appropriately assigned the walk and bike mode availability An important point to note here is that only 54 observations trips have accessibility to transit The transit availability was determined using the condition that both origin and destination TAZs were accessible to transit 57 and the access and egress time to transit was less than 15 minutes The results indicate a limited transit service in terms of area coverage To be precise Lubbock only has nine active transit routes with five stops per route Table 8 4 provides the mode share obtained from the survey data for a sample size of 1975 for the HB work trip Table 8 4 Mode share for Lubbock area Mode Frequency Percentage Walk 1 0 05 Drive Alone 1871 94 73 Shared Ride 99 5 01 Transit 3 0 15 Bike 1 0 05 Table 8 4 makes clear that almost all of the trips were made via drive alone or shared ride modes The share for transit walk and bike is negligible A mode share distribution of this kind makes it infeasible to estimate a mode choice model with all five modes due to insufficient number of observations for transit walk and bike mode Hence we borrow the level of service variable coefficient values from the Bhat and Sardesai 2006 study and adjust the alternati
128. rip based approach for the TDM with TAZs as the unit of analysis and their TDM has a feedback process between the traffic assignment and trip distribution steps A summary of the MPOs TDMs is presented in Table 2 2 detailed descriptions are available in Appendix D The five MPOs listed in Table 2 2 disaggregate trips by purpose because as mentioned before travelers may have different mode preferences in different choice occasions The small MPO CCRPC considers five trip purposes while the medium and large MPOs consider seven to nine trip purposes Trip production models are similar among MPOs All MPOs develop a cross classification model to estimate trip productions although each MPO uses different explanatory variables see Appendix D for details Methodological differences arise for trip attractions Trips attractions are computed based on other estimates NCHRP Report 365 previous models or linear regressions Gravity models are used by all MPOs in the trip distribution step except for Metro MPO which does not have an independent trip distribution model Table 2 2 TDMs of MPOs outside of Texas Base Trip Trip generation models Trip f Mee year purposes Trip Trip attraction Sua production model CCRPC 2002 HB work HB school HB Cross NCHRP Report Gravity tod l small MPO 2003 shopping HB other NHB classification 365 y HB work HB shop HB Lincoln MPO j i p p il 2009 recreation HB university Cro
129. rips The non motorized shares were estimated using a distance based algorithm model with data from the 2000 Census Transportation Planning Package CTPP The transit shares were obtained from transit ridership data census journey to work data and a sensitivity analysis of data from other areas Finally an auto occupancy model was used to separate the remaining trips into auto driver or passenger driver trips based on the data from the CTPP This last step was taken to convert person trips from the trip generation and distribution models into vehicle trips for assignment to the roadway network GCMPC GCMPC 2009 A three level NL model is used for the mode choice step in the GCMPC area see Figure 2 1 The model divides the person trips into the five modes shown in the figure only three trip purposes were estimated HB work HB other and NHB Travel counts household travel survey data and the 2007 transit on board survey data were used to obtain the data inputs The 2000 CTPP data was used as a reference for HB work trip as well The entire TDM was validated using traffic counts The nested structure was revisited and corrected to reach the acceptable error standards defined by the Michigan DOT After the validation process transit ridership estimates differed from the ridership counts by 25 Non Motorized Source GCMPC 2009 Figure 2 1 NL model structure of GCMPC s mode choice model e AMBAG AMBAG 2011 An NL
130. s 0 0 482 0 482 0 212 0 Travel Cost Dollars o 0 o 0 0959 0 gt REFRESH ALL 1 x CLICK TO CHECK DATA 2 L CLICK FOR MODE SHARE 3 J Annn SCENARIO RUN 4 _INPUT__INDMDUALRECORDS TAZ HHJNCOME DATA MIT DA AREATYPE DISTANCEDA IVTT TRANSIT OVTT TRANSIT DISTANCE TRANSIT DISTANCE_BIKE DISTANCE WALK Sign in Figure B 1 Forecasting Tool Input Sheet 108 Input Requirements Table B 1 provides the name and the type of data required for entry into the input sheets Table B 1 Sheet Name and Data Requirement Sheet Name Description Functionality User needs to provide the mode choice model and various other INPUT inputs as mentioned in the sheet refer to Table B 3 for a detailed discussion INDIVIDUAL_RECORDS If user has the individual level trip distribution records provide them in this sheet in the format specified at the top of the sheet TAZ_HH_INCOME_DATA IVTT_DA AREA_TYPE DISTANCE_DA IVTT_TRANSIT OVTT_TRANSIT This sheet accepts the TAZ level household split in percentage based on household size and household annual income refer to the sheet in the tool for a sample input This sheet accepts the in vehicle travel time matrix for Drive Alone mode i e TAZ TAZ in vehicle travel time This sheet accepts the area classification indicator variable for each TAZ refer to Table B 2 for area classification code This is used to calcu
