Submission number-146 Optimization of signal timings at signalized intersections in Indian Scenario By G.KARTHIK-Assistant professor at GRIET, Hyderabad R.V.RALEGOANKAR-Associate professor at VNIT, Nagpur Abstract: In order to achieve the comfort, safety and to organize the traffic flow in an organized manner, traffic signal timings play a vital role. Optimal signal timing is the most effective and economic method for mitigating vehicular delays in urban areas. In this paper, an in-deterministic way of signal optimization technique is proposed, which is suitable for the Indian road conditions. The road network of the study area was coded in the Microsimulation software called VISSIM with the existing signal timings and field delays were compared with the simulated delays for software validation. Sensitive analysis was carried out by varying the parameters in the software and cycle time was adjusted with the parameter contributing less amount of delay. The proposed model has given good results by reducing delays up to 29%. Keywords: VISSIM, signalized intersection, delays, optimization 1. Introduction: Traffic flows through signalized intersections are filtered by the signal system (stopping of vehicles during red time) causing vehicular delays. Vehicular delays at signalized intersections will increase the total travel time through an urban road network and ultimately results in the reduction of speed, reliability and cost-effectiveness of the transportation system. The uncontrolled and ill planned growth of urban centers has resulted in a number of problems like traffic congestion, shortage of water and electricity, deteriorating environment and public health. The growing cities have generated the high levels of demand for travel by motor vehicles in the cities. The automobile population in India has increased from a mere 0.3 million in 1951 to more than 45 million in 2001. The registered two wheelers constitute nearly seventy percent of the vehicle population in almost all the cities. More than 90 percent of the automobiles are located in urban centers. By considering the above factors traffic signals has to be designed in such a way that the flow is organized with less amount of delays. Various simulation programs and optimization techniques have evolved that aids us in the optimization process. Delay and its derivative are used as the objective function in most optimization software. For example, SYNCHRO optimizes signal timings based on the percentile delay while TRANSYT-7F performs optimization based on factor that involves the average delay called the disutility index. However, as mentioned earlier delay is a stochastic variable and optimizing a signalized intersection for an average value fails the system performance for extreme demand conditions. Thus, a methodology that optimizes the intersection considering stochastic variability would be useful. VISSIM is a microscopic, time step and behavior based simulation model developed to model urban traffic and public transit operations. VISSIM is a program with abilities to display and visualize complex traffic flow in a graphical way. It is able to cope with the analyses of various traffic and transit operations under various conditions and aid the assessment of traffic impacts of physical and operational alternatives in transportation planning Multimodel simulation (Aleksandar Z. Stevanovic and Peter T. Martin, 2008) is the ability to simulate more than one type of traffic. In VISSIM different types of traffic can be simulated like vehicles (cars, buses and tricks), public transit (trams, buses), cycles (bicycles, motor bikes) and pedestrians etc. Output of VISSIM is dependent on the parameters and data input used the network coding. The basic data set needed for a VISSIM network is shown in the table 1. TABLE 1: BASIC DATA REQUIRED FOR A VISSIM NETWORK General Data Network Data Traffic flow Data Signal control Data Geometrical data of the road network Digital images of the plan of entire study area Location of bus stops Input flow in vehicles per hour Vehicle speed at free flow Traffic composition Cycle time Green, Amber and Red time for each signal group
2. Methodology: Stochastic way of optimizing the signals considers delay as an objective function. Optimization of signals involves decreasing the unwanted delays at an intersection. So it is necessary to calculate the field delays manually using the methods suggested by highway capacity manual (HCM). Microsimulation software VISSIM is selected for simulation of road network. The road network is formulated and simulated by specifying the input as field values. Field delay values were compared with the simulated delay values. Software results are considered as valid for the delays obtained in the field and simulation do not differ more than 15 %. Sensitive analysis is to be done after validating the software i.e. by verifying which input parameter has the most influence on the output delay values. Optimal cycle time for a signalized intersection will be selected by considering less amount of delay after running n number of simulations. A brief methodology is proposed for the optimization of signal timings is presented in flow chart figure 1. FIGURE 1: PROCEDURE FOR OPTIMIZATION OF SIGNAL TIMINGS 2.1 Calculation of delays in field: The below figure 2 represents the schematic diagram for the measurement of field delays. Measurements for stopped delay as well as deceleration and acceleration delays were obtained in the field by tracing individual vehicle trajectories between a point at 235 m upstream of the intersection stop line and a point at 120 m downstream of the intersection (Cesar A. Quiroga and Darcy Bullock,1995). The roadway between these two points was referred to in the current study as the system. Tracing vehicles in the system was achieved by measuring the vehicle crossing times at 12 screen lines. Figure 2 depicts the system and the screen lines, where the first station or screen line represents the system entry point and is used as a global reference for all distances and times measured in the system. Station 8 is typically the stop line at the traffic signal, and Station 12 is the system exit point and is selected at the site where all vehicles have achieved a speed closer to the average speed at the system entry point. The part of the system upstream of the intersection was divided into seven sections formed by eight control stations or screen lines while the part of system downstream of the stop line was divided into four sections. Although this process is very laborious and time consuming, it allows for calculating the speed and acceleration rate of vehicles at any point in the system. Measured delays of vehicles that stopped and vehicles that did not stop were analyzed separately and then combined to give delays for the average vehicle.
