SUTRA : Sustainable Urban Transportation for the City of Tomorrow

Preliminary Report SUTRA : Sustainable Urban Transportation for the City of Tomorrow WP 03: Multi-modal Transportation Modelling D03.3 User Manual and Example Test Data Sets First Draft Prepared for: Commission of the European Communities Research Directorate-General Prepared by: PTV Planung Transport Verkehr AG Author: Josef Janko Karlsruhe, January 2002 PTV Planung Transport Verkehr AG

Executive Summary Executive Summary Transportation problems are among the most pressing strategic development problems in many cities, often a major constraint for long-term urban development in general. In SUTRA these problems are addressed with a consistent and comprehensive approach and planning methodology that helps to design strategies for sustainable cities. This includes an integration of socio-economic, environmental and technological concepts to improve forecasting, assessment and strategic policy level decision support. It uses traffic equilibrium modelling to evaluate alternative transportation policies, including multi-modal systems and their relations to land use, technological development, socio-economic development, and spatial and structural urban development in general. The Traffic Assignment software package VISUM has been enhanced with modules to allow the modelling of particular scenarios of sustainable transportation, namely Park+Ride, High Occupancy Vehicles, and Road User Charging. The new methods and modules are available to the project partners. As a number of decision support indicators will be estimated from dedicated models (for emissions, air quality, public health, economic and energy system analysis) relying on the output of the transport model their relevant input data hav to be made available. Specifications and interface requirements have been defined between the developers of the models. The data structures of the transport models have been adapted to enable data exchange in both directions between the transport model and the emission model TREM as well as post-processing of emission data through the transport model. Data exchange with the other models is guaranteed by the standard VISUM output functions in the form of text files or through the Windows clipboard. PTV AG 01/02 page 3/22

Contents Contents Executive Summary...3 Contents...4 1 Introduction...5 2 Park+Ride...6 2.1 Introduction...6 2.2 Modelling Procedure...7 2.3 Park+Ride An Example...10 3 High-Occupancy Vehicles...15 3.1 Introduction...15 3.2 Modelling Procedure...15 4 Road User Charging...17 5 Interfaces between the Transport Model and the other Models..18 5.1 Emission model interface...18 5.2 Energy model interface...20 5.3 Public health model interface...20 5.4 Environmental impacts model interface...21 5.5 Economy impacts model interface...21 6 References...22 PTV AG 01/02 page 4/22

Introduction 1 Introduction This user manual is the third in the set of deliverables produced for WP03: D03.1 Multi-modal transportation modelling (Implementation Report) D03.2 Transportation Model prototype D03.3 Transportation model: User Manual and example test data sets It explains the steps necessary to model Park+Ride and High-Occupancy Vehicles and describes the defined interfaces between the traffic assignment software and other models applied within SUTRA. The application of the Road User Charging Module is documented in the VISUM Manual. PTV AG 01/02 page 5/22

Park+Ride 2 Park+Ride 2.1 Introduction Park+Ride describes travels in conurbations from the outskirts to the centre, which are undertaken partly by private and partly by public transport. The first part in the less dense outer areas with usually less density of public transport is made by car. The car is parked at a station or a stop of a bus or rail service, from where the second part of the journey is continued to the destination in the centre of the city. In this way one tries to combine the advantages and to avoid the disadvantages of both transport systems. Figure 2.1 shows an example for Park+Ride in an urban area with the private transport part of travel in red and dedicated P+R services in blue. To the potential clients four facilities are available, which are linked to the city centre with bus lines. Figure 2.1 : Example for a P+R system Modelling of Park+Ride trips requires a sequence of steps. The approach can be outlined as follows: Definition of P+R sites PTV AG 01/02 page 6/22