131. s for drive alone mode sssssneseesossossesoseeeessssssosereesessssosereesessssssesre 38 Table 6 1 Transit fares for Bryan College Station MPO esssssssssssssesssesssesssssessseesseesseesseeesseee 42 Table 6 2 Transit fares for San Angelo MPO sssssssssssssesssesessseessresserssesssseessseesssresseesseesseeesseee 43 Table 6 3 Transit fares for Longview MPO sessssssssssessessesssseessseessresseesseessseeesseesssresseesseesseee 43 Table 6 4 Transit fares for Lubbock MPO sesesessesesessssessesssesreesessreseresreseresresseseresresseseresreeseese 44 Table 7 1 List of demographic and trip variables esseeeeseeeseeeesseseresressrserssressesersrressessrerrensesse 49 Table 7 2 Mode share for the HB work trip based on survey data ssonsseeesseeesseesseesserssesesseee 50 Table 8 1 Distribution of sample based on household siZe eseseseseeeeesrseresressessrerrerseseresrerseese 56 Table 8 2 Distribution of sample based on household income eeseeeseeeresseeressrrsrerressessresreesesse 57 Table 8 3 Descriptive statistics for level of service variable sseseseeesesereeserererresserererrerseese 57 Table 8 4 Mode share for Lubbock area 5 cen2tostes cot ccancsticavenciisl acccnseaeislainietentiteceusien 58 Table 8 5 Mode choice model coefficients ssesesessesseseesseseesresseserssresstseresressesersrressessresresseese 58 Table 8 6 Implied money value of travel tiMe eee eeseecee
132. s which could help to understand and predict mode choices more accurately 25 Table 3 4 Recommendation on incorporating a mode choice model in the TDM Non Policy Recommendation on MPO Population insignificant evaluation incorporating a growth share of non needs mode mode choice model auto modes choice analysis in the TDM Small MPOs Texarkana MPO Small Yes Recommended San Angelo MPO Small Yes Recommended Victoria MPO Small Yes Yes Recommended Sherman Denison MPO Small Not recommended Wichita Falls MPO Small Yes Recommended Abilene MPO Small Yes Recommended Harlingen San Benito MPO Medium Yes Recommended Medium small sized MPOs Tyler Area MPO Medium Yes Recommended Longview MPO Small Yes Recommended Bryan College Station MPO Large Yes Yes Highly recommended Waco MPO Small Yes Recommended Brownsville MPO Large Yes Yes Highly recommended Amarillo MPO Medium Not recommended Laredo MPO Large Yes Yes Highly recommended Midland Odessa MPO Medium Not recommended Lubbock MPO Medium Yes Recommended South East Texas RPC Small Yes Recommended Killeen Temple MPO Large Yes Yes Highly recommended Corpus Christi MPO Small Yes Yes Recommended Medium large sized MPOs Hidalgo County MPO Large Yes Recommended El Paso MPO Medium Yes Recommended The development of a mode choice model is a process that requires time Most MPOs in and outside Texas have been developing mode choice model and reevaluating them for several years Some MPOs such as CAMP
133. s the unique number of the person who participated in the survey trip number mode of trip and purpose of the trip e The determination of trip origin and destination is slightly complicated In the survey data the trip number is recorded as follows The first trip for each person is recorded as zero for where their day began Each subsequent trip is numbered sequentially 1 2 3 etc Hence the sequence of trips forms a chain starting with trip zero and the previous trip location serves as an origin for the next trip To avoid any mistake in recording origin destination zone pairs we checked the arrival and departure time for each of the trips along with the trip number to ensure that proper ordering was maintained e Once the trip characteristics were appended appropriately for each of the household members we appended the demographic variables household size and household income for each of the individuals by matching the unique household number Tables 8 1 and 8 2 provide the sample distribution based on household size and income Table 8 1 Distribution of sample based on household size Household Size persons 1 2o0r3 4or more Frequency 95 998 882 Percentage 4 81 50 53 44 66 56 Table 8 2 Distribution of sample based on household income Income Frequency Percentage Less than 25K 236 11 95 Between 25K amp 50K 585 29 62 Greater than 50K 1154 58 43 Table 8 1
134. s to indicate that the majority of the population is in the high income category greater than 50 000 In Table 8 10 we provide the descriptive statistics for level of service variables for all five modes 60 Table 8 10 Descriptive statistics for level of service variable Variables Minimum Maximum Average Standatrd Frequency Deviation IVTT for DA amp SR Min 0 4 58 2 13 1 8 5 1189 OVTT for DA amp SR Min 2 0 2 8 2 1 0 2 1189 Travel Distance for DA amp SR Miles 0 2 35 7 8 2 5 7 1189 Travel Cost for DA Dollars 0 1 8 9 2 0 1 4 1189 Travel Cost for SR Dollars 0 1 4 5 1 0 0 7 1189 IVTT for Transit Min 5 0 25 0 9 7 4 9 15 OVTT for Transit Min 7 8 18 3 14 7 3 3 15 Travel Cost for Transit Dollars 1 3 1 3 1 3 0 15 Trip time for Walk Min 3 8 29 2 21 4 5 8 66 Trip Distance for Walk Miles 0 2 1 5 1 1 0 3 66 Trip time for Bike Min 1 40 22 2 9 8 671 Trip Distance for Bike Miles 0 2 7 3 4 1 1 8 671 DA drive alone SR shared ride The walk and bike mode availability was determined using the same conditions as in the Longview area An important point to note here is that once again only 15 observations trips have accessibility to transit indicating a limited transit service in terms of area coverage To be precise Longview has only five active transit routes with five stops per route Table 8 11 provides the mode share obtained from the surv
135. sed on maximum walk time Put 1 if you want limit the availability of Bike mode based on maximum Bike time Individual level estimates TAZ level estimates Number of Alternatives Number of Explanatory Variables Including Constants Create TAZ Configuration TAZ Level Mode Share Estimate Household Cate gory Income Category Scenario Alternative Related Variables Reduce Transit IVTT by Percentage Increase Drive Alone IVTT by Percentage Increase Drive Alone OVTT by Percentage DEVELOPER M perties 3p Import rt 72 Expansion Packs F Expor Document Panel XML Modify SEG 2 Copy of SubodhMacrol Copy Excel E 20 Area Type Wait Time 0 25 1 40 2 10 3 1 75 4 2 L o oowu 1 50 1 25 1 00 1 00 I J K E M N Parameters Drive Shared Transit Bike Alone Ride Alternative Specific Constant 0 0 52 1 2 0 055 1 43 Household Size Single member Household is the base category Household Size two or three 0 0 374 Household Size four or more 0 0 0 836 Household Annual Income Less than 20K is the base category Income between 20K and 50K 0 486 0 486 0 0 Income greater than 50K 0 0 847 0 847 0 0 74 In Vehicle Travel Time Min 0 0 000 0 000 0 389 0 Out of Vehicle Travel Time Min 1 37 0 0 0 569 1 37 Travel Distance Mile
