|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 177
APPENDIX B
FIELD STUDY DATA COLLECTION AND REDUCTION
Several analytic models were developed during this study to facilitate the evaluation of
interchange ramp teIIIiinal capacity and level of service. This appendix provides a description of the
database used to calibrate these models. Specifically, it describes the composition of the database,
the field study sites, the methods used to collect the data, and finally, some summary statistics of the
reduced data. Subsequent appendices describe the development and calibration of the analytic
models as well as Weir application to the interchange evaluation process.
B.1 DATABASE COMPOSITION
This section describes the traffic flow problems associated with interchange areas, as they
relate to the objectives ofthis research. The problems of primary interest are those occuning on the
arsenal cross street at or between Me interchange ramp terminals and any adjacent, closely-spaced
intersections. Initially, these flow problems are described in the context of models needed to
describe Me problem's elect on arterial performance. Then, the variables included in these models
are identified in the context of defining a data collection plan.
B.~.1 Traffic Flow Problems Associated With Interchange Ramp Terminals
The findings from the survey of practitioners (see Appendix A) indicated Mat there were
several types of traffic flow problems associated with signalized interchange ramp terminals. The
impact of these flow problems was typically amplified by relatively short distances (as measured
along Me arsenal cross street) between Me terminals or between a tenninal and an adjacent signalized
intersection. These flow problems were broadly categonzed as: (~) midblock turbulence (i.e.,
weaving) and unbalanced lane volumes Mat stem from high-volume turn movements in the
interchange vicinity; and (2) flow restriction or impediment to discharging queues due to a relatively
near downstream traffic queue. Four models were developed to facilitate Me evaluation of these
flow problems. The vanables included in these models were used to identify the data needed for
mode! calibration. The four models include:
1. Capacity Model. This mode! quantifies the effect of downstream traffic conditions on the
traffic characteristics used to estimate the capacity of lefr-tum and through movements at
interchange ramp terminals arid adjacent intersections. These characteristics include start-up lost
time, saturation flow rate, arid clearance lost time. The capacity of an upstream signal phase has
been found to be adversely affected by Me close proximity of a downstream queue; particularly
when Me queue spills back into He upstream intersection. The form of this mode} is described
in Appendix C.
2. Approach Lane Utilization Model. This model quantifies Me extent of unbalanced lane use
in multi-lane larle groups. On a cycle-by-cycle basis, marry drivers in the interchange area tend
B-!
OCR for page 178
to use one lane of a multi-lar~e lane group more than the others; they rarely choose the lane wad
the fewest vehicles in it. One possible reason for this clustering in interchange areas may be
driver desire to 'preposition" for a downstream turn. In this situation, drivers in interchange
areas position their vehicles in the appropriate inside (or outside) lane at an upstream ramp
terminal or adjacent signalized intersection in anticipation of a turn at the next terminal or
intersection. The frequency of this behavior is increased when turn volumes are high or the
distance between terminals is too short for "comfortable" lane changing. The form of the lane
utilization mode! is descnbed in Appendix C.
3. Queue Length Model. This model can be used to convert a predicted queue length from the
number of equivalent passenger cars to units of distance (e.g., meters). This queue length
conversion model was found be an essential component of He capacity model. This model was
also incorporated into Be queue interaction model developed for this research, as descnbed in
Appendix D.
4. Arterial Weaving Motlel. This mode! quantifies the effect of weaving activity on the efficiency
Of arsenal traffic flow. The weaving maneuver that is predominate in interchange areas is the
off-ramp right-turn movement that weaves across the arsenal to make a left-turn at the
downstream signalized intersection. This maneuver has been observed to cause significant
turbulence In He arsenal traffic flow resulting In significant increases In travel time and, In some
cases, lengthy queues on the off-ramp. The form of this model is described In Appendix E.
B.~.2 Database Elements
The data needed to calibrate He aforementioned models can be categorized as: (~) basic
traffic characteristics (model inputs), (2) traffic performance measures (model outputs), (3) signal
controller settings, (4) traffic control features, and (5) geometric data. The elements that comprise
the first two categories are dynamic and were collected continuously during the field studies. These
elements are listed in Table Be-.
The latter three categories represent static data types. They were often measured prior to the
start of the study. Elements of each of these three categories are listed below.
· Signal Controller Settings. This category includes the traffic signal controller settings and
operation. In particular, cycle times, coordination onsets, signal phase sequence, change
interval, and phase splits were collected during each study. If the controller had one or more
actuated phases, then the individual actuated phase durations were recorded with a computer
connected to the signal controller.
Traffic Control Features. This category includes speed limit, traffic control signs, and
pavement markings. This information was of a general nature and was used to describe the
character of the interchange area.
B-2
OCR for page 179
Table By-. Traff~c-related database elements
Mo' lel
Category ~ Data Type use ~ Lane ~ Queue
. Capacity Utilization Length
Traffic Discharge Headway Cat
haractenstics | Discharge Speed | C l l l
Start-Up Lost Time C
Count & Location of Cars on Be Downstream
Link at He 5~ of Offs''
Queued Driver ' Wing Reaction Time ~| | V
| Queued Vehicle Storage Length l l | V l
Traffic Demands by Lane and Movement
| TravelPa~Mat ix(O-D's) | | V I ;
Weave & Non-Weave Vehicle Travel Tune
| Weave & Non-1 'cave Flow Rate per Lane l l l
Performance Saturation Flow Rate C
.
Men cures | Lane Utilization Percentage l l V |
Queue Length V ~
| Weaving Vehicl Speed l l l l u
Notes:
1 - Data collection method: C - computer-monitored tape switches; V - video tape records.
Weaving
V
V
V
V
· Geometric Data. This category includes geometric infonnation along the arsenal and at each
intersection. Artenal information includes cross section, distance between intersections, turn bay
lengths, and lane assignments. Intersection information includes approach grade, skew angle,
and turn radii.
B.2 STUDY SITE DESCRIPTION
B.2.l Study Site Characteristics
A list of desirable characteristics for the field study sites was prepared based on information
obtained from the survey of practitioners and the insights obtained while formulating the
aforementioned models. These site characteristics are described in the following paragraphs.