FIGURE 2: SCREEN LINES USED FOR SPEED DATA COLLECTION 2.2 Sensitive Analysis: Next step involves in the identification of the sensitive parameters. Sensitive analysis involves in identifying the parameters, of which changes may leads to the change of delays and queue length. In general with the increase in the volume at intersection congestion also increases, since the capacity of the intersection may reach its saturation point. So volume should be definitely taken into account as the sensitive parameter (Ragab M. Mousa, 2003). With the increase in the speed of the vehicle, there will be changes in the acceleration as well as deceleration which ultimately results in the fast moving of vehicles at the signal point when the signal turns to green. So speed is considered to be the sensitive parameter for the analysis. In the default simulation of the software there may be the vehicles which are not seen often in the real world scenario. Vehicles of that category can be neglected for the further simulation runs. Change in the driver behavior involves in the following: deceleration of the vehicle after the signal turns to red or green, behavior at amber (go as green or stop as red), maintaining the gap with the front vehicle and overtaking sight distance. So the change in the driver behavior may results in the variation of delays. 3. Case Study: The methodology designed was applied to a case study of Swatantraveer sawarkar chowk, Nagpur. It is a 5 road intersection out of which four are main roads and one is an arterial road. It is situated between two minor signalized intersections. It is one of the busiest intersections in Nagpur. The aerial photograph and simulation screen shot of the site is shown in figures 3 and 4. FIGURE3. AERIAL PHOTOGRAPH OF SWATANTYAVEER SAWARKAR CHOWK, NAGPUR
FIGURE 4. CENTRE LINE VIEW OF VISSIM SIMULATION SCREEN SHOT Geometry of the study area involves five links and each link is of 2 lanes. In addition to the geometrical characteristics one has to do survey for volume characteristics which involves no. of vehicles per hour, driver characteristics which involves average speed of the system and minimum heady, and as our aim is to do signal optimization, signal characteristics are also taken into consideration. Signal characteristics for one phase of the road network of the study area are cycle time 135sec, green time 25sec, red time 105 sec and amber 5sec. 3.1 VISSIM inputs and outputs: For the purpose of simulation of the selected road network one has to input the field data into the software to get the desired output. As mentioned in the previous chapter one has to input geometrical, driver, vehicular, signal characteristics to get the outputs like travel time, delays and queue lengths. A flow chart which represents the process of stochastic optimization in VISSIM has been shown in figure 5. FIGURE 5: PROCEDURE FOR SIMULATION IN VISSIM SOFTWARE 4. Results and Discussion: Simulation had been done with the default parameters of VISSIM and with the input of field data. First objective of the simulation run is to check whether the software considered is valid or not. To validate the software, field delays are compared with the first simulation run i.e. using default parameters of the software and the obtained results holds good if converge up to 85% or more. The following table 2 shows the default parameters of the VISSIM.
