ISSN 1055- 1425
April 2010
This work was performed as part of the California PATH Program of the
University of California, in cooperation with the State of California Business,
Transportation, and Housing Agency, Department of Transportation, and the
United States Department of Transportation, Federal Highway Administration.
The contents of this report reflect the views of the authors who are responsible
for the facts and the accuracy of the data presented herein. The contents do not
necessarily reflect the official views or policies of the State of California. This
report does not constitute a standard, specification, or regulation.
Final Report for Task Order 6304
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
Weaving Analysis, Evaluation and
Refinement
UCB- ITS- PRR- 2010- 19
California PATH Research Report
Alexander Skabardonis, Amy Kim
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
ii
ACKNOWLEDGEMENTS
This work was performed by the California PATH Program at the University of California at
Berkeley, in cooperation with the State of California Business, Transportation and Housing Agency,
Department of Transportation ( Caltrans), Division of Research and Innovation ( DRI) ( under
Interagency Agreement # 65A0208, Task Order 6304). The contents of this report reflect the views of
the authors, who are responsible for the facts and the accuracy of the data presented herein. The
contents do not necessarily reflect the official views or policies of the State of California.
The authors thank Fred Yazdan contract manager of Caltrans DRI for his support and advice during
the project. We also thank the project technical advisory committee members Fred Yazdan, Tam
Nguyen, Sam Toh, Scott Eades, Rod Otto, Steve Hague, and Zhongren Wang of Caltrans for their
comments and suggestions throughout the study. Sam Toh and Scott Eades of Caltrans District 5 also
provided data from freeway weaving sites in District 5. Katherine Lo and Brian Park assisted with the
data analysis and the application of the weaving methods. Professors Roger Roess and Jose Ulerio
from the Polytechnic University in New York provided the data from the NCHRP 3- 75 Weaving
Analysis project.
iii
STATE OF CALIFORNIA DEPARTMENT OF TRANSPORTATION
TECHNICAL REPORT DOCUMENTATION PAGE
TR0003 ( REV. 10/ 98)
1. REPORT NUMBER
CA10- 0982
2. GOVERNMENT ASSOCIATION NUMBER
3. RECIPIENT’S CATALOG NUMBER
5. REPORT DATE
February 2010
4. TITLE AND SUBTITLE
Weaving Analysis, Evaluation and Refinement
6. PERFORMING ORGANIZATION CODE
7. AUTHOR( S)
Alexander Skabardonis, Amy Kim
8. PERFORMING ORGANIZATION REPORT NO.
UCB- ITS- PRR- 2010- 19
10. WORK UNIT NUMBER
193
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Institute of Transportation Studies
University of California, Berkeley
Berkeley, CA 94720
11. CONTRACT OR GRANT NUMBER
65A0208
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
October 2005 to June 2008
12. SPONSORING AGENCY AND ADDRESS
California Department of Transportation
Division of Research and Innovation, MS- 83
1227 O Street
Sacramento CA 95814
14. SPONSORING AGENCY CODE
15. SUPPLEMENTAL NOTES
16. ABSTRACT
Weaving sections are common design elements on freeway facilities such as near ramps and freeway to- freeway
connectors. When the traffic demands exceed the capacity at weaving areas congestion may occur, which affects
the operation of the entire freeway section. Traffic operational problems also may exist at weaving areas even
when traffic demands are less than capacity because of the complexity of vehicle interactions, resulting in poor
level of service ( LOS) and potential safety problems. Existing procedures for the design and analysis of freeway
weaving sections have several shortcomings, and their practical application often produces inconsistent results.
This report describes the work performed under PATH Task Order 6304. The objective of the study was to
evaluate the existing weaving analysis procedures to determine under which conditions the “ best available” tools
are most effective. The HCM2000, Leisch and Level D methods were evaluated using field data from 36 weaving
sections for a total of 189 data points of speeds and volumes. A weaving performance matrix was developed to
assist in determining which method to be used for a given mix of design and operational characteristics in a
weaving section.
17. KEY WORDS
Weaving, traffic flow, mathematical models
18. DISTRIBUTION STATEMENT
No restrictions
19. SECURITY CLASSIFICATION ( of this report)
None
20. NUMBER OF PAGES
155
21. PRICE
production of completed page authorized
iv
DISCLAIMER STATEMENT
This document is disseminated in the interest of information exchange. The contents of this
report reflect the views of the authors who are responsible for the facts and accuracy of the data
presented herein. The contents do not necessarily reflect the official views or policies of the State
of California or the Federal Highway Administration. This publication does not constitute a
standard, specification or regulation. This report does not constitute an endorsement by the
Department of any product described herein.
For individuals with sensory disabilities, this document is available in Braille, large print,
audiocassette, or compact disk. To obtain a copy of this document in one of these alternate
formats, please contact: the Division of Research and Innovation, MS- 83, California Department
of Transportation, P. O. Box 942873, Sacramento, CA 94273- 0001.
v
EXECUTIVE SUMMARY
Objectives and Methodology
Several methods exist for the design and analysis of freeway weaving sections. However, the existing
procedures have several shortcomings, and their practical application often produces inconsistent
results. This is mostly due to the lack of empirical data on weaving operations. Most of the existing
methods are based on limited data that are not representative of the entire range of the geometric
characteristics and traffic patterns in weaving areas, especially for California conditions. The
systematic evaluation of existing weaving methods and the development of an improved analysis
method have been recognized as high priority research needs.
The objectives of this research project are a) evaluation of the existing weaving analysis procedures to
determine under which design and operating conditions the “ best available” tools are most effective,
and b) development of an improved procedure either by modification of existing approaches or a new
method as appropriate.
We reviewed the literature on existing weaving analysis methodologies. We selected the Highway
Capacity Manual 2000 ( HCM2000), Leisch and Level D methods for evaluation with field data from
real- world weaving sections. We assembled a database of 36 real- world weaving sections from
California and the rest of the country for a total of 189 data points of operating conditions ( traffic
volumes and speeds). The analysis of the results identified the strengths and limitations of each
method in determining the performance of a freeway weaving section for a range of operating
conditions. Additional analyses were performed by applying the selected analysis methods to
synthetic datasets for the design and operating conditions that field data were not available. A total of
339 datasets were created. The analysis of the results focused on the consistency of the predictions
from each analysis method.
The research team also developed a new weaving analysis model based on empirical study of
bottleneck activations in two California freeway weaving sections. A theory was formulated for
mandatory lane changing ( i. e., lane changes required of a desired origin- destination pattern) based on
the empirical findings. The theory was used to enhance an existing simulation model of car- following
and lane changing. The model successfully reproduces field operating conditions in weaving sections.
Recommendations
We developed a performance matrix for each weaving analysis method to serve as a guide for
Caltrans staff when choosing the “ best” analysis method for the weaving section under study. Each
cell of the matrix represents a distinct design and operating condition. There are a total of 144 cells
for typical weaving sections of two, three, four and five lanes wide. Based on the comparison of the
model prediction with field and synthetic data, we show on each cell the performance of the particular
method as good ( or “ green light”), or partially good or often inconsistent ( or “ yellow light”) or poor
( or “ red light”) for a particular design and operating condition.
The proposed performance matrix for each analysis method ( HCM2000, Leisch and Level D) is
included in Appendix C of the report. Also, included in the Appendix is a single weaving analysis
performance matrix that shows the recommended methodology for each design and operating
condition. These matrices will be continually updated should more field data and/ or results from the
methods’ applications become available. It is envisioned that the proposed performance matrices will
be incorporated in the Caltrans Highway Design Manual.
vi
The new weaving analysis model was coded into an executable standalone computer program written
in MATLAB for use by Caltrans engineers for analysis and design of weaving sections. The inputs of
the program include the weaving section’s geometrics, free- flow speed, and traffic demands by
vehicles’ origin- destination. The software outputs include total delays as well as delay for each O- D
pair, and plots of cumulative vehicle count curves that display discharge flows and average speeds.
The model is documented in detail in Appendix D.
vii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ....................................................................................................................... ii
TECHNICAL REPORT DOCUMENTATION PAGE ......................................................................... iii
EXECUTIVE SUMMARY ...................................................................................................................... v
TABLE OF CONTENTS ........................................................................................................................ vii
LIST OF FIGURES........................................................................................................................ .......... ix
LIST OF TABLES......................................................................................................................... ........... ix
CHAPTER 1. INTRODUCTION............................................................................................................... 1
1.1 Problem Statement ............................................................................................................................... 1
1.2 Objectives of the Study ........................................................................................................................ 1
1.2.1 Evaluation of Existing Methods ........................................................................................................ 1
1.2.2 Development of a New Analysis Procedure...................................................................................... 1
1.3 Organization of the Report .................................................................................................................. 2
CHAPTER 2. LITERATURE REVIEW .................................................................................................. 3
2.1 The Level D Method......................................................................................................................... .... 5
2.2 The Leisch Method ............................................................................................................................... 8
2.3 The HCM2000 Method ......................................................................................................................... 9
2.4 Other Methods ............................................................................................................................... ..... 13
2.5 Measuring the Weaving Section Length............................................................................................ 13
CHAPTER 3. METHODOLOGY ........................................................................................................... 15
3.1 Weaving Sections Performance Matrix ........................................................................................... 15
3.2 Weaving Sections Classification ......................................................................................................... 16
3.2.1 Geometric Characteristics ............................................................................................................... 16
3.2.2 Operational Conditions.................................................................................................................... 17
CHAPTER 4. THE STUDY DATABASE .............................................................................................. 19
4.1 Data Sources ............................................................................................................................... ........ 19
4.1.1 California Studies— Major Weaving Sections ............................................................................... 19
4.1.2 California Studies— Ramp Weaves................................................................................................. 20
4.1.3 The NGSIM Data Sets...................................................................................................................... 22
4.1.4 The NCHRP 3- 75 Database ............................................................................................................. 22
viii
4.1.5 Caltrans District 5 Data .................................................................................................................. 23
4.2 The Study Database....................................................................................................................... ..... 24
CHAPTER 5. EVALUATION OF EXISTING METHODS ............................................................... 27
5.1 Application of Existing Methods to Field Data................................................................................. 27
5.2 Santa Barbara Data Sets .................................................................................................................. 28
5.3 Synthetic Data Sets ............................................................................................................................. 28
CHAPTER 6. CONCLUSIONS ............................................................................................................... 32
REFERENCES..................................................................................................................... .................... 34
APPENDIX A. CALIFORNIA STUDIES MAJOR WEAVING SECTIONS.................................... 35
APPENDIX B. NCHRP 3- 75 TEST SITES ............................................................................................ 36
APPENDIX C. WEAVING ANALYSIS PERFORMANCE MATRIX ............................................. 40
C1. Weaving Analysis Performance Matrix – HCM2000 Method................................................... 41
C2. Weaving Analysis Performance Matrix – Leisch Method ........................................................ 42
C3. Weaving Analysis Performance Matrix – Level D Method ....................................................... 43
C4. Weaving Analysis Performance Matrix – Recommended Methodology.................................. 44
APPENDIX D. Lee J, H., and M. C. Cassidy, “ An Empirical And Theoretical
Study Of Freeway Weave Bottlenecks,” PATH Research Report UCB- ITS- PRR- 2009- 13.......... 45
ix
LIST OF FIGURES
Figure 2.1 Weaving Chart – 1965 HCM ....................................................................................................... 5
Figure 2.2 Level D Method: Distribution of On- and Off- Ramp Traffic in Lane 1 and Auxiliary Lane ..... 6
Figure 2.3 Leisch Method: Nomograph for One Sided Weaving Sections .................................................. 8
Figure 2.4 Configurations of Freeway Weaving Sections ( 1985/ 2000 HCM)............................................. 9
Figure 2.5 Measuring the Weaving Section Length ( HCM2000 Exhibit 13- 11) ..................................... 14
Figure 2.6 NCHRP 3- 75 Measurement of Weaving Section Length ([ 2]) ................................................. 14
Figure 3.1 Typical Balanced ( HCM2000 Type B) Weaving Sections ....................................................... 16
Figure 3.2 Typical Unbalanced ( HCM2000 Type C) Weaving Sections................................................... 16
Figure 5.1 Measured vs. HCM2000 Predicted Weaving Speed ................................................................ 27
Figure 5.2 Comparison of Weaving Analysis Methods— Caltrans District 5 Data ................................... 28
Figure 5.3 Comparison of Leisch vs. HCM2000— Synthetic Data Sets .................................................. 30
LIST OF TABLES
Table 2.1 Existing Procedures for the Analysis of Freeway Weaving Sections .......................................... 3
Table 2.2 Level D Method: Proportion of Through Traffic Remaing in Outer Through Lane .................... 7
Table 2.3 The PINY Weaving Analysis Method........................................................................................ 10
Table 2.4 Constants for Computing Weaving Intensity Factors ( HCM2000, Exhibit 24- 6)...................... 12
Table 2.5 LOS Criteria for Freeway Weaving Sections ( HCM2000, Exhibit 24- 2) .................................. 12
Table 3.1 Proposed Weaving Analysis Performance Matrix ..................................................................... 15
Table 4.1 California Weaving Studies— Major Weaving Sections ........................................................... 20
Table 4.2 California Weaving Studies— Ramp Weaves ............................................................................ 21
Table 4.3 The NCHRP 3- 75 Database ....................................................................................................... 23
Table 4.4 Caltrans District 5 Database ....................................................................................................... 24
Table 4.5 The Study Database: Geometric Characteristics ....................................................................... 25
Table 4.6 Mapping the DataSets to the Perormance Matrix ...................................................................... 26
Table 5.1 Synthetic Data Sets ................................................................................................................... 29
Table 5.2 Difference in LOS Prediction— Synthetic Data Sets ................................................................ 31
1
CHAPTER 1
INTRODUCTION
1.1 Problem Statement
Weaving sections are common design elements on freeway facilities such as near ramps and freeway-to-
freeway connectors. When the traffic demands exceed the capacity at weaving areas congestion
may occur, which affects the operation of the entire freeway section. Traffic operational problems
also may exist at weaving areas even when traffic demands are less than capacity because of the
complexity of vehicle interactions, resulting in poor level of service ( LOS) and potential safety
problems.