Park+Ride Determination of P+R demand Split of P+R demand into private transport and public transport legs Assignment of P+R trips as parts of the demand segments for private and public transport 2.2 Modelling Procedure The general modelling assumption is, that P+R trips are genuine private transport travels, which can be shifted to public transport as a consequence of an improved service. While the first and the third step are made by the assignment model, the second step has to be made externally. According to the availability of data different levels of model detailing are possible. Road Network PrT C2 Public Transport to the City Centre PuT S PuT C P+R L PrT C1 Figure 2.2 : P+R L PuT S PuT C PrT C1 PrT C2 Transport model elements of a P+R facility Park+Ride Link Public Transport Stop Public Transport Connector Private Transport Connector Private Transport Connectors between the P+R zone and the destinations (in the city centre) for the determination of P+R potential In the network model a P+R site has to be represented as a zone and a link (Figure 2.2). Zones are necessary as elements, where travels within a transport system start and end. A change between private and public transport therefore requires a PTV AG 01/02 page 7/22

Park+Ride zone. In the P+R links of the model two properties can be included, the capacity of the P+R facility and the costs for using P+R. The P+R links, like all the other links in the network model, have a limited capacity, which is here defined by the number of parking places and number of changes during a day. This allows to include different trip purposes with different durations of stay (commuters, shoppers). Also costs for parking and bus fares can be defined in the impedance of these P+R links. In a first step the impedances for the public transport legs of P+R trips have to be defined with the Assignment model. This requires a selection of potential destination zones of P+R trips by the modeller. Parameters for this determination would be the destinations and journey times of the available public transport services for the existing or potential P+R sites. The destination zones have to be connected to the P+R sites. A first assignment run is made in which potentials of clients using the P+R sites are estimated for the whole modelling area in competition to private transport on the whole journey. Figure 2.3 depicts the transport model representation of a P+R facility with the real public transport service and the modelled connections to some of the zones in the city centre. Figure 2.3 : Transport model representation of a P+R facility in the urban context PTV AG 01/02 page 8/22

Park+Ride The maximum figures of potential clients for a P+R system have to be reduced to actual ones. This can be done depending on the availability of data. If for a city no data at all are at hand, data from other cities should be adapted. If there already P+R sites are in operation, usually data about their usage are available. In an optimum situation data for user behaviour are available (from stated preference and/or revealed preference surveys). Depending on the character of these input data demand determinations are possible ranging from simple percentage estimations to detailed logit models for mode choice. In the third step another assignment run is necessary to assign these actual P+R demands to the public transport leg and to provide their correct return trip. As it is not mandatory in the trip assignment that for both directions of a journey the same connector of a zone is used, it has to be guaranteed, that P+R customers start their private transport return trip from the same place where they left their car. Therefore the P+R return trips have to be removed from the original demand matrix for the private transport segment and have to be entered as new relations between the P+R site and their final destination. In this same model run the P+R clients have to be assigned to the public transport services between the P+R facilities and the city centre. In detail this requires the following steps: 1. Preparation of the network model Definition of the P+R sites as links and zones Definition of the public transport services for these sites. Estimation of potential P+R destinations through determination of isochrones for the relevant public transport services, or based on local knowledge, or based on survey results Definiton of the relevant connectors between the P+R sites and the destinations 2. First assignment to estimate the demand matrix for the P+R clients; validation against reference data. 3. Modification of travel demand Select link analyses for the P+R site links generate the matrices of P+R trips from the origins to the final destinations in the city centre (original P+R trips). These trips are removed from the private transport matrix. The P+R facilities are defined as destinations of the original P+R trips, and these trips are added to the private transport demand. The P+R facilities are defined as origins of the original P+R trips, and these trips are added to the public transport demand. PTV AG 01/02 page 9/22

Park+Ride The return trips have to be treated in the same way. The single steps of this modification are best done with Scripting. This allows for a planned and documented handling of the matrices and further an easy and safe version management. 4. Second assignment with modified demand for the public and private transport systems. The whole procedure has to be calibrated in a possible iterative process, if the shift from private to public transport is of a magnitude which has influence on the mode choice decision: if the reduction of private traffic is so big, that congestion is relieved, car usage might be encouraged again. 2.3 Park+Ride An Example This example is drawn from a multi-modal traffic study for the area of Nottingham, where the implementation of the P+R model could already be tested and applied. A number of scenarios have been evaluated with a range from four to twelve P+R sites, from which one has been selected here for demonstration. In a first stage an assignment is made without modelling P+R trips Results are given in Figure 2.4. Displayed are the Park+Ride facilities, the road network (black), railway connections (magenta) and LRT lines (cyan). The private transport vehicle flows are coloured in red, the public transport passenger figures in blue. In a second stage an assignment determines the potential maximum of P+R clients (Figure 2.5). These are all trips in private transport, which might have an advantage from using the P+R facilities. PTV AG 01/02 page 10/22

