Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Report
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-08-051
Date: June 2008

Surrogate Safety Assessment Model and Validation: Final Report

PDF Version (3.39 MB)

PDF files can be viewed with the Acrobat® Reader®

Chapter 3. Theoretical Validation (Continued)

Case 5: Three-Phase Interchange Versus Four-Phase Interchange (VISSIM)

This case study compares the safety performance of the two primary types of traffic control logic for diamond interchanges. A diamond interchange, as shown in figure 60, is composed of two closely spaced signalized intersections that connect the surface street system to the freeway system. One controller is most often used to control both intersections of a diamond interchange. Because of this, both intersections of the interchange are dealt with together as a single entity.

Two alternatives for interchange traffic control are typically used: three-phase signal control and four-phase signal control.

Figure 60. Illustration. Diamond Interchange. This is an illustration of a diamond interchange. A diamond interchange is composed of two closely spaced signalized intersections that connect the surface street system to the freeway system. One controller is most often used to control both intersections of a diamond interchange.

Figure 60. Illustration. Diamond Interchange.

Three-phase is a legacy term that implies the concurrent service of the two cross-road movements, the two left-turn movements, and the two frontage-road movements during common phases as well as a cross-road left-turn phase that lags (or follows) that of the conflicting cross-road through movement.(7) The interior left turns lag the arterial through movements. The disadvantage of the three-phase sequencing for diamond interchange is that the vehicles on the interior of the diamond cannot be guaranteed to clear during one cycle. Three-phase sequencing is applicable for very wide interchanges over 183 m (600 ft) wide and for traffic demands that are directionally balanced and not too heavy. Figure 61 illustrates the time sequencing for three-phase control.(18)

Figure 61. Screen Capture. Three-Phase Sequencing Signal Control for Diamond Interchange. This is an illustration of a three-phase sequencing signal control for a diamond interchange. The three-phase sequencing signal control means concurrent service of the two crossroad movements, the two left-turn movements, and the two frontage road movements during common phases as well as a crossroad left-turn phase that lags (or follows) the conflicting crossroad through movement. The interior left turns lag the arterial through movements.

Figure 61. Screen Capture. Three-Phase Sequencing Signal Control for Diamond Interchange.

Four-phase sequencing uses an overlap phase. The overlap period provides extra time for the interior area to clear and provide better progression out of the interchange. This helps to minimize interior delay and queuing even though a longer cycle length may be necessary. The four exterior movements are serviced sequentially. This control logic is suitable for narrow interchanges less than 122 m (400 ft) wide and for traffic demands that are high and/or directionally unbalanced. Figure 62 illustrates the time sequencing for the four-phase control.(18)

According to the available crash prediction models for diamond interchanges, the type of signal-phase timing plan will have no impact on crash frequency. Thus, the crash prediction model will generate the same crash results for both three-phase and four-phase signal control.

Figure 62. Screen Capture. Four-Phase Sequencing Signal Control for Diamond Interchange. This is an illustration of a four-phase sequencing signal control for a diamond interchange. Four-phase sequencing uses an overlap phase. The overlap period provides extra time for the interior area to clear and provide better progression out of the interchange. The four exterior movements are serviced sequentially.

Figure 62. Screen Capture. Four-Phase Sequencing Signal Control for Diamond Interchange.

Intersection Description

The intersection used to test control logic for a diamond interchange is shown in figure 63 and figure 64. Because of the two closely spaced signalized intersections, the left-turn bay for each intersection is 45.75 m (150 ft) long, less than the normal 76.25-m (250-ft) length.

Figure 63. Screen Capture. Intersection for Diamond Interchange with Three-Phase Test. This is a screen capture of a diamond interchange model in VISSIM. The diamond interchange is going to apply a three-phase sequencing signal control. The left-turn bay for each intersection is 45.75 m (150 ft) long.

Figure 63. Screen Capture. Intersection for Diamond Interchange with Three-Phase Test.

Figure 64. Screen Capture. Intersection for Diamond Interchange with Four-Phase Test. This is a screen capture of a diamond interchange model in VISSIM. The diamond interchange is going to apply a four-phase sequencing signal control. The left-turn bay for each intersection is 45.75 m (150 ft) long.

Figure 64. Screen Capture. Intersection for Diamond Interchange with Four-Phase Test.

Table 35 shows the traffic volumes arriving to each approach of the intersection. Fully-actuated traffic control is applied in this test. Figure 65 through figure 70 illustrate the timing plans for each volume-level test scenario.

Table 35. Case 5 Service Flow by Each Approach.
Approach
Southbound
Northbound
Eastbound
Westbound
L
TH
R
L
TH
R
L1
TH
R
L2
TH
R
Phase ID (Three Phase)
4
4
 
8
8
 
1
2
 
5
6
 
Phase ID
(Four Phase)
4
4
 
8
8
 
2
2,4
 
6,8
6
 
Low Volumes
100
200
100
100
200
100
100
200
100
100
200
100
Medium Volumes
400
100
50
350
200
200
200
400
150
300
300
300
High Volumes
400
200
300
400
400
200
500
400
200
500
300
300

Note: L, TH, and R correspond to vehicles proceeding left, through, or right at the intersection.

Figure 65. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in Low Volumes. This is a screen capture of the timing plan for a three-phase diamond interchange in low volumes condition. The split time for E-W through movements (phase 2 and phase 6) is 20 seconds. The split time for E-W left-turn movements (phase 1 and phase 5) is 20 seconds. The split time for S-N movement (phase 4 and phase 8) is 20 seconds.

Figure 65. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in Low Volumes.

Figure 66. Screen Capture. Timing Plan for Four-Phase Diamond Interchange in

Figure 66. Screen Capture. Timing Plan for Four-Phase Diamond Interchange in Low Volumes.

