quang le june 25, 2012 performance evolution of the diverging diamond interchange cal poly pomona
TRANSCRIPT
Quang LeJune 25, 2012
PERFORMANCE EVOLUTION OF THE
DIVERGING DIAMOND INTERCHANGE
Cal Poly Pomona
Project ObjectiveDDI BackgroundProject MethodologyResultsConclusionsFurther ResearchQuestions
OVERVIEW
Project Objectives
Determine the behavior of several performance measures, such as delay, stop time, average speed, and queue length as the spacing between the two crossovers is increased or decreased.
Compare several performance measures using different volumes scenarios under fixed distances
Run models under a different signal timing parameter (not part of original proposal)
DDI OVERVIEW
The diverging diamond interchange (DDI), also known as the double crossover diamond (DCD) interchange, provides an alternative design solution to mitigate traffi c congestion.
Allows Crossover of traffi c to left side of the road to reduce traffi c conflicts
1ST DDI in Versailles, France (1970)
DDI IN THE UNITED STATES
Eleven (11) DDI can be found in the USMissouri (5), Utah (4), Tennessee (1), and Kentucky (1).
The first DDI opened to traffic in the United States was the interchange connecting MO-13 with I-44 in Springfield, Missouri on June 21, 2009.
DDI at MO-13 and I-44
ADVANTAGES
Vehicle Safety Ex: The average number of reported traffi c accidents
before the DDI was fifty-three. Reduced to twenty-five in the one year after the DDI was constructed and open to the public (Missouri DOT).
Pedestrian safety is increased as accidents resulting from turning movements are eliminated.
The DDI has been shown to increase the capacity of the system.
CONFLICT POINTS IN A DDI
Type Diamond SPUI DDI
Diverging 10 8 8
Merging 10 8 8
Crossing 10 8 2
Total 30 24 18
Reduction of total delay in the system. The DDI reduces the number of signals that are required in the system The delay is decreased because the two interchanges
are reduced to a two phase signal. This creates shorter cycle lengths and allows for the loss time to be saved from having fewer signal phases, which can be transferred to the green time.
Construction costs in a DDI have proven to be more economical than other conversions (Missouri DOT 2010).
ADVANTAGES
DISADVANTAGES
Lack of Sample Size (first one built in 2009) Traffi c restrictions at the off-ramps to either a
left or right turn Public Perception: General public, engineers,
political The DDI is designed to work with heavy left
turning volumes. Under free-flowing traffi c conditions along the corridor, this design will become redundant, and actually increase the delay of the intersection with confl icts between opposing direction of travel.
Limited resources to reference when designing a new diverging diamond interchange. As a result, design criteria, such as signal timing and signal warrants, level of service criteria have not been generalized for the DDI.
PROJECT METHODOLOGY
Creation of initial model at L=550’ crossover spacing under low traffi c volumes using same traffi c parameters (same signal timing).
Calibrate model within 10% accuracy of the Bared paper.
Create model and simulate results for the remaining traffi c volumes at L=550’ to obtain results as in the paper.
Increase crossover spacing to L=1,100ft, 1,650ft, and 275ft to develop further models.
Evaluate data output from VISSIM.Alter signal timing to obtain better
performance results in each DDI and compare the results
INITIAL HYPOTHESES
Currently, it can be deduced that larger interchange spacing between the on ramps and off ramps will be able to accommodate a larger traffi c volumes in the system, resulting in fewer overall delays. Developed during the literature review stage.
DIVERGING DIAMOND INTERCHANGE
Three lanes of traffi c in each direction of the cross streets. The right most lane merges into the freeway onramps, and
the interchange becomes a two lane DDI between the two crossovers.
The roadway then remains two lanes leaving the DDI on the cross streets.
The northbound and southbound both have one lane for right turn traffi c and two lanes for left turn traffi c.
