Efficiency and Safety Evaluation of Unconventional Intersections

Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

Pooya Najaf1, Srinivas S. Pulugurtha2 1- Research and Teaching Assistant, INES Ph.D. Candidate, The University of North Carolina at Charlotte, NC, USA 2- Associate Professor and Graduate Program Director, Department of Civil & Environmental Engineering, The University of North Carolina at Charlotte, NC, USA

1- [email protected] 2- [email protected]

Abstract The conventional countermeasures such as pre-timed, semi-actuated, and actuated signals, signal coordination systems, multiple left-turn lanes, and intelligent transportation systems are not able to overcome the safety and operational problems of traffic congestion (e.g., delay, fuel consumption, crashes, and pollution) in many urban and suburban areas. The unconventional designs have therefore attracted engineers’ attention as an alternative to solve these problems. Two main objectives of the unconventional designs are reducing delay and the number of conflict points. The unconventional alternatives may reduce the number of conflict points by re-routing some left-turns. It is, however, vital to know whether they can improve safety by reducing the number and intensity of traffic conflicts between vehicles. This research focuses on evaluating the efficiency and estimating the number and intensity of conflicts between vehicles (as a surrogate safety measure) for six most common unconventional intersection (unsignalized median U-turn, signalized median U-turn, super street, bowtie, forward jughandle, and reverse jughandle) and three conventional intersection (pre-timed, optimized, and actuated) designs by simulating the real- world traffic information using VISSIM, Synchro and Surrogate Safety Assessment Model (SSAM) software. According to the final results, pre-timed conventional intersection has the worst performance, while unsignalized and signalized median U-turn intersections have the best performance (i.e., lowest average delay, highest average speed, and lowest number of conflicts, except their high conflicts’ intensity and their long travelled distances). Furthermore, forward and reverse jughandles are safest designs. In addition, super streets cannot be appropriate design alternatives due to their low average speed, high average delay, and high number of conflicts. Keywords: Traffic Simulation, Unconventional Intersections, Traffic Conflicts, Measures of Effectiveness

1. INTRODUCTION

As the growth rate of population and auto ownership is higher than construction of new roadways and transportation facilities, traffic congestion is a major problem in many urban and suburban areas. Congestion leads to several operational, environmental, and safety problems as well as increase in fuel consumption. Since transportation engineers have not overcome these problems using demand management policies and other conventional countermeasures, unconventional designs have gained their attention [1]. Two main principles of unconventional alternatives are reducing delay to through vehicles and the number of conflict points at intersections [2]. However, it is still important to study how much they can reduce the delay and what is their effect on the number and intensity of traffic conflicts. Six types of unconventional intersections are studied in this research, including unsignalized median U-turn, signalized median U-turn, super street, bowtie, forward and reverse jughandle. Their effectiveness is compared after simulation in VISSIM [3] using the real-world information. The number and intensity of conflicts between vehicles is also studied using Surrogate Safety Assessment Model (SSAM) software [4]. Comparing these unconventional designs with the basic conventional intersections including pre-timed, optimized (simulated in Synchro [5]), and actuated (simulated in VISSIM) signals helps to

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

understand how effective and safe are the different types of unconventional intersections. A brief description of aforementioned unconventional designs is presented next.

a) Unsignalized and Signalized Median U-Turn In the case of unsignalized and signalized median U-turn, the left-turns to and from the arterial use median crossovers (Figure 1-a). This unconventional design helps eliminate left-turns, while a combination of right-turn and U-turn replaces each left-turn [2].

b) Super Street The super street is an extension of the median U-turn in which through arterial movements have higher priority. As it is shown in the Figure 1-b, the four intersection approaches are designed as two independent three- approach intersections [2].

c) Bowtie The bowtie (Figure 1-c) is another type of the median U-turn design with the median and directional crossovers on the cross-street. This eliminates left-turn at the main intersection. The bowties could be roundabouts on the cross-street. This overcomes the disadvantage of requiring a wide right-of-way (ROW) on the cross-street [2].

d) Jughandle The forward jughandle (Figure 1-d) uses ramps prior to the intersection. This design helps to diverge the left- turns from the right side of the arterial [2]. Reverse jughandle (Figure 1-e), on the other hand, uses roundabouts for arterial left-turns.