Interchange Types. The study sites were selected to collectively include Me two basic
forms of service interchange commonly used in suburban and urban areas: the diamond and the
partial cloverleaf (or parclo) interchanges. Variations of these two interchange forms stem from
variations in the distance between We ramp terminals and the routing of the left and right-turn
movements (i.e., Trough the signal or via a loop ramp). Further assessment of the correlation
between interchange type, the extent of its operational problems, and its frequency of application in
B-3
OCR for page 180
urban areas led to the following six interchange types being identified as the most appropriate
candidates for the field studies:
Diamond Interchange
I. Compressed Diamond
2. Tight Urban Diamond (without frontage roads)
3. Tight Urban Diamond twin frontage roads)
4. Single Point Urban Diamond
Partial Cloverleaf (Parclo)
5. Parclo B (2-quad)
6. Parclo AB (2-quad)
Study Types. The development of We data collection procedure was strongly influenced by:
(~)theneedto optimize the qualify end quantify oftraffic-related characteristics (see Table B-~), and
(2) the capabilities of the computer and video data collection equipment available to the study team.
Based on a thorough examination of all feasible collection procedures, it was concluded that two
different types of field study were needed.
One type of study required the equipment to be deployed for the purpose of collecting the
capacity and lane utilization model data. The data needs of these models required the concurrent
study of one direction of travel along two successive, arterial street segments. These segments
included the section between the two ramp terminals and the section between the ramp terminal and
adjacent intersection. This study type was referred to as a Capacity Study. A typical data collection
setup for this study is shown in Figure B-1 . This figure shows the setup for the diamond interchange
form, a similar setup was used for the parclo form.
As shown In Figure B-1, there are two possible study "cases" at an interchange. These cases
are named the "downstream" and "upstream" cases. The natne of each case identifies the orientation
of the adjacent intersection with respect to the travel direction studied. The field studies were
designed to include a mixture of both cases. This approach permitted an examination of the effects
of the adjacent intersection on bow interchange inflow and outflow.
The second type of study required the equipment to be deployed for the purpose of collecting
the weaving data. This study focused on the off-ramp right-turn movement weaving across the
through traffic to make a left-turn at the adjacent, downstream intersection. To collect this data, the
equipment was deployed at each end of the arterial weaving segment. This study type was referred
to as a Weaving Stucly. A typical data collection setup for this study is shown in Figure B-2.
Queue length arid starting reaction time data, needed for the Queue Length Model, were
collected with both study types. The data needs of this model required only a view of the front and
rear of the traffic queue. As this type of view was generally available from either study type, a third
study type was not developed to collect this data.
c' ~
B-4
OCR for page 181
Downstream Case
Zip
U pstrea m Case
Camera 2
\\ ~Boundary of/
I Study Zone
~'1 NY ~
Camera ~17 .
_ _ _
_ ~l ~ ~/ ~ A. . it::
Boundary of /
Study Zone
LEGEN D
"< ond field of view ,~ Tape Switch Sensor
7 Photocell Sensor ~ Tape Switch Speed Trap
-
Figure B-1. Capacity study data collection setup for a diamond interchange.
B-S
. l l Camera ~
... ... ................ ~
. _ _ i., ·.-.- _
-- ~ ~ ~ r
- =_ ~
· ^.B. : ~
. = . . ._ . .
..~' 'I
. . .~;¢,.. A.
.......... =-
OCR for page 182
an/
~ - -
~ video Camaro
\2 ~
LEGEND
Comer
Bounds of/
Study Zone f
~e -. If afar ~ co~ deft.
B-6
~ ,
_ ^
_ , Hi.
. ~ ~ .
Comes 2
OCR for page 183
The road segments and corresponding traffic movements considered dunng each study are
shown (by number or letter) in Figures B-l and B-2. These figures also show the locations of the
tape switch sensors, photocell sensors, and video cameras. The sensors were monitored by a
computer, which also served as the data processing and recording device. This computer processed
the sensor input and automatically converted it into more desirable fonns such as phase duration and
discharge headway. The video cameras were used to record Me vehicle locations on the downstream
segments, travel paths, and weaving/non-weav~ng vehicle characteristics. The actual data were
manually extracted from videotapes dunog playback in the laboratory, subsequent to the field
studies, and processed Into more usefid forms at that time. The specific types of data collected by
each collection system are listed In Table By-.
Geometric and Traffic Demand Criteria. The selection of specific field study sites (i.e.,
interchanges) was based on their degree of compliance with the following geometric and traffic
demand criteria.
Geometric Cntena
I. Cross Section:
2. Ramp to Ramp Distance:
3. Ramp to Intersection Distance:
4. Adjacent Land Access:
Traffic Demand Criteria
I. Arterial AADT:
2. Ramp Terminal v/c Ratio:
Traffic Control Cr~tena
I. Arterial Speed Limit: 48 to 80 km/in
2. Ramp & Intersection Signalization: pretimed or sem~-actuated/coordinated
3. Ramp & Intersection Cycle Length: preferably the same at both
4. Artenal Left-Turn Phasing: protected
5.
4 or 6 through traffic lanes
60 to 275 meters (stop line to stop line)
60 to 275 meters (stop line to stop line)
no parking and preferably no driveways on arterial
20,000 or more
0.7 to I.0 during peak hours
Arterial Left-Turn Phasing:
Arterial alignment:
~.... . ..
· . -
preferably less than 2° horizontal curvature, less than 2%
grade, and negligible skew
In audition to these cntena, the study sites had to have frequent and recharting traffic queues on We
arterial during the peak traffic periods. The intent of these criteria was to insure that We database
would be representative of Interchanges in urban areas with fairly typical geometries. The
requirement of "hequent and recuTnug queues" was a recognized deviation Tom the characterization
of "being representative" and "Wpical;" however it was a necessary extension as it produced the
_ . . ~ , ., .
· ~ ~ . · . · ~ ~ · ~ ~ ~ . . e . ~ - - ~ _ _
number or observations necessary to yield meaningful arid statistically soured models.