Figure 67. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in Medium Volumes. This is a screen capture of the timing plan for a three-phase diamond interchange in medium volumes condition. The split time (all in seconds) for westbound left-turn movement (phase 1), eastbound through movement (phase 2), southbound movements (phase 4), eastbound left-turn movement (phase 5), westbound through movement (phase 6), and northbound movements are 18, 20, 22, 14, 22, and 24, respectively.

Figure 67. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in Medium Volumes.

Figure 68. Screen Capture. Timing Plan Four-Phase Diamond Interchange in Medium Volumes. This is a screen capture of the timing plan for a four-phase diamond interchange in medium  volumes condition. The split time (all in seconds) for eastbound movements (phase 2), southbound movements (phase 4), westbound movements (phase 6), and northbound movements (phase 8) are 26, 26, 28, and 20, respectively.

Figure 68. Screen Capture. Timing Plan Four-Phase Diamond Interchange in Medium Volumes.

Figure 69. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in High Volumes. This is a screen capture of the timing plan for three-phase diamond interchange in high volumes condition. The split time (all in seconds) for westbound left-turn movement (phase 1), eastbound through movement (phase 2), southbound movements (phase 4), eastbound left-turn movement (phase 5), westbound through movement (phase 6), and northbound movements are 23, 27, 20, 23, 26, and 21, respectively.

Figure 69. Screen Capture. Timing Plan for Three-Phase Diamond Interchange in High Volumes.

Figure 70. Screen Capture. Timing Plan for Four-Phase Diamond Interchange in High Volumes. This is a screen capture of the timing plan for a four-phase diamond interchange in high volumes condition. The split time (all in seconds) for eastbound movements (phase 2), southbound movements (phase 4), westbound movements (phase 6), and northbound movements (phase 8) are 35, 26, 33, and 26, respectively.

Figure 70. Screen Capture. Timing Plan for Four-Phase Diamond Interchange in High Volumes.

Data Analysis and Comparison Results

Ten replications were performed for each simulation scenario, and the resulting output trajectory data were analyzed by SSAM. The F-test and t-test were applied to compare surrogate measures of safety and the aggregations of those measures. Table 36 through table 39 list the values of all surrogate measures of safety and corresponding t-test results for different types of aggregations with the low-speed events and crash data excluded (TTC ? 0 and MaxS = 16.1 km/h (10 mi/h)).

Table 36. Case 5 Comparison Results for Total Conflicts.
Total
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
3 phase 4 phase 3 phase 4 phase 3 phase 4 phase
Low Volume Mean
1.20
1.60
3.60
4.20
24.80
36.30
Variance
0.62
2.04
3.82
3.29
22.18
59.12
t-value(95%), difference (%)
-0.775
-0.712
-4.033 , -46.4%
Medium Volume Mean
3.20
1.60
7.40
3.90
63.30
36.90
Variance
3.29
4.49
12.71
6.54
35.57
60.32
t-value(95%), difference (%)
1.814
2.522, 47.3%
8.525 , 41.7%
High Volume Mean
1.80
1.00
6.20
4.30
69.80
47.90
Variance
2.62
0.89
9.07
2.68
214.62
32.54
t-value(95%), difference (%)
1.350
1.753
4.405 , 31.4%

Note: Shaded cells indicate statistically significant differences between the two alternatives. The tan and blue colors indicate extreme values to the right and left columns respectively.

This table shows that the three-phase timing plan shows fewer total conflicts at low-traffic volumes but higher total conflicts at medium- and high-traffic volumes. Also, very few conflict events occur with TTC = 1.0 s.

Table 37. Case 5 Comparison Results for Rear-End Conflicts.
Rear End
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
3 phase 4 phase 3 phase 4 phase 3 phase 4 phase
Low Volume Mean
N/A
N/A
N/A
N/A
13.50
21.10
Variance
N/A
N/A
N/A
N/A
13.39
15.43
t-value(95%), difference (%)
N/A
N/A
-4.477 , -56.3%
Medium Volume Mean
N/A
N/A
N/A
N/A
22.80
16.80
Variance
N/A
N/A
N/A
N/A
41.29
6.62
t-value(95%), difference (%)
N/A
N/A
2.741 , 26.3%
High Volume Mean
N/A
N/A
N/A
N/A
35.90
18.20
Variance
N/A
N/A
N/A
N/A
8.54
9.51
t-value(95%), difference (%)
N/A
N/A
13.172 , 49.3%

Note: Shaded cells indicate statistically significant differences between the two alternatives. The tan and blue colors indicate extreme values to the right and left columns respectively.

Consistent with the result for total conflicts, the three-phase timing plan shows fewer rear-end events for low-traffic volumes and more events for medium- and high-traffic volumes. Rear-end events make up approximately 50 percent of the total conflicts.

Table 38. Case 5 Comparison Results for Lane-Change Conflicts.
Lane Change
TTC = 0.5 TTC =1.0 TTC = 1.5
3 phase 4 phase 3 phase 4 phase 3 phase 4 phase
Low Volume Mean
1.10
1.60
3.30
3.90
11.20
14.90
Variance
0.54
2.04
2.68
3.21
12.62
32.54
t-value(95%), difference (%)
-0.983
-0.782
-1.741
Medium Volume Mean
3.20
1.60
5.90
3.60
27.20
18.70
Variance
3.29
4.49
8.32
6.49
28.62
30.23
t-value(95%), difference (%)
1.814
1.890
3.504, 31.3%
High Volume Mean
1.80
1.00
5.50
3.80
46.70
30.90
Variance
2.62
0.89
8.94
2.40
95.79
23.66
t-value(95%), difference (%)
1.350
1.596
4.572 , 33.8%

Note: Shaded cells indicate statistically significant differences between the two alternatives.

Consistent with the result for total conflicts, the three-phase timing plan shows fewer lane-change events for low-traffic volumes (although this result is not statistically significant) and more events for medium- and high-traffic volumes.

