DETERMINACIÓN DEL VALOR OPERATIVO DE SISTEMAS DE CONTROL Y MANEJO DE TRÁFICO URBANO EN FUNCIÓN DE LA EFICIENCIA OPERATIVA Y EL NIVEL DE ADAPTACIÓN (CASO DE ESTUDIO BOGOTÁ) DETERMINATION OF THE OPERATIONAL VALUE OF URBAN TRAFFIC CONTROL AND MANAGEMENT SYSTEMS DEPENDING ON THEIR LEVEL OF TRAFFIC ADAPTION (CASE STUDY BOGOTA)
Prof. –Univ. Klaus Banse Presidente Ejecutivo, Fundación Intelligent Transport Systems Colombia (ITS Colombia) Calle 23A No 30A – 22, Bogotá D.C., Colombia, Tel.: +57 (1) 244 4788 k.banse@its-colombia.org Intelligent Transportation Society of America (ITS America), Washington D.C., EE.UU. Transportation Research Board of the Academies (TRB), Washington D.C., EE.UU. Verband Deutscher Elektrotecniker (VDE), Frankfurt, Alemania
RESUMEN La mayoría de los sistemas de control y manejo de tráfico en América Latina han sido implementados en etapas gubernamentales, no técnicas, lo cual ocasiona sistemas difíciles de ampliar o modernizar y muchas veces costosos en su operación y planificación permanente. En el presente artículo se presenta una metodología para ver como se evalúan sistemas de control y manejo de tráfico no solo bajo la óptica de los costos de inversión sino tomando en cuenta costos operativos y costos externos relacionados con la eficiencia del flujo de tráfico. Esos costos externos son determinados mediante el uso de modelos de simulación microscópica de una red parcial de la ciudad. La misma metodología puede ser usada para justificar la modernización de sistemas existentes y para determinar la escala y el horizonte de la modelización. PALABRA CLAVE: Control y manejo de tráfico, control responsivo, control adaptativo ABSTRACT The majority of traffic control and management systems in Latin America have been implemented in political rather than technical phases. As a result many existing systems are difficult to extend or modernize and system operation and planning results unnecessarily expensive to the public administrations. The following article presents a methodology how to evaluate existing traffic management and control systems not only under the investment perspective but also taking into account operating costs and the external costs related with the efficiency in the traffic flow. These external costs are being calculated using microscopic simulation of a part of the network of the city. The same methodology can be applied to justify the modernization of existing system as well as the scale of the modernization and its horizon. KEY WORDS: Traffic management and control, responsive control, adaptive control
In case of short- and mid term benefits the Municipality has to act immediately, in case of long term benefits the planning of modernization should be restudied in 5 years. In case of taking the decision towards modernization a project budget has to be developed and taken into consideration in the final decision taking process. 3. TOOLS AND METHODOLOGY 3.1 Tool choice for microscopic simulation As microscopic simulation tool for this investigation PTV VISSIM® in its current Version 5.00 was chosen due to the following reasons: 1) The versatility in the adjustment of driver behavior model based on the Wiedemann car following models developed in 1974 and 1993 2) The availability of a driver behavior model partly calibrated to Latin American conditions developed by the Universidad Nacional de Colombia, Medellín, Colombia. 3) The availability of the lane change model developed by PTV AG. 4) The ease of configuring and editing traffic dependency logics through the included VisVAP® module as well as preconfigured SCATS® and SCOOT® logics. 5) The possibility to simulate public transport routes according to the local operation scheme and boarding and unboarding behavior of passengers. 6) The ease of report creation with the VISSIM Analyzer module.
4.2 Simulation area, input and conditions For the simulation a network of 24 intersections from one of the most congested areas of Bogotá was chosen. The area was limited by 7th and 10th Street West and 72nd, 82nd and 127th Street North (Image 3), (Table 2).
