10th ITS European Congress, Helsinki, Finland 16 19 June 2014 TP-SP 0051 Multi-level transport systems model for traffic management activities Jacek Oskarbski 1, Kazimierz Jamroz 2 1. Gdansk University of Technology, Poland, 80-233 Narutowicza 11/12, +48 604475876; jacek.oskarbski@pg.gda.pl 2. Gdansk University of Technology, Poland Abstract Intelligent Transportation Systems offer tools for strengthening transport systems in cities by affecting the transport behaviour change of inhabitants to rationalize the use of existing infrastructure and increase reliability while reducing operating costs of transport. The Tricity Agglomeration is currently under implementation of the first stages of the transport management system TRISTAR. One of the TRISTAR elements will be Transport Planning System (TPS) supported with information obtained from other modules of the TRISTAR system. Transport Engineering Department in Gdansk University of Technology initiated the development and implementation of an integrated, hierarchical forecast and analysis system MST (Multilevel Model of Transport Systems). A key element of such a model will be the possibility to acquire and use data from the TRISTAR. Models of transport systems (in the first step for the city of Gdynia) are developed within the EU project CIVITAS DYN@MO. Keywords: transport management, traffic control, transport planning Introduction The modern city management both at the level of planning and strategic decision-making in the development of the transport system as well as at the level of operational decisions concerning for example traffic organization requires the use of better and more modern tools. The use of modern technologies can improve the operational area of traffic management, but can also support activities related to the planning and design of transport systems solutions to current traffic organization. Given the above benefits the Tri- City authorities have taken steps to implement the measures of the ITS. Gdynia, Sopot, and Gdansk are three Polish cities very close to one another at the Baltic Sea coast. They form an agglomeration referred to as
Tricity whose total surface area is 4152 and whose population is 750,000 persons. The Tricity has nearly 1,400 km of roads, 300 signalized intersections and over 2,000 daytime public transport lines (buses, trams, trolleybuses and trains). The Tricity faces transport-related problems that are typical of large agglomerations. In the modal split the share of public transport has decreased in the recent years and the share of cars has increased. Currently the modal split in non-pedestrian travel is 46%/54% (public transport/private transport) and the car owner ship rate is approx. 500 cars/1,000 inhabitants. Each city has its own administration, to include road management administration and, therefore, cooperation in such a complex area as traffic management is a true challenge [1]. The first conceptual work on the structure of the transport management system TRISTAR began in 2002 with development of the concept of an integrated system for the Tricity ring road and the city of Gdynia. In subsequent years the Highway Engineering Department in Gdansk University of Technology has developed a traffic management system concepts for the cities of Sopot and Gdansk. Developed concepts contributed to the continuation of actions to implement the system. In 2006 the mayors of Gdansk, Gdynia and Sopot signed an agreement to undertake joint activities aimed to prepare a proposal for external funding for the construction of the system, which formed the basis for the start of the cooperation of the various cities in order to achieve the objective on the basis of consensus the detailed concept of TRISTAR system was developed in 2007 [2]. In the same year, a demo project was implemented in Gdynia along Morska Street, over the distance of 3.6 km which included 9 signalized intersections. As a part of the project, the SCATS system (a traffic control system delivered by Roads and Maritime Services from Australia) and the RAPID system (public transport management system delivered by Sigtec from Australia) were installed; these systems were the initial elements of a priority system for public transport vehicles. The results of the demo project were encouraging because implementation of the SCATS system improved the traffic conditions on the Morska Street despite a small increase in traffic. Improvement was noticed in travel time both by car (a reduction by 12%) and, to a greater extent, by public transport (a reduction by 18.5%). In the 4 years prior to the demo project, the share of public transport travel in non-pedestrian travel decreased by 4.3%. Despite this downward trend, after the system was implemented, the number of passengers in the demo project area increased by 6%. The results enabled defining realistic objectives for the construction of the TRISTAR traffic management project in the Tricity. The objectives are: to improve the traffic conditions by providing traffic management tools and increasing the share of travel by public transport by improving its competitiveness thanks to the use of ITS technologies. In the case of the TRISTAR system, a unique method was selected: the works would be divided into two parts, namely construction works and other works (IT and ITS), and performed under different types of contracts. The design documentation is very extensive: it comprises over 9,200 single designs. Work on the documentation started in the second half of 2009 and ended in 2010. In August 2011 an unlimited tender was announced for the implementation of the TRISTAR [1]. 2