B.2.2 Study Site Locations
In addition to Me aforementioned criteria, there was a need for geographic diversity in the
collective list of study sites. Study sites were identified in six geographic regions ofthe U.S. These
regions included Me Northwest, Southwest, Upper Midwest, Lower Midwest, Northeast, and
B-7
OCR for page 184
Southeast. Within these regions, highway agencies in Me states with large metropolitan areas were
contacted and inquiry was made as to potential study locations. Interchanges that most nearly
complied with the desired criteria were identified as candidates for a preliminary site visit.
Based on We results of We preliminary visit to the candidate sites, twelve interchanges were
identified as being most suitable for field study. Every effort was made to identify two interchanges
for each of the six types identified in a previous section; however, this goal could not be achieved
In some instances. Table B-2 descnbes We distribution of the twelve study sites, as categonzed by
nterchar~ge type arid study location.
Table B-2. Interchanges studied by tone and location
~,, ,,, _
l _
l Interchange | Study Location l
Type Nebraska Arizona Texas Kansas
Compressed Diamond 1 2
right Urban Diamond (no frontage) | | 1
Tight Urban Diamond (Wow homage)
Single Point Urban Diamond 7 T ~
RarcloB (2-quad) l l l I _
Rarclo AB (2-quad) ~I T I I =
I Total: 1 2 1 3 1 2 1 ~ ~- .1
2
3
California
Total
- 1 1
3
1
3
2
2
1
12
The traffic and geometric characteristics of each site are listed in Table B-3. In general, the
study sites satisfied almost all ofthe geometric and traffic demand criteria previously described. In
a few instances, the distance to the adjacent Intersection exceeded the desired 275-meter maximum
distance; however, the traffic demands at these sites were sufficiently high as to precipitate the
extensive queuing considered desirable for study purposes.
A capacity study was conducted at each of the twelve study sites. In addition, weaving
studies were conducted at six of the sites. All total, eighteen studies were conducted in eight cities
and five states.
The traffic signal characteristics of each study site are listed in Table B-4. It is interesting
to note that a few of the interchanges are not coordinated with the adjacent intersection. The reason
for this lack of coordination is different in each case. The Peona Road site is not coordinated
because it is currently standard practice in Arizona for the state DOT to operate the interchanges and
the city to operate the adjacent intersections. The Towneast Boulevard site is not coordinated due
to a lack of funds for coordination hardware. The Stevenson Boulevard site is not coordinated
because the existing coordination hardware failed in service and resources were not available to
replace it.
B-8
.
OCR for page 185
Table B-3. Traffic and geometric character~sffcs of the shady sites
Ramp to Ramp to
Interchange Arterial City, Arterial Arterial Ram p Intersection Speed
Type State AADT Thru Distance Distance Limit
l ~ | ~ Lanes ~ (meters)I ~ (meters~l ~ (l =~
Compressed |MetcalfAve |Overland Perk, | 58,600 | 6 | 200 | 204 | 72 ||
Diamond 1110~ to I-435 IKansac l l l I I 1
|75th Street 1 Overland Park, | 32,000 1 4 174 1 155 1 56
|I-35toFronta ;e |Kansas l l l I I 4
|Maple Street 1 Omaha, 1 34,200 1 4 268 1 198 1 72
102nd to I-680 Nebraska
Fight Urban |Peoria Road 1 Phoenix, | 34,400 | 6 107 | 276 | 64 ||
diamond |25~Ave. to I 17 1 Arizona I I l l I 41
|MathildaAv |Sunnyvale, 1 34,540 1 6 1 87 1 110 1 72 11
| SR-237 to Rot ~| California l l l I I :1
Fexas |Arapaho Roa |Richardson, | 39,000 | 6 | 99 | 265 | 64 ||
Diamond US75 to Greenville Texas
.. .
|Towneast Blv 1 Mesquite, 1 35,000 1 6 137 1 223 1 56
I Emponum to ] -635 I Texas l l l l l
arclo AB |60th Street 1 Omaha, 1 31,800 1 4 259 1 216 1 64 11
2 quad) |I-80to Grover INebraska l l l I I ~1
_ _
Parclo B Somersville Rd Antioch, 39,700 4 265 119 56
2 quad) | Delta Fair to S Rat | California l l l I I 41
Istevenson Bh d 1 Newark, | 55,600 | 4 264 | 157 | 56 ||
|Balentine to I- ;80 |California l l l I I 4
. ingle Point |7th Street |Phoenix, 1 42,000 1 6 1 78 1 331 1 56
1 Jrban |I-10toMcDo,rell IArizona l l l I I 1
Diamond
(SPUI) Indian School Rd Phoenix, 54,500 6 91 316 56
16th St. to SR-51 Arizona
Notes:
1
Distance measured from stop line to stop line in the same d~rection, except at SPUI's. At SPUI's, ~e "same
direction" concept is also applied but the opposing direction through stop line is used as the reference point at the
second ramp terminal (since the Trough stop line at the second ramp terminal does not exist at the SPUI).
B-9
OCR for page 186
Table B-4. Signal charactenstics of the study sites
. ,
Interchange | Adjacent Intersection
Interchange Arterial Signal Control Signal Control
Type _
Arterial No. of Control Cycle Control Coordi
Lett-Turn control- Type ~Length2 Type ~nation3
Protection lers (see)
Compressed Metcalf Ave Protected 2 SA 126,1 05 SA C
D~amond
75~ S~eet Protected 1 SA 82, 90 SA C
~ ~ Maple Street TProtlPenn ~1 T SA ~ 90T SA ~
~ ight Urban TPeoria Road ~Prot./Pe~m. ~ 1 ~ FA ~166T SA ~
.
amonc . _
Ma~ildla Ave Protected 2 SA100 SA C
1
Texas Arapaho Road Protected 1 SA90,120, 128 SA C
Diamond
Tou neast Blvd Prot./Pem~. 1 FA1 1 8 FA N
Parclo AB 60~ S~eet Prot./Pe~m. 2 SA90 SA C
(2 quad)
1
Parclo B Somersville Rd Protected 2 SA110 SA C
(2 quad)
Stevenson Blvd Protected 2 SA110, 100 FA N
Single Point 7~ StreetProtected 1 SA 90 SA C
Urban
diamond | Indian School~d |Protected | 1 | SA | 8, 90, 102 | SA | (
~.