Major Major

Minor Minor Figure 1-b- Super Street Figure 1-a- Median U-turn

Major Major

Minor Minor

Figure 1-c- Bowtie Figure 1-d- Forward Jughandle

Major

Minor

Figure 1-e- Reverse Jughandle Figure 1- Different Types of Unconventional Designs

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

Some of the advantages and disadvantages of the aforementioned designs is as follows [2, 6, 7 and 8]. The advantages of the median U-turn intersection include reduced delay for through arterial movements, increased capacity, easier progression for through arterial traffic and fewer stops, fewer threats to pedestrians, and fewer and more separated conflict points. Its disadvantages include driver confusion, increased delay for left-turns, increased travel distances and stops for left-turns, and need for larger ROW. The advantages of the super street over a conventional intersection include reduced delay for through arterial movements and also for left-turns from the arterial, reduced stops for through arterial traffic, fewer threats to pedestrians, and reduced and separated conflict points. Its disadvantages include driver confusion, pedestrian confusion, and increased delay, travel distances and stops for cross-street movements and also for left-turns to the arterial. The advantages of the bowtie over the conventional signalized intersections include reduced delay for through arterial movements, increased capacity, reduced stops and easier progression for through arterial movements, fewer threats to pedestrians, and reduced and more separated conflict points. Its disadvantages include driver confusion, possible driver disregard for the left-turn prohibition, increased delay, travel distance and number of stops involving left-turn traffic and possibly cross-street through traffic, and confusing arterial U-turn. The advantages of the jughandle over the conventional signalized intersections include reduced delay and stops and easier progression for through arterial movements, narrower ROW, and reduced and more separated conflict points. Its disadvantages include driver confusion, increased delay, travel distances and stops for left-turns from the arterial, more difficult movements for pedestrians to cross ramps and the main intersection, and lack of access to arterial for parcels next to ramps.

2. DATA DESCRIPTION

Studied corridor, in this research, is extracted from Moon et al. [1] paper, in which the number of conflicts and average travel time is compared between a real conventional intersection and a simulated super street. The study area is an arterial in a rural area on the four-lane divided National Highway 38, Gyeonggi-do, South Korea. The corridor section with a length of 1,670 meters and speed limit of 80 km/h contains three signalized intersections, spaced by a distance of about 400 meters (Figure 2). In this research, all vehicles including all traffic movements on the corridor are studied (i.e., network performance) rather than comparison between through major arterial and minor cross-street traffic flows. The values in parenthesis, in the figure, show the percentage of heavy vehicles and yellow time for each phase is 3 seconds. Based on the real field data, the signalized intersections are uncoordinated and with different pre-timed cycle lengths. The figure shows 15 min peak hour (18:00-18:15) turning volumes for each intersection along with the proportion of heavy vehicles for the corridor and signal phases [1].

363 m 375 m

Minor Street 1 Minor Street 2 Minor Street 3

6 (0) 6 (0) 74 (35) 1 (0) 4 (25) 6 (0) 7 (14) 6 (10) 10 (0) 89 (10) 95 (8) 22 (4) 121 (5)

141 (3) 87 (0) 145 (3 9 (0) 4 (0) 10 (0) 13 (0) 203 (5) 199 (5) 233 (8)

153 (14) 153 (0) 175 (13) 6 (17) 10 (0) Phase Phase 14 (0) 3 (0) Phase Time 120 24 Time 17 90 25 Time 13 22 12 30

Figure 2- Study Area [1]

Since the real volumes are not balanced, in this research new balanced volume values are used in the simulation procedures (Figure 3).

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

6 6 6 74 22 4

11 10 4 6 9 12 238 121 204 152 203 110 145 139 178 153 154 6 175 13

10 14 Figure 3- Balanced Volumes 3

3. SIMULATION PROCESS

a) Simulating the Conventional Intersection First, the aforementioned corridor with the balanced volumes, same geometric characteristics, and existing pre-timed signalization is simulated in VISSIM.

b) Calibration Using Field Data Calibration is needed after simulating the main network in the VISSIM. This step is strongly dependent on the availability of the data. Traffic volumes, travel time, delay, and average speed are most common parameters for calibration. However the only available parameter in this research is the number of vehicles for each movement (i.e., turning volumes). Calibration process can be done by changing several parameters such as desired speed, speed distribution, car following model, minimum look ahead distance, maximum look ahead distance, the number of observed vehicles, look back distance, general lane change behavior, maximum deceleration, accepted deceleration, minimum headway, waiting time before diffusion, safety distance reduction factor, desired position at free flow, observation of vehicles on next lanes, lateral distances, signal control decision model, reduction safety factor close to stop line, start upstream of stop line, end upstream of stop line, and so on. The total number of data collectors in the simulated network is 36, including 6 inputs, 6 outputs and 24 inter- network collectors. The final goal is obtaining results from collectors as close as possible to the real field volumes. Figure 4 shows the accuracy of simulated volumes versus field volumes after calibration.