Geometrics
LocationSpacing (ft)
Utah1Main and I-15 8502500 East and I-15 9003Timpanogos Highway and I-15 6504Highway 201 and Bangerter Rd. 800
Tennessee5US 129 and Bessemer St. 650
Kentucky
6Harrodsburg Road and New Circle Road 700Missouri
7I435 and Front Street 4508US 65 and Missouri 248 7509I44 and MO-13 550
10I270 and Dorsett Ave 50011Mo 60 and National Ave 700
CURRENT DDI SPACING
Avg~680ft
DDI SPACING SCENARIOS
L=550 ft
L=1,100 ft
L=1,650 ft
L=275 ft
Data taken from: “Design and Operational Performance of a Double Crossover Intersection and Diverging Diamond Interchange” by Joe Bared
Read only fi le from Bared with Input values
DATA DESCRIPTION
Northbound Off
Ramp (vph)Southbound Off
Ramp (vph)Eastbound (vph) Westbound (vph) Total Flow
Traffic Scenario L T R L T R L T R L T R (vph)High 2 700 0 400 700 0 400 400 800 500 400 800 500 5600High 1 650 0 350 650 0 350 350 750 450 350 750 450 5100
Medium 400 0 200 400 0 200 200 500 300 200 500 300 3200Low 200 0 100 200 0 100 100 300 150 100 300 150 1700
TRAFFIC VOLUMES
TRAFFIC VOLUMES
MICROSCOPIC SIMULATIONS
VISSIM is a microscopic, time step and behavior-based simulation model Each entity is simulated individually Macro: averages simulation
Microscopic simulation is a safer, less expensive, and faster than field implementation and testing.
Provides the ability to control factors not easily measured in the field.
Cannot obtain field data for this project
VISSIM Developed by PTV from Germany Diffi culties modeling DDI with other software, such as
SYNCHRO
MODEL CONSTRAINTS
Existing Geometry / Lane configurationTraffi c parameters from Bared Study were used to
obtain similar resultsAdjacent Intersections/ramps (focus on DDI network)No Pedestrian traffi c were modeledOne signal timing from Bared Study was used
Model Calibration/Verification
Model calibration is defined as the process by which the individual components of the simulation model are adjusted or fine tuned so the model will accurately represent field measured or observed traffi c conditions.
Usually performed with real life data, in this case, data generated from the Bared Study.
Goal: 10% of the MOEs as determined in the Bared Study. Delay Stop Times
Traffic Scenario
Model Delay
Base Delay Low
Base Delay
Base Delay High
% Error Low
% Error % Error High
550 Low 18.5 16.5 17 17.4 89.19 91.89 94.05550 Median 21.8 19.5 20 20.4 89.45 91.74 93.58550 High 1 32.7 31.5 32 32.4 96.33 97.86 99.08550 High 2 41.5 39.5 40 41.4 95.18 96.39 99.76
COMPARISON OF BASE MODEL (L=550 FT) WITH BARED PAPER
MODEL PARAMETERS
Signal TimingDriver Behavior SettingsRoute AssignmentsRouting DecisionsAreas of Reduced SpeedPriority RulesTravel Time LocationsQueue Counter Locations
LOCATION OF TRAFFIC SIGNALS
8 Total Signals and 6 phases were used.
1
12
53 4
46
Phase 2=Phase 5Phase 3=Phase 6
Signal Timing Diagram from Bared Study
Model from the Bared StudyYell=4 secAll Red =1 sec
INITIAL SIGNAL TIMING PARAMETERS
UPGRADED SIGNAL TIMING PARAMETERS FOR L=1100 FEET
SCENARIOS18 second diff erence between signals
UPGRADED SIGNAL TIMING PARAMETERS FOR L=1650 FEET
SCENARIOS28 second diff erence between signals
RESULTS
ScenarioTotal Vehicles
(Initial)275Low 1524275Med 2880
275High1 4908275High2 5244550Low 1572550Med 2880
550High1 4908550High2 55441100Low 14881100Med 28801100High
1 49081100High
2 53821650Low 15721650Med 28801650High
1 49081650High
2 5382
MODEL THROUGHPUT
DELAY RESULTS (BARED TIMING)
Traffic Scenario
Delay (sec/veh)
)
Stop Time
(sec/veh)
# Stop
s
%Diff Delay
%Diff Stop Time
% Diff #Stop
s275Low 19.8 14.8 0.74 275Med 23 15.8 0.87 16.16 6.76 17.57
275High1 34.9 22.1 1.21 76.26 49.32 63.51
275High2 47.8 31 1.6141.4
1109.46
116.22550Low 18.5 13.1 0.71 550Med 21.8 14.3 0.82 17.84 9.16 15.49
550High1 32.7 19.9 1.05 76.76 51.91 47.89
550High2 41.5 24.3 1.3124.3
285.50
83.101100Low 20.8 15 0.81 1100Med 24 15.4 0.86 15.38 2.67 6.17
1100High1 31 17.4 1.01 49.04 16.00 24.691100High2 35.6 19.7 1.11 71.15 31.33 37.041650Low 19.9 13.7 0.85 1650Med 21.8 13.5 0.82 9.55 -1.46 -3.53
1650High1 28.7 15.8 0.95 44.22 15.33 11.761650High2 31.9 17.2 1.01 60.30 25.55 18.82
Consistent trend that delay increases with increasing traffi c volumes and is reduced as crossover spacing is increased
Same trend for number of stops and stop timesA greater percent delay diff erence is found in lower