Notes:
1 - FA = fillly actuated, SA = semi-actuated.
2 - Values listed are ~ose observed dunng ~e study periods. Underline values are averages.
3 - C = coordinated wi~ interchange s~al~s), N = not coordinated wi~ interchange si~al~s)
B.3 DATA COLLECTION
B.3.1 Approach
Cycle
Length2
(see)
26, 105
82, 90
90
90
100
90,120, 128
3
90
0
24
90
78,90, 102
The data collection equ~pment used to collect the field data included v~deo cameras and
computer-monitored tape switch sensors placed in the traf Eic lanes. As described in a preceding
section, the equ~pment deployment followed one of two study types (i.e., a capacity or weaving
study). Data collected dur~ng these studies is descr~bed in this section.
All data were collected dunng weekday, daytime penods between the hours of 7:00 am and
7:00 pm. The study penod generally included the hours of peak traffic demand at ~e study site.
Data were not collected dunng ~nclement weather nor dunng unusual traffic conditions (e.g., a traffic
accident).
B-10
OCR for page 187
B.3.2 Capacity and Lane Utilization Data
During each ofthe capacity studies, the equipment was deployed in a manner consistent with
that shown in Figure B-] . At some sites, the location of power lines, fences, or private property
required slight deviations from Me desired camera position. Typical camera positions are shown in
Figure B-3 for one study site. The corresponding fields of view obtained wad We video cameras are
shown in Figure B-4.
Data collected dunng the capacity studies were used to calibrate the capacity and lane
utilization models. Data for the capacity mode} were collected using bow the computer-mon~tored
tape switches arid the video recorders at each of the twelve Interchange study sites. In general, three
traffic movements were monitored during each study. These movements included an interchange
off-ramp or arterial left-turn, an interchange arsenal through, and either the second interchange
through movement (for the downstream case) or the upstream intersection through movement (for
Me upstream case).
The tape switches were used to record traffic flow behavior in two traffic lanes for each
movement monitored. A speed trap consisting of two parallel tape switches was typically located
in the inside-most lane. This trap provided information on vehicle headway, speed, acceleration, and
wheelbase. A single tape switch was located in the adjacent lane. This single tape switch provided
additional headway information. The computer monitoring the tape switches was also connected
to Me signal controller (using photo-cell sensors) and used to monitor the signal indication status.
The video recorders were positioned to provide a visual record of traffic crossing the taDe switches
as well as a view of the downstream street segment.
4, - -- ~
Data for the large utilization mode} were collected using two video cameras and recorders at
each ofthe twelve interchange study sites. All traffic movements entenug Me study boundary were
tracked as Hey traveled Trough the study area. The data collected Included Me approach lane used
at each intersection or terminal and the travel time through the system.
B-~]
OCR for page 194
Table B-5. Capacity mode! database - format
Sample Data
field:
A _ B C D _ E F _ G H I _ ~ _ K _ L M_ N_ O_ P_ Q R S T U_ V_
K1 S 3 62 5 16 9 10.59 51.8 19 19 1 4.332 4.332 2.8 5.80 1.63 0 1 2 2 4 4 3 x x x x G 12 78.2 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 2 5.926 1.594 2.8 7.53 0.81 0 1 2 2 4 4 3 x x x x G 12 72.5 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 3 8.768 2.843 2.9 7.03 0.99 0 1 2 2 4 4 3 x x x x G 12 73.5 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 4 10.194 1.425 2.7 8.40 0.95 0 1 2 2 4 4 3 x x x x G 12 71.5 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 5 11.654 1.461 2.6 8.31 0.05 0 1 2 2 4 4 3 x x x x G 12 72.9 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 6 12.850 1.196 2.8 8.28 -0.13 0 1 2 2 4 4 3 x x x x G 12 74.2 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 7 14.085 1.235 2.7 7.78 -0.75 0 1 2 2 4 4 3 x x x x G 12 76.2 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 8 16.486 2.401 2.6 6.66 -0.01 0 1 2 2 4 4 3 x x x x G 12 78.0 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 9 18.328 1.842 2.7 7.13 -1.26 0 1 2 2 4 4 3 x x x x G 12 77.0 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 10 20.803 2.475 2.7 7.50 -1.18 0 1 2 2 4 4 3 x x x x G 12 77.4 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 11 23.695 2.892 2.6 5.17 -1.60 0 1 2 2 4 4 3 x x x x G 12 76.2 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 12 31.333 7.638 2.7 1.19 2.51 0 1 2 2 4 4 3 x x x x G 12 74.1 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 13 33.776 2.442 2.8 6.09 1.73 0 1 2 2 4 4 3 x x x x G 12 60.3 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 14 35.207 1.431 2.8 7.50 1.47 0 1 2 2 4 4 3 x x x x G 12 59.2 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 15 38.657 3.450 2.7 10.81 0.15 0 1 2 2 4 4 3 x x x x G 12 60.6 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 16 41.163 2.506 2.8 11.29 0.30 0 1 2 2 4 4 3 x x x x G 12 57.0 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 17 43.250 2.087 2.7 10.27 0.87 0 1 2 2 4 4 3 x x x x G 12 55.9 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 18 44.350 1.100 2.7 10.84 0.82 0 1 2 2 4 4 3 x x x x G 12 55.6 205
K1 S 3 62 5 16 9 10.59 51.8 19 19 19 47.462 3.113 3.5 10.48 -1.45 0 1 2 2 4 4 3 x x x x G 12 57.8 205
.
Field Description
A | Site Co e - indicates city and interchange location
B | Junctio Type - S=Intersection, C=Interchange
C | Time ol Day Code - I=first two-hour study period, 2=second..., etc.
