Operational and Safety Performance of a Non-Traditional Intersection Design: The Superstreet Taehyeong Kim, Department of Civil and Environmental Engineering, University of Maryland College Park, MD 20740 Phone: (301) 405-6550 Fax: (301) 405-2585
[email protected] Praveen K. Edara, Virginia Transportation Research Council, 530 Edgemont Rd, Charlottesville, VA 22903 Phone:(434) 293-1996 Fax : (434)-293-1990
[email protected] Joe G. Bared, Federal Highway Administration, Turner-Fairbank Highway Research Center, McLean, VA Phone: (202) 493-3314 Fax: (202) 493-3417
[email protected]
Submission date: Nov. 15, 2006 Word Count = 3065 (text, including abstract and references) + 4000 (figures and tables) = 7065
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
1
ABSTRACT Due to heavy traffic volumes, signalized intersections in many regions across United States are becoming increasingly congested. Operational and safety performance of these intersections have drastically decreased. Alternate intersection designs that may perform better than the conventional signalized intersections are being explored. The results of studies conducted on one non-traditional intersection, the superstreet, are presented in this paper. The superstreet design is similar to the median u-turn design but has some additional features that allow for perfect progression of through traffic on the major road in both directions by preventing the minor road traffic from crossing the major road. Three different cases of superstreet design were studied. For each case, experiments using microscopic traffic simulation were conducted for varying traffic volumes and their performances were compared to the conventional designs. Limited safety evaluation was also carried out for these designs using a surrogate safety assessment tool (SSAM) based on simulation. Simulation results showed that the superstreet design with one u-turn lane outperforms the comparable conventional intersection under high traffic volumes. For the two u-turn lanes case, smaller increase in throughput was observed. SSAM results suggest that the superstreet design with one u-turn lane is safer than the comparable conventional design. For the two u-turn lanes design the superstreet is not safer than the comparable conventional design.
BACKGROUND A four-leg intersection that allows all 12 movements (left turn, right turn, and through movements from each approach) is the prevalent design in urban areas. Any design that significantly deviates from this conventional design (CD) can be termed as a non-traditional design (ND). Non-traditional designs that have the potential to offer better level of service or higher safety (or both) deserve to be considered as an alternative to CD. It is necessary for traffic engineers to know the traffic conditions under which a given ND performs better than a corresponding CD, to make informed decisions. In this research, one such design: The superstreet signalized intersection was studied. As shown in Figure 1, the superstreet design prevents minor road traffic from crossing the major road by diverting all movements from the minor road to turn right followed by a u-turn located downstream of the intersection. Traffic movements from the major road remain the same as in the conventional intersection. Traffic signals require only two phases instead of the four or more phases. Phase 1 allows the major road through movements and phase 2 allows the major road’s left turn movements and the minor road through and left movements. Also, the superstreet is likely to be safer than conventional intersection since it has 14 conflict points compared to 32 for a conventional intersection. A possible disadvantage of the superstreet design is the need for one additional lane (for u-turn traffic) than the corresponding conventional design, hence increasing the intersection width by at least one lane width, and the need for space to accommodate u-turn maneuvers for design vehicles.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
2
FIGURE 1 Traffic movements at a Superstreet.
Richard Kramer (1), a traffic engineer in Huntsville, Alabama invented the superstreet design in 1987. The design was further studied primarily by Hummer and Reid (2, 3, 4, and 5). Hummer and Reid (2) employed microscopic traffic simulation (primarily CORSIM) to model travel time, average stops per vehicle, and average speed for a 2.5 mile arterial in suburban Detroit that consisted of five large irregularly spaced signalized intersections. Their results showed that the superstreet produced a 10 % corridor-wide travel time savings and a 15 % increase in average speeds over a comparable conventional design. But, the average number of stops was higher for the superstreet design. In a following study (3) they compared travel times between seven unconventional arterial intersection designs: the median u-turn, superstreet median, quadrant roadway intersection, bowtie, jughandle, split intersection, and continuous flow intersection designs. Comparisons were made with seven existing, high-volume, conventionally designed intersections in Virginia and North Carolina. The results of the CORSIM simulation experiments indicated that the superstreet was only competitive with the conventional design at intersections with two-lane cross streets. Based on their simulation studies in (4) Hummer and Reid suggest considering a superstreet where high-arterial through volumes conflict with moderate to low cross-street through volumes. This will be the case for many suburban arterials where roadside development generates most of the conflicting traffic. One should also consider a superstreet where close to 50/50 arterial through-traffic splits exist for most of the day, but uneven street spacings remove any chance of establishing two-way progression. As for median U-turns, arterials with narrow medians and no prospects for obtaining extra rights-of-way for widening are poor candidates for the superstreet. In a recent study, Hummer (5) commented that safety is impossible to simulate at this point in our profession, and full superstreet installations are limited in the U.S. thus far so to estimate the safety potential of this design we must turn to the literature on related designs. The Maryland State Highway Administration uses the term ‘J-turn’ instead of superstreet for exactly the same intersection design with non-signalized u-turns. A mini summary report (6) of NCHRP 15-30 (Median Intersection Design for Rural High-Speed Divided Highways)
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
3
presents the tremendous safety improvements of the J-turn design as compared to the conventional intersection design obtained at a site in Maryland. At one site where it was implemented, crash statistics were analyzed for three years. The number of crashes decreased from an average of nine crashes per year to an average of less than one crash per year. The intent of this study was to come up with a range of traffic volumes for which the superstreet designs perform better than comparable conventional geometries. The findings of this study should assist traffic engineers faced with the challenge of when to implement the superstreet design.
ANALYSIS METHODOLOGY Performances of the superstreet designs and the comparable conventional designs were obtained using traffic simulation. VISSIM, a microscopic, time-step, and behavior based simulation tool was used for this analysis. Three superstreet designs (see Table 1) were simulated: Case 1) One left lane and two through lanes on the major road, Case 2) One left lane and three through lanes on the major road, Case 3) Two left lanes and three through lanes on the major road. The minor road always had one lane (case 1 and 2) and two lane (case 3) for through and left turns and one right turn lane. VISSIM layouts of these designs and the corresponding conventional designs are shown in Figures 2, 3, and 4. TABLE 1 Cases studied for the Superstreet Cases Case 1 Case 2 Case 3
Road Major Road Minor Road Major Road Minor Road Major Road Minor Road
Number of Lanes Left Through 1 2 1 (shared lane: left and through) 1 3 1 (shared lane: left and through) 2 3 2 (shared lane: left and through)
Right 1 1 1 1 1 1
The first case was simulated for three major traffic scenarios: high, medium, and low; the remaining two designs were not studied for medium and low volumes as their application is mainly intended for sites operating under heavy traffic conditions. Further, these traffic scenarios are subdivided into a minimum of three levels of traffic volumes. Case 1 For the ‘low to medium volume’ scenarios, levels 1, 2, 3, and 4, the proportion of minor road’s green time to major road’s green time (ratio) was 0.3. On the major road left turn traffic varied from 100 vphpl to 250 vphpl and through traffic varied from 350 vphpl to 850 vphpl; on the minor road left turn traffic was in the range of 40 vphpl to 100 vphpl and through traffic varied between 60 vphpl to 130 vphpl. Table 2 has the detailed traffic volumes for each scenario. For the ‘high volume’ scenarios, three different timing plans were used: ratio of 0.4 for scenario 5, 0.3 for scenario 6, and 0.2 for scenario 7. For each scenario, near saturation flows were computed. Left turn traffic on the major road varied from 260 vehicles per hour per lane (vphpl) to 340 vphpl and through traffic varied from 900 vphpl to 1150 vphpl; on the minor road both left turn and through traffic were in the range of 100 vphpl to 180 vphpl (see Table 2).
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
4
TABLE 2 Directional traffic volumes for all scenarios of Case 1 Minor road Southbound (1 lanes) R TH L 50 60 40 80 70 70 120 130 90 100 100 100 150 180 180 150 140 140 120 100 100
Minor road Northbound (1 lanes) R TH L 50 60 40 80 70 70 50 60 50 100 100 100 150 180 180 150 140 140 120 100 100
Level 1 2 3 4 5 6 7 Unit: veh/hour R: Right Turn, TH: Through, L: Left Turn
Major road Eastbound (2 lanes) R TH L 100 700 100 170 1000 170 250 1500 250 200 1700 200 350 1800 320 350 2000 340 350 2300 260
R 100 170 100 200 350 350 350
Major road Westbound (2 lanes) TH 700 1000 1000 1700 1800 2000 2300
L 100 170 300 200 320 340 260
FIGURE 2 Case 1: Superstreet (left) and Corresponding Conventional Intersection (Right).
Case 2 Cases 2 and 3 are representative of intersections operating under heavy traffic volumes. Therefore, their performance was measured only for the high volume scenarios. Case 2 is similar to case 1 with only one difference; case 2 has one additional through lane on the major road. Three different timing plans were used: ratio of 0.2 for scenario 1, 0.3 for scenario 2, and 0.4 for scenario 3. Table 3 has the detailed traffic volumes for each scenario. TABLE 3 Directional traffic volumes for all scenarios of Case 2 Minor road Southbound (1 lanes) R TH L 120 100 100 150 140 140 150 180 180
Minor road Northbound (1 lanes) R TH L 120 100 100 150 140 140 150 180 180
Level 1 2 3 Unit: veh/hour R: Right Turn, TH: Through, L: Left Turn
TRB 2007 Annual Meeting CD-ROM
R 350 350 350
Major road Eastbound (3 lanes) TH L 3450 260 3000 340 2700 320
Major road Westbound (3 lanes) R TH 350 3450 350 3000 350 2700
L 260 340 320
Paper revised from original submittal.
Kim, Edara, and Bared
5
FIGURE 3 Case 2: Superstreet (left) and Corresponding Conventional Intersection (Right).
Case 3 On the major road, left turn traffic varied from 250 vphpl to 300 vphpl and through traffic varied from 850 vphpl to 1100 vphpl; on the minor road left turn traffic was in the range of 50 vphpl to 70 vphpl and the through traffic varied between 150 vphpl to 210 vphpl (see Table 4). TABLE 4 Directional traffic volumes for all scenarios of Case 3 Minor road Southbound (2 lanes) R TH L 120 300 100 150 360 120 150 420 140
Minor road Northbound (2 lanes) R TH L 120 300 100 150 360 120 150 420 140
Level 1 2 3 Unit: veh/hour R: Right Turn, TH: Through, L: Left Turn
R 350 350 350
Major road Eastbound (3 lanes) TH L 3250 500 2850 600 2550 600
Major road Westbound (3 lanes) R TH 350 3250 350 2850 350 2550
L 500 600 600
FIGURE 4 Case 3: Superstreet (left) and Corresponding Conventional Intersection (Right).
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
6
Simulation Model The traffic signal control at the superstreet operates as follows: the main intersection operates as a two phase signal – phase 1 allows for through movements on the major road, phase 2 allows for left turn movements from the major road concurrently with all traffic movements from the minor road (as left turning and through movements will also turn right at the main intersection). The upstream and down stream signal (about 400ft from the main intersection) also operates as a two phase signal with the same timings as the phases at the main intersection. The U-turn signal (where left and through movements of the minor road make a U-turn) has the same timing plan as phase 2 of the main intersection signal. The signal for the through movement at the u-turn on the major road is the same as phase 1 at the main intersection. The signal timing plans for all superstreet designs were mainly based on traffic volumes. A cycle length of 80 seconds was found appropriate for all traffic scenarios of the superstreet. Depending on the scenario, three different timing plans were used for the high volume scenarios: Scenario 1 - Ratio of minor road’s green time (phase 2) to major road’s green time = 0.2, Scenario 2 - Ratio of minor road’s green time to major road’s green time = 0.3, Scenario 3 - Ratio of minor road’s green time to major road’s green time = 0.4. For medium and low volume scenarios at case 1, a ratio of 0.3 was used. Corresponding conventional intersection designs for each of these three designs were also simulated in VISSIM as shown in Figure 5. Signal timing plans for each traffic scenario were obtained using TRANSYT 7F. A cycle length of 120 seconds was found to be the optimal for all the scenarios of the conventional design. All designs were simulated for a 1-hour time period. In VISSIM, each traffic scenario was run five times with different random seeds and the average values were calculated for the performance measures. Other design and driver behavior assumptions used in the VISSIM model included the following: • Lane width of 12 ft • 400 ft offset in the superstreet design • In the signal setting, yellow and all red signal timings were 3 s and 3 s respectively • No right turn on red for all approaches • 5 % trucks were assumed both on major and minor road • Desired speeds for cars and trucks on major road were 42.3 mph to 48.5 mph and 36.0 mph to 42.3 mph respectively. For minor road, desired speeds for cars and trucks were 29.8 mph to 36.0 mph and 24.9 mph to 28.0 mph respectively. • Wiedemann 74(7) car following model was selected based on its aptness to urban traffic • General lane change behavior was assumed to be free lane selection which means that vehicles are allowed to overtake in any lane. • The desired lateral position of a vehicle at free flow was assumed to be middle of the lane
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
7
FIGURE 5 3-D Visualization of the superstreet design(left) and conventional intersection(right) simulation in VISSIM (Case 1).
Safety Analysis Safety analysis was carried out using the Surrogate Safety Assessment Methodology (SSAM) tool. SSAM, which is currently under development by FHWA (an alpha version used in this study), computes safety measures from traffic simulation models (VISSIM, AIMSUN, and TEXAS Model) by analyzing the trajectories of vehicles and estimating their proximity to each other. Among several safety measures computed by SSAM, two measures were identified to be crucial for this study - Time to Collision (TTC) and Maximum Speed (MaxS). TTC is the time after which a collision could occur between two approaching vehicles, if they continue to maintain the same speed differential and proceed with their course. MaxS is the maximum of the speeds of the two vehicles involved in the conflict. Two cases of the superstreet design, 1) one u-turn lane (case 2) and 2) two u-turn lanes (case 3) were studied using SSAM. Table 5 shows the traffic volumes for three scenarios (high, medium, and low) for these two cases. Each scenario was simulated in VISSIM for a 1-hour time period for ten different random seeds and trajectories of vehicles were recorded. These trajectories were then input to SSAM tool. All events with TTC values less than the threshold value of 1.5 seconds (and greater than 0.1 seconds) and MaxS greater than 10 mph were recorded as potential conflicts. Conflicts were further categorized by their type as crossing conflicts, rear end conflicts, and lane change conflicts. TABLE 5 Volume scenarios of one u-turn lane and two u-turn lanes case for safety analysis Case 2
Scenario
Low Medium High 3 Low Medium High Unit: veh/hour
Left 25 50 100 25 50 100
TRB 2007 Annual Meeting CD-ROM
Minor road Through 25 50 100 75 150 300
Right 30 60 120 30 60 120
Left 67 132 200 167 330 500
Major road Through 700 1,386 2,100 950 1,881 2,850
Right 67 132 200 117 231 350
Paper revised from original submittal.
Kim, Edara, and Bared
8
RESULTS AND DISCUSSION Simulation results for the three design cases and the corresponding conventional designs are shown in Tables 6, 7, and 8. The following performance measures were computed for all vehicles passing through the intersection (both major road and minor road): Delay per vehicle, Queue length, Number of stops, Throughput, Average travel time per vehicle, and Average speed. Throughput denotes the number of serviced vehicles at intersection, and average travel time per vehicle is defined as the time elapsed from the instant it enters the network to the instant it leaves the network. Table 6 shows the results for case 1 scenarios. For high volume scenarios (levels 5, 6, and 7), the superstreet design produced 30% to 41% savings in the average travel time and a 23% to 43% increase in throughput. Superstreet’s performance at medium to low volumes was not very different than the conventional design’s performance. TABLE 6 Performance Measures for Case 1 Level
Intersection
Delay (sec/veh)
Queue (ft)
No. of Stops
Serviced Vehicles
1
Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement
25.8 17.9 31% 30.4 21.4 30% 38.1 25.0 34% 72.9 25.2 65% 130.4 71.8 45% 115.3 57.5 50% 110.1 44.4 60%
11.9 6.7 44% 21.9 11.5 47% 39.7 18.3 54% 127.4 26.7 79% 390.8 350.1 10% 319.8 231.7 28% 193.6 156.2 19%
0.5 0.6 1% 0.6 0.6 0% 0.7 0.7 0% 1.4 0.7 53% 2.3 1.6 31% 2.1 1.2 41% 2.1 0.9 58%
2057 2065 0% 3054 3083 1% 3817 3826 0% 4583 4707 3% 4628 5672 23% 4772 5914 24% 4344 6223 43%
2
3
4
5
6
7
Average Travel Time (sec/veh) 74.7 70.0 6% 79.5 74.0 7% 86.9 77.3 11% 123.2 76.9 38% 186.3 129.6 30% 169.2 111.5 34% 161.9 96.2 41%
Average Speed (mph) 27.03 29.46 9% 25.40 27.89 10% 23.25 26.68 15% 16.59 26.83 62% 10.92 16.12 48% 12.02 18.60 55% 12.52 21.37 71%
Average travel time and throughput values for case 1 are also illustrated in Figures 6 and 7. Entering volume in these charts denotes the total demand or the total number of vehicles that would like to pass through the intersection. However, the actual number of vehicles that are able to pass through the intersection in the simulation time period is denoted by throughput. Therefore the values of entering volumes are always greater than the throughput. It is clear from these figures that as the entering volumes increase the one u-turn superstreet design performs better than its comparable conventional design.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
9
200
Average Travel Time(sec/veh)
180 160 140 120 100 80 60 2100(level 1) 3140(level 2) 3900(level 3) 4800(level 4) 5960(level 5) 6240(level 6) 6460(level 7) Entering Volumes(Veh/hour) Conventional
Superstreet
FIGURE 6 Comparison Plot: Average Travel Time (Case 1).
Number of Serviced Vehicles(Veh/hour)
7000
6000
5000
4000
3000
2000 2100(level 1) 3140(level 2) 3900(level 3) 4800(level 4) 5960(level 5) 6240(level 6) 6460(level 7) Entering Volumes(Veh/hour) Conventional
Superstreet
FIGURE 7 Comparison Plot: Throughput (Case 1).
Simulation results for case 2 are shown in Table 7. The superstreet design produced 15% to 43% savings in the average travel time and a 22% to 34% increase in throughput. TABLE 7 Performance Measures for Case 2 Level
Intersection
Delay (sec/veh)
Queue (ft)
No. of Stops
Serviced Vehicles
1
Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement
92.2 30.8 67% 57.1 40.3 29% 136.8 42.4 69%
212.0 66.8 68% 172.7 110.3 36% 480.7 144.7 70%
1.8 0.7 61% 0.9 0.9 0% 2.5 1.0 59%
6985 8542 22% 6028 8068 34% 6169 7549 22%
2
3
TRB 2007 Annual Meeting CD-ROM
Average Travel Time (sec/veh) 142.4 81.0 43% 108.7 92.3 15% 191.7 83.4 56%
Average Speed (mph) 14.25 25.21 77% 18.75 22.32 19% 10.63 21.57 103%
Paper revised from original submittal.
Kim, Edara, and Bared
10
Results of case 3 are shown in Table 8. Travel time savings of 28% to 31% were achieved with the superstreet design. Throughput for the superstreet was 12% to 23% higher than the conventional design. TABLE 8 Performance Measures for Case 3 Level
Intersection
Delay (sec/veh)
Queue (ft)
No. of Stops
Serviced Vehicles
1
Conventional Superstreet Improvement Conventional Superstreet Improvement Conventional Superstreet Improvement
86.5 45.7 47% 90.7 47.5 48% 97.3 47.2 52%
199.6 124.5 38% 229.2 181.8 21% 271.3 152.3 44%
1.7 1.0 40% 1.7 1.1 35% 1.8 1.2 36%
7295 8946 23% 7346 8573 17% 7277 8174 12%
2
3
Average Travel Time (sec/veh) 136.7 99.0 28% 142.2 101.9 28% 149.7 103.6 31%
Average Speed (mph) 14.38 20.57 43% 13.67 20.05 47% 13.18 20.00 52%
SSAM results for the safety analysis are shown in Table 9. Mean and variance of number of conflicts (categorized by type) are shown. To test whether the superstreet design is safer than conventional design, t-test was performed at 95% confidence level. For 1 u-turn lane case (case 2), only mean values of rear end and total conflict were significantly different. This means that the total number of conflicts in a superstreet is about 80% lower than the number of conflicts at a comparable conventional intersection at 95% confidence level. Thus, we can say that the superstreet design with one u-turn lane is safer than a comparable conventional design. For the 2 u-turn lanes case (case 3), t-test results show that the superstreet is less safe than the comparable conventional design. This could happen due to minor road vehicles trying to position themselves or change lanes for left-turn and through before or after the u-turn movement: resulting in higher number of lane-change and rear-end crashes when simulated. TABLE 9 SSAM Results: 1 u-turn lane and 2 u-turn lanes cases 1 u-turn lane case Mean Variance t-test value (95%) Improvement Result 2 u-turn lanes case
Crossing Conflicts
Rear End Conflicts
Conv. Super. 0.40 0.00 0.93 0.00 1.312 (1.812) 100.00% Not significant
Conv. Super. 100.70 0.00 360.01 0.00 16.783 (1.812) 100.00% Significant
Crossing Conflicts
Rear End Conflicts
Conv. Super. Conv. Super. Mean 0.00 0.00 15.00 27.20 Variance 0.00 0.00 12.22 21.07 t-test value(95%) 0.000 (1.812) -6.687 (1.812) Improvement 0.00% -81.33% Result Not significant Significant Conv. – Conventional Intersection, Super. – Superstreet Intersection Unit: conflicts per hour
TRB 2007 Annual Meeting CD-ROM
Lane Change Conflicts Conv. Super. 24.60 27.00 28.49 35.11 -0.952 (1.812) -9.76% Not significant Lane Change Conflicts Conv. Super. 18.70 32.70 14.90 34.68 -6.288 (1.812) -74.87% Significant
Total Conflicts Conv. Super. 125.70 27.00 501.57 35.11 13.473 (1.812) 78.52% Significant Total Conflicts Conv. Super. 33.70 59.90 41.12 71.88 -7.794 (1.812) -77.74% Significant
Paper revised from original submittal.
Kim, Edara, and Bared
11
CONCLUSIONS •
At high volumes, the traffic performance of superstreet designs with one u-turn lane, and two u-turn lanes are better in terms of throughput, travel time, and delay than the comparable conventional intersections. For low to medium volumes, the superstreet’s performance is not much better than the conventional design’s performance.
•
The superstreet design with one u-turn lane may be safer than the comparable conventional design. The total number of conflicts in the superstreet is about 80% less than the conflicts at a conventional intersection.
•
The superstreet design with two u-turn lanes may be less safe than the comparable conventional design. It is possibly due to minor road vehicles trying to position themselves or change lanes for left-turn and through movements before or after the u-turn movement: resulting in higher number of lane-change and rear-end crashes when simulated.
•
Operationally, the superstreet design should be considered as an alternative to the conventional design when: the major road has 1 left-turn lane and 2 or 3 through lanes with left-turn traffic between 260 vphpl to 340 vphpl and through traffic between 900 vphpl to 1150 vphpl; and the minor road has 1 shared lane for left-turn and through traffic with each in the range of 200 vphpl to 360 vphpl.
REFERENCES 1. Kramer, R. P. New Combinations of Old Techniques to Rejuvenate Jammed Suburban Arterials. In Strategies to Alleviate Traffic Congestion: Proc. 1987 National Conference, Institute of Transportation Engineers, Washington D.C., 1988. 2. Reid, J.D., and J.E. Hummer. Analyzing System Travel Time in arterial Corridors with Unconventional Designs Using Microscopic Simulation. In Transportation Research Record: Journal of the Transportation Research Board, No. 1678, TRB, National Research Council, Washington, D.C., 1999, pp. 208-215. 3. Reid, J.D., and J.E. Hummer. Travel Time Comparisons between Seven Unconventional Arterial Intersection Designs. In Transportation Research Record: Journal of the Transportation Research Board, No.1751. TRB, National Research Council, Washington, D.C., 2001, pp. 56-66. 4. Hummer, J.E., and J.D. Reid. Unconventional Left-Turn alternatives for Urban and Suburban arterials: An Update. In Transportation Research Circular E-C019: Urban Street Symposium Conference Proceedings, Dallas, TX, June 28-30, 1999, TRB, National Research Council, Washington, D.C., December 2000. 5. Hummer, J. E. From Clark Street to Super ST. TM+E, Vol. 10, No. 4, Oct. 2005, pp. 22-24. http://www.itsworld.com/articles/IntersectionSafety.pdf. Accessed July 5, 2006.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.
Kim, Edara, and Bared
12
6. Thomas Maze. Median Intersection Design for Rural High-Speed Divided Highways. NCHRP 15-30, Iowa State University, Ames, 2006. 7. VISSIM User Manual, Version 4.10. PTV, Germany, March 2005.
TRB 2007 Annual Meeting CD-ROM
Paper revised from original submittal.