crossover spacing and higher volume scenariosShows that the DDI is more sensitive under these
conditions
DELAY RESULTS (BARED TIMING)
DELAY RESULTS (IMPROVED SIGNAL TIMING)
Traffic Scenario
Delay (sec/ve
h)
Stop Time
(sec/veh)# Stops
%Diff Delay
%Diff Stop Time
% Diff #Stop
s275Low 19.8 14.8 0.74 275Med 23 15.8 0.87 16.16 6.76 17.57
275High1 34.9 22.1 1.21 76.26 49.32 63.51275High2 47.8 31 1.6 141.41 109.46 116.22550Low 18.5 13.1 0.71 550Med 21.8 14.3 0.82 17.84 9.16 15.49
550High1 32.7 19.9 1.05 76.76 51.91 47.89550High2 41.5 24.3 1.3 124.32 85.50 83.101100Low 17.6 12.1 0.63 1100Med 18.4 11.6 0.62 4.55 -4.13 -1.59
1100High1 28.2 16.5 0.85 60.23 36.36 34.921100High2 34 18.8 0.97 93.18 55.37 53.971650Low 17.3 11.9 0.67 1650Med 18.7 11.6 0.67 8.09 -2.52 0.00
1650High1 26.3 14.9 0.82 52.02 25.21 22.391650High2 30.9 16.9 0.91 78.61 42.02 35.82
Updated signal timing shows a greater delay savings overall
Improved overall performance, Same behavior where percent diff erences are higher
for lower cross over spacing
DELAY RESULTS (IMPROVED SIGNAL TIMING)
DIFFERENCE IN DELAY BETWEEN TWO SIGNAL TIMING
PARAMETERS
Traffic Scenario
Delay (sec/veh)
Stop Time (sec/veh)
# Stops
% Delay Savings
1100Low 3.2 2.9 0.18 15.38
1100Med 5.6 3.8 0.24 23.33
1100High1 2.8 0.9 0.16 9.03
1100High2 1.6 0.9 0.14 4.49
1650Low 2.6 1.8 0.18 13.07
1650Med 3.1 1.9 0.15 14.22
1650High1 2.4 0.9 0.13 8.36
1650High2 1 0.3 0.1 3.13
Greater delay savings shown for lower crossover spacing.
HCM DELAY FOR SIGNALIZED INTERSECTIONS
LOS Delay (seconds)
A 0-10
B 10-20
C 20-35
D 35-55
E 55-80
F >80
• Applying values to DDI are over exaggerated, but gives a starting point for a comparative analysis.
• HCM does not have dedicated table for DDI, used for relative comparison purposes only.
Movement Delay (HCM)
Left from Corridor
Avg.Left from Fwy
Avg.
Traffic Scenario Delay LOS Delay LOS
275Low 20.3 C 29.6 C
275Med 23.3 C 34.95 C
275High1 33.9 C 65 E
275High2 38.75 D 101.9 F
550Low 23.7 C 24.25 C
550Med 29.8 C 28.6 C
550High1 51.8 D 34.75 C
550High2 62.9 E 45.9 D
1100Low 21.1 C 21.5 C
1100Med 23.55 C 24.2 C
1100High1 34.6 C 41.3 D
1100High2 43.55 D 47.2 D
1650Low 17.9 B 24.7 C
1650Med 22.2 C 28.8 C
1650High1 37.25 D 30 C
1650High2 46.55 D 32.85 C
DELAY VS. TRAFFIC VOLUME (BARED SIGNAL TIMING)
Delay vs Traffic Volume (L=275ft)
0
10
20
30
40
50
60
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(se
c)
Delay
Stop Time
Delay vs Traffic Volume (L=550ft)
05
1015202530354045
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(se
c)
Delay
Stop Time
DELAY VS. TRAFFIC VOLUME (BARED SIGNAL TIMING)
Delay vs Traffic Volume (L=1100ft)
0
5
10
15
20
25
30
35
40
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(se
c)
Delay
Stop Time
Delay vs Traffic Volume (L=1650ft)
0
5
10
15
20
25
30
35
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(s
ec
)
Delay
StopTime
DELAY VS crossover DISTANCES (BARED SIGNAL TIMING)
Delay vs Cross Over Distance (Low)
0
5
10
15
20
25
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
Delay vs Cross Over Distance (Med)
0
5
10
15
20
25
30
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
DELAY VS crossover DISTANCES (BARED SIGNAL TIMING)
Delay vs Cross Over Distance (High1)
0
5
10
15
20
25
30
35
40
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
Delay vs Cross Over Distance (High2)
0
10
20
30
40
50
60
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
DELAY VS. TRAFFIC VOLUME (MPROVED SIGNAL TIMING)
Delay vs Traffic Volume (L=1100ft)
0
5
10
15
20
25
30
35
40
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(se
c)
Delay
Stop Time
Delay vs Traffic Volume (L=1650ft)
0
5
10
15
20
25
30
35
275Low 275Med 275High1 275High2
Traffic Volume
De
lay
(se
c)
Delay
Stop Time
DELAY VS crossover DISTANCES (IMPROVED SIGNAL TIMING)
Delay vs Cross Over Distance (Low)
0
5
10
15
20
25
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
Delay vs Cross Over Distance (Med)
0
5
10
15
20
25
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
DELAY VS crossover DISTANCES (IMPROVED SIGNAL TIMING)
Delay vs Cross Over Distance (High1)
0
5
10
15
20
25
30
35
40
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
Delay vs Cross Over Distance (High2)
0
10
20
30
40
50
60
0 200 400 600 800 1000 1200 1400 1600 1800
Cross Over Distance (Ft)
De
lay
(se
c)
Delay
Stop Time
DELAY SUMMARY
Delays consistently increase as volumes increase in each scenario being modeled.
