This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore. Title Modelling and simulation of vehicle movements at signalised road junctions Author(s) Liu, Jialin Citation Liu, J. (2014). Modelling and simulation of vehicle movements at signalised road junctions. Student research paper, Nanyang Technological University. Date 2014 URL http://hdl.handle.net/10220/26020 Rights 2014 The Author(s).
Modelling and Simulation of Vehicle Movements at Signalised Road Junctions Liu Jialin School of Civil and Environmental Engineering Abstract - Signalised road junctions is widely used in urban area to increase safety level and minimise conflicts between vehicles. Modelling and simulating the vehicle movements is able to help understanding the vehicle behaviour at the road junction for optimising the junction layout and signal set up. This paper presents a study about vehicle movements at different types of signalised road junctions. The results about the safety level of the road and vehicle travel time are tested when signal control strategies are changed. Field observation is conducted to record realistic traffic movements at junctions and the traffic volume and signal timing data are extracted. With the help of computer simulation, junctions are modelled for capacity and safety assessment by changing signal timing type from permissive to protected right-turn mode. Keywords - Signalised Road Junction; Permissive Right-Turn; Right-Turn; Travel Time; Traffic Conflicts; Safety Assessment. (key words) 1 INTRODUCTION Transportation system is increasingly important nowadays, especially in cities with high population density like Singapore. The residents rely on transportation system a lot in daily life. According to Singapore Land Transport Statistics in Brief [1], the average daily traffic volume entering the city in 2013 is 292,000. The average annual kilometres travelled per car in 2012 is 18,200. In this huge urban transportation system, the traffic situation at signalised road junctions plays an important role. There are 2,185 traffic lights in Singapore in 2013 [1]. As a result, it is necessary to understand vehicle behaviour at road junctions to optimise the transportation system. This study aims to model and simulate movements of vehicles at signalised road junctions (see Figures 1, 2 and 3) in residential area in Singapore, and compare the result with a similar junction in Tangshan, China. Firstly, field work was conducted at two signalised junctions in Singapore. Recordings of traffic volume and signal timing for each junction were collected. Then, the data for traffic volume and signal timing was extracted from the recordings. After that, models of junctions were built using software and the extracted data was put in as parameters. Next, the type of the signal cycle was changed from permissive to protected right-turn mode to get traffic time and conflicts in each Assoc Prof Wong Yiik Diew Ms Chai Chen, Dora School of Civil and Environmental Engineering situation. Finally, the results for three junctions in different situations were compared. Figure 1 in Singapore 2 LITERATURE REVIEW A signalised road junction is a location that roads intersect with each other and vehicles pass through which is controlled by traffic lights. It mainly aims to increase level of safety and convenience, and minimise delay of vehicles [2]. 2.1 JUNCTION TYPE There are right-hand traffic and left-hand traffic which are regulations for vehicles and passengers to follow. In Singapore, left-hand traffic is used, but China uses right-hand traffic which is an opposite one. At each junction, vehicle movements including straight through, left turn, right turn, and U-turn shall occur in timeseparated signal phases. Most of the vehicles are separated by direction of movements in different lanes, while some may also share a lane although directed to different directions. There are 3-way, 4-way, 5-way and 6-way junctions that involve 3 to 6 approaches [3]. 3- way and 4-way junctions are the focus in this study. 2.2 VEHICLE TYPE Vehicles are motor machines that transport passengers or goods. 7 types of vehicles are defined in this study: passenger cars, motorcycles, small buses (up to 30 seats), buses (more than 30 seats), light goods vehicles with laden weights up to 3 tonnes, heavy goods vehicles with 2 axles, and heavy goods vehicles with 3 or more axles [4]. Passenger car equivalent (PCE) is a metric to evaluate traffic flow rate in transportation area [5]. Every type of vehicles has different abilities of carrying passengers and goods, so they have different values of 1
