H. Taale and W.J.J.P. Schouten J. van Kooten

Transcription

1 DESIGN OF A COORDINATED RAMP-METERING SYSTEM NEAR AMSTERDAM H. Taale and W.J.J.P. Schouten J. van Kooten Dutch Ministry of Transport, Public Works and Water Management AGV Consultancy, The Netherlands INTRODUCTION In the Netherlands ramp-metering is applied since 1989, when the first ramp-metering system was installed on the on-ramp Hemhavens (S101), near the Coentunnel, of the A10-West, part of the ringroad around Amsterdam. Due to the good results obtained on this location, two other systems were implemented, one on the A13, near Delft and one on the A12, near Zoetermeer. A fourth system will come in operation this year. These systems are all local systems which meter the traffic on one on-ramp. In the framework of the DRIVE-II project EUROCOR (European Urban Corridor Control) it was decided to investigate coordinated ramp-metering. EUROCOR is a project that deals with ramp-metering and Variable Message Signs (VMS). For this project there are two test-sites: the Corridor Périphérique round Paris and the A10-West. On the Corridor Périphérique control strategies for VMS and urban traffic control are tested and on the A10 strategies for coordinated ramp-metering. This paper describes the project for the evaluation of strategies for coordinated ramp-metering on the A10. It focuses on the simulation studies, the design of the communication network and the assessment study. SIMULATION STUDIES Before it was decided to install the system for coordinated ramp-metering, two simulation studies were carried out. Use was made of the assignment and simulation program SATURN to analyze the impacts of ramp-metering on several on-ramps on network level, and of the simulation program FLEXSYT to analyze the impacts on local level. These simulation studies are discussed below. Figure 1. Modelled road network

2 SATURN Simulation Study SATURN is described in Hall et al (1) and Van Vliet (2). This model was chosen, because the input was available from another study dating from Of course the input had to be modified. The network used, is pictured in Figure 1. In the study a distinction was made between restrictive control and dispersive control. Restrictive control restricts the number of vehicles that is allowed to enter the motorway. Dispersive control does not restrict the number of vehicles, but smooths the on-ramp volume in time. First the input was calibrated for the current situation, based on the use of the on-ramps, the use of the A10-West, the location of congestion points and the use of the urban network. Then a number of alternatives was studied: - restrictive control on the Hemhavens (S101) and Nieuwe Havens (S102) ramps; - restrictive control on the S101 and S102 ramps and dispersive control on the Bos en Lommer (S104) and Geuzenveld (S105) ramps. - restrictive control on the S101 and S102 ramps and dispersive control on the S104, S105, Osdorp (S106) and Sloten (S107) ramps to the A10-West and the Osdorp (S107/to the A4) and Amstelveen (S108/to the A10-South) ramps. The results can be summarized in the main conclusion that ramp-metering will have a positive effect on the traffic flow on the A10-West: congestion will reduce and travel times will shorten. The urban network shows some changes in usage, but the traffic situation will not deteriorate. The first alternative leads to an increase of the volume at the on-ramp S104, which is undesirable, because of the complex situation there. In addition: the application of ramp-metering on the S104 and S105 ramps (the second alternative) leads to a much more stable traffic situation. Extending the application of rampmetering to the on-ramps south of the S105 (the third alternative) gives no additional advantage for the traffic flow on the A10-West. These results have led to the recommendation to apply ramp-metering on the on-ramps S102, S104 and S105. The study is reported by AGV (3). FLEXSYT Simulation Study The FLEXSYT study was conducted to determine the effects of ramp-metering on local level, that is on one on-ramp and the junction with the urban network. FLEXSYT is described in Middelham (4) and was used for its abilities to simulate traffic and traffic control programs in a most accurate way. Two junctions with the A10-West were selected: Nieuwe Havens (S102) and Geuzenveld (S105). The lay-out of these junctions are pictured in the Figures 2 and 3. The study focused on the influence of certain Figure 2. Lay-out of the S102 junction Figure 3. Lay-out of the S105 junction variables on the traffic situation on the motorway and the urban intersection, namely the capacity of the motorway, the on-ramp volume, the metering algorithm and the total demand. The results show that the situation for the S102 junction is very critical. Ramp-metering on the S102 causes traffic to choose an on-ramp upstream. This leads to more traffic on the A10 near the S102 onramp. This in turn causes a considerable deterioration of the situation on the intersection, even if the total traffic flow on the intersection is less. This is due to the longer metering cycles on the S102 on-ramp, calculated by the metering algorithm based on the actual situation on the motorway. On the other hand, if the assumption is made that the capacity of the motorway increases as a result of metering, it has a very favourable effect on traffic flows, particularly at the intersection. Simulations done with different percentages for the total traffic flows, showed that the situation is extremely sensitive to traffic flow. On the S105, ramp-metering starts to operate only if an artificial bottleneck is introduced on the motorway downstream. Blocking back then causes the switching on of the metering system. Once metering operates, almost immediately the maximum metering cycle is applied. This has consequences for waiting times at the on-