Figure 4- Simulated Volumes vs. Field Volumes after Calibration

c) Simulating the Optimized and Actuated Conventional Intersection Since the signal timing for the studied corridor is pre-timed, optimized conventional intersection is simulated in Synchro and then VISSIM. In addition, the actuated signal timing is simulated in VISSIM to compare with other conventional and unconventional intersections.

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

d) Simulating Unconventional Intersections After calibrating the conventional intersection, unconventional designs are simulated in VISSIM using same values for simulation parameters. In addition to simulation parameters, other criteria such as area type (i.e., rural or urban), roadway characteristics (e.g., median type and width),the number of lanes on major and minor roads, turning volumes for major and minor roads, and speed limits should be same for all simulated corridors.

e) Signal Phasing How to signalize the intersection is one of the most important issues in simulating the unconventional designs. Most appropriate actuated signalization and phasing for every type of unconventional intersections is extracted from the Federal Highway Administration (FHWA) suggested procedures [10] and other design guides [2, 11]. Also, detectors are used to simulate these actuated signals. Figure 5 shows the phasing for signalized U- turn, super street, bowtie and forward jughandle intersections. Phasing for reverse jughandle is similar to bowtie’s signalization. The numbers on top of the rings show the order of the phases (e.g., signalized U-turn intersection has two phases) [2].

2 3 1

1 1 Major 2 4 2 Major 4

2 1 3 Minor Minor Figure 5-a- Median U-turn Signalization Figure 5-b- Super Street Signalization

2 1

3 Major 4 Major 1 3

Minor 1 Minor 2 Figure 5-c- Bowtie Signalization Figure 5-d- Forward Jughandle Signalization Figure 5- Signal Phasing of Unconventional Designs

f) Validation Before the analysis (i.e., comparison between the intersections), simulated corridors must be validated considering the traffic volumes. Table 1 represents the field and simulated output and input volumes. As it is clear in the table, bowtie and conventional actuated designs simulate volumes close to the field observations.

Table 1- Traffic Volume Validation for Simulated Corridors Real Simulated Real Simulated Intersection Type Input Input Output Output Basic 796 793 727 676 Optimized 796 793 727 701 Actuated 796 793 727 715 Unsignalized U-turn 796 793 727 707 Signalized U-turn 796 793 727 666 Super Street 796 793 727 677 Bowtie 796 793 727 717 Forward Jughandle 796 793 727 702 Reverse Jughandle 796 793 727 690

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

4. SAFETY ANALYSIS

Conflicts are one of the commonest surrogate safety measures which are compatible with micro-simulation. Conflict is defined for a situation in which two or more vehicles approach each other on a collision course, and a crash is going to happen if neither of them takes any evasive action [9]. Time to collision (TTC), post encroachment time (PET), time gap, time headway, and initial deceleration rate are main indicators to detect the conflict. TTC is defined as the expected time for two vehicles to collide if they remain at their present speed and on the same trajectory (i.e., the minimum time-to-collision value observed during the conflict for the current location, speed, and trajectory of two vehicles at a given instant). PET is defined as the time between when the first vehicle last occupied a position and the time when the second vehicle subsequently arrived at the same position (i.e., the minimum post encroachment time observed during the conflict) [4, 9]. In this research, micro- simulated trajectories, which are extracted from the VISSIM are used to model the conflicts between vehicles, using TTC and PET as the surrogate safety measures. SSAM can compute the number of conflicts using VISSIM trajectory outputs. Rear-end, lane change and crossing vehicle-to-vehicle interactions are three major types of conflicts, shown in Figure 6 [4]. In this research, 1.5 s and 4.8 s are respectively used as maximum values for TTC and PET.

