Abstract
Analyzing the capacity of a signalized circular intersection is an essential aspect of traffic flow management. With the increased number of vehicles at the intersection, it is preferable to examine ways to increase capacity without altering the existing geometric features. A signalized circular intersection on the national highway in Győr, Hungary, between 47° 40′ 43.7988″ N and 17° 39′ 37.6668″ E is chosen and analyzed for capacity enhancement. The survey is conducted using 360-degree cameras. The PTV Vissim software is then used to construct a model based on the current and projected vehicle counts, as well as the current and proposed options. The result shows that it is possible to increase the capacity of signalized circular intersections without altering the geometric features.
1 Introduction
Studying traffic flow is becoming a significant issue for nations as the volume of traffic increases and quality-of-service declines [1, 2]. Assuming that vehicles can arrive at the roundabout separately from the major stream and the minor stream, a new capacity model for multilane roundabouts was developed based on capacity evaluation [3]. In terms of safety, signalized circular intersections (often called signalized roundabouts, but according to Hungarian guidelines and throughout this research paper, they are called signalized circular intersections or used interchangeably) have the advantages of being able to handle large numbers of commuters and requiring vehicles entering the intersection to slow down for a yield sign [4]. According to Barna et al. [5], the Hungarian signal control system consists of three components: “leg-by-leg” control, two-phase control, and an all-phase control system. The “leg-by-leg” program controls the green time individually.
In the end, capacity models have their limits when it comes to figuring out the capacity of multilane signalized circular intersections in cities with heavy traffic and traffic congestion. Instead of using these models to figure out how many vehicles can go through the proposed intersection, it would be better to use a more advanced simulation method to test different options. This will aid in determining the delays and entry flow. However, the goal of this research is to analyze how to improve the capacity by providing different options without changing the existing geometric features and by using simulation methods.
2 Data and methods
In Hungary, there are about 5 signalized circular intersections, and among them, the research focuses on the one with multilane entry, which is located in Győr, between 47° 40′ 43.554″ N and 17° 39′ 37.3284″ E. The intersection has 3 entry lanes in Fehervari Street named South-East (SE); in Fehervari Street named North-West (NW); and 4 entry lanes in Ipar Street named North-East (NE); and 4 lanes in Szigethy Attila Street named South-West (SW) at entry lanes. In addition to this, the geometric feature of the intersection is shown in Table 1.
Geometry parameter of a signalized circular intersection
| Approach's | SW | SE | NW | NE |
| Entry width (m) | 18.00 | 18.00 | 15.00 | 15.00 |
| Approach width (m) | 8.00 | 8.00 | 4.00 | 8.00 |
| Number of lanes at the approach | 2 | 2 | 1 | 2 |
| Give way length (m) | 18.00 | 20.00 | 15.00 | 14.00 |
| Circulation road width (m) | 9.00 | 13.00 | 13.50 | 13.50 |
| Radius of the inscribed circle (m) | 38.50 | 38.50 | 38.50 | 38.50 |
The “leg-by-leg” principle governs the six different signal groups for incoming flows at the signalized circular intersection, which turn green one after another. For this study, the signal groups are named P1, P2, P3, P4, P5, and P6, and happen once every 24 h. The signal groups have different cycle lengths and signal coordination. Then, among the signal groups, P1, with a cycle length of 64 s, is chosen to analyze the capacity since it has congestion and an unbalanced traffic flow compared to others in the morning session between 7:30 and 8:30. The survey was conducted in 2022 on April 14, 16, 19, and 22 on weekdays and May 6, 7, 13, and 14 on weekends, with the help of the video-capture method in addition to personal counting to make data-gathering procedures, operations, and accuracy easier.
The researchers counted the number of vehicles on the road every 15 min by first classifying them by size and then assigning each size to a set of categories [6]. Then the Average Daily Traffic (ADT) is calculated for entering vehicles into circulation in terms of Passenger Car Unit (PCU) using the Peak Hour Factor (PHF) for each lane. For current conditions, the ADT is about 4287 PCU/h in the morning session, and it is forecasted for the goal of this research for the next ten years and was calculated to be 5,573 PCU/h [7]. It is observed that the Origin-Destination (OD) matrix for the current peak hour situation is shown in Table 2, and the posted limit speed was 50 km h−1, but vehicles were observed entering the roundabout at speeds between 55 km h−1 and 65 km h−1. It was found that the critical gap is 3 s, and the follow-up time is 2 s.