Image 3 - Simulation network area in the north of Bogotá
• Mean velocity [km/h]: Median velocity of all vehicles that traveled in the simulated network during the simulation period. • Total delay time [h]: Total delay time accumulated by all vehicles that traveled in the simulated network during the simulation period. • Mean delay time per vehicle [s]: Mean delay time per vehicle from all vehicles that traveled in the simulated network during the simulation period. • Total stop by stopping [h]: Time during which vehicles were stopped accumulated by all vehicles that traveled in the simulated network during the simulation period. • Mean stop time per vehicle [s]: Mean stop time per vehicle from all vehicles that traveled in the simulated network during the simulation period. • Total number of stops [n]: Number of stops accumulated by all vehicles that traveled in the simulated network during the simulation period. • Mean stops per vehicle [n]: Number of stops per vehicle from all vehicles that traveled in the simulated network during the simulation period. • Fuel consumption [gal]: Fuel consumption generated by all vehicles that traveled in the simulated network during the simulation period. • CO2 emissions [g]: Theoretical CO2 emissions generated by all vehicles that traveled in the simulated network during the simulation period. The following intersection measurements were taken and generated into ASCII files for processing and detailed analysis:
!
Table 2 - List of simulated intersections in the network
3.2 Simulation methodology The methodology is based on a simplified three step approach: 1) Simulation and calibration of a base line scenario 2) Creation and simulation of traffic adaptive scenario 3) Evaluation and analysis of simulation output information 4. CASE STUDY BOGOTA The Colombian capital Bogotá with its 7 million inhabitants (2007) is located in the centre of the country (Image 1). The city has close to 1.200 signalized intersections controlled from 3 traffic control systems located in the north, centre and south of the city (Image 2). Only very few intersections have traffic detectors and even less work on traffic dependent logics. The vast majority of intersections are fix time controlled with signal program selection depending on the time of day by the central systems, which means that there is no life interaction between traffic and the signal programs whatsoever.
Image 1 - Map of Colombia
1. PARTS OF A TRAFFIC MANAGEMENT AND CONTROL SYSTEMS First of all it has to be clear how to separate the different parts of the traffic management and control systems. Therefore the architecture has to be split into three main subsystems (Figure 1): • Central control • Communications • Peripherals
Figure 1 – Main subsystems of traffic management and control systems
Internal number 1009 1808 1438 1259 1258 1257 1256 1255 1230 1165 1157 1008 1007 1770 1420 1279 1250 1118 1006 1005 1004 1003 1156 1107
Intersection Avenida Carrera 7 x Calle 72 Avenida Circumvalar x Calle 83 Avenida Carrera 7 x Calle 112 Avenida Carrera 7 x Calle 106 Avenida Carrera 7 x Calle 109 Avenida Carrera 7 x Avenida Calle 116 Avenida Carrera 7 x Calle 119 Avenida Carrera 7 x Avenida Calle 127 Avenida Carrera 7 x Calle 121 Avenida Carrera 7 x Avenida Calle 92 Avenida Carrera 7 x Calle 94 Avenida Carrera 7 x Calle 75 & Calle 76 Avenida Carrera 7 x Calle77 Avenida Carrera 11 x Calle 74 Avenida Carrera 11 x Calle 73 Avenida Carrera 11 x Calle 76 Y CL 77 Avenida Carrera 11 x Calle 78 Avenida Carrera 11 x Calle 81 & Calle 82 Avenida Carrera 7 x CL 79 B Avenida Carrera 7 x Avenida Calle 82 Avenida Carrera 7 x Avenida Circunvalar & Calle 84 Avenida Carrera 7 x Avenida Calle 85 Avenida Carrera 11 x Avenida Calle 72 Carrera 9 x Avenida Calle & Calle73
• • • • •
Level of Service [A-F]. Median queue length [m]. Number of stops at the intersection approaches [n]. Fuel consumption at the intersection approaches [gal]. CO2 emissions for at the intersection approaches [g].
4.5 Yearly savings on external costs The network performance enhancements through implementation of adaptive traffic control technology are shown in Table 5. The diminution of travel time per intersection is shown in Figure 7.
Table 5 - Overall comparative network analysis
Indicator Number of vehicles [n] Total distance traveled [km] Total travel time [h] Mean velocity [km/h] Total delay time [h] Mean delay time per vehicle [s] Total stop time [h] Mean stop time per vehicle [s] Number of stops [n] Mean stops per vehicle [n] Fuel consumption [gal] CO2 emissions [g] Mean queue length []m] Base line (fix time) 29.788 48.391 2.519 19 1.528 185 774 94 117.627 4 2.843 198.746 57 Projected (adaptive) 30.897 51.502 2.173 24 1.117 130 520 61 92.160 3 2.716 189.867 42
MAIN SUBSYSTEMS
Central control Communications Peripherals
The base line simulation and calibration was done with the most recent traffic data and signal timing available. A relative calibration error of 40% (Table 3) in some intersection accesses had to be accepted, as the traffic data was from so called typical traffic weeks from different periods and years. Figure 6 shows a typical example for the inconsistency in the traffic data provided and used for traffic planning purposes till date.