Figure 6- Rear-end, Lane Change and Crossing Conflicts [4]

5. RESULTS

The efficiency of the simulated designs is compared through VISSIM results after calibration and validation steps. Since the studied corridor contains one 4-legged intersection and two 3-legged intersections, measures of effectiveness (MOEs) are compared for the entire simulated network (network performance comparison). Table 2 shows average delay per vehicle, total delay, average stopped delay per vehicle, average number of stops per vehicle, total number of stops, average speed, total distance travelled, and total travel time as primary MOEs to compare the performance of simulated corridors. Highlighted cells represent two best designs.

Table 2- Measures of Effectiveness for Simulated Corridors Average Average Average Total Total Total Average Total Delay Time Stopped Number of Distance Intersection Type Delay Number Speed Travel per Vehicle Delay per Stops per Traveled Time (h) of Stops (km/h) Time (h) (s) Vehicle (s) Vehicles (km) Basic 51.225 11.284 31.507 1.586 1258 41.361 1111.719 26.879 Optimized 20.248 4.46 6.412 1.04 825 55.716 1136.311 20.395 Actuated 20.835 4.59 7.126 0.987 783 55.381 1137.999 20.548 Unsignalized U-turn 9.436 2.079 0.123 0.583 462 63.62 1247.218 19.604 Signalized U-turn 15.241 3.357 2.85 0.779 618 59.595 1248.527 20.95 Super Street 47.304 10.42 21.178 2.132 1691 42.243 1083.797 25.656 Bowtie 21.626 4.764 5.75 1.216 964 54.725 1154.088 21.089 Forward Jughandle 20.472 4.51 7.62 0.909 721 55.273 1138.304 20.594 Reverse Jughandle 24.44 5.384 8.759 1.156 917 53.263 1131.679 21.247

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

Table 3 shows the number of different types of conflicts and the intensity of the conflicts between the vehicles for all simulated corridors. Higher average values of TTA and/or PET demonstrate high conflicts intensity. Highlighted cells represent the best designs regarding the number and intensity of conflicts.

Table 3- Number and Intensity of Conflicts Number of Conflicts Intensity of Conflicts Intersection Type Lane Mean Variance Mean Variance Crossing Rear-end Total Change TTC TTC PET PET Basic 3 146 8 157 0.93 0.34 1.58 2.25 Optimized 15 152 5 172 0.92 0.37 1.34 1.89 Actuated 0 143 9 152 0.96 0.33 1.16 1.50 Unsignalized U-turn 12 79 21 112 0.42 0.34 0.43 0.65 Signalized U-turn 4 138 26 168 0.71 0.39 0.84 1.14 Super Street 1 309 38 348 0.80 0.37 0.99 1.60 Bowtie 2 203 37 242 0.85 0.33 1.18 1.63 Forward Jughandle 0 103 6 109 1.03 0.28 1.35 1.49 Reverse Jughandle 1 162 15 178 1.04 0.27 1.32 1.76

Results from Tables 1 to 3 are summarized as follow: - Bowtie and actuated intersections have the highest capacity. - Signalized U-turn and the basic conventional intersections have the lowest capacity. - Unsignalized and signalized U-turn intersections have the lowest average delay, total delay, average stopped delay, average number of stops, and total number of stops values, and the highest average speed values (i.e., highest efficiency). - Basic conventional design has the highest average delay, total delay, average stopped delay, average number of stops, total number of stops and total travel time, and the lowest average speed values (i.e., lowest efficiency). - Super street intersection has the lowest total distance traveled, very low average speed, and very high average delay, total delay, average number of stops, and total number of stops values. - Unsignalized U-turn design and optimized conventional intersection have lowest total travel time values. - Unsignalized and signalized U-turn intersections have the highest total distance traveled values. - Optimized conventional intersection has more number of stops but lower delay (higher average speed) values in comparison with the actuated conventional intersection. - Forward jughandle has lower number of stops and delay (higher average speed) values in comparison with the reverse jughandle. - Forward jughandle and actuated conventional intersection have the lowest number of crossing conflicts. - Optimized conventional and unsignalized U-turn intersections have the highest number of crossing conflicts. - Unsignalized U-turn and forward jughandle have the lowest number of rear-end conflicts. - Super street has the highest number of rear-end conflicts. - Optimized conventional and forward jughandle intersections have the lowest number of lane change conflicts. - Super street and bowtie intersections have the highest number of lane change conflicts. - Unsignalized U-turn and forward jughandle intersections have the lowest number of total conflicts. - Super streets have the highest number of total conflicts. - Unsignalized and signalized U-turn intersections have the lowest values of both TTC and PET (the highest intensity of conflicts). - Forward and reverse jughandle have the highest values of both TTC and PET (the lowest intensity of conflicts).