3 ramp and the intersection. If the metering algorithm is adjusted in such a way that the metering system is switched off when there is a severe congestion on the motorway, this brings about a great improvement in comparison to the algorithm without this facility. So, on this location the best thing to do will be to meter only when congestion starts to occur and when congestion is beginning to solve. The simulation study is reported by Taale (5). Result of the Simulation Studies Based on both simulation studies the Regional Directorate North-Holland of the Ministry of Transport decided to install metering systems on the S102, S104 and S105 ramps. For the DRIVE-II project EUROCOR it was decided to implement a coordination system for these metering systems and the existing metering system on the S101 ramp. There are three ways to implement coordination: - collect data from all on-ramps in each local system and calculate the metering rate in the local system; - collect data from all on-ramps centrally and calculate the metering rate centrally; - collect data from all on-ramps centrally and calculate an adjustment to the metering rate calculated locally. The third possibility was chosen, because a central coordination system was necessary anyway (for data collection and processing) and the local systems had to be adjusted only slightly in this case. An additional advantage is that the central system can easily be disconnected, while the local systems can continue to operate, which is important in case of a break down of the central system. For the implementation a communication network was necessary. The next chapter describes the development of such a communication network. DESIGN OF COMMUNICATION NETWORK A central control concept has been chosen to execute the coordination between the local ramp-metering systems and a computer system has been designed. This system, called "Central Traffic Management System" (CTMS), consists of a communication and file server, various work stations, and local systems which are linked with each other by a communication network. The main functions of the CTMS are traffic, operational and electro-technical management and control of various Dynamic Traffic Management applications on and along the motorway network, for example route information signs and ramp-metering systems. The link between the central system, the workstations and the local systems has been realised with a communication network which is based on national and international developments. Existing and future communication networks The communication network has been based on the future situation in which use will be made of the Roads Telecommunication Network (Wegen Telecommunicatie Netwerk, WTN) along the Dutch motorway network. The WTN is a network of cables along the motorways. This network was originally installed for voice applications (emergency telephones), but the cables will be increasingly used for data communication in the future. For the management and standardization of the existing data communication networks (e.g. WTN) the Traffic Information and Communication Network (Verkeers Informatie en Communicatie Netwerk, VIC- Net) is being developed. This network is a telematics infrastructure along motorways, which will meet the needs for increasing data communications for dynamic traffic management applications. The standardized VIC-Net offers the possibility of exchanging information between different applications and results in an optimal use of the capacity of existing cable networks along the motorway network. At this moment the form in which this network will be realized is studied. To meet the need for reliable data communication, the Ministry of Transport, Public Works and Water Management has established a communication protocol for the VIC-Net on the basis of the international standard TCP/IP, the protocol recommended by the DRIVE-II project GERDIEN (General European Road Data Information Exchange Network). The software package VIC-Net Open Protocol Package (VOPP) has been developed for the users of VIC-Net to provide easy access to the VIC-Net. VOPP contains the four lower layers of the OSI-model (Open Systems Interconnection). These support data link protocol connections via the WTN, Ethernet, PSTN (Public Telephone Network) and can easily be expanded for application on other network media. VOPP contains, as an extension to the TCP/IP standard, a priority scheduling mechanism to handle critical-time applications such as coordinated ramp-metering. VOPP is being developed for systems which are to be installed along roads and, thanks to its independence of hardware, is suitable for various hardware environments. Communication network A10-West The realization of a communication network on the basis of VOPP network software, has not proven possible, given the short implementation period for the A10-West project (April 1994). To be able to migrate to the VIC-Net at a later stage, a temporary solution has been chosen which fits in with future developments. To set up the migration path, use has been made of the standard reference model for networks (OSI, see Figure 4).