D Movement Code - indicates movement type (i.e., left or through) and lane number
E Cycle Count - signal cycle number since the start of the study period
_
F,G,H Start of Phase - start of the subject phase green
~| Phase ~ Iration - duration ofthe phase green and yellow indications
J | Numbe in Queue - number of queued vehicles served during initial portion of phase
K | Numbe in Phase - number of vehicles served during phase (No. in queue & arrivals after queue)
L Vehicle Count - number of queued vehicles served since start of phase
M Discharge Time - time back axle of vehicle crosses stop line/tape switch relative to start of phase
N Discharge Headway -headway between subject vehicle back axle and preceding ~rehicle back axle
O Wheelbase - subject vehicle wheelbase
P Discharge Speed - subject vehicle speed when crossing the stop line/tape switch
Q Acceleration/Deceleration - subject vehicle acceleration when crossing the stop line/tape switch
R Zone Counts - number of vehicles in each 30-meter zone in downstream lane (left to right - up to
downstream) at the start of subject phase
S | Downst ~am Indication -status of downstream through signal indication (G=green/yellow, R=red)
T | Spillbac i Number - queue position experiencing spillback while crossing stop line/tape switch
U Downstream Density - density in the downstream lane at the time of subject vehicle discharge
V | Segmen Leng~b - length of the downstream street segment (stop line to stop line) , '
Notes:
1 - MT (Military Time): Field F = hours, Field G = minutes, Field H = seconds.
B-18
Units
na
na
na
na
na
MT
sec
na
na
na
sec
sec
meters
mls
mls2
na
na
na
v~
meters
OCR for page 195
B.4.3 Lane Uff~zation Mode' Database
Traffic events recorded on video tape during the capacity studies were used to create a
database of traffic characteristics arid performance measures necessary for calibrating the large
utilization model. The specific data collected included the large volume per cycle, the number of
approach traffic lanes, the distribution of traffic volumes to downstream turns, arid the type of
nterchar~ge. The lane utilization analysis focussed on Me large use of through traffic movements
served by multi-lar~e large groups.
Data reduction required the use of two video cameras to track vehicles through the study
boundaries. The two cameras were synchronized in time and the recorded unages played back
simultaneously. This synchronization facilitated the tracking of each vehicle from entry point to exit
point along the arterial. Dunng video playback Me ent~y/ex~t locations and times were recorded for
each tracked vehicle. The time required for tracking each vehicle was somewhat lengthy and
required the use of a sampling technique. Specifically, it was determined that only vehicles entering
the study boundaries during Tree five-minute periods would be tracked. These five-minute penods
were allocated to each of the three highest traffic hours studied. This approach typically yielded
data for 250 or more vehicles for each entry movement at each ofthe twelve sites.
The assembled lane utilization model database includes the entry and exit time and location
for 8,198 vehicles observed at twelve sites for 32 traffic movements. Of these vehicles, about 65
percent (or 5,292) were recorded as being a through movement on at least one of the studied
intersection approaches.
B.4.4 Queue Length Mode' Database
The data reduction procedure for the queue length database required a camera view of the
front and back of Me through movement traffic queue on an intersection approach. The front view
was used to measure the distance-to-stop-line and starting-reaction tune of the first queued driver.
The back view was used to measure the same statistics for the last queued vehicle. When the last
queued vehicle extended beyond the field of view or was too distant to precisely ascertain, a vehicle
in a lower numbered queue position, nearer to the camera, was used instead. This technique of
selecting a vehicle nearer to the camera maximized the precision of the queue length and reaction
time measurements. All distance measurements were made at the start of the phase. Queues with
trucks or motorcycles were not considered. Left-turn queues were not studied.
The assembled queue length database contains queue length and reaction time measurements
for 122 Brst-in-queue passenger cars and 1,053 last-~n-queue passenger cars. This data was obtained
at eight of Me twelve study sites. Studies were not conducted at four sites because of the lack of an
adequate view of the traffic queue.
B.4.5 Weaving Mode!
Data reduction required the use of bow camera views to track vehicles through the weaving
section. The weaving maneuver examined ~ this study was the off-ramp-right-turn-to-downs~eam
B-19
OCR for page 196
ntersection-left-turn. The two camera recordings were synchronized in time and played back
simultaneously to obtain Me travel time and stopping location of weaving and non-weav~ng vehicles.
During video playback, vehicles entering the study section were tracked as they proceeded
Tom the upstream ramp terminal to the adjacent, downstream intersection. This tracking required
recording the manner in which the vehicles entered (i.e., off-ramp right-turn or arterial through
movement) and their arrival time to various points downstream (i.e., back of queue, stop line). A
sampling technique was used to select the tracked vehicles as the lengthy tracking tune for each
~ ~ . ~ ~ ~ ~ ~ . . . . ~ ~ . ~ . ~ ~ ~
vehicle precluded the collection of a 1 (J()-percent sample. 1 his technique speckled that three through
vehicles and three on-ramp nght-turn vehicles be picked randomly at five-m~nute intervals for the
purpose of tracking. This pattern was followed during We I.S-hour period that bracketed the peak
hour at each site. After this initial sampling effort, it was determined that additional data were
needed for the weaving category. Thus, a second review of the videotape was conducted and data
for an additional 332 weaving vehicles were added to the database.
The composition of We weaving mode] database is shown ~ Table B-6. As this table
indicates, We database contains entry dines for 17,939 vehicles. Of these vehicles, 980 vehicles were
hacked Tough We study segment. About one-half of We hacked vehicles (i.e., 421 of 980) were
observed to complete a weaving marleuver.
Table B-6. Weaving mode' database - sample size
Attribute or l Study Site |
Statistic | Metcalf | 75th Stre' t | Arapaho | Towneast | 60th |
Avenue Road Blvd. Street
Interchange Type Compressed CompressedTexasTexas Parclo AB
Diamond Diamond Diamond Diamond
Total Entries3,457 2,761 3,272 3,389 2,007
Artenal Enhies ~2,722 ~1,671 ~2,195 ~2,568 ~1,432
Ramp Envies735 1083 1,077 821 575
Envies Sampled162 159 146 159 192
Weaves in Sample ~68 ~5' 1 48 1 67 1 109 L - - · --- ~
, .
B.S SUMMARY STATISTICS
B.5.1 Approach
7th Street
SPUI
3,053
2,093
960
162
Total
17,939
12,688
5,251
980
1
71
421
The model calibration activity began with a review of relevant summary statistics for each
of Me data bases. Specifically, this review included Me computation of average values for selected
traffic characteristics and performance measures (as described in Table Bob. These averages are
categorized by site, interchange type, and traffic movement where appropriate. A more detailed
B-20
OCR for page 197
examination of the database was conducted during the model calibration activities; the findings from
this examination are described in Appendix C.
