Modelling Speeds of Arterial Weaving Sections in Metro Manila ATRF 2009 Auckland, New Zealand September 29, 2009 Ronald John Galiza School of Civil Engineering University of Queensland
Modelling Speeds of Arterial Weaving Sections in Metro Manila
ATRF 2009ATRF 2009Auckland, New ZealandSeptember 29, 2009
Ronald John GalizaSchool of Civil EngineeringUniversity of Queensland
Outline
• Problem Motivation
• Review of Existing Methods
• Survey Methodology
2
• Survey Methodology
• Results and Discussions
• Conclusions
• Recommendations
Signalized Intersections (before)
3
After Closure DLT from Arterial
4
•Left-turn from signalized intersection
•Left-turn diversion to U-turn plus right-turn (UTRT)
Through from Side Street
5
•Through movement at signalized intersection
•Through diversion to right-turn plus U-turn plus right-turn (RTUTRT)
Left-Turn from Side Street
6
•Left-turn movement at signalized intersection
•Left-turn diversion to right-turn plus U-turn (RTUT)
Weaving Areas Formed
7
Definition of a Weaving Area
�crossing of two or more traffic streams
traveling in the same general direction
without the aid of traffic control devices
(HCM 2000)
8
(HCM 2000)
Weaving Section in an Arterial
Exit Road (U-turn)
� Freeway 2-sided
weaving section…A
B
C
D
9
A
B
C
D
Minor Road (Driveway)
Exit Road (U-turn)
�The HCM Type C (2-sided) weaving configuration best describes arterial weaving…
Arterial weaving section… �
B
Capacity at Weaving Section
Flow Rate, q
Density, k
skuq =
10
Density, k
Speed, u
�Densities and speeds are difficult to determine w/o proper equipment.
Messer & Bonneson’s Ramp Weave Model (NCHRP Web Doc 12)
11
Ramp weave on arterial cross streets in interchange areas
Ramp Weave Model’s Weaving Path
C
�Path B – C
12
B
�Path B – C
Ramp Weave Model’s Non-weaving Paths
C
D
�Path A – C
�Path A – D
13
A
B
D �Path A – D
�Path B – D
Ramp Weave Models)3600/)1(( 21 wU VPbb
aow eUbU−−
=Weaving Speed:
Uw - average speed of weaving vehicles, m/s
Ua - average speed of entering vehicles, m/s
14
Ua - average speed of entering vehicles, m/s
Pu - probability of a weaving vehicle being unblocked
Vw - weaving flow rate, vph
21 ,, bbbo - regression coefficients
Ramp Weave Models
Non-weaving Speed:)3600/(
354 aVbb
anw eUbU−
=
Unw - average speed of non-weaving vehicles, m/s
Ua - average speed of entering vehicles, m/s
15
Ua - average speed of entering vehicles, m/s
Va - arterial flow rate, vph
543 ,, bbb - regression coefficients
�The weaving and non-weaving speed models need to be calibrated to get coefficients.
Survey Methodology
• Data collected using video recording stationed high vantage point;
• Recorded videos converted into digital
16
• Recorded videos converted into digital format;
• Data were extracted by repeated viewing of the recording on a computer.
Site Selection
• Must have three or more lanes in each direction
• Traffic volume on the driveway should be relatively high
17
be relatively high
• Weaving length should be less than 305 m
• Presence of signalized intersections within the vicinity should be minimal
Site 1 (Banaue – Quezon Ave)
18
Site 1: Banaue –Quezon Ave
Site 2 (Scout Borromeo-Quezon Ave.)
19
Site 2: Scout Borromeo –Quezon Ave
Model SummaryRegression coefficients
Site 1 Site 2
Weaving Model
bo 3.1140 1.8363
b1 0.4675 0.5418
b2 4.7944 0.6163
Weaving & Nonweaving Speed Models
(Site 1 vs Site 2 Model)
4.000
6.000
8.000
10.000
12.000
Sp
ee
d,
m/s
Site 2 - W
Site 1 - W
Site 2 - NW
Site 1 - NW
Ua = 10.0 m/s
SL = 1700 pcphpl
20
)3600/)1(( 21 wU VPbb
aow eUbU−−
=Weaving Speed:
Non-weaving Speed:)3600/(
354 aVbb
anw eUbU−
=
b2 4.7944 0.6163
non-weaving
Model
b3 3.0603 4.2738
b4 0.5261 0.4253
b5 0.3339 0.5568
0.000
2.000
Va20
050
080
011
0014
0017
0020
0023
0026
0029
00
Arterial Flow Rate, pcph
Sp
ee
d,
m/s
SL = 1700 pcphpl
Nt = 4
Vw = 350 pcph
lL = 0
Site 1 and 2 Calibration Results
Site 1 manoeuvre speed model statistics
Maneuver Speed
Model
Observations Root Mean Square Error R2
Weaving 35 0.490 m/s 0.44997 LARGE
21
non-weaving 35 0.447 m/s 0.54191 VERY LARGE
Site 2 manoeuvre speed model statistics
Maneuver Speed
Model
Observations Root Mean Square Error R2
Weaving 18 0.444 m/s 0.44019
non-weaving 18 0.425 m/s 0.66295
LARGE
VERY LARGE
Site 1 Model Testing
Parity Plot Site 1 Weaving (Testing)
y = 0.4516x + 3.2301
R2 = 0.45017.00
8.00
Pre
dic
ted
, m
/s
Parity Plot Site 1 Nonweaving (Testing)
y = 0.5335x + 3.5985
R2 = 0.5428.000
9.000
Pre
dic
ted
, m
/s
22
4.00
5.00
6.00
4.00 5.00 6.00 7.00 8.00
Observed, m/s
Pre
dic
ted
, m
/s
5.000
6.000
7.000
5.000 6.000 7.000 8.000 9.000
Observed, m/s
Pre
dic
ted
, m
/s
Site 1 Validation
Parity Plots (Validation Data)
y = 0.4006x + 3.4678
R2 = 0.43997.000
8.000
Pre
dic
ted
, m
/s
Parity Plot Site 1 Nonw eaving (Validation)
y = 0.4948x + 3.9681
R2 = 0.4718.000
9.000
Pre
dic
ted
, m
/s
23
4.000
5.000
6.000
4.000 5.000 6.000 7.000 8.000
Observed, m/s
Pre
dic
ted
, m
/s
5.000
6.000
7.000
5.000 6.000 7.000 8.000 9.000
Observed, m/s
Pre
dic
ted
, m
/s
Site 1 Summary
Model Process
Maneuver
Type Comparison Correlation t-test Goodness-of-fit
Site 1
TestingWeaving Obs - Pred OK
OK LARGE
Non-
weaving Obs - Pred
OK OK VERY LARGE
24
Site 1 weaving Obs - Pred
Validation Weaving Obs - Pred OK OK LARGE
Non-
weaving Obs - Pred
OK OK LARGE
Site 2 Model Testing
Parity Plot Site 2 Weaving (Testing)
y = 0.4303x + 3.4524
R2 = 0.4405
6.000
6.500
7.000
7.500
8.000
Pre
dic
ted
, m
/s
Parity Plot Site 2 Nonweaving (Testing)
y = 0.6613x + 2.7438
R2 = 0.6629
8.000
9.000
10.000
Pre
dic
ted
, m
/s
25
4.000
4.500
5.000
5.500
6.000
4.000 5.000 6.000 7.000 8.000
Observed, m/s
Pre
dic
ted
, m
/s
5.000
6.000
7.000
5.000 6.000 7.000 8.000 9.000 10.000
Observed, m/s
Pre
dic
ted
, m
/s
Site 2 Validation
Parity Plot Site 2 Weaving (Validation)
y = 0.3938x + 3.6832
R2 = 0.61997.000
8.000
Pre
dic
ted
, m
/s
Parity Plot Site 2 Nonweaving (Validation)
y = 0.3834x + 4.9606
R2 = 0.3956
7.000
8.000
9.000
10.000
Pre
dic
ted
, m
/s
26
5.000
6.000
5.000 6.000 7.000 8.000
Observed, m/s
Pre
dic
ted
, m
/s
5.000
6.000
7.000
5.000 6.000 7.000 8.000 9.000 10.000
Observed, m/s
Pre
dic
ted
, m
/s
Site 2 Summary
Model Process Maneuver Type Comparison Correlation t-test Goodness-of-fit
Site 2
Testing Weaving Obs - Pred OK OK LARGE
non-weaving Obs - Pred OK OK VERY LARGE
Validation Weaving Obs - Pred OK OK VERY LARGE
non-weaving Obs - Pred OK OK LARGE
27
Conclusions
�Ramp Weave Models
• Arterial and weaving flow rates affect weaving and non-weaving speeds.
• Weaving and non-weaving speeds decrease
28
• Weaving and non-weaving speeds decrease as the arterial flow rate increases.
• Weaving and non-weaving speeds decrease with increased weaving flow rates.
• Weaving speeds are lower than non-weaving speeds.
Conclusions
• Approach speeds are higher than non-weaving speeds.
• Weaving length significantly affects the weaving speed.
29
weaving speed.
• Shorter weaving length yields lower weaving speeds.
• The models developed can accurately predict weaving and non-weaving speeds.
Recommendations
• More extensive data collection from other sites.
• Incorporate weaving length in the
30
• Incorporate weaving length in the models.
• Use computer simulation to evaluate the speeds.
• Identify and develop a LOS criteria.
�Questions
�Comments
That ends the presentation… Thank you for listening…
31
�Comments
�Discussion
Click to see first ever elevated U-turn slot in the Philippines/World?