136. seeceseeceeaeeceeaceceeaeeceseeceeeeeeeneeeees 59 Table 8 7 Implied mode share for Lubbock Area based on estimated model eee 59 Table 8 8 Distribution of sample based on household SiZe cceeseecesseceeeeceeeeeceeeeeceeeeeeeteeeees 60 Table 8 9 Distribution of sample based on household income eee eeeeeeceeeeeereeeeeeeeseeeeeeeenees 60 Table 8 10 Descriptive statistics for level of service variable 0 ec eeeeseeseeeseessseeeseeneeeeeaees 61 Table 8 11 Mode share for Longview area fac 2G soe e25ceoks eceseesuc taste asics ses saedieatsesonaes saenayeuvtecqowveaaes 61 Table 8 12 Mode choice model coefficients ai 1c ene ee Nee eine 62 Table 8 13 Implied mode share for Longview area based on estimated model e cesses 62 Table A 1 Out of vehicle travel time based on area type eeecceesscecesneeceeeeeceeeeeceeeeeceeeeeenteeeees 80 Table B 1 Sheet Name and Data Requirement csseveccsssicsqraszevabecciategeudaeeouvedasy teavaasantunagaosededevayeass 109 Table B 2 Out of Vehicle Travel Time Based on Area Type eccceescecesnceceeececeeeeeceeeeeenaeees 110 Table B 3 INPUT Sheet Detail ot isn siete toate nna te Gelinas nit cotta as ater at een cine tata 110 Table B 4 Individual Level Mode Summary ccceeesccesssecesnceceseeeceeneeceeaceceeceeceeeeecsteeeeneees 114 xi xii Chapter 1 Introduction 1 1 Background Urban travel demand results from a complex mul
137. sification Type Wait Time table see Figure B 1 or refer to this sheet in the tool 110 Input Name Description Number of Passengers in Car for Shared Ride Provide the number of passengers for Shared Ride mode Maximum Walk Time min Maximum Bike Time min Provide maximum walk time acceptable to the user It is used to determine the walk availability between TAZ pairs Provide maximum bike time acceptable to the user This figure is used to determine the bike availability between TAZ pairs Put if you want to limit the availability of Walk mode based on maximum walk time Make this 1 if you want to limit the availability of Walk mode based on maximum walk time during mode share calculation otherwise enter 0 Put 1 if you want to limit the availability of Bike mode based on maximum bike time Individual level estimates TAZ level estimates Make this 1 if you want to limit the availability of Bike mode based on maximum bike time during mode share calculation otherwise enter 0 Make this 1 if you want to run the individual level estimates This is applicable only when the user provides individual level trip distribution data in the sheet named INDIVIDUAL_RECORDS otherwise enter 0 Make this 1 if you want to run the TAZ level estimates This is applicable under any circumstance because TAZ level household information is available readily from the Texas Package However on
138. sign recommendations to incorporate a mode choice component into a TAZ based TDM The task looked at the inputs and outputs of the Texas Package and documented the implications for the way the mode choice model is specified and applied The proposed recommendations are based on data currently available for small and medium sized MPOs 33 34 Chapter 5 Procedure to Develop Skims This chapter will outline guidelines on how to obtain the travel system attributes required to estimate mode choice models based on the data already available in the Texas Package The CTR research team contacted four urban regions that are best positioned to benefit from including a mode choice model component based on conversations with the project monitoring committee PMC and the findings of Task 1 to obtain and assemble all the data needed To complete its three steps trip generation trip distribution and traffic assignment the Texas Package receives demographic data and trip characteristics data from the MPOs However to model mode choice an additional data component is required skims Skims are a set of matrices that show travel times and costs for each mode and for each O D zone pair representing the level of service of each mode Currently TxDOT only develops skims for the auto mode known as network skims However the issue of skim development is not specific to the auto mode but also relevant to other modes This chapter provides step by step gu
139. sorereessssssesereesesss 13 2 3 3 San Antonio Bexar County MPO SABCMPO ecccsssccecsseceesseeeesteeeeseeeeneeeeaaees 15 2 3 4 North Central Texas Council of Governments NCTCOG ccccccccccccceesessnseceeeeees 16 Chapter 3 Incorporating a Mode Choice Component for Small and Medium Sized MPOs in TOMAS conessisoseberassosecsososcssoescrouscsiuo bossos sprees sopesar sotos tos sosna se csabsuasvaaebpeveaueleossbesnssiansbes 19 3 1 1 Population GroWwth icne innsin ae an E AE nish E EEV 19 312 M d CHOICE Share Sgrena e a a N N 20 Sl 3 Stratesic Planning Goal Seni n a a E A aia 22 3 2 Recommendations s einai aE TEE E AEE AA E E A S TS 25 Chapter 4 Develop a Forecasting Approach and Model Design esssesssecssecsssccsscocesocessossose 29 4 1 The Texas Package s asswecetsiavaj sccvvapaeaaqacaydacaisstaceuaedeaaraauadadsssuedeaes E N A G a 29 4 2 Mode Choice Model Recommendations ssseesssessessseeseeessstessetsseesseesseeessseesseesseesseesseee 30 AS NIGEL SPECIC ALON masensi a cue deGpaasuse i ameded E A E vara 31 44 Forec sting Approach n a E a A E AE A A A A En stents 33 AD Next Steps repenre e a e ER e E Ee SA 33 Chapter 5 Procedure to Develop Skims e sesesssesssesssecssocesocesoocesoeessecesocesocssoocesocsssecesocssocsssossose 35 Dl SKIN C mponentSso eane e A o e R a EEAS eE E ASi 35 5 2 Mod Availability rsen n ik ee oe eel oat a SE ties 37 DF RATE ISVS OPENS NL 1055 cs paz E naasse pase Gewcus TA A