Origin destination matrix for surveyed traffic volume
| Approaches | Turn to | ||||
| U | Left | Through | Right | ||
| From | SW | 0.00 | 0.41 | 0.47 | 0.12 |
| SE | 0.00 | 0.57 | 0.33 | 0.10 | |
| NW | 0.00 | 0.06 | 0.51 | 0.43 | |
| NE | 0.00 | 0.17 | 0.65 | 0.18 | |
Then PTV Vissim software is used to simulate and create a new model. In the PTV Vissim software, the survey hourly vehicle volume and the future hourly vehicle volume on each entry approach based on their respective time intervals will be used for comparison purposes. To comply with the existing system, PTV Vissim software is calibrated as follows:
Then the new cycle length is 62 s with a 2 s reduction in cycle time to be on the safe side of capacity determination; because of this, a readjustment is made to the signal time for each approach. The width of the road lane is 4.0 m, and the length of the passenger cars ranges from 3.75 to 15 m on average. The width of the average car ranges from 1.7 to 1.9 m.
Each stage has a different inter-green period because it is the time between the end of one green light on one of the legs and the start of another green light on another leg. The Longest waiting length (LW) between the stop line and the giveaway line is 15 m on the SE approach. Based on the rule of thumb, the safest trailing distance (δ) is 3 m, and the queue between two vehicles is kept to a minimum (ΔN) of 3 s for any speed. The simulation runs for 1 h, dividing it into 15-min intervals, like 0–15 min; 15–30 min; 30–45 min; and 45–60 min. This study used Δ (the maximum headway) of 3.0 m as a gap between vehicles since one vehicle may be singled out, whereas vehicles with headways above Δ are made up of more than one vehicle, and then modeled the intersection with PTV Vissim software.
For this study to come to a conclusion, the simulation needs to be run about 11 times.
3 Result and discussion
Using the survey traffic volume with the current options and the proposed options, a comparison is made. On the other hand, the comparison was also made using the projected traffic volume for the next ten years with the current options and with the proposed options. For the proposed plan, the options provided are:
Option 1: Use the minimum speed of 30 km h−1 as the reduced speed and the maximum speed of 60 km h−1 as the desired speed;
Option 2: The cycle length (C) for P1 is reduced from 64 to 62 s;
Option 3: A signal's timing is rearranged with a corresponding sequence and new vehicle composition.
From the current signal time for P1, it is found that the maximum green entry time for SE is 18 s, while the maximum red entry time for NW and NE is 47 s. The green times for pedestrians at the NW, NE, SE, and SW approaches are 27, 37, 29, and 32 s, respectively. The inter-green time between the NW and NE entries is 2 s; 1 s between the NE and SE entries; 0 s between the SE entry and the SW entry; and 4 s between the SW entry and the NW.
On the other hand, the proposed options can change all signal timing parameters at the entry and exit points and be related to the number of future hourly volumes. The red times for the NW and NE have decreased from 47 to 43 and 41 s, respectively. In contrast, the green time for the SW is reduced from 15 to 11 s, and the green time for the SE is reduced from 18 to 15 s. The inter-green time between NW and NE entries has decreased by 2 s; the inter-green time between NE and SE entries has remained unchanged; and the inter-green time between SW and NW entries has decreased by 3 s; however, the inter-green time for SE and SW entries has increased by 3 s due to the change in signal timings for entry approaches. The new vehicle composition increases the number of public transports like a single bus and an articulated bus in the given system.
Figures 1 and 2 show how the delay is simulated for P1 using data from a survey of traffic volume during the morning peak hour before and after using the proposed options for each approach. This gave the average vehicle delay and the average stopped delay for each approach and its turns. Since a “stop delay” is a delay per vehicle in seconds absent public transit stops (which is a pursuit intervention technique maneuver) and in parking lots, it is a pursuit intervention technique maneuver. A vehicle delay is the difference between the theoretical (ideal) travel time and the actual travel time. The vehicle delay and stop delay are different based on time intervals.
The delay caused by P1's traffic volume survey compared to the current options
Citation: Pollack Periodica 18, 3; 10.1556/606.2023.00766
The delay caused by P1's traffic volume survey compared to the proposed options
Citation: Pollack Periodica 18, 3; 10.1556/606.2023.00766
The result shows, for example, that the maximum vehicle and stop delay is between 15 and 30 min at the entry of the SE approach, with delay values of 10.91 and 2.87 s. When comparing each approach, the delay spontaneously increases from SW to the other approaches. Specifically, at the SE approach, the delays are higher than other consecutive approaches. The lengthened intervals between green lights and the lengthy red light duration are to blame for this delay.
On the other hand, the survey traffic volume with the proposed options has a vehicle and stop delay at a time interval between 15 and 30 min on SE of 11.10 and 2.74 s, respectively. As it is shown in Figs 1 and 2, based on the proposed options, the average vehicle and stop delay can decrease or increase at the bounds because the newly adjusted inter-green time with signal time adjustment between the bounds influences delay at all bounds. Comparing the two results, the delay with the proposed one is higher for the same traffic volume and entry speed, but the adjusted signal timing with reduced cycle length in the model shows that there is an increase in delay in all approaches, which may lead to a long queue. But in the SW approach, the delays are less comparatively, so it may happen that the length of entry for this approach is longer, and it has an exclusive lane for right turning vehicles, which helps them approach faster than others.