B.5.2 Capacity Mode! Database Summary
As indicated in a previous section, the capacity model datable contair q ~ ~ v~riPh~ of
traffic characteristics and signal timing data. Specifically, the database contains information on
signal cycle length, phase duration, the frequency and extent of drivers entering the intersection after
the yellow indication, and the discharge headway of each queued passenger car. This last
characteristic was used to compute average saturation flow rate and start-up lost tune. Driver use
of the yellow indication was used to estimate the lost time at the end of the signal phase.
~^~- ~ ~J
In general' cycle lengths ranged Dom 78 to 166 seconds among the twelve study sites. The
sites had either sem~-actuated or fill-actuated signal control. The cycle length was found to vary by
+10 to +20 seconds at any one site. The sites with larger variations in cycle length are characterized
as having semi-actuated control, time-of-day signal timing plan selection, and frequent pearl
implementation. The transition periods between signal timing Claris at these sites tended to produce
a few, extremely long cycle lengths that led to the high variability in cycle length that was observed
at the study sites.
The average minimum discharge headway and corresponding saturation flow rate are
summarized ~ Table B-7 for He two junction and movement types studied. The mourn discharge
headway H is computed as the average of all headways observed for the fifth and higher queue
positions. The reciprocal of H. with proper unit conversion, is the saturation flow rate s (i.e., s =
3600/H, vphgpl). This technique for computing the saturation flow rate is consistent with the
procedure described in the ~ 994 Highway Capacity Manual (1, Chapter 99.
In general, the saturation flow rate is very similar among the Interchanges and intersections
studied. An examination ofthe saturation flow rates categorized by interchange type indicated that
there were no distinct differences In flow rate among types. On the other hand, the data in Table B-7
indicate that the left-turn movements may be discharging more efficiently than the through
movements at He study sites, however, the difference is relatively small.
The database was segregated into "with" and "without" spillback categories. The "with"
spillback category relates to the headways observed during signal phases that experienced queue
spillback from the downstream intersection. The headways included in this category represent only
those vehicles able to discharge before the onset of spilIback. Vehicles that discharge prior to
spillback were found to have low saturation flow rates. They had lithe incentive to discharge at
higher rates because they were essentially discharging into the back of He downstream queue.
The average start-up lost time is shown in Table B-8 for the two junction and movement
types studied. This characteristic was computed as the sum of the portion of the headway for each
ofthe first four queued passenger cars Hat was in excess ofthe rniliirnllm discharge headway. While
He magnitude of the start-up lost time is largely dependent on the minimum discharge headway, it
is also dependent on the distance between the first queued vehicle and the reference line used to
B-21
OCR for page 198
demarcate entry to Me ~ntersechon (typically this is We stop lined. To maze Me influence of this
latter factor, Me reference line location was detennined for each study site. Specifically, Me
reference Ime was established as Spout I.0 to 1.5 meters downstream of Me Font of Me first queued
vehicle. Typically, this technique resulted in the reference line being located near Me stop line.
Table B-7. Capacity mode! database - summary saturation flow statistics
Ir . - ~ - - ~ - ~
Junction
Type
interchange
Intersection
Movement
Types
Left-Turn
Through
Leh-Turn
Through
Average:
Min. Discharge Headway2
: Without Spillback (sec/veh)
No.
_ 5,285
9,292
42
5,971
20,590
l
Avg.
l
1.84
1.87
1.83
. 1.88
1.86
l
Std Dev
.
0.45
0.52
0.40- _
0.52 _
0.47
Sat. Flow
Rates
(pcphgpl)
1,957
1,925
1~967
1,915
1,935
Min. Discharge Headway4
With Spillback (sec/veh)
.
No. | Avg. .
-- 1 -
_
408 2.17
_ 20 2.22
_
_ 786 2.16
_ 1,214 ~ ~2.18
Std Dev
0.95
0.44
0.94
0.78
Sat. Flow
Rate3
(pcphgpl)
1~659
1~622
1~667
1 651
~e.
Notes:
1 - Le0-turn movements Dom bow the off-ramp and He arsenal. Through movements along He arsenal.
2 - Based on the average headway of the fit through last queued passenger car.
3 - Computed as 3,600 divided by the minimum discharge headway.
4 - Based on tibe average headway of He fit through last queued passenger car able to discharge prior to queue
Spillback Dom He downstream intersection.
"--" no data available.
Junction
Type
~ eft-T~n'
Intersection
Left-Turn
Table By-. Capacity mode! database - summary start-up lost time statistics
1. .
Movement
Types
Start-Up Lost Timer
Without Spillback (see)
Start-Up Lost Time3
With Spillback (see)
No.
Std Dev
Average
2 65
4.40
1.04
1 08
1.07
1 07
1.07
No.
108
166
1.69
3.04
1.87
2.20
Std Dev l
1.58
0.98
1.85
1.47 .
-
Notes:
1 - Leflc-turn movements from bow the off-ramp and He arterial. Through movements along He arsenal.
2 - Based on He average headway of He fifth Trough last queued passenger car.
3 - Based on He average headway of He fit Trough last queued passenger car able to discharge prior to queue
Spillback Dom He do~ms~eam intersection.
"--" no data available.
As with the headway and saturation flow rate fatal the start-up lost times shown In Table B-8
were categorized as being "with" arid "without" spiliback. In general, start-up lost time in Me
B-22
OCR for page 199
"without" spillback category tends to be higher as a consequence of the extra time lost by the
discharging traffic queue as it accelerates to the higher speeds associated with the higher saturation
flow rates. Typical values for the "without" category range from 2.46 to 2.80 seconds (excluding
the data in the "~ntersection/left-turn" category) whereas values for the "with" category range from
1.69 to 1.87 seconds. The difference between these two ranges suggests that the more normal,
"without" spillback condition is associated with about 1.0 seconds more start-up lost time than the
"with'' spillback condition. The data in Table B-8 also suggest that the start-up lost time may be
higher for a left-turn movement than a through movement.