Table 39. Case 5 Comparison Results for Average Surrogate Measures of Safety.
  T3 T4 C3 C4 RE3 RE4 LC3 LC4
TTC (low)
1.29
1.29
N/A
N/A
1.39
1.39
1.17
1.16
t -value, diff(%)
0.000
N/A
0.000
0.208
TTC (med)
1.29
1.29
N/A
N/A
1.37
1.37
1.18
1.2
t -value, diff(%)
0.000
N/A
0.000
-0.580
TTC (high)
1.29
1.3
N/A
N/A
1.38
1.39
1.26
1.25
t -value, diff(%)
-0.725
N/A
-0.695
0.531
PET(low)
2.24
2.31
N/A
N/A
2.72
2.63
1.64
1.87
t -value, diff(%)
-0.955
N/A
0.938
-1.507
PET(med)
2.47
2.43
N/A
N/A
2.88
3.15
1.93
1.74
t -value, diff(%)
0.520
N/A
-3.382, -9.38%
1.740
PET(high)
2.28
2.27
N/A
N/A
3.14
3.18
1.86
1.77
t -value, diff(%)
0.154
N/A
-0.479
1.310
MaxS(low)
8.17
7.93
N/A
N/A
6.62
6.78
10.06
9.57
t -value, diff(%)
0.719
N/A
-0.631
0.796
MaxS(med)
7.33
6.8
N/A
N/A
6.8
6.46
8.05
7.13
t -value, diff(%)
3.108, 7.23%
N/A
2.037, 5% 3.259, 11.43%
MaxS(high)
6.48
6.32
N/A
N/A
6.37
6.29
6.54
6.34
t -value, diff(%)
1.637
N/A
0.544
1.566
DeltaS(low)
5.48
5.49
N/A
N/A
5.45
5.57
5.5
5.36
t -value, diff(%)
-0.072
N/A
-0.768
0.576
DeltaS(med)
3.69
2.91
N/A
N/A
3.37
2.36
4.1
3.44
t -value, diff(%)
5.966, 21.14%
N/A
5.067, 29.97% 4.031, 16.1%
DeltaS(high)
3.08
2.8
N/A
N/A
2.3
1.76
3.45
3.35
t -value, diff(%)
3.024, 9.09%
N/A
3.086, 23.48%
1.133
DR(low)
-2.98
-3.09
N/A
N/A
-2.86
-2.84
-3.11
-3.45
t -value, diff(%)
0.872
N/A
-0.190
1.312
DR(med)
-2.32
-1.8
N/A
N/A
-2.06
-1.54
-2.65
-2.06
t -value, diff(%)
-4.073, 22.41%
N/A
-4.570, 25.24% -2.479, 22.26%
DR(high)
-2.01
-1.57
N/A
N/A
-1.57
-1.23
-2.21
-1.73
t -value, diff(%)
-3.598, 21.89%
N/A
-2.928, 21.66% -2.719, 21.72%
MaxD(low)
-3.75
-3.62
N/A
N/A
-3.25
-3.1
-4.36
-4.37
t -value, diff(%)
-0.929
N/A
-1.379
0.036
MaxD(med)
-4.05
-4.27
N/A
N/A
-3.2
-2.8
-5.17
-5.7
t -value, diff(%)
1.486
N/A
-3.575, 12.5% 2.362, -10.25%
MaxD(high)
-4.78
-4.7
N/A
N/A
-3.05
-2.85
-5.62
-5.7
t -value, diff(%)
-0.563
N/A
-1.479
0.472
MaxDeltaV(low)
2.88
2.88
N/A
N/A
2.85
2.89
2.89
2.84
t -value, diff(%)
0.000
N/A
-0.466
0.394
MaxDeltaV(med)
1.94
1.52
N/A
N/A
1.77
1.24
2.16
1.8
t -value, diff(%)
6.026, 21.65%
N/A
5.064, 29.94% 4.047, 16.67%
MaxDeltaV(high)
1.61
1.46
N/A
N/A
1.22
0.92
1.8
1.75
t -value, diff(%)
3.059, 9.32%
N/A
3.232, 24.59%
1.052

Note: Shaded cells indicate statistically significant differences between the two alternatives.

This table indicates that four-phase control reduces the severity of conflict events.

Correlations with Predicted Crash Frequency

The predicted crash rates for all scenarios in this test are listed in table 40 with the corresponding average number of conflict events of each type. The Spearman rank correlation coefficients are calculated for each test and listed in the right-most column. Note that because there is no term in the crash prediction model to discriminate between three-phase and four-phase control, the predicted number of crashes for each case is the same. The results from SSAM would indicate that a three-phase design results in more conflict events than a four-phase design and is thus a less safe intersection.

Table 40. Case 5 Spearman Rank Correlations Between Conflicts and Crash Frequency.
AADT
Low
Medium
High
Rs
3 4 3 4 3 4
Crash frequency M
3.6
3.6
7.1
7.1
10.2
10.2
1
R
1
1
3
3
5
5
Total Conflict M
24.8
36.3
63.30
36.90
69.80
47.90
0.91
R
1
2
4
3
6
5
Crossing Conflict M
 
 
 
 
 
 
N/A
R
 
 
 
 
 
 
Rear-End Conflict M
13.5
21.1
22.80
16.80
35.90
18.20
0.46
R
1
4
5
2
6
3
LC Conflict M
11.20
14.90
27.20
18.70
46.70
30.90
0.94
R
1
1
4
3
6
5

Note: Rows labeled "M" provide mean values and rows labeled "R" provide the ranking of each alternative. The Rs column provides Spearman rank correlation coefficients indicating agreement with theoretical crash estimates.

Findings and Conclusions

Based on the observation on the safety surrogate data obtained from the test, the following conclusions can be drawn:

  • There is little appreciable difference for any of the safety measures between three-phase and four-phase control logic. In general, three-phase control produces more conflicts than four-phase control for high-traffic volumes.
  • TTC values for rear-end conflicts are typically larger than 1.0s.
  • There are only a few crossing conflicts for both designs.