In addition to the above results, Table 3 shows that the simulation for P1 with the survey traffic volume and proposed option yields the same number of vehicles. This means that the signalized circular intersection can handle the total number of vehicles that are expected to arrive, which is about 4,287 PCU/h, no matter, which of the proposed options is used without changing the shape of the intersection.
Traffic volume of a survey period with the current and proposed options
| Time interval | SW | SE | NW | NE | Total |
| 7:30–7:45 | 367 | 481 | 219 | 147 | 1,214 |
| 7:45–8:00 | 364 | 442 | 221 | 190 | 1,217 |
| 8:00–8:15 | 268 | 316 | 229 | 188 | 1,001 |
| 8:15–8:30 | 211 | 289 | 205 | 150 | 855 |
| Total | 1,210 | 1,528 | 874 | 675 | 4,287 |
So, the capacity of the roundabout can be the same as either the current survey volume or the proposed options. This could mean that the capacity of the roundabout can be slowly increased by adding new options.
Then, based on the above results for improvement scenarios, the shorter cycle length and higher number of vehicles for each approach of the improved plan are put into a PTV Vissim simulation program. But the projected traffic volume led to a new OD matrix, as it is shown in Table 4. Figure 3 shows what happens when improvements are made to vehicles in P1 in their original conditions. Figure 4 shows what happens when improvements are made to vehicles in P1 with a desired speed of 60 km h−1 and a reduced speed of 30 km h−1 as proposed options for their delays.
Origin destination matrix for proposed traffic volume
| Approaches | Turn to | ||||
| U | Left | Through | Right | ||
| From | SW | 0.00 | 0.51 | 0.33 | 0.16 |
| SE | 0.00 | 0.39 | 0.46 | 0.15 | |
| NW | 0.00 | 0.13 | 0.47 | 0.41 | |
| NE | 0.00 | 0.18 | 0.71 | 0.11 | |
Delay caused by a projected traffic volume in comparison to the current options
Citation: Pollack Periodica 18, 3; 10.1556/606.2023.00766
Delay caused by a projected traffic volume in comparison to the proposed
Citation: Pollack Periodica 18, 3; 10.1556/606.2023.00766
The results demonstrate, for instance, that the maximum vehicle and stop delay at the SE approach entry is between 30 and 45 min, with delay values ranging from 11.00 to 2.77 s. When comparing SW to other methods, the delay grows on its own without any intervention. In particular, delays are longer at the SE approach compared to other subsequent approaches. This holdup is due to the increased length of red lights and decreased frequency of green ones.
However, at 15–30 min on SE, the predicted traffic flow with the proposed choices results in a vehicle and stop delay of 11.27 and 3.21 s, respectively. As can be seen in Figs 3 and 4, the delay across all bounds is affected by the adjusted inter-green time and the signal time adjustment across the bounds. This indicates that the proposed options may lower or increase the typical wait time for automobiles and pedestrians at the crosswalk. At the same volume and rate of entry, the model shows that adjusting the signal timing to a lower cycle length causes an increase in delay at all approaches, which could lead to a long queue. Right-turning vehicles, however, have their own dedicated lane on the SW approach, allowing them to get there significantly faster than those coming in from any other direction.
When the survey traffic volume was compared to the projected traffic volume using the current options, the new option changed the average time it took for a vehicle to move and the time it took for a vehicle to stop. The speed and type of vehicles, as well as the timing of the lights, have a significant impact on the delays in both scenarios. Based on this, it is better to check whether the projected number of vehicles is still present or not in the given roundabout. Tables 5 and 6 illustrate the number of vehicles entering the roundabout after using the projected traffic flow with the current and proposed options for P1 using PTV Vissim code. Based on the current options, the result shows that there are approximately 5,573 PCU/h entering the roundabout. Hence, the given signalized circular intersection can accommodate the projected traffic volume with an enhanced balanced flow pattern, and there are no residue vehicles at this time without changing the geometric features but by using the proposed options.
Vehicle measurement for a projected traffic volume with current options
| Time interval | SW | SE | NW | NE | Total |
| 7:30–7:45 | 457 | 541 | 294 | 253 | 1,545 |
| 7:45–8:00 | 456 | 502 | 299 | 295 | 1,552 |
| 8:00–8:15 | 360 | 356 | 303 | 292 | 1,311 |
| 8:15–8:30 | 303 | 329 | 278 | 255 | 1,165 |
| Total | 1,576 | 1,728 | 1,174 | 1,095 | 5,573 |
Vehicle measurement for a projected traffic volume with a proposed option
| Time interval | SW | SE | NW | NE | Total |
| 7:30–7:45 | 380 | 430 | 291 | 265 | 1,366 |
| 7:45–8:00 | 382 | 427 | 289 | 270 | 1,368 |
| 8:00–8:15 | 374 | 425 | 290 | 272 | 1,361 |
| 8:15–8:30 | 374 | 424 | 288 | 271 | 1,357 |
| Total | 1,510 | 1,706 | 1,158 | 1,078 | 5,452 |
Using the current options, comparing the survey traffic volume to the projected traffic volume shows that there could be a 23% increase in total traffic flow. On the other hand, if both the survey traffic volume and the projected traffic volume are used, both with the proposed options, about 21% more vehicles can use the signalized circular intersection, no matter how long their delays are.