Table 3 - Example for velocity calibration differences
Figure 7 - Travel time enhancements per intersection
TOTAL DELAY TIME PER INTERSECTION [h]
Source: Banse K. (2007) Traffic Management and Control Systems , Cartagena, Colombia
1.1 Central control The central control subsystem includes all hardware and software technically related with the operation and maintenance of the system. The related hard- and software can be installed in the traffic control centre itself or in the planning and maintenance department of the municipality. 1.2 Communications The communication subsystem includes all hardware and software installed or leased for the purpose of communication between the central control and peripherals. This subsystem includes also all civil works used for communications wirings as well as wireless infrastructure if owned by the municipality and used for the communication of the traffic management and control system. If the infrastructure is shared with other urban system the percentage has to be established. 1.3 Peripherals The peripherals subsystem includes all hardware and software installed in intersection control as well as other applications as monitoring, variable message signs, bus priority systems, etc. This subsystem includes also all civil works on intersection and final application level. 2. DESCRIPTION OF COSTS The costs have to be divided in three main areas due to their character (Figure 2): • Investment • Operational costs • External costs All calculations have to be based on the same currency.
Figure 2 – Cost classification areas
Central control – Center (Paolquemao) Central control – South (Muzu)
Intersection (acces) Carrera 7 Calle 116 (N) Carrera 7 Calle 116 (S) Carrera 7 Calle 106 (N) Carrera 7 Calle 106(S) Carrera 7 Calle 92 (N) Carrera 7 Calle 92 (S) Carrera 7 Calle 72 (N) Calle 72 x Carrera 11 (N) Calle 76 x Carrera 11 (N) Calle 82 x Carrera 11 (N) Carrera 7 Calle 84 (S)
Median velocity simluated [km/h] 16,8 31,7 15 30 23,1 25,1 22,7 25,8 22 24,8 24,5
Median velocity calibrated [km/h] 10 32,7 10 32,7 22,6 15,5 22,6 22,6 27 27 15,5
Difference [%] 40 3 33 9 2 38 0 12 23 9 37
Image 2 - Aerial image of the city of Bogotá with location of the 3 traffic control centers
Fix time
Adaptive
Figure 6 - Example for inconsistency in the provided traffic count data
Volume difference !v = 255 eq.veh.
Calle 106 Calle 109
Central control - North (Chicó)
Quantifying travel time and fuel usage, applying the nationally approved values (2007) and taking into account that the peak hour in Bogotá represents only about 10% of the daily traffic volume, the cost of travelling the network under fix time conditions is calculated in 44.843.238 COP/day, while with the use of adaptive traffic control the total cost decreases to 38.683.743 COP/day representing savings of about 6´150.000 COP/day. The simulated network represents only about 2,5% of today’s signalized intersections in Bogotá The projected savings for the whole city would be 246.379.800 COP/day or 87.711.208.800 COP/year, which using the current exchange rate (05/2008) are about 50´000.000 USD/year. 5. CONCLUSIONS The main conclusions are that: 1) Traffic adaptive control increases traffic flow considerably compared to fix time control decreasing travel time, fuel usage, environmental impacts and the related operation costs. 2) Microscopic simulation is the only reliable way to determine the real savings potential in function of the traffic signal timing mechanism. 3) Bad or incongruent traffic data is being used for the majority of planning process. This has to be improved by the massive deployment of automated full time traffic measurement systems. For further questions please feel free co contact ITS Colombia.
Avenida Carrera 7
Volume input vi = 4.112 eq.veh.
Volume output vo 3.859 eq.veh.
The public transport routes were configured according to the information provided by the authorities and areas of informal boarding and unboarding processes were configured with equivalent speed reductions. 4.3 Traffic dependency logics 4.1 Description of fix time, responsive and adaptive systems Fix time control (Figure 3) switches between different pre established signal programs that have been developed based on one week traffic counts in what is considered a typical traffic week. The switching decision is based depending on day and time. There is no life interaction between the traffic and the signal timing.