PCE which has an important effect on mode of transport [6]. 2.3 SIGNAL ARRANGEMENT In order to minimise conflicts and vehicles can go through easily, traffic signals are set to follow a programmed cycle. A typical signal cycle for one approach consists of straight-through green, amber, all red, green arrow, and flashing green arrow for each approach [7]. The simplest signal phase is two-phase operation. It is commonly used in junction with few right-turn movements or there are large gaps between straightthrough vehicles. One way to facilitate right-turn movements is to add a right-turn phase such as when number of right-turn vehicles is large or gaps between straight-through vehicles are small such that serious traffic delay arises while waiting to make a right turn. 2.4 TRAFFIC SAFETY Surrogate Safety Assessment Model (SSAM) is a useful technique for analysing collisions between vehicles in traffic streams. The two parts of it, simulation and conflict analyses, allow it to assess traffic safety in the real world. Conflicts occur when two vehicles do not evade each other such that collision will likely happen. At signalised road junction, conflicts between vehicles exist not only in rear-end direction but also in crossflow and lane change movements, which makes junction more complex than normal roads. SSAM identifies interaction between vehicles and analyses different kinds of conflicts. After that, SSAM is able to calculate safety indicators including time-to-collision (TTC) and post-encroachment time (PET) [9]. Time- To-Collision (TTC) measures the severity of conflicts. Hayward (1972) defined it as the time remaining until two vehicles to collide if they keep at the same speeds and on the same paths [10]. Post-Encroachment Time (PET) is the time difference between two vehicles passing the same zone. It is an indicator to measure the propensity of crashes [11]. 3 DATA COLLECTION & ANALYSES This part aims to collect the data from real signalised junctions for further usage of simulation. The choice of junctions should be typical and able to be modelled. The traffic volume (classified in vehicle type) and signal timing for each junction is required. Figure 2 in Singapore One type of right-turn movement is permissive rightturn mode, without an exclusive (protected) right-turn phase. Under permissive-only arrangement, vehicles turn right under circular green indication by relying on large gaps between straight-through vehicles. An alternative arrangement is the Red-Amber-Green (RAG) Arrow, in which vehicles turn right during the provided right-turn phase, but not permitted during circular green indication. The advantage of right-turn movements under RAG arrangement is that the conflicts between straightthrough vehicles and right-turn vehicles are eliminated. However, the signal cycle length under RAG arrangement can be longer than one for permissive right turn without a protected right-turn phase. There is another type of signal arrangement called permissive/protected right-turn phase. It is a type with permissive right turn under circular green followed by protected right turn with green arrow. Permissive/ protected right-turn arrangement is safer than permissive-only right-turn arrangement but can be more risky than RAG arrangement in general case [8]. Figure 3 Junction 3 in Tangshan, China To fulfil this inquiry, the traffic volumes are recorded from the top level of HDB (Housing and Development Board) buildings. The signal timings are recorded near the traffic lights. For the observation time, week days evening peak hours (Monday to Friday, approximately 17:00-19:00) were chosen, because the maximum traffic volume often occurs during that time, and it may induce more traffic delay (see Tables 1, 2, and 3). From the recording, both the traffic volume and signal timing for each junction can be extracted. The traffic volumes are cumulated every 5 minutes. The number of vehicles that go through the junction is recorded separately by types of vehicles and different lanes of roads and by directions. 2