Currently the implementation of the first stages of the system is going on, including the basic layout of the Tricity street with about 150 intersections and pedestrian crossings equipped with traffic signals, communication infrastructure (including fiber optic cable) with a length of about 100 km, installation of 60 video surveillance cameras (CCTV), 60 cameras to identify vehicles (ANPR), about 70 variable message signs at the public transport stops, more than 20 variable message signs for drivers and equipment nearly 700 public transport vehicles in position transmitters and on-board computers [3]. The target system will be expanded and will include all intersections equipped with traffic signals and will be integrated with the planned systems on express roads (Tricity Bypass, South Bypass) as well as airports, rail, Tricity sea harbours. Assumptions for the implementation of the MST Planning of transport networks is a fairly complicated process, because for its construction and expansion weighing decisions are needed. Building of efficient transport networks requires to solve complex optimization problems in transportation. The results of analysis carried out by using various tools, which include various types of forecasting software are used to facilitate the process of making planning decisions. Forecasting is based on scientific analysis, to formulate opinions about the future state of the studied phenomenon on the basis of knowledge about its current course [4]. In the case of transportation systems forecasting mostly concerns the prediction estimation of the future (medium or long term) traffic volume or volume of passenger travel or freight for existing or planned transport network. Need for use of traffic models to forecast and test transport networks is obvious due to the inability to survey the events and relationships occurring in the transport networks in an empirical way [5]. Experience of conducting analysis and forecasting of traffic in the Pomeranian Region, indicate the need to organize the approach to develop predictive models for the analysis of transport. In taking up this challenge, the given below assumptions for system design concept of forecasting and traffic analysis should be foreseen. According to these assumptions the system of forecasting and traffic analysis should [5]: - cover an area of Pomeranian Region (first step is to complete the model for the city of Gdynia within CIVITAS DYN@MO), but to be fed with data from regional and national model - have a hierarchical structure consisting of layers of different management levels - enable the efficient exchange of information and data between layers of management - enable the provision of data to the various tasks arising from the process of functioning of the transport systems and facilities, which will facilitate optimal decision-making - provide data for planning work, feasibility studies of transport facilities, projects of changes in the traffic arrangement plans, taking into account geometric solutions at 3
intersections and interchanges and advanced traffic control - set up data and information for updating Gdynia s (in the first step) Sustainable Urban Mobility Plan (SUMP) and provide detailed analyses and verification of effects on mobility management initiated by SUMP proposals and measures - provide simulation tools to persuade inhabitants to proposed solutions Assumptions given above cannot be achieved with a single traffic or transport model. This will be possible, however, by building an integrated, multi-level model of traffic forecasting and analysis (Table 1). The approach will be multi-level and multi-layer model based on the idea adopted by the Department of Planning and Transportation of the City of London [6]. The levels are determined by type of administrative area and the layers are determined by kind of transport analysis or transport management. MST (Multilevel Model of Transport Systems) will be fed with data from detection systems located in the architecture and infrastructure of the TRISTAR system and ultimately also using localization devices in vehicles and mobile phones. Strategic level of MST includes the provision of data to develop a transport policy, the implementation of planning studies and network studies. At the strategic level macroscopic model (based on the VISUM program [7]) will be applied. The model requires disaggregation of districts within the area of the City of Gdynia together with detailed road networks and elaboration of transport lines and running of public transport within the city. The inputs to the model (for the calibration of the private and public transport model, calibration of resistance functions on street sections) will be information obtained from TRISTAR system such as traffic volumes, vehicle classes, average speed on sections of the street network, network element capacity, public and private vehicles travel times and time of public transport passengers service. The model will be used to collect the data for the mezoscopic model, complying with the planned routes located in the vicinity of the city [5]. Examples of the use of a macroscopic model are to estimate the effectiveness of the introduction or changes in public transport lines and the closure of chosen streets in the city centre. Tactical level includes the provision of data to develop a decision-making papers (network and corridor studies, feasibility studies), development projects of traffic arrangement, traffic control and evaluation of planning solutions effectiveness as well as for traffic management purpose (for example for testing new signal programmes, evaluating the effectiveness of chosen streets closure or introduction of bus lanes). The object of research will be the transport network, a string of roads or streets, the segment of public transport line in this case. At the tactical level mezoscopic model (based on the VISUM and SATURN programs [8] as well as BALANCE offline program [9] implemented within TRISTAR system) will be applied. This model will be used to analyse the scenarios of traffic arrangement modifications as well as to estimate the efficiency measures of planned modifications. The model will derive the results from the macroscopic model within the scope of demand modelling with a parallel division into particular modes of transport. This will enable calibration with taking into account the road network within the Pomeranian Region 4
