Estimating the Impacts of Access Management
with Micro-simulation: Lessons Learned
Bill Eisele
Bill Frawley
Casey Toycen, Roelof Engelbrecht, Kristin
Turner, Anna Griffin Martin
Sponsored by:
6th National Conference on Access Management
8-30-2004
Objectives
„
„
„
„
Investigate the impacts of raised median
installation and driveway density by traffic
volume
‹ 3 case studies
‹ 3 theoretical corridors
Investigate the use of a surrogate safety
measure in micro-simulation
‹ Time-to-collision
Selecting a micro-simulation tool for
access management alternatives analysis
‹ Input and output characteristics
Concluding thoughts and future work
Methodology
„
„
„
„
„
3 case studies and 3 theoretical corridors
Three simulation runs of each traffic volume
‹ Each run provides a random estimate of the measure
‹ Analyzed peak hour
‹ Maintain O-D patterns
Reduction in conflict points, travel time, speed and delay were
analyzed
‹ Travel time and speed of vehicles traversing the corridor
‹ System delay
Used VISSIM model
Also ran time-to-collision analysis within VISSIM (1 run)
‹ Proof-of-concept
Three Case Studies (Characteristics)
Corridor
Length
Location (miles)
Signals per
Mile /
Access
Points per
Mile
Median
Opening
Spacing
(feet)
Number of
Lanes Each
Direction
Bryan,
Texas
0.55
3.0 / 91
690 to 1,320 2
Temple,
Texas
0.71
5.6 / 66
350 to 850
Tyler,
Texas
1.47
4.1 / 46
500 to 1,500 3
„
Median opening spacing for selected alternative
2
Texas Case Study (Results)
Location
Percent
Difference
in Conflict
Points
Bryan,
Texas
-60
Temple,
Texas
-56
Tyler,
Texas
-60
„
Estimated
Existing
ADT
Future
Estimated Percent
Difference in
Future
Travel Time
ADT
Future
Actual
Difference in
Speed (mph)
Large reduction in conflict points with raised median installation
Texas Case Study (Results)
Percent
Difference
in Conflict
Points
Estimated
Existing
ADT
Future
Estimated Percent
Difference in
Future
Travel Time
ADT
Bryan,
Texas
-60
18,200
21,800
Temple,
Texas
-56
13,300
16,000
Tyler,
Texas
-60
24,400
29,300
Location
„
„
48,000
48,000
Lower ADT is “existing +20%”
Higher volume selected for further analysis
Future
Actual
Difference in
Speed (mph)
Texas Case Study (Results)
Percent
Difference
in Conflict
Points
Bryan,
Texas
-60
Temple,
Texas
-56
Tyler,
Texas
-60
Location
„
„
Estimated
Existing
ADT
Future
Estimated Percent
Difference in
Future
Travel Time
ADT
Future
Actual
Difference in
Speed (mph)
18,200
21,800
-11
4 (increase)
48,000
-38
11 (increase)
13,300
16,000
3
1 (decrease)
24,400
29,300
2
<1 (decrease)
48,000
57
4 (decrease)
Case-specific results
Function of traffic patterns, median opening locations, etc.
Bryan, Texas Speed Results
Texas Avenue (Speed Results)
40
36
35
33
28
30
Speed (mph)
32
28
25
20
17
15
10
5
0
ADT~18,200
ADT ~21,800
Approximate ADT
ADT ~48,000
TWLTL
Raised Median
Tyler, Texas Speed Results
Broadway Avenue (Speed Results)
18
16
16
TWLTL
14 14
Speed (mph)
14
12
11 11 11
Raised Median (U-turns
at Signals)
Raised Median (U-turns
at Signals and mid-block)
11
10
7 7
8
6
4
2
0
ADT ~24,000 ADT ~29,000 ADT ~48,000
Approximate ADT
Theoretical Scenarios (Characteristics)
Driveway
Number of Spacing
Driveways (feet)
Raised
Median
Opening
Spacing
(feet)
2
18
660
660
2
42
330
660
3
42
330
660
3
84
165
660
Number
of Lanes
Theoretical Median
in Each
Corridor
Treatment Direction
Scenario 1
Scenario 2
TWLTL
and
Raised
TWLTL
Raised
TWLTL
Scenario 2
Scenario 3
Raised
TWLTL
Raised
„
ITE trip generation, driveways across from each other
Theoretical
Corridor
Estimated
Future ADT
Scenario 1
18,000 to 28,000
18,000
Scenario 2
(2 lanes)
23,000
28,000
18,000
Scenario 2
(3 lanes)
23,000
28,000
48,000
18,000
Scenario 3
(3 lanes and
higher
driveway
density)
23,000
28,000
33,000
38,000
48,000
Future Percent
Difference in Travel
Time
Future Actual
Difference in
Speed (mph)
Future Actual
Difference in
Speed (mph)
Theoretical
Corridor
Estimated
Future ADT
Future Percent
Difference in Travel
Time
Scenario 1
18,000 to 28,000
Not Applicable
Not Applicable
18,000
2
0 (same)
23,000
6
2 (decrease)
28,000
31
7 (decrease)
18,000
7
2 (decrease)
23,000
8
3 (decrease)
28,000
11
3 (decrease)
48,000
44
9 (decrease)
18,000
6
2 (decrease)
23,000
1
1 (decrease)
28,000
2
0 (same)
33,000
6
2 (decrease)
38,000
23
6 (decrease)
48,000
10
2 (decrease)
Scenario 2
(2 lanes)
Scenario 2
(3 lanes)
Scenario 3
(3 lanes and
higher
driveway
density)
Conflict and Safety Analysis
„
„
„
„
„
„
„
Crash data suspect, incomplete, or
unavailable
Micro-simulation as a tool
Distance between vehicles divided by
speed difference is time-to-collision (TTC)
Rear-end only evaluated here (others later)
TTC threshold of 4 and 10 seconds
Number of assumptions—real drivers not
“perfect”
Relative safety performance of alternatives
‹ Access, medians, signals….
Time-to-collision Illustration
Speed1
(mph)
Speed2
(mph)
D
(feet)
D x 3,600
Time-to-collision =
(seconds)
(Speed1 – Speed2) x 5,280
Example:
Time-to-collision
=
(seconds)
100 feet x 3,600 seconds/hour
(30 mph – 15 mph) x 5,280 feet/mile
= 4.5 seconds
Conflict and Safety Analysis
TWLTL generally lower harmonic mean
(as expected for increased conflict
points)…..
Texas Avenue Harmonic Mean of TTC
TTC Harmonic Mean
(Seconds)
60
52
50
40
50
44
38 36
34
TWLTL
30
Raised Median
20
10
0
ADT ~ 17,400
ADT ~20,900
ADT Level
ADT ~ 48000
…which equates to higher proportion of
vehicle time at ≤ 10 seconds for the
TWLTL alternative.
Texas Avenue Proportion of Vehicle Time with
TTC <=10 Seconds
Percentage
10.0
8.9
7.7 7.4
8.0
6.0
6.2
4.9
5.0
TWLTL
Raised Median
4.0
2.0
0.0
ADT ~ 17,400
ADT ~20,900
ADT Level
ADT ~ 48000
Similar result on Tyler corridor…
Broadway Avenue
Proportion of Vehicle Time with TTC <= 10 Seconds
Percentage
TWLTL
10.0%
9.0%
Raised Median (U-turns at
Signalized)
8.0%
7.0%
6.0%
5.0%
Raised Med (U-turns at
Signalized and mid-block)
4.0%
3.0%
2.0%
9.6%
8.2%
7.7%
6.7%
6.1% 5.8%
3.0% 2.9% 3.1%
1.0%
0.0%
ADT ~24,000
ADT ~29,000
ADT Level
ADT ~48,000
…and on the theoretical corridors.
Percentage
Scenario 2 and Scenario 3
(3 Lanes Each Direction)
Proportion of Vehicle Time with
TTC <= 10 Seconds
14.00%
Scenario 2, TWLTL
12.00%
Scenario 2, Raised
Median
Scenario 3, TWLTL
10.00%
8.00%
Scenario 3, Raised
Median
6.00%
4.00%
2.00%
0.00%
ADT ~18,000 ADT ~23,000 ADT ~28,000 ADT ~48,000
ADT Level
Selecting a Micro-simulation Tool
„
Input characteristics
‹ Complex situations, individual drivers
‹ Geometric inputs
 Scale, auxiliary lanes, turning radii, lane
width
‹ Operational inputs
 Gap acceptance, speeds, accelerations
 Signal optimization
‹ Calibration (evaluate default values)
‹ O-D, and other underlying theory
Selecting a Micro-simulation Tool
„
Output characteristics
‹ Analysis at the individual vehicle level
‹ Spatial
 By location—intersections, median openings
‹ Temporal
 Over time—platooning, queuing
‹ Animation features
 Visual consistency
 3-dimensions
Concluding Thoughts
„
„
„
„
Results from raised median
installation are case-specific
Caution should be used when
generalizing AM impacts across
corridors
Function of traffic volumes,
driveway density, weaving (o-d
patterns), median opening location
and density, decel lane length,
signal coordination, speed
distribution, driver behavior, etc.
Micro-simulation allows detailed
corridor analysis
Concluding Thoughts
„
„
„
Relatively small increases in travel time are
likely offset by the well-documented increase in
safety
‹ NCHRP 395, NCHRP 420
‹ Bill Frawley’s talk tomorrow
TTC appears to be a promising method for
assessing safety in the micro-simulation
environment
‹ Indexed ranking of alternatives
Must coordinate access management analysis
needs with micro-simulation tool
Future Work
More runs that vary median opening
number, median opening spacing and
location, driveway density, traffic
volume, decel lane length, etc. to
populate larger matrix
„ Expansion of TTC to angle crashes
„ Additional TTC runs
„
Contact Information
Bill Eisele, Ph.D., P.E.
979/845-8550
[email protected]
Bill Frawley, AICP
817/462-0533
[email protected]
Can’t get enough of this?!?....
See poster at the break!
Full paper on CD (or contact me)!