Efforts to develop procedures for the design and analysis of freeway weaving sections begun in the
50’ s. However, the existing procedures have several shortcomings, and their practical application
often produces inconsistent results. This is mostly due to the lack of empirical data on weaving
operations. Most of the existing methods are based on limited data that are not representative of the
entire range of the geometric configurations and traffic volumes and patterns in weaving areas,
especially for California conditions. The systematic evaluation of existing weaving methods and the
development of an improved analysis method have been recognized as high priority research needs.
1.2 Objectives of the Study
The objectives of this research project are a) evaluation of the existing weaving analysis procedures to
determine under which design and operating conditions the “ best available” tools are most effective,
and b) development of an improved procedure either by modification of existing approaches or a new
method as appropriate. The tasks performed and the end products for each research objective are
briefly described below.
1.2.1 Evaluation of the Existing Methods
We reviewed existing weaving analysis procedures. Next, we assembled existing data on real- world
weaving sections from several sources. We applied the existing methods to the field data and
analyzed the results. Based on the evaluation of the results we developed recommendations regarding
the use of existing methods for design and analysis of weaving sections.
The end product of this research effort are guidelines documenting under what conditions which of
the “ best available” weaving analysis tools are most effective, and how and under what conditions
these tools can be properly applied. The guidelines are in the form of weaving analysis performance
matrices, which describe the operating conditions under which each analysis method is most
effective. It is envisioned that the proposed guidelines will be incorporated in the Caltrans Highway
Design Manual ( Chapter 500-- Section 504.7) [ 1].
1.2.2 Development of a New Weaving Analysis Procedure
The research first investigated what triggered the bottleneck activations in two California freeway
weaving sections, SR- 55N and SR22W in Orange County, and I- 210W and Lake Avenue in
Pasadena. Both sites are recurrent bottlenecks during the rush, and investigations revealed that
changes in the spatial patterns of vehicular lane- changes, especially among freeway- to- ramp ( F- R)
maneuvers, caused variations in bottleneck discharge flow. It was also found that the spatial
distributions of these lane changes, in turn, were dictated by the traffic conditions in the auxiliary lane
( i. e., the lane connecting the off- ramp to the upstream on- ramp). Reductions in on- ramp flows
increased the attractiveness of the auxiliary lane, thus motivating F- R drivers to perform their
2
maneuvers nearer the onramp. Conversely, increases in on- ramp flows motivated F- R drivers to
perform their maneuvers over a wider stretch of the weaving section.
Next, a theory was formulated for mandatory lane changing ( i. e., lane changes required of a desired
origin- destination pattern) based on the empirical findings. The theory was used to enhance an
existing simulation model of car- following and lane changing. With this new theory, the driver’s
decision to attempt a lane change is determined by the vehicle’s distance from the downstream end of
the weaving section’s diverge area, the number of lanes to be crossed in reaching the desired
destination, and the difference in densities between the driver’s target lane and her current one. The
model successfully reproduces the observed mechanisms of bottleneck activation and discharge flows
in weaving sections.
The model was developed into an executable standalone computer program written in MATLAB to be
used by Caltrans engineers for analysis and design of weaving sections. The inputs of the program
include the weaving section’s geometrics ( e. g., length of the weaving section of interest, number of
lanes), free- flow speed, and traffic demands by vehicles’ origin- destination. The software outputs
include total delays as well as delay for each O- D pair, and plots of cumulative vehicle count curves
that display discharge flows and average speeds.
1.3 Organization of the Report
This document is the final report for the study. Chapter 2 reviews existing methods for the analysis of
freeway weaving sections. The study methodology and the approach for developing the weaving
analysis performance matrix are described in Chapter 3. Chapter 4 describes the development of the
study database. The application of the existing methods to the field data and the analysis of the
results are presented in Chapter 5. Chapter 6 summarizes the study findings and recommendations.
The proposed guidelines in the form of weaving analysis performance matrix for each method are
included in Appendix C. Appendix D includes the research report1 documenting in detail the new
weaving analysis procedure and software.
1 Lee, J. H., and M. C. Cassidy, “ An Empirical and Theoretical Study of Freeway Weave Bottlenecks,” PATH
Research Report UCB- ITS- PRR- 2009- 13, February 2009.
3
CHAPTER 2
LITERATURE REVIEW
Efforts to develop procedures for the design and analysis of freeway weaving sections begun in the
1950’ s. Table 2.1 provides an overview of existing analysis procedures; of those methods, the Leisch
Method and the Moskowitz and Newman ( Level D Method) are included in the Caltrans Highway
Design Manual ( shown as shaded cells in Table 2.1).
Table 2.1 Existing Procedures for the Analysis of Freeway Weaving Sections [ 2]
Model Basic
Type
Address
Capacity?
Address
LOS?
MOE Comments
HCM 1965
( Normann)
1965
Macroscopic,
Equivalent Non-
Weaving
Vehicles
Not directly.
Yes
Approx.
Speed
Based on very sparse data.
Quality of Flow used to map into
LOS.
Hess
1963
Macroscopic,
Lane Distribution
Yes
Freeway
Capacity
Controls
Yes
Merge,
Diverge, and
Freeway
Volume
Regression- based model, focuses
on lane 1 of the freeway and the
ramp, general LOS criteria based
on flow rates
Moskowitz &
Newman
( Level D
method)
1963
Microscopic,
Lane Distribution
and Lane-
Changing by Cell
Yes
Freeway
Capacity
Controls
Yes
Merge,
Diverge,
Weaving, and
Freeway
Volume
Focus on high- volume cell among
freeway lane 1 and auxiliary lane,
general LOS criteria based upon
flow rates
Polytechnic
Method
1973- 1980
Macroscopic,
Regression
Based,
Speed Prediction
Not directly.
Yes
Average
Speed of
Weaving and
Non- Weaving
Vehicles
Iterative approach. Introduced
weaving section configuration
and type of operation into the
analysis process.
Leisch
1983
Macroscopic,
Equivalent Non-
Weaving
Vehicles
Not directly.
Yes
Average
Speed
A re- calibration of the HCM 1965
Leisch/ Normann work.
Nomographs used.
JHK method
1984
Macroscopic,
Regression-
Based,
Speed Prediction
Not directly.
Yes
Average
Speed of
Weaving and
Non- Weaving
Vehicles
Introduced a different “ density”
concept tied to weaving intensity,
introduced basic model form still
used in HCM 2000.
1985 HCM
Roess et al
Macroscopic
Not directly.
Yes
Average
Speed of
Weaving &
Non- Weaving
Vehicles
Developed as a merger of the
earlier Polytechnic and JHK
models. The JHK model form
was stratified to consider
configuration and type of
operation.
Fazio
1985
Macroscopic,
Regression- Based
Not directly.
Yes
Average
Speed of
Weaving and
Non- Weaving
Vehicles
Added lane- changing parameter
to Reilly- type model, eliminating
the need for different
configuration types to be
considered.
ITS UC
Berkeley
1988- 1995
Microscopic,
Lane- distribution
Lane- changing by
Cell
Yes, Based on
max flows
and max lane-changing
per
Cell
Yes
Density
A modern look at the Level D
method including major weaving
sections. Lane distribution
modeled for each component flow
of the weaving section.
HCM 2000
Roess et al
Macroscopic Yes Yes Density Addition of density model and
capacity predictions to 1985
HCM methodology.
Lertworawanic
& Elefteriadou
2001- 2002
Microscopic,
Gap Acceptance
Yes
No
N/ A
Capacity model based upon gap
acceptance and linear
programming optimization
4
The first formal procedure for analysis of weaving sections appeared in the 1965 edition of HCM [ 3],
based on research conducted by O. K. Normann [ 4]. It was intended to cover both simple and
multiple weaving areas, one- sided and two- sided weaving areas. The basic model in the 1965 HCM is
shown in Figure 2.1. It graphically depicts a relationship between weaving length, total weaving
volume, and Quality of Service. The latter is an operational measure that is later mapped into the
HCM Level of Service ( LOS) definitions. Associated with the five defined qualities of flow ( I – V
from best to worst) and intermediate points was a weaving equivalence factor, k.
The k- factor essentially converted the total volume in the weaving section to an equivalent non-weaving
volume using the following equation:
SV
v k v
SV
N vo1 vo2 vw1 k vw2 w2 + ( − 1)
=
+ + +
=
where:
N = number of lanes in the weaving section,
vo1, o2 = larger and smaller non- weaving volume respectively, veh/ h,
vw1, w2 = larger and smaller weaving volume respectively, veh/ h,
k = weaving equivalence factor,
SV = service volume per lane, veh/ h ( volume for specified quality of flow, max
value: 2,000 veh/ h/ lane)
The k- factor is drawn from Figure 1 using the weaving length and the total weaving volume, vw1+ vw2.
Its value ranged from 1.0 for Quality of Flow I to 3.0 at Quality of Flow III and above. The value of
k= 1 also provides a boundary for “ out of the realm of weaving,” i. e., the point at which length is
sufficient for the section to operate as isolated merge and diverge points with a basic freeway section
in between. There was no empirical evidence regarding these boundary k- values. The value of 1.0
was logical, given that this was the boundary beyond which weaving operations were equivalent to
basic freeway operations. The maximum value of 3.0 was based on the assumption that a weaving
vehicle would need a gap of approximately 3 vehicle- lengths to successfully execute a weaving
maneuver. Intermediate values of the k- factor were developed using interpolation process without
empirical data.
The 1965 HCM method was widely used and brought some national consistency to the analysis and
design of weaving areas. The methodology covered a wide range of situations and configurations in
which weaving could exist. However, the method was based on very limited few field data.
The 1965 HCM Chapter on ramp junctions contains another procedure for analysis of ramp- weaving
configurations, i. e., one lane ramps followed by off- ramp with a continuous auxiliary lane. It was
recommended that a procedure developed by Hess [ 5] will be used to analyze ramp junctions for LOS
A to C ( free- flow conditions), and the methodology developed in California by Moskowitz &
Newman [ 6] when the LOS is D or E ( heavy traffic conditions).