Table 1 Site Description for Site 1 (see Figure 1) Location Date Time Camera s Location Intersection of Woodlands Ave 9 and Woodlands Dr 81 29-Aug-13 17:15-17:45 Blk 809 02-Sep-13 16:45-17:15 Blk 809 16-Sep-13 16:45-17:00 Woodlands Dr 81 24-Sep-13 17:00-17:45 Woodlands Dr 81 Table 2 Site Description for Site 2 (see Figure 2) Location Date Time Camera s Location Intersection of Woodlands Ave 7 and Woodlands St 83 05-Sep-13 16:50-17:50 Blk 852 24-Oct-13 24-Oct-13 16:45-17:25, 17:27-17:47 16:45-17:17, 17:35-18:03 Woodlands Ave 7 Woodlands St 83 Table 3 Site Description for Site 3 (see Figure 3) Location Date Time View Intersection of Guangming Road and Xingyuan Road 29-Sep-13 16:45-17:45 Top of building under construction The signal timing is determined by using Avidemux 2.5. Avidemux can be used to edit video for cutting and encoding [12]. The frame that the traffic light changes was extracted. Using these frames, one can calculate the durations of each signal phase (see Table 4). 4 SIMULATION EXPERIMENT To simulate the junction, PTV Vissim was used. PTV Vissim is a software package that assists to simulate and control traffic. It is developed by Planung Transport Verkehr AG in Germany. Road and junction can be designed using Vissim [13]. By inputting the parameters, a real junction is modelled (see Figures 4 and 5). Figure 4 Simulation Experiment for In this study, three sites are modelled using Vissim. The previously extracted data were put in to complete the modelling of real junctions. The modelled junction is run to get results about the travel time of each vehicle, number and type of conflicts that occur at the junction, and time-to-collision and post encroachment time for two neighbouring vehicles. Table 4 Signal Timing Arrangement (seconds) Green Red Amber Flashing Green Approach 1 72 51 3 0 Approach 1 Right 85 38 0 3 Approach 2 24 99 3 0 Approach 2 Right 31 92 0 3 Approach 1 72 51 3 0 Approach 1 Right 10 113 0 3 Approach 2 24 99 3 0 Approach 2 Right 4 119 0 3 Approach 1 62 61 3 0 Approach 1 Right 20 103 0 3 Approach 2 19 104 3 0 Approach 2 Right 10 113 0 3 Approach 1 50 37 3 0 Approach 2 49 38 3 0 Approach 2 Right 52 36 0 2 Approach 3 28 59 3 0 Approach 1 33 54 3 0 Approach 2 33 54 3 0 Approach 2 Right 10 54 3 0 Approach 3 Right 33 54 3 0 Approach 3 Left 10 54 3 0 Junction 3 (in Tangshan) Approach 1 40 38 2 2 Approach 2 30 48 2 2 Junction 3 (in Tangshan) Approach 1 32 46 2 2 Approach 1 Left 6 74 2 2 Approach 2 22 56 2 2 Approach 2 Left 6 74 2 2 Next, the signal phase of the junction is changed from permissive to protected one. Keeping other data unchanged, the junction model is run again to collect new results. However, for sites 2 and 3, serious traffic jam appears when running the revised model. As a result, the signal phases are further modified to minimise the traffic jam. 3
is 59.42 seconds, but it is still longer than existing permissive/protected junction. For, which is a T-Junction, the travel time of permissive junction is 40.68 seconds per vehicle (see Table 6). After modification to protected junction, it changes to 53.82 seconds per vehicle. Table 6 Travel Time Figure 5 Simulation Experiment for Finally, the signal phase of junction 1 is also modified to get better result. The model is run again to get the new data for protected modified. 5 RESULTS AND DISCUSSION 5.1 TRAVEL TIME From the simulation experiment, several results can be obtained. From the travel time point of view, the simulation experiments of two junctions in Singapore that have been changed to protected right turn take longer time to travel the same route. Lane Table 5 Travel Time Travel Time Vehicle No. Travel Time Vehicle No. Travel Time Vehicle No. No.1 45.3 979 66.5 973 52.2 974 No.2 30.8 197 194 152 76.4 193 No.3 23.8 41 30.4 40 28.4 41 No.4 45.2 1060 45.2 1057 51.2 1060 No.5 30.1 20 77.4 20 72.5 18 No.6 38 119 39.2 120 46 119 No.7 71.2 2 51.1 2 64.8 2 No.8 55.7 35 92.9 41 67.9 40 No.9 68.3 99 65.5 93 73.2 96 No.10 62.6 33 67.6 34 78.5 30 No.11 59.8 175 130.2 162 78.9 172 No.12 22.5 42 22.3 48 23 74 Average 46.11 73.53 59.42 For, the result of simulation of the real junction is an average of 46.11 seconds of travel time per vehicle (see Table 5). If the junction is changed to a protected one (that is, junction under RAG arrangement and without changing the signal timings), the travel time comes to 73.53 seconds per vehicle. After refinement in the signal timings ( modified