In general, when traffic volume is low, either logic- for diamond-interchange control has similar results for surrogate measures of safety. When traffic volume increases, three-phase control logic generates more conflicts than four-phase control logic and thus could be considered a less safe control strategy. In addition, the average values of the surrogate measures of safety for three-phase control are all consistently worse, reinforcing the determination that three-phase control is a less safe intersection design than four-phase control.

Case 6: Single-Point Urban Interchange Versus Diamond interchange (VISSIM)

The Single-Point Urban Interchange (SPUI) is a relatively new variant of the diamond (shown in figure 71). Where a diamond has two ramp intersections at the surface street (one on each side of the freeway), the ramps of a SPUI are placed so close together to make them effectively part of the same intersection. This allows one traffic signal to control all crossing movements and enables concurrent opposing left turns, which increases the capacity of the interchange above that of the three- and four-phase diamond control schemes.

Figure 71. Screen Capture. Single-Point Urban Interchange. This is an illustration of a single-point urban interchange. The ramps of a SPUI are placed so close together to make them effectively part of the same intersection. Only one traffic signal controls all crossing movements and enables concurrent opposing left turns.

Figure 71. Screen Capture. Single-Point Urban Interchange.

The traffic signal control for the SPUI has three phases, as shown in figure 72, figure 73, and figure 74.(19)

PHASE 1: Cars on the surface street (shown in red) are allowed to drive straight through only (no turns). The yellow cars waiting to turn onto the freeway must wait.

PHASE 2: All cars on the surface street proceeding straight through or turning left onto freeway are stopped. Cars exiting the freeway to enter the street (green) are allowed to turn left.

PHASE 3: Left-turning vehicles from the freeway (green) and cars proceeding straight through on the surface street (red) are stopped. Cars on the surface street are allowed to turn onto the on-ramp for the freeway (yellow).

Figure 72. Screen Capture. Phase 1. This is an illustration of phase 1 of 3 traffic signal control for a single-point urban interchange. Cars on the surface street (shown in red) are allowed to drive straight through only (no turns at all). The yellow cars waiting to turn onto the freeway must wait.

Figure 72. Screen Capture. Phase 1.

Figure 73. Screen Capture. Phase 2. This is an illustration of phase 2 of 3 traffic signal control for a single-point urban interchange. All cars on the surface street proceeding straight through or turning left onto the freeway are stopped. Cars exiting the freeway and entering the street (shown in green) are allowed to turn left.

Figure 73. Screen Capture. Phase 2.

Figure 74. Screen Capture. Phase 3. This is an illustration of phase 3 of 3 traffic signal control for a single-point urban interchange. Left-turning vehicles from the freeway (shown in green) and cars proceeding straight through on surface street (shown in red) are stopped. Cars on the surface street wishing to turn onto the on-ramp for freeway (shown in yellow) may do so.

Figure 74. Screen Capture. Phase 3.

The disadvantages of a SPUI include the following:

  • Complex intersection and signal phases may be unfamiliar to drivers.
  • Distance between stop bars on the surface street creates problems for bicycles, which need more time to clear the area between them.
  • More free-flow motor vehicle movements (part of what increases the SPUI's capacity) makes it more difficult for pedestrians to safely cross.
  • Vehicle clearance time (where all lights must be red) must be longer than three- or four-phase control.

The goal of the comparison is to identify any differences in the safety performance use of the SPUI intersection with traditional diamond-signal control. Comparison of conventional crash prediction models cited earlier (each applied at the same volume) suggests that the SPUI averages slightly more crashes than a conventional diamond interchange at low and moderate volumes, and the trend is reversed at higher volumes (where SPUIs have fewer predicted crashes than diamond-interchange geometry). However, a recent comparison of SPUI and tight diamond-interchange crashes suggests no significant differences in total crashes, though SPUIs incurred fewer injuries/fatalities than comparable diamond interchanges.(20)

Intersection Description

The intersections used to test the diamond interchange versus SPUI are shown in figure 75, figure 76, and figure 77, respectively. All left-turn bays in the networks are 76.25 m (150 ft) long.

Figure 75. Screen Capture. Diamond Interchange in VISSIM. This is a screen capture of a diamond interchange model in VISSIM. All left-turn bays are 45.75 m (150 ft) long.

Figure 75. Screen Capture. Diamond Interchange in VISSIM.

Figure 76. Screen Capture. Link-Connector View of Diamond Interchange Model in VISSIM. This is a screen capture of an intersection model in VISSIM with a link-connector view. This view portrays links as blue lines and connectors between link segments as purple lines. All lines in this conceptual view have the same thin profile, regardless of how many lanes are associated with the link or connector line. All left-turn bays are 45.75 m (150 ft) long.

Figure 76. Screen Capture. Link-Connector View of Diamond Interchange Model in VISSIM.

Figure 77. Screen Capture. SPUI in VISSIM. This is a screen capture of a single- point urban interchange model with two different types of view in VISSIM. All left-turn bays are 45.75 m (150 ft).

Figure 77. Screen Capture. SPUI in VISSIM.

Table 41 lists the traffic volumes applied for each approach of the intersection. Fully-actuated traffic control is applied in this test. Figure 78 though figure 83 indicate the timing plans for each testing scenario.

Table 41. Case 6 Service Flow by Each Approach.
Approach
Southbound Northbound Eastbound Westbound
L TH R L TH R L TH R L TH R
Phase ID (Three Phase)
4
4
 
8
8
 
1
2
 
5
6
 
Low Volumes
400
0
100
300
0
200
100
300
100
100
300
100
Medium Volumes
450
0
100
400
0
350
200
400
150
300
300
300
High Volumes
700
0
500
700
0
500
600
600
300
600
600
300

Note: L, TH, and R correspond to vehicles proceeding left, through, or right at the intersection.

Figure 78. Illustration. Timing Plan for Diamond Interchange in Low Volumes. This is a screen capture of the timing plan for diamond interchange in low volumes condition. The split time for E-W through movements (phase 2 and phase 6) is 20 seconds. The split time for E-W left-turn movements (phase 1 and phase 5) is 10 seconds. The split time for S-N movement (phase 4 and phase 8) is 20 seconds.