5
Figure 2.1 Weaving Chart-- 1965 HCM
( Source: Highway Capacity Manual, 2nd Edition, Special Report 87, Highway Research Board, Washington
DC, 1965, pg 166)
2.1 The Level D Method
The Level D method was developed in California by Moskowitz & Newman to analyze weaving
sections under heavy traffic conditions ( LOS is D or E) [ 6]. The method applied to weaving sections
with one lane ramps followed by off- ramp with a continuous auxiliary lane. The method was
included in the ramp junctions Chapter of the 1965 HCM.
Figure 2.2 illustrates the Level D method. It shows, for various weaving lengths, percentages of on-ramp
and off- ramp traffic remaining in the auxiliary lane and the right- most through lane at 500 ft
intervals through the weaving section. This was augmented by a Table that provides the proportion of
the freeway through traffic remaining in outer through lane in the weaving section ( Table 2.2). Table
2.2 only applies to traffic not involved in a ramp movement within 4,000 ft.
The analyst estimates the traffic volumes in the right most through lane and the auxiliary lane at 500
ft intervals using the values in Figure 2.2 and Table 2.2. These values are compared against the lane
capacities in the weaving section. The method identifies the segment and lane of the weaving section
with the highest volume ( and highest amount of lane- changing activity).
6
Figure 2.2 Level D Method: Distribution of on- and off- Ramp Traffic in Lane 1 and Auxiliary Lane
( Source: Caltrans, Highway Design Manual, and Highway Capacity Manual, 2nd Edition, Special Report 87,
Highway Research Board, Washington DC, 1965).
7
Table 2.2 Level D Method: Proportion of Through Traffic Remaining in Outer Through Lane
( Source: Caltrans Design Manual, and Highway Capacity Manual, 2nd Edition, Special Report 87, Highway
Research Board, Washington DC, 1965.)
The Level D method was extended for other types of weaving sections in a series of studies
conducted at the Institute for Transportation Studies ( ITS) at the University of California at Berkeley,
in cooperation with Caltrans [ 7,8,9]. The overall capacity and level of service in a weaving area was
most heavily influenced by the flow and lane- changing activity of critical cells ( a particular lane at a
particular distance from the entry gore area) within the section. Thus, the most effective models
should predict the activity in the critical cell( s) of the section, and from that, make overall predictions
of section capacity and level of service. The recommended procedures consist of the following steps
( Lanes were numbered 1 through n starting with the right- most lane of the weaving section):
1. Predict the proportion ( and then the flow rate) of ramp- to- freeway ( RF) traffic in lanes 1 and 2 at
various distances from the entry gore area.
2. Predict the proportion ( and then the flow rate) of freeway- to- ramp ( FR) traffic in lanes 1 and 2 at
various distances from the entry gore area.
3. Predict the proportion ( and then the flow rate) of freeway- to- freeway ( FF) traffic in the right- most
through at various distances from the entry gore area.
4. Determine the flow rates in the critical cell( s) of the weaving section.
5. Determine the density or lane- changing rate of the critical cell( s); establish capacity and LOS
8
2.2 The Leisch Method
This method was developed by J. Leisch based on data from 48 weaving sections around the country
[ 10,11]. The method used concepts similar to the 1965 HCM and a nomograph approach. Two sets
of nomographs were created: one for one- sided weaving sections, and the other for two- sided
weaving areas. A sample of the Leisch nomographs for one- sided weaving sections is shown in
Figure 2.3.
In calibrating his nomographs, Leisch retained some key elements of the 1965 HCM weaving chart.
The primary relationship is still between the length of the weaving section and the total weaving
volume, except that LOS curves replace Quality of Flow curves. The weaving intensity factor ( k)
continues to range between 1.0 and 3.0 for one- sided weaving sections. Solution of the nomographs
results in determination of either the LOS of a weaving section with known design characteristics, or
the number of lanes needed to obtain a specified LOS. The method also produces estimates of the
average speeds of all vehicles and the weaving vehicles. The method accounts for the difference in
operational characteristics between lane- balanced and unbalanced weaving sections. Lane balanced
sections have one more lane going away, such as an optional lane at exit; i. e., one weaving movement
is not required to change lanes.
The advantage of the Leisch method is that it is relatively easy to apply, and could be manipulated to
produce design and/ or operational analysis results. However the development and calibration of
nomographs was mostly based on experience and judgment with very limited field data.
Figure 2.3 Leisch Method: Nomograph for One- Sided Weaving Sections
( Source: Leisch, J., Completion of Procedures for Analysis and Design of Traffic Weaving Areas, Final Report,
Vols 1 and 2, Federal Highway Administration, Washington DC, 1983.)
9
2.3 The HCM2000 Method
The origin of the HCM2000 method is the weaving analysis method developed by the Polytechnic
Institute of New York ( PINY method) [ 12,13]. This method is based on the same field data as the
Leisch method, but it explicitly recognizes the geometric configuration of the weaving section,
depending on the minimum number of lane changes required by the weaving vehicles.
Table 2.3 summarizes the PINY method. The LOS is defined based on the speeds of weaving and
non- weaving vehicles. Freeway weaving sections are classified into three configurations, depending
on the minimum number of lane changes required by weaving vehicles ( Figure 2.4).
Type A: each weaving vehicle must make one lane- change ( ramp weaves)
Type B: major weaving configurations requiring one lane change for the one weaving
movement and none for the other weaving movement
Type C: major weaving configurations requiring two or more lane changes for one weaving
movement and none for the other weaving movement
Figure 2.4 Configurations of Freeway Weaving Sections ( 1985/ 2000 HCM)
( Source; Highway Capacity Manual, 2000, Transportation Research Board, Washington DC, 2000.)
The concept of constrained vs. unconstrained operations was also introduced. Constrained operations
occur when the geometry of the section constrains weaving vehicles from using certain freeway lanes.
Under constrained operations weaving vehicles occupy a smaller proportion of the roadway than they
would without the constraint of geometry; non- weaving vehicles occupy more space, and the
difference between non- weaving and weaving vehicle speeds increases.
The PINY methodology was complex, because of the inter- relationships of values of Sw, Snw, Nw, and
Nnw. A solution is started by assuming a high value of Sw ( e. g., 55 mph) and iterating through the
process until the resulting Snw agrees with the starting assumption. The type of operation ( constrained
or unconstrained) is determined by comparing Nw to Nw ( max). The logic of the equations is not
intuitively obvious, and many users had difficulty implementing the procedures when they were
published.
Type C
Type A
Type B
10
Table 2.3 The PINY Weaving Analysis Method
Weaving
Type
Equation
Maximum Number of Weaving Lanes, Nw ( max)
A Nw ( max) = 2.0
B Log Nw ( max) = 0.741+ 0.480 log R
C Log Nw ( max) = 0.896+ 0.186 log R – 0.402 log LH
Speed Relationship Between Sw and Snw
A log Sw = 0.142+ 0.692 log Snw + 0.313 log LH
B Sw = 15.031+ 0.819 Snw – 23.527 VR **
C Sw = 2.309 + 0.871 Snw + 4.579 VR **
Portion of Total Lanes Used by Weaving Vehicles, Nw/ N
A log Nw/ N = 0.340 + 0.571 log VR – 0.438 log Sw + 0.234 log LH **
B Nw/ N = 0.761- 0.011 LH – 0.005 ( Snw – Sw)
C Nw/ N = 0.085+ 0.703 VR + ( 234.763/ L) – 0.018 ( Snw – Sw)
Speed – Flow Relationship for Non- Weaving Vehicles
A
B
C
Vnw = 1500 Nnw – 50.0 Snw + 1900
** Secondary equations only applying when the section is unconstrained, i. e., Nw < Nw ( max)
Notes: 1. All volumes expressed as equivalent passenger cars/ hour ( pch) under ideal conditions
S w
, Snw = average speed of weaving, non- weaving traffic, mph
Vw , Vnw = weaving and non- weaving volume ( pch)
N w
= number of lanes used by weaving vehicles under unconstrained operation;
N = Number of lanes in weaving section
VR = ratio of weaving volume to total volume;
R = ratio of smaller weaving volume to total weaving volume.
L = length of weaving section, ft; ( LH = length of weaving section, hundreds of ft.)
The JHK Methodology[ 14]: This method was developed as part of a research effort sponsored by
FHWA to evaluate the accuracy of the PINY and Leisch methodologies against field data. often
produced significantly different results. The algorithms were developed based upon the “ density”
within the weaving section, which was calculated by dividing the ratio of weaving to total volume by
the length of the section. The procedures developed were universally applied to all weaving sections,
regardless of configuration or type of operation. The procedure consists of two equations to predict
the average speeds of weaving and non- weaving vehicles:
⎥⎦
⎤
⎢⎣
⎡ +
+
−
= +
d
w nw b c
L
a VR v N
S S S S
1 ( 1 ) ( / )
max min
, min ( 2- 1)
where all variables are as previously defined. Smax and Smin are the maximum and minimum speeds
expected for weaving or non- weaving vehicles as appropriate.
The 1985 HCM method: This is a modification of the JHK method to incorporate the concepts of
weaving configuration and constrained vs. unconstrained operation [ 15, 16]. This was accomplished
by calibrating the constants ( a, b, c, d) in Equation ( 2- 1) for the three different weaving configurations
11
( A, B, and C), for unconstrained and constrained operations, and for weaving and non- weaving
speeds. The result was 12 different equations, all in the form Equation ( 2- 1).
Several concerns have been expressed by transportation researchers and professionals regarding the
HCM1985 method because a) it could not provide capacity estimates; b) it uses rather complex
equations for estimating weaving and non- weaving vehicle speeds to determine LOS, and the logic of
these formulae is not readily apparent, and d) often inappropriately reflects impacts created by
changes in geometric configuration of the weaving areas.
Another weaving operations study ( NCHRP 3- 55) [ 17] was undertaken as part of the research for the
2000 edition of HCM. The study relied heavily on simulation, due to the high cost of data collection.
The study did not produce a satisfactory procedure to replace the existing HCM85 weaving analysis
methodology. It yielded a number of trends that were judged to be counter- intuitive, e. g., it proposed
that the operation of weaving areas was not influenced by length of the section.
The HCM2000 methodology [ 18] is a judgmental modification of the previous methods to provide
improved consistency among the freeway- related methodologies in the HCM. These modifications
included a) recalibration of the constants ( a, b, c, d) to reflect further changes in other freeway
analysis related chapters of the Manual, and b) determination of LOS based upon the density in the
weaving section eliminating the practice of assigning separate levels of service to weaving and non-weaving
vehicles.
The HCM2000 methodology first calculates the speeds of weaving and non- weaving vehicles as
follows:
d
b c
w nw
w nw
w nw
L
W a VR v N
W
S S S S
( 1 ) ( / )
1
,
,
max min
, min
+
=
+
−
= +
where: Sw = average speed of weaving vehicles, mph
Snw = average speed of non- weaving vehicles, mph
Smin = minimum average speed for weaving section, mph
Smax = maximum average speed for weaving section, mph
Ww = weaving intensity factor for weaving vehicles
Wnw = weaving intensity factor for non- weaving vehicles
VR = volume ratio; ratio of weaving flow rate to total flow rate
v = total demand flow rate under equivalent ideal conditions, pc/ h
N = number of lanes in the weaving section
L = length of the weaving section, ft
a- d = constants of calibration
In the HCM2000, the minimum average speed for all weaving sections ( Smin) is set at 15 mph. The
maximum average speed ( Smax) is defined as FFS+ 5, where FFS is the freeway free- flow speed. The
additional 5 mph corrects for the shape of the algorithm, which tends to under- predict higher speeds.
With these assumptions, the equations for the speed of weaving and non- weaving vehicles become:
w nw
n nw W
S FFS
,
, 1
15 10
+
−
= +
12
The calibration constants ( a, b, c, d) are given below ( Table 2.4):
Table 2.4 Constants for Computing Weaving Intensity factors ( HCM2000, Exhibit 24- 6)
The HCM2000 methodology then converts the component speeds to an average flow weighted speed
for all vehicles. Next, the density for the section is computed from the average speed and flow rate.