Table B-9 summarizes Me extent of driver use of We charge interval (i.e., "green extension")
and the implications of this use on lost time at the end of the phase. The green extension reported
In this table represents Me average time after the yellow indication was presented that the last vehicle
entered Me intersection. Subtraction ofthis time from the yellow-plus-red-clearance interval yields
Me lost time at the end of the phase.
Table B-9. Capacity mode' database - summary clearance lost time statistics
l Junction ~ Movement | ~ reen extension (see) | Clearance Lost Time ksec)
~ Type ~ Typel ~i;; ~ Average ~ SldDev ~Vo. ~ Average ~
?
I ~ t e r c h a ~ g e L e ~ c - T u ~4 9 1 2 . 8 2 1 . 4 8 4 9 1 2 . 7 7
Through 675 2.65 1.46 675 2.60
ntersection ~Left-Tu~n I 'S T 2.45 1 1.52 1 58 1 2.55
Through 369 2.54 1.31 369 3.1 1
Average: 1 1,593 1 2.67 1 1.44 1 ,593 1 2.77
Std Dev
l.9g
2.15
.52
1
1.43
1.94
Notes:
1 - LeR-turn movements Mom bow Me off-ramp and Me arsenal. Through movements along Me arterial.
The frequency of green extension was also examined using the capacity database. In general,
it was fourld that drivers entered the intersection after the yellow was presented in about one-half of
the phases studied. Although the average amount of green extension is relatively constant at about
2.67 seconds across the twelve study sites, the frequency of green extension was fourld to vary with
geographic region and with cycle length. Specifically, drivers in some regions of Me country were
found to be more frequent users than in others. This finding may relate to local traffic laws
regarding use of the yellow interval and to the level of enforcement of these laws. Drivers were also
found to be more likely to enter Me intersection when the cycle length was long. Presumably, these
drivers were willing to accept the risks associated with entry on yellow in order to avoid the lengthy
delays associated with long cycle lengths.
The data in Table B-9 indicate Mat the average clearance lost time is about 2.77 seconds.
This value is within the range of I.2 to 2.8 seconds recognized by the 1994 Highway Capacity
Manual (1, Chapter 2), although it is near the upper limit. This trend is likely due to Me longer
change intervals used at some of the interchanges and intersections studied. The charge intervals
B-23
OCR for page 200
at Me study locations are likely to be longer than those at more typical intersections because of the
larger conflict area and the higher speeds associated with intersections near or at interchanges.
B.5.3 Lane Utilization Mode' Database Summary
The analysis of the lane utilization mode! database focused on the computation of lane
utilization factors for each of Me approaches studied. The utilization factor was computed for each
of Me multi-lane approaches using the following equation:
v' N
U man
~ vi'
where:
U= lane utilization factor for We lane group;
v ',~ = maximum demand flow rate in any of N lanes, vpcpl,
v, =
demand flow rate in lane i, i = I, 2, ... N. vpcpl; and
N= number of lanes in the lane group.
(Ball)
The variables for this equation were computed as follows: demand flow rate for lane i was computed
as the count of vehicles In Mat lane for a given cycle. The largest of these lane counts was recorded
as the maximum demand flow rate for the same cycle. These counts were taken for each signal cycle
that occulted during a 5-minute period and used to compute average values of vi ' and v ',~ for Mat
5-minute period. The lane coinciding with Me maximum demand flow rate was often found to vary
each cycle.
The lane utilization factors computed for the through movement lane groups at the twelve
study sites are shown in Table B-IO. As the data in this table indicate, the factors range from 1.12
to 1.72, with a strong sensitivity to the number of I=es in Me lane group. Specifically, Me factor
is lowest for the two-lane lane group and increases web increasing number of lanes. This trend
suggests Mat traffic volumes become more unbalanced as the number of available lanes increases.
Table B-IO. Lane utilization mode! database - summary staffstics1
l Number of Lanes in the Lane Group
Movement
Type 2 Lanes 3 Lanes 4 Lanes 5 Lanes
No. ~ Avg ~ Std Dev T 3 o. | Avg. | S1 d Dev7 No. ~ Avg. T Sty Dev T r 0. fAvg.
,eft-Turn ~14 ~ I.1 1 ~ 0.~8 ~ 2 ~ I.28~ 0.~ --I -. ~ ~ ~ . ,
~ tough T 651 1.1!] 009T 26t 126t 0.21] ~1.3 ~0.084 2T 1.72, ~
Std Dev
0.47
Notes:
1 - Observations represent five-minute averaging intervals.
"--" no data available.
B-24
OCR for page 201
The factors for the two and three-lane lane groups were compared with the values
recommended in the 1994 Highway Capacity Manual (1, Chapter 9J for application at isolated
intersections. These recommended values (i.e., 1.05 for two lanes and 1.10 for three lanes) confirm
the trend noted above regarding larger values for lane groups with more lanes. On the other hand,
the recommended values tend to be much lower than the values reported in Table B-10. This trend
suggests that lane utilization in interchange areas tends to be more unbalanced than at isolated
intersections. This result was anticipated because of the significant fuming activity in interchange
areas and the resultant need for drivers to preposition themselves in the left-most (or r~ght-most) lane
on the street segment prior to the segment Tom which the turn wall be made.
B.5.4 Queue Length Mode! Database Summaty
The analysis of the queue length mode] database focussed on We computation of the average
lane length occupied by a queued passenger car and the average queued driver starting reaction time.
These average values are listed In Table B-! 1 .
Table By-. Queue length model database - summary statistics
First Queued Passenger Car Subsequent Queued Passenger Car |
Variable ~ Average ~ Min. ~ Max. Average ~ Min. -
Lane length (storage), meters/car 5.0 1.6 7.5 7.0 6.8
, reaction time, seconds ~1.52 ~1.17 ~1.73 1.06 ~0.9~_
Max.
7.5
1.19
As the statistics in Table B-1 1 indicate, the average lane length occupied by Me first queued
passenger car was found to be 5.0 meters. This length was found to vary from 1.6 to 7.5 meters
among the study sites. Further examination of this variability indicates that it correlates with
differences in the location of the stop line relative to the travel path of Me nearest conflicting traffic
movement. Sites wad low average lane lengths for the first queue position tended to have stop lines
that were more distant from the intersection conflict area. As a result, drivers at these sites would
frequently stop downstream ofthe stop line, nearer to the conflict area, presumably to maze Bed
travel time to the intersection once the green indication is presented.