140. sportation service Increasing the share of non drive alone modes of quality of life 1 2 Objective of Research Project In the above context of focusing on solutions to manage growing travel demand in urban areas in Texas the Texas Department of Transportation TxDOT Transportation Planning and Programming TPP Division is initiating another enhancement of their travel demand modeling system so that they can analyze alternative transportation modes carpooling public transportation bicycling walk modes and evaluate and prioritize multimodal projects at the regional level TxDOT created a standardized approach for travel demand modeling called the Texas Package Suite Sener et al 2009 of Travel Demand Models referred to as the Texas Package The Texas Package in conjunction with TransCAD is a three step daily vehicle trip based model The three steps included in the Texas Package are trip generation trip distribution and traffic assignment The Texas Package has been used since the late 1990s statewide but TxDOT is looking into the possible inclusion of mode choice models for MPOs with the need for one Currently TxDOT TPP is responsible for TDM development to support the regional long range plan update and associated long range planning activities within 22 of the 25 Texas urban areas The three Texas metropolitan planning organizations MPOs that TxDOT TPP does not assist with model development are the Dallas Fort Worth El Paso a
141. ss Earlier model Gravity modi me e en HB other work based other classification results y sized MPO NHB HB work low income HB GCMPC work high income HB Cx Linear medium small 2005 shopping HB other HB cl ssific tion regression Gravity model sized MPO school HB university NHB model other NHB work From survey HB work HB maintenance data if AMBAG 4 dium 2005 HB discretionary work Cross available Gravity model one a nad ha based HB school visitors classification otherwise y sized MPO others NCHRP Report 365 HB work HB shopping HB No longer Destination Metro MPO 2008 recreation HB other NHB Cross eae aa dae NE Mange nO ja PA He elseincaugn work and HB trip distribution ge college model A summary of the mode choice models of MPOs outside of Texas is presented in Table 2 3 Several differences in both methodology and data usage are clear from the information presented in the table First all MPOs disaggregate trips by purpose however these purposes are not the same as those used in the previous TDM steps generation and distribution steps The MPOs with a large number of trip purposes in the earlier steps aggregate the trips in only three purposes in the mode choice step HB work HB other and NHB Additionally two MPOs disaggregate trips by either transit availability scores Lincoln MPO classifies zones based on transit coverage and operations or time periods Second smaller MPOs tend to use fewer data inpu
142. t has a bus stop and the adjacent TAZ on the other side of the road are only accessible by bus Next bus stops were appropriately mapped on the TAZ configuration and the transit travel time between accessible zones was calculated The procedure for construction of transit routes mapping of stops on a TAZ and construction of a distance band for identification of a transit accessible TAZ is documented in the TransCAD user guide chapters 16 and 22 Interested readers are referred to TransCAD user guide for detailed information on transit route construction and calculation of travel time Out of vehicle travel time To generate the transit OVTT skim the distance from the centroid of the TAZ only for the TAZs found to be accessible via transit during the IVTT calculation step for transit to the bus stop location was calculated and a walking speed of 3 miles per hour mph was used to calculate the access time at the origin A similar procedure is used to calculate the egress time at the destination as well TAZ TAZ distance The distance between each transit accessible TAZ pair was calculated using the route link length Travel cost To estimate the cost of travel via transit we used the fares available from the respective MPO transit agency websites Longview and Lubbock Transit Agencies not dated The regular cost of travel in the Longview area is 1 25 and 0 60 for senior citizens persons 65 years of age or older The fare structur
143. t line distance the code provides two sets of outputs The first output provides the TAZ numbers based on user criteria corresponding to each stop and the second output provides the corresponding distances between each stop and TAZ The script requires two sets of input 1 a csv file containing TAZ numbers and corresponding centroid X and Y coordinates the code accepts both latitude and longitude as X and Y coordinates and X and Y coordinates in any other units i e feet or meters The latitude and longitude of a TAZ can be obtained from a Google map or sometimes are available in the TAZ shapefiles provided by MPOs Similarly the X and Y coordinates can be obtained in other units of measurement such as feet or meters by processing the TAZ shapefile in ArcGIS or TransCAD Once a shapefile is opened in ArcGIS or TransCAD the user can simply point the cursor at the top of TAZ centroid and record the X and Y coordinates by noting the values shown at the lower toolbar The unit of X and Y coordinates provided by ArcGIS or TransCAD are generally in feet but users can easily change the default setting 104 2 similar to the first file the second file contains the stop number and corresponding X and Y coordinates The user can provide the X and Y coordinates in both latitude and longitude or in any other units as discussed above However the X and Y coordinates in both the files should be in the same units i e either latitude and longi