6. CONCLUSION

This research has two major goals: 1) evaluating the performance (efficiency), and 2) analyzing the number and intensity of conflicts between vehicles in six most common unconventional intersections (unsignalized median U-turn, signalized median U-turn, super street, bowtie, forward jughandle, and reverse jughandle). In

www.SID.ir Archive of SID

8th National Congress on Civil Engineering, 7-8 May 2014 Babol Noshirvani University of Technology, Babol, Iran

addition to these unconventional designs, pre-timed, optimized and actuated conventional intersections are studied. Measures of effectiveness are compared through VISSIM simulations, after calibration and validation. The number and intensity of conflicts were obtained from the SSAM using the simulated trajectories. According to the final results: - Pre-timed conventional intersection does not have satisfying performance. - Optimized and actuated conventional intersections perform equally well, while optimized intersection has a very high capacity. - Unsignalized median U-turn intersection has the lowest average delay, highest average speed, lowest number of conflicts (specially, rear-end conflicts) values. It also has a satisfying capacity. Its only disadvantages are high conflicts intensity and long travelled distances. - Signalized median U-turn intersection has low average delay and high average speed values. However, it does not have a capacity as high as the unsignalized U-turn’s. In addition, its travelled distance and conflict intensity are high. - Super street has low capacity and average speed, and high average delay and number of conflicts. So it does not have a high efficiency or safety. - Bowtie performs moderately, however it has the highest capacity among all studied designs. - Forward jughandle has the lowest number of conflicts with the lowest intensity. It is the safest studied design, and, has a moderate capacity, average delay and average speed. - Reverse jughandle does not show better performance than forward jughandle with respect to the number of conflicts, average speed, average delay and capacity. It is concluded that pre-timed conventional intersections are obviously the worst intersection designs among the studied designs. Furthermore, it seems that unsignalized and signalized U-turn intersections have the best performance except their high conflicts intensity and their large travelled distance. In addition, forward and reverse jughandles are safest designs among studied corridors. Also, it seems that super streets cannot be appropriate alternatives for conventional intersections.

7. REFERENCES

1- Moon, J.P., Kim, Y.R., Kim, D.G., and Lee, S.K. (2011), “The Potential to Implement a Superstreet as an Unconventional Arterial Intersection Design in Korea”, KSCE Journal of Civil Engineering, Vol. 15, Issue 6, pp.1109-1114. 2- Hummer, G.E., and Reid, J.D. (2000), “Unconventional Left-Turn Alternatives for Urban and Suburban Arterials”, TRB Circular E-C019: Urban Street Symposium. 3- PTV, (2011), “VISSIM 5.30-05 User Manual”, Karlsruhe, Germany, http://www.ptv-vision.com/en- uk/products/vision-traffic-suite/ptv-vissim/overview/ 4- Surrogate Safety Assessment Model (SSAM). (2008), “Software User Manual”, FHWA-HRT-08-050, http://www.fhwa.dot.gov/publications/research/safety/08049/ 5- Synchro Studio 7. (2007), “Synchro User Manual”, Trafficware, http://www.trafficwareinc.com 6- Furtado, G., Tencha, G., and Devos, H. (2003), “Unconventional Arterial Design, Jughandle Intersection Concept for McKnight Boulevard in Calgary”, Transportation association of Canada. 7- Jagannathan, R., Gimbel, M.A., Bared, G.J., Hughes, W.E., Persaud, B., and Lyon, C. (2006), “Safety Comparison of New Jersey Jug Handle Intersections and Conventional Intersections”, Transportation Research Record: Journal of the Transportation Research Board, No. 1953, Transportation Research Board of the National Academies, Washington D.C., pp. 187–200. 8- Wisconsin Department of Transportation. (2008), “Environmental Evaluation of Facilities Development Actions”. 9- Caliendo, C., and Guida, M. (2012), “A Micro-Simulation Approach for Predicting Crashes at Un-signalized Intersections Using Traffic Conflicts”, Journal of Transportation Engineering, Accepted June 13, 2012. 10- U.S. Department of Transportation. (2010), “Alternative Intersections/Interchanges: Informational Report (AIIR)”, FHWA-HRT-09-060. 11- Wilbur Smith Associates. (2008), “Innovative Intersections: Overview and Implementation Guidelines”, Community Planning Association of Southwest Idaho.