TABLE 2 : DEFAULT PARAMETERS OF VISSIM Link data No. of lanes on each link=2 Behavior type=urban(motorized) Lane width= 3.5m Lane change= 200m back Vehicle types Car, HGV, Tram, Bus, Pedestrian, Bike Signal control Amber time= 3sec Car following model Wiedemann 74 Look ahead distance= 250m(max) Driving Behavior Look back distance=150m(max) Max deceleration=-4m/sec 2 Min headway= 0.5m Max longitudinal speed=3.6km/hr Behavior at Amber= Go as green Delay is obtained by marking the travel time sections randomly in a road network and queue lengths can be obtained by queue counters. Below table3 shows the value of delays for simulation using default value TABLE 3: DELAY VALUES FOR DIFFERENT GREEN TIMES WITH FIELD DATA Green time (Sec) Delays for simulation using default values (Sec) 34 149.90 33 144.60 32 140.90 31 130.86 30 128.80 29 122.28 28 120.96 27 116.18 26 112.76 25 108.00 24 100.38 23 93.60 22 89.20 21 81.78 20 76.71 Field delays for the green time of 25 is 114, where as the simulated delays for the green time of 25 is 108 which has a difference of less than 15% which shows that we can proceed with the software for the further analysis. Considering all the sensitive parameters delays and queue lengths are calculated with different green times and tabulated below in table 4.
delays(sec) S.NO TABLE 4: DELAY AND QUEUE LENGTH VALUES FOR DIFFERENT GREEN TIMES WITH SENSITIVE PARAMETERS Green time (sec) Simulation results with change of different sensitive parameters Speed variation Vehicular change Volume change Delay (sec) Queue Length (m) Delay (Sec) Queue Length (m) Delay (sec) Queue Length (m) Driver behavior Change Delay (sec) Queue Length (m) 1 34 134.24 86.00 147.16 81.33 135.9 75.33 106.5 87.33 2 33 124.80 82.33 146.10 86.00 126.37 74.67 92.43 87.33 3 32 122.63 82.67 136.36 87.00 122.03 69.33 94.00 85.67 4 31 116.44 81.67 133.93 85.33 119.90 71.00 90.03 82.33 5 30 110.30 82.33 134.40 83.67 114.00 68.00 85.70 84.33 6 29 108.42 80.00 124.16 83.00 113.17 65.00 85.70 83.67 7 28 102.61 80.33 111.46 81.00 107.73 57.67 80.10 83.67 8 27 92.93 78.67 115.53 81.33 101.73 62.00 74.26 75.67 9 26 88.73 77.00 109.80 79.33 97.73 61.00 75.50 81.66 10 25 86.00 78.00 105.43 79.67 93.27 63.33 76.40 78.33 11 24 83.21 73.00 100.60 77.67 85.73 60.33 71.63 79.33 12 23 78.59 74.00 92.10 77.33 79.76 56.33 65.70 80.00 13 22 75.72 71.33 88.10 78.00 77.23 53.33 75.00 73.33 14 21 69.91 68.67 79.93 72.67 69.83 51.00 72.56 72.33 15 20 65.17 65.67 76.13 81.33 67.53 50.33 67.46 71.67 Variations of the delays with the change in sensitive parameters using different green times are graphically represented in figure 6. delay variations with green time 160 140 120 100 80 60 40 15 20 25 30 35 green time(sec) default speed vehicular volume behaviaral FIGURE 6: VARIATION OF DELAYS FOR DIFFERENT GREEN TIMES WITH SENSITIVE PARAMETERS The above graphs and simulated values shows that change of green times has a significant impact in the delays. But that variation is different in consideration of different parameters. Figure 6 indicates that the delays obtained with the field data is very high compared to that of simulated values. After varying the sensitive parameters it is clear that deduction of heavy good vehicles may not have any effect on the delays. But the volumetric change and the change in speed have a considerable impact, that s why the line passing through the points obtained by the change in volume and vehicle speed is below the vehicular change. Most importantly driver behavioral change seems to have a huge impact on the vehicular delays, which is why the line passing through the points obtained by driver behavioral
changes is far below the other sensitive parameters, which clearly suggests a better traffic flow at road intersection mainly depends upon the behavior of driver. 5. Conclusions: 1.) System delays are reduced to 29% at the same cycle time as that of field by considering driving behavior as a sensitive parameter. It is therefore concluded the proposed methodology for setting timing plans work well in improving performance. 2.) That 29% of reduction in delays will save about 8419 hours per year for all vehicles in system during the peak hours. Delay savings were calculated by taking the difference of average delay per vehicle for the base and field conditions. 