Citation
Determining criteria for selecting red light camera locations

Material Information

Title:
Determining criteria for selecting red light camera locations
Creator:
Alturki, Mansour Abdulhamid ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic filer (303 pages). : ;

Subjects

Subjects / Keywords:
Photography in traffic engineering ( lcsh )
Electronic traffic controls ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
The objective of this dissertation is to develop a systematic method and criteria for selecting effective (i.e., severe crash reducing) red light camera locations among all signalized intersections of a given jurisdiction. Another objective is to develop criteria that can be implemented using accessible data while maintaining the comprehensiveness feature of the criteria. Selecting locations for red light cameras has received less attention by researchers of transportation engineering than assessing their effectiveness in reducing crashes. However, better site selection rules can result in greater effectiveness, which is the main goal of installing red light cameras. The methodology was divided into two phases that is mostly based on statistical criteria, but with more field investigations in the second phase. The first phase includes five criteria, which are, (i) crash severity level, (ii) normalized crash severity level, (iii) potential for improvement in terms of crash rate, (iv) potential for improvement in terms of crash frequency, and (v) crash types. The second phase includes six other criteria, which are, (i) fluctuation of crashes, (ii) vehicle types, (iii) economic evaluation, (iv) intersection characteristics, (v) approach determination, and (vi) red light locations. The study applies its methodology to three major cities in Colorado; these are Colorado Springs, Fort Collins, and Denver. The study found red light camera candidate intersections that are very consistent with the city engineers' opinions of potentially effective locations and the history of crash data from Denver since 2003.
Thesis:
Thesis (Ph.D.)--University of Colorado Denver. Civil engineering
Bibliography:
Includes bibliographic references.
System Details:
System requirements: Adobe Reader.
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Mansour Abdulhamid Alturki.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
902650006 ( OCLC )
ocn902650006

Downloads

This item has the following downloads:


Full Text
DETERMINING CRITERIA FOR SELECTING RED LIGHT CAMERA LOCATIONS
by
MANSOUR ABDULHAMID ALTURKI
MEng, University of Colorado Denver, 2008
MBA, University of Colorado Denver, 2008
BS, King Saud University, 2005
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Civil Engineering Program
2014


This thesis for the Doctor of Philosophy degree by
Mansour ALTurki
has been approved for the
Civil Engineering Program
by
Bruce Janson, Chair
Wesley Marshall, Advisor
Juan Robles
Gary Kochenberger
Bob Kois
May 2, 2014


Mansour AbdulHamid ALTurki (Ph.D., Civil Engineering)
Determining Criteria for Selecting Red Light Camera Location
Thesis directed by Professor Bruce Janson
ABSTRACT
The objective of this dissertation is to develop a systematic method and criteria for
selecting effective (i.e., severe crash reducing) red light camera locations among all
signalized intersections of a given jurisdiction. Another objective is to develop criteria
that can be implemented using accessible data while maintaining the comprehensiveness
feature of the criteria. Selecting locations for red light cameras has received less attention
by researchers of transportation engineering than assessing their effectiveness in reducing
crashes. However, better site selection rules can result in greater effectiveness, which is
the main goal of installing red light cameras. The methodology was divided into two
phases that is mostly based on statistical criteria, but with more field investigations in the
second phase. The first phase includes five criteria, which are, (i) crash severity level, (ii)
normalized crash severity level, (iii) potential for improvement in terms of crash rate, (iv)
potential for improvement in terms of crash frequency, and (v) crash types. The second
phase includes six other criteria, which are, (i) fluctuation of crashes, (ii) vehicle types,
(iii) economic evaluation, (iv) intersection characteristics, (v) approach determination,
and (vi) red light locations. The study applies its methodology to three major cities in
Colorado; these are Colorado Springs, Fort Collins, and Denver. The study found red
light camera candidate intersections that are very consistent with the city engineers
opinions of potentially effective locations and the history of crash data from Denver since
2003.
111
The form and content of this abstract are approved. I recommend its publication.
Approved: Bruce Janson


DEDICATION
This dissertation is lovingly dedicated to my mother, Mrs. Eman ALTurki, for her
encouragement, and constant love that have sustained me throughout my life, Without her,
I wont be at this level of education.
To my father Mr. AbdulHamid ALTurki who has been my silent inspiration and my
support when hard times come around.
IV


ACKNOWLEDGEMENTS
I am most grateful to the members of my committee, Mr. Bob Kois, Prof. Gary
Kochenberger, and Mr. Juan Robles for their time, encouragement, and expertise
throughout this project. Special thanks to Prof. Bruce Janson (the Chairman of the
Committee) and Prof. Wesley Marshall (my advisor), for their exquisite attention to detail,
patience and for their continuous demand for excellence. Prof. Bruce and Prof. Wes have
been more than advisors to me.
There are people in everyones lives who make success both possible and rewarding. My
wife, Ahoud ALSharaia, my children, Eman ALTurki, and Nawaf ALTurki steadfastly
supported and encouraged me.
Dr. Saleh ALSoghair, Eng, Dino Bakkar, and Eng. Andy Richter I will never forget the
support and encouragement you provided to me by facilitating many obstacles that came
on my way to this accomplishment.
My friend Eng. Ziyad ALBathi helped, cajoled, and prodded me when I needed it the
most.
For my uncle Abdullah ALTurki, my father in law Mr. Ahmad ALSharaia, and my
neighbors Mr. Tim Garduno and Mrs. Wendy Garduno for their support and effort that
they made sure to give to me in many occasions.
I also like to give special thanks to Anderson Academic Commons Library at the
University of Denver for providing me with the all the resources I needed during my
research time.
v


Without the support of my siblings Malath Malak, Maram, Nourh, Hamad, and
Abdullah, pursuit of this advanced degree would never have been started.
Thank you, ALL, now and always.
vi


TABLE OF CONTENTS
CHAPTER
I. PROBLEM STATEMENT......................................................
Introduction............................................................
Statement of Problem....................................................
Main Questions..........................................................
Study Objectives........................................................
Hypotheses and Contribution to the Transportation Engineering Industry and Public
Safety..................................................................
Limitations to the Study................................................
II. BACKGROUND............................................................
Traffic Safety Overview.................................................
History of Red Light Camera Systems.....................................
Glossary of Terms.......................................................
Vehicle Detection and Surveillance Technologies.........................
Implications for Public Privacy.........................................
Impact on Revenue.......................................................
Study Timeline..........................................................
Dissertation Structure..................................................
III. LITERATURE REVIEW....................................................
Introduction............................................................
Effectiveness of RLC on Safety..........................................
.. 1
.. 1
..2
.. 3
.. 3
..4
.. 5
.. 6
.. 6
.. 8
11
12
18
19
21
22
24
24
24
vii


Effectiveness of RLC on Type of Crashes.......................................33
Effectiveness of RLC on Crashes Severity......................................38
Characteristics of Red Light Runners..........................................43
RLC and Signal Timings........................................................49
Methodologies and Procedures Used for RLC Analysis............................50
RLC Spillover Effect (Halo Effect)............................................53
RLC Site Selections...........................................................57
IV. METHODOLOGY................................................................62
Introduction.................................................................62
Why These Locations as Case Studies?..........................................62
Data Required and Field Investigation.........................................63
Methodology..................................................................66
Phase I Includes Four Criteria..............................................66
Phase II Includes Seven Criteria............................................74
Expected Findings.............................................................87
V. ANALYSES AND FINDINGS......................................................88
Section I: Analyses of RLC Sites Selection for Colorado Springs...............88
Section II: Further Analysis and Field Investigation of Top 10 RLC Candidates in
Colorado Springs..............................................................95
Section I: Analyses of RLC Sites Selection for Fort Collins................106
Section I: Analyses of RLC Sites Selection for Denver......................122
viii


Section II: Further Analysis and Field Investigation of Top 10 RLC Candidates in
Denver...........................................................129
Recommednations and Conclusions..................................144
WORKS CITED........................................................151
APPENDIX...........................................................157
IX


LIST OF FIGURES
FIGURE
1. Number and rank of motor vehicles traffic fatalities as a cause of death in the United
States. 1981-2009 (Subramanian, 2009)..........................................7
2. The Hague traffic police put a sort of monocular. (Gatsometer, 2010).............9
3. Older RLC in Ludwigsburg, Germany. (Lowe, 2006).................................. 10
4. Distribution of the loops electromagnetic field. (Hockaday, 1991)..............13
5. Loop location at the intersection. (Kell, 1990)................................. 14
6. Loop sensors reflect damages to the asphalt. (Kell, 1990)....................... 15
7. Speed limit cameras can take shapes of normal road elements. (Klein, Millimeter-
Wave and Infrared Multi sensor Design and Signal Processing, 1997)............. 16
8. Intrusive sensors and camera requires less effort and no damages. (Klein, Final Report:
Mobile Surveillance and Wireless Communication Systems Field Operational Test -
Vol. 2: FOT Objectives, Organization, System Design, Results, Conclusions, and
Recommendations, 1999).................................................. 17
9. The proportion of crashes occurring at monitored approaches vs. non-monitored
approaches. (Dahnke, Stevenson, Stein, & Lomax, 2008)...................27
10. Percentage of crash type in Scottsdale for 14-year period. (Shin & Washington, 2007)
.........................................................................37
11. Percentage of crashes per year by crash type and severity (PDO vs. injury and fatal).
(Shin & Washington, 2007)................................................41
12. Percentage of crashes per year by crash type and severity (minor vs. major). (Shin &
Washington, 2007)........................................................42
x


13. Normalized red light violation values by age group (Yang & Najm, 2006)......47
14. Distributions of red light violation records by vehicle speed (Yang & Najm, 2006). 47
15. Distribution of red light violation by time of day (Yang & Najm, 2006)......48
16. A photo taken from a camera for an accident involving RLR (Administration, 2005)
..............................................................................51
17. Intersections studied in Arlington Virginia (McCartt & Hu, 2013)...............55
18. Illustration of the term potential for improvement.............................69
19. Calculation of crash type rate (ALTurki, 2013).................................72
20. A vandalized RLC in Phoenix Arizona. (Garrett, 2011)..........................81
21. Colorado Springs reported crashes in relation to annual average daily traffic.91
22. Briargate Py & N Powers B1 (Google Maps)......................................98
23. Airport Rd & S Academy B1 (Google Maps).......................................99
24. E Woodmen Rd/I-25 (Google Maps)...............................................99
25. E Platte Av & N Academy B1 (Google Maps).....................................100
26. Barnes Rd & N Powers Blvd....................................................100
27. E Platte Ave & N Union Blvd..................................................101
28. N Academy Blvd & Vickers Dr (Google Maps)....................................101
29. N Powers Blvd & Stetson Hills Blvd...........................................102
30. Maizeland Rd & N Academy Blvd................................................102
31. Dublin Blvd & N Union Blvd. (Google Maps)....................................103
32. Final RLC locations (Colorado Springs).......................................105
33. Fort Collins reported crashes in relation to annual average daily traffic....109
34. College Ave & Monroe. (Google Maps)..........................................114
xi


35. Timberline Rd & Horsetooth Rd. (Google Maps)..............................115
36. Lemay & Harmony. (Google Maps)............................................115
37. College Ave & Tribly Rd. (Google Maps).....................................116
38. College Ave & Horsetooth Rd. (Google Maps).................................116
39. S Shields St & W Plum St. (Google Maps)....................................117
40. Timberline Rd & Drake Rd. (Google Maps)...................................117
41. Shields St & Mulberry St. (Google Maps)....................................118
42. Shields St & Elizabeth St. (Google Maps)...................................118
43. Ziegler Rd & Rock Creek Dr. (Google Maps)..................................119
44. Final RLC locations (Fort Collins).........................................121
45. Denver's reported crashes in relation to annual average daily traffic.....125
46. E Alameda Ave & Leetsdale Dr. (Google Maps)...............................132
47. W Colfax Ave & N Kalamath St. (Google Maps)...............................133
48. Leetsdale Dr & Quebec St. (Google Maps)...................................133
49. S Monaco St & Leetsdale Dr. (Google Maps)..................................134
50. E 6th Ave & N Lincoln Blvd. (Google Maps)..................................134
51. W Mississippi Ave & S Platte River Dr. (Google Maps).......................135
52. N Colorado Blvd & E Colfax Ave. (Google Maps)..............................135
53. S Federal Blvd & W Alameda Ave. (Google Maps)..............................136
54. E Alameda Ave & S Monoco St (Google Maps)..................................136
55. S University Blvd & E Evans Ave. (Google Maps).............................137
56. Final RLC locations (Denver)..............................................139
xii


57. Trend of total crashes before and after the year of RLC installation at four signalized
intersections in Denver...............................................................140
58. Trend of front to side type of crashes before and after the year of RLC installation at
four signalized intersections in Denver...............................................141
59. Trend of rear end type of crashes before and after the year of RLC installation at four
signalized intersections in Denver....................................................142
60. City of Denver warns drivers to drive safely as they approach the intersection of S
University Blvd & E Evans Ave.........................................................146
xiii


LIST OF TABLES
TABLE
1. RLC effectiveness on safety at Fairfax County, Virginia (Hobeika & Yaungyai, 2006)
.............................................................................29
2. Summary of the recent studies of RLC effectiveness on safety.................32
3. Results of one-year before/after study Sacramento California (McGee & Eccles, 2006)
.............................................................................34
4. Before and after changes in crashes, Sydney, Australia (Hillier, Ronczka, & Schnerring,
1993).......................................................................... 35
5. Results for individual jurisdictions for total crashes (Administration, 2005).... 36
6. The distribution of crashes by severity for all signalized intersections 1997 (McGee &
Eccles, 2006).................................................................. 38
7. Percent of last drivers running a red light by demographic category. (Martinez & Porter,
2006)..........................................................................44
8. Cinit crash cost estimates by severity level used in the economic effects analysis.
(Federal Highway Administration, 2005)...................................... 52
9. Observed red light violation rates per 10,000 vehicles by time into red signal phase and
percentage changes 1 month and 1 year after red light camera ticketing began,
compared with warning period. (McCartt & Hu, 2013)..........................56
10. Data required for RLC sites selection Criterion...............................65
11. Weighting percentages for criterions in Phase I. (Colorado Springs)...........73
12. Weighting percentages for criterions in Phase I. (Fort Collins)...............73
13. Weighting percentages for criterions in Phase I. (Denver).....................73
xiv


14. Pre-calculated yellow intervals at various speeds..............................80
15. Sample of field evaluation table used to evaluate intersection characteristics.81
16. Table used for determining numbers of at-fault vehicles in each approach.....82
17. Formulas used to obtain final findings.........................................83
18. Ranking of top 10 RLC candidates in Colorado Springs based on normalized crash
severity level................................................................89
19 Ranking of top 10 RLC candidates in Colorado Springs based on crash severity....89
20. Ranking of top 10 RLC candidates in Colorado Springs based on potential for
improvement in relation to crash rate.........................................90
21. Ranking of top 10 RLC candidates in Colorado Springs based on potential for
improvement in relation to crash Frequency....................................90
22. Ranking of top 10 RLC candidates in Colorado Springs based on crash type......91
23. Final top 10 RLC candidates in Colorado Springs for all criteria in phase 1...93
24. Intersections field evaluation of Colorado Springs top 10 RLC candidates......96
25. Number of at fault vehicles per approach (Colorado Springs)...................104
26. Ranking of top 10 RLC candidates in Fort Collins based on normalized crash severity
level........................................................................107
27. Ranking of top 10 RLC candidates in Fort Collins based on crash severity level.... 107
28. Ranking of top 10 RLC candidates in Fort Collins based on potential for
improvement in relation to crash rate........................................108
29. Ranking of top 10 RLC candidates in Fort Collins based on potential for
improvement in relation to crash Frequency...................................108
30. Ranking of top 10 RLC candidates in Fort Collins based on crash type..........109
xv


31. Final top 10 RLC candidates in Fort Collins for all criteria in phase 1.......110
32. Intersection evaluation table (Fort Collins)..................................112
33. Number of at fault vehicles per approach (Fort Collins).......................120
34. Ranking of top 10 RLC candidates in Denver based on normalized crash severity
level...........................................................................123
35. Ranking of top 10 RLC candidates in Denver based on normalized crash severity
level...........................................................................123
36. Ranking of top 10 RLC candidates in Fort Collins based on potential for
improvement in relation to crash rate...........................................124
37. Ranking of top 10 RLC candidates in Fort Collins based on potential for
improvement in relation to crash rate...........................................124
38. Ranking of top 10 RLC candidates in Denver based on crash type.................125
39. Final top 10 RLC candidates in Denver for all criteria in phase 1..............127
40. Intersection evaluation table (Denver)..........................................130
41. Number of at fault vehicles per approach (Denver)...............................138
42. Total crashes by year in current RLC locations in Denver........................140
43. Front to side type of crashes by year in current RLC locations in Denver........141
44. Rear end type of crashes by year in current RLC locations in Denver...........141
45. Analysis of Colorado Springs intersections based on crash severity level and
normalized crash severity level.................................................157
46. Colorado Springs intersections ranked based on normalized crash severity level... 161
47. Colorado Springs intersections ranked based on crash severity level.............164
xvi


48. Analysis of Colorado Springs Intersections based on potential for improvement in
relation to crash rate and crash frequency.......................................167
49. Colorado Springs intersections ranked based on potential for improvement in relation
to crash rate....................................................................171
50. Colorado Springs intersections ranked based on potential for improvement in relation
to crash frequency...............................................................174
51. Analysis of Colorado Springs intersections based on crash types..................177
52. Colorado Springs intersections ranked based on front to side rate................180
53. Analysis of Fort Collins intersections based on crash severity level and normalized
crash severity level...............................................................183
54. Fort Collins intersections ranked based on normalized crash severity level...........187
55. Fort Collins intersections ranked based on crash severity level......................190
56. Analysis for potential for improvement for all intersections of Fort Collins in relation
to crash rate and frequency........................................................193
57. Fort Collins intersections ranked based on potential for improvement in relation to
crash rate.........................................................................198
58. Fort Collins intersections ranked based on potential for improvement in relation to
crash frequency....................................................................201
59. Analysis of Fort Collins intersections based on crash types........................204
60. Fort Collins intersections ranked based on front to side crashes...................208
61. Analysis of Denver intersections based on crash severity level and normalized crash
severity level.....................................................................211
62. Denver intersections ranked based on normalized crash severity level...............222
xvii


63. Denver intersections ranked based on crash severity level.........................230
64. Analysis of potential for improvement for Denver intersections based on crash rate
and frequency......................................................................238
65. Denver intersections ranked based on potential for improvement in relation to crash
rate...............................................................................250
66. Denver intersections ranked based on potential for improvement in relation to crash
frequency..........................................................................258
67. Analysis for Denver intersections based on crash types.............................266
68. Denver intersections ranked based on front to side crashes.........................277
xviii


LIST OF EQUATIONS
EQUATION
1. Normalized- crash serverity level............................................67
2. Crash severity level.........................................................67
3. PFI in crash rate............................................................69
4. Annual crash rate............................................................70
5. Average crash rate...........................................................70
6. Crash frequency..............................................................71
7. Proportionality to obtain relative weights...................................72
8. collision cofefficieient of variation........................................74
9. Sample mean..................................................................74
10. Fluctuation of crashes by calculating the standard mean....................75
11. Type of vehciles by calculating Chi-square test............................75
12. RLC Economic evaluation....................................................76
xix


CHAPTER I
PROBLEM STATEMENT
Introduction
The condition of being protected against physical, economic, emotional,
educational, political, occupational, or any other aspects that could be damaged or
harmed is the definition of safety. (Federal Flighway Administration, 2012) Recently,
public safety, as one of the major safety categories, has received more attention due to the
fact that it is directly related to humans lives and health, which is considered as a
significant indication of better developments and communities.
Keeping in mind all the developments and advancements associated with todays
technologies and environmental regulations, public safety has become even more
challenging to achieve. Promoting public safety in systems like medical and health safety,
building safety, and so on is very important, but it is even more important when it relates
to the transportation system.
The transportation system requires the highest level of safety due to the number of
users involved in the system every day, as well as the nature of risks people can suffer as
a result of the system being unsafe. One of the most dangerous and risky traffic related
violations is red light running, which is a behavior that can cause some of the most
serious injuries and fatalities the transportation system may generate.
In the United States and during the year of 2010 alone, almost 50 percent of all
crashes reported to the police occurred at intersections. In fact and according to the
Insurance Institute for Highway Safety, signalized intersections accounted for more than
1


68,000 serious non-fatal injuries and 7707 deaths in 2010 alone. (Insurance Institute for
Highway Safety, 2012).
As a result, many transportation agencies, organizations, departments and
communities across the nation like the Federal Highway Administration (FHWA), the
National Highway Traffic Safety Administration (NHTSA), seek to address crashes and
reduce both injuries and fatalities by increasingly looking for tools to supplement
traditional enforcement resources. One of the safety tools that over 550 US communities
have employed is a red light camera (RLC). (National Safety Council, 2009)
The first chapter of this study starts by explaining the statement of problem the
study addresses in addition to representing the main questions, and research objectives.
This chapter will also demonstrate how this study could contribute to the civil
engineering science in general and more specifically to the transportation engineering
field despite the limitations that are usually associated with similar studies.
Statement of Problem
Many of the post-implementation evaluations that were conducted to measure the
effectiveness of red light cameras (RLC) have shown an overall effectiveness in reducing
the frequency of crashes at intersections where red light cameras are operated, although
there are exceptions in some cases.
As it will be illustrated in the literature review, most studies that researched the
effectiveness of RLC on safety were mostly making before/after crash comparisons.
Other studies investigated more details regarding the types and severity of crashes
associated with RLC. In comparison, fewer studies discussed other important areas of
2


research that could show significant indication of RLC effectiveness on safety such as the
RLC sites selection.
This study will examine one of the critical elements that is usually associated with
the installation or expansion of RLC systems, which is the selection of RLC sites that
have the greatest potential to improve safety. The study will also demonstrate its
practically by applying the methodology to three major cities of Colorado; which are
Colorado Springs, Fort Collins, and Denver.
Main Questions
This study will try to answer the following questions in order to achieve the study goals:
1) In Colorado Springs, the City discontinued the RLC program after one year of
installation (2010) due to unsuccessful results. If we go back to 2010, what kind
of criteria could be used to make selection of specific intersections within the city
limit and therefore could possibly make the RLC program more effective and
show successful results?
2) In Denver and Fort Collins, the costly system has been under operation for at
least 10 years. Are these cities making the best choices when selecting the
locations of their RLC systems? Can that be supported in a scientific way?
Study Objectives
This study aims to provide a RLC site selection methodology based on analytical
procedures that require accessible data that will be obtainable by any community to select
RLC sites with greatest potential so the selection becomes more systematic. This study
intends to use some statistical models that will be presented in more detail as part of the
methodology chapter. Additionally, this study aims to form a more obvious picture of the
3


effectiveness of RLC programs on reducing red light running crashes and their potential
safety improvement when comparing the current RLC sites to the candidate sites
concluded in the analysis chapter.
Hypotheses and Contribution to the Transportation Engineering Industry and
Public Safety
The following are two primary motivations and potential benefits of this study:
1) An analytical-based site selection methodology increases the effectiveness of red
light camera programs.
2) Comprehensive and scientific RLC criteria can positively impact public opinion
about RLC system.
This study aims to contribute to the transportation industry from different points
of view. The following bullets describe these contributions:
1) This study will provide transportation agencies, planners, engineers, and
researchers with statistical figures and findings related to one of the least
researched areas, which is RLC sites selection (according to the literature review).
2) This study will contribute to the field of civil engineering and transportation by
reviewing the RLC experiences in Denver and Fort Collins.
3) This study provides an analytical-based methodology for RLC site selection that
can be used by any city in implementing a RLC program to potentially improve
public safety.
4) The study will apply its methodology to three major cities of Colorado; which are
Colorado Springs, Fort Collins, and Denver.
4


5) It is also important to note that the analytical-based methodology mentioned
above is formed based on the sort of data that most of the cities around the world
have access to, which makes it a more usable methodology.
Limitations to the Study
Accurately assessing candidates with potential to be equipped with RLC is
challenging for several reasons:
1) Many safety related factors are uncontrolled and/or confounded during the
periods of observation.
2) Availability and accuracy of data may not be accessible at the needed level.
3) The variety and number of agencies involved in such programs can make it
more challenging to find accurate and consistent data.
5


CHAPTER II
BACKGROUND
Traffic Safety Overview
Road traffic safety means reducing accident causes on the road through improved
vehicles, facilities, and driving practices. Road and vehicle design, driver impairment,
speed of operation, and other factors like proper signal timing, better signal design,
improved intersection design, and many more are all considered factors that could
decrease or increase the level of safety on the road. (Road Safety, 2010)
According to the World Health Organization (WHO), more than a million people
are killed on the worlds roads each year. A report published by the WHO in 2004
estimated that 1.2 million people were killed and 50 million injured in traffic crashes
around the world each year and that traffic crashes are the leading cause of death among
children 10-19 years of age. The report also noted that the problem was most severe in
developing countries and that simple prevention measures could halve the number of
deaths. (World Health Organization, 2010)
Because of these facts, road traffic crashes are one of the worlds largest public
health and injury prevention problems. The problem is more acute because victims are
overwhelmingly healthy prior to their crashes.
In 2009, motor vehicle traffic crashes were among the top 10 causes of death in
the United States for the first time since 1981. In 2008, vehicle traffic crashes were 11th.
(See Figurel)
In 2009, when ranked by specific ages, motor vehicle traffic crashes were the
leading cause of death for age 4 and every age 11 through 27, while motor vehicle traffic
6


crashes were the leading cause of death for each age 13 through 30 the year before.
(Subramanian, 2009)
Figure 1: Number and rank of motor vehicles traffic fatalities as a cause of death in the United States.
1981-2009 (Subramanian, 2009)
Note The coding of mortality data changed significantly in 1999, so comparisons of the number
of deaths and death rates from 1998 and before with data from 1999 and after may not be advisable
(Subramanian, 2009)
In the United States, three acts were announced to seek better and safer
transportation systems, starting with the Intermodal Surface Transportation Efficiency
Act, which was signed by President Bush back in 1991.
The act provides funding to continue the provisions of the National Traffic and
Motor Vehicle Safety Act of 1966, and the Motor Vehicle Information and Cost Savings
Act. The act includes a number of motor vehicle safety rulemaking requirements and
additional directions, including rollover protection for occupants of passenger cars,
multipurpose passenger vehicles, and light trucks, side impact protection for occupants of
multipurpose passenger vehicles, improved head impact protection (from interior
components) for occupants of passenger cars, and airbag crash protection systems for
7


drivers and right front passengers in new passenger cars, new light trucks (including light
buses), and multipurpose passenger vehicles.
On June 9, 1998, the president signed the Transportation Equity Act for the 21st
Century (TEA-21). This act paid major attention to safety, strengthening the safety
programs across the US Department of Transportation that aim to save road users lives
and property. (TheU.S. Department of Transportation, 1998)
The third act, The Safe, Accountable, Flexible, and Efficient Transportation
Equity Act (SAFETEA) was announced formally in 2005. The act provides
comprehensive attention to the safety associated with the transportation system. The act
establishes a new core Highway Safety Improvement Program that aims to make
significant progress in reducing fatalities that take place on the highways. It concentrates
on several areas of concern in the system like work zones, children walking to school,
and older drivers. It doubled the funds to improve the infrastructure and implement
strategic highway safety planning to ensure accommodation of the safety requirements.
(The National Tranportation Library, 1991)
History of Red Light Camera Systems
Historically, traffic enforcement cameras can be dated back to 1905 where the
popular machines were used to record motorists speeds by taking time-stamped images
of vehicles moving across the start and end point of the road. By using the popular
machine system, authorities were able to calculate the vehicle speed and identify the
driver by referring to the time-stamps and images respectively.
Gatsometer BV was a company founded back in 1958 by rally driver Maurice
Gatsonides. It produced a monitor device to track the average speed in order to improve
8


his lap times. Later, the company started supplying police radars, red light cameras, and
mobile speed traffic cameras. (Gatsometer, 2010)
Figure 2: The Hague traffic police put a sort of monocular. (Gatsometer, 2010)
Worldwide, red light cameras have been in use since the 1960s, and were used for
traffic enforcement in Israel as early as 1969. The first red light camera system was
introduced in 1965, using tubes stretched across the road to detect the violation and
subsequently trigger the camera. Red light cameras were first developed in the
Netherlands. One of the first developers of these red light camera systems was
Gatsometer BV. (Gatsometer, 2010)
The cameras first received serious attention in the United States in the 1980s
following a highly publicized crash in 1982 involving a red-light runner who collided
with an 18-month-old girl in a stroller (or "push-chair") in New York City. Subsequently,
a community group worked with the city's Department of Transportation to research
9


automated law-enforcement systems to identify and ticket drivers who run red lights.
New York's red-light camera program went into effect in 1993. From the 1980s onward,
red light camera usage expanded worldwide, and one of the early camera system
developers, Poltech International, supplied Australia, Britain, South Africa, Taiwan, the
Netherlands and Hong Kong. American Traffic Systems (subsequently American Traffic
Solutions) (ATS) and Redflex Traffic Systems emerged as the primary suppliers of red
light camera systems in the US, while Jenoptik became the leading provider of red light
cameras worldwide. (Lowe, 2006)
Initially, all red light camera systems used film, which was delivered to local law
enforcement departments for review and approval. The first digital camera system was
introduced in Canberra, Australia in December 2000, and digital cameras have
increasingly replaced the older film cameras in other locations since then.
Figure 3: Older RLC in Ludwigsburg, Germany. (Lowe, 2006)
10


Glossary of Terms
Traffic Enforcement Camera (TEC): An automated ticketing machine that could be
mounted beside or over the road to observe traffic violators. (Wilson C, 2010)
Red Light Camera (RLC): is a traffic enforcement camera that captures an image of a
vehicle which has entered an intersection against a red traffic light. By automatically
photographing vehicles that run red lights, the camera produces evidence that assists
authorities in their enforcement of traffic laws. (Insurance Institute for Highway Safety,
2010)
Red Light Runner (RLR): The simplest definition of red-light running (RLR) is the act
of entering, and proceeding through, a signalized intersection after the traffic signal has
turned red. (National Committee on Uniform Traffic Laws and Ordinances., 2000)
Infraction: In 1981, the legislature of the US decriminalized many minor traffic offenses
to promote public safety and to facilitate the implementation of a uniform and
expeditious system for the disposition of such offenses.
Common traffic infractions are speeding as well as seat belt and liability
insurance violations. These offenses are called infractions and are considered civil cases.
(Grays Harbor County, 1981)
Violation: to break, disregard, or infringe a law or a certain agreement. (In this study: to
break a traffic law)
Citation: is another word for a traffic ticket. It is a notice issued by a law enforcement
official to a motorist or other road user, accusing violation of traffic laws. It could be
cited as a moving vehicle which includes but not limited to violations such as exceeding
11


the speed limit or running red light or non-moving violation (illegally parked vehicle).
(Grays Harbor County, 1981)
Halo Effect spillover: Refers to the ability of an intersection safety camera to have a
positive effect at nearby, untreated intersections because of a longer term influence on
driver behavior (For example, driver will not run red lights at intersections near
intersections equipped with a red light camera). (NHCRP, 2003)
Intrusive sensors: record vehicle count and classification data with some lane closure
and drilling into the roadway. (Federal Highway Administrations Intelligent
Transportation Systems Joint Program Office, 2000)
Non-intrusive sensors: record vehicle count and classification data without interruption
to traffic flow. Installation of non-intrusive detection systems usually involves no
requirement for road closure or traffic management and deployment includes utilizing
existing roadside infrastructure. (Federal Highway Administrations Intelligent
Transportation Systems Joint Program Office, 2000)
The Kangaroo effect: A kangaroo effect is created when drivers decelerate suddenly
when they notice a speed camera or red light camera, and then quickly accelerate again.
This is thought to have an adverse effect on traffic flow and the environment, as well as
road safety. (Federal Highway Administrations Intelligent Transportation Systems Joint
Program Office, 2000)
Vehicle Detection and Surveillance Technologies
Vehicle detection and surveillance technologies can be categorized into two major
types: intrusive and non-intrusive sensors. These types of technologies go through three
main processes: the transducer, which detects the presence of a vehicle or its axles; the
12


signal-processing device, which then converts the transducer data into an electrical
signal; and, finally, a data-processing device that converts the electrical signal into traffic
parameters. There are several traffic parameters that might be included like speed,
vehicles count, occupancy, gap, weight, and many others. (Bailly, 1998)
In this section of the study, more information related to the intrusive and non-
intrusive sensors will be provided. The information will include the operating principle,
sensor measurement accuracy, costs, advantages, and disadvantages of these technologies.
Intrusive sensor (in-ground inductive loop). These types of sensors are usually
installed into the surface of the pavement by tunneling under the surface, in saw-cuts or
holes on the surface, or by anchoring directly into the surface. Intrusive sensors can be
micro-loop probes, pneumatic road tubes, or piezoelectric cables and other weight-in-
motion sensors. (Hockaday, 1991)
Figure 4: Distribution of the loops electromagnetic field. (Hockaday, 1991)
13


There are many advantages to the intrusive sensor like unlimited number of speed
measurements, the ability to specify the lane where the violation has occurred, and also
the level of accuracy it provides when recording the speed and the location of a vehicle.
(Kell, 1990)
Figure 5: Loop location at the intersection. (Kell, 1990)
The main disadvantages and drawbacks that are mainly associated with the
intrusive sensors are the disruption they can cause to traffic operation during the
installation processes and road closures, or when maintenance is required, whether that
type of maintenance is related to the sensor or other applications. They can also cause
damage to the surface of the road, especially when substandard drilling and cutting
activities are used when attaching the sensor to the roadway.
14


Figure 6: Loop sensors reflect damages to the asphalt. (Kell, 1990)
As far as non-intrusive sensors (loopless trigger radar), most studies show the
need for a more reliable and cost-effective method that could be applied to the same
applications as the intrusive sensor, but with fewer disadvantages. Non-intrusive sensors
came to be the solution since the installation of these sensors does not require the amount
of cutting and drilling the intrusive sensors do, and therefore cause less traffic disruption
and no damage to the surface at all. (Kell, 1990)
15


At the same time, non-intrusive sensors (aboveground sensors) have met many of
the applications required by surface streets and freeways. The non-intrusive sensors can
be mounted above or to the side of the roadway that needs monitoring. Many
technologies are currently used for this application like laser radar, video images,
microwave radars, and passive infrared, or a combination of two or more of them. The
system is also able to record speed, vehicles weight, vehicle categories, and vehicle
count. (Klein, Millimeter-Wave and Infrared Multisensor Design and Signal Processing,
1997)
16


The sensor can be mounted in a position perpendicular or oblique to the traffic
flow to allow the system to monitor each lane. In comparison to the intrusive sensors,
studies show that aboveground sensors are less affected by weather change and ambient
lights, are faster and easy to install, have an accuracy of speed detection that ranges +/- 2
mph, and monitor the configuration of each lane individually. (Klein, Final Report:
Mobile Surveillance and Wireless Communication Systems Field Operational Test Vol.
2: FOT Objectives, Organization, System Design, Results, Conclusions, and
Recommendations, 1999)
Figure 8: Intrusive sensors and camera requires less effort and no damages. (Klein, Final Report: Mobile
Surveillance and Wireless Communication Systems Field Operational Test Vol. 2: FOT Objectives,
Organization, System Design, Results, Conclusions, and Recommendations, 1999)
17


Implications for Public Privacy
Practically, RLC systems work by capturing the image of the vehicle, its driver,
and the vehicle license plate number as that vehicle goes through a red light. These
photographs provide evidence to authorities in order to assist with traffic law
enforcement and, therefore, the issuing of tickets to the violator. Typically, law
enforcement officials will review the photographs and determine whether a violation has
occurred. The next step is the infraction, which will be mailed to the mailing address
registered under the license plate. In some cases, photographs are not clear and therefore
officials cannot make a final decision. As a result, officials will either dismiss the citation
or mail the violator a notice requesting identification information to assist in making their
decision. (Insurance Institute for Highway Safety, 2010)
Using red-light camera systems is associated with several legal and privacy
concerns, including concerns about citation distribution, types of penalties, and the right
of authorities to issue a ticket based on a photograph. Before implementation, the public
should be educated on how the system works to ensure that the public understands that
the citations are only issued after photographs are reviewed by a police officer. (Elmitiny
& Radwan, 2008)
Another issue related to RLC that the public is frequently complaining about the
availability of signage and in particular, the messages that need to be given to drivers
about what is actually being monitored and enforced. A study in the UK discussed the
issues facing UK agencies responsible for implementing and operating camera-based
enforcement programs in relation to signage, as the camera signs can be located
differently depending on whether or not they are funded by a Safety Camera Partnership.
18


The studys major finding was that signs must be used consistently across the UK in
general, the boundaries of geo-political regions, of police force authority and of road
network operators' responsibility. (Wilson, 2007)
Impact on Revenue
The public argues that the main purpose behind RLCs is revenue. Several studies
and researches have found much evidence towards this being the case. An article titled
Big Brother is Ticking You, published as part of Popular Mechanics magazine,
emphasizes on the fierce opposition to RLC by citizens and organizations such as the
American Automobile Association and National Motorists Association.
The article referred to the Washington, DC experience with RLC, as the increased
number of crashes at approaches where RLC is installed, especially rear-end ones, have
been associated with an increased number of revenue for the city. Adding to the general
and growing discontent is the fact that a few towns have been caught shortening yellow
signal timing, thereby catching more red light runners and generating more revenue but
also inadvertently increasing accident rates. (Reynolds, 2006)
Tom Brodbeck, the Suns City columnist, argued whether the RLC program in the
city of Winnipeg is really aimed at safety and not revenue. He reevaluated the 50 most
dangerous intersections around the city in terms of crashes and wondered why only 7 of
the 31 red light cameras throughout the city are located at those intersections. What is
more interesting is the fact that there were no cameras at all among the top 10 most
dangerous intersections. If the main purpose of the cameras is to increase safety, then
why they are not placed in the locations with the least safety? Tom asked.
19


In fact, in two of those top 50 locations where the cameras were installed, the rate of
crashes has risen around 20 percent. (Brodbeck, 2012)
According to a study conducted by OpEdnews.com, police unions and for-profit
camera companies have teamed up on several occasions to defeat laws that proposed to
ensure traffic cameras are designed for public safety rather than to collect revenue. For
example, in Connecticut, police unions and traffic light camera companies opposed
efforts to expand the length of yellow lights despite the fact that implementing that would
reduce red light violations by 90 percent. (Fang, 2012)
In Florida last year, American Traffic Solutions, one of the largest for-profit
camera corporations, hired 17 lobbyists to defeat a similar bill. The company circulated a
letter signed by police chiefs and worked closely with officials from the Florida Sheriffs
Association, a labor group, to pressure legislators.
In California, a bill by State Sen. Joseph Simitian to ensure that traffic cameras
can only be set up to promote public safety rather than collect revenue was opposed by
the California Police Chiefs, a law enforcement labor union group. (Tucker, 2009)
20


Study r Imeline
Steps Description Timeline
1 Complete the preliminary examination
2 Defining the committee members
3 Searching for dissertation topics
4 Choosing dissertation topic Spring
5 Develop a topic 2012
6 Begin meetings with the advisor
7 Find out requirement for proposal submission
8 Outline proposal
9 Complete the theoretical framework Summer 2012
10 Determine the methodology of chapters
11 Submit proposal to advisor and committee to review
12 Edit proposal according to the review
13 File paper work and schedule defense
14 Defend proposal
15 Revise proposal if necessary Fall 2012
16 Meet advisor to draw up a research schedule
17 Conduct research for study
18 Analyze data from research
19 Outline dissertation
20 Meet with advisor to discuss outline and preliminary data analysis
21 Update proposal chapters for dissertation Spring 2013
22 Write results and findings chapter
23 Write summary and conclusion chapter
24 Find out requirements for final submission
25 Find out requirements for final oral defense
26 Polish writing and meeting with editor if necessary
27 Submit draft dissertation to committee
28 Edit dissertation according to the committee review
29 Submit final dissertation paperwork and schedule for oral defense Fall 2013
30 Have final dissertation typed professionally by word processor
31 Defend dissertation
32 Make final revision if needed
33 Graduate
21


Dissertation Structure
The dissertation has been divided into five chapters:
Chapter 1: Introduction including a statement of the problem, study objectives and main
questions, the study hypothesis, and how it contributes to the transportation engineering
industry.
Chapter 2: Background overview of the topic, traffic safety overview, history of red
light cameras, glossary of terms, vehicle detection and surveillance technologies a brief
discussion of two of the major concerns that associated with RLC systems, implication on
public privacy and impact on revenue. The chapter ends by presenting the study timeline,
and dissertation structure.
Chapter 3: Presents the literature review, which includes a comprehensive review of the
most recent studies and articles on the subject of RLC systems. This chapter is
categorized and divided according to the most recent research topics as it starts first with
the technical part of the RLC detection types, going through studies related the
of RLC on safety (mostly before/ after comparison), RLC collision types, RLC crashes
severity, Red light runner characteristics, RLC and signal timings, RLC spillover effect,
and most recent models and procedures used to conduct RLC related studies. This chapter
also reviews the recent researches that discussed the importance of RLC site selection,
which is the focus of this dissertation.
Chapter 4: Presents the procedures, models, and the phases of the methodology chapter
that will be used for RLC sites selection. It also represents the analytical-based
methodology that will be used for RLC site selection using case studies from Colorado
22


Springs, Denver, and Fort Collins. This chapter also presents the data required for each
case study.
Chapter 5: Presents the analytical findings for all intersections within each case study
city limit and conclusions from Colorado Springs, Denver, and Fort Collins, which
eventually show the top 10 candidate intersections that have priority over other signalized
intersections for RLC installation. This chapter also includes recommendations that
support future studies on this subject.
23


CHAPTER III
LITERATURE REVIEW
Introduction
Red light running is a significant public health concern, killing more than 800
people and injuring more than 200,000 in the United States per year. It is a significant
safety problem as drivers become more aggressive on city roads, and become impatient
waiting for traffic signals to change. RLC programs are considered one of the most
controversial topics facing traffic engineers, city councils, and public awareness groups.
Red light running cameras systems are automated enforcement systems that detect and
capture vehicles that run a red light and issue a citation. RLC systems are becoming
widely used in the United States to reduce the number and severity of red light running
crashes. (Fitzsimmons, Hallmark, McDonald, Orellana, Matulac, & Pawlovich, 2008)
In this chapter of the research, many of the recent studies and researches related to
the red light camera programs will be presented by discussing major areas of previous
researches such as the effectiveness of RLC on safety, effectiveness of RLC on type of
crashes, effectiveness of RLC on crashes severity, characteristics of red light runners,
RLC and signal timings, RLC spillover effect, and methodologies and analysis
procedures used to measure effectiveness of RLC. Finally, the literature review chapter
will finally focus on the most recent studies related to the main objective of this research,
which is the RLC sites selection.
Effectiveness of RLC on Safety
Studies evaluating the effectiveness of red light cameras on safety mostly suggest
that they are effective in reducing red light violations and injury crashes. A four-year
24


analysis (2004-2007) of the effectiveness of the RLC program in Raleigh, North Carolina,
which was a follow-up study to an earlier one made before 2004 but with a smaller
sample size (5 months), both showed that the program is producing positive safety results.
(Hummer & Cunningham, 2010)
In San Francisco, California, with its compact driving environment and dense
network of signalized intersections, red-light running reached a political crisis in 1994.
The city and county of San Francisco recently completed a pilot red-light photo-
enforcement program. The number of vehicles photographed violating red lights at the
photo-enforced locations dropped by more than 40% just 6 months into the pilot. Recent
statistics indicate that San Francisco's combined efforts to combat red-light running have
resulted in a significant decrease in the number of annual crashes caused by red-light
violators citywide. Based on the success of the pilot and supportive state legislation, San
Francisco is moving forward to expand the red-light photo-enforcement program to make
it one of the largest programs in the United States with 26 cameras rotating in 35
locations. (Fleck. J, 1999)
Iowa is another state that has a serious safety problem with red light running that
accounts for 35% of fatal and major injuries plus 21% of total crashes at signalized
intersections. The state has adopted the program in three communities; one of the
communities is Davenport that had installed the program back in 2004. Two years of
crash data after installation were available for analysis, which included 4 RLC locations
and 5 control intersections as part of it. The results of the analysis indicated that the
cameras were effective in reducing total crashes and RLR related-crashes on average of
20% and 40%, respectively. In the other hand, there was an increase of total crashes,
25


RLR related-crashes, and RLR rear-end related crashes of about 7%, 20%, and 33%,
respectively. (Hallmark, Orellana, Fitzsimmons, McDonald, & Matulac, 2010)
A comprehensive study conducted by the Center of Civic Engagement at Rice
University from September 2006 to August 2008, which included 70 monitored and non-
monitored approaches and six years of crashes data, concluded that the proportion of
crashes occurring at monitored approaches decreased significantly relative to the non-
monitored approaches, as Figure (9) shows below. The comparison of data between
monitored and non-monitored approaches supports the conclusion that red light cameras
are mitigating a general, more severe increase in collisions. Although this study supports
the idea that red light cameras have a positive effect in reducing crashes at monitored
approaches in comparison with non-monitored approaches, several questions have been
raised by these findings. The most important of these is Why have crashes at non-
monitored approaches increased so dramatically in the past year? The study suggested
that these results could be evidence of an increase in crashes across the city. The selection
in 2006 of intersections with high rates of crashes could be serving to magnify this effect.
(Dahnke, Stevenson, Stein, & Lomax, 2008)
26


70
Monitored vs, Unmonitored Approaches by Implementation Month
Implementation Months of Red Light Cameras
Figure 9 The proportion of crashes occurring at monitored approaches vs. non-monitored approaches.
(Dahnke, Stevenson, Stein, & Lomax, 2008)
An additional study was conducted to estimate the safety impacts of RLCs on
traffic crashes at signalized intersections in the cities of Phoenix and Scottsdale, Arizona.
Twenty-four RLC equipped intersections in both cities were examined in detail. The
evaluation results indicated that both Phoenix and Scottsdale are operating cost-effective
installations of RLCs, which show positive safety improvement: however, the variability
in RLC effectiveness within jurisdictions is larger in Phoenix (Shin & Washington, 2007).
A paper is to evaluate the safety effectiveness of automated traffic enforcement
systems, that is, red light cameras, installed at 254 signalized intersections in 32
jurisdictions in Texas. A before-after study by the empirical Bayesian methodology was
performed to remove the regression-to-mean bias during the evaluation of treatments.
The results indicate significant decreases in the incidences of all types of red light
running (RLR) crashes and right-angle RLR crashes by 20% and 24%, respectively. A
significant increase of 37% for rear-end RLR crashes was discovered. The study results
27


suggest that a significant safety benefit for red light cameras is achieved if intersections
have four or more RLR crashes per year or have two or more RLR crashes per 10,000
vehicles. Red light cameras show counterproductive results if intersections experience
fewer than two RLR crashes per year or have one crash per 10,000 vehicles per year (Ko,
2013).
In Virginia, a study included six jurisdictions (Alexandria, Arlington, Fairfax City,
Fairfax County, Falls Church, Vienna) that deployed red light cameras. It documented the
safety impacts of those cameras based on 7 years of crash data for the period January 1,
1998, through December 31, 2004. The results show that cameras were associated with a
modest reduction in comprehensive injury crashes. (Garber, Miller, Abel, Eslambolchi, &
Korukonda, 2007)
A study that evaluated the Red Light Camera (RLC) program in Fairfax County,
Virginia was conducted back in 2003 and covered 13 cameras after 2 years of operation.
In conducting the analysis, violation results were grouped into two distinct periods: 1)
initial period (1st three months) and 2) after initial period. These two distinct periods
were also grouped into five periods for each, and there were as follows: 1) initial period,
2) fourth to ninth month period, 3) 10th to 15th, 3) 16th to 21st, 4) 22nd to 27th, 5) after 27th
month. The study reported that the RLC program reduced the traffic signal violation rate
by up to 63% in the 22nd to 27th month period of its operation (see Table 1). The results
also show that the increase of the intersection amber time, combined with RLC, produced
a higher reduction of up to 72% in violation rate. The crash rate was reduced by 27%
after 2 years of RLC operation; however, this reduction was not statistically significant.
(Hobeika & Yaungyai, 2006)
28


Table 1 RLC effectiveness on safety at Fairfax County, Virginia (Hobeika & Yaungyai, 2006)
Camera Intersections Average number of violations/10,000 vehicles % Change in violations per 10,000 vehilces
Initial Period 4-9 mo 10-15 mo 16-21 mo 22-27 mo After 27 4-9 mo 10-15 mo 16-21 mo 22-27 mo After 27
1 2.86 2.7 -5.50%
2 1.59 0.59 0.26 0.44 62.80% -83.60% -72.60%
3
4 2.33 1.19 0.88 0.97 1.03 1.2 49.10% -62.40% -58.30% -56.10% -48.70%
5 8.68 6.73 2.67 2.09 2.67 2.03 22.50% -69.30% -75.90% -69.20% -76.60%
6 2.13 2.34 2.56 2.01 10.10% 20.50% -5.70%
7
8 2.15 2.21 2.29 1.25 2.70% 6.50% -41.80%
9 2.44 2.26 2.78 2 -7.20% 14.00% -18.10%
10 2.21 2.15 1.83 -2.40% -17.20%
Average 3.05 2.56 1.9 1.46 1.85 1.615 17.09% -27.36% -45.40% -62.65% -62.65%
to
VO


A very interesting pilot study was conducted in Maine, which is considered as one
of the states that have a major problem with red light running. Maine is one of the states
that does not allow issuing citations based on photographic evidence, so only warning
letters were issued to violators. Therefore, the study covering the period from September
2004 to August 2005 was mainly concerned with the reduction of red light running
violations as a result of warning letters only. Observations of red-light running indicate
that the violation rate dropped by around 28% between December 2004 (when the system
was first installed) and May 2005 (when the system had been operational for several
months). However, it was the infractions that occurred at low speeds and within the first
second or so that were reduced. Infractions more than 3 seconds into red and at speeds
above 35 mph actually increased. It was interpreted that these later infractions were not
caused by the enforcement, but rather by other factors like weather and roadway
conditions. Conflict and crash data indicate that there were no great improvements in
safety between the before period and the period when the system was in operation. Actual
fines and RLC systems rather than warning tickets may have produced greater safety
effects. (Garder, 2006)
Another study was mainly aimed at estimating the RLR problem in Indiana. The
other objectives of the research included: (1) learning drivers' opinions on the problem,
(2) studying the effectiveness of selected countermeasures, (3) studying the legal issues
related to photo-enforcement. A crash statistics study, telephone survey, and extended
monitoring of a selected intersection were the three major investigations chosen to
estimate the magnitude of the problem. The crash statistics for the 1997-1999 period
showed that 22% of signalized intersection crashes in Indiana resulted from RLR. RLR
30


preceded 50% of fatal crashes at these intersections. The telephone survey showed that
67% of Indiana drivers felt that RLR was a problem in the state, and 12% of them
claimed to have been involved in a RLR crash. The extended monitoring of the through
movements at the study intersection also recorded a considerable violation rate. Traffic at
a selected intersection in West Lafayette, Indiana, was videotaped and the video material
was used to detect the red light violations. The expected number of drivers arriving at the
start of the red signal has been proposed as a true measure of exposure to RLR. The
authors call it an opportunity for RLR. This exposure was used to estimate the RLR rate.
The statistical significance of the difference in the RLR rates between different periods
was estimated using binomial distribution. Photo-enforcement reduced the violation rate
by 62% during the week of enforcement and by 35% during the week immediately
following the start of enforcement. (Tarko & Reddy, 2003)
As a conclusion of this section of the chapter, it seems clear that most studies
agreed on a certain level of improvement associated with the installation and operation of
RLC programs. Studies were made using different methodologies, time periods, data,
and locations; however, they all concluded that there were positive implications of RLC.
Table 2 briefly presents all of these studies and their findings.
31


Table 2 Summary of the recent studies of RLC effectiveness on safety
Study Title Location Time Period/ Data Type Findings
Evaluating the Effectiveness of Red-Light Running Camera Enforcement in Raleigh, North Carolina (Eiummer & Cunningham, 2010) Raleigh, North Carolina 4 years (2004-2007) Positive safety results. (Significant in 2 groups)
Analysis of the RLC effectiveness on reducing red light violations and injury crashes.
5 groups data sets
Can we make red light runners stop? Red light photo enforcement in San Francisco, California (Fleck. J, 1999) San Francisco. California 1994 Vehicles violating RLC decreased by 40%
a pilot red-light photo- enforcement program analysis + Intention for future expansion of the program
Red Light Running in Iowa: Automated Enforcement Program Evaluation with Bayesian Analysis (Hallmark, Orellana, Fitzsimmons, McDonald, & Matulac, 2010) Davenport. Iowa 2004 RLC us effective in reducing total crashes and RLR crashes
Two years of after installation data including control intersections
Evaluation of the City of Houston digital automated red light camera program (Dahnke, Stevenson, Stein, & Lomax, 2008) The Center of Civic Engagement at Rice University 2001-2006 of crashes data included 70 of monitored and non-monitored approaches Monitored approaches crashes decreased significantly relative to the non-monitored approaches
Houston. Texas
The impact of red light cameras on safety in Arizona (Shin & Washington, 2007) Phoenix and Scottsdale, Arizona 2000-2005 of before data crashes Positive safety improvement + more effectiveness results in Phoenix.
The Impact of Red Light Cameras (Photo-Red Enforcement) on Crashes in Virginia (Garber, Miller, Abel, Eslambolchi, & Korukonda, 2007) Six jurisdictions in Virginia (Alexandria, Arlington, Fairfax City, Fairfax County, Falls Church, Vienna) 1998-2004 of crashes data Modest reduction on comprehensive injury crashes
Traffic Conflict Studies Before and After Introduction of Red- Light Running Photo Enforcement in Maine (Garder, 2006) Maine September 2004 to August 2005 28% decrease of low speeds and within the first second Infractions + increase of infractions at more than 3 seconds into red and at speeds above 35 mph.
The reduction of red light running violations as a result of warning letters only.
Evaluation of safety enforcement on changing driver behavior (Tarko & Reddy, 2003) West Lafayette, Indiana Crash statistics for the 1997- 1999 period The photo-enforcement reduced the violation rate by 62% during the week of enforcement and by 35% during the week immediately following.
32


Effectiveness of RLC on Type of Crashes
When reviewing studies concerned about the type of crashes at signalized
intersections, it looks obvious that the crash type that is targeted in the analyses is the
right-angle (T-bone) crashes, which involve a violating vehicle colliding with another
vehicle crossing the intersection legally on a green signal display. Another crash type
likely to be investigated is a vehicle turning left colliding with a vehicle moving through
the intersection from the opposite approach direction. For this later scenario, the turning
vehicle could be violating the red when the opposite direction has a green, or vice-versa.
On the other hand, there is a concern that rear-end crashes of vehicles approaching the
intersection will increase with RLC enforcement. Knowing that there is a camera system,
and on seeing the yellow display, a more cautious motorist may stop more abruptly,
causing the following motorist, not anticipating the need to stop and likely to be
following too closely, to hit the lead vehicle from behind. Assuming that these crash
types produce equal crash severity, then a net benefit would accrue if the crash reductions
of the angle type exceeded any crash increases of the rear-end type. In general, angle
crashes are usually more severe and, therefore, even a zero change in total crashes may
prove to be safer, if there is a smaller proportion of angle to rear-end crashes with the use
of cameras.
Red-light camera enforcement offers potential as a cost-effective, powerful tool in
reducing red-light running and associated crashes. However, studies on the effectiveness
of the red-light camera system have shown mixed results in terms the of types of crashes
associated with the system, with some studies showing a reduction in T-bone red-light
related crashes, while others report no significant improvement. Furthermore, most
33


studies have shown that red-light camera systems increased rear-end crashes. (Elmitiny &
Radwan, 2008)
As with all synthesis documents, a comprehensive report published by the
National Cooperative Highway Research Program was performed that relied exclusively
on available information; no new data collection or analysis. The information came from
published literature, various websites, and from a questionnaire sent to more than 50
jurisdictions nationwide and some foreign countries known or believed to have installed
red light running camera systems. The findings that can be drawn from the information
complied by that study are as follows. There is a preponderance of evidence, albeit
inconclusive, indicating that red light running camera systems improve the overall safety
of intersections where they are used. As expected, angle crashes are usually reduced and,
in some situations, rear-end crashes increase, but to a lesser extent. (The National
Cooperative Highway Research Program, 2003)
As an example, before-after crash results for Sacramento California are shown on
Table (3) below. (McGee & Eccles, 2006)
Table 3 Results of one-year before/after study Sacramento California (McGee & Eccles, 2006)
Crashes No. of Crashes 12 Months No. of Crashes 12 Months Change (%)
Before Installation After Installation
Total number of crashes 81 73 -10
Injury crashes 60 44 -27
Right-angle crashes 42 31 -26
Rear-end crashes 32 28 -12
Red light crashes 28 17 -39
34


The effectiveness of a group of red light camera installations in Sydney in
reducing right angle and right- (left) turn opposed crashes was analyzed using crash data
from 2 years before and 2 years after the cameras were installed (See Table 4). The study,
published in 1993, had 6 cameras circulating in 16 intersections with cameras and
covered another 16 intersections as control (the control sites were matched on the basis of
crash history, traffic volume, and intersection configuration). The camera (treatment) and
control sites were grouped as follows: Eight most-used camera sites, eight least-used
camera sites, eight most-used control sites, and eight least-used control sites. The study
concluded that red light cameras reduce target crashes and increase rear end crashes with
an overall reduction in accident numbers and severity that was similar to other
engineering countermeasures. (Hillier, Ronczka, & Schnerring, 1993)
Table 4 Before and after changes in crashes, Sydney, Australia (Hillier, Ronczka, & Schnerring, 1993)
Intersection Group (%) Change in Target Crashes (%) Change in Rear-End Crashes (%) Change in Overall Casualty Crashes
Most-used camera sites -48 +62 -28
Least-used camera sites -49 +27 -33
Most-used control sites +2 -29 +17
Least-used control sites other countermeasures -52 -18 -39
According to one of the most comprehensive studies to date on RLCs, a report
from FHWA titled as Safety Evaluation of Red-Light Cameras, which included data from
seven jurisdictions (Baltimore, MD; Charlotte, NC; El Cajon, CA; Howard County and
Montgomery County, MD; and San Diego; and San Francisco, CA) and 132 intersections,
35


concluded that the use of RLCs led to the following: 25 percent decrease in total right-
angle crashes, 16 percent reduction in injury right-angle crashes, 15 percent increase in
total rear-end crashes, and 24 percent increase in injury rear-end crashes. As Table (5)
below shows, the direction of these effects was remarkably consistent across jurisdictions.
The analysis indicated a modest spillover effect on right-angle crashes; however, this was
not mirrored by the increase in rear end crashes seen in the treatment group, which
detracts somewhat from the credibility of this result as evidence of a general deterrence
effect. (Administration, 2005)
Table 5 Results for individual jurisdictions for total crashes (Administration, 2005)
Jurisdiction Number (%) Change in Right Angle (%) Change in Rear End Crashes
Crashes (Standard Error) (Standard Error)
1 -40.0 (5.4) 21.3 (17.1)
2 0.8 (9.0) 8.5 (9.8)
3 -14.3(12.5) 15.1 (14.1)
4 -24.7 (8.7) 19.7(11.7)
5 -34.3 (7.6) 38.1 (14.5)
6 -26.1 (4.7) 12.7 (3.4)
7 -24.4(11.2) 7.0(18.5)
The standard error is the standard deviation of the sampling distribution of a statistic. The tenn may also be used to refer to an
estimate of that standard deviation, derived from a particular sample used to compute the estimate. (Everitt, 2003)
Consistent with findings in other regions, the study that was conducted in Arizona
has concluded that angle and left-turn crashes are reduced in general, while rear-end
crashes tend to increase as a result of RLCs. In Scottsdale, for instance, the crash trends
suggest that an effort to reduce angle crashes through the use of RLCs may be
36


worthwhile, since angle crashes are generally more severe than rear-end crashes (See
Figure 10 below). (Shin & Washington, 2007)
Single Vehicle
8.78%
/ 8.44%
Sideswipe-same
direction
OTHER
11.47%
/
I
/
/
Sideswipe opposite
direction
0.83%
Angle
16.42%
Rear-end /
39.06% "
\
\_ Lett-tum
14.28%
Figure 10 Percentage of crash type in Scottsdale for 14-year period. (Shin & Washington, 2007)
The Virginia study (presented earlier as part of RLC effectiveness on safety
section) that includes six different jurisdictions found that cameras are associated with an
increase in rear-end crashes (about 27% or 42% depending on the statistical method used)
and a decrease in red light running crashes (about 8% or 42% depending on the statistical
method used). It also shows that there is significant variation by intersection and by
jurisdiction: one jurisdiction (Arlington) suggests that cameras are associated with an
increase in all six crash types that were explicitly studied (rear-end, angle, red light
running, injury red light running, total injury, and total) whereas two other jurisdictions
saw decreases in most of these crash types. (Garber, Miller, Abel, Eslambolchi, &
Korukonda, 2007)
37


Effectiveness of RLC on Crashes Severity
According to the Insurance Institute for Highway Safety, during the period from
1992 to 1998, almost 6,000 people (approximately 850 per year) died in RLR crashes in
the Unites States, and another 1.4 million (approximately 200,000 per year) were injured
in crashes that involved red light running.
Using 1997 data from the General Estimates System and a narrower definition of
RLR crashes, Smith, et. al, estimated that approximately 97,000 crashes, resulting in 961
fatalities, could be attributed to red light running in the United States per year during this
same period. Table 6 shows the distribution of crashes by severity for all signalized
intersections, those involving angle crashes, and those considered to be the result of red
light running. As seen, slightly more than 44% of the fatalities at signalized intersections
were attributed to red light running. (McGee & Eccles, 2006)
Table 6 The distribution of crashes by severity for all signalized intersections 1997 (McGee & Eccles,
2006)
Crashes Measure Signalized Intersections Angle Crashes at Red Light Running
Signalized Intersections
Fatal crashes 2,176 1,587 961 (44%)
Injury crashes 318,000 261,000 51,000(16%)
PDO crashes 469,000 361,000 45,000 (9.5%)
Total crashes 789,000 623,000 97,000 (12%)
Fatalities 2,344 1,729 1,059(45%)
Injuries 543,000 464,000 91,000(16%)
Note: Percentage is calculated out of total crashes at signalized intersections.
A study was done to evaluate the crash effects of 87 signed fixed digital speed
and red light cameras and accompanying warning signs placed at 77 signalized
intersections across Victoria, Australia. Across the 77 intersections where the cameras
38


evaluated were installed, it was estimated that 17 serious or fatal crashes per year and 36
minor injury crashes would be prevented, representing crash cost savings to the
community of over $8 million per year. Based on the outcomes of the evaluation,
continued and expanded use of combined fixed red-light and speed cameras in Victoria is
expected to improve driver safety, save lives and reduce crash related costs. Analysis
results estimated large decreases in casualty crashes associated with the FDSRL cameras
and their associated signage. When only the crashes involving vehicles travelling from
the approach intersection leg where the camera was placed are considered, the estimated
casualty crash reduction was 47%. When crashes involving vehicles from all approaches
are compared, the estimated casualty crash reduction was 26%. A 44% reduction in right
angle and right turn against crashes, those particularly targeted by red light enforcement,
was also estimated. While use of the FDSRL cameras was associated with a reduction in
overall casualty crash risk, there was no evidence for a reduction in relative crash severity
meaning the cameras were associated equally with reductions in minor injury crashes as
serious injury and fatal crashes. (Budd, Scully, & Newstead, 2011)
An article examines the effectiveness of red-light cameras at reducing the rate of
violations as well as the level and severity of intersection-related crashes. Although the
evaluations differ in sample size, type of intersection and evaluation methods, several
trends emerge. The findings suggest that if installed at locations with significant red-light
running crashes and/or violations, red-light cameras substantially reduce red-light
violation rates and reduce crashes that result from red-light running. Although they may
not reduce total crashes, they usually are effective at reducing crash severity. The author
39


finally suggested that red-light cameras enforcement should not be seen as a substitute for
proper traffic engineering of signalized intersections. (Bochner & Walden, 2010)
A study developed a Bayesian HBL (hierarchical binomial logistic) model to
identify the risk factors on individual severity of driver injury and vehicle damage at
urban intersections of Singapore. For the study to conclude significant findings, it was
helpful to account for the severity correlation of driver-vehicle units involved in the same
multi-vehicle crashes. The study included various geometric features, traffic conditions,
and driver-vehicle characteristics, as well as nine variables identified as significant using
95% BCI (Bayesian credible interval). Among these, the crash-level significant factors
are Time of Day, Intersection Type, Nature of Lane, Street Lighting, Presence of Red
Light Camera, and Pedestrian Involved. In particular, it was found that crashes occurring
in peak time, in good street-lighting condition, and in the case of pedestrians involved are
associated with lower severity, while those occurring in night time, at T/Y type
intersections, on right-most lane, and in the presence of red light cameras have larger
odds of being severe. Vehicle type, Driver Age and Involvement of Offending Party were
also found to affect severities of driver injury and vehicle damage significantly.
Specifically, results indicated that heavy vehicles have a better resistance to serious
injury or extensive damage, while two-wheel vehicles, young or aged drivers, with the
involvement of offending party have a higher risk of being high severity. (Helai, Chor, &
Haque, 2008)
The study that included Phoenix and Scottsdale (Arizona) also investigated the
severity of crashes occurred at RLC intersections. It concluded that injury and fatal
crashes of approximately 16.95 per year occurred at RLC intersections of Phoenix,
40


compared to 10.43 per year in Scottsdale. Additionally, the number of rear-end crashes
resulting in injuries or fatalities (5.67/year) is higher than that of angle crashes
(2.41/year), as was found previously. Further examinations however, again show that
angle crashes are more serious than rear-end crashes.
Figures (11) and (12) below show the proportion of crashes by severity. The
percentage of PDO crashes and minor crashes for rear-end crashes is higher than the
percentage of injury/fatal crashes. (Shin & Washington, 2007)
100.0%
00.0%
B0.0%
70.0% -
60.0%
50.0% -
40.0%
30.0% -
20.0%
10.0% -
0.0%
PDO lnjury+ Fatal

AX
A1
o
Figure 11 Percentage of crashes per year by crash type and severity (PDO vs. injury and fatal). (Shin &
Washington, 2007)
41


100.0%
90.0%
S0.0%
70.0%
50.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%





>'

(S'
' ,-sT
r


Figure 12 Percentage of crashes per year by crash type and severity (minor vs. major). (Shin &
Washington, 2007)
A study shows the attitude of people toward red light cameras in 14 cities with red light
camera programs concluded that two thirds favor the use of cameras for red light
enforcement and 42 percent strongly favor it. The chief reasons for opposing cameras
were the perceptions that cameras make mistakes and that the motivation for installing
them is revenue, not safety. Forty-one percent of drivers favor using cameras to enforce
right-tum-on-red violations. Nearly 9 in 10 drivers were aware of the camera
enforcement programs in their cities, and 59 percent of these drivers believed that the
cameras have made intersections safer. Almost half know someone who received a red
light camera citation, and 17 percent had received at least one ticket themselves. When
compared with drivers in the 14 cities with camera programs, the percentage of drivers in
Houston who strongly favored enforcement was about the same (45%), but strong
42


opposition was higher in Houston than in the other cities (28 versus 18%) (Mccartt,
2012)
Characteristics of Red Light Runners
Knowing the characteristics of the red light runners has been another point of
interest to many researchers. It is another way to mitigate this serious problem that is
considered among the most risky behaviors in the transportation system by defining the
characteristics of those drivers who run red lights more frequently comparing to others.
A study was conducted in Southeast Virginia that includes eight intersections and
covers an 8-month period during which photo enforcement cameras were installed at
three sites (Al, A2, and A3). As Table (7) shows, data collectors observed 1765 light
cycles. Overall, 18.8% of last drivers entered intersections on green lights, 68.4% on
yellow, and 12.7% on red. Demographics were recorded for 1433 drivers (only the
yellow and red light runners). Demographics of red light runners across the five data
collection periods are provided in table 7. The numbers represent the percent of red light
runners out of all yellow and red light runners during that observation period broken
down by subcategories for each demographic variable. Overall, men had higher raw red
light running rates than women; however, the only significant difference between men
and women occurred in Phase 1. Red light running rates for both men and women
declined from baseline levels and reached their lowest levels during Phase 4. The only
significant difference in red light running as a function of ethnic group classification was
during Phase 2 when non-whites were more likely to run red lights than whites. Note that
numbers in parentheses are the sample sizes for categories each phase of the project. The
43


percent represents those who ran the red light as opposed to the yellow light. (Martinez &
Porter, 2006). The phases (collection periods) are:
Phases 1 and 2: These observations took place in June and July 2004, respectively,
before any cameras were installed and served as baseline measures of red light running
behavior.
Phase 3: In September 2004, observations occurred again. Intersection A1
received cameras and was in the 30-day warning period (i.e., when warning letters were
mailed to the registered owners of vehicles that ran the red light).
Phase 4: This observation took place in November 2004. Intersection A1 was in
the actual citation phase, A2 was in the warning phase, and cameras at A3 were being
tested to go operational the day after it was was observed. (Note that this observation
phase took place in November when it would get dark about 5 p.m. and that the camera
flash allowed for no mistake that it was functional.)
Phase 5: The fifth observation phase occurred in January 2005 when Al, A2, and
A3 were issuing citations.
Table 7 Percent of last drivers running a red light by demographic category. (Martinez & Porter, 2006)
Demographics Phase 1 Phase 2 Phase 3 Phase 4 Phase 5
Gender
Female 13.0 (108) 16.7 (96) 17.0 (88) 8.0 (50) 10.1 (69)
Male 25.6 (207) 18.5 (162) 19.5 (128) 10.8 (93) 14.4(111)
Race
White 20.3 (177) 14.1 (149) 17.0 (147) 11.0(100) 15.4(123)
44


Demographics Phase 1 Phase 2 Phase 3 Phase 4 Phase 5
Non-White 24.3 (103) 21.5(93) 18.1 (83) 9.8(41) 10.4 (47)
Safety Belt Use
Yes 20.4(113) 10.5 (114) 12.4(113) 4.3 (70) 15.7(102)
No 27.3 (77) 22.6 (62) 23.4 (77) 13.9 (36) 10.4 (48)
Age Group
25 or younger 24.5 (102) 23.1(65) 22.2 (54) 11.1 (36) 15.9(44)
26-35 19.0 (84) 16.8(107) 26.8(71) 6.4 (47) 12.9 (62)
36 and older 19.7 (76) 11.3 (62) 9.0 (78) 10.9 (46) 16.0 (50)
Number of People in vehicle
1 17.3 (260) 18.5 (211) 20.4 (201) 7.9(126) 11.4(158)
2 or more 32.4 (68) 14.3 (63) 13.6 (59) 15.6 (32) 19.5 (41)
Note: The word phase refers to project phases
Another report conducted by the Volpe National Transportation Systems Center
presents results from an analysis of about 47,000 red light violation records collected
from 11 RLC equipped intersections in the City of Sacramento, California, between May
1999 and June 2003. The report used seven different variables to study the
characteristics of red light runners, these variables are:
Age of the violator
Gender of the violator
Time (in hours) when the violation occurred
Model Year of the vehicle driven by the violator
45


Measured vehicle speed at the time of the violation
Elapsed time from the onset of red signal until the time of the violation
The distribution of repeat red light offenders
The report suggests that younger drivers under 30 years of age are more likely to run red
lights than drivers in other age groups (See Figure 13) (Note: Distribution of red light
violators that is shown in the figure is categorized by 7 age groups and included data on
the number of licensed drivers (LDs) in California, the total million vehicle miles
traveled (MVMT) and relative ratios of red light violation (RLV) percentages by licensed
% of RLV
driver percentages and total MVMT percentages). Relative ratios for % f LB ancj
% of RLV
% of MVMT
were plotted and presented in Figure 13. Additionally, the report indicates
that about 56 percent of the violators were traveling at or below the posted speed limit
(See Figure 14). Moreover, 94 percent of the violations occurred within 2 seconds after
the onset of red light, and only 3 percent of the violations were recorded 5 seconds after
the onset of red light. Approximately 4 percent of the violators were repeat offenders.
46


Number of Red Light Violation Record!
Figure 13 Normalized red light violation values by age group (Yang & Najm, 2006)
Figure 14 Distributions of red light violation records by vehicle speed (Yang & Najm, 2006)
Figure 15 illustrates the distribution of Sacramentos red light violations by
violation time (in hours). The overall trend shown in this figure is consistent with the
47


expectation most of red light violations occurred during the daytime hours when most
urban driving is done (i.e., 7 AM to 7 PM). However, the highest count of red light
violations during the time period from 2:00 PM to 2:59 PM is somewhat surprising.
Finally, red light violations rates are estimated between 6 and 29 violations per 100,000
intersection-crossing vehicles. (Yang & Najm, 2006)
AM .AM AM AM AM AM AM AM AM AM AM AM PM PM PM PM PM PM PM PM PM PM PM PM
Time of ihe Day
Figure 15 Distribution of red light violation by time of day (Yang & Najm, 2006)
A study introduced by the National Highway Council concluded that 96 percent
of drivers in a recent survey fear they will get hit by another vehicle running a red light
when they enter an intersection. Some 800 licensed drivers aged 18-65 were polled. Two-
thirds of the respondents see other drivers run red lights every day, with 54% speculating
that the culprits were in a hurry. The National Highway Traffic Safety Administration
48


counted 1,114 traffic deaths in 1997 in intersections where drivers failed to heed red-light
signals. (Karr, 1999)
RLC and Signal Timings
Two principal methods used to reduce red light running involve lengthening the
duration of yellow change intervals and automated red light enforcement. These two
types of countermeasures were usually supported by studies from different points of view
that tried to conclude which of them is more efficient.
A study evaluated the incremental effects on red light running of first lengthening
yellow signal timing, followed by the introduction of red light cameras. At six
approaches to two intersections in Philadelphia, Pennsylvania, yellow change intervals
were increased by about Is, followed several months later by red light camera
enforcement. The number of red light violations was monitored before changes were
implemented, several weeks after yellow timing changes were made, and about 1 year
after commencement of red light camera enforcement. Similar observations were
conducted at three comparison intersections in a neighboring state where red light
cameras were not used and yellow timing remained constant. Results showed that yellow
timing changes reduced red light violations by 36%. The addition of red light camera
enforcement further reduced red light violations by 96% beyond levels achieved by the
longer yellow timing. As a conclusion, the study shows that the provision of adequate
yellow signal timing reduces red light running, but longer yellow timing alone does not
eliminate the need for better enforcement, which can be provided effectively by red light
cameras. (Retting & Williams, 1996)
49


The City Council for the City of Springfield, Missouri, approved a contract to
install up to sixteen cameras for automated red light enforcement in the spring of 2006.
During the implementation phase of the program, test sampling of potential intersections
for placement of the cameras revealed significant differences in yellow timings and red
light running at city signals compared to Missouri DOT signals inside the city. This
difference prompted city and state traffic engineers to review their respective methods of
calculating the yellow and all-red timings. Despite using the same equation recommended
by ITE, the agencies used different assumptions for perception-reaction time and how to
interpret and use the results. City and state traffic engineers came to agreement and
documented the assumptions to be used in a Memo of Understanding (MOU) to bring
consistency to the yellow and all-red timings throughout the city. The result was that
yellow time at all city signals was increased and yellow time at nearly all state signals
was decreased. All signals were retimed in conformance with the MOU in the spring of
2008 and in conformance to ITE recommended practice, three months prior to the first
red light camera startup and 18 months prior to the installation of a camera on an
intersection where the yellow time had been reduced. The result of the signal retiming
has brought credibility to the red light camera program for the public and media with a
reduction in rear-end crashes in addition to a reduction in total crashes at traffic signals.
(Newman, 2010)
Methodologies and Procedures Used for RLC Analysis
Several types of methodologies and analysis procedures were used such as the
Binary model, which was preliminarily developed to examine how the stopping-crossing
50


decision of drivers at the onset of amber is affected by geometric, traffic, and situational
variables.
Results showed that the presence of RLCs is one of the five significant factors
affecting a drivers decision to cross at the onset of amber. A Multinomial logic model
further confirmed that RLCs are effective in reducing RLR frequency. Further analysis
on the fitted models revealed that while the presence of RLCs is effective in reducing risk
of right-angle crashes, it has a mixed effect on the risk of rear-end crashes. Whether the
RLC reduces or increases the possibility of rear-end crashes depends on the speed of the
trailing vehicle and the headway between vehicles. (Helai, Chor, & Haque, 2008)
Another study conducted by the Federal Highway Administration used the
Empirical bays for before and after crashes data from 132 treatment sites. Crash effects
detected were consistent in direction with those found in many previous studies:
51
TB!MAPCAJor*htop ijrunman iVfi's&bri Systems


decreased right-angle crashes and increased rear end ones. The economic analysis
examined the extent to which the increase in rear end crashes negates the benefits for
decreased right-angle crashes. There was indeed a modest aggregate crash cost benefit of
RLC systems. The study concluded that economic benefits (see Table 8) could go to its
highest level during the occurrence of the highest total entering average annual daily
traffic, the largest ratios of right-angle to rear end crashes, and the presence of protected
left-turn phases.
Note that FHA used samples (K+B+C+A) to refer to different crash cost levels. A
refers to cost estimate of fatal and serious crash levels, K for injury estimate of right
angle crashes, and B- and C- level refer to injury estimate of rear end and left turn crashes.
(Federal Highway Administration, 2005)
Table 8 Unit crash cost estimates by severity level used in the economic effects analysis. (Federal
Highway Administration, 2005)
Crash severity level Right-angle crash cost Rear end crash cost
O (Standard deviation) $ 8673 (1285) $11463 (3338)
K+A+B+C (Standard deviation) $64468 (11919) $53659 (9276)
Note: In this study, (K+A+B+C) are all combined to refer to injury level crashes due to inconsistent sample
size while 0 refers to non-injury level crashes. (Administration, 2005)
A meta-analysis was used to determine the effects of red-light cameras (RLCs) on
intersection crashes. The study shows that the size and direction of results reported from
studies included in the meta-analysis are strongly affected by study methodology. The
studies that have controlled for most confounding factors yield the least favorable results.
Based on these studies, installation of RLCs leads to an overall increase in the number of
52


crashes by about 15%. Rear-end crashes increase by about 40% and right angle crashes,
which are the target crashes for RLC, are reduced by about 10%. All effects are, however,
non-significant. Meta-regression analysis shows that results are more favorable when
there is a lack of control for regression to the mean (RTM). (Erke, 2009)
RLC Spillover Effect (Halo Effect)
There is some but not much evidence that RLR cameras will not only deter
motorists from violating a signal at intersections equipped with cameras, but will also
modify driver behavior at other intersections. If cameras do have an effect on driver
behavior beyond those intersections where the cameras are used, then the other
intersections in the area will likely also experience a decrease in angle crashes. This is a
spillover effect or a halo effect.
A study of an RLR camera program in Oxnard, California, found a decrease in
crashes at intersections with cameras and intersections without cameras. The studys
authors attributed this reduction to spillover. (Retting R. A., 2002). On the other hand, an
evaluation of cameras in Sydney, Australia, did not find a significant reduction in RLR-
related crashes at intersections without cameras. The authors concluded that spillover did
not occur at non-camera intersections used as control group intersections. (IMBERGER,
2003)
A national study involving multiple jurisdictions has yet to prove that this red
light camera spillover effect does or does not occur. Consequently, the studies suggested
that agencies should consider the possibility of this spillover in their evaluation of RLR
cameras and modify their methodology or conclusions accordingly. Also, agencies
53


(according to the author) may want to evaluate and quantify the spillover effect in
addition to the effect at intersections equipped with cameras. (McGee & Eccles, 2006)
The effectiveness of red light running cameras in reducing the number of drivers
who run the red light in Clive, Iowa was evaluated. The number of red light running
violations at camera-enforced intersection approaches were compared to violations at
approaches at intersections where cameras were not used within the same metropolitan
area using a cross-sectional analysis. A Poisson lognormal regression was used to
evaluate the effectiveness of the cameras in reducing violations. Results indicated that red
light running cameras substantially reduced the number of violations at camera-enforced
approaches as compared to control approaches. (Fitzsimmons, Hallmark, Orellana,
McDonald, & Matulac, 2009)
In June 2010, Arlington County, Virginia, installed red light cameras at four
heavily traveled signalized intersections. A study examined the effects of the camera
enforcement on red light violations. Traffic was videotaped during the one month
warning period and both one month and one year after ticketing began at 12 signalized
intersections, including the four camera intersections, four spillover intersections
without cameras in Arlington County (two on the same travel corridors as the camera
intersections and two on different travel corridors), and four control intersections
without cameras in adjacent Fairfax County. Rates of red light violations per 10,000
vehicles were computed.
54


Figure 17 Intersections studied in Arlington Virginia (McCartt & Hu, 2013)
Consistent with prior research, there were significant reductions in red light
violations at camera-enforced intersections. These reductions were greater the more time
that passed since the light turned red, when violations are more likely to result in crashes.
Spillover benefits were observed only for nearby intersections on the same travel
corridor, and these were not always statistically significant. At intersections on other
travel corridors, red light running increased compared with expected rates based on the
control intersections. (See Table 9)
The study concluded that this evaluation examined the first year of Arlington
Countys red light camera program only, which was modest in scope and without
ongoing publicity. A larger, more widely publicized program likely is needed to achieve
community-wide effects. (McCartt & Hu, 2013)
55


Table 9 Observed red light violation rates per 10,000 vehicles by time into red signal phase and percentage changes 1 month and 1 year after red light
camera ticketing began, compared with warning period. (McCartt & Hu, 2013)
Violation rates per 10,000 vehicles by time (seconds) into Percent change in violation rates
red compared with warning period
1 Month after 1 Year after 1 Month after 1 Year after
Warning Period ticketing ticketing ticketing ticketing
>0.5 >1 >1.5 >0.5 >1 >1.5 >0.5 >1 >1.5 >0.5 >1 >1.5 >0.5 >1 >1.5
Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec Sec
Arlington County
Camera intersections 11.7 5.8 3 11.6 4.7 1.6 8.9 4.1 1.5 -l -20 -47 -24 -30 -50
Corridor spillover intersections 19.3 10.3 4.7 12.6 6.7 3.2 20.1 10.2 6.1 -35 -35 -31 4 -1 30
Non-corridor spillover intersections 1.7 0.4 0.4 4.3 2 1.3 4.8 2.9 1.6 159 434 240 184 688 343
Fairfax County control intersections 6.9 2.8 0.5 8.6 2.8 1.5 8.9 4 1.8 25 -2 228 30 44 283
Os


RLC Site Selections
Although there are many studies that have investigated the safety improvement of
the RLC system, there are relatively very few studies that have covered the RLC sites
selection and where/ when to implement them especially when considering the cost
associated with the system.
A general study provides a tool for identifying and priority-ranking problem
intersections with respect to red light running within the entire roadway network under
the jurisdiction of a particular agency. The tool includes three steps as a guide to
estimate the safety changes upon installation of a red light camera at a signalized
intersection. These three steps are: empirical Bayes method, collision prediction models,
and collision modification factors (the assumption of negative binomial error distribution
was used for developing the last step). (Hadayeghi, Malone, Suggett, & Reid, 2007)
The city of Durham wished to explore the feasibility of implementing a red light
camera program. Particularly, they wanted to ensure that the sites were selected in an
objective and defensible manner based on sound traffic engineering judgment. The study
concluded that RLC sites selection criteria could be based on two main elements: the
overrepresentation of angle crashes and the higher than expected number of crashes.
Additionally, the study used a more detailed field investigation to observe things like the
signal timings, intersection layout, traffic signal type and placement, prevailing traffic
patterns and operating speeds, and the suitability of each approach for a red light camera.
Based on the review, a short list of candidate sites/approaches was developed. For
approaches remaining on the short list, it was suggested that the occurrence of red light
running be confirmed through a detailed violation records, and rear end crashes be
57


closely monitored in the post-implementation period. (Suggett, Malone, & Borchuk,
2005)
The Intersection Safety Camera Program (ISCP) in British Columbia, Canada, has
proved effective in reducing the frequency of crashes at locations where red light cameras
have been deployed. Post-implementation evaluations of ISCP conducted by the
Insurance Corporation of British Columbia detected a 14% reduction in crashes resulting
in injuries 18 months after the program was implemented. A follow-up study conducted
36 months after ISCP implementation examined the safety performance of ISCP and
found that the rate of crashes resulting in injuries was reduced by 6.4%. Given the
ongoing and long-term success of ISCP at reducing crashes, it was decided that the
program should be expanded. To support ISCP expansion, it was necessary to examine
how the program had been implemented and to learn from the results of the previous
program evaluations. A critical element of ISCP is the selection of sites to be targeted for
deployment of intersection safety cameras. The sites selected should have a demonstrated
safety problem, such as results from previous evaluations of intersection safety camera
after installation. In addition, sites should be selected such that the life-cycle cost of
deployment of the intersection safety camera will be less than the safety benefits that will
accrue from reduced numbers of crashes and the associated costs, (de Leur & Milner,
2011)
The twelfth offering of a Mentors Program at Texas A&M University on
Advanced Surface Transportation Systems presented a document in 2002 by the
Advanced Institute in Transportation Systems Operations and Management. One of the
papers that was discussed and presented was about the criterion of sites selection. The
58


paper has an introduction that indicates the lack of papers and studies related to RLC site
selection and the actual need for a uniform set of criteria to aid traffic engineers and cities
in the site selection process. From a survey that included most of US cities that have
RLC installed in their transportation system, results showed that the three most
commonly used criteria are the history of red light crashes, red light citations, and
engineering issues associated with intersection.
The author of the paper suggested a set of major and minor guidelines that should
be used. The guidelines are developed using successful experiences with similar
situations. Guidelines are used when a policy would be too limiting or confining, or for
situations that are highly variable. They allow careful assessment of intersection
conditions that are indicators for the need of traffic control devices or engineering
countermeasures. These guidelines were presented as follows:
Major guidelines
1. Accident History
The use of accident statistics can be helpful in this area, though the author disagrees with
the total reliance on them alone. Accident statistics should be used to identify problem
areas that need to be investigated further.
2. Red Light Citation History
The evidence that there is a problem is a good indicator that countermeasures need to be
implemented. Usually the presence of these citations is more of an indicator of allocation
of available police resources and the relative safety of enforcing the law. Nevertheless,
this is a red flag that should alert one to possible problem intersections though others may
exist.
59


3. Approach Speeds
As the speed increases, the severity of resulting crashes increase as well. There may be
situations where there are high approach speeds and high violations, but few crashes.
These areas need to be further evaluated and closely monitored for possible development
of crashes.
Minor guidelines
1. Traffic and Pedestrian Volumes
Generally higher traffic volumes relate to a greater probability of violations and crashes.
This is of particular concern when the cross-street traffic volumes are also high as well.
While the crash of two vehicles can result in either injury or death, the same is not true of
a vehicle-pedestrian collision. Intuitively, the pedestrian is almost always killed or
severely injured when a vehicle runs a red light and collides with them. As a result,
intersections with high pedestrian volumes and/ or traffic volumes need to be closely
examined.
2. Intersection Degree of Saturation
With higher degrees of saturations at intersections the headway gaps between vehicles are
smaller. Consequentially, there is a greater probably of a vehicle running a red light
whether intentionally or unintentionally. The difference between these two intentions
needs to be recognized and quantified.
3. Perceived Benefit to Cost
As with all engineering countermeasures, the costs associated with the installation of a
system should be evaluated. These costs do not only include the costs of construction and
equipment, but also the beneficial costs received from the reduction in crashes due to red
60


light violators. (Qu, 2003)
From analysis of the data, there are intersections in Rhode Island where RLR is a
problem. A model is developed to prioritize intersections based on a Composite
Intersection Index (CII), where the highest score indicates the most problematic
intersection. The CII is based on a comprehensive set of variables including the
following: (1) the entering average daily traffic (ADT) (in 10,000s of vehicles) per
number of lanes entering the intersection; (2) the rate of RLR violations occurring after 1
second; (3) the number of phases; and the (4) average approach speed (based on approach
speed limits). (Hunter, 2003)
61


CHAPTER IV
METHODOLOGY
Introduction
Although red light cameras are widely used to discourage red light running,
relatively few studies have been done on how to best deploy these cameras to maximize
their efficacy. Due to their high price tags, many jurisdictions will purchase a relatively
smaller number of cameras and rotate them among a larger number of camera-ready
intersections. This methodology chapter is divided into two sections. The first section is a
brief description of the reasons behind choosing these cities and a representation of the
data required to complete the study. The second section covers the analytical-based
methodology, which will be used to determine the number of cameras needed to
effectively enforce locations within a certain city limit. The study will use data from the
cities of Colorado Springs, Fort Collins, and Denver to be able to implement the
methodology and derive the findings.
Why These Locations as Case Studies?
There are several reasons for choosing these locations, which I divided into
general and specific reasons:
1) General Reasons
There is an ongoing controversy in the state of Colorado regarding the
effectiveness of red light cameras, which generated two bills to the State Senate during
the years of 2012 and 2013. Areas like RLC is main objective, types of crashes it is
causing, and the involvement in public privacy are some of concerns citizens hold on
RLC, and therefore studies like this will clarify some argumentative points.
62


The case studies represent three cities with different characteristics, which is
another concern associated with implementing RLC systems, and therefore results should
indicate whether the system can or cannot be impacted by different city characteristics.
Additionally, it is obvious that such a topic is highly based on the data collection
stage (as mentioned earlier), which may require several calls, field investigations, and
physical visits to the data sources, besides the need to physically commute to these cities
several times to collect and update some of the collected data. Therefore, data
accessibility is another reason why these cities were chosen.
2) Specific Reasons (effective and ineffective programs)
In Colorado Springs, the RLC program started back in 2010 and was shut down
one year after that for ineffective results. It seems obvious that there was something
implemented or managed differently than in the Fort Collins and Denver cases that have
been running their programs for more years (over 10 years already). Consequently, the
study can make a very solid comparison between current actual RLC locations and the
ones suggested by the study.
Finally, availability of data for a good length of time in the city of Colorado
Springs, Fort Collins, and Denver is a plus for choosing the cities as case studies,
especially with RLC site selection studies that require at least a period of 3 years data to
show meaningful results.
Data Required and Field Investigation
In order to implement the criterion of RLC sites selection, specific data are
required. Most of the data were obtained from the traffic engineering office, police
63


department, and from some field investigation of each city, while other types of data were
not available and therefore not included in the analysis chapter.
The following data were collected for 82 signalized intersections in Colorado
Springs (from 2007-2009), 106 signalized intersections in Fort Collins (from 2010-2012),
and 309 signalized intersections in Denver (from 2010-2012). (Note: all data are for a 3-
year period).
Traffic volume/ approach/ intersection.
Crash Types which were categorized into front to side, rear-end crashes and other.
Crash severity that is divided into fatal crashes, injury crashes, and property
damage crashes.
Approach or direction of the At-fault vehicle.
Intersection characteristics.
Overall final locations of RLC.
Table 10 illustrates the type of data, its source, and time period where data was
collected.
64


Table 10 Data required for RLC sites selection Criterion
Data Name Data Types Source Time Period
Traffic Volume / Approach / Intersection EB, WB, NB, SB City Traffic Engineering Office. 3 Years
Crash Types Front to side, Rear-end City Police Dept. 3 Years
Crash Severity Fatal, Injury, PDO City Police Dept. 3 Years
Vehicle Types Commercial, Pass, Trucks City Police Dept. 3 Years
Crash History Date of Crashes City Police Dept. 3 Years
Economic Evaluation Cost per Crash types, Cost of RLC. City Traffic Engineering Office, Insurance agencies Current
Intersection Characteristics Intersection Layout, Approach Speed, City Traffic Engineering office and Field Investigation Current
Social Structure, and Traffic Signal Timing
Direction of At-Fault Vehicle EB, NB, WB, SB City Police Dept. 3 Years
RLC Selected Locations Overall RLC locations Field Investigation Current
On
Ol


Methodology
There are several possible criteria and procedures that could be used for RLC site
selection, but the availability of data must be taken into account in developing the criteria
and procedures. In the study, there are two main phases, each having specific criteria that
could be followed in order to identify candidate RLC sites.
All signalized intersections with available data in each city will be screened and
tested using the criteria in Phase I. Before moving to Phase II, candidate RLC sites from
each test will be scaled by weighting factors determined by the city stakeholders and
decision makers to make the final ranking list. Finally, qualified intersections will be
evaluated by conducting a comprehensive field investigation in Phase II.
The following provides a detailed description of the two phases and tests under
each of them.
Phase I Includes Four Criteria
1- Criterion of Crashes Severity
Although crash frequency has often been the primary consideration in the
implementation of RLC, crashes differ in severity. There are several levels of crash
severity which should be considered when choosing RLC location. From the literature
review, crash severity is mostly divided into three different levels: crashes resulting in
fatalities, injuries, and property damage only. Each of these crash severities was given a
relative weight representing its impact level.
100 for crashes resulting in fatalities.
10 for crashes resulting in injuries.
66


1 for crashes resulting in property damage only.Crash Severity could be ranked for
all intersections within the city limit by dividing the total number of crashes that are
categorized by severity level and then normalized by the total number of crashes using
the following formula:
Equation 1 Normalized- Crash Serverity Level (Colorado Springs Traffic Engineering Office, 2011)
Where:
N-CSL = Normalized Crashes Severity Level
F = Crashes Resulting in Fatalities.
I = Crashes Resulting in Injuries.
PD = Crashes Resulting in Property Damage.
TC = Total Crashes.
It is important to note that the crash severity level equation may identify sites with
high crash severity but few crashes. Thus, it is also preferred to calculate the crash
severity level without normalizing the equation by the total number of crashes using the
following equation, using the following equation:
(100F) + (10/)+ (l PD)
N-CSL =
TC
Equation 2 Crash Severity Level (Colorado Springs Traffic Engineering Office, 2011)
67


The higher crash severity level is for a specific intersection, the higher that intersection is
ranked among the list of the citys intersections. (This applies to both equations).
Where:
CSL = Crash Severity Level
2- Potential For Improvement (PFI)
The term potential for improvement (PFI) is used as a measure for identification
of the criteria crash rate and crash frequency for all locations included in a city, based on
a certain parameter or benchmark. Figure (19) shows a graphical illustration of the
concept of PFI, and why some values are below the line (negative). In the graphic, sites 1
and 2 have a positive value for the PFI, as they are above the blue line. Conversely, site 3
has a negative value for the PFI, as it is below the line.
In each of the case studies, potential for improvement will be measured in crash
rate and crash frequency. Since average rate was used as a parameter, the locations will
be divided into positive and negative values. Negative values mean that there is no
potential for improvement. In fact, these locations are performing better than normal /
average. The negative crash rate (or frequency) value means that this is the number of
crashes below an average crash rate (or frequency) level and since it is below, there is no
PFI. In the other hand, the positive crash rate and frequency value means that this is the
number of crashes above an average crash rate (frequency) level and since it is above,
there is a PFI.
Normally, collision prediction model (CPM) is used as a parameter to measure
potential for improvement, however, in this study the normal average rate will be used
instead. (See equation 5)
68


Use Potential for Improvement (PFI) as the
measure for identification
Collisions
CPM for 2-lane,
rural highway
* Volume
Figure 18 Illustration of the term potential for improvement.
From three years crashes and volume per approach data that each of the cities has
thankfully provided, potential for improvement can be obtained from the two criteria as
follows:
PFI in Crash Rate (Crash/Movement)
It is a model that has become a standard method for measurement of road safety
performance and especially for RLC candidate sites selection. This criterion can be
calculated by subtracting estimated crash rate (the parameter) from annual crash rate
of each site. As shown in equation 3, this criterion will provide PFI in relation to
crash rate for each intersection based on three years data and it can be mathematically
expressed as follows:
PFI in Crash Rate (Crash / Movement) per intersection =
Annual Crash Rate per intersection Estimated Crash Rate per intersection
Equation 3 PFI in crash rate (ALTurki, 2013)
69


\Y li ere Annual Crash Rate per intersection and Estimated Crash Rate per intersection are
calculated using the following equations:

lOOOOOOJ
Annual Crash Rate per intersection/million vehicles [AADTj X X 3J
Equation 4 Annual crash rate
where,
TC = Total Crashes per intersection.
AADT; = Annual Average Daily Traffic per intersection.
And,
Where:
Estimated Crash Rate per intersection/million vehicles/ year
(Estimated Annual Crashes X 1000000)
(AADT; X 3 65 )
Equation 5 Estimated crash rate
Estimated Annual Crashes Obtained by performing regression analysis.
PFI in Crash Frequency (Crash/ Year)
While PFI in crash rate of each intersection is an important step towards selecting
RLC candidate sites, PFI in crash frequency is another important criterion to consider
when making the candidate lists.
70


Once PFI in collision rate is calculated for each intersection, PFI in crash
frequency can also be calculated by multiplying annual crashes per intersection by
estimated annual crashes. It can be expressed as follows:
PFI in Crash Frequency (Crash/Year) =
Annual Crashes Estimated Annual Crashes
Equation 6 PFI in Crash frequency (ALTurki, 2013)
where,
Annual Crashes = TC/3
As a result, both PFI in crash rate and PFI in crash frequency will make another
two lists of candidate RLC sites based on the potential for improvement criterion. Note
that the higher crash rate for a specific intersection, the higher that intersection is ranked
among the list of the candidate intersections. Similarly, this applies to the crash frequency
candidate list.
3- Criterion of Crashes Types
One of the major argument points with regard to RLC programs is the type of
crashes that usually decrease or increase due to the implementation of the RLC. As
indicated in the literature review chapter, most studies have shown an overall decrease on
the front to side type of crashes and contrarily an increase on rear-end type crashes
wherever an RLC is implemented.
This is an important criterion that could be used as part of RLC sites selection
phase one especially with the availability of such data. From all signalized intersections
within the city limit, crash types are available and the idea here is to avoid locations with
high proportion of rear-end crashes and target those with high proportion front to side
71


type of crashes. The following formula is used to calculate the front to side type of
crashes rate:
X
n
c
72
Ave TC
Figure 19 Calculation of crash type rate (ALTurki, 2013)
where
X Rate of front to side crahes at intersection (n J.
C = Number of front to side crahes at intersection (ti).
4- Weighting Scale for RLC Intersection Candidates
It is very important to note that all criteria from phase I could result in different
top 10 lists of candidate intersections, and according to many previous studies and
projects in the same field, this is a normal scenario.
This step requires a subjective judgment on the part of the group making the
evaluation which is in this case would the city engineering office represented by their city
engineers (See tables 11,12,and 13). Next, each intersection will be ranked in order of
importance and then by using a formula of proportionality to obtain relative ratio weights
(Nicholas J. Garber, 2015).
Wf
i? _______j
J Y" IV-
Equation 7 Proportionality to obtain relative weights (Nicholas J. Garber, 2015)
Where
Kj weighting factor of objective j
Wj relative weight for objective j
72


Next, weighted scores from all top 10 locations from each of the criteria will be
added up to make the final top 10 that combined all the criteria in phase I.
The final top 10 should be processed and moved on towards Phase II for further
analysis to conclude the final RLC candidate list for each city.
Table 11 Weighting percentages for criterions in Phase I for Colorado Springs. (Colorado Springs Traffic
Engineering Office, 2011)
Criterion Name Weight
Normalized Crashes Severity 15%
Crashes Severity 20%
Crashes Rate 20%
Crashes Frequency 30%
Crashes Type 15%
Table 12 Weighting percentages for criterions in Phase I for Fort Collins (Fort Collins Traffic Engineering
Office, 2012).
Criterion Name Weight
Normalized Crashes Severity 20%
Crashes Severity 15%
Crashes Rate 35%
Crashes Frequency 25%
Crashes Type 5%
Table 13 Weighting percentages for criterions in Phase I for Denver (Denver Traffic Engineering Office,
2012).
Criterion Name Weight
Normalized Crashes Severity 15%
Crashes Severity 5%
Crashes Rate 35%
Crashes Frequency 40%
Crashes Type 5%
73


Phase II Includes Seven Criteria
1- Fluctuation of Crashes
For the final top 10 RLC candidate locations, some traditional statistical measures
can be used to see if crash frequencies are historically fluctuated or stable. Performing
this criterion over the final top 10 list from phase I will help eliminate sites where data
are abnormally fluctuated and excessive variability in crashes and violations are found.
This can be calculated using the coefficient of variation (V) formula:
s
V =-
x 100
Equation 8 collision cofefficieient of variation (de Leur & Milner, 2011)
Where:
S = The standard deviation.
= Sample mean (Annual collision).
As known, the sample mean ( ) can be calculated by using the following formula:
IX
X =71
Equation 9 Sample mean
Where:
x = Sample data (Annual collision),
n = Sample size (Number of years of data).
And, the standard deviation (s) can be calculated using the following formula:
74


2
S = n-1
Equation 10 Fluctuation of crashes by calculating the standard mean
2- Type of Vehicles
From the final top 10 candidate RLC sites, type of vehicles that have been
atypically involved in crashes or violations can be screened to consider during the
installation of the RLC.
RLC benefits might be limited at these locations because they cannot photograph
license plates of tractor unit of multiunit vehicles or are not capable of photographing
license plates of very large trucks. As a result, cameras with features that are capable of
photographing multiunit vehicles are recommended at these locations.
Type of vehicles can be screened and analyzed using the chi-square test )
shown in the following formula:
Equation 11 Type of vehciles by calculating Chi-square test, (de Leur & Milner, 2011)
Where
/j = Frequency of type i vehicles at a site.
fgi = Frequency of type i vehicles at a site (f si = f X p J where is the pr0portion of a
vehicle type in reference population and/is the total vehicle types at a site
75


3- Economic Evaluation
As discussed in the literature review chapter, a big chunk of the argument related
to the RLC systems is based on the opinion that it is more about increasing cities
revenue rather than improving the safety level. Therefore, considering this element while
selecting RLC locations is a plus toward the success of the program in any community.
An economic evaluation of each of the final top 10 RLC candidates will evaluate
these locations after installation in an annual basis. Cost of RLC, revenue generated from
the system, and safety benefits are the three elements involved to execute this criterion.
Total Cost of RLC per year + Total Revenue of RLC per year < Total Safety Benefits per year
Equation 12 RLC Economic evaluation. (ALTurki, 2013)
Where:
Total Cost of RLC = Overall cost of RLC installation, operation, maintenance in a given location
/ Year. (This can be obtained from RLC providers).
Total Revenue of RLC = Overall revenue generated by RLC in a given location / Year. (Total of
RLC tickets value)
Ave Safety Benefits = Average cost of a crash producing PDO and injuries. (Determined by
major auto insurance companies)
According to the city of Fort Collins around 80-100 tickets/month are generated by RLC
which means a revenue of $120,000/year. If quick assumptions are made, a RLC costs
$40,000 which means $480,000/year, that includes the process of installation, operation,
and maintenance. This will total up to $600,000 (if we assume city and operator are
lobbying together as claimed by parties standing against the system).
76


A study published by the National safety council estimated motor-vehicle crashes as
(National Safety Council, 2000):
Death = $1M
Injuries= $35,500
PDO= $6,500
The estimates were based on wage and productivity losses, medical expenses,
administrative expenses, motor vehicle damaged, and employer costs. If estimates are
used as average safety estimate, then there is no doubt that average safety benefits will
exceed total cost and revenue of RLC, because if RLC prevents 2 crashes resulting in
injuries/ month = $71,000 (which is more than $850,000 per year. Injuries level of
severity alone will exceed RLC cost and revenue combined.
By the end of each year, locations where safety benefits exceed total RLC cost
and revenue are proofing their economic effectiveness and should remain under operation.
In contrast, locations where total RLC cost and revenue are more than its safety benefits
are not considered economically effective and should be eliminated.
4- Intersection Characteristics
During sites selection criterion Phase II, more variables are used to eliminate
these qualified intersections even further. One of the steps is to visit the sites and
evaluate their characteristics and suitability for RLC based on (PASS and FAIL) scoring
system. Locations with high PASS score indicate a need for further action such as a RLC
system, while locations with low PASS score indicate a need to fix these characteristics
before installing a RLC.
77


There are four characteristics that should be included as part of the field
evaluation, these are:
Intersection Layout: It includes four guidelines:
Lane width: Lane widths are commonly narrower on low volume roads and wider
on higher volume roads. According to ITE, 12- foot lanes are desirable, although
widths as narrow as 10 feet have been used in severely constrained situations
unless large trucks and buses are using the lane. Therefore, the range from 10-12
will be sat as a parameter for all final top 10 sites.
Lighting: Statistics indicate that the non-daylight accident rate is higher than that
during daylight hours. This fact, to a large degree, may be attributed to impaired
visibility. In urban and suburban areas where there are concentrations of
pedestrians and roadside and intersectional interferences, fixed-source lighting
tends to reduce crashes (American Association of State Highway and
Transportation Officials (AASHTO), 2001).
Clear Signage: Roadway signs in the United States increasingly use symbols
rather than words to convey their message. Symbols provide instant
communication with roadway users, overcome language barriers, and are
becoming standard for traffic control devices throughout the world. Familiarity
with symbols on traffic signs is important for every road user in order to maintain
the safety and efficiency of our transportation facilities. Proper and clear signs
associated with the nature of the intersection design is a must for each of the final
top 10 sites to pass the field investigation.
78


Channelization: One of the most effective and efficient methods of controlling
the traffic on a highway is the adoption of high intersection geometric design
standards. Channelization is an integral part of at grade intersections and is used
to separate turning movements from through movements where this is considered
advisable and hence helps reduce the intensity and frequency of loss of life and
property due to accidents to a large extent. Proper Channelization increases
capacity, improves safety, provides maximum convenience, and instills driver
confidence. Improper Channelization has the opposite effect and may be worse
than none at all. Over Channelization should be avoided because it could create
confusion and worsen operations. Channelization is defined as the separation or
regulation of conflicting traffic movements into delineated paths of travel by
traffic islands or pavement marking to facilitate the safe and orderly movements
of vehicles, bicycles, and pedestrians.
All guidelines should meet CDOT requirements for signalized intersections (
Colorado Department of Transportation, 2000).
Approach Speed: Speed limits are set by each state or territory. Speed limits are
always posted in increments of five miles per hour. Some states have lower limits
for trucks and at night, and occasionally there are minimum speed limits. Most
speed limits are set by state or local statute, although each state allows various
agencies to set a different, generally lower, limit. However in this study, the pass
parameter is that to have the average speed limit less or equal to the posted speed
limit in order to pass the field investigation.
79


Yellow Phase Change Interval: As discussed earlier, excessively short or long
yellow change intervals may encourage driver disrespect and unsafe operating
practices. Therefore, it is important to confirm that all top 10 RLC candidates are
within the suggested ITE yellow interval values (See table 14).
Table 14 Pre-calculated yellow intervals at various speeds.
Posted Speed Limit (mph) Minimum Yellow Vehicle Change Interval (sec)
15 3
20 3
25 3
30 3.2
35 3.6
40 3.9
45 4.3
50 4.7
55 5
60 5.4
65 5.8
Social Structure: Another worthwhile evaluation characteristic here is to make
sure the area where RLC will be installed is suitable and has no excessive
vandalism that would target the camera. This can be measured by reviewing
criminal history and income level of the area, which is normally provided by the
city police department.
80


Figure 20 A vandalized RLC in Phoenix Arizona. (Garrett, 2011)
Table 15 Sample of field evaluation table used to evaluate intersection characteristics. (ALTurki, 2013)
Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7)
Intersection Name Intersection Layout Lane Width
Lightening
Channelization
Signage
Yellow Change Interval Meet ITE guidelines
Approach Speed Ave Speed < Posted Speed
Social Structure Criminal history and Income level
81


Full Text

PAGE 1

DETERMINING CRITERIA FOR SELECTING RED LIGHT CAMERA LOCATIONS by MANSOUR ABDULHAMID ALTURKI MEng, University of Colorado Denver, 2008 MBA, University of Colorado Denver, 2008 BS, King Saud University, 2005 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Civil Engineering Program 2014

PAGE 2

This thesis for the Doctor of Philosophy degree by Mansour ALTurki has been approved for the Civil Engineering Program by Bruce Janson, Chair Wesley Marshall, Advisor Juan Robles Gary Kochenberger Bob Kois May 2, 2014

PAGE 3

Mansour AbdulHamid ALTurki (Ph.D., Civil Engineerin g) Determining Criteria for Selecting Red Light Camera Location Thesis directed by Professor Bruce Janson ABSTRACT The objective of this dissertation is to develop a systematic method and criteria for selecting effective (i.e., severe crash reducing) r ed light camera locations among all signalized intersections of a given jurisdiction. A nother objective is to develop criteria that can be implemented using accessible data while maintaining the comprehensiveness feature of the criteria. Selecting locations for re d light cameras has received less attention by researchers of transportation engineering than a ssessing their effectiveness in reducing crashes. However, better site selection rules can r esult in greater effectiveness, which is the main goal of installing red light cameras. The methodology was divided into two phases that is mostly based on statistical criteria but with more field investigations in the second phase. The first phase includes five criteri a, which are, (i) crash severity level, (ii) normalized crash severity level, (iii) potential fo r improvement in terms of crash rate, (iv) potential for improvement in terms of crash frequen cy, and (v) crash types. The second phase includes six other criteria, which are, (i) f luctuation of crashes, (ii) vehicle types, (iii) economic evaluation, (iv) intersection charac teristics, (v) approach determination, and (vi) red light locations. The study applies its methodology to three major cities in Colorado; these are Colorado Springs, Fort Collins, and Denver. The study found red light camera candidate intersections that are very consistent with the city engineersÂ’ opinions of potentially effective locations and the history of crash data from Denver since 2003. The form and content of this abstract are approved I recommend its publication. Approved: Bruce Janson

PAGE 4

DEDICATION This dissertation is lovingly dedicated to my mothe r, Mrs. Eman ALTurki, for her encouragement, and constant love that have sustaine d me throughout my life, Without her, I wonÂ’t be at this level of education. To my father Mr. AbdulHamid ALTurki who has been my silent inspiration and my support when hard times come around.

PAGE 5

ACKNOWLEDGEMENTS I am most grateful to the members of my committee, Mr. Bob Kois, Prof. Gary Kochenberger, and Mr. Juan Robles for their time, e ncouragement, and expertise throughout this project. Special thanks to Prof. Br uce Janson (the Chairman of the Committee) and Prof. Wesley Marshall (my advisor), for their exquisite attention to detail, patience and for their continuous demand for excell ence. Prof. Bruce and Prof. Wes have been more than advisors to me. There are people in everyoneÂ’s lives who make succe ss both possible and rewarding. My wife, Ahoud ALSharaia, my children, Eman ALTurki, a nd Nawaf ALTurki steadfastly supported and encouraged me. Dr. Saleh ALSoghair, Eng, Dino Bakkar, and Eng. And y Richter I will never forget the support and encouragement you provided to me by fac ilitating many obstacles that came on my way to this accomplishment. My friend Eng. Ziyad ALBathi helped, cajoled, and p rodded me when I needed it the most. For my uncle Abdullah ALTurki, my father in law Mr. Ahmad ALSharaia, and my neighbors Mr. Tim Garduno and Mrs. Wendy Garduno fo r their support and effort that they made sure to give to me in many occasions. I also like to give special thanks to Anderson Acad emic Commons Library at the University of Denver for providing me with the all the resources I needed during my research time.

PAGE 6

Without the support of my siblings Malath Malak, Maram, Nourh, Hamad, and Abdullah, pursuit of this advanced degree would nev er have been started. Thank you, ALL, now and always.

PAGE 7

TABLE OF CONTENTS CHAPTER I. PROBLEM STATEMENT............................... ................................................... ............ 1 Introduction ...................................... ................................................... ............................ 1 Statement of Problem .............................. ................................................... ..................... 2 Main Questions .................................... ................................................... ........................ 3 Study Objectives .................................. ................................................... ........................ 3 Hypotheses and Contribution to the Transportation E ngineering Industry and Public Safety ............................................ ................................................... ............................... 4 Limitations to the Study .......................... ................................................... ..................... 5 II. BACKGROUND .................................... ................................................... ..................... 6 Traffic Safety Overview ........................... ................................................... ................... 6 History of Red Light Camera Systems ............... ................................................... ......... 8 Glossary of Terms ................................. ................................................... ..................... 11 Vehicle Detection and Surveillance Technologies ... ................................................... 12 Implications for Public Privacy ................... ................................................... .............. 18 Impact on Revenue ................................. ................................................... ................... 19 Study Timeline .................................... ................................................... ....................... 21 Dissertation Structure............................. ................................................... .................... 22 III. LITERATURE REVIEW ............................ ................................................... ............ 24 Introduction ...................................... ................................................... .......................... 24 Effectiveness of RLC on Safety .................... ................................................... ............. 24

PAGE 8

Effectiveness of RLC on Type of Crashes ........... ................................................... ...... 33 Effectiveness of RLC on Crashes Severity .......... ................................................... ...... 38 Characteristics of Red Light Runners .............. ................................................... .......... 43 RLC and Signal Timings ............................ ................................................... ............... 49 Methodologies and Procedures Used for RLC Analysis .............................................. 50 RLC Spillover Effect (Halo Effect) ................ ................................................... ........... 53 RLC Site Selections ............................... ................................................... .................... 57 IV. METHODOLOGY ................................... ................................................... ............... 62 Introduction ...................................... ................................................... .......................... 62 Why These Locations as Case Studies? .............. ................................................... ....... 62 Data Required and Field Investigation ............. ................................................... ......... 63 Methodology ....................................... ................................................... ....................... 66 Phase I “Includes Four Criteria” .................. ................................................... .............. 66 Phase II “Includes Seven Criteria” ................ ................................................... ............ 74 Expected Findings ................................. ................................................... ..................... 87 V. ANALYSES AND FINDINGS .......................... ................................................... ...... 88 Section I: Analyses of RLC Sites Selection for Colo rado Springs ............................... 88 Section II: Further Analysis and Field Investigatio n of Top 10 RLC Candidates in Colorado Springs .................................. ................................................... ..................... 95 Section I: Analyses of RLC Sites Selection for Fort Collins. ..................................... 106 Section I: Analyses of RLC Sites Selection for Denv er ............................................. 12 2

PAGE 9

Section II: Further Analysis and Field Investigatio n of Top 10 RLC Candidates in Denver ............................................ ................................................... .......................... 129 Recommednations and ConclusionsÂ…Â…Â…Â…Â…Â…Â…Â…Â…..Â…Â…Â…Â…Â…Â…Â…Â…Â… ..144 WORKS CITED ....................................... ................................................... ................... 151 APPENDIX .......................................... ................................................... ........................ 157

PAGE 10

LIST OF FIGURES FIGURE 1. Number and rank of motor vehicles traffic fatali ties as a cause of death in the United States. 1981-2009 (Subramanian, 2009) ............. ................................................... ..... 7 2. The Hague traffic police put a sort of monocular (Gatsometer, 2010) .......................... 9 3. Older RLC in Ludwigsburg, Germany. (Lowe, 2006) .................................................. 10 4. Distribution of the loopÂ’s electromagnetic field (Hockaday, 1991) ............................ 13 5. Loop location at the intersection. (Kell, 1990) ................................................... .......... 14 6. Loop sensors reflect damages to the asphalt. (Ke ll, 1990) ......................................... .. 15 7. Speed limit cameras can take shapes of normal ro ad elements. (Klein, MillimeterWave and Infrared Multisensor Design and Signal Pro cessing, 1997) .................... 16 8. Intrusive sensors and camera requires less effor t and no damages. (Klein, Final Report: Mobile Surveillance and Wireless Communication Syst ems Field Operational Test Vol. 2: FOT Objectives, Organization, System Design Results, Conclusions, and Recommendations, 1999) ............................ ................................................... .......... 17 9. The proportion of crashes occurring at monitored approaches vs. non-monitored approaches. (Dahnke, Stevenson, Stein, & Lomax, 200 8) ....................................... 27 10. Percentage of crash type in Scottsdale for 14-y ear period. (Shin & Washington, 2007) ................................................... ................................................... ............................. 37 11. Percentage of crashes per year by crash type an d severity (PDO vs. injury and fatal). (Shin & Washington, 2007) ......................... ................................................... .......... 41 12. Percentage of crashes per year by crash type an d severity (minor vs. major). (Shin & Washington, 2007) ................................. ................................................... ................ 42

PAGE 11

13. Normalized red light violation values by age gr oup (Yang & Najm, 2006) ............... 47 14. Distributions of red light violation records by vehicle speed (Yang & Najm, 2006) 47 15. Distribution of red light violation by time of day (Yang & Najm, 2006) ................... 48 16. A photo taken from a camera for an accident inv olving RLR (Administration, 2005) ................................................... ................................................... ............................. 51 17. Intersections studied in Arlington Virginia (Mc Cartt & Hu, 2013)............................ 55 18. Illustration of the term potential for improvem ent. .............................................. ...... 69 19. Calculation of crash type rate (ALTurki, 2013) ................................................... ....... 72 20. A vandalized RLC in Phoenix Arizona. (Garrett, 2011) ............................................ 81 21. Colorado Springs reported crashes in relation t o annual average daily traffic ........... 91 22. Briargate Py & N Powers Bl (Google Maps)...... ................................................... .... 98 23. Airport Rd & S Academy Bl (Google Maps) ....... ................................................... ... 99 24. E Woodmen Rd/I-25 (Google Maps) ............... ................................................... ........ 99 25. E Platte Av & N Academy Bl (Google Maps) ...... ................................................... 100 26. Barnes Rd & N Powers Blvd ..................... ................................................... ............ 100 27. E Platte Ave & N Union Blvd.................... ................................................... ............ 101 28. N Academy Blvd & Vickers Dr (Google Maps) ..... ................................................. 1 01 29. N Powers Blvd & Stetson Hills Blvd ............ ................................................... ......... 102 30. Maizeland Rd & N Academy Blvd ................ ................................................... ....... 102 31. Dublin Blvd & N Union Blvd. (Google Maps)...... ................................................... 103 32. Final RLC locations (Colorado Springs) ........ ................................................... ....... 105 33. Fort Collins reported crashes in relation to an nual average daily traffic .................. 109 34. College Ave & Monroe. (Google Maps) ........... ................................................... .... 114

PAGE 12

35. Timberline Rd & Horsetooth Rd. (Google Maps) .. .................................................. 115 36. Lemay & Harmony. (Google Maps) ................ ................................................... ...... 115 37. College Ave & Tribly Rd. (Google Maps) ........ ................................................... .... 116 38. College Ave & Horsetooth Rd. (Google Maps)..... ................................................... 116 39. S Shields St & W Plum St. (Google Maps) ....... ................................................... .... 117 40. Timberline Rd & Drake Rd. (Google Maps) ....... ................................................... .. 117 41. Shields St & Mulberry St. (Google Maps)........ ................................................... ..... 118 42. Shields St & Elizabeth St. (Google Maps)....... ................................................... ...... 118 43. Ziegler Rd & Rock Creek Dr. (Google Maps) ..... ................................................... .. 119 44. Final RLC locations (Fort Collins) ............ ................................................... ............ 121 45. Denver's reported crashes in relation to annual average daily traffic. ...................... 125 46. E Alameda Ave & Leetsdale Dr. (Google Maps) ... .................................................. 132 47. W Colfax Ave & N Kalamath St. (Google Maps) ... ................................................. 1 33 48. Leetsdale Dr & Quebec St. (Google Maps) ....... ................................................... .... 133 49. S Monaco St & Leetsdale Dr. (Google Maps) ..... ................................................... .. 134 50. E 6th Ave & N Lincoln Blvd. (Google Maps) .............. ............................................. 134 51. W Mississippi Ave & S Platte River Dr. (Google Maps) ......................................... 135 52. N Colorado Blvd & E Colfax Ave. (Google Maps) ................................................ 13 5 53. S Federal Blvd & W Alameda Ave. (Google Maps) ............................................... 136 54. E Alameda Ave & S Monoco St (Google Maps) ..... ................................................. 1 36 55. S University Blvd & E Evans Ave. (Google Maps) ................................................. 1 37 56. Final RLC locations (Denver) .................. ................................................... .............. 139

PAGE 13

57. Trend of total crashes before and after the yea r of RLC installation at four signalized intersections in Denver ........................... ................................................... ............. 140 58. Trend of front to side type of crashes before a nd after the year of RLC installation at four signalized intersections in Denver............ ................................................... .... 141 59. Trend of rear end type of crashes before and af ter the year of RLC installation at four signalized intersections in Denver ................ ................................................... ....... 142 60. City of Denver warns drivers to drive safely as they approach the intersection of S University Blvd & E Evans Ave. .................... ................................................... ..... 146

PAGE 14

LIST OF TABLES TABLE 1. RLC effectiveness on safety at Fairfax County, V irginia (Hobeika & Yaungyai, 2006) ................................................... ................................................... ............................. 29 2. Summary of the recent studies of RLC effectivene ss on safety ................................... 32 3. Results of one-year before/after study Sacrament o California (McGee & Eccles, 2006) ................................................... ................................................... ............................. 34 4. Before and after changes in crashes, Sydney, Aus tralia (Hillier, Ronczka, & Schnerring, 1993) ............................................. ................................................... ......................... 35 5. Results for individual jurisdictions for total c rashes (Administration, 2005) ............... 36 6. The distribution of crashes by severity for all signalized intersections 1997 (McGee & Eccles, 2006) ..................................... ................................................... ..................... 38 7. Percent of last drivers running a red light by d emographic category. (Martinez & Porter, 2006) ............................................. ................................................... ......................... 44 8. Unit crash cost estimates by severity level used in the economic effects analysis. (Federal Highway Administration, 2005) ............ ................................................... .. 52 9. Observed red light violation rates per 10,000 ve hicles by time into red signal phase and percentage changes 1 month and 1 year after red lig ht camera ticketing began, compared with warning period. (McCartt & Hu, 2013) ........................................... 56 10. Data required for RLC sites selection Criterion .................................................. ....... 65 11. Weighting percentages for criterions in Phase I (Colorado Springs) ........................ 73 12. Weighting percentages for criterions in Phase I (Fort Collins) ................................. 73 13. Weighting percentages for criterions in Phase I (Denver) ........................................ 73

PAGE 15

14. Pre-calculated yellow intervals at various spee ds. ............................................... ....... 80 15. Sample of field evaluation table used to evalua te intersection characteristics. .......... 81 16. Table used for determining numbers of “at-fault vehicles” in each approach ............ 82 17. Formulas used to obtain final findings......... ................................................... ............ 83 18. Ranking of top 10 RLC candidates in Colorado Sp rings based on normalized crash severity level. ................................... ................................................... ...................... 89 19 Ranking of top 10 RLC candidates in Colorado Spr ings based on crash severity. ..... 89 20. Ranking of top 10 RLC candidates in Colorado Sp rings based on potential for improvement in relation to crash rate. ............ ................................................... ....... 90 21. Ranking of top 10 RLC candidates in Colorado Sp rings based on potential for improvement in relation to crash Frequency. ....... ................................................... 90 22. Ranking of top 10 RLC candidates in Colorado Sp rings based on crash type. .......... 91 23. Final top 10 RLC candidates in Colorado Springs for all criteria in phase I. ............. 93 24. Intersections field evaluation of Colorado Spri ngs top 10 RLC candidates. .............. 96 25. Number of at fault vehicles per approach (Color ado Springs) ................................. 104 26. Ranking of top 10 RLC candidates in Fort Collin s based on normalized crash severity level. ............................................ ................................................... ......................... 107 27. Ranking of top 10 RLC candidates in Fort Collin s based on crash severity level.... 107 28. Ranking of top 10 RLC candidates in Fort Collin s based on potential for improvement in relation to crash rate. ............ ................................................... ..... 108 29. Ranking of top 10 RLC candidates in Fort Collin s based on potential for improvement in relation to crash Frequency. ....... .................................................. 108 30. Ranking of top 10 RLC candidates in Fort Collin s based on crash type. ................. 109

PAGE 16

31. Final top 10 RLC candidates in Fort Collins for all criteria in phase I. .................... 110 32. Intersection evaluation table (Fort Collins) ................................................... ......... 112 33. Number of at fault vehicles per approach (Fort Collins) .......................................... 120 34. Ranking of top 10 RLC candidates in Denver base d on normalized crash severity level ............................................. ................................................... ......................... 123 35. Ranking of top 10 RLC candidates in Denver base d on normalized crash severity level ............................................. ................................................... ......................... 123 36. Ranking of top 10 RLC candidates in Fort Collin s based on potential for improvement in relation to crash rate. ............ ................................................... ..... 124 37. Ranking of top 10 RLC candidates in Fort Collin s based on potential for improvement in relation to crash rate. ............ ................................................... ..... 124 38. Ranking of top 10 RLC candidates in Denver base d on crash type.......................... 125 39. Final top 10 RLC candidates in Denver for all c riteria in phase I. ........................... 127 40. Intersection evaluation table (Denver) ........ ................................................... ........... 130 41. Number of at fault vehicles per approach (Denve r) ................................................ .. 138 42. Total crashes by year in current RLC locations in Denver. ...................................... 1 40 43. Front to side type of crashes by year in curren t RLC locations in Denver. .............. 141 44. Rear end type of crashes by year in current RLC locations in Denver. .................... 141 45. Analysis of Colorado Springs intersections base d on crash severity level and normalized crash severity level. .................. ................................................... ......... 157 46. Colorado Springs intersections ranked based on normalized crash severity level. .. 161 47. Colorado Springs intersections ranked based on crash severity level. ..................... 164

PAGE 17

48. Analysis of Colorado Springs Intersections base d on potential for improvement in relation to crash rate and crash frequency......... ................................................... ... 167 49. Colorado Springs intersections ranked based on potential for improvement in relation to crash rate. .................................... ................................................... ..................... 171 50. Colorado Springs intersections ranked based on potential for improvement in relation to crash frequency. ............................... ................................................... ................ 174 51. Analysis of Colorado Springs intersections base d on crash types ............................ 177 52. Colorado Springs intersections ranked based on front to side rate. .......................... 180 53. Analysis of Fort Collins intersections based on crash severity level and normalized crash severity level. ............................. ................................................... ................. 183 54. Fort Collins intersections ranked based on norm alized crash severity level. ........... 187 55. Fort Collins intersections ranked based on cras h severity level. .............................. 19 0 56. Analysis for potential for improvement for all intersections of Fort Collins in relation to crash rate and frequency. ...................... ................................................... ........... 193 57. Fort Collins intersections ranked based on pote ntial for improvement in relation to crash rate. ....................................... ................................................... ...................... 198 58. Fort Collins intersections ranked based on pote ntial for improvement in relation to crash frequency ................................... ................................................... ................. 201 59. Analysis of Fort Collins intersections based on crash types ..................................... 204 60. Fort Collins intersections ranked based on fron t to side crashes. ............................. 20 8 61. Analysis of Denver intersections based on crash severity level and normalized crash severity level. ................................... ................................................... .................... 211 62. Denver intersections ranked based on normalized crash severity level. ................... 222

PAGE 18

63. Denver intersections ranked based on crash seve rity level. ...................................... 230 64. Analysis of potential for improvement for Denve r intersections based on crash rate and frequency. .................................... ................................................... .................. 238 65. Denver intersections ranked based on potential for improvement in relation to crash rate............................................... ................................................... ......................... 250 66. Denver intersections ranked based on potential for improvement in relation to crash frequency.......................................... ................................................... .................... 258 67. Analysis for Denver intersections based on cras h types. .......................................... 266 68. Denver intersections ranked based on front to s ide crashes. .................................... 2 77

PAGE 19

LIST OF EQUATIONS EQUATION 1. Normalizedcrash serverity level .............. ................................................... ................ 67 2. Crash severity level ........................... ................................................... ......................... 67 3. PFI in crash rate .............................. ................................................... ........................... 69 4. Annual crash rate .............................. ................................................... ......................... 70 5. Average crash rate.............................. ................................................... ........................ 70 6. Crash frequency ................................ ................................................... ......................... 71 7. Proportionality to obtain relative weights ..... ................................................... ............. 72 8. collision cofefficieient of variation .......... ................................................... .................. 74 9. Sample mean .................................... ................................................... .......................... 74 10. Fluctuation of crashes by calculating the stand ard mean .......................................... .. 75 11. Type of vehciles by calculating Chi-square test ................................................. ....... 75 12. RLC Economic evaluation. ...................... ................................................... ................ 76

PAGE 20

1 CHAPTER I PROBLEM STATEMENT Introduction The condition of being protected against physical, economic, emotional, educational, political, occupational, or any other aspects that could be damaged or harmed is the definition of safety. (Federal Highwa y Administration, 2012) Recently, public safety, as one of the major safety categorie s, has received more attention due to the fact that it is directly related to humansÂ’ lives a nd health, which is considered as a significant indication of better developments and c ommunities. Keeping in mind all the developments and advancemen ts associated with todayÂ’s technologies and environmental regulations, public safety has become even more challenging to achieve. Promoting public safety in systems like medical and health safety, building safety, and so on is very important, but i t is even more important when it relates to the transportation system. The transportation system requires the highest leve l of safety due to the number of users involved in the system every day, as well as the nature of risks people can suffer as a result of the system being unsafe. One of the mos t dangerous and risky traffic related violations is red light running, which is a behavio r that can cause some of the most serious injuries and fatalities the transportation system may generate. In the United States and during the year of 2010 al one, almost 50 percent of all crashes reported to the police occurred at intersec tions. In fact and according to the Insurance Institute for Highway Safety, signalized intersections accounted for more than

PAGE 21

2 68,000 serious non-fatal injuries and 7707 deaths i n 2010 alone. (Insurance Institute for Highway Safety, 2012). As a result, many transportation agencies, organiza tions, departments and communities across the nation like the Federal High way Administration (FHWA), the National Highway Traffic Safety Administration (NHT SA), seek to address crashes and reduce both injuries and fatalities by increasingly looking for tools to supplement traditional enforcement resources. One of the safet y tools that over 550 US communities have employed is a red light camera (RLC). (Nationa l Safety Council, 2009) The first chapter of this study starts by explainin g the statement of problem the study addresses in addition to representing the mai n questions, and research objectives. This chapter will also demonstrate how this study c ould contribute to the civil engineering science in general and more specificall y to the transportation engineering field despite the limitations that are usually asso ciated with similar studies. Statement of Problem Many of the post-implementation evaluations that we re conducted to measure the effectiveness of red light cameras (RLC) have shown an overall effectiveness in reducing the frequency of crashes at intersections where red light cameras are operated, although there are exceptions in some cases. As it will be illustrated in the literature review, most studies that researched the effectiveness of RLC on safety were mostly making b efore/after crash comparisons. Other studies investigated more details regarding t he types and severity of crashes associated with RLC. In comparison, fewer studies discussed other important areas of

PAGE 22

3 research that could show significant indication of RLC effectiveness on safety such as the RLC sites selection. This study will examine one of the critical element s that is usually associated with the installation or expansion of RLC systems, which is the selection of RLC sites that have the greatest potential to improve safety. The study will also demonstrate its practically by applying the methodology to three ma jor cities of Colorado; which are Colorado Springs, Fort Collins, and Denver. Main Questions This study will try to answer the following questio ns in order to achieve the study goals: 1) In Colorado Springs the City discontinued the RLC program after one y ear of installation (2010) due to unsuccessful results. I f we go back to 2010, what kind of criteria could be used to make selection of spec ific intersections within the city limit and therefore could possibly make the RLC pro gram more effective and show successful results? 2) In Denver and Fort Collins, the costly system has been under operation for at least 10 years. Are these cities making the best ch oices when selecting the locations of their RLC systems? Can that be suppor ted in a scientific way? Study Objectives This study aims to provide a RLC site selection met hodology based on analytical procedures that require accessible data that will b e obtainable by any community to select RLC sites with greatest potential so the selection becomes more systematic. This study intends to use some statistical models that will be presented in more detail as part of the methodology chapter. Additionally, this study aims to form a more obvious picture of the

PAGE 23

4 effectiveness of RLC programs on reducing red light running crashes and their n nr when comparing the current RLC sites to the candid ate sites concluded in the analysis chapter. Hypotheses and Contribution to the Transportation E ngineering Industry and Public Safety The following are two primary motivations and poten tial benefits of this study: 1) An analytical-based site selection methodology incr eases the effectiveness of red light camera programs. 2) Comprehensive and scientific RLC criteria can posit ively impact public opinion about RLC system. This study aims to contribute to the transportation industry from different points of view. The following bullets describe these contr ibutions: 1) This study will provide transportation agencies, pl anners, engineers, and researchers with statistical figures and findings r elated to one of the least researched areas, which is RLC sites selection (acc ording to the literature review). 2) This study will contribute to the field of civil en gineering and transportation by reviewing the RLC experiences in Denver and Fort Co llins. 3) This study provides an analytical-based methodology for RLC site selection that can be used by any city in implementing a RLC progr am to potentially improve public safety. 4) The study will apply its methodology to three major cities of Colorado; which are Colorado Springs, Fort Collins, and Denver.

PAGE 24

5 5) It is also important to note that the analytical-ba sed methodology mentioned above is formed based on the sort of data that most of the cities around the world have access to, which makes it a more usable method ology. Limitations to the Study Accurately assessing candidates with potential to b e equipped with RLC is challenging for several reasons: 1) Many safety related factors are uncontrolled and/or confounded during the periods of observation. 2) Availability and accuracy of data may not be access ible at the needed level. 3) The variety and number of agencies involved in such programs can make it more challenging to find accurate and consistent da ta.

PAGE 25

6 CHAPTER II BACKGROUND Traffic Safety Overview Road traffic safety means reducing accident causes on the road through improved vehicles, facilities, and driving practices. Road a nd vehicle design, driver impairment, speed of operation, and other factors like proper s ignal timing, better signal design, improved intersection design, and many more are all considered factors that could decrease or increase the level of safety on the roa d. (Road Safety, 2010) According to the World Health Organization (WHO), m ore than a million people are killed on the worldÂ’s roads each year. A report published by the WHO in 2004 estimated that 1.2 million people were killed and 5 0 million injured in traffic crashes around the world each year and that traffic crashes are the leading cause of death among children 10-19 years of age. The report also noted that the problem was most severe in developing countries and that simple prevention mea sures could halve the number of deaths. (World Health Organization, 2010) Because of these facts, road traffic crashes are on e of the worldÂ’s largest public health and injury prevention problems. The problem is more acute because victims are overwhelmingly healthy prior to their crashes. In 2009, motor vehicle traffic crashes were among t he top 10 causes of death in the United States for the first time since 1981. In 2008, vehicle traffic crashes were 11th. (See Figure1) In 2009, when ranked by specific ages, motor vehicl e traffic crashes were the leading cause of death for age 4 and every age 11 t hrough 27, while motor vehicle traffic

PAGE 26

7 crashes were the leading cause of death for each ag e 13 through 30 the year before. (Subramanian, 2009) nrrr rrrrrrr !rrr" # Note “ The coding of mortality data changed significantly in 1999, so comparisons of the number of deaths and death rates from 1998 and before with data from 1999 and after may not be advisable” (Subramanian, 2009) In the United States, three acts were announced to seek better and safer transportation systems, starting with the Intermoda l Surface Transportation Efficiency Act, which was signed by President Bush back in 199 1. The act provides funding to continue the provisions of the National Traffic and Motor Vehicle Safety Act of 1966, and the Motor Veh icle Information and Cost Savings Act. The act includes a number of motor vehicle saf ety rulemaking requirements and additional directions, including rollover protectio n for occupants of passenger cars, multipurpose passenger vehicles, and light trucks, side impact protection for occupants of multipurpose passenger vehicles, improved head impa ct protection (from interior components) for occupants of passenger cars, and ai rbag crash protection systems for

PAGE 27

8 drivers and right front passengers in new passenger cars, new light trucks (including light buses), and multipurpose passenger vehicles. On June 9, 1998, the president signed the Transport ation Equity Act for the 21st Century (TEA-21). This act paid major attention to safety, strengthening the safety programs across the US Department of Transportation that aim to save road usersÂ’ lives and property. (The U.S. Department of Transportatio n, 1998) The third act, The Safe, Accountable, Flexible, and Efficient Transportation Equity Act (SAFETEA) was announced formally in 2005 The act provides comprehensive attention to the safety associated wi th the transportation system. The act establishes a new core Highway Safety Improvement P rogram that aims to make significant progress in reducing fatalities that ta ke place on the highways. It concentrates on several areas of concern in the system like work zones, children walking to school, and older drivers. It doubled the funds to improve the infrastructure and implement strategic highway safety planning to ensure accommo dation of the safety requirements. (The National Tranportation Library, 1991) History of Red Light Camera Systems Historically, traffic enforcement cameras can be da ted back to 1905 where the popular machines were used to record motoristsÂ’ spe eds by taking time-stamped images of vehicles moving across the start and end point o f the road. By using the popular machine system, authorities were able to calculate the vehicle speed and identify the driver by referring to the time-stamps and images r espectively. Gatsometer BV was a company founded back in 1958 by rally driver Maurice Gatsonides. It produced a monitor device to track t he average speed in order to improve

PAGE 28

9 his lap times. Later, the company started supplying police radars, red light cameras, and mobile speed traffic cameras. (Gatsometer, 2010) : The Hague traffic police put a sort of monocular. (Gatsometer, 2010) Worldwide, red light cameras have been in use since the 1960s, and were used for traffic enforcement in Israel as early as 1969. The first red light camera system was introduced in 1965, using tubes stretched across th e road to detect the violation and subsequently trigger the camera. Red light cameras were first developed in the Netherlands. One of the first developers of these r ed light camera systems was Gatsometer BV. (Gatsometer, 2010) The cameras first received serious attention in the United States in the 1980s following a highly publicized crash in 1982 involvi ng a red-light runner who collided with an 18-month-old girl in a stroller (or "push-c hair") in New York City. Subsequently, a community group worked with the city's Department of Transportation to research

PAGE 29

10 automated law-enforcement systems to identify and t icket drivers who run red lights. New York's red-light camera program went into effec t in 1993. From the 1980s onward, red light camera usage expanded worldwide, and one of the early camera system developers, Poltech International, supplied Austral ia, Britain, South Africa, Taiwan, the Netherlands and Hong Kong. American Traffic Systems (subsequently American Traffic Solutions) (ATS) and Redflex Traffic Systems emerge d as the primary suppliers of red light camera systems in the US, while Jenoptik beca me the leading provider of red light cameras worldwide. (Lowe, 2006) Initially, all red light camera systems used film, which was delivered to local law enforcement departments for review and approval. Th e first digital camera system was introduced in Canberra, Australia in December 2000, and digital cameras have increasingly replaced the older film cameras in oth er locations since then. $ : Older RLC in Ludwigsburg, Germany. (Lowe, 2006)

PAGE 30

11 Glossary of Terms Traffic Enforcement Camera (TEC): An automated ticketing machine that could be mounted beside or over the road to observe traffic violators. (Wilson C, 2010) Red Light Camera (RLC): is a traffic enforcement camera that captures an im age of a vehicle which has entered an intersection against a red traffic light. By automatically photographing vehicles that run red lights, the cam era produces evidence that assists authorities in their enforcement of traffic laws. ( Insurance Institute for Highway Safety, 2010) Red Light Runner (RLR): The simplest definition of red-light running (RLR) is the act of entering, and proceeding through, a signalized i ntersection after the traffic signal has turned red. (National Committee on Uniform Traffic Laws and Ordinances., 2000) Infraction : In 1981, the legislature of the US decriminalized many minor traffic offenses to promote public safety and to facilitate the impl ementation of a uniform and expeditious system for the disposition of such offe nses. Common traffic infractions are speeding as well as seat belt and liability insurance violations. These offenses are called inf ractions and are considered civil cases. (Grays Harbor County, 1981) Violation : to break, disregard, or infringe a law or a certa in agreement. (In this study: to break a traffic law) Citation : is another word for a traffic ticket. It is a not ice issued by a law enforcement official to a motorist or other road user, accusing violation of traffic laws. It could be cited as a moving vehicle which includes but not li mited to violations such as exceeding

PAGE 31

12 the speed limit or running red light or non-moving violation (illegally parked vehicle). (Grays Harbor County, 1981) Halo Effect “spillover” : Refers to the ability of an intersection safety c amera to have a positive effect at nearby, untreated intersections because of a longer term influence on driver behavior (For example, driver will not run r ed lights at intersections near intersections equipped with a red light camera). (N HCRP, 2003) Intrusive sensors: record vehicle count and classification data with some lane closure and drilling into the roadway. (Federal Highway Adm inistration’s Intelligent Transportation Systems Joint Program Office, 2000) Non-intrusive sensors: record vehicle count and classification data withou t interruption to traffic flow. Installation of non-intrusive dete ction systems usually involves no requirement for road closure or traffic management and deployment includes utilizing existing roadside infrastructure. (Federal Highway Administration’s Intelligent Transportation Systems Joint Program Office, 2000) The Kangaroo effect: A kangaroo effect is created when drivers decelerat e suddenly when they notice a speed camera or red light camera and then quickly accelerate again. This is thought to have an adverse effect on traffi c flow and the environment, as well as road safety. (Federal Highway Administration’s Inte lligent Transportation Systems Joint Program Office, 2000) Vehicle Detection and Surveillance Technologies Vehicle detection and surveillance technologies can be categorized into two major types: intrusive and non-intrusive sensors. These t ypes of technologies go through three main processes: the transducer, which detects the p resence of a vehicle or its axles; the

PAGE 32

13 signal-processing device, which then converts the t ransducer data into an electrical signal; and, finally, a data-processing device that converts the electrical signal into traffic parameters. There are several traffic parameters th at might be included like speed, vehicles count, occupancy, gap, weight, and many ot hers. (Bailly, 1998) In this section of the study, more information rela ted to the intrusive and nonintrusive sensors will be provided. The information will include the operating principle, sensor measurement accuracy, costs, advantages, and disadvantages of these technologies. Intrusive sensor (in-ground inductive loop). These types of sensors are usually installed into the surface of the pavement by tunne ling under the surface, in saw-cuts or holes on the surface, or by anchoring directly into the surface. Intrusive sensors can be micro-loop probes, pneumatic road tubes, or piezoel ectric cables and other weight-inmotion sensors. (Hockaday, 1991) % : Distribution of the loopÂ’s electromagnetic field. (Hockaday, 1991)

PAGE 33

14 There are many advantages to the intrusive sensor l ike unlimited number of speed measurements, the ability to specify the lane where the violation has occurred, and also the level of accuracy it provides when recording th e speed and the location of a vehicle. (Kell, 1990) & : Loop location at the intersection. (Kell, 1990) The main disadvantages and drawbacks that are mainl y associated with the intrusive sensors are the disruption they can cause to traffic operation during the installation processes and road closures, or when m aintenance is required, whether that type of maintenance is related to the sensor or oth er applications. They can also cause damage to the surface of the road, especially when substandard drilling and cutting activities are used when attaching the sensor to th e roadway.

PAGE 34

15 : Loop sensors reflect damages to the asphalt. (Kel l, 1990) As far as non-intrusive sensors (loopless trigger r adar), most studies show the need for a more reliable and cost-effective method that could be applied to the same applications as the intrusive sensor, but with fewe r disadvantages. Non-intrusive sensors came to be the solution since the installation of t hese sensors does not require the amount of cutting and drilling the intrusive sensors do, a nd therefore cause less traffic disruption and no damage to the surface at all. (Kell, 1990)

PAGE 35

16 ( : Speed limit cameras can take shapes of normal roa d elements. (Klein, Millimeter-Wave and Infrared Multisensor Design and Signal Processing, 1997) At the same time, non-intrusive sensors (abovegroun d sensors) have met many of the applications required by surface streets and fr eeways. The non-intrusive sensors can be mounted above or to the side of the roadway that needs monitoring. Many technologies are currently used for this applicatio n like laser radar, video images, microwave radars, and passive infrared, or a combin ation of two or more of them. The system is also able to record speed, vehicleÂ’s weig ht, vehicle categories, and vehicle count. (Klein, Millimeter-Wave and Infrared Multise nsor Design and Signal Processing, 1997)

PAGE 36

17 The sensor can be mounted in a position perpendicul ar or oblique to the traffic flow to allow the system to monitor each lane. In c omparison to the intrusive sensors, studies show that aboveground sensors are less affe cted by weather change and ambient lights, are faster and easy to install, have an acc uracy of speed detection that ranges +/2 mph, and monitor the configuration of each lane ind ividually. (Klein, Final Report: Mobile Surveillance and Wireless Communication Syst ems Field Operational Test Vol. 2: FOT Objectives, Organization, System Design, Res ults, Conclusions, and Recommendations, 1999) : Intrusive sensors and camera requires less effort and no damages. (Klein, Final Report: Mobile Surveillance and Wireless Communication Systems Fie ld Operational Test Vol. 2: FOT Objectives, Organization, System Design, Results, Conclusions, and Recommendations, 1999)

PAGE 37

18 Implications for Public Privacy Practically, RLC systems work by capturing the imag e of the vehicle, its driver, and the vehicle license plate number as that vehicl e goes through a red light. These photographs provide evidence to authorities in orde r to assist with traffic law enforcement and, therefore, the issuing of tickets to the violator. Typically, law enforcement officials will review the photographs a nd determine whether a violation has occurred. The next step is the infraction, which wi ll be mailed to the mailing address registered under the license plate. In some cases, photographs are not clear and therefore officials cannot make a final decision. As a result officials will either dismiss the citation or mail the violator a notice requesting identifica tion information to assist in making their decision. (Insurance Institute for Highway Safety, 2010) Using red-light camera systems is associated with s everal legal and privacy concerns, including concerns about citation distrib ution, types of penalties, and the right of authorities to issue a ticket based on a photogr aph. Before implementation, the public should be educated on how the system works to ensur e that the public understands that the citations are only issued after photographs are reviewed by a police officer. (Elmitiny & Radwan, 2008) Another issue related to RLC that the public is fre quently complaining about the availability of signage and in particular, the mess ages that need to be given to drivers about what is actually being monitored and enforced A study in the UK discussed the issues facing UK agencies responsible for implement ing and operating camera-based enforcement programs in relation to signage, as the camera signs can be located differently depending on whether or not they are fu nded by a Safety Camera Partnership.

PAGE 38

19 The study’s major finding was that signs must be us ed consistently across the UK in general, the boundaries of geo-political regions, o f police force authority and of road network operators' responsibility. (Wilson, 2007) Impact on Revenue The public argues that the main purpose behind RLC’ s is revenue. Several studies and researches have found much evidence towards thi s being the case. An article titled “Big Brother is Ticking You”, published as part of Popular Mechanics magazine, emphasizes on the fierce opposition to RLC by citiz ens and organizations such as the American Automobile Association and National Motori sts Association. The article referred to the Washington, DC experien ce with RLC, as the increased number of crashes at approaches where RLC is instal led, especially rear-end ones, have been associated with an increased number of revenue for the city. Adding to the general and growing discontent is the fact that a few towns have been caught shortening yellow signal timing, thereby catching more red light runn ers and generating more revenue but also inadvertently increasing accident rates. (Reyn olds, 2006) Tom Brodbeck, the Sun’s City columnist, argued whet her the RLC program in the city of Winnipeg is really aimed at safety and not revenue. He reevaluated the 50 most dangerous intersections around the city in terms of crashes and wondered why only 7 of the 31 red light cameras throughout the city are lo cated at those intersections. What is more interesting is the fact that there were no cam eras at all among the top 10 most dangerous intersections. “If the main purpose of th e cameras is to increase safety, then why they are not placed in the locations with the l east safety?” Tom asked.

PAGE 39

20 In fact, in two of those top 50 locations where the cameras were installed, the rate of crashes has risen around 20 percent. (Brodbeck, 201 2) According to a study conducted by OpEdnews.com, pol ice unions and for-profit camera companies have teamed up on several occasion s to defeat laws that proposed to ensure traffic cameras are designed for public safe ty rather than to collect revenue. For example, in Connecticut, police unions and traffic light camera companies opposed efforts to expand the length of yellow lights despi te the fact that implementing that would reduce red light violations by 90 percent. (Fang, 2 012) In Florida last year, American Traffic Solutions, o ne of the largest for-profit camera corporations, hired 17 lobbyists to defeat a similar bill. The company circulated a letter signed by police chiefs and worked closely w ith officials from the Florida Sheriff's Association, a labor group, to pressure legislators In California, a bill by State Sen. Joseph Simitian to ensure that traffic cameras can only be set up to promote public safety rather than collect revenue was opposed by the California Police Chiefs, a law enforcement lab or union group. (Tucker, 2009)

PAGE 40

21 Study Timeline ) *) + nnn r nrn n n !n #$%$rn$ & '$n ( nrn!) $ rn $nnn! *nn! #n #n!)n$r rn +nrn ,nn!$nn$ $nr$ & -n.nnrn ( '$n ,!n$$nn nnnn /nnnrn 0$nrn 0$nn$n #$%$rrn$ #$%$rrnnr 1!n!rn #n $nrn & *nn! ( $rnnn!)n$r nr 2nrnnrn! rn ,n)rnr 3n$n

PAGE 41

22 Dissertation Structure The dissertation has been divided into five chapter s: Chapter 1: Introduction including a statement of the problem, study objectives and main questions, the study hypothesis, and how it contrib utes to the transportation engineering industry. Chapter 2: Background overview of the topic, traffic safety o verview, history of red light cameras, glossary of terms, vehicle detection and surveillance technologies a brief discussion of two of the major concerns that associ ated with RLC systems, implication on public privacy and impact on revenue. The chapter e nds by presenting the study timeline, and dissertation structure. Chapter 3: Presents the literature review, which includes a c omprehensive review of the most recent studies and articles on the subject of RLC systems. This chapter is categorized and divided according to the most recen t research topics as it starts first with the technical part of the RLC detection types, goin g through studies related the of RLC on safety (mostly before/ after comparison) RLC collision types, RLC crashes severity, Red light runner characteristics, RLC and signal timings, RLC spillover effect, and most recent models and procedures used to condu ct RLC related studies. This chapter also reviews the recent researches that discussed t he importance of RLC site selection, which is the focus of this dissertation. Chapter 4: Presents the procedures, models, and the phases of the methodology chapter that will be used for RLC sites selection. It also represents the analytical-based methodology that will be used for RLC site selectio n using case studies from Colorado

PAGE 42

23 Springs, Denver, and Fort Collins. This chapter als o presents the data required for each case study. Chapter 5: Presents the analytical findings for all intersect ions within each case study city limit and conclusions from Colorado Springs, D enver, and Fort Collins, which eventually show the top 10 candidate intersections that have priority over other signalized intersections for RLC installation. This chapter al so includes recommendations that support future studies on this subject.

PAGE 43

24 CHAPTER,,, LITERATURE REVIEW Introduction Red light running is a significant public health co ncern, killing more than 800 people and injuring more than 200,000 in the United States per year. It is a significant safety problem as drivers become more aggressive on city roads, and become impatient waiting for traffic signals to change. RLC programs are considered one of the most controversial topics facing traffic engineers, city councils, and public awareness groups. Red light running cameras systems are automated enf orcement systems that detect and capture vehicles that run a red light and issue a c itation. RLC systems are becoming widely used in the United States to reduce the numb er and severity of red light running crashes. (Fitzsimmons, Hallmark, McDonald, Orellana Matulac, & Pawlovich, 2008) In this chapter of the research, many of the recent studies and researches related to the red light camera programs will be presented by discussing major areas of previous researches such as the effectiveness of RLC on safe ty, effectiveness of RLC on type of crashes, effectiveness of RLC on crashes severity, characteristics of red light runners, RLC and signal timings, RLC spillover effect, and m ethodologies and analysis procedures used to measure effectiveness of RLC. F inally, the literature review chapter will finally focus on the most recent studies relat ed to the main objective of this research, which is the RLC sites selection. Effectiveness of RLC on Safety Studies evaluating the effectiveness of red light c ameras on safety mostly suggest that they are effective in reducing red light viola tions and injury crashes. A four-year

PAGE 44

25 analysis (2004-2007) of the effectiveness of the RL C program in Raleigh, North Carolina, which was a follow-up study to an earlier one made before 2004 but with a smaller sample size (5 months), both showed that the progra m is producing positive safety results. (Hummer & Cunningham, 2010) In San Francisco, California, with its compact driv ing environment and dense network of signalized intersections, red-light runn ing reached a political crisis in 1994. The city and county of San Francisco recently compl eted a pilot red-light photoenforcement program. The number of vehicles photogr aphed violating red lights at the photo-enforced locations dropped by more than 40% j ust 6 months into the pilot. Recent statistics indicate that San Francisco's combined e fforts to combat red-light running have resulted in a significant decrease in the number of annual crashes caused by red-light violators citywide. Based on the success of the pil ot and supportive state legislation, San Francisco is moving forward to expand the red-light photo-enforcement program to make it one of the largest programs in the United States with 26 cameras rotating in 35 locations. (Fleck. J, 1999) Iowa is another state that has a serious safety pro blem with red light running that accounts for 35% of fatal and major injuries plus 2 1% of total crashes at signalized intersections. The state has adopted the program i n three communities; one of the communities is Davenport that had installed the pro gram back in 2004. Two years of crash data after installation were available for an alysis, which included 4 RLC locations and 5 control intersections as part of it. The resu lts of the analysis indicated that the cameras were effective in reducing total crashes an d RLR related-crashes on average of 20% and 40%, respectively. In the other hand, there was an increase of total crashes,

PAGE 45

26 RLR related-crashes, and RLR rear-end related crash es of about 7%, 20%, and 33%, respectively. (Hallmark, Orellana, Fitzsimmons, McD onald, & Matulac, 2010) A comprehensive study conducted by the Center of Ci vic Engagement at Rice University from September 2006 to August 2008, whic h included 70 monitored and nonmonitored approaches and six years of crashes data, concluded that the proportion of crashes occurring at monitored approaches decreased significantly relative to the nonmonitored approaches, as Figure (9) shows below. Th e comparison of data between monitored and non-monitored approaches supports the conclusion that red light cameras are mitigating a general, more severe increase in c ollisions. Although this study supports the idea that red light cameras have a positive eff ect in reducing crashes at monitored approaches in comparison with non-monitored approac hes, several questions have been raised by these findings. The most important of the se is “Why have crashes at nonmonitored approaches increased so dramatically in t he past year?” The study suggested that these results could be evidence of an increase in crashes across the city. The selection in 2006 of intersections with high rates of crashes could be serving to magnify this effect. (Dahnke, Stevenson, Stein, & Lomax, 2008)

PAGE 46

27 +))rr r))rr))r !*r"""-.r/" # An additional study was conducted to estimate the s afety impacts of RLCs on traffic crashes at signalized intersections in the cities of Phoenix and Scottsdale, Arizona. Twenty-four RLC equipped intersections in both citi es were examined in detail. The evaluation results indicated that both Phoenix and Scottsdale are operating cost-effective installations of RLCs, which show positive safety i mprovement: however, the variability in RLC effectiveness within jurisdictions is larger in Phoenix (Shin & Washington, 2007). A paper is to evaluate the safety effectiveness of automated traffic enforcement systems, that is, red light cameras, installed at 2 54 signalized intersections in 32 jurisdictions in Texas. A before-after study by the empirical Bayesian methodology was performed to remove the regression-to-mean bias dur ing the evaluation of treatments. The results indicate significant decreases in the i ncidences of all types of red light running (RLR) crashes and right-angle RLR crashes b y 20% and 24%, respectively. A significant increase of 37% for rear-end RLR crashe s was discovered. The study results

PAGE 47

28 suggest that a significant safety benefit for red l ight cameras is achieved if intersections have four or more RLR crashes per year or have two or more RLR crashes per 10,000 vehicles. Red light cameras show counterproductive results if intersections experience fewer than two RLR crashes per year or have one cra sh per 10,000 vehicles per year (Ko, 2013). In Virginia, a study included six jurisdictions (Al exandria, Arlington, Fairfax City, Fairfax County, Falls Church, Vienna) that deployed red light cameras. It documented the safety impacts of those cameras based on 7 years of crash data for the period January 1, 1998, through December 31, 2004. The results show that cameras were associated with a modest reduction in comprehensive injury crashes. ( Garber, Miller, Abel, Eslambolchi, & Korukonda, 2007) A study that evaluated the Red Light Camera (RLC) p rogram in Fairfax County, Virginia was conducted back in 2003 and covered 13 cameras after 2 years of operation. In conducting the analysis, violation results were grouped into two distinct periods: 1) initial period (1st three months) and 2) after initial period. These two distinct periods were also grouped into five periods for each, and t here were as follows: 1) initial period, 2) fourth to ninth month period, 3) 10th to 15th, 3) 16th to 21st, 4) 22nd to 27th, 5) after 27th month. The study reported that the RLC program redu ced the traffic signal violation rate by up to 63% in the 22nd to 27th month period of it s operation (see Table 1). The results also show that the increase of the intersection amb er time, combined with RLC, produced a higher reduction of up to 72% in violation rate. The crash rate was reduced by 27% after 2 years of RLC operation; however, this reduc tion was not statistically significant. (Hobeika & Yaungyai, 2006)

PAGE 48

29 +r0.1r2rrr/1 2"3r!4r-5r2r" '# Camera Intersections Average number of violations/ 10,000 vehicles % Cha nge in violations per 10,000 vehilces Initial Period 4-9 mo 10-15 mo 16-21 mo 22-27 mo After 27 4-9 mo 10-15 mo 16-21 mo 22-27 mo After 27 n nnnrr nnn rnnn rnrnnnrn nn nnnnn rnnnnnn nnrn rrnrnnn nrnn nrrr nrnrrrr

PAGE 49

30 A very interesting pilot study was conducted in Mai ne, which is considered as one of the states that have a major problem with red li ght running. Maine is one of the states that does not allow issuing citations based on phot ographic evidence, so only warning letters were issued to violators. Therefore, the st udy covering the period from September 2004 to August 2005 was mainly concerned with the r eduction of red light running violations as a result of warning letters only. Ob servations of red-light running indicate that the violation rate dropped by around 28% betwe en December 2004 (when the system was first installed) and May 2005 (when the system had been operational for several months). However, it was the infractions that occur red at low speeds and within the first second or so that were reduced. Infractions more th an 3 seconds into red and at speeds above 35 mph actually increased. It was interprete d that these later infractions were not caused by the enforcement, but rather by other fact ors like weather and roadway conditions. Conflict and crash data indicate that t here were no great improvements in safety between the before period and the period whe n the system was in operation. Actual fines and RLC systems rather than warning tickets m ay have produced greater safety effects. (Garder, 2006) Another study was mainly aimed at estimating the RL R problem in Indiana. The other objectives of the research included: (1) lear ning drivers' opinions on the problem, (2) studying the effectiveness of selected counterm easures, (3) studying the legal issues related to photo-enforcement. A crash statistics s tudy, telephone survey, and extended monitoring of a selected intersection were the thre e major investigations chosen to estimate the magnitude of the problem. The crash st atistics for the 1997-1999 period showed that 22% of signalized intersection crashes in Indiana resulted from RLR. RLR

PAGE 50

31 preceded 50% of fatal crashes at these intersection s. The telephone survey showed that 67% of Indiana drivers felt that RLR was a problem in the state, and 12% of them claimed to have been involved in a RLR crash. The e xtended monitoring of the through movements at the study intersection also recorded a considerable violation rate. Traffic at a selected intersection in West Lafayette, Indiana, was videotaped and the video material was used to detect the red light violations. The ex pected number of drivers arriving at the start of the red signal has been proposed as a true measure of exposure to RLR. The authors call it an opportunity for RLR. This exposu re was used to estimate the RLR rate. The statistical significance of the difference in t he RLR rates between different periods was estimated using binomial distribution. Photo-en forcement reduced the violation rate by 62% during the week of enforcement and by 35% du ring the week immediately following the start of enforcement. (Tarko & Reddy, 2003) As a conclusion of this section of the chapter, it seems clear that most studies agreed on a certain level of improvement associated with the installation and operation of RLC programs. Studies were made using different me thodologies, time periods, data, and locations; however, they all concluded that the re were positive implications of RLC. Table 2 briefly presents all of these studies and t heir findings.

PAGE 51

32 +rr20.1 r2 Study Title Location Time Period/ Data Type Findings Evaluating the Effectiveness of Red-Light Running Camera Enforcement in Raleigh, North Carolina (Hummer & Cunningham, 2010) Raleigh, North Carolina 4 years (2004-2007) Positive safety results. (Significant in 2 groups) Analysis of the RLC effectiveness on reducing red light violations and injury crashes. 5 groups data sets Can we make red light runners stop? Red light photo enforcement in San Francisco, California (Fleck. J, 1999) San Francisco. California 1994 Vehicles violating RLC decreased by 40% a pilot red-light photoenforcement program analysis + Intention for future expansion of the program Red Light Running in Iowa: Automated Enforcement Program Evaluation with Bayesian Analysis (Hallmark, Orellana, Fitzsimmons, McDonald, & Matulac, 2010) Davenport. Iowa 2004 RLC us effective in reducing total crashes and RLR crashes Two years of after installation data including control intersections Evaluation of the City of Houston digital automated red light camera program (Dahnke, Stevenson, Stein, & Lomax, 2008) The Center of Civic Engagement at Rice University 2001-2006 of crashes data included 70 of monitored and non-monitored approaches Monitored approaches crashes decreased significantly relative to the non-monitored approaches Houston. Texas The impact of red light cameras on safety in Arizona (Shin & Washington, 2007) Phoenix and Scottsdale, Arizona 2000-2005 of before data crashes Positive safety improvement + more effectiveness results in Phoenix. The Impact of Red Light Cameras (Photo-Red Enforcement) on Crashes in Virginia (Garber, Miller, Abel, Eslambolchi, & Korukonda, 2007) Six jurisdictions in Virginia (Alexandria, Arlington, Fairfax City, Fairfax County, Falls Church, Vienna) 1998-2004 of crashes data Modest reduction on comprehensive injury crashes Traffic Conflict Studies Before and After Introduction of RedLight Running Photo Enforcement in Maine (Garder, 2006) Maine September 2004 to August 2005 28% decrease of low speeds and within the first second Infractions + increase of infractions at more than 3 seconds into red and at speeds above 35 mph. The reduction of red light running violations as a result of warning letters only. Evaluation of safety enforcement on changing driver behavior (Tarko & Reddy, 2003) West Lafayette, Indiana Crash statistics for the 19971999 period The photo-enforcement reduced the violation rate by 62% during the week of enforcement and by 35% during the week immediately following.

PAGE 52

33 Effectiveness of RLC on Type of Crashes When reviewing studies concerned about the type of crashes at signalized intersections, it looks obvious that the crash type that is targeted in the analyses is the right-angle (T-bone) crashes, which involve a viola ting vehicle colliding with another vehicle crossing the intersection legally on a gree n signal display. Another crash type likely to be investigated is a vehicle turning left colliding with a vehicle moving through the intersection from the opposite approach directi on. For this later scenario, the turning vehicle could be violating the red when the opposit e direction has a green, or vice-versa. On the other hand, there is a concern that rear-end crashes of vehicles approaching the intersection will increase with RLC enforcement. Kn owing that there is a camera system, and on seeing the yellow display, a more cautious m otorist may stop more abruptly, causing the following motorist, not anticipating th e need to stop and likely to be following too closely, to hit the lead vehicle from behind. Assuming that these crash types produce equal crash severity, then a net bene fit would accrue if the crash reductions of the angle type exceeded any crash increases of t he rear-end type. In general, angle crashes are usually more severe and, therefore, eve n a zero change in total crashes may prove to be safer, if there is a smaller proportion of angle to rear-end crashes with the use of cameras. Red-light camera enforcement offers potential as a cost-effective, powerful tool in reducing red-light running and associated crashes. However, studies on the effectiveness of the red-light camera system have shown mixed res ults in terms the of types of crashes associated with the system, with some studies showi ng a reduction in T-bone red-light related crashes, while others report no significant improvement. Furthermore, most

PAGE 53

34 studies have shown that red-light camera systems in creased rear-end crashes. (Elmitiny & Radwan, 2008) As with all synthesis documents, a comprehensive re port published by the National Cooperative Highway Research Program was p erformed that relied exclusively on available information; no new data collection or analysis. The information came from published literature, various websites, and from a questionnaire sent to more than 50 jurisdictions nationwide and some foreign countries known or believed to have installed red light running camera systems. The findings tha t can be drawn from the information complied by that study are as follows. There is a p reponderance of evidence, albeit inconclusive, indicating that red light running cam era systems improve the overall safety of intersections where they are used. As expected, angle crashes are usually reduced and, in some situations, rear-end crashes increase, but to a lesser extent. (The National Cooperative Highway Research Program, 2003) As an example, before-after crash results for Sacra mento California are shown on Table (3) below. (McGee & Eccles, 2006) +r$02r6r2r r1rr!78-9" '# Crashes No. of Crashes 12 Months Before Installation No. of Crashes 12 Months After Installation Change (%) Total number of crashes 81 73 -10 Injury crashes 60 44 -27 Right-angle crashes 42 31 -26 Rear-end crashes 32 28 -12 Red light crashes 28 17 -39

PAGE 54

35 The effectiveness of a group of red light camera in stallations in Sydney in reducing right angle and right(left) turn opposed crashes was analyzed using crash data from 2 years before and 2 years after the cameras w ere installed (See Table 4). The study, published in 1993, had 6 cameras circulating in 16 intersections with cameras and covered another 16 intersections as control (the co ntrol sites were matched on the basis of crash history, traffic volume, and intersection con figuration). The camera (treatment) and control sites were grouped as follows: Eight most-u sed camera sites, eight least-used camera sites, eight most-used control sites, and ei ght least-used control sites. The study concluded that red light cameras reduce target cras hes and increase rear end crashes with an overall reduction in accident numbers and severi ty that was similar to other engineering countermeasures. (Hillier, Ronczka, & S chnerring, 1993) +r%:rrrr"22 ";rr!4"0
PAGE 55

36 concluded that the use of RLCs led to the following : 25 percent decrease in total rightangle crashes, 16 percent reduction in injury right -angle crashes, 15 percent increase in total rear-end crashes, and 24 percent increase in injury rear-end crashes. As Table (5) below shows, the direction of these effects was rem arkably consistent across jurisdictions. The analysis indicated a modest spillover effect on right-angle crashes; however, this was not mirrored by the increase in rear end crashes se en in the treatment group, which detracts somewhat from the credibility of this resu lt as evidence of a general deterrence effect. (Administration, 2005) +r&0r= rr!;r" &# Jurisdiction Number (%) Change in Right Angle Crashes (Standard Error) (%) Change in Rear End Crashes (Standard Error) 1 -40.0 (5.4) 21.3 (17.1) 2 0.8 (9.0) 8.5 (9.8) 3 -14.3 (12.5) 15.1 (14.1) 4 -24.7 (8.7) 19.7 (11.7) 5 -34.3 (7.6) 38.1 (14.5) 6 -26.1 (4.7) 12.7 (3.4) 7 -24.4 (11.2) 7.0 (18.5) The standard error is the standard deviation of the sampling distribution of a statistic. The term may also be used to refer to an estimate of that standard deviation, derived from a particular sample used to compute the estimate. ( Everitt, 2003) Consistent with findings in other regions, the stud y that was conducted in Arizona has concluded that angle and left-turn crashes are reduced in general, while rear-end crashes tend to increase as a result of RLCs. In S cottsdale, for instance, the crash trends suggest that an effort to reduce angle crashes thro ugh the use of RLCs may be

PAGE 56

37 worthwhile, since angle crashes are generally more severe than rear-end crashes (See Figure 10 below). (Shin & Washington, 2007) >rr2)r %2r)!-?r" (# The Virginia study (presented earlier as part of RL C effectiveness on safety section) that includes six different jurisdictions found that cameras are associated with an increase in rear-end crashes (about 27% or 42% depe nding on the statistical method used) and a decrease in red light running crashes (about 8% or 42% depending on the statistical method used). It also shows that there is signific ant variation by intersection and by jurisdiction: one jurisdiction (Arlington) suggests that cameras are associated with an increase in all six crash types that were explicitl y studied (rear-end, angle, red light running, injury red light running, total injury, an d total) whereas two other jurisdictions saw decreases in most of these crash types. (Garber Miller, Abel, Eslambolchi, & Korukonda, 2007)

PAGE 57

38 Effectiveness of RLC on Crashes Severity According to the Insurance Institute for Highway Sa fety, during the period from 1992 to 1998, almost 6,000 people (approximately 85 0 per year) died in RLR crashes in the Unites States, and another 1.4 million (approxi mately 200,000 per year) were injured in crashes that involved red light running. Using 1997 data from the General Estimates System a nd a narrower definition of RLR crashes, Smith. et. al, estimated that approxim ately 97,000 crashes, resulting in 961 fatalities, could be attributed to red light runnin g in the United States per year during this same period. Table 6 shows the distribution of cras hes by severity for all signalized intersections, those involving angle crashes, and t hose considered to be the result of red light running. As seen, slightly more than 44% of t he fatalities at signalized intersections were attributed to red light running. (McGee & Eccl es, 2006) +r'+r22 rr<(!78-9" '# Crashes Measure Signalized Intersections Angle Cras hes at Signalized Intersections Red Light Running Fatal crashes 2,176 1,587 961 (44%) Injury crashes 318,000 261,000 51,000 (16%) PDO crashes 469,000 361,000 45,000 (9.5%) Total crashes 789,000 623,000 97,000 (12%) Fatalities 2,344 1,729 1,059(45%) Injuries 543,000 464,000 91,000(16%) Note: Percentage is calculated out of total crashes at signalized intersections. A study was done to evaluate the crash effects of 8 7 signed fixed digital speed and red light cameras and accompanying warning sign s placed at 77 signalized intersections across Victoria, Australia. Across th e 77 intersections where the cameras

PAGE 58

39 evaluated were installed, it was estimated that 17 serious or fatal crashes per year and 36 minor injury crashes would be prevented, representi ng crash cost savings to the community of over $8 million per year. Based on the outcomes of the evaluation, continued and expanded use of combined fixed red-li ght and speed cameras in Victoria is expected to improve driver safety, save lives and r educe crash related costs. Analysis results estimated large decreases in casualty crash es associated with the FDSRL cameras and their associated signage. When only the crashes involving vehicles travelling from the approach intersection leg where the camera was placed are considered, the estimated casualty crash reduction was 47%. When crashes invo lving vehicles from all approaches are compared, the estimated casualty crash reductio n was 26%. A 44% reduction in right angle and right turn against crashes, those particu larly targeted by red light enforcement, was also estimated. While use of the FDSRL cameras was associated with a reduction in overall casualty crash risk, there was no evidence for a reduction in relative crash severity meaning the cameras were associated equally with re ductions in minor injury crashes as serious injury and fatal crashes. (Budd, Scully, & Newstead, 2011) An article examines the effectiveness of red-light cameras at reducing the rate of violations as well as the level and severity of int ersection-related crashes. Although the evaluations differ in sample size, type of intersec tion and evaluation methods, several trends emerge. The findings suggest that if install ed at locations with significant red-light running crashes and/or violations, red-light camera s substantially reduce red-light violation rates and reduce crashes that result from red-light running. Although they may not reduce total crashes, they usually are effectiv e at reducing crash severity. The author

PAGE 59

40 finally suggested that red-light cameras enforcemen t should not be seen as a substitute for proper traffic engineering of signalized intersecti ons. (Bochner & Walden, 2010) A study developed a Bayesian HBL (hierarchical bino mial logistic) model to identify the risk factors on individual severity of driver injury and vehicle damage at urban intersections of Singapore. For the study to conclude significant findings, it was helpful to account for the severity correlation of driver–vehicle units involved in the same multi-vehicle crashes. The study included various geometric features, traffic conditions, and driver–vehicle characteristics, as well as nine variables identified as significant using 95% BCI (Bayesian credible interval). Among these, the crash-level significant factors are Time of Day, Intersection Type, Nature of Lane, Street Lighting, Presence of Red Light Camera, and Pedestrian Involved. In particul ar, it was found that crashes occurring in peak time, in good street-lighting condition, an d in the case of pedestrians involved are associated with lower severity, while those occurri ng in night time, at T/Y type intersections, on right-most lane, and in the prese nce of red light cameras have larger odds of being severe. Vehicle type, Driver Age and Involvement of Offending Party were also found to affect severities of driver injury an d vehicle damage significantly. Specifically, results indicated that heavy vehicles have a better resistance to serious injury or extensive damage, while two-wheel vehicle s, young or aged drivers, with the involvement of offending party have a higher risk o f being high severity. (Helai, Chor, & Haque, 2008) The study that included Phoenix and Scottsdale (Ari zona) also investigated the severity of crashes occurred at RLC intersections. It concluded that injury and fatal crashes of approximately 16.95 per year occurred at RLC intersections of Phoenix,

PAGE 60

41 compared to 10.43 per year in Scottsdale. Addition ally, the number of rear-end crashes resulting in injuries or fatalities (5.67/year) is higher than that of angle crashes (2.41/year), as was found previously. Further exami nations however, again show that angle crashes are more serious than rear-end crashe s. Figures (11) and (12) below show the proportion of crashes by severity. The percentage of PDO crashes and minor crashes for rea r-end crashes is higher than the percentage of injury/fatal crashes. (Shin & Washing ton, 2007) >rr)2r2r 2)r2!>*@=2rrr#! ?r" (#

PAGE 61

42 >rr)2r2r 2)r2!r=#!?r" (# A study shows the attitude of people toward red lig ht cameras in 14 cities with red light camera programs concluded that two thirds favor the use of cameras for red light enforcement and 42 percent strongly favor it. The c hief reasons for opposing cameras were the perceptions that cameras make mistakes and that the motivation for installing them is revenue, not safety. Forty-one percent of d rivers favor using cameras to enforce right-turn-on-red violations. Nearly 9 in 10 driver s were aware of the camera enforcement programs in their cities, and 59 percen t of these drivers believed that the cameras have made intersections safer. Almost half know someone who received a red light camera citation, and 17 percent had received at least one ticket themselves. When compared with drivers in the 14 cities with camera programs, the percentage of drivers in Houston who strongly favored enforcement was about the same (45%), but strong

PAGE 62

43 opposition was higher in Houston than in the other cities (28 versus 18%) (Mccartt, 2012) Characteristics of Red Light Runners Knowing the characteristics of the red light runner s has been another point of interest to many researchers. It is another way to mitigate this serious problem that is considered among the most risky behaviors in the tr ansportation system by defining the characteristics of those drivers who run red lights more frequently comparing to others. A study was conducted in Southeast Virginia that in cludes eight intersections and covers an 8-month period during which photo enforce ment cameras were installed at three sites (A1, A2, and A3). As Table (7) shows, d ata collectors observed 1765 light cycles. Overall, 18.8% of last drivers entered inte rsections on green lights, 68.4% on yellow, and 12.7% on red. Demographics were recorde d for 1433 drivers (only the yellow and red light runners). Demographics of red light runners across the five data collection periods are provided in table 7. The num bers represent the percent of red light runners out of all yellow and red light runners dur ing that observation period broken down by subcategories for each demographic variable Overall, men had higher raw red light running rates than women; however, the only s ignificant difference between men and women occurred in Phase 1. Red light running ra tes for both men and women declined from baseline levels and reached their low est levels during Phase 4. The only significant difference in red light running as a fu nction of ethnic group classification was during Phase 2 when non-whites were more likely to run red lights than whites. Note that numbers in parentheses are the sample sizes for cat egories each phase of the project. The

PAGE 63

44 percent represents those who ran the red light as o pposed to the yellow light. (Martinez & Porter, 2006). The phases (collection periods) are : Phases 1 and 2: These observations took place in June and July 200 4, respectively, before any cameras were installed and served as bas eline measures of red light running behavior. Phase 3: In September 2004, observations occurred again. In tersection A1 received cameras and was in the 30-day warning peri od (i.e., when warning letters were mailed to the registered owners of vehicles that ra n the red light). Phase 4: This observation took place in November 2004. Inte rsection A1 was in the actual citation phase, A2 was in the warning ph ase, and cameras at A3 were being tested to go operational the day after it was was o bserved. (Note that this observation phase took place in November when it would get dark about 5 p.m. and that the camera flash allowed for no mistake that it was functional .) Phase 5: The fifth observation phase occurred in January 20 05 when A1, A2, and A3 were issuing citations. +r(>rr 2r)r2!7r<->" '# Demographics Phase 1 Phase 2 Phase 3 Phase 4 Phase 5 Gender Female 13.0 (108) 16.7 (96) 17.0 (88) 8.0 (50) 10.1 (69) Male 25.6 (207) 18.5 (162) 19.5 (128) 10.8 (93) 14.4 (11 1) Race White 20.3 (177) 14.1 (149) 17.0 (147) 11.0 (100) 15.4 (1 23)

PAGE 64

45 Demographics Phase 1 Phase 2 Phase 3 Phase 4 Phase 5 Non-White 24.3 (103) 21.5 (93) 18.1 (83) 9.8 (41) 10.4 (47) Safety Belt Use Yes 20.4 (113) 10.5 (114) 12.4 (113) 4.3 (70) 15.7 (102 ) No 27.3 (77) 22.6 (62) 23.4 (77) 13.9 (36) 10.4 (48) Age Group 25 or younger 24.5 (102) 23.1(65) 22.2 (54) 11.1 (36) 15.9 (44) 26-35 19.0 (84) 16.8 (107) 26.8 (71) 6.4 (47) 12.9 (62) 36 and older 19.7 (76) 11.3 (62) 9.0 (78) 10.9 (46) 16.0 (50) Number of People in vehicle 1 17.3 (260) 18.5 (211) 20.4 (201) 7.9 (126) 11.4 (15 8) 2 or more 32.4 (68) 14.3 (63) 13.6 (59) 15.6 (32) 19.5 (41) Note: The word phase refers to project phases Another report conducted by the Volpe National Tran sportation Systems Center presents results from an analysis of about 47,000 r ed light violation records collected from 11 RLC equipped intersections in the City of S acramento, California, between May 1999 and June 2003. The report used seven differen t variables to study the characteristics of red light runners, these variabl es are: Age of the violator Gender of the violator Time (in hours) when the violation occurred Model Year of the vehicle driven by the violator

PAGE 65

Measured vehicle speed at the time of the violation Elapsed tim e from the onset of red signal until the time of th e violation The distribution of repeat red light offenders The report suggests that younger drivers under 30 y ears of age are lights than drivers in other age groups (See Figure 13) violators that is shown in the figure is categorized by 7 age groups and the number of licensed drivers (LDs) in California, the total million traveled (MVMT) and relative ratios of red light driver percent ages and total MVMT percentages were plotted and presented in Figure that about 56 percent of the violators (See Figure 14) Moreover, 94 percent of the violations occurred w ithin 2 seconds after the onset of red light, and only 3 percent of the v iolations were recorded 5 seconds after the onset of red light. App Measured vehicle speed at the time of the violation e from the onset of red signal until the time of th e violation The distribution of repeat red light offenders The report suggests that younger drivers under 30 y ears of age are more likely to run red lights than drivers in other age groups (See Figure 13) (Note: Distribution of red light that is shown in the figure is categorized by 7 age groups and included the number of licensed drivers (LDs) in California, the total million vehicle miles and relative ratios of red light violation (RLV) percentages by licensed ages and total MVMT percentages ). Relative ratios for were plotted and presented in Figure 13. Additionally, the report indicates that about 56 percent of the violators were traveling at or below the posted speed limit Moreover, 94 percent of the violations occurred w ithin 2 seconds after the onset of red light, and only 3 percent of the v iolations were recorded 5 seconds after the onset of red light. App roximately 4 percent of the violators were repeat o ffenders. 46 e from the onset of red signal until the time of th e violation more likely to run red (Note: Distribution of red light included data on vehicle miles violation (RLV) percentages by licensed and Additionally, the report indicates were traveling at or below the posted speed limit Moreover, 94 percent of the violations occurred w ithin 2 seconds after the onset of red light, and only 3 percent of the v iolations were recorded 5 seconds after roximately 4 percent of the violators were repeat o ffenders.

PAGE 66

47 $nr
PAGE 67

48 expectation – most of red light violations occurred during the daytime hours when most urban driving is done (i.e., 7 AM to 7 PM). However the highest count of red light violations during the time period from 2:00 PM to 2 :59 PM is somewhat surprising. Finally, red light violations rates are estimated b etween 6 and 29 violations per 100,000 intersection-crossing vehicles. (Yang & Najm, 2006) &*r2 r2!5r-nr=" '# A study introduced by the National Highway Council concluded that 96 percent of drivers in a recent survey fear they will get hi t by another vehicle running a red light when they enter an intersection. Some 800 licensed drivers aged 18-65 were polled. Twothirds of the respondents see other drivers run red lights every day, with 54% speculating that the culprits were in a hurry. The National Hig hway Traffic Safety Administration

PAGE 68

49 counted 1,114 traffic deaths in 1997 in intersectio ns where drivers failed to heed red-light signals. (Karr, 1999) RLC and Signal Timings Two principal methods used to reduce red light runn ing involve lengthening the duration of yellow change intervals and automated r ed light enforcement. These two types of countermeasures were usually supported by studies from different points of view that tried to conclude which of them is more effici ent. A study evaluated the incremental effects on red li ght running of first lengthening yellow signal timing, followed by the introduction of red light cameras. At six approaches to two intersections in Philadelphia, Pe nnsylvania, yellow change intervals were increased by about 1s, followed several months later by red light camera enforcement. The number of red light violations was monitored before changes were implemented, several weeks after yellow timing chan ges were made, and about 1 year after commencement of red light camera enforcement. Similar observations were conducted at three comparison intersections in a ne ighboring state where red light cameras were not used and yellow timing remained co nstant. Results showed that yellow timing changes reduced red light violations by 36%. The addition of red light camera enforcement further reduced red light violations by 96% beyond levels achieved by the longer yellow timing. As a conclusion, the study sh ows that the provision of adequate yellow signal timing reduces red light running, but longer yellow timing alone does not eliminate the need for better enforcement, which ca n be provided effectively by red light cameras. (Retting & Williams, 1996)

PAGE 69

50 The City Council for the City of Springfield, Misso uri, approved a contract to install up to sixteen cameras for automated red lig ht enforcement in the spring of 2006. During the implementation phase of the program, tes t sampling of potential intersections for placement of the cameras revealed significant d ifferences in yellow timings and red light running at city signals compared to Missouri DOT signals inside the city. This difference prompted city and state traffic engineer s to review their respective methods of calculating the yellow and all-red timings. Despite using the same equation recommended by ITE, the agencies used different assumptions for perception-reaction time and how to interpret and use the results. City and state traff ic engineers came to agreement and documented the assumptions to be used in a Memo of Understanding (MOU) to bring consistency to the yellow and all-red timings throu ghout the city. The result was that yellow time at all city signals was increased and y ellow time at nearly all state signals was decreased. All signals were retimed in conforma nce with the MOU in the spring of 2008 and in conformance to ITE recommended practice three months prior to the first red light camera startup and 18 months prior to the installation of a camera on an intersection where the yellow time had been reduced The result of the signal retiming has brought credibility to the red light camera pro gram for the public and media with a reduction in rear-end crashes in addition to a redu ction in total crashes at traffic signals. (Newman, 2010) Methodologies and Procedures Used for RLC Analysis Several types of methodologies and analysis procedu res were used such as the Binary model, which was preliminarily developed to examine how the stopping–crossing

PAGE 70

51 decision of drivers at the onset of amber is affect ed by geometric, traffic, and situational variables. Results showed that the presence of RLCs is one of the five significant factors affecting a driverÂ’s decision to cross at the onset of amber. A Multinomial logic model further confirmed that RLCs are effective in reduci ng RLR frequency. Further analysis on the fitted models revealed that while the presen ce of RLCs is effective in reducing risk of right-angle crashes, it has a mixed effect on th e risk of rear-end crashes. Whether the RLC reduces or increases the possibility of rear-en d crashes depends on the speed of the trailing vehicle and the headway between vehicles. (Helai, Chor, & Haque, 2008) ';)rrrrrr 0.0!;r" &# Another study conducted by the Federal Highway Admi nistration used the Empirical bays for before and after crashes data fr om 132 treatment sites. Crash effects detected were consistent in direction with those fo und in many previous studies:

PAGE 71

52 decreased right-angle crashes and increased rear en d ones. The economic analysis examined the extent to which the increase in rear e nd crashes negates the benefits for decreased right-angle crashes. There was indeed a m odest aggregate crash cost benefit of RLC systems. The study concluded that economic bene fits (see Table 8) could go to its highest level during the occurrence of the highest total entering average annual daily traffic, the largest ratios of right-angle to rear end crashes, and the presence of protected left-turn phases. Note that FHA used samples (K+B+C+A) to refer to di fferent crash cost levels. A refers to cost estimate of fatal and serious crash levels, K for injury estimate of right angle crashes, and Band Clevel refer to injury estimate of rear end and left turn crashes. (Federal Highway Administration, 2005) +rrr22 rr2!r 4Ar2;r" &# Crash severity level Right-angle crash cost Rear en d crash cost O (Standard deviation) $ 8673 (1285) $11463 (3338) K+A+B+C (Standard deviation) $64468 (11919) $53659 (9276) Note: In this study, (K+A+B+C) are all combined to refer to injury level crashes due to inconsistent s ample size while O refers to non-injury level crashes. (A dministration, 2005) A meta-analysis was used to determine the effects o f red-light cameras (RLCs) on intersection crashes. The study shows that the size and direction of results reported from studies included in the meta-analysis are strongly affected by study methodology. The studies that have controlled for most confounding f actors yield the least favorable results. Based on these studies, installation of RLCs leads to an overall increase in the number of

PAGE 72

53 crashes by about 15%. Rear-end crashes increase by about 40% and right angle crashes, which are the target crashes for RLC, are reduced b y about 10%. All effects are, however, non-significant. Meta-regression analysis shows tha t results are more favorable when there is a lack of control for regression to the me an (RTM). (Erke, 2009)RLC Spillover Effect (Halo Effect) There is some but not much evidence that RLR camera s will not only deter motorists from violating a signal at intersections equipped with cameras, but will also modify driver behavior at other intersections. If c ameras do have an effect on driver behavior beyond those intersections where the camer as are used, then the other intersections in the area will likely also experien ce a decrease in angle crashes. This is a spillover effect or a halo effect. A study of an RLR camera program in Oxnard, Califor nia, found a decrease in crashes at intersections with cameras and intersect ions without cameras. The studyÂ’s authors attributed this reduction to spillover. (Re tting R. A., 2002). On the other hand, an evaluation of cameras in Sydney, Australia, did not find a significant reduction in RLRrelated crashes at intersections without cameras. T he authors concluded that spillover did not occur at non-camera intersections used as contr ol group intersections. (IMBERGER, 2003) A national study involving multiple jurisdictions h as yet to prove that this red light camera spillover effect does or does not occu r. Consequently, the studies suggested that agencies should consider the possibility of th is spillover in their evaluation of RLR cameras and modify their methodology or conclusions accordingly. Also, agencies

PAGE 73

54 (according to the author) may want to evaluate and quantify the spillover effect in addition to the effect at intersections equipped wi th cameras. (McGee & Eccles, 2006) The effectiveness of red light running cameras in r educing the number of drivers who run the red light in Clive, Iowa was evaluated. The number of red light running violations at camera-enforced intersection approach es were compared to violations at approaches at intersections where cameras were not used within the same metropolitan area using a cross-sectional analysis. A Poisson l ognormal regression was used to evaluate the effectiveness of the cameras in reduci ng violations. Results indicated that red light running cameras substantially reduced the num ber of violations at camera-enforced approaches as compared to control approaches. (Fitz simmons, Hallmark, Orellana, McDonald, & Matulac, 2009) In June 2010, Arlington County, Virginia, installed red light cameras at four heavily traveled signalized intersections. A study examined the effects of the camera enforcement on red light violations. Traffic was vi deotaped during the one month warning period and both one month and one year afte r ticketing began at 12 signalized intersections, including the four camera intersecti ons, four “spillover” intersections without cameras in Arlington County (two on the sam e travel corridors as the camera intersections and two on different travel corridors ), and four “control” intersections without cameras in adjacent Fairfax County. Rates o f red light violations per 10,000 vehicles were computed.

PAGE 74

55 (,;3 r!71r-4" $# Consistent with prior research, there were signific ant reductions in red light violations at camera-enforced intersections. These reductions were greater the more time that passed since the light turned red, when violat ions are more likely to result in crashes. Spillover benefits were observed only for nearby in tersections on the same travel corridor, and these were not always statistically s ignificant. At intersections on other travel corridors, red light running increased compa red with expected rates based on the control intersections. (See Table 9) The study concluded that this evaluation examined t he first year of Arlington CountyÂ’s red light camera program only, which was m odest in scope and without ongoing publicity. A larger, more widely publicized program likely is needed to achieve community-wide effects. (McCartt & Hu, 2013)

PAGE 75

56 +r@rr) 2r)rr) rrr2rr rrr")rAAr) 4,n52$67 nnnn !"#$ % $ & "$''$ n ! n ! n ! n ! n ! "#$%& $'(( rrnrr nn $()##*((nrnn rn +()##*((nrnrrrrrrnr r ,-$%&#(( nrnrr

PAGE 76

57 RLC Site Selections Although there are many studies that have investig ated the safety improvement of the RLC system, there are relatively very few studi es that have covered the RLC sites selection and where/ when to implement them especia lly when considering the cost associated with the system. A general study provides a tool for identifying an d priority-ranking problem intersections with respect to red light running wit hin the entire roadway network under the jurisdiction of a particular agency. The tool includes three steps as a guide to estimate the safety changes upon installation of a red light camera at a signalized intersection. These three steps are: empirical Ba yes method, collision prediction models, and collision modification factors (the assumption of negative binomial error distribution was used for developing the last step). (Hadayeghi, Malone, Suggett, & Reid, 2007) The city of Durham wished to explore the feasibili ty of implementing a red light camera program. Particularly, they wanted to ensure that the sites were selected in an objective and defensible manner based on sound traf fic engineering judgment. The study concluded that RLC sites selection criteria could b e based on two main elements: the overrepresentation of angle crashes and the higher than expected number of crashes. Additionally, the study used a more detailed field investigation to observe things like the signal timings, intersection layout, traffic signal type and placement, prevailing traffic patterns and operating speeds, and the suitability of each approach for a red light camera. Based on the review, a short list of candidate site s/approaches was developed. For approaches remaining on the short list, it was sugg ested that the occurrence of red light running be confirmed through a detailed violation r ecords, and rear end crashes be

PAGE 77

58 closely monitored in the post-implementation period (Suggett, Malone, & Borchuk, 2005) The Intersection Safety Camera Program (ISCP) in Br itish Columbia, Canada, has proved effective in reducing the frequency of crash es at locations where red light cameras have been deployed. Post-implementation evaluation s of ISCP conducted by the Insurance Corporation of British Columbia detected a 14% reduction in crashes resulting in injuries 18 months after the program was impleme nted. A follow-up study conducted 36 months after ISCP implementation examined the sa fety performance of ISCP and found that the rate of crashes resulting in injurie s was reduced by 6.4%. Given the ongoing and long-term success of ISCP at reducing c rashes, it was decided that the program should be expanded. To support ISCP expansi on, it was necessary to examine how the program had been implemented and to learn f rom the results of the previous program evaluations. A critical element of ISCP is the selection of sites to be targeted for deployment of intersection safety cameras. The site s selected should have a demonstrated safety problem, such as results from previous evalu ations of intersection safety camera after installation. In addition, sites should be se lected such that the life-cycle cost of deployment of the intersection safety camera will b e less than the safety benefits that will accrue from reduced numbers of crashes and the asso ciated costs. (de Leur & Milner, 2011) The twelfth offering of a Mentors Program at Texas A&M University on Advanced Surface Transportation Systems presented a document in 2002 by the Advanced Institute in Transportation Systems Operat ions and Management. One of the papers that was discussed and presented was about t he criterion of sites selection. The

PAGE 78

59 paper has an introduction that indicates the lack o f papers and studies related to RLC site selection and the actual need for a uniform set of criteria to aid traffic engineers and cities in the site selection process. From a survey that included most of US cities that have RLC installed in their transportation system, resul ts showed that the three most commonly used criteria are the history of red light crashes, red light citations, and engineering issues associated with intersection. The author of the paper suggested a set of major an d minor guidelines that should be used. The guidelines are developed using success ful experiences with similar situations. Guidelines are used when a policy would be too limiting or confining, or for situations that are highly variable. They allow car eful assessment of intersection conditions that are indicators for the need of traf fic control devices or engineering countermeasures. These guidelines were presented a s follows: Major guidelines 1. Accident History The use of accident statistics can be helpful in th is area, though the author disagrees with the total reliance on them alone. Accident statisti cs should be used to identify problem areas that need to be investigated further. 2. Red Light Citation History The evidence that there is a problem is a good indi cator that countermeasures need to be implemented. Usually the presence of these citation s is more of an indicator of allocation of available police resources and the relative safe ty of enforcing the law. Nevertheless, this is a red flag that should alert one to possibl e problem intersections though others may exist.

PAGE 79

60 3. Approach Speeds As the speed increases, the severity of resulting c rashes increase as well. There may be situations where there are high approach speeds and high violations, but few crashes. These areas need to be further evaluated and closel y monitored for possible development of crashes. Minor guidelines 1. Traffic and Pedestrian Volumes Generally higher traffic volumes relate to a greate r probability of violations and crashes. This is of particular concern when the cross-street traffic volumes are also high as well. While the crash of two vehicles can result in eithe r injury or death, the same is not true of a