(results from the macroscopic model). The model will be powered and calibrated with data from TRISTAR (in the future the implementation of the public transport passenger counters in vehicles to update and calibrate the models will also use the data to statistical analysis, respectively) such as traffic volumes, vehicle classes, average speed on sections of the street network, vehicles travel times, saturation flows, split, offset and cycle length (when needed) of traffic signals and headways between vehicles. The macroscopic and the mezoscopic models will provide the basis for the elaboration of the microscopic model (simulations) which will be powered by current or historical data from the data warehouse of TRISTAR system and traffic control system BALANCE [9]. Table 1- Use of transport modelling tools depending on the level of management and area Area Country Level of management Object Model kind Strategic Tactical Operational Transport network Dedicated - prognostic Transport network/street and PT lines Advanced Street, PT line, set of junctions, juction Simulation Model type Macroscopic Mezoscopic Microscopic VISUM VISUM Region VISUM VISUM Agglomeration Tool VISUM VISUM, SATURN City/cordon/ local VISUM SATURN, TRANSYT, DRACULA VISSIM, DRACULA Operational level includes the provision of data to develop specific projects of traffic arrangement, traffic control programs, projects of transportation services for the selected objects and primarily visualize (simulate) the operation of transport facilities. The object of research will be a section of road or street, section of the public transport line or junction. At the operational level microscopic model (based on the VISSIM program [7] or DRACULA program [8]) will be applied. This model will allow for verification and demonstration of results obtained from the macro and mezoscopic models and will provide important input for the community projects and the planned internet platform (Mobility 2.0) within CIVITAS DYN@MO to communicate with inhabitants. As part of the first stages of the TRISTAR system implementation the macroscopic model will be used - for the development of traffic management scenarios in the case of incidents occurrence, road works, public events as well 5
as for planning of new public transport routes (lines) or change routes (lines) of public transport vehicles with the possibility to analyze changes in the timetable (the VISUM). Microscopic model will be used for the simulation of changes of traffic control parameters and analyzes of traffic control strategies (VISUM, BALANCE offline, VISSIM). After implementation of TRISTAR system the models will be supplied with the information on traffic parameters in real-time [5]. Multi-level Transport Systems Model in Transport Planning System In the area of Gdynia models will be extended to the entire city and developed within the EU's CIVITAS DYN@MO project (CIVITAS II Plus). The delivered software enables simulation of the effects of road incidents (collisions, accidents, congestion) and mass events into transport operations quality. Described models will derive the necessary data for calibration or simulation directly from the databases of TRISTAR system. For the preliminary analysis, aimed at optimizing the parameters of traffic control can be used TRANSYT [10], Vistro [7] or Saturn programs. TRANSYT program can only be used for the primary analyzes, due to the static - fixed distribution of traffic in the network model, which does not reflect actual travel behaviour of drivers, due to the fact that any change of signalling parameters will change the loss of time, capacity and queue lengths on individual intersections, which in turn changes the choice of routes. It is therefore appropriate to extend the analysis using packages which allow to take into account the dynamic distribution of traffic (e.g. SATURN program). The Transport Planning System also involves the use of CROSSIG program [9] to assist in the preparation of traffic signal projects that data on traffic streams will be derived from the BALANCE traffic control system and other elements of a multilevel model. Implementation of traffic control system consists in moving into the controller operating system environment the so-called TRENDS Kernel. The TRENDS Kernel enables embedding in the controller a local Entire Priority Intersection Control System (EPICS). The EPICS system collects traffic data from inductive loops (or other detection systems) at intersections. On the other hand, the Balancing Adaptive Network Control Method (BALANCE) system ensures optimum operation of controllers in groups (control areas). The BALANCE system uses a traffic model, e.g. from the VISUM software, or the integrated DRIVERS and optimizing algorithms. It is possible to change the target functions in the optimization performed by the BALANCE system to reach the best Performance Index. The following factors can be considered: the vehicle delays/the number of vehicle stops/the queue length. The factors can be assigned different weights. The BALANCE system communicates with the traffic signal controllers through a control module [1]. Data from detection systems of Urban Traffic Management System (component of TRISTAR), such as volumes and speeds, number of stops, queue length and the control parameters will be obtained through a traffic control system BALANCE / DRIVERS. 6