Park+Ride Figure 2.4 : Original Flows and Passenger Figures without P+R traffic PTV AG 01/02 page 11/22

Park+Ride Figure 2.5 : Potential Vehicle Flow Changes caused by P+R (direction to the city centre only This figure displays only the one direction of the trips, which is important for the choice situation of the travellers. Dark red are flow increases, light red flow reductions. This potential demand between the P+R sites and the city centre has to be reduced to actual figures. This can be done based on survey data or through the application of utility models. Following the determination of the actual P+R demand a final assignment run is necessary to obtain the actual flows in private transport and passenger figures in public transport. PTV AG 01/02 page 12/22

Park+Ride Figure 2.6:Actual Flows and Passenger Figures with P+R Modelling A more detailed figure of the volume changes shows the effects of the individual P+R facilities. PTV AG 01/02 page 13/22

Park+Ride Figure 2.7:Changes in the Flows and Passenger Figures through P+R Private transport flows are given in red (dark: growing light: reduced), passenger figures are given in blue (dark:reduced light: increasing). PTV AG 01/02 page 14/22

High-Occupancy Vehicles 3 High-Occupancy Vehicles 3.1 Introduction The term High-occupancy vehicle (HOV) can include buses, vanpools, carpools, and other authorised vehicles. The focus of projects to promote HOV lies on carpools; most facilities use a two person (2+) per vehicle carpool definition, but some require three person (3+) carpools. Incentives for participating in a carpool can be Availability of dedicated HOV lanes Parking spaces at convenient locations only for HOV Exemption from road user charging for HOV In the transport model high-occupancy vehicles also are regarded as carpools. They have to be defined as a transport system of their own. Dedicated HOV links can be prepared in the network model which are blocked for other vehicles. Where no particular HOV links are available, these vehicles join the other vehicles in common links. Demand for carpooling has to be determined in a demand model. These trips may be defined in two categories: The whole trip is done in a carpool. An estimated share of demand for private travel has to be split from the private transport matrix and moved to a HOV matrix. Carpool members start their trips separately, meet at an agreed place (e.g. beginning of a HOV lane), and travel then together. In this case the treatment of HOV demand is similar to the processing of P+R trips. The demand has to be removed from the private transport matrices first and then added back again for the private transport legs of the trips, while the HOV legs are part of the HOV demand. 3.2 Modelling Procedure The modelling procedure is similar to Park+Ride: 1. Preparation of the network model Definition of HOV as a new transportation system Definition of the HOV meeting sites as links and zones PTV AG 01/02 page 15/22

High-Occupancy Vehicles Addition of HOV incentives to the network, e.g. additional lanes dedicated to transport system HOV and closed for other systems or common lanes for bus and HOV only Estimation of potential destinations for the High Occupancy Vehicles, based on local knowledge, or based on survey results Definiton of the relevant connectors between the HOV meeting sites and the destinations 2. First assignment to estimate the demand matrix for the HOV clients; validation against reference data. 3. Modification of travel demand Select link analyses for the HOV site links generate the matrices of HOV trips from the origins to the final destinations in the city centre (original HOV trips). These trips are removed from the private transport matrix. The HOV meeting sites are defined as destinations of the original HOV trips, and these trips are added again to the private transport demand. The HOV meeting sites are defined as origins of the original HOV trips, and these trips are considered as part of the demand within the transport system HOV. The return trips have to be treated in the same way. The single steps of this modification are best done with Scripting. This allows for a planned and documented handling of the matrices and further an easy and safe version management. 4. Second assignment with modified demand for the private transport systems and HOV. The whole procedure has to be calibrated in a possible iterative process, if the shift from private transport to HOV is of a magnitude which has influence on the mode choice decision: if the reduction of private traffic is so big, that congestion is relieved, car usage might be encouraged again. PTV AG 01/02 page 16/22