4 Processor (RCP) is the interface between the standard communication network and the local ramp-metering system (see Figure 6). Figure 4. Migration path to VICNET The setup of the communication network for the A10- West project looks conventional, as far as topology is concerned (star network), and is contrary to the principles for the use of WTN. The advantage of this setup is that the migration to the VIC-Net will be easy to implement and will only lead to limited loss of capital. The communication network consists of point-to-point connections between the central computer and the local ramp-metering systems and a PSTN or Ethernet connection between the central computer and the work stations (see Figure 5). Figure 6. System parts The RCP contains standard network software and is expanded with a simple protocol for communication with a local ramp-metering system. The Virtual Terminal Service protocol (VTS) has been taken as an example for this. With the RCP, it becomes possible to link various types of systems to the standard communication network in a simple way. For the migration to the VIC-Net, only modems will have to be exchanged and the network software of the RCP will have to be replaced by VOPP network software. The RCP also offers the possibility to store data from the local systems, which are not directly linked with the CTMS, in logfiles. These logfiles can be read via the public telephone network (PSTN). In the future, the suppliers of traffic systems will integrate the VOPP software with their own control software, in such a way that a separate communication processor will no longer be necessary. The communication network used, is specified in AGV (6) and (7). ASSESSMENT STUDY Figure 5. Communication network The network software is based on TCP/IP and contains the data link protocol SLIP or PPP to maintain the point-to-point connections. The communication between the central computer and the local systems is, in the first place, executed by a separate communication processor. This Road-side Communication The assessment study is split into two parts. One part focuses on the impacts of ramp-metering in general and the other on the impacts of the different rampmetering control strategies. The latter one is part of the EUROCOR project. In the following the different strategies for ramp-metering and the assessment plan are discussed.

5 Strategies for Ramp-Metering Strategies for local ramp-metering. In the Netherlands two strategies for local ramp-metering control have been used and compared: the RWS strategy, developed and used by the Ministry of Transport, and the ALINEA strategy, developed and used in the framework of the DRIVE-I project CHRISTIANE. The RWS strategy calculates the metering cycle based on the real-time flow and speed data upstream and downstream of the on-ramp, while the ALINEA strategy calculates the metering cycle based on the realtime measurement of the density downstream the onramp. Comparison of both strategies led to the conclusion that the ALINEA strategy gives a better performance in terms of average speed and flow, but a higher deviation in speeds, so a less homogeneous traffic flow. The comparison is reported by Smulders and Middelham (8). Strategies for coordinated ramp-metering. There are two forms of coordination that will be implemented. The first form only consists of the centrally commanded on and off switching of the local systems. So for example, if a local system decides to start operation, based on local, actual measurements, the central system can decide that also the local systems upstream of the first one should meter. The second form of coordination also adjusts the local metering rate, based on the measurements for the entire motorway stretch. If the local systems use the ALINEA strategy, the first form is called ALICEN and the second form METALINE. If the local systems use the RWS strategy, the first form is called RWSCEN and the second form RWSCOR. Assessment Plan Situations. The assessment will be done for seven different situations: the current situation (only rampmetering on the S101), local ramp-metering with the ALINEA strategy, local ramp-metering using the RWS strategy, coordinated ramp-metering with ALICEN, coordinated ramp-metering with RWSCEN, coordinated ramp-metering with METALINE, coordinated ramp-metering with RWSCOR. Aspects. The following aspects will be considered: - traffic volumes on motorway and urban network; - speed on the motorway; - capacity of the motorway, derived from volume; - travel times from the beginning of the on-ramps to a point downstream of the Coentunnel; - delay, derived from travel time; - queues on motorway, on-ramps and urban network; - travelled distance on motorway and urban network; - red light obedience; - weather conditions; - incidents and accidents. Data collection. The data are collected automatically as much as possible. Data collection is done by the local metering systems and forwarded to the central system, where access to the data is possible. For the measurement of travel times a large number plate survey will be carried out. On several locations observers note down the number of the registration plate and the passage time of vehicles, for a sample of the passing cars. These data are matched afterwards. Also the measurement of the length of queues is done by observers. The gathering of weather conditions and the occurence of incidents and accidents is necessary for data filtering. The assessment will be carried out during the evening peak hours from 15:00 until 19:00 hours and during at least eight days for every situation. Planning. The implementation of the local systems and the central system will be finished April Then the different strategies have to be tuned, so the assessment (data collection, data analysis en report) will take place from April until November 1994 REFERENCES 1. Hall, M.D., Van Vliet, D. and Willumsen, L.G., 1980, Traf.Eng.+Control, 21, Van Vliet, D., 1982, Traf.Eng.+Control, 23, AGV Consultancy, 1993, "Analysis of the effect of ramp-metering on the A10 West using SATURN", Ministry of Transport, The Netherlands 4. Middelham, F., 1983, Compendium of Technical Papers,29/5-10, ITE, 53rd Annual Meeting, London 5. Taale, H., 1993, "Analysis of the local effects of ramp-metering on two junctions with the motorway A10-West using FLEXSYT", Ministry of Transport, The Netherlands 6. AGV Consultancy, 1993, "Communication network coordinated ramp-metering A10-West", Ministry of Transport, The Netherlands 7. AGV Consultancy, 1993, "Functional design coordinated ramp-metering A10-West", Ministry of Transport, The Netherlands (only in Dutch) 8. Smulders, S.A. and Middelham, F., 1991, "Isolated Ramp-metering: Real life Study in The Netherlands", Deliverable number 7a of the DRIVE-I project CHRISTIANE