The LOS is determined from the computed density value based on the following table ( Table 2.5):
Table 2.5 LOS Criteria for Freeway Weaving Sections ( HCM2000, Exhibit 24- 2)
The capacity of a weaving section is established as the minimum of three values:
1. The total flow rate that results in a density of 43 pc/ mi/ ln, assumed to result in breakdown
2. The total flow rate that results in a weaving flow rate equal to the maximum allowable value
( 2,800 pc/ h for Type A sections; 4,000 pc/ h for Type B sections; 3,500 pc/ h for Type C sections).
3. The total flow rate equal to the basic freeway capacity of all lanes in the weaving section.
13
The application of the HCM2000 methodology consists of the following steps:
1. Input: geometric data, traffic volumes per movement, free flow speed of freeway segment
2. Volume adjustment: peak- hour factor, heavy vehicles, driver population
3. Compute flow rates
4. Establish weaving segment configuration type
5. Compute unconstrained weaving and non- weaving speed
6. Check for constrained- flow operation
• If constrained, compute constrained weaving and non- weaving speeds
• Otherwise, use the unconstrained parameter
7. Compute average space mean speed within weaving segment
8. Compute density within the weaving segment
9. Determine LOS
2.4 Other Methods
Fazio model: This is a modification of the JHK model. It includes a variable to account for the lane-changing
activity in the weaving section [ 19]. Insertion of such a variable allowed for development
of speed- prediction equations ( one for weaving speed, one for non- weaving speed) without reference
to weaving section configuration categories. The proposed model was based on limited data and
assumed an entering lane- distribution pattern for weaving vehicles. It was also assumed that all
weaving vehicles left the section on the lane closest to their weaving maneuver.
Penn State model: This methodology for estimating the capacity of ramp- weave and major weave
sections is based upon linear optimization and gap acceptance modeling [ 20]. The methodology
defines the sum of weaving section capacity as the sum of two components: a) the capacity of
weaving lanes and b) the capacity of non- weaving lanes ( all other lanes), assumed to be equivalent to
the basic freeway capacity per HCM 2000. The maximum values of each weaving flow rate are based
upon gap acceptance modeling of the necessary lane- changing maneuvers made by the weaving
vehicles. The study had to assume the values of gap acceptance parameters and the validation data
base was relatively sparse, and not microscopically detailed.
Simulation Studies: Several studies have been undertaken using simulation to evaluate the operation
of existing weaving sections and to evaluate alternative designs [ 21]. Most of these studies focused
on developing procedures for successfully simulating weaving operations, and evaluating the
effectiveness of existing simulation models to simulate weaving sections.
NCHRP 3- 75 Study: The objective of this national study is to develop a new procedure for analysis of
weaving sections to be included in the 2010 edition of the HCM [ 2]. The procedures is based on field
data on weaving operations collected at 14 sites throughout the country. The new procedure has been
included in the 2010 edition of HCM which expected to be available in late 2010.
2.5 Measuring the Weaving Section Length
The measurement of weaving section length according to the HCM2000 is shown in Figure 2.5. The
length is measured from a point at the merge gore where the right edge of the freeway shoulder lane
and the left edge of the merging lane( s) are 2 ft apart to a point at the diverge gore where the two
edges are 12 ft apart. This definition was originally introduced in the 1965 HCM [ 3], and it also the
same for the Leisch method [ 11] and the Level D method as described in Chapter 8 of the 1965 HCM
[ 3]. This definition appears that is based on the ramp geometry between loops of a cloverleaf
interchange, where the exit loop generally diverted at a sharper angle than the entry loop merged.
14
Figure 2.5 Measuring the Weaving Section Length ( HCM2000 Exhibit 13- 11)
The Caltrans Highway Design Manual defines the length from a point at the merge gore where the
right edge of the freeway shoulder lane and the left edge of the merging lane( s) are 6 ft apart to the
diverge gore where the two edges meet [ 1].
In the recently completed NCHRP 3- 75 project [ 2], several ways of measuring weaving length were
considered, and the selected ones are shown in Figure 2.6 below:
Figure 2.6 NCHRP 3- 75 Measurement of Weaving Section Length ([ 2])
Ls = the distance between the end points of any barrier markings ( solid white lines) that prohibit or
discourage lane‐ changing ( ft)
LB= the distance between points in the gore areas where the left edge of the ramp traveled way
and the right edge of the freeway traveled way meet ( ft)
Field observations indicate that LB defines the length used for lane- changing, but the use of Ls
improved the statistical fit of models to the field data, and will be used in the 2010 HCM. In the 3- 75
database it was found that Ls = 77% of LB , which can be used as a default value when the details of
striping are not known.
In the test sites included in our study, the weaving length was measured as the distance between
points in the gore areas where the left edge of the ramp traveled way and the right edge of the freeway
traveled way meet ( ft).
15
CHAPTER 3
METHODOLOGY
Currently, the Caltrans Highway Design Manual includes two methodologies for determining the
capacity and/ or Level of Service of weaving sections: the Level D method and the Leisch method.
Although the HCM2000 method is not officially recommended for use, it is often applied to check
whether other analysis results are reasonable.
To determine how well each of the above methods predicts operations at weaving sections, for each
study site the analysis results of the three methods were compared to the actual operating conditions
that correspond to each data set. The results were then further analyzed to determine which of three
existing methods predicts best the operating characteristics of a weaving section under certain
geometric and operational conditions.
3.1 Weaving Methods Performance Matrix
The weaving analysis performance matrix was created to serve as a guide for Caltrans design
engineers when choosing the “ best” weaving analysis method for the weaving section under study,
based on comparisons with field data. A method that works well for a given geometric/ operational
mix will be given a “ green light”, a satisfactory method a “ yellow light” and poor one a “ red light”.
Table 3.1 shows the proposed weaving analysis performance matrix. A performance matrix will be
developed for each analysis method.
Table 3.1 Proposed Weaving Analysis Performance Matrix
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy
Weaving: Heavy
Non- Weaving: Heavy
Weaving: Mid to Low
Non- Weaving: Mid to Low
Weaving: Heavy
Non- Weaving: Mid to Low
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy
Weaving: Heavy
Non- Weaving: Heavy
Weaving: Mid to Low
Non- Weaving: Mid to Low
Weaving: Heavy
Non- Weaving: Mid to Low
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy
Weaving: Heavy
Non- Weaving: Heavy
Weaving: Mid to Low
Non- Weaving: Mid to Low
Weaving: Heavy
Non- Weaving: Mid to Low
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy
Weaving: Heavy
Non- Weaving: Heavy
Weaving: Mid to Low
Non- Weaving: Mid to Low
Weaving: Heavy
Non- Weaving: Mid to Low
Weaving: Mid to Low
16
Each cell represents a distinct design and operating condition. For a given number of lanes in the
weaving section, the matrix has 48 cells; operational characteristics are reported by row while
geometric characteristics are reported by column. There are a total of 192 possible cells for typical
weaving sections of two, three, four and five lanes wide. Shaded cells indicate infeasible conditions.
For example, it is not possible to have a two lane weaving section with more than one on- or off-ramps.
Therefore, the proposed matrix includes a total of 144 cells.
The following sections describe the classification of design and operational conditions for developing
the performance matrix.
3.2 Weaving Section Classification
3.2.1 Geometric Characteristics
The weaving sections geometric characteristics include the total number of lanes in the weaving
section, the number of auxiliary lanes, and the length of weaving section. First, we consider the total
number of lanes in the weaving section: two, three, four, or five lanes wide. Next, for a given number
of lanes, we consider the presence and number of auxiliary lanes:
1. No Auxiliary Lane, single lane on- & off- ramps
2. With Auxiliary Lane, single - lane on & off- ramps ( Type A, HCM 2000): These are
weaving sections consisting of two- lane on or off- ramps in which each weaving
movement is required to make one lane change. These are also called ramp weaves.
3. Balanced, > 1 lane on- & off- ramps ( Type B): These are weaving sections consisting of
two- lane on or off- ramps in which one weaving movement is not required to make a lane
change, and the other weaving movement is required to make one lane change. It also
includes balanced sections, i. e., weaving sections with an optional lane at exit, i. e., “ one
more lane going away. Note balanced sections include weaving sections with a single
lane on- or off- ramp ( Figure 3.1).
Figure 3.1 Typical Balanced ( HCM2000 Type B) Weaving Sections
4. Unbalanced, > 1- lane on-& off- ramps ( Type C): These are weaving sections consisting of
two- lane on or off- ramps in which one weaving movement is required to make two lane
changes, and the other weaving movement is not required to make a lane change. It also
includes unbalanced sections, i. e., weaving sections without an optional lane at exit
( Figure 3.2)
Figure 3.2 Typical Unbalanced ( HCM2000 Type C) Weaving Sections
NB101
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
WB10SB
1 1 1
2 2 2
3 3 3
4
17
Under each of these groups, a weaving section can further be classified according to its length as
short, medium or generous, as follows:
1. Short Weave Length (< 1,000 ft)
2. Medium Weave Length ( 1,000- 2,500 ft)
3. Generous Weave Length (> 2,500 ft)
These thresholds were assumed to be reasonable in that weaving sections in each group would exhibit
similar traffic behavior given certain traffic volumes.
3.2.2 Operational Conditions
The operational conditions are grouped based on the total weaving and non- weaving traffic volumes
in the weaving section as follows:
1. Non- Weaving Volumes: Heavy, Weaving Volumes: Heavy
2. Non- Weaving Volumes: Heavy, Weaving Volumes: Mid to Low
3. Non- Weaving Volumes: Mid to Low, Weaving Volumes: Heavy
4. Non- Weaving Volumes: Mid to Low, Weaving Volumes: Mid to Low
The non- weaving volume includes all traffic traveling through a weaving section ( freeway- to-freeway)
and from the on- ramp to off- ramp. The weaving volume consists of the on- ramp to freeway
volume and the volume from the freeway to the off- ramp. It was determined that volumes could be
grouped in this way because it does not appear that performance estimates from the existing analysis
methods would differ if, for instance, one weaving section had high on- ramp to freeway volumes and
another had high freeway to off- ramp volumes. The analysis methods do not recognize the difference
between these two groups of traffic, and two scenarios would yield the same analysis results.
The non- weaving and weaving volumes are classified as “ heavy” or “ mid to low” based on the
number of lanes in the “ conflict area” of the weaving section. The term conflict area is used to
indicate the travel lanes where most of the turbulence occurs due to merging and diverging traffic.
Most turbulence occurs in the lanes adjacent to the on- and off- ramps, and as a result the conflict area
is defined as follows, based on “ A Proposed Analytical Technique for the Design and Analysis of
Major Freeway Weaving Sections” [ 7]:
1. ( Conflict Area 1) The area of the weaving section extending from the right- most auxiliary
lane to the lane directly to the left of the diverge gore, or
2. ( Conflict Area 2) The area of the weaving section extending from the right- most auxiliary
lane to the lane directly to the left of the merge gore.
1 2
Whichever of the above descriptions encompasses more lanes of the weaving area will govern as the
conflict area. The lanes in the conflict area are those “ reserved” for weaving volumes, and the
Conflict Area Conflict Area
18
remaining lanes of the weaving section are those “ reserved” for non- weaving volumes. The table
below indicates the criteria by which weaving and non- weaving volumes are classified as “ heavy” or
“ mid to low”.
Volume ( vph) ( A)
N – [# lanes in conflict area] “ Heavy” Criteria “ Mid to Low” Criteria
Non- weaving 1 lane Heavy = > 1,800; Mid to Low < 1,800
2 lanes Heavy = > 3,600; Mid to Low < 3,600
3 lanes Heavy = > 5,400; Mid to Low < 5,400
4 lanes Heavy = > 7,200; Mid to Low < 7,200
Volume ( vph) ( B)
# lanes in conflict area “ Heavy” Criteria “ Mid to Low” Criteria
All weaving 1 Heavy = > 1,000; Mid to Low < 1,000
2 Heavy = > 2,000; Mid to Low < 2,000
3 Heavy = > 3,000; Mid to Low < 3,000
( A) + ( B) = N ( number of lanes in weaving section)
For instance, for the second “ conflict area” figure, there are two lanes in the conflict area and ( 4- 2) =
two lanes designated for non- weaving traffic.