Road User Charging 4 Road User Charging Road user charging has been applied on motorways in a number of countries for a long time. In cities or dense urban areas however road user charging has been not been applied widely. Prominent exception is the city of Singapore, where a charging scheme is effective now since 30 years. Few cities in Europe have installed similar projects as case studies during the last years. Genoa, one of the SUTRA city partners, prepares an area wide scheme for its centre. Conventional methods to model road user charges apply a constant value of time which can be included in a generalised cost function of a monocriterion assignment method [1]. TRIBUT is a bicriterion traffic assignment method which equally considers travel time and cost. The trip choice between different paths is modeled by defining the value of time as a random variable with a distribution of the log-normal type, thus considering that each trip has a specific willingness to pay toll for travel time reduction. This approach offers a significantly better price elasticity than monocriterion methods. Its most prominent features are randomly distributed values of time, the principles of path search and path choice. Furthermore it presents different aspects of the application in practice, in particular the definition of different demand classes, the modelling of linear or non-linear pricing schemes and the value of time estimation. A detailed description of the TRIBUT method can be found in [2]. Within the transport model a wide range of scenarios can be tested: Charging for entering areas (city centre) Charging for use of single parts of the road network (bridges, motorways) Charging of particular users (goods vehicles, passenger cars with only one person in it) Charging at peak hours Combinations of all these An example data set will follow as soon as scenarios from the SUTRA cities are available. PTV AG 01/02 page 17/22

Interfaces between the Transport Model and the other Models 5 Interfaces between the Transport Model and the other Models 5.1 Emission model interface Estimations of emission productions are carried out in the framework of this project with TREM (Transport Emission Model for Line Sources). A description of this model can be found in Deliverable D04.1 [3]. It requires as input from the transport model the numbers of vehicles in each link of the network, and the information, which share of these vehicles is driven under cold engine conditions. These conditions are defined as so-called cold distances individually for different pollutants and vehicle categories. Table 5.1 shows the formulae of cold distances for three vehicle engine technologies and five pollutants. Engine Technology Pollutant Cold distance d c Boundary distance Gasoline cars with catalyst CO 2 d c = 0.29V 0.05 d c = 0 CO d c = 0.24V 0.14 d c = 0 HC d c = 0.06V + 2.19 NO x d c = 0.19V + 3.40 FC d c = 0.24V + 0.54 Diesel cars without catalyst CO 2 d c = 0.24V + 0.09 CO d c = 0.08V + 4.83 HC d c = 0.08V + 4.83 NO x d c = -0.07V + 7.50 d c = 0 FC d c = 0.13V + 3.42 Gasoline cars without catalyst CO 2 d c = 0.15V + 2.68 CO d c = 0.04V + 5.42 HC d c = 0.09V + 1.94 NO x d c = 0.02V + 2.83 FC d c = 0.28V + 0.47 Table 5.1: Cold Distance d c [km] as a function of the average speed V [km/h] PTV AG 01/02 page 18/22

Interfaces between the Transport Model and the other Models Software has been developed, which enhances the transport model data structures with additional variables to store these data within the network assignment results and which makes these data available to the emission model. An example for possible graphical output is given in Figure 5.1, which shows the shares of vehicles with cold engine for one vehicle category and one pollutant (gasoline cars, CO 2 ). In the ideal case for each vehicle category an individual demand segment is defined. If this is not possible, percentages must be given for splitting single demand segments into these categories. Share of vehicles with cold engines 0... 10 % 11... 20 % 21... 30 % 31... 40 % 41... 50 % 51... 60 % 61... 70 % 71... 80 % 81... 90 % 91... 100 % Figure 5.1:Distribution of Gasoline Vehicles with CO 2 related Cold Engines The calculated data are stored to an Excel sheet. An example of this data extraction with the three vehicle categories and five pollutants is given below (Table 5.2). From the VISUM data structures all kinds of output or post-processing are possible as for any other link attribute. The transport model data structures are designed flexible, so that a treatment of additional vehicle categories and pollutants is possible. PTV AG 01/02 page 19/22