In addition, information will be used from traffic measurement stations which will allow detection of the instantaneous speed, traffic density and the traffic structure (class vehicles according to EUR 6 on sections between intersections or according to EUR 3 within junctions) which will allow e.g. trip matrices estimation and calibration for different types of vehicles in a multi-level model. Figure 1- Transport Planning System diagram process cycle The use of ANPR cameras in addition to the function of determining the travel time between cameras will also allow for the development of applications, enabling the estimation of trip matrix in real-time for use by SATURN, VISUM or BALANCE offline. Calibrated in the SATURN or VISUM matrices allow analysis of the impact of changes of control parameters on the efficiency of the transport system and selection of traffic control strategies. The SATURN program takes into account the reduction of transport demand and the accessibility of specific areas by using elastic assignment methods. Transport Planning System will be equipped with additional tools provided by GEVAS, enabling monitoring and analysis of quality of implemented priorities for public transport vehicles (VTmonitor / FAS) and the quality of the traffic control system operations (VTnet / pcoq), which will provide additional information on the performance of transport systems operation on individual intersections 7
within the TRISTAR implementation area, in addition to the analyzes carried out by means of analytical models of MST. Transport Planning System diagram is shown in Figure 1. Conclusions Modern transport systems management requires the use of better tools for forecasting and analysis of transport. Used computer programs enable the development of the transport system model for the selected object or area of street network contain the basic procedures of the mathematical model of the transport system. The role of the analyst is the selection of appropriate transport parameters of these models. Selected parameters should reflect actual travel behaviour of inhabitants and the characteristics of the area and transport networks in the current state as well as the rules for their development. Negative experiences with conducting analysis and forecasting of traffic in the Pomeranian Region require arrangement through the development of an integrated system for forecasting and analysis of transport. This should include specific areas of administrative management and the levels of traffic management. The basic idea is the possibility of cooperation between different systems and the ability to use their results by different institutions for different purposes of transport management. In addition, it also should be included some thoughts of a more general nature: 1) A very important part of many regional and local transport analysis are forecasting data obtained from the traffic model for national roads. This model, developed and distributed by GDDKiA (National Road Admininistration), however, has several drawbacks (lack of network updates, no taking into account the public transport lines and demands) and therefore requires regular updating and further development. 2) In the development of regional traffic forecasting models, a big obstacle is the lack of reliable data about the transport behavior of residents. A good solution would be to include some of the most important questions to regular surveys conducted by the Regional Statistical Offices or incorporated them into the Census conducted by the Central Statistical Office. 3) Development of reliable forecasting transport models requires a number of complex procedures, so a good option would be to develop guidelines for development of such models, which is lacking in Poland 4) Implementation of new technologies in transport provides access to data, which can power transport models. It will contribute to a reduction in expenses which are incurred for transport research and data collecting as well as contributes to increase the accuracy of modeling and improving the quality of transport models. 8
References 1. Jamroz, K., Oskarbski, J., Krukowski P. in The role of the Gdansk Science and Technology Park (Poland) in the creation of technologies that influence the development of cities. In Proceedings 30th IASP World Conference on Science and Technology Parks 2013, Brazil. IASP. 2. Oskarbski, J. (2011). Intelligent transport systems for agglomerations on the example of the Tricity, Komunikacja Publiczna 1/2011. 3. Oskarbski, J. (2011). Functional structure of the transport management system in the Tricity TRISTAR, Przegląd Komunikacyjny 7-8/2011. 4. Klóska, R., Hundert, M., Czyżycki, R. (2007). Wybrane zagadnienia prognozowania. Economicus, Szczecin. 5. Oskarbski J. (2011). Perspectives of Telematics Implementation in Tricity Transport Systems Management and Planning. Communications in Computer and Information Science. (Modern Transport Telematics), 239/2011. Springer Berlin Heidelberg. 6. Smith, J., Blewitt, R. at al. (2010). Traffic Modelling Guidelines. Traffic Manager and Network Performance Best Practice. Version 3.0. Transport for London. 7. PTV AG, http://www.ptvag.com/software 8. ATKINS, http://www.saturnsoftware.co.uk 9. GEVAS, http://www. gevas.eu 10. TRL, www.trlsoftware.co.uk 9