The thresholds for “ heavy” and “ mid to low” traffic for non- weaving volumes were determined by
assuming that a freeway lane is operating at or near capacity if volumes are 1,800 vehicles per hour
( vph) or greater. The thresholds for “ heavy” and “ mid to low” traffic for weaving volumes were
determined by assuming that a freeway on- or off- ramp lane is operating at or near capacity if
volumes are 1,000 vph or greater.
19
CHAPTER 4
THE STUDY DATABASE
The evaluation of the existing methodologies for the analysis of weaving sections should be based on
empirical data representing a wide range of conditions. This Chapter describes the development of
the study database. We will select the test sites to be used in the project from the above database that
are most typical of California conditions and represent a range of design and traffic characteristics.
4.1 Data Sources
The following data sources were identified from previous studies at UC Berkeley, Caltrans staff,
other UC Berkeley studies and contacts with researchers and practitioners:
• California Weaving Studies Database [ 7]: As it was mentioned in the literature review, a
series of studies were undertaken by the Institute of Transportation Studies at University of
California, Berkeley to develop new weaving analysis procedures for major weaving sections.
Data from eight weaving sections were collected and analyzed.
• The California Ramp Weaving Studies Database [ 22]: Caltrans collected data in the early
90’ s on about 20 weaving sites consisting of a one lane on- ramp followed by a single lane
off- ramp with a continuous auxiliary lane.
• The NGSIM Data [ 23]: Detailed data on vehicle trajectories were collected on two California
weaving sites as part of the Next Generation Simulation ( NGSIM) program.
• The NCHRP 3- 75 Database [ 2]: This data were collected as part of a national project to
develop a new weaving analysis procedure for the next edition of the Highway Capacity
Manual ( HCM 2010).
• The Caltrans District 5 Database [ 24]: Caltrans District 5 staff provided to the research team
data on weaving sections mostly along US- 101 in Santa Barbara.
Each of the databases identified are described below:
4.1.1 California Studies — Major Weaving Sections
Table 4.1 shows the eight test sites where data were collected as part of the University of California
weaving studies in the early 90s. The schematics for each site are shown in Appendix A. The data
were collected using video and processed to obtain volumes per traffic movement, speeds of weaving
and non- weaving vehicles and lane distribution of component flows.
All the sites are major weaving sections with more than one on or off- ramps, typical of urban freeway
weaving sites. The I- 10EB_ LA was excluded from the database because it is six lane wide, and the
proposed matrix considers up to five lanes. The rest of the sites are all five lanes, except the SR-
92WB site which is three lanes wide.
The data were reviewed for accuracy and coded into the study database for further analysis. There
are a total of seven test sites and 27 data points of volume/ speed conditions.
20
Table 4.1 California Weaving Studies-- Major Weaving Sections
N: Number of lanes in the weaving section
L: length of the weaving section ( ft)
C: Weaving section configuration per HCM2000
V: Total volume in the weaving section ( vph)
VR: Ratio of weaving volume to total volume
S: Average speed in the weaving section ( mph)
4.1.2 California Studies — Ramp Weaves
Caltrans staff collected data on weaving sections in the early 90’ s using video recordings, as part of a
study to evaluate the accuracy of the Level D methods. All the data were collected on urban freeways
with a one lane on- and off- ramp connected with an auxiliary lane ( Type A per HCM2000). A total
of 20 weaving sections.
Site Location N L ( ft) C
V
( vph) VR
S
( mph)
US- 101SB Los Angeles 5 792 A 5909 0.29 53.5
5 792 A 5534 0.32 54.9
5 792 A 6463 0.29 60.4
I- 805NB San Diego 5 1371 B 7197 0.22 60.9
5 1371 B 6663 0.23 60.9
5 1371 B 6903 0.25 61.2
5 1371 B 6909 0.23 56.9
I- 10WB_ LA Los Angeles 5 1690 B 7751 0.31 58.0
5 1690 B 5986 0.31 62.9
5 1690 B 5941 0.32 62.1
5 1690 B 5832 0.33 62.3
5 1690 B 6427 0.33 60.5
I- 10WB_ SB San Bernardino 5 1989 B 4020 0.25 59.0
5 1989 B 3822 0.25 60.2
5 1989 B 4612 0.25 65.6
SR- 92WB San Mateo 3 1400 B 3221 0.43 52.1
3 1400 B 2760 0.41 53.6
3 1400 B 3035 0.35 59.7
3 1400 B 4033 0.33 57.6
I- 10EB_ LA Los Angeles 6 1437 C 4622 0.37 53.0
6 1437 C 4389 0.40 51.9
6 1437 C 5800 0.34 57.4
6 1437 C 6411 0.34 57.0
6 1437 C 10102 0.37 45.7
US- 101NB Los Angeles 5 787 C 9684 0.43 49.2
5 787 C 9202 0.38 49.0
I- 280SB San Jose 5 1347 C 5665 0.30 67.8
5 1347 C 5130 0.32 67.1
5 1347 C 4720 0.31 62.7
5 1347 C 4997 0.31 65.9
5 1347 C 7092 0.27 64.2
5 1347 C 7391 0.28 61.4
21
Most of the data in each study site consisted of 5 minute volumes per movement. Speeds of weaving
and weaving vehicles were extracted for seven sites. Following review of the data we selected the
weaving sections where performance measures ( speeds) were available. Table 4.2 shows the selected
test sites from the Caltrans ramp weaves study.
Note that all weaving sections are five lanes wide typical of urban freeways in Southern California.
Also, at the time of the data collection there was a 55 mph posted speed limit on all locations. The
final ramp weaves database consists of seven sites and 28 data points of volumes and speeds
Table 4.2 California Weaving Studies— Ramp Weaves
I- 5SB San Diego 5 1255 5868 0.21 55.0
5 1255 6132 0.21 54.6
5 1255 6240 0.20 54.6
5 1255 6156 0.16 54.7
5 1255 5940 0.17 54.8
5 1255 6192 0.16 54.8
SR- 60EB Los Angeles 5 1100 9240 0.08 57.6
5 1100 8784 0.09 56.6
5 1100 8568 0.09 59.5
5 1100 5400 0.08 60.7
5 1100 5388 0.10 61.9
5 1100 5052 0.11 60.3
SR- 91WB Los Angeles 5 1895 5448 0.13 58.7
5 1895 5124 0.12 57.1
5 1895 5592 0.11 53.6
I- 10WB Los Angeles 5 777 4428 0.17 56.9
5 777 4524 0.15 58.0
5 777 4800 0.16 57.1
SR- 91EB Los Angeles 5 845 6612 0.13 59.0
5 845 6084 0.14 58.4
5 845 6396 0.10 58.6
US- 101NB Los Angeles 5 808 6600 0.23 50.2
5 808 6744 0.19 51.6
5 808 6636 0.18 55.2
I- 110SB Los Angeles 5 610 7716 0.07 54.8
5 610 7488 0.07 54.0
5 610 7440 0.09 53.8
Site Location N V
L ( pcph) S
VR ( mph)
N: Number of lanes in the weaving section
L: length of the weaving section ( ft)
V: Total volume in the weaving section ( passenger cars/ hr)
VR: Ratio of weaving volume to total volume
S: Average speed in the weaving section ( mph)
22
4.1.3 The NGSIM Data Sets
Detailed data on weaving sections have been collected as part of the of NGSIM sponsored by FHWA.
The NGSIM database consists of vehicle trajectories and aggregate loop detector data from two
freeway sites in California: I- 80EB in San Francisco Bay area and US- 101NB in Los Angeles. The
objective of the data collection is to obtain highly detailed data to study vehicle interactions in car-following
and lane changing. A total of 11, 779 vehicle trajectories are available. The format is
vehicle ID, lane and position at 0.1 sec intervals.
The I- 80EB site is part of the Berkeley Highway Laboratory ( BHL). The site is Type B weaving
section per HCM2000 with a length of 1,650 ft; there are six freeway lanes entering the weaving
section and five freeway lanes leaving it. Lane 1 is an HOV lane. Data on freeway operations are
collected from multiple video cameras located on top of a 300 ft building adjacent to the freeway, and
loop detectors placed approximately at 0.3 mile intervals on each freeway lane. The video data are
processed through a machine vision system that tracks each vehicle as travels through the section and
produces vehicle trajectories.
The US- 101NB has five through lanes with a continuous auxiliary lane ( Type A weaving
section). The data include 45 minutes of vehicle trajectories in transition ( 7: 50 to 8; 05 am)
and congestion ( 8: 05 – 8: 35 am).
4.1.4 The NCHRP 3- 75 Database
The data base for NCHRP Project 3- 75 consists of 14 sites in four different regions of the country.
The data on traffic volumes and speeds were collected using video recordings. The data from the two
California sites ( I- 80EB in Emeryville and US- 101NB in Los Angeles) were provided by the NGSIM
program ( described in Section 4.1.3).
Table 4.3 shows the basic characteristics of the test sites, and Appendix B includes the schematics of
each test site. A total of 157 5- minute data periods exist, which are also configured as 52 15- minute
data periods. Most of the weaving sections are Type B per HCM2000 with five lanes.
Following review of the sites and the data provided, the following sites were excluded from the study
database ( shown as shaded cells in Table 4.3):
Site # 3 I- 270WB: This is essentially a one- lane connector plus an auxiliary lane, with a speed
of 35 mph. It is not a typical freeway weaving site.
Site # 4 Los Angeles US- 101NB ( NGSIM 2): This is a six lane weaving section. The
evaluation of existing methods for developing the performance matrix includes weaving
sections with up to five lanes.
Site # 6 I- 95NB: This is a two sided weaving section. The scope of the study includes only
one sided weaving section.
The final NCHRP data sets included in the study database consist of 11 sites with 113 data points of
volumes and speeds.
23
Table 4.3 The NCHRP 3- 75 Database [ 2]
Type: Weaving section configuration per HCM2000
L: length of the weaving section ( ft)
N: Number of lanes in the weaving section
Data Periods: Number of 5 minute data periods available
4.1.5 Caltrans District 5 Data
Caltrans District 5 provided data on freeway weaving sections mostly along US- 101 in Santa Barbara.
The data include geometrics and traffic volumes. Speed data are not available. Most of the sites are
three lanes wide and do not include auxiliary lanes. Table 4.4 shows basic information of the District
5 datasets. There are a total of 11test sites with 22 data points.
Site Location Type
Length
( ft)
# Lanes
( N)
Data
Periods
1 Emeryville, CA
I- 80 EB ( NGSIM1) B 1,605 6* 6
2 Portland, OR
I- 405 EB B 650 4 6
3 Ohio
I- 270 WB A 540 2 24
4 Los Angeles, CA
US- 101 NB ( NGSIM2) A 698 6 9
5 Miami, FL
I- 95 SB B 1,120 5 7
6 Miami, FL
I- 95 NB C 1,380 4 12
7 Baltimore, MD
MD- 100 EB A 465 3 12
8 Baltimore, MD
MD- 100 EB B 1,085 3 12
9 Phoenix, AZ
SR 202 EB B 2,110 5* 12
10 Phoenix, AZ
SR 101 EB C 2,235 5 12
11 Phoenix, AZ
SR 101 WB B 2,010 4 12
12 Portland, OR
SR 217 SB B 2,820 3 12
13 Portland, OR
I- 5 SB B 1,565 4 12
14 Portland, OR
I- 5 SB B 2,060 5 12
* Includes an HOV lane
Excluded from the study database
24
Table 4.4 Caltrans District 5 Database
N: Number of lanes in the weaving section
Nb: Number of approaching freeway lanes
L: length of the weaving section ( ft)
V: Total volume in the weaving section ( passenger cars/ hr)
VR: Ratio of weaving volume to total volume
4.2 The Study Database
The final database consists of 36 test sites and a total of 189 data points. Table 4.5 shows the
available datasets per geometric characteristics ( number of lanes and configuration), and Table 4.6
shows a mapping of the data available to the cells in the weaving performance matrix.