Figure 3 - Description of fix time traffic control
COST CLASSIFICATION
Investment Operational costs External costs
In order to avoid the creation of any manufacturer links in the traffic adaptive simulation it was decided to simulate a two stage adaption based only on phase selection and signal timing. The tactical detectors were located at the stop line of each access and lane (Image 4). The two stage logics were developed with the graphical editor VisVAP (Image 5) and later converted into traffic controller language (Table 4).
Image 4 – Location of tactical detectors for traffic adaption on Calle 127 x Carrera 7
FIX TIME SIGNAL CONTROL
Central System Operation Fuel usage
6. SPECIAL THANKS Special thanks to:
No life interaction Traffic situation Fix time control Active signal state
Communications
Maintenance
Enfironment
Empresa de Telecomunicaciones de Bogotá; for kindly allowing to use information from the study: Apoyo en la determinación de la estrategia para la modernización del sistema de control y manejo de tráfico actualmente en operación en la ciudad de Bogotá, D.C. Sistemas Inteligentes de Transporte Limitada (SIT Ltda); for actively supporting this ongoing investigation through sponsoring and access to their hardware and software infrastructure. 7. SOURCES
Peripherals
Planning
Time
Signal Program 1 Signal Program 2
Data acquisition
Signal Program 3
Source: Banse K. (2007) Traffic Management and Control Systems , Cartagena, Colombia
Monday
Signal Program N
23
2.1 Investment Investment includes all hardware and software acquired and in use. Their analysis value has to be calculated based on local or national legal parameters for accounting and capital depreciation. For the analysis three values have to be taken into account. • Year 0 • Year 5 • Year 10 (value to date) (median innovation cycle for traffic management technology) (median usual life-cycle of technology)
Source: Banse K. (2007) Traffic Management and Control Systems , Cartagena, Colombia
!
Image 5 – Edition of traffic adaptive logics in VisVAP®
Banse K. (2006) Introducción a la simulación microsópica multimodal, Training material, Universidad de Cartagena, Cartagena, Colombia Banse K. (2007) Sistemas de control y manejo de tráfico, Lecture, Universidad de Cartagena, Cartagena, Colombia Sistemas Inteligentes de Transporte Ltda (2007) Apoyo en la determinación de la estrategia para la modernización del sistema de control y manejo de tráfico actualmente en operación en la ciudad de Bogotá, D.C., Study, Empresa de Telecomunicaciones de Bogotá D.C., Bogotá, Colombia Planung Transport Verkehr AG (2008) Vissim 5 Handbook, User manual, Planung Transport Verkehr AG, Karlsruhe, Germany.
Responsive control (Figure 4) also switches between different pre established signal programs that have been developed based on one week traffic counts in what is considered a typical traffic week. The switching decision is based on traffic data acquired by tactical sensors on intersection level in case of local responsive control or strategic traffic sensors in case of central responsive control. As the number of signal programs is limited, the selected program only suites the traffic situation in a certain percentage. The reaction to changes in the traffic pattern takes usually at least one cycle time.
Figure 4 - Description of responsive traffic control
!
The investment in place can easily be determined either based on the acquisition value out of contracts or through a field visit and evaluation by neutral experts. 2.2 Operational costs Operational costs include all cyclic costs such as staff, rent, lease, regularly outsourced activities for traffic counts, planning, maintenance, etc. The operational costs for the base line scenario can also easily be calculated by gathering the information on all current contracts and other kind of obligations. The total amount should be calculated to a yearly cost. For the traffic adaptive scenario the planning costs and costs for traffic counts can be diminished. 2.3 External costs External costs include all costs related with the traffic flow efficiency such as fuel usage and emissions, time consumptions by travelers, etc. These costs for the base line scenario as well as for the traffic adaptive scenario have to be determined through microscopic simulation of a partial network as shown later in the case study Bogotá in Chapter 4. 2.4 Simplified cost analysis The resulting costs have to be projected into Table 5 identifying the quantified benefits on short term (0 years), mid term (5 years) and long term (10 years).