Again, the difference between the projected traffic flow, with the current, and the proposed options is about 2.17%. Due to the proposed options and projected traffic volumes, certain vehicles will be unable to enter a signalized circular intersection's circulation. This may occur if some vehicles are stopped inside the circle for an indicated green time before being released to exit the circle immediately, or if some vehicles are making the right turn due to the assigned desired speed, priority rules, or reduced speed for all flows; however, the proposed plans have a larger number of vehicles accessing the roundabout than during the survey period, with adjustments to cycle time and signal timings.
Based on the results of the simulation, the proposed options make the roundabout more useful by letting more vehicles enter from each approach. This means that the current roundabout can handle more vehicles and have a better flow pattern without changing its shape. This can be done by changing the cycle time and the timing of the traffic lights. This means that over the next ten years, this roundabout will be able to fit an average of 23% more vehicles on all approaches if a projected traffic volume is used with the current situation, and 21% more vehicles on all approaches if a projected traffic volume is used with the proposed options. The number of cars that can enter the multi-lane signalized circular intersection every hour can be increased if the proposed options are used over the next ten years at the best time for P1 (the morning peak hour).
4 Conclusion
In this research, it is said that studying how traffic flows at signalized circular intersections is an important part of designing traffic flow in the transportation field. In this research, the ways to improve the capacity of the intersection without changing the existing geometric features of the intersection are studied. For this aim, one of the signalized circular intersections in Győr, Hungary, is selected and analyzed. The flow of traffic is studied by collecting data with the help of a 360° camera and defining the geometric parameters. Based on this, options are proposed, and traffic volumes are forecasted for the next ten years, like altering the cycle length by modifying the entry speed and turning speed to 60 km h−1 and 30 km h−1. Then PTV Vissim software is used after calibrating at the intersection to analyze the capacity.
The result shows that using the current options, the difference between surveyed and projected traffic volume is about 23%, and using the proposed options, the difference between traffic volume is about 21%, no matter how long their delays are. Since the goal of this research is to increase capacity, it has been shown that if the new model with the projected traffic volume is used along with the proposed option, the current infrastructure can handle more vehicles without changing the shape of the roundabout. This can be an input for decision-makers to see options on how to increase capacity. In the future, the capacity can be set by changing the speed limits in cities based on how safe they are for pedestrians and cyclists. Also, in the future, the capacity of a signalized circular intersection can be studied by changing the geometry.
Acknowledgements
The authors would like to thank PTV Group for the thesis license of PTV Vissim microsimulation software and Dr. Balázs Horváth for making it easy to use PTV Vissim at Széchenyi Istvan University.
References
- [1]↑
W. Cheng, X. Zhu, and X. Song, “Research on capacity model for large circular signalized intersections,” Proced. Eng., vol. 137, pp. 352–361, 2016.
- [2]↑
Roundabouts: an informational guide, 2000. [Online]. Available: https://www.fhwa.dot.gov/publications/research/safety/00067/00067.pdf. Accessed: Dec. 14, 2022.
- [3]↑
A. Šarić and I. Lovrić, “Multi-lane roundabout capacity evaluation,” Front. Built Environ., vol. 3, pp. 1–12, 2017.
- [4]↑
Z. Magyari and C. Koren, “Visibility studies at roundabouts entries,” Pollack Period., vol. 14, no. 3, pp. 63–74, 2019.
- [5]↑
S. Barna and G. Schuchmann, “Traffic performance of signalized circular intersections,” Pollack Period., vol. 12, no. 2, pp. 67–78, 2017.
- [6]↑
EU, Vehicle definitions, Transport policy. [Online]. Available: https://www.transportpolicy.net/standard/eu-vehicle-definitions/. Accessed: Dec. 14, 2022.
- [8]↑
F. V. Webster, “Traffic Signal Settings,” Road Research Technique, Road Research Laboratory, London, Paper no. 39, 1958.
- [10]↑
Protocol for VISSIM Simulation, Washington State Department of Transportation, 2014. [Online]. Available: https://wsdot.wa.gov/sites/default/files/2021-03/TrafficOps-VISSIM-Protocol.pdf. Accessed: Dec. 14, 2022.