Site HCM
Type N Nb L V
( pcph) VR
US- 101SB_ SB_ 1 A 3 2 2379 3242 0.61
3 2 2379 2451 0.66
US- 101SB_ SB_ 4 A 4 3 1181 5093 0.32
4 3 1181 5658 0.28
US- 101NB_ SB_ 1 3 3 3199 5617 0.32
3 3 3199 6091 0.31
US- 101NB_ SB_ 2 3 3 3084 4957 0.16
3 3 3084 5376 0.15
US- 101NB_ SB_ 3 3 3 3986 5136 0.26
3 3 3986 5094 0.29
US- 101NB_ SB_ 4 3 3 3773 3134 0.46
3 3 3773 3867 0.39
US- 101SB_ SB_ 2 2 2 3199 3422 0.31
2 2 3199 3412 0.44
US- 101SB_ SB_ 3 3 3 3281 4852 0.27
3 3 3281 5809 0.28
US- 101SB_ SB_ 5 3 3 1558 5243 0.38
3 3 1558 5102 0.40
US- 101SB_ SB_ 6 3 3 3445 4623 0.43
3 3 3445 4426 0.44
US- 101SB_ SB_ 7 3 3 1312 3958 0.24
3 3 1312 3665 0.25
25
Table 4.5 The Study Database: Geometric Characteristics
XXX: NCHRP/ NGSIM Data XXX: California Major Weaving Sites Data
XXX: California Ramp Weaves Data
XXX: Caltrans District 5 Data
Number of Lanes in the Weaving Section
CONFIGURATION N= 2 N= 3 N= 4 N= 5
NO AUXILIARY LANE US- 101SB_ SB3
US- 101SB_ SB2 US- 101SB_ SB4
US- 101SB_ SB5
US- 101SB_ SB6
US- 101SB_ SB7
US- 101NB_ SB1
US- 101NB_ SB2
US- 101NB_ SB3
RAMP WEAVE US- 101SB
HCM TYPE A MD- 100EB_ A US- 101SB_ SB4 SR- 91EB
US- 101SB_ SB1 SR- 91WB
I- 110EB
US- 101NB
I- 10WB
SR- 60EB
I- 5SB
MAJOR WEAVE MD- 100EB_ B I- 405EB I- 80EB
BALANCED SR- 92WB SR- 101WB I- 95SB
HCM TYPE B SR- 217SB I- 5SB_ A SR- 202EB
I- 5SB_ B
I- 805NB
I- 10WB_ SB
I- 10WB_ LA
MAJOR WEAVE US- 101NB
UNBALANCED SR- 101EB
HCM TYPE C I- 280SB
26
Table 4.6 Mapping the Datasets to the Performance Matrix
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy US- 101SB_ SB2 ( 2)
Weaving: Heavy
Non- Weaving: Heavy
Weaving: Mid to Low
Non- Weaving: Mid to Low
Weaving: Heavy
Non- Weaving: Mid to Low
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy US- 101SB_ SB3 ( 1)
Weaving: Heavy US- 101NB_ SB1 ( 2)
US- 101NB_ SB3 ( 2)
Non- Weaving: Heavy US- 101NB_ SB2 ( 2) MD- 100EB_ A( 10) SR- 92WB ( 3) SR- 217SB ( 12)
Weaving: Mid to Low MD- 100EB_ B( 11)
Non- Weaving: Mid to Low US- 101SB_ SB5 ( 2) US- 101SB_ SB4 ( 2)
Weaving: Heavy US- 101SB_ SB3 ( 1)
US- 101SB_ SB6 ( 2)
Non- Weaving: Mid to Low US- 101SB_ SB7 ( 2) MD- 100EB_ A( 2) US- 101SB_ SB1 ( 2) SR- 92WB ( 1)
Weaving: Mid to Low MD- 100EB_ B( 1)
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy I- 5SB_ A ( 2)
Weaving: Heavy SR- 101WB ( 7)
Non- Weaving: Heavy US- 101SB_ SB4 ( 1) I- 405 EB ( 3) I- 5SB_ A ( 5)
Weaving: Mid to Low SR- 101WB ( 3)
SR- 202EB ( 12)
Non- Weaving: Mid to Low I- 5SB_ A ( 3)
Weaving: Heavy
Non- Weaving: Mid to Low US- 101SB_ SB4 ( 1) I- 405 EB ( 3) I- 5SB_ A ( 2)
Weaving: Mid to Low SR- 101WB ( 2)
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy I- 80EB ( 5) US- 101NB ( 2)
Weaving: Heavy
Non- Weaving: Heavy US- 101SB ( 2) SR- 60EB ( 3) I- 805NB ( 1) I- 280SB ( 3)
Weaving: Mid to Low SR- 91EB ( 2) I- 95SB ( 6) SR- 101EB ( 5)
I- 110SB ( 3)
US- 101NB ( 1)
Non- Weaving: Mid to Low I- 10WB_ LA ( 3)
Weaving: Heavy I- 5SB_ B ( 1)
I- 80EB ( 1)
Non- Weaving: Mid to Low US- 101SB ( 1) SR- 60EB ( 3) I- 805NB ( 3) I- 280SB ( 3)
Weaving: Mid to Low SR- 91EB ( 1) SR- 91WB ( 3) I- 10WB_ LA ( 2) SR- 101EB ( 7)
I- 10WB ( 3) I- 5SB ( 6) I- 10 WB_ SB ( 3)
US- 101NB ( 2) I- 5SB_ B ( 9)
I- 95SB ( 1)
XXX: NCHRP/ NGSIM Data XXX: California Major Weaving Sites Data XXX: California Ramp Weaves Data XXX: Caltrans District 5 Data
27
CHAPTER 5
EVALUATION OF EXISTING METHODS
The existing weaving analysis tools were applied to the all the datasets shown in Table 4.6. The
predictions from each method were compared to the field measurements within a site and across all
sites to determine the strengths and limitations of each analysis tool. The following sections show
sample results from the extensive analyses performed.
5.1 Application of Existing Methods to Field Data
Figure 5.1 shows a comparison of the field measured vs. HCM200 predicted weaving speeds by
configuration of the weaving section. The data are from the California major weaving sites. It can
be seen that HCM2000 under- predicts the weaving speeds in all the sites, especially for Type A ( ramp
weaves) sections. HCM2000 predictions are close to the field values for Type B weaving sections.
In terms of overall average speeds in the weaving section, the mean speed difference between field
and HCM2000 estimates are 25% for Type A, 2% for Type B and 10% for Type C weaving section
configurations respectively.
Figure 5.1 Measured vs. HCM2000 Predicted Weaving Speed
Comparisons with the NCHRP datasets show that both HCM2000 and Leisch methods predict a LOS
in the weaving area different than the field in about 40 % of the test cases. The Level D method
predicts that approximately 50% of the weaving data sets will “ fail” ( Level of Service D or worse);
field data indicate that LOS D or worse occurs in 23% of the cases.
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
measured Sw
estimated Sw
Sw type A Sw type B Sw type C
28
5.2 Santa Barbara Data Sets
The Caltrans District 5 datasets do not have field measured speeds for comparison with the model
predictions. Therefore, we compared the predictions of the methods with the same data. Figure 5.2
shows the difference in LOS predicted by Level D vs. HCM2000 and Leisch methods for each
dataset. It can be seen that Level D and Leisch methods are in agreement in 16 of the cases, i. e., in
73% of the cases Level D and Leisch predict the same LOS. Only in six datasets ( 27%) the two
methods vary by one LOS designation. However, the differences are greater in the case of Level D
vs. HCM2000. The two methods predict different LOS in 14 out of 22 datasets ( 64%); four datasets
have a difference in traffic performance by two LOS designations.
Figure 5.2 Comparison of Weaving Analysis Methods— Caltrans District 5 Data
5.3 Synthetic Data Sets
It can be seen from Table 4.6 that significant gaps in field data exist. Despite the efforts of the
research team and the project advisory committee it was not possible to obtain additional real- world
data sets. Therefore, we had to develop synthetic data sets representing operating conditions lacking
field data. Table 5.1 shows the scenarios for which synthetic data were created. There are a total of
113 scenarios. Under each scenario, three synthetic datasets were created for a total of 339 datasets.
For example, under scenario 20 ( A three lane weaving section Type A with medium length and heavy
non- weaving and weaving volumes), we generated three datasets for the given configuration and total
volumes by assuming: i) balanced weaving volumes, ii) unbalanced weaving volumes with high on-ramp
to freeway volume and low freeway to off- ramp volume, and iii) unbalanced weaving volumes
with low on- ramp to freeway volume and high freeway to off- ramp volume.
0
1
2
3
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
DATA SET #
LOS DIFFERENCE
LEVEL D vs. HCM
LEVEL D vs. LEISCH
29
XXX: NCHRP/ NGSIM Data XXX: California Major Weaving Sites Data XXX: California Ramp Weaves Data XXX: Caltrans District 5 Data XX Synthetic Data
Table 5.1 Synthetic Data Sets No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy 98 102 US- 101SB_ SB2 ( 2)
Weaving: Heavy
Non- Weaving: Heavy 99 103 106
Weaving: Mid to Low
Non- Weaving: Mid to Low 100 104 107
Weaving: Heavy
Non- Weaving: Mid to Low 101 105 108
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy US- 101SB_ SB3 ( 1)
Weaving: Heavy 78 2 US- 101NB_ SB1 ( 2) 18 20 23 39 43 45
US- 101NB_ SB3 ( 2)
Non- Weaving: Heavy 79 46 US- 101NB_ SB2 ( 2) MD- 100EB_ A( 10) 21 24 40 SR- 92WB ( 3) SR- 217SB ( 12)
Weaving: Mid to Low MD- 100EB_ B( 11)
Non- Weaving: Mid to Low US- 101SB_ SB5 ( 2) US- 101SB_ SB4 ( 2)
Weaving: Heavy 80 US- 101SB_ SB3 ( 1) 19 22 25 41 44 47
US- 101SB_ SB6 ( 2)
Non- Weaving: Mid to Low 81 US- 101SB_ SB7 ( 2) 1 MD- 100EB_ A( 2) US- 101SB_ SB1 ( 2) 26 42 SR- 92WB ( 1) 48
Weaving: Mid to Low MD- 100EB_ B( 1)
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy I- 5SB_ A ( 2)
Weaving: Heavy 82 12 86 27 14 31 16 SR- 101WB ( 7) 49 59 63 67
Non- Weaving: Heavy US- 101SB_ SB4 ( 1) I- 405 EB ( 3) I- 5SB_ A ( 5)
Weaving: Mid to Low 83 109 87 28 32 SR- 101WB ( 3) 50 60 64 68
SR- 202EB ( 12)
Non- Weaving: Mid to Low 84 13 88 29 15 33 17 I- 5SB_ A ( 3) 51 61 65 69
Weaving: Heavy
Non- Weaving: Mid to Low 85 110 89 30 US- 101SB_ SB4 ( 1) 34 I- 405 EB ( 3) I- 5SB_ A ( 2) 52 62 66 70
Weaving: Mid to Low SR- 101WB ( 2)
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000- 2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy I- 80EB ( 5) US- 101NB ( 2)
Weaving: Heavy 90 3 94 4 6 35 53 55 8 74
Non- Weaving: Heavy US- 101SB ( 2) SR- 60EB ( 3) I- 805NB ( 1) I- 280SB ( 3)
Weaving: Mid to Low 91 10 95 SR- 91EB ( 2) 36 112 I- 95SB ( 6) 56 71 SR- 101EB ( 5) 75
I- 110SB ( 3)
US- 101NB ( 1)
Non- Weaving: Mid to Low I- 10WB_ LA ( 3)
Weaving: Heavy 92 111 96 5 7 37 54 I- 5SB_ B ( 1) 57 72 9 76
I- 80EB ( 1)
Non- Weaving: Mid to Low US- 101SB ( 1) SR- 60EB ( 3) I- 805NB ( 3) I- 280SB ( 3)
Weaving: Mid to Low 93 11 97 SR- 91EB ( 1) SR- 91WB ( 3) 38 113 I- 10WB_ LA ( 2) 58 73 SR- 101EB ( 7) 77
I- 10WB ( 3) I- 5SB ( 6) I- 10 WB_ SB ( 3)
US- 101NB ( 2) I- 5SB_ B ( 9)
I- 95SB ( 1)
30
Since there are no field data available for comparison with each method’s predictions, emphasis is
placed on comparing the methods’ results with the same input data, similar to the analysis performed
for the Caltrans District 5 datasets.