Table 1 - Calculation of overall benefits for decision taking
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
RESPONSIVE SIGNAL CONTROL
$SIGNALGRUPPEN CROSSIG K1 K2 K3 K6 k91 $ PÜ-Name PÜ-Kommentar PÜ-Kommentar PÜ_Nummer Übergangsdauer [s] Von Phase Nach Phase $ K2 K5 $PÜ PÜ-Name PÜ-Kommentar PÜ-Kommentar PÜ_Nummer Übergangsdauer [s] Von Phase Nach Phase $ K1 K6 $PÜ PÜ-Name PÜ-Kommentar PÜ-Kommentar PÜ_Nummer Übergangsdauer [s] Von Phase Nach Phase $ K1 K2 K4 : : : : : : : --- PÜ von 1 nach 2 (10 s) --1 10 1 2 -127 3 : : : : : : : 0 127
Wiedemann R. (1974) Microscopic Traffic Simulation, Universität Karlsruhe, Germany Wiedemann R., Reiter U. (1993) Microscopic Traffic Simulation, Universität Karlsruhe, Germany 8. ABBREVIATIONS ASCII: American Standard Code for Information Interchange; ASCII is a character encoding based on the English alphabet used for digital computer communication on basic level. SCATS®: Sydney Coordinated Adaptive Traffic System; Adaptive traffic control and management system, developed by the Roads and Traffic Authority (RTA), New South Wales, Australia. SCOOT®: Split Cycle Offset Optimization Technique; Adaptive traffic control and management system, developed by TRL Limited, London, England. VisVAP®: Vissim Verkehrsabhängige Programmierung; Traffic adaptive language vor PTV VISSIM, developed by Planung Transport Verkehr AG, Karlsruhe, Germany. 9. ITS COLOMBIA ITS Colombia is a non profit organization for the promotion of intelligent transport systems as means of improving traffic safety, efficiency of the mobility systems and and quality of mobility in all modes. ITS Colombia membership is open for the private and public sector. ITS Colombia is co-organizing the yearly international technology event ANDINATRAFFIC.
Table 4 - Example of traffic language script developed out of VisVAP
VISSIM K1 K2 K3 K6 k91 K1 2 3 4 5 VISSIM Phase_1 Phase_2 Phase_3 K1 K5 K1 K2 K2 K1 K4 K1 K2 K6 K5 K6 k6 K5 K2 K4 k4 K4 K 1
Traffic situation
Responsive control
Active signal state
Situation 1 Situation 2 Situation 3
Signal Program 1 Signal Program 2 Signal Program 3
Situation N
Signal Program N
K2 K5 K6 k4 $PHASEN CROSSIG 1 2 3 $ Phase_1 rot Phase_2 rot Phase_3 rot Phase_4 rot 5 $STARTPHASE $ Phase_1 $PÜ
--- PÜ von 1 nach 3 (14 s) --2 14 1 3 -127 7 0 127
k6
: : : : : : :
--- PÜ von 1 nach 4 (11 s) --3 11 1 4 -127 -127 4 2 0 127
Source: Banse K. (2007) Traffic Management and Control Systems , Cartagena, Colombia
4.4 Evaluation and analysis Measurement points for volume, velocity and queue length were located on every access of all signalized intersection. Preliminary observation of results could be done life during the simulation process (Image 6).
Image 6 - Life display of traffic dependent signal program development in PTV VISSIM®
Adaptive control (Figure 5) bases its changes in green time distribution, cycle time, coordination and zone assignment based on traffic data acquired by tactical detectors on intersection level in case of a local adaptive control or based on sensors from multiple intersections in case of a central adaptive control. There are no pre established signal programs necessary for adaptive control. The signal timing is calculated every second according to the traffic situation and reaction to changes in the traffic pattern is nearly immediate.
Figure 5 - Description of adaptive traffic control
ADAPTIVE SIGNAL CONTROL
!
Traffic situation
Adaptive control
Active signal state
The following network measurements were taken and generated into ASCII files for processing and detailed analysis: • Number of vehicles [n]: Total number of vehicles that traveled in the simulated network during the simulation period. • Distance traveled [km]: Total distance travelled by all vehicles in the simulated network during the simulation period from their origin to their destination. • Travel time [h]: The total time traveled by all vehicles in the simulated network during the simulation period from their origin to their destination. Information: www.its-colombia.org www.andinatraffic.com
Traffic data Flow Count Occupation …
Adaptive traffic logics & strategies Distribution Cycle time Coordination Zones …
&'" >3&*+(30$1*23"(?($;&$1"*+(,3%&%(
Source: Banse K. (2007) Traffic Management and Control Systems , Cartagena, Colombia