protected junction), travel time becomes shorter which Lane Travel Time Vehicle No. Travel Time Vehicle No. No.1 41.2 551 51.1 545 No.2 38.4 261 50.3 271 No.3 39.8 602 49.1 598 No.4 34.9 37 68.1 37 No.5 47.8 143 46.3 143 No.6 42 13 58 13 Average 40.68 53.82 For Junction 3 (site in Tangshan, China), the modified junction performs better overall. The permissive junction travel time is 423.89 seconds per vehicle, while the protected junction travel time is 220.81 seconds per vehicle (see Table 7). However upon examining the data in more detail, the average travel time of all routes in permissive junction except approach 2 left-turn is 59.88 seconds. Only the approach 2 left-turn vehicles take really long time to travel which is 4536 seconds. After changing to protected junction, the travel time of vehicles in all routes except approach 2 left-turn vehicles need longer time. As a result, generally speaking, junction 3 s modified result still involves longer travelling time. Table 7 Junction 3 Travel Time Junction 3 (intangshan) Lane Travel Time Vehicle No. Travel Time Vehicle No. No.1 71.7 834 401.1 470 No.2 65.6 312 514.3 168 No.3 48 268 285.8 147 No.4 59.4 517 214.1 474 No.5 67.6 271 465.8 227 No.6 4536 124 117.5 117 No.7 68.7 547 112.3 532 No.8 61.2 188 193.5 168 No.9 46 114 56.6 113 No.10 60.1 523 70.1 530 No.11 64.9 205 173 181 No.12 45.5 188 45.6 185 Average 432.89 220.81 5.2 TRAFFIC SAFETY Junction 3 (in Tangshan) From the safety point of view, three different kinds of results, time-to-collision (TTC), post encroachment time and number of conflicts, are used to present the 4
safety situation (see Table 8). The average TTC of vehicles in is getting longer with the modification. The average TTC of permissive Junction 1 is 0.2454 seconds. Changing to protected junction, it becomes 0.2935 seconds. After modification, the TTC becomes 0.3673 seconds for vehicles on average, which is getting better gradually. For, the permissive junction s TTC is 0.5773 seconds, while the modified junction s TTC is 0.3763 seconds, which is getting worse. For the third junction, the TTC of permissive junction is 0.0434 seconds which is really short. Changing to protected junction, the TTC becomes 0.0606 which is a little bit better. Generally, from TTC result, modifications of both the first and the third junctions help reduce danger, but not for the second junction. Table 8 SSAM Results the protected junction eliminates nearly all the highseverity crossing conflicts (from 187 counts to 6 counts). Although the numbers of rear-end conflicts and lane change conflicts increase (from 45 counts to 230 counts for rear-end conflicts, and from 21 counts to 106 counts for lane change conflicts), the modification eliminates a large number of them induced by the protected junction (from 230 counts to 149 counts for rear-end conflicts, and from 106 counts to 53 counts for lane change conflicts). The modification of is effective in reducing conflicts. Total Crossing RearEnd Lane Change AveTTC AvePET 253 187 45 21 0.2454 0.494 341 5 230 106 0.2935 0.6563 208 6 149 53 0.3673 0.725 75 32 31 12 0.5773 1.0827 152 0 48 104 0.3763 0.7289 Junction 3 3807 3695 104 8 0.0434 0.092 Junction 3 6937 6185 479 273 0.0606 0.1413 From PET perspective, the modified also becomes better. The PETs of permissive, protected, and protected modified are 0.494s, 0.656s, and 0.725s, respectively. For, the permissive one has 1.083s PET, while vehicles in the modified Junction 2 have PET of 0.729s which is shorter. The PET of vehicles in permissive Junction 3 is 0.092 seconds. After modification, the PET increases to 0.141 seconds. From the PET point of view, the result is the same as one based on the TTC, that is the modification of Junctions 1 and 3 decreases the level of danger, but the modification of increases the level of danger. From the number of conflicts point of view, the situation is more complex (see Figure 5). For Junction 1, the total number of conflicts of the protected junction (341 counts) is larger than the total number of conflicts of the permissive junction (253 counts). After modification, the protected junction (208 counts) induces fewer conflicts than the permissive case. Also, Figure 6 Conflict Map for For, the result is in a negative way. The total number of conflicts increases from 75 counts to 152 counts due to modification. Although the