144. t of zone centroids and displays the set in the drop down view in the toolbar Figure A 13 Haas All Records W Selection E E Em PREFE 38 B S FF ALA 2 5 S O S 2 Figure A 13 Viewing the centroid set e If you are not able to view the centroid set Select the Node layer from the drop down list on the toolbar and click on to open the Node layer table All the Centroid Nodes now should be shown with a dot of a certain color If not scroll down further in the drop down list on toolbar and you should see the centroid set it will be given the name you provided during centroid creation step Creating the Binary Network File e Select the Network layer from the drop down list on the toolbar e Select Network Paths gt Create from the toolbar If the Network Paths option is not visible on the toolbar select Procedures gt Network Paths to add the Network Paths option on the toolbar e Select the parameters as needed under the Other Link Fields It is good to include at least the length speed and travel time from the line layer to provide the option of updating the travel time after any editing has taken place The user should also include any field which will be used later to disable some links in order to obtain the travel time matrix for a given scenario The example in Figure A 14 shows all attributes being included e Once selection is over click OK and provide the appropriate name
145. t the tool right click on it This display the snapping toolbar see Figure A 52 as a floating toolbar so drag it to reposition it Snapping oja o alz Figure A 52 Snapping toolbar 99 e To set the snapping tolerance click on the snapping toolbar and select Options see Figure A 53 This action opens a window as shown in Figure A 54 Enter the tolerance value in pixel Unfortunately we can set the tolerance only in pixels Generally the images have about a 60 cm resolution so for 4 meters we can specify 7 pixels To be on the safe side use a value between and 4 Snapping 33 jo a Use Snapping Intersection Snapping A Midpoint Snapping O Tangent Snapping Options Figure A 53 Snapping toolbar options Snapping Options General Tolerance Symbol color E Snap to basemap layers V Snap to feature service layers Snap tips V Show tips Layer name Snap type Background Figure A 54 New snapping tolerance setting window 100 Classic snapping tool e To use the classic snapping tool click on the Editor tool then select Options from the bottom of the dropdown menu then check the option Use Classic Snapping see Figure A 55 f Editing Options Display measurements using 3 decimal places Sticky move tolerance o pixels Stretch geometry proportionately when moving a vertex v Use symbolized feat
146. tem modes environment options connectivity Small MPOs Texarkana Urban Texarkana f MPO Transportation Study x x x x 2035 Plan San Angelo MTP Fiscal Years 2 z 2 MPO 2010 2035 ee Victoria Urbanized Victoria MPO Area MTP 2035 X X X X X X X Sherman Transportation Denison MPO Outlook 2035 Wichita Falls MPO 2010 2035 MTP x x Abilene Metropolitan Area Abilene MPO MTP 2010 2035 x X X X x Harlingen San Benito MPO 2010 2035 MTP X X X X x X Medium small sized MPOs Tyler Area MPO MTP 2035 X X X x Longview MPO MTP 2035 X X X x Bryan College Bryan College Station Station MPO MPO 2010 2035 MTP i x j i Connections 2035 Waco MPO The Waco MTP X X X x x Brownsville MPO 2010 2035 Brownsville MTP X X x x X 23 Strategic planning goals related to mode choices continuation Improve _Incentivize Reduce Provide Enhance Improve FPE Reduce expand public non emissions multimodal integration MPO Source f A public energy 3 transportation motorized protect the transportation and health consumption cos system modes environment options connectivity Medium small sized MPOs Amarillo MTP Amarillo MPO 2010 2035 X X X Laredo MPO 2010 2035 MTP X X X X X Midland 2035 Transportation x a E Odessa MPO Plan Update Lubbock MPO Lubbock MPO 2012 2040 MTP X X x X South East Texas RPC MTP 2030 X X X x Killeen Killeen Temple Temple MPO Urban Transportation xX xX xX X X xX p Plan Mobility 2035 Corpus Christi Corpus Christi
147. the CBD area can be set to 5 while the parking cost in residential areas can be null 38 Parking cost is usually divided by two so that the cost is allocated to each leg of a round trip home workplace and workplace home 5 3 2 Transit To create the transit skims the first step is to create the transit network A transit network is a spatial representation of bus routes available in a region In a transit network each transit route is coded with links representing the path it follows and nodes representing the stops along the path These links and nodes are shared with those of the highway system The transit network is used to generate the transit skims including IVTT OVTT and cost To create the transit network MPOs need to collect information on the bus service This information can usually be obtained from the service operator and includes the bus routes the location of the bus stops fares headways schedules and speeds To be incorporated into the Texas Package the route system is coded as a TransCAD route system using the TransCAD route system editing toolkit Because transit network creation and maintenance can involve several routes and stops detailed cross checking of the transit network coding against available transit maps and schedules should be performed The transit networks periodically need checks and updates as transit systems evolve and routes change Note that the route service information can vary by time of