3.) From the above results, it is clear that the delay values over a green time of 27, 26 and 25 do not differ much. So the proposed optimal signal timings for studied area are the cycle length of 135, 130 and 125 for the respective green timings. 4.) As discussed earlier, driving behavior has a huge impact on delays that are occurring at intersections. So in addition to the better signal timings, driving behavior at signal points should be maintained better. 5.) Driving behavior involves better headway maintenance, acceleration and deceleration at signal points and behavior at amber signals (whether go as green or stop as red). 6.) Better maintenance of headway between the vehicles may also reduce accidents at the signal point or at left turn lanes 7.) Vehicles will consume less amount of fuel if they don t have to wait at signalized intersections for a longer time. It also helps in reducing annual fuel consumption cost and environmental pollution. 7. References: 1.)Aleksandar Z. Stevanovic and Peter T. Martin. Assessment of the Suitability of Microsimulation as a Tool for the Evaluation of Macroscopically Optimized Traffic Signal Timings, Journal of Transportation Engineering, Vol. 134, No. 2, February 1, 2008. 2.) Tom V. Mathew and Padmakumar Radhakrishnan. Calibration of Microsimulation Models for Nonlane-Based Heterogeneous Traffic at Signalized Intersections, Journal of Urban Planning and Development, Vol. 136, No. 1, March 1, 2010. 3.) Pan Liu, Xu Qu, Hao Yu, Wei Wang and Bing Cao. Development of a VISSIM Simulation Model for U-Turns at Unsignalized Intersections, Journal of Transportation Engineering, Vol. 138, No. 11, November 1, 2012. 4.) Mark C. Smith, Adel W. Sadek, and Shan Huang. Large-Scale Microscopic Simulation: Towards an Increased Resolution of Transportation Models, Journal of Transportation Engineering, Vol. 134, No. 7, July 1, 2008. 5.) Jeffery Archer and William Young. Signal Treatments to Reduce the Likelihood of Heavy Vehicle Crashes at Intersections: Microsimulation Modeling Approach, Journal of Transportation Engineering, Vol. 136, No. 7, July 1, 2010. 6.) Fang Clara Fang and Lily Elefteriadou. Some Guidelines for Selecting Microsimulation Models for Interchange Traffic Operational Analysis. Journal of Transportation Engineering, Vol.131, No. 7, July 1, 2005. 7.) Ilsoo Yun, and Byungkyu (Brian) Park. Stochastic Optimization for Coordinated Actuated Traffic Signal Systems, Journal of Transportation Engineering, Vol. 138, No. 7, July 1, 2012 8.) Fang Clara Fang and Lily Elefteriadou. Development of an Optimization Methodology for Adaptive Traffic Signal Control at Diamond Interchanges, Journal of Transportation Engineering, Vol. 132, No. 8, August 1, 2006 9.) Aleksandar Stevanovic, Jelka Stevanovic and Cameron Kergaye. optimizing signal timings to improve safety of signalized arterials. 3rd International Conference on Road Safety and Simulation, September 14-16-2011, Indianapolis, USA.
10.) Zong Tian and Thomas Urbanik. System Partition Technique to Improve Signal Coordination and Traffic Progression. Journal of Transportation Engineering, Vol. 133, No. 2,February 1, 2007. 11.) Ragab M. Mousa. Simulation Modeling and Variability Assessment of Delays at Traffic Signals. Journal of Transportation Engineering, Vol. 129, No. 2, March 1, 2003. 12.) Satish Chandra, Abhishek Agrawal and Ashalatha Rajamma (2009) Microscopic Analysis of Service Delay at Uncontrolled Intersections in Mixed Traffic Conditions. 13.) Cesar A. Quiroga and Darcy Bullock. measuring control delay at signalized intersections Journal of Transportation Engineering, Vol. 125, No. 4, July/August, 1999. 14.) Kamran Ahmed, Ghassan Abu-Lebdeh and Bashar Al-Omari. Estimation of Delay Induced by Downstream Operations at Signalized Intersections over Extended Control Time, Journal of Transportation Engineering, Vol. 139, No. 1, January 1, 2013 15.) Ragab M. Mousa. Analysis and Modeling of Measured Delays at Isolated Signalized Intersections, Journal of Transportation Engineering, Vol.128, No. 4, July 1, 2002 16.) Aleksandar Stevanovic, Jelka Stevanovic, Kai Zhang, and Stuart Batterman. Optimizing Traffic Control to Reduce Fuel Consumption and Vehicular Emissions. Journal of the Transportation Research Board, No. 2128, Transportation Research Board of the National Academies, Washington, D.C., 2009