Figure 78. Illustration. Timing Plan for Diamond Interchange in Low Volumes.

Figure 79. Illustration. Timing Plan for SPUI in Low Volumes. This is a screen capture of the timing plan for the SPUI in low volumes condition. The split time for E-W through movements (phase 2 and phase 6) is 24 seconds. The split time for E-W left-turn movements (phase 1 and phase 5) is 12 seconds. The split time for S-N movement (phase 4 and phase 8) is 24 seconds.

Figure 79. Illustration. Timing Plan for SPUI in Low Volumes.

Figure 80. Illustration. Timing Plan for Diamond Interchange in Medium Volumes. This is a screen capture of the timing plan for the diamond interchange in medium volumes condition. The split time (all in seconds) for westbound left-turn movement (phase 1), eastbound through movement (phase 2), southbound movements (phase 4), eastbound left-turn movement (phase 5), westbound through movement (phase 6), and northbound movements (phase 8) are 18, 22, 20, 14, 24, and 22, respectively.

Figure 80. Illustration. Timing Plan for Diamond Interchange in Medium Volumes.

Figure 81. Illustration. Timing Plan SPUI in Medium Volumes. This is a screen capture of the timing plan for the SPUI in medium volumes condition. The split time (all in seconds) for westbound left-turn movement (phase 1), eastbound through movement (phase 2), southbound movements (phase 4), eastbound left-turn movement (phase 5), westbound through movement (phase 6), and north bound movements are 17, 24, 24, 14, 27, and 24, respectively.

Figure 81. Illustration. Timing Plan SPUI in Medium Volumes.

Figure 82. Illustration. Timing Plan for Diamond Interchange in High Volumes. This is a screen capture of the timing plan for the diamond interchange in high volumes condition. The split time for E-W through movements (phase 2 and phase 6) is 25 seconds. The split time for E-W left-turn movements (phase 1 and phase 5) is 15 seconds. The split time for S-N movement (phase 4 and phase 8) is 20 seconds.

Figure 82. Illustration. Timing Plan for Diamond Interchange in High Volumes.

Figure 83. Illustration. Timing Plan for SPUI in High Volumes. This is a screen capture of the timing plan for the SPUI in high volumes condition. The split time for E-W through movements (phase 2 and phase 6) is 25 seconds. The split time for E-W left-turn movements (phase 1 and phase 5) is 40 seconds. The split time for S-N movement (phase 4 and phase 8) is 25 seconds.

Figure 83. Illustration. Timing Plan for SPUI in High Volumes.

Data Analysis and Comparison Results

Ten replications were performed for each simulation scenario, and the resulting output trajectory data were analyzed by SSAM. F-test and t-tests were applied to identify the statistical significance of each surrogate measure of safety. Table 42 through table 45 list the values of all surrogate measures of safety and corresponding t-test results for different types of aggregations with the low-speed events and crash data excluded (TTC ? 0 and MaxS = 16.1 km/h (10 mi/h)).

Table 42. Case 6 Comparison Results for Total Conflicts.
Total
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
DIA SPUI DIA SPUI DIA SPUI
Low Volume Mean
2.1
1
4
3.2
79.6
35
Variance
3.433
1.778
8.667
3.289
104.489
34.444
t-value(95%), difference (%)
1.524
0.732
11.966, 56.0%
Medium Volume Mean
2.5
2.3
7.6
5.5
143.1
64.1
Variance
1.833
3.567
11.156
4.722
549.211
89.656
t-value(95%), difference (%)
0.272
1.667
9.884, 55.2%
High Volume Mean
0.6
3.4
5.1
10.2
140.7
117.2
Variance
0.933
4.711
5.878
8.178
218.233
130.400
t-value(95%), difference (%)
-3.727, -466.7%
-4.302, -100.0%
3.980, 16.7%

Note: Shaded cells indicate statistically significant differences between the two alternatives. The tan and blue colors indicate extreme values to the right and left columns respectively.

This table indicates, when considering a threshold value of 1.5 s for TTC, that the SPUI geometry reduces the number of conflict events by 20 percent to 60 percent.

Table 43. Case 6 Comparison Results for Rear-End Conflicts.
Rear End
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
DIA SPUI DIA SPUI DIA SPUI
Low Volume Mean
N/A
N/A
N/A
N/A
53
13.4
Variance
N/A
N/A
N/A
N/A
87.778
23.378
t-value(95%), difference (%)
N/A
N/A
11.878, 74.7%
Medium Volume Mean
N/A
N/A
N/A
N/A
100.9
31.2
Variance
N/A
N/A
N/A
N/A
353.433
48.178
t-value(95%), difference (%)
N/A
N/A
10.998, 69.1%
High Volume Mean
N/A
N/A
N/A
N/A
98.5
65.3
Variance
N/A
N/A
N/A
N/A
208.056
54.678
t-value(95%), difference (%)
N/A
N/A
6.477, 33.7%

Note: Shaded cells indicate statistically significant differences between the two alternatives.

This table indicates, when considering a threshold value of 1.5 s for TTC, that the SPUI geometry reduces the number of rear-end conflict events by 30 percent to 75 percent.

Table 44. Case 6 Comparison Results for Lane Change Conflicts.
Lane Change
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
DIA SPUI DIA SPUI DIA SPUI
Low Volume Mean
2.1
0.9
3.7
2.9
26.3
21.6
Variance
3.433
1.211
7.344
3.211
26.456
30.044
t-value(95%), difference (%)
1.761
0.779
1.977
Medium Volume Mean
2.5
2.3
6.4
5.1
42.1
32.4
Variance
1.833
3.567
6.711
4.100
43.211
54.933
t-value(95%), difference (%)
0.272
1.250
3.096, 23%
High Volume Mean
0.6
3.4
3.4
8.6
42
51.5
Variance
0.933
4.711
4.711
6.711
28.000
101.611
t-value(95%), difference (%)
-3.727, -466.7%
-4.866, -152.9%
-2.639 -22.6%

Note: Shaded cells indicate statistically significant differences between the two alternatives. The tan and blue colors indicate extreme values to the right and left columns respectively.