148. the survey data We extracted the trip information for each of the households This information includes unique household number person number indicates the unique number of the person who participated in the survey trip number mode of trip and purpose of the trip The determination of trip origin and destination is slightly complicated In the survey data the trip number is recorded as follows The first trip for each person is recorded as zero for where their day began Each subsequent trip is numbered sequentially 1 2 3 etc Hence the sequence of trips forms a chain starting with trip zero and the previous trip location serves as an origin for the next trip To avoid any mistake in recording O D zone pairs we checked the arrival and departure time for each of the trips along with the trip number to ensure that proper ordering was maintained Once the trip characteristics were appended appropriately for each of the household members we appended the demographic variables household size and household income for each of the individuals by matching the unique household number We also provide a list of alternatives available for mode of the trip purpose of the trip and household income in the survey data e Mode of the trip It contains the following alternatives Walk Auto Carpool Vanpool Commercial Vehicle Bus School Bus Taxi Bicycle Motorcycle and others e Purpose of the trip It includes the following alternatives M
149. the user to delete the old sheet Click on the Delete option as shown in Figure B 4 if the program asks 114 you to do so and then re click the same button Accepting the Delete option removes any unnecessary files A You can t undo deleting sheets and you might be removing some data If you don t need it click Delete Figure B 4 Delete the Old Sheets Also you will need to enable the Macro option available in Excel in order to run this tool By default when you open the tool it will ask you to enable the Macro Please accept the option If the macro enable option does not pop up in the beginning follow this procedure gt Click the Microsoft Office Button and then click Excel Options gt Click Trust Center click Trust Center Settings and then click Macro Settings gt Enable all macros not recommended potentially dangerous code can run Click this option to allow all macros to run This setting makes your computer vulnerable to potentially malicious code and is not recommended TIP You can open the macro security settings dialog box from the Developer tab in the Ribbon which is part of the Microsoft Office Fluent user interface If the Developer tab is not available Oa click the Microsoft Office Button amp and then click Excel Options Click Popular and then select the Show Developer tab in the Ribbon check box 115 116 Appendix C The Multinomial Logit MNL Model Consider a de
150. tidimensional choice process which includes residential location vehicle ownership time of day destination mode and route framework is difficult and the choice process is usually compartmentalized into simpler sub processes in a logical and tractable way see Koppelman and Bhat 2006 Pinjari et al 2011 Within this context the models used today in most of the metropolitan areas of Texas and other trip based approach is used The typical trip based approach uses a trip as the unit of analysis and usually includes four sequential steps as Trip Generation choice and traffic assignment The trip generation step p involves the estimation of the number of home based HB on and non home based NHB person trips typically Trip Distribution classified by trip purpose produced from and attracted to l each traffic analysis zone TAZ in the study area The trip distribution Step determines the trip interchanges G e Mode Choice number of trips between each zone The third mode choice step splits the person trips between each pair of zones by i trips and number of transit trips between zones The traffic assignment step assigns the vehicle trips to the roadway network to obtain link level vehicle volumes and travel Figure 1 1 Four step trip different time periods in the day based approach The results of trip based travel demand models TDMs are used to make important investment and policy decisions which is the main reason f
151. timate walk time with a maximum of 10 minutes Access to transit by walking is based on market segmentation at the TAZ level Households that live within quarter mile short walk and half mile long walk buffers around available transit stops have access to transit while households located further than a half mile are assumed to have no transit stop available within the allowed walking distance no walk Households are then segmented in seven distance based categories at both the origin and destination of the trip short walk short walk short walk long walk long walk short walk long walk long walk no walk short walk no walk long walk no walk no walk CAMPO is in the process of updating its TDM Some improvements considered for the mode choice model are to 1 use three HB work trip purposes employing on board surveys designed to collect the required data 2 use household income variables instead of auto ownership to address environmental justice issues and 3 incorporate new transit modes such as bus rapid transit commuter rail and light rail The Travel Model Improvement Program FHWA 2010 recommended that CAMPO validate the mode choice model to restructure it in 12 accordance with the FTA requirements and to collect data on the commuter rail service for future usage 2 3 2 Houston Galveston Area Council HGAC The study area for HGAC encompasses eight counties Montgomery Liberty Chambers Galveston Brazoria