Figure 5.3 shows the difference between the predicted LOS from HCM2000 and Leisch methods for
each dataset. The value of (- 1) in the Figure means that HCM2000 predicts better performance by
one LOS than the Leisch method. It can be seen that Leisch in general predicts worse Level of
Service than HCM2000. HCM2000 predicts worse performance by two LOS designations for short
Type A weaving sections ( ramp weaves) with heavy volumes.
Figure 5.3 Comparison of Leisch vs. HCM2000— Synthetic Data Sets
Comparisons of Level D method with the Leisch and HCM2000 methods shows similar patterns as
in the case of Santa Barbara data sets, but there are significant variations among different cases. The
results are summarized in Table 5.2 in the same scenario groupings as in Table 5.1. It can be seen
that Level D predicts different LOS than HCM2000 and Leisch methods in the cases of ramp weaves
with auxiliary lanes. Note that scenarios with multiple on and off- ramps are not included because they
cannot be analyzed with the Level D method.
- 2
- 1
0
1
2
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97 100 103 106
DATA SET
DIFFERENCE IN PREDICTED LOS
31
Table 5.2 Differences in LOS Prediction -- Synthetic Data Sets
SCENARIOS *
Level D
vs. HCM
Level D
vs. Leisch
N= 2 98 102
No Auxiliary Lane 99 103 106
100 104 107 0 0
101 105 108
N= 3 78 2
No Auxiliary Lane 79 46 0 0
80
81 1
N= 4 82 12 86
No Auxiliary Lane 83 109 87 0 0
84 13 88
85 110 89
N= 5 90 3 94
No Auxiliary Lane 91 10 95 0 0
92 111 96
93 11 97
N= 3 18 20 23
Auxiliary Lane 21 24 3 0
19 22 25
26
N= 4 27 14 31
Auxiliary Lane 28 32 2 3
29 15 33
30 34
N= 5 4 6 35
Auxiliary Lane 36 2 2
5 7 37
38
* Data Sets with multiple on/ off ramps not included.
Level D method not designed for multiple on/ off ramps
32
CHAPTER 6
CONCLUSIONS
6.1 Summary of the Study Findings
The objectives of the study were to a) evaluate the existing weaving analysis procedures to determine
under which conditions the “ best available” tools are most effective, and b) develop a new weaving
analysis method. The end product of the study is an analysis performance matrix that gives guidance
on which of the existing analysis methods should be applied for a particular weaving section under
study. A new simulation model was also developed to predict the performance of weaving areas.
The HCM2000, Leisch and Level D methods were evaluated using field data from 36 real- world
weaving sections for a total of 189 data points of speed and volumes. The analysis of the results
identified the strengths and limitations of each method in determining the performance of a freeway
weaving section for a range of operating conditions.
Additional analyses were performed by applying the selected analysis methods to synthetic datasets
for the geometric and operating conditions that field data were not available. A total of 339 datasets
were created. The analysis of the results focused on the consistency of the predictions from each
analysis method.
6.2 Recommendations
The proposed weaving analysis performance matrix for each method are shown in Appendix C. The
performance matrix is based on the comparison between observed ( when available) and predicted
conditions from each method. For those operating conditions that field data were not available, the
recommended indicators of performance (“ good”, “ inconsistent”, “ and poor”) are based on the
comparison of the results from the different method. Also, shown are cells with synthetic data
( indicated by X), cells with limited data ( indicated XX, typically one real- world data set), and cells
with multiple field data sets ( indicated by XXX). The performance matrix for each method is also
submitted as an Excel spreadsheet and it can be readily updated should more data and analyses
become available.
Note that the existing methods appear to have the same performance on several design and operating
conditions. In such situations, Caltrans engineers should follow existing guidance as in the Caltans
Design Manual, i. e., apply the Leisch method and check operating conditions with Level D method if
applicable, because of the complexity in the HCM2000 method relative to the other methods.
6.3 Future Research
The following research activities are suggested towards improving the design and analysis methods
for freeway weaving sections:
a) Update and refinement of the proposed performance matrices: Several cells in the proposed
performance matrix for each method are lacking field data on traffic performance. There is a
need to obtain additional data and update these matrices.
b) Weaving section capacity: The existing methods do not directly provide estimates of the
capacity of the weaving section. There is a need to evaluate the accuracy of the methods
regarding capacity prediction, by selecting weaving sites that are bottlenecks and comparing
predicted and observed queue discharge flows.
33
c) Evaluation of New HCM Weaving Analysis Method: A new weaving analysis method has
been developed as part of the NCHRP study and has been adopted for the next edition of the
HCM in late 2010. This method is simpler to use than the existing HCM2000 method
because it does not have separate procedures per each weaving configuration and constrained
vs. unconstrained operation. This method should be evaluated using the same data used in
this study to determine if it an appropriate analysis tool to be used by Caltrans staff.
d) Evaluation and Refinement of the Weaving Software Tool: a simulation model was developed
in this study to analyze weaving sections based on field data from two weaving bottlenecks.
This tool can be potentially used on several freeway operations analyses ( e. g., auxiliary lane
lengths, ramp metering) provided that accurately represents real- world operating conditions.
There is a need for systematic model testing with field data and refinement in order to be a
practical analysis tool by Caltrans staff.
34
e) REFERENCES
1. California Department of Transportation “ Highway Design Manual,” Sixth Edition, Sacramento, CA, 2007.
2. Roess, R. et al, “ Analysis of Freeway Weaving Sections,” Final Report, NCHRP Project 3- 75,
Transportation Research Board, January 2008.
3. Highway Capacity Manual,” Special Report 87, Transportation Research Board, Washington DC, 1965.
4. Normann, O. K., “ Operation of Weaving Areas,” Highway Research Bulletin 167, Transportation Research
Board, Washington DC, 1957.
5. Hess, J., “ Capacities and Characteristics of Ramp- Freeway Connections,” Highway Research Record 27,
Highway Research Board, National Academy of Sciences, Washington DC, 1963.
6. Moskowitz, K., and Newman, L., Notes on Freeway Capacity, Traffic Bulletin 4, California Department of
Highways, Sacramento CA, July 1962.
7. Cassidy, M., et al, A Proposed Technique for the Design and Analysis of Major Freeway Weaving Sections,
Research Report UCB- ITS- RR- 90- 16, Institute of Transportation Studies, University of California
Berkeley, Berkeley CA, 1990.
8. Cassidy M, and May, A. D., “ Proposed Analytic Technique for Estimating Capacity and Level of Service of
Major Freeway Weaving Sections,” Transportation Research Record # 1320, Washington DC, 1991.
9. Windover, J., and May, A. D., “ Revisions to Level D Methodology of Analyzing Freeway Ramp- Weaving
Sections,” Transportation Research Record # 1457, Washington DC, 1995.
10. Leisch, J., unpublished studies, 1958- 1964.
11. Leisch, J., “ Completion of Procedures for Analysis of and Design of Traffic Weaving Areas,” Final Report,
Vols 1 and 2, U. S. Department of Transportation, Federal Highway Administration, Washington DC, 1983.
12. Pignataro, L, et al, “ Weaving Areas – Design and Analysis,” National Cooperative Highway Research
Report 159, Transportation Research Board, Washington DC, 1975.
13. Roess, R., et al, “ Freeway Capacity Analysis Procedures,” Final Report, Project No. DOT- FH- 11- 9336,
Polytechnic University, Brooklyn NY, 1978.
14. Reilly, W., et al, “ Weaving Analysis Procedures for the New Highway Capacity Manual,” Technical
Report, Contract No. DOT- FH- 61- 83- C- 00029, Federal Highway Administration, Washington DC, 1984.
15. Highway Capacity Manual, Special Report 209, Transportation Research Board, Washington DC, 1985.
16. Roess, R, " Development of Weaving Area Analysis Procedures for the 1985 Highway Capacity Manual,"
Transportation Research Record 1112, Transportation Research Board, Washington DC, 1987.
17. “ Weaving Zones,” Draft Report, NCHRP Project 3- 55( 5), Viggen Corporation, Sterling VA, 1998.
18. Highway Capacity Manual, 4th Edition, Transportation Research Board, Washington DC, 2000.
19. Fazio, J., “ Development and Testing of a Weaving Operational Design and Analysis Procedures,” M. S.
Thesis, University of Illinois at Chicago Circle, Chicago IL, 1985.
20. Lertworawanich, P., and Elefteriadou, L., “ Capacity Estimations for Type B Weaving Areas Based Upon
Gap Acceptance,” Transportation Research Record # 1776, Washington DC, 2002.
21. Skabardonis, A., “ Simulation of Freeway Weaving Areas,” Transportation Research Record 1802,
Transportation Research Board, Washington DC, 2002.
22. Fong, H. K., and F. D. Rooney, “ Weaving Areas Near One Lane Ramps,” Division of Traffic Operations,
California Department of Transportation, Sacramento, 1990.
23. Skabardonis A., and V. Alexiadis, “ Traffic Data through the Berkeley Highway Laboratory,” Proceedings,
Traffic Modeling Workshop, Federal Highway Administration, Sedona, AZ, September 2005.
24. Toh, S., and S. Eades, “ Data on Weaving Areas: Santa Barbara,” Unpublished Communication, 2007.
35
APPENDIX A: CALIFORNIA STUDIES- MAJOR WEAVING SECTIONS
WB10LA
Los Angeles, WB I- 10
Garvey on – I605 Off
WB10SB
San Bernardino, WB I- 10
Etawanda on – I15 Off
EB10LA
Los Angeles, EB I10
I605 on – Frazer Off
NB805
San Diego, NB I- 805
University on – El Cajon Off
NB101
Los Angeles, NB US 101
Los Angeles on – I110 Off
SB280
San Jose, SB I- 280
I880 on – Bascom off
WB92
San Mateo, WB SR92
Ralston on – I280 NB off
SB101
Los Angeles, SB US 101
I110 on – Broadway Off
36
APPENDIX B: NCHRP 3- 75 TEST SITES
Figure B1. Site 1: I- 80EB; Powell St to Ashby Avenue, Emeryville, CA
Figure B2. Site 2: I- 405EB; 6th Avenue to 12th Avenue, Portland, OR
Figure B3. Site 3: I- 270WB; I- 270 & US23, Franklin Co, OH
Figure B4. Site 4: US- 101NB; Ventura Blvd to Cahuenga Blvd, Los Angeles, CA
1 HOV HOV 1 HOV HOV 1
2 2 2
3 3 3
4 4 4
5 5 5
6 6
1605'
1 1 1
3 2 2
3 3 3
4
650'
1 1 1
2
570'
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
6
698'
37
APPENDIX B: NCHRP 3- 75 TEST SITES ( Cont.)
Figure B5. Site 5: I- 95SB; NW 135th St to N. Miami Blvd., Miami, FL
Figure B6. Site 6: I- 95NB; SE 8th St to SE 1st St., Miami, FL
Figure B7. Site 7: MD- 100EB; I- 95SB to I- 95NB, Baltimore, MD
Figure B8. Site 8: MD- 100EB; I- 95NB to US Rte 1, SB, Baltimore, MD
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
6
1,120
1 1 1
2 2 2
3 3 3
4 4 4
5
1,380
1 1 1 1
2 2 2 2 2
3 3 3
465
1 1 1
2 2 2
3 3 3
4
1,085
38
APPENDIX B: NCHRP 3- 75 TEST SITES ( Cont.)
Figure B9. Site 9: SR202 EB; 32nd St to 40th St, Phoenix, AZ
Figure B10. Site 10: SR201 EB; SR 51 to Tatum Blvd, Phoenix, AZ
Figure B11. Site 11: SR101 WB; Tatum Blvd to SR 51, Phoenix, AZ
Figure B12. Site 12: SR217SB; SW Pacific Hwy to SW 72nd Ave, Portland, OR
0 HOV HOV 0 HOV HOV 0
1 1 1
2 2 2
3 3 3
4 4 4
5
2,110
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
2,235
1 1 1
2 2 2
3 3 3
4 4 4
5
2,015
1 1 1
2 2 2
3 3 3
4
2,820
39
APPENDIX B: NCHRP 3- 75 TEST SITES ( Cont.)