The results shown in this table for lane-change conflicts are inconclusive. For high volumes, there is a definite trend for the SPUI to increase the number of lane-change conflicts.

Table 45. Case 6 Comparison Results for Average Surrogate Measures of Safety .
  TDIA TSPUI CDIA CSPUI REDIA RESPUI LCDIA LCSPUI
TTC (low)
1.36
1.32
N/A
N/A
1.41
1.39
1.26
1.28
t -value, diff(%)
2.687, 2.94%
N/A
1.542
-0.730
TTC (med)
1.36
1.33
N/A
N/A
1.4
1.41
1.27
1.25
t -value, diff(%)
2.721, 2.21%
N/A
-1.544
0.881
TTC (high)
1.35
1.32
N/A
N/A
1.37
1.38
1.31
1.24
t -value, diff(%)
3.523, 2.22%
N/A
-1.401
3.955, 5.34%
PET(low)
2.61
2.23
N/A
N/A
2.83
2.6
2.17
2
t -value, diff(%)
6.209, 14.56%
N/A
3.081, 8.13%
1.744
PET(med)
2.7
2.37
N/A
N/A
2.88
2.68
2.27
2.06
t -value, diff(%)
7.078, 12.22%
N/A
3.903, 6.94% 2.465, 9.25%
PET(high)
3.01
2.59
N/A
N/A
3.13
3.04
2.76
2.03
t -value, diff(%)
10.373, 13.95% N/A 2.265, 2.88% 9.440, 26.45%
MaxS(low)
7.33
8.1
N/A
N/A
6.9
6.65
8.16
9.01
t -value, diff(%)
-3.035, -10.50%
N/A
1.046
-2.095, -10.42%
MaxS(med)
7.51
7.64
N/A
N/A
7.1
6.55
8.51
8.7
t -value, diff(%)
-0.796
N/A
3.986, 7.75%
-0.618
MaxS(high)
7.4
7.67
N/A
N/A
7.15
7.1
7.99
8.39
t -value, diff(%)
-2.597, -3.65%
N/A
0.476
-1.945
DeltaS(low)
5.45
6.15
N/A
N/A
5.34
5.85
5.63
6.34
t -value, diff(%)
-8.186, -12.84%
N/A
-5.411, -9.55%
-5.171, -12.61%
DeltaS(med)
5.03
5.89
N/A
N/A
4.97
5.59
5.17
6.17
t -value, diff(%)
-12.617, -17.10%
N/A
-7.585, -12.47%
-8.841, -19.34%
DeltaS(high)
3.11
3.93
N/A
N/A
2.91
3.56
3.55
4.39
t -value, diff(%)
-10.447, -26.37%
N/A
-6.433, -22.34% -6.850, -23.66%
DR(low)
-2.94
-3.11
N/A
N/A
-2.76
-2.87
-3.31
-3.25
t -value, diff(%)
2.460, -5.78% N/A 2.322, -3.99%
-0.439
DR(med)
-2.83
-3.02
N/A
N/A
-2.68
-2.78
-3.2
-3.26
t -value, diff(%)
3.580, -6.71% N/A 3.173, -3.73% 0.509
DR(high)
-2.03
-2.41
N/A
N/A
-2
-2.13
-2.1
-2.77
t -value, diff(%)
6.323, -18.72%
N/A
2.621, -6.50% 5.122, -31.90%
MaxD(low)
-3.25
-3.39
N/A
N/A
-2.95
-3.01
-3.83
-3.62
t -value, diff(%)
1.746
N/A
0.977
-1.360
MaxD(med)
-3.25
-3.4
N/A
N/A
-2.97
-2.97
-3.92
-3.82
t -value, diff(%)
2.494, -4.62%
N/A
0.000
-0.772
MaxD(high)
-3.27
-3.76
N/A
N/A
-2.88
-2.91
-4.18
-4.83
t -value, diff(%)
6.896, -14.98%
N/A
0.607
4.366, -15.55%
MaxDeltaV(low)
2.86
3.21
N/A
N/A
2.81
3.05
2.94
3.3
t -value, diff(%)
-7.722, -12.24%
N/A
-4.511, -8.54%
-5.071, -12.24%
MaxDeltaV(med)
2.63
3.09
N/A
N/A
2.6
2.94
2.72
3.24
t -value, diff(%)
-11.532, -17.49%
N/A
-7.386, -13.08%
-8.384, -19.12%
MaxDeltaV(high)
1.63
2.06
N/A
N/A
1.52
1.86
1.86
2.31
t -value, diff(%)
-10.406, -26.38%
N/A
-6.411, -22.37%
-6.99, -24.19%

Note: Shaded cells indicate statistically significant differences between the two alternatives. The tan and blue colors indicate extreme values to the right and left columns respectively.

This table shows that the average values of surrogate measures of safety (in terms of average severity per conflict) are consistently worse for the SPUI design than the three-phase diamond interchange.

Correlations with Predicted Crash Frequency

The predicted crash rates for all scenarios in this test are in table 46 with the corresponding average number of conflict events of each type. Rank orders for each category of data are also listed in the table. The Spearman rank correlation coefficients are calculated for each test and listed in the right-most column of the table.