152. trix generation for each TAZ pair The travel cost can now simply be calculated by assuming a Per Mile Gas cost and multiplying this with the corresponding travel distance The out of vehicle travel time is generated based on TAZ area type The typical value assumed by CAMPO is provided in Table A 1 79 Table A 1 Out of vehicle travel time based on area type Area Type Typical CAMPO Value Central business district CBD 1 5 minutes CBD Fringe 1 25 minutes Urban and Suburban 1 00 minutes Development of Skims for Bike and Walk To generate the skim distance matrix for bike and walk modes we assume that people tend to avoid freeway and highway segments when commuting in bike and walk modes Also biking and walking are generally not allowed on freeways and highways So to incorporate these effects in the network the user needs to disable the freeway and highway links for these two modes e To disable the links select Network Paths gt Settings on the toolbar TransCAD opens a window as shown in Figure A 20 novo Sete File C dlongviewmpodata Combine netfinet Based on C G40_NET LNG40_NET LNG40_NET dbd Description C Users caee4p226874a Desktop fwdlongview General Centroids Do not use centroids e node C Create from selection set Link Type Type None S Figure A 20 Network setting window e Click on Update and select Disable Links and By Expression as shown in Figure A 21
153. ts and choice alternatives than do larger MPOs Finally Table 2 3 notes the use of a variety of mode choice model structures including the MNL and NL models see Appendices A and B and a simple fixed percentage mode split model Table 2 3 Mode choice models of MPOs outside of Texas Disaggregation MIE level Data inputs Choice alternatives CCRPC 5 trip purposes 4 Transit network characteristics transit impedance mode attributes 1 Drive alone 2 Shared ride 3 Transit 4 Bike 5 Walk small MPO area types Lincoln MPO 1 EPU 3 transit availability ime demnierial zone scores for sized MPO transit only Trip distance boarding data auto occupancy 1 Non motorized 2 Transit 3 Auto GCMPC 3 trip purposes HB medium small work HB other sized MPO NHB IVTT OVTT transit fare trip distance socio economic characteristics 1 Drive alone 2 Share ride 3 Transit 4 Bike 5 Pedestrian AMBAG 3 trip purposes HB medium large work HB other In vehicle time walk time wait time fare value of time trip 1 Drive alone 2 Share ride 2 person 3 Shared ride 3 person 4 Premium transit service 5 Local transit service sized MPO NHB distance number of 6 Park and ride PNR transfers transit fare i 7 Kiss and ride KNR 8 Non motorized In vehicle time walk i Davedione perm apnea baba 2 Drive with passenger number of transfers trip a Adio
154. tude or feet or meters etc The MATLAB script is provided below Proper comments text in green color are provided at the top of each line in the code to help user run the code efficiently clear all cle Earth s radius in km R 6371 Make this 1 if you are using latitude and longitude as X and Y coordinate otherwise 0 Lat_Lng 1 Input file for TAZ number and coordinates This file should have 3 columns with the number of rows equal to total number of TAZs in an area The first column contains the TAZ numbers 1 to N the second column contains the X coordinates or latitude and the third column contains Y coordinates or longitude User can change the name of the file as per his her requirement However make sure that the format is same An example file is provided along with this code TAZ XY csvread LUBBOCK TAZ XY csv j Input file for Stop number and coordinates This file should have 3 columns with a number of rows equal to the total number of stops depending on the number of routes considered in the study The first column contains the stop numbers 1 to N the second column contains the X coordinates or latitude and the third column contains Y coordinates or longitude User can change the name of the file as per his her requirement However make sure that the format is the same An example file is provided along with this code Stop _ XY csvread LUBBOCK Stops csv j row_TAZ col_TAZ size TAZ_XY
155. ure during editing Phemseesene Show warnings and information on start editing lt S Stream Mode Stream tolerance 0 mapunrits Group 50 points together when streaming Edit Sketch Symbology Unselected Selected Vertex a a a a Current Vertex Segment Figure A 55 Classic snapping option window e To set the snapping tolerance click on Editor Snapping and then Options see Figure A 56 Set the snapping tolerance to around 4 map units 101 Classic Snapping Options Figure A 56 Classic snapping tolerance setting window o Note use either one of the snapping methods do not use both We recommend using the classic snapping option e To digitize the roads click on Start Editing under the Editor Toolbox Option and this time select Final_bus_links see Figure A 57 and click OK e This action opens up a pane on the right side Select the Final_bus_links to highlight the Construction Tools at the bottom Select the Line option as shown in Figure A 57 eat Features ax T z lt Search gt Q O o Final_Bus_Links EF Construction Tools E Final_Bus_Links 7 a Rectangle D Q Circle S Ellipse Ce Freehand Figure A 57 Create feature window line option e Select the Line option from the Editor Toolbox see the square blue box in Figure A 58 Editory gt Z 41 i N PA g Eg Figure A
156. utput2 i 2 Num_TAZ_ Req 1 clear Temp_Dist end end Temp_Dist 1 Num_TAZ Req 1 Temp_Dist 1 Num_TAZ Req 2 First output file providing the list of TAZs TAZ numbers for each of the stops The first column contains the stop number in each row and the remaining columns contain the TAZ numbers That is if the first row of the file has a value of 1 115 118 117 then the 1 indicates the stop number and 115 118 and 117 indicate the TAZ numbers in the ascending order of distance dimwrite Stop TAZ Map csv Outputl1 delimiter precision 15 Second output file providing the distance between the stop and the TAZs Please note that the distances reported in this file are in the same order as the TAZs reported in first file That is if the first row of the file has a value 1 70 45 200 32 550 43 then the 1 indicates the stop number and 70 45 200 32 and 550 43 indicate the distance between stop 1 and TAZs 115 118 and 117 respectively All the other rows can be interpreted in the same way as discussed Remember that if the X amp Y coordinates are latitude and longitude then the unit of distance is kilometer otherwise it is the same as the unit of the X amp Y coordinates dimwrite Stop TAZ Dist csv Output2 delimiter precision 15 disp Done Appendix B Forecasting Tool User Manual Introduction The Excel based forecasting tool allows users to provide a mode choice model along with various skims in ve
157. ve specific coefficients of the multinomial logit MNL model to represent the market share as obtained from survey data Appendix C provides insight into MNL models The data for the Bhat and Sardesai study was drawn from the web based survey of Austin area commuters The idea behind borrowing the level of service variable coefficients and adjusting the alternative specific constants is that Austin and Lubbock residents share the same underlying sensitivity to travel time and cost which is a realistic assumption Table 8 5 provides the estimated parameter values for the Lubbock area Table 8 5 Mode choice model coefficients Drive Shared Variables Alene Ride Transit Walk Bike Alternative Specific Cpe es cera 3 780 2 86 4 520 4 65 4 620 7 05 7 540 7 68 IVTT min 0 035 0 035 0 035 lt OVTT min 0 070 0 070 0 070 lt Travel Cost Income less than 25K cents ee EOP ve a erseee Travel Cost Income between 25K and 50K 0 0026 0 0026 0 0026 lt cents Travel Cost Income greater than SOK cents 0 00095 0 00095 0 00095 58 All the estimates are intuitive and consistent in direction sign of coefficients The ratio between OVTT and IVTT is 2 0 exactly the same as indicated in the CAMPO mode choice model see the 2013 CAMPO Travel Demand Model 2013
158. vides information on area type population for each of the block etc Next we combined the TAZ file with the network file in order to map network streets over the TAZ configuration By mapping the network configuration on to TAZ configuration the analyst can easily identify the TAZ centroid and perform 10 A centroid is a point that represents the center of a TAZ zone and all the traffic is assumed to be generated from and attracted to this point It serves as a virtual link which loads the traffic from a TAZ to the main network In general the assumption is that no time is consumed in traversing the centroid link but one can assume a speed for a 50 further calculations using the TAZ centroid as a starting point In our skim generation we assumed a zero travel time for the centroid link Then all the travel times were calculated from TAZ centroid to TAZ centroid To generate the TAZ pair IVTT we used the multiple shortest path module of TransCAD software The steps involved in creating skims in TransCAD are not described here Interested readers are referred to the TransCAD user guide Chapters 3 and 13 for the complete set of steps involved in creating multiple shortest paths TransCAD Out of vehicle travel time To generate the OVTT skim we used the assumptions documented in CAMPO s demand modeling document CAMPO 2013 Specifically CAMPO uses an area classification scheme to determine the OVTT Under this scheme the OVT
159. y Thanksgiving and Christmas Eve and Day The fares for the bus service are listed in Table 6 1 Table 6 1 Transit fares for Bryan College Station MPO Cash fares Tickets and passes Regular fare 1 50 MultiRide pass 42 one way trips 55 00 Children 6 12 0 75 Ticket book 40 one way trips 60 00 Seniors and disabled pass 40 one way Children under 6 with paying customer Free trips 30 00 Seniors 65 and over 0 75 Monthly summer pass kids 6 18 25 00 Disabled 0 75 Semester pass for college students 70 00 Medicare with Medicare Card 0 75 with proof of registration Transfers one per trip Free Source http vww btd org FixedRoutes htm The District consists of seven bus routes that run every hour The routes along with the number of stops the number of stops does not count the ending stop which is the beginning point are the following e Purple Route 38 stops e Blue Route 32 stops e Green Route 29 stops e Maroon Route 27 stops e Yellow Route 22 stops e Red Route 28 stops e Orange Route 25 stops 6 2 2 San Angelo MPO The San Angelo MPO transit system is TRANSA Urban TRANSA Urban has six fixed routes that operate Monday through Friday from 6 AM to 6 PM and on Saturday from 7 30 AM to 6 30 PM The fares are presented in Table 6 2 gt The District route map can be found at http www btd org images B CS 20Map 20SEPTEMBER 2017 202012 2011x17 20WEB pdf 42 Table 6 2 Transit fares fo
160. y of TxDOT or TxDOT TPP are used to obtain information on trip mode choice traveler characteristics and trip purpose characteristics while supplementary land use and transportation system data are used to generate O D characteristics and transportation system characteristics Both MNL and NL models are derived from random utility maximizing behavior at the disaggregate level Formally the utility is as shown in Equation 4 1 30 Equation 4 1 Utility function U true utility of mode i to the individual q deterministic or observable portion of the utility Va U p SVa t Ey estimated by the analyst for mode i and individual q error or the portion of the utility unknown to the analyst for mode i and individual q In the MNL model the error term Eji is an unobserved term associated to alternative i In the NL model the error term can also be associated to the nest to which the alternative i belongs see Koppelman and Bhat 2006 The systematic portion of utility can have any mathematical form but the function is most generally formulated as additive to simplify the estimation process as shown in Equation 4 2 Equation 4 2 Deterministic component of the utility function yV deterministic or observable portion of the utility ai estimated by the analyst for mode i and individual q K parameter which defines the direction and importance of Vi gt Bix X gik Ban the effect of attribute k on the utility of

Developing a Mode Choice Model for Small and

Contents

Download Pdf Manuals

Related Search

Related Contents