Figure B13. Site 13: I- 5SB; SW Nyberg Rd to I- 205, Portland, OR
Figure B14. Site 14: I- 5SB; SR- 217 to Upper Boones Ferry Rd, Portland, OR
1 1 1
2 2 2
3 3 3
4 4 4
5
1,565
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
6
2,060
40
APPENDIX C
WEAVING ANALYSIS PERFORMANCE MATRIX
C1. HCM2000 Method
C2. Leisch Method
C3. Level D Method
C4. Weaving Analysis Performance Matrix – Recommended Methodology
41
APPENDIX C
C1. Weaving Analysis Performance Matrix— HCM2000 Method
Weaving Analysis Performance Matrix
Methodology: HCM2000
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XX
Weaving: Heavy
Non- Weaving: Heavy X X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XXX X X X X X X
Weaving: Heavy
Non- Weaving: Heavy X X XX XX X X X XXX XX
Weaving: Mid to Low
Non- Weaving: Mid to Low X XX XXX X X X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X XX X XX XX X X XXX X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X XXX X X X X
Weaving: Heavy
Non- Weaving: Heavy X X X X XX X XX XXX X X X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X X XX X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X X XX X XX XXX X X X X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X X XX X XX X X
Weaving: Heavy
Non- Weaving: Heavy X X X XXX XX X X XXX X X XXX X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X X XXX X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X XXX XXX X X XXX X X XXX X
Weaving: Mid to Low
Notes
LEGEND: Methodology's prediction of performance * All weaving sections considered are single side, right side configurations
( i. e. does not include left side or two sided configurations)
= Poor, inconsistent results
= Fair; sometimes inconsistent results ** Non- weaving Vols in vph: N – [# lanes in conflict area] =
= Good and consistent results 1 lane: Heavy = > 1,800; Mid to Low < 1,800
2 lanes: Heavy = > 3,600; Mid to Low < 3,600
3 lanes: Heavy = > 5,400; Mid to Low < 5,400
4 lanes: Heavy = > 7,200; Mid to Low < 7,200
** Weaving Vols in vph: [# lanes in conflict area] =
1 lane: Heavy = > 1,000; Mid to Low < 1,000
2 lanes: Heavy = > 2,000; Mid to Low < 2,000
3 lanes: Heavy = > 3,000; Mid to Low < 3,000
X Synthetic Data
XX Limited Data
XXX Multiple Data Sets
42
APPENDIX C
C2. Weaving Analysis Performance Matrix— Leisch Method
Weaving Analysis Performance Matrix
Methodology: LEISCH
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XX
Weaving: Heavy
Non- Weaving: Heavy X X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XXX X X X X X X
Weaving: Heavy
Non- Weaving: Heavy X X XX XX X X X XXX XX
Weaving: Mid to Low
Non- Weaving: Mid to Low X XX XXX X X X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X XX X XX XX X X XXX X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X X XXX X X X X
Weaving: Heavy
Non- Weaving: Heavy X X X X XX X XX XXX X X X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X X XX X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X X XX X XX XXX X X X X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X X XX X XX X X
Weaving: Heavy
Non- Weaving: Heavy X X X XXX XX X X XXX X X XXX X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X X XXX X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X XXX XXX X X XXX X X XXX X
Weaving: Mid to Low
Notes
LEGEND: Methodology's prediction of performance * All weaving sections considered are single side, right side configurations
( i. e. does not include left side or two sided configurations)
= Poor, inconsistent results
= Fair; sometimes inconsistent results ** Non- weaving Vols in vph: N – [# lanes in conflict area] =
= Good and consistent results 1 lane: Heavy = > 1,800; Mid to Low < 1,800
2 lanes: Heavy = > 3,600; Mid to Low < 3,600
3 lanes: Heavy = > 5,400; Mid to Low < 5,400
4 lanes: Heavy = > 7,200; Mid to Low < 7,200
** Weaving Vols in vph: [# lanes in conflict area] =
1 lane: Heavy = > 1,000; Mid to Low < 1,000
2 lanes: Heavy = > 2,000; Mid to Low < 2,000
3 lanes: Heavy = > 3,000; Mid to Low < 3,000
X Synthetic Data
XX Limited Data
XXX Multiple Data Sets
43
APPENDIX C
C3. Weaving Analysis Performance Matrix— Level D Method
Weaving Analysis Performance Matrix
Methodology: LEVEL D
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XX
Weaving: Heavy
Non- Weaving: Heavy X X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X XXX X X X
Weaving: Heavy
Non- Weaving: Heavy X X XX XX X X
Weaving: Mid to Low
Non- Weaving: Mid to Low X XX XXX X X X
Weaving: Heavy
Non- Weaving: Mid to Low X XX X XX XX X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X
Weaving: Heavy
Non- Weaving: Heavy X X X X XX X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X X XX X
Weaving: Mid to Low
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous
Weave Length
(> 2500')
Non- Weaving: Heavy X X X X X X
Weaving: Heavy
Non- Weaving: Heavy X X X XXX XX X
Weaving: Mid to Low
Non- Weaving: Mid to Low X X X X X X
Weaving: Heavy
Non- Weaving: Mid to Low X X X XXX XXX X
Weaving: Mid to Low
Notes
LEGEND: Methodology's prediction of performance * All weave sections considered are single side, right side configurations
( i. e. does not include left side or two sided configurations)
= Poor, inconsistent results
= Fair; sometimes inconsistent results ** Non- weaving Vols in vph: N – [# lanes in conflict area] =
= Good and consistent results 1 lane: Heavy = > 1,800; Mid to Low < 1,800
2 lanes: Heavy = > 3,600; Mid to Low < 3,600
3 lanes: Heavy = > 5,400; Mid to Low < 5,400
4 lanes: Heavy = > 7,200; Mid to Low < 7,200
** Weaving Vols in vph: [# lanes in conflict area] =
1 lane: Heavy = > 1,000; Mid to Low < 1,000
2 lanes: Heavy = > 2,000; Mid to Low < 2,000
3 lanes: Heavy = > 3,000; Mid to Low < 3,000
METHOD NOT DESIGNED FOR MULTIPLE ON/ OFF RAMPS
X Synthetic Data
XX Limited Data
XXX Multiple Data Sets
44
Weaving Analysis Performance Matrix
Recommended Methodology
No. of Lanes in Weaving Section, N = 2
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D
No. of Lanes in Weaving Section, N = 3
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 LEISCH LEISCH LEISCH HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEVEL D LEVEL D LEVEL D LEISCH LEISCH LEISCH
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH HCM2000 HCM2000
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH LEISCH LEISCH
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH HCM2000 LEISCH
No. of Lanes in Weaving Section, N = 4
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 LEISCH LEISCH LEISCH HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEVEL D LEVEL D LEVEL D HCM2000 LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D HCM2000 LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D HCM2000 LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D HCM2000 HCM2000 HCM2000 LEISCH LEISCH LEISCH
No. of Lanes in Weaving Section, N = 5
Configuration* ---> No Auxiliary Lane, 1- lane on & off ramps With Aux. Lane, 1- lane on/ off ramps ( Type A) Balanced > 1- lane on & off ramps ( Type B) Unbalanced > 1- lane on & off ramps ( Type C)
Operational
Conditions( vols) **
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Short Weave
Length (< 1000')
Medium Weave
Length ( 1000-
2500')
Generous Weave
Length (> 2500')
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 LEISCH LEISCH HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEVEL D LEVEL D LEISCH/ LEVEL D LEISCH LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Heavy HCM2000 HCM2000 HCM2000 HCM2000 LEISCH HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEVEL D LEISCH/ LEVEL D LEISCH LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 LEISCH HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Heavy LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEVEL D LEISCH/ LEVEL D LEISCH LEISCH HCM2000 LEISCH LEISCH LEISCH
Non- Weaving: Mid to Low HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000 HCM2000
Weaving: Mid to Low LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH/ LEVEL D LEISCH LEISCH HCM2000 LEISCH LEISCH LEISCH
C4. Weaving Analysis Performance Matrix— Recommended Methodology
45
APPENDIX D
Lee, J. H., and M. C. Cassidy,
“ AN EMPIRICAL AND THEORETICAL STUDY OF FREEWAY WEAVE
BOTTLENECKS”
PATH Research Report UCB- ITS- PRR- 2009- 13
February, 2009.
         	  
  
            
                                          !     "    # $  %  #   &           
' "  (              !     "   )   "  *          "                       !     "    +    "    ) 
   "          " )   "   %     " &  $ &  " *  )  ,        "         "          " )   "      
' "             ,        "         "          " )        !  %  &      $    "         " -
   . *  "   "   .   .     .       .   /  *  .    . (     .   .    .        .    .    .      "    !  .
          *     "        * *    *                    "          " -      *  "   "       "   
"  *       !  .   /  *  .    .   0 *   ! . (     .   .   !  *    .   .    .      .   .    !     "   1 .     .
      .     . "   . *  "        .  .    "     2 .    *  0 *     " 2 .   .   &  !     " 1
  "  !  3                4      5 6  

CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
       	  
            	    
             
 	                     
  
     !     " # #   $ % % &   ' (
      	     "   )  #     	 
   #    	 
*        +   ,  -  
      * .        
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
APPENDIX D
D- 1
APPENDIX D
D- 2
􀀁
MOU/ TO 6304:
Part I: An Empirical and Theoretical Study
of
Freeway Weave Bottlenecks
by
Joon ho Lee
and
Professor Michael J. Cassidy
APPENDIX D
D- 3
APPENDIX D
D- 4
i
ACKNOWLEDGMENTS
The research team is especially grateful to those Caltrans personnel who offered guidance
as members of the Advisory Committee. Those members were Sam Toh, Scott Eades, St
eve Hague, Zhongren Wang, Rodney Oto, Jose Mujica, Vu H. Nguyen, Fred Yazden and
Tam Nguyen.
APPENDIX D
D- 5
APPENDIX D
D- 6
ii
ABSTRACT
The present research performed an empirical and theoretical analysis on what triggers
bottleneck activations and discharge flow changes in weaving sections. Investigations
revealed that changes in the spatial distributions of mandatory lane changes, especially
for Freeway- to- Ramp ( F- R) maneuvers, led to variations in bottleneck discharge flows.
When the F- R maneuvers were concentrated near on- ramp, they became more disruptive,
resulting in bottleneck activations with reductions in discharge flows. Findings further
indicate that the spatial distributions of these lane changes, in turn, were dictated by the
traffic conditions in the auxiliary lane. On- ramp flow reductions increased the
attractiveness of the auxiliary lanes, thus motivating F- R drivers to perform their
maneuvers nearer the on- ramp, and vice versa. A micro- simulation model was developed
based on the observed lane- changing behaviors, and it successfully reproduced the
observed mechanisms of weaving bottleneck flows.
Keywords: Weaving, Weaving sections, Simulation, Discrete choice modeling
APPENDIX D
D- 7
APPENDIX D
D- 8
iii
EXECUTIVE SUMMARY
Though there have been numerous studies of freeway weaving sections ( i. e., segments in
which an on- ramp is followed by an off- ramp), there remains a significant lack of
empirical and theoretical understanding of the traffic behavior that causes weaving
sections to become bottlenecks with varying discharge flows. The present research
entails empirical analysis and theoretical modeling of what triggered the bottleneck
activations and discharge flow changes in two freeway weaving sections. Both sites were
recurrent bottlenecks during the rush, and investigations revealed that changes in the
spatial patterns of vehicular lane- changes, especially among Freeway- to- Ramp ( F- R)
maneuvers, caused variations in bottleneck discharge flow. When the F- R maneuvers
were concentrated near a weaving section’s on- ramp, they became more disruptive,
resulting in bottleneck activations with diminished discharge flows. Findings further
indicated that the spatial distributions of these lane changes, in turn, were dictated by the
traffic conditions in the auxiliary lane ( i. e., the lane connecting the off- ramp to the
upstream on- ramp). Reductions in on- ramp flows increased the attractiveness of the
auxiliary lane, thus motivating F- R drivers to perform their maneuvers nearer the on-ramp.
Conversely, increases in on- ramp flo