The average lane length occupied by the second and subsequent queued vehicles was found
to vary from 6.S to 7.5 meters among the study sites. The overall average was fourld to be 7.0 meters
per passenger car.
The overall average of 7.0 meters/car is slightly less than that reported In the literature.
Specifically, Messer and Farnbro (2) studied Me through movements at two intersections and found
averages of 7.3 and 7.7 meters/car. In an earlier study, Herman et al 63J examined Me lane length
occupied by queued vehicles using a test track facility and several full-size 1970 Chevrolet sedans.
They reported finding an average lane length of 7.9 meters/car. As both of these studies were
B-25
OCR for page 202
conducted in the 1 970's, it is likely that the shorter storage length found in this research is due to
reductions in average vehicle length during the last 20 years.
The average reaction time for the firsts queue drivers was found to vary from 1 .17 to 1 .73
seconds among the study sites. The overall average reaction time was fourth to be ~ .52 seconds. In
contrast, the average reaction time for the subsequent queued drivers was found to be ~ .06 seconds,
with a range of 0.96 to 1. 1 9 seconds among sites. This trend was expected because the first driver
has more of a Surprise situation (i.e.' the signal indication changing from red to green) than the
subsequent queued drivers who can look ahead, see that the indication is green, and anticipate their
tune of departure. As a result, the first drivers require slightly more reaction time than subsequent
queued drivers.
The average reaction times floured in this research are generally consistent with hose reported
in We literature. Specifically, the study by Messer and Fambro 629 found that the first queued driver
required about 2.0 seconds of reaction time and that subsequent queued drivers required about ~ .0
seconds. An earlier study by George and Heroy 649 at five intersections found that the first queue
driver required about I.8 seconds and that subsequent queued drivers required about I.3 seconds.
B.5.5 Weaving Mode! Database Summary
The analysis of the weaving mode] database focussed on the volume and speed of the
weaving and non-weav~ng traffic streams. These data were collected because it was hypothesized
that Me volume ofthe two conflicting streams would affect Weir individual running speeds through
Me weaving section. It was theorized that these speeds would decrease with Increasing volume. The
weaving movement that is the subject of this analysis is the off-ramp r~ght-turn movement that
weaves across the arterial to make a left-tu~n at the downstream signalmen Intersection. The average
volumes and speeds through the weaving section for the six study sites are listed in Table B-12.
Table B-12. Weaving mode! database - summary statistics
Statistic
Variable | No. | Average | Minimum | Maximum |
Volume
otal arterial volume entering weaving section, vph ~6 ~1,409 ~954
I=. v~ mends weaving section, vphpl |6 | 575 | 465
| Veavingvolume(off-r~Tnp right todownstrea~nleft),vph |6 | 151 | 100
| speed
| Arterial vein. spot speed at entry to weaving section, m/s |324 | 14.1 | 10.3
Artenal vein. running speed Trough weaving section, m/s 324 10.6 8.3
| ~rterialveh.speed reduction due to weaving activity,rn/s |324 | 3.4 | 1.4
| Weaving vein. running speed Cough weaving section, m/s 421 8.0 6.6 ,
,813
8s6
230
9.3
12.1
7.2
10 3
B-26
OCR for page 203
As the volumes in Table B-12 indicate, the six study sites have relatively high weaving
volumes. On average, the weaving vehicles accounted for about one-half of the off-ramp right-turn
volume at any one site. The arterial lane volumes were also relatively high such that weaving
opportunities were limited during a significant portion of the signal cycle. It should be noted that
the off-ramp right-turn movement at three of the sites was signalized (with right-turn on red
allowed), the over three were yield-controlled.
Two types of speed statistic were reported for the arterial vehicles. One statistic is the spot
speed of the arterial vehicles at a point just upstream of the off-ramp. The second statistic is the
running speed of the same arterial vehicles. This speed related the distance traveled through the
weaving section to the corresponding travel time. The distance and time were measured from the
point of entry to the weaving section to the downstream intersection stop line or to the first point of
joining the stopped queue associated with the downstream signal, whichever was reached first. The
running speed measured in this manner was reasoned to be the better measure of weaving vehicle
Impact because it excluded the effect of downstream signal delay on the speed estunate.
The difference between the arterial spot speed and Me running speed is an indicator of a
speed reduction in the weaving area due to weaving activity. The average speed reduction at the
study sites was 3.4 m/s. This statistic is more useful than the spot or running speeds alone because
it eliminates the effect of differing speed limits among the sites. A preliminary examination of this
speed difference indicates a strong correlation between it and the total arterial and weaving volumes.
Increases In either volume level tended to Increase the speed reduction.
We average weaving vehicle speed is also shown In Table B-12. The weaving vehicle speed
tends to be lower than that of the arterial vehicles because the weaving vehicle enters the weaving
section at a relatively slow speed due to the ramp control (i.e., signal or yield sign). Some
preliminary analysis of this speed indicates that it decreases with increasing arterial lane volume.
B.6 APPENDIX B REFERENCES
1. TAB Special Report 209: Highway Capacity Manual, 3rd ed. TRB, National Research Council,
Washington,D.C. (1994~.
2. Messer, C.J., and Fambro, D.B. "Effects of Signal Phasing and Length of Left-Turn Bay on
Capacity. In Transportation Research Record 644, TRIP National Research Council,
Washington, D.C. (1977) pp. 95-101.
3. Herman, R., Larn, T., and Rothery, R.W. "The Starting Characteristics of Automobile
Platoons." Proc., 5th International Symposium on the Theory of Traffic Flow and
Transportation, American Elsevier Publishing Co., New York (1971) pp. 1-17.
4. George, E.T., and Heroy, F.M. "Starting Response of Traffic at Signalized Intersections. -
Traffic Engineering, Institute of Transportation Engineers, Washington, D.C. (July 1966) pp.
39-43.
B-27
OCR for page 204
Representative terms from entire chapter:
queue length