Interfaces between the Transport Model and the other Models Petrol-Cat Vehicle Flows with Cold Engine for Diesel Vehicle Flows with Cold Engine for Petrol-without-Cat Vehicle Flows with Cold Engine for Link From To nr Node Node Flow CO 2 CO HC NO x FC CO 2 CO HC NO x FC CO 2 CO HC NO x FC 3 46 10172 2542 1106 639 0 367 639 183 0 0 0 26 34 0 0 0 139 3 10172 46 1366 894 894 894 894 894 255 255 255 0 255 128 128 128 128 128 5 62 10161 163 102 102 13 66 102 29 7 7 0 8 5 2 3 0 15 5 10161 62 136 85 85 85 85 85 24 24 24 0 24 12 12 12 12 12 6 100 716 390 239 239 239 239 239 68 68 68 0 68 34 34 34 34 34 6 716 100 494 265 237 19 230 237 68 24 24 0 34 30 3 17 0 38 7 24 716 1607 919 892 106 874 892 255 79 79 0 125 122 9 63 0 131 7 716 24 866 559 559 509 559 559 160 160 160 0 160 80 73 80 30 80 8 24 10150 4642 3095 3095 774 2878 3095 884 370 370 0 423 376 68 212 0 442 8 10150 24 1917 1165 1159 554 1072 1159 331 210 210 0 273 137 61 115 61 166 9 25 10150 1954 1165 1159 736 1072 1159 331 224 224 0 273 141 61 121 61 166 9 10150 25 4692 3124 3124 803 2906 3124 893 378 378 0 431 338 72 216 0 446 10 716 10150 50 29 29 29 29 29 8 8 8 0 8 4 4 4 0 4 10 10150 716 37 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11 45 716 526 349 349 299 349 349 100 100 100 0 100 50 43 50 0 50 11 716 45 1150 655 655 81 644 655 187 55 55 0 91 78 6 46 0 94 12 62 10156 4213 2659 2659 877 2436 2659 760 295 295 0 349 285 125 172 0 380 12 10156 62 5584 3300 2797 975 2120 2797 799 279 279 0 414 225 67 169 0 471 13 25 10151 4168 2745 2743 1901 2654 2743 784 554 554 0 659 356 256 298 0 392 13 10151 25 2057 1331 1291 659 903 1291 369 231 231 0 251 129 94 116 73 190 14 25 11195 7585 4870 4870 3027 4715 4870 1391 935 935 0 1038 522 372 519 0 696 14 11195 25 7063 4052 3296 699 2657 3296 942 262 262 0 492 281 34 147 0 578 Table 5.2: Output table from transport model results (extraction) 5.2 Energy model interface In the MARKAL-LITE specification as input is required the total distance covered by the vehicles of the different transport systems over the period of one year. The transport model produces link volumes as constant values over the given simulation period. The link volumes can be disaggregated into the different demand segments. The required data can be calculated by first summing up the figures for the whole network and then expanding them to the desired time period. These expansion factors have to be provided by the city partners. 5.3 Public health model interface No written specification is available for the public health model input data requirements. It is assumed that the required data can be derived from the parameters link volumes, link lengths, and journey times. PTV AG 01/02 page 20/22

Interfaces between the Transport Model and the other Models 5.4 Environmental impacts model interface No specification is available for the environmental impacts model input data requirements. It is assumed that the required data can be derived from the parameters link volumes, link lengths, and journey times. 5.5 Economy impacts model interface No specification is available for the economy impacts model input data requirements. It is assumed that the required data (if any) can be derived from the parameters link volumes, link lengths, and journey times. PTV AG 01/02 page 21/22

References 6 References [1] U.S. Department of Commerce, Bureau of Public Roads (ed.): Traffic Assignment Manual. Washington, D.C. (1964) [2] Barbier-Saint-Hilaire, F., M. Friedrich, I. Hofsäß, W. Scherr : TRIBUT a Bicriterion Approach for Equilibrium Assignment. Traffic Engineering & Control, April 2000. [3] anon., Department of Environment and Planning, University of Aveiro [ed.] : TREM Transport Emission Model for Line Sources Methodology. SUTRA Deliverable D04.1 PTV AG 01/02 page 22/22