crossing conflicts are all eliminated, the other two types of conflicts increase by a large amount (rear-end conflict: from 31 counts to 48 counts; lane change conflict: from 12 counts to 104 counts). The modification for Junction 2 does not help to reduce conflicts efficiently. For Junction 3, the total number of conflicts is doubled, from 3807 counts to 6937 counts, mainly due to the doubling of crossing conflicts, from 3695 counts to 6185 counts. The number of rear-end conflicts also increases from 104 counts to 479 counts. For lane change conflicts, it also changes from 8 counts to 273 counts. The modification for Junction 3 in Tangshan, China, increases the number of conflicts. 6 CONCLUSION The junction in Tangshan would experience increases in the number and severity of traffic conflicts. According to on-site data and results from simulation experiment, several conclusions are made. The protected junction causes longer travel time for vehicles. Although from the literature review the protected junction should be easier for vehicles to pass through, from the experiments one can see that the severity of traffic conflicts is reduced, but number of conflicts are dependent. For which is an X-junction in Singapore, number of conflicts is reduced. For Junction 5
2 (T-Junction in Singapore) and Junction 3 (X-Junction in Tangshan which has larger traffic volume than Singapore), the number of conflicts increases. The site situation affects whether a protected junction is better or not. For junction with smaller traffic volume, a protected junction can be used for lower severity of traffic conflicts and fewer conflicts. However for junctions with larger traffic volume, a protected junction may cause more conflicts and more serious traffic jam. ACKNOWLEDGMENTS The author would like to express her heartfelt gratitude to her project supervisor, A/P Wong Yiik Diew and cosupervisor, Ms. Chai Chen, Dora (Research Associate) for their constant guidance and support, as well as the technicians from Transport & Geospatial Laboratory for their technical support, leading to the completion of this study. In addition, the author wishes to acknowledge the funding support for this project from Nanyang Technological University under the Undergraduate Research Experience on CAmpus (URECA) programme. Environmental Engineering, Nanyang Technological University, Singapore. [9] Federal Highway Administration (2008). Surrogate Safety Assessment Model. Federal Highway Administration, U.S. Department of Transportation. FHWA-HRT-08-049 [10] Hayward, J. Ch. (1972). Near miss determination through use of a scale of danger. TTSC 7115, the Pennsylvania State University, Pennsylvania [11] Laureshyn, A. (2006). Assessment of traffic safety and efficiency with the help of automated video analysis. Department of Technology & Society, Traffic & Road, Lund University, Sweden. [12] Mean (2013). Avidemux wiki documentation. Retrieved on June 14, 2014, from http://www.avidemux.org/admwiki/doku.php?id=s tart [13] PTV Group (2013). PTV Vissim. Retrieved from http://vision-traffic.ptvgroup.com/enus/products/ptv-vissim/ REFERENCES [1] Land Transport Authority (2013). Singapore Land Transport Statistics in Brief. Land Transport Authority, Singapore. [2] Fitzpatrick, K., Wooldridge, M.D., and Blaschke, J.D. (2005). Urban Intersection Design Guide: Volume 1 Guidelines. Texas Transportation Institute. FHWA/TX-05/0-4365-P2 Vol. 1 [3] Lim, Z.L. (2010). Traffic Characteristics on the Road Network in NTU Campus. Final Year Report, School of Civil and Environmental Engineering, Nanyang Technological University, Singapore. [4] Shruti, S. B. (2013). Study of Traffic Delay at Signalised Junctions. Final Year Report, School of Civil and Environmental Engineering, Nanyang Technological University, Singapore. [5] Rahka, H., Trani, A. and Ahn, K. (2004). Development of passenger car equivalents for freeway merging sections. ProQuest. ISBN 0-549- 24044-6. [6] Amanpreet, S. A. (2007). Development of Passenger Car Equivalents for Freeway Merging Sections. ProQuest. ISBN 0549240446, 9780549240440 [7] Koonce, P. (2008). Traffic Signal Timing Manual. Federal Highway Administration, U.S. Department of Transportation. FHWA-HOP-08-024 [8] Seet, Q. Y. (2010). Study of Left-Turn-Green- Arrow (LTGA) Traffic Scheme at Signalised Junctions. Final Year Report, School of Civil and 6