Delays consistently increase as crossover spacing is decreased.
Greater percent diff erence in delays and stop times as crossover is decreased and volumes is increased Sensitivity of the models.
Using an updated signal timing increases the delay savings.
Isolating the delays for left turning movements allows us to examine performance of key turning movements.
Same results were found for stop time and number of stops
DDI shows poor performance in the L=275 scenario, best as cross over spacing is increased.
Traffic ScenarioAverage Speed
(MPH)
% Speed Diff.
275Low 16.354 275Med 15.005 -8.25
275High1 12.171 -25.58275High2 9.676 -40.83550Low 15.737 550Med 14.718 -6.48
550High1 11.937 -24.15550High2 10.352 -34.221100Low 17.826 1100Med 16.975 -4.77
1100High1 15.128 -15.141100High2 14.082 -21.001650Low 20.711 1650Med 19.995 -3.46
1650High1 18.003 -13.081650High2 17.187 -17.02
AVERAGE SPEED (BARED SIGNAL TIMING)
Similar behavior as delay analysis.
Scenario Average Speed (MPH) % Speed Diff.
1100Low 19.234
1100Med 18.436 -4.15
1100High1 15.852 -17.58
1100High2 14.41 -25.08
1650Low 21.438
1650Med 20.739 -3.26
1650High1 18.446 -13.96
1650High2 17.407 -18.80
AVERAGE SPEED (IMPROVED SIGNAL TIMING)
Overall higher average speeds are found.
Traffic ScenarioSpeed Difference
(mph) % Difference
1100Low 1.408 7.32
1100Med 1.461 7.92
1100High1 0.724 4.57
1100High2 0.328 2.28
1650Low 0.727 3.39
1650Med 0.744 3.59
1650High1 0.443 2.4
1650High2 0.22 1.26
DIFFERENCES OF SPEED IN TWO TIMINGS
Average Speed Summary
Average speed consistently decreases as volumes are increased.
Average speed consistently increases as crossover spacing is increased.
Similar results as delay concerning increased sensitivity at higher volumes and lower crossover spacing.
SIMULATION OF MODEL AT L=275 FEET AND VOLUME=HIGH2
PROJECT SUMMARY
DDI exhibits better performance in average speed, delay, and stop time in large crossover spacing
DDI exhibits better performance in average speed, delay, and stop time under lower traffi c volumes
No significant advantages at lower traffi c volumesPerformance decreases as a faster rate as volume is
increased at lower crossover spacingOptimizing the signal timing on the main corridor
serves to significantly increase network performanceRecommended spacing not be less than ~550ft
DDI begins to show excessive delays
FUTURE RESEARCH
More specifi c traffi c patterns that focus on special cases where individual travel movements govern the system.
Compare individual approaches and traffi c movements, rather than the performance of the entire network.
The task of updated the signal timing was limited in this study as to maintain signal timing as a constant in throughout the sixteen models.
Research may be performed to compare the results of a 4-lane DDI with a 6-lane DDI.
Pedestrian performance. They were omitted in this study as adding pedestrians to the system will have no impact to the vehicular traffi c as long as the signal timing parameters for the pedestrians coincides with that of the vehicles.
Further analysis on LOS requirements for DDI.
ACKNOWLEDGEMENTS
Xudong Jia, CSU PomonaSyed Raza, CSU Pomona Joe Bared, FHWA
QUESTIONS?