Table 46. Case 6 Spearman Rank Correlations Between Conflicts and Crash Frequency.
AADT
Low
Medium
High
Rs
DIA SPUI DIA SPUI DIA SPUI
Crash Frequency M
4.8
5.5
7.3
7.9
14.1
12.8
1
R
1
2
3
4
6
5
Total Conflict M
79.6
35
143.1
64.1
140.7
117.2
0.43
R
3
1
6
2
5
4
Crossing Conflict M
 
 
 
 
 
 
N/A
R
 
 
 
 
 
 
Rear-End Conflict M
53
13.4
100.9
31.2
98.5
65.3
0.43
R
3
1
6
2
5
4
LC Conflict M
26.3
21.6
42.1
32.4
42
51.5
0.69
R
1
1
5
3
4
6

Note: Averages that are not significantly different are assigned the same rank. Rows labeled "M" provide mean values and rows labeled "R" provide the ranking of each alternative. The Rs column provides Spearman rank correlation coefficients indicating agreement with theoretical crash estimates.

Findings and Conclusions

Based on the observation on the safety surrogate data obtained from the test, the following conclusions can be drawn:

  • SPUI shows significant reductions in total and rear-end conflicts compared to the conventional diamond interchange at most levels of volume. This finding is perhaps explained by the reduction from two stopping points to a single stopping point for arterial traffic crossing through the interchange and by the known correlation of vehicle stops with rear-end crashes.
  • Crossing events are rare for either intersection design.
  • Almost all average values of surrogate measures of safety indicate that events that occur at SPUI intersections will be more severe than those at diamond interchanges. However, given that rear-end events are generally less severe that and the SPUI featured a dramatic reduction in rear-end events, the increase in average severity measures per conflict is consistent and does not conclusively suggest that SPUIs are more dangerous.
  • Correlation of the conflict averages for all volume levels is not consistent with crash prediction models

In general, these results indicate that a SPUI-intersection design will reduce the total number of conflicts that occur (primarily the rear-end conflicts), but when conflicts do occur, they will be more severe on average.

Case 7: Cross Four-Leg Intersection Versus Offset T-Intersection (VISSIM)

This comparison is between an unsignalized four-leg intersection and an offset T-intersection. According to Vogt's theory in the development of the crash prediction models for two-lane roads, T-intersections with an obtuse angle from the minor road were found to have fewer crashes than four-legged rural intersections. (21) From the results of recent studies by Bared, on an aggregate level, the expected benefit of converting a cross intersection to an offset T-intersection is a reduction in total crashes of 20 percent to 30 percent for rural 2 x 2-lane, two-way, stop-controlled intersections.(14) The reduction in fatal/injury crashes is expected to be approximately 40 percent for 2 x 2-lane intersections. One reason for this, as shown in figure 84 , is the noticeable reduction in conflict points (from 32 to 22) by converting a cross intersection to an offset T-intersection.

Figure 84. Illustration. Potential Conflict Points for 2 x 2-Lane Intersections. This is an illustration of potential conflicts for 2 x 2-lane intersections. There is a noticeable reduction in conflict points (from 32 to 22) by converting a cross intersection to an offset T intersection.

Figure 84. Illustration. Potential Conflict Points for 2 x 2-Lane Intersections.(14)

Intersection Description

The intersections used for this test are shown in figure 85 and figure 86, respectively. All left-turn bays are 76.25 m (250 ft) long.

Figure 85. Screen Capture. Conventional Nonsignalized Intersection. This is a screen capture of a nonsignalized intersection model in VISSIM. All left-turn bays are 76.25 m (250 ft) long.

Figure 85. Screen Capture. Conventional Nonsignalized Intersection.

Figure 86. Screen Capture. Offset T Intersection. This is a screen capture of an offset T intersection model in VISSIM. All left-turn bays are 76.25 m (250 ft) long.

Figure 86. Screen Capture. Offset T-Intersection.

Table 47 shows traffic volumes applied for each approach of the intersection. No traffic control is used in this test.

Table 47. Case 7 Service Flow by Each Approach.
Approach
Southbound
Northbound
Eastbound
Westbound
L
TH
R
L
TH
R
L1
TH
R
L2
TH
R
Low Volumes
25
50
25
25
50
25
60
180
60
60
180
60
Medium Volumes
50
100
50
50
100
50
100
300
100
100
300
100
High Volumes
88
175
88
88
175
88
140
420
140
140
420
140

Note: L, TH, and R correspond to vehicles proceeding left, through, or right at the intersection.

Data Analysis and Comparison Results

Ten replications were performed for each simulation scenario, and the resulting output trajectory data were analyzed by SSAM. The F-test and t-test were applied to compare the average number of conflict events of each event type and surrogate measures of safety between the two intersection designs. Table 48 through table 51 list the values of all surrogate measures of safety and corresponding t-test results for different types of aggregations with the low-speed events and crash data excluded (TTC ? 0 and MaxS = 16.1 km/h (10 mi/h)).

Table 48. Case 7 Comparison Results for Total Conflicts.
Total
TTC = 0.5 TTC = 1.0 TTC = 1.5
 
CON OFFT CON OFFT CON OFFT
Medium Volume Mean
N/A
N/A
N/A
N/A
3.5
5.7
Variance
N/A
N/A
N/A
N/A
3.167
7.122
t-value(95%), difference (%)
N/A
N/A
-2.169, -62.86%
High Volume Mean
N/A
N/A
N/A
N/A
4.4
8.7
Variance
N/A
N/A
N/A
N/A
2.933
7.122
t-value(95%), difference (%)
N/A
N/A
-4.288, -97.73%

Note: CON indicates conventional intersection cross and OFFT indicates an offset T-intersection. Shaded cells indicate statistically significant differences between the two alternatives.

This table indicates that the total conflicts are increased with the offset T, although as shown in the following tables, those conflicts are comprised of lane changes and rear ends only.

Table 49. Case 7 Comparison Results for Rear-End Conflicts.
Rear End
TTC = 0.5 TTC = 1.0 TTC = 1.5
  CON OFFT CON OFFT CON OFFT
Low Volume Mean
N/A
N/A
N/A
N/A
N/A
N/A
Variance
N/A
N/A
N/A
N/A
N/A
N/A
t-value(95%), difference (%)
N/A
N/A
N/A
Medium Volume Mean
N/A
N/A
N/A
N/A
1.3
1.9
Variance
N/A
N/A
N/A
N/A
2.900
1.433
t-value(95%), difference (%)
N/A
N/A
-0.911
High Volume Mean
N/A
N/A
N/A
N/A
1.4
3.9
Variance
N/A
N/A
N/A
N/A
1.600
1.878
t-value(95%), difference (%)
N/A
N/A
-4.239, -178.57%

Note: Shaded cells indicate statistically significant differences between the two alternatives.

Table 50. Case 7 Comparison Results for Lane-Change Conflicts.
LC
TTC = 0.5 TTC = 1.0 TTC = 1.5
CON OFFT CON OFFT CON OFFT
Medium Volume Mean
N/A
N/A
N/A
N/A
1.8
3.7
Variance
N/A
N/A
N/A
N/A
1.956
4.678
t-value(95%), difference (%)
N/A
N/A
-2.333, -105.56%
High Volume Mean
N/A
N/A
N/A
N/A
2.7
4.6
Variance
N/A
N/A
N/A
N/A
2.233
3.156
t-value(95%), difference (%)
N/A
N/A
-2.588, -70.37%

Note: Shaded cells indicate statistically significant differences between the two alternatives.

Table 51. Case 7 Comparison Results for Average Surrogate Measures of Safety .
  TCON TOFFT CCON COFFT RECON REOFFT LCCON LCOFFT
TTC (med)
1.37
1.45
N/A
N/A
1.4
1.42
1.41
1.46
t-value, diff(%)
-2.203, -5.84%
N/A
-0.556
-1.774
TTC (high)
1.45
1.45
N/A
N/A
1.42
1.46
1.46
1.45
t-value, diff(%)
0.000
N/A
-1.497
0.000
PET(med)
2.34
2.89
N/A
N/A
2.85
3.17
2.32
2.71
t-value, diff(%)
-2.457, -23.50%
N/A
-0.960
-1.377
PET(high)
2.92
2.86
N/A
N/A
2.88
3.04
2.97
2.73
t-value, diff(%)
0.468
N/A
-0.780
1.417
MaxS(med)
7.37
6.01
N/A
N/A
7.46
6.58
7.07
5.74
t-value, diff(%)
2.626, 18.45%
N/A
1.088
1.670
MaxS(high)
5.88
6.13
N/A
N/A
6.33
6.52
5.71
5.81
t-value, diff(%)
-1.204
N/A
-0.378
-0.995
DeltaS(med)
6.13
5.16
N/A
N/A
5.83
4.03
5.93
5.74
t-value, diff(%)
2.713, 15.8%
N/A
3.481, 30.9%
0.493
DeltaS(high)
5.38
5.53
N/A
N/A
4.76
5.17
5.71
5.81
t-value, diff(%)
-0.680
N/A
-0.764
-0.995
DR(med)
-3.12
-2.81
N/A
N/A
-2.88
-2.64
-3.2
-2.89
t-value, diff(%)
-1.370
N/A
-1.222
-1.048
DR(high)
-2.75
-2.77
N/A
N/A
-2.49
-2.62
-2.9
-2.89
t-value, diff(%)
0.230
N/A
0.611
-0.323
MaxD(med)
-3.63
-3.37
N/A
N/A
-3.19
-3.22
-3.92
-3.46
t-value, diff(%)
-1.036
N/A
0.131
-1.375
MaxD(high)
-3.22
-3.28
N/A
N/A
-3.04
-3.16
-3.28
-3.4
t-value, diff(%)
0.467
N/A
0.550
0.724
MaxDeltaV(med)
3.17
2.64
N/A
N/A
3.06
2.05
3.05
2.93
t-value, diff(%)
2.915, 16.7%
N/A
3.763, 33.0%
0.598
MaxDeltaV(high)
2.81
2.87
N/A
N/A
2.46
2.65
3
3.06
t-value, diff(%)
-0.491
N/A
-0.686
-0.679

Note: Shaded cells indicate statistically significant differences between the two alternatives.

This table indicates that for the statistically significant measures, the offset T design results in a reduction in both the crash probability and the severity for all conflict types.

Correlations with Predicted Crash Frequency

The predicted crash rates for all scenarios in this test are in table 52 with the corresponding surrogate measures of safety. Rank orders for each category of data are also listed in the table. The Spearman rank correlation coefficients are calculated for each test.

Table 52. Case 7 Spearman Rank Correlations Between Conflicts and Crash Frequency.
AADT
Medium
High
Rs
CON OFFT CON OFFT
Crash Frequency M
2.9
2.0
5.0
3.4
1
R
2
1
4
3
Total Conflict M
3.5
5.7
4.4
8.7
0
R
1
3
2
4
Crossing Conflict M
 
 
 
 
N/A
R
 
 
 
 
Rear-End Conflict M
1.3
1.9
1.4
3.9
0.7
R
1
1
3
4
LC Conflict M
1.8
3.7
2.7
4.6
0
R
1
3
2
4

Note: Averages that are not significantly different are assigned the same rank. Rows labeled "M" provide mean values and rows labeled "R" provide the ranking of each alternative. The Rs column provides Spearman rank correlation coefficients indicating agreement with theoretical crash estimates.

Since the offset T increases the number of conflicts that occur in the simulation model, the correlation with the reduction in crashes that is expected with an offset T is very poor.

Findings and Conclusions

Based on the above observations, the following conclusions can be drawn:

  • The number of total conflicts for both cases are low because of the low traffic volumes, which are typical for uncontrolled intersections.
  • The conventional intersection generates fewer total and lane-change conflicts than the offset T-intersection.
  • The offset T-intersection exhibits lower average values for all severity and probability of crash surrogate measures.

In general, the conventional intersection shows fewer total and lane-change conflicts than the offset T-intersection. The conflicts that occur at the offset T, however, have lower average values of surrogate measures of severity and probability of collisions.

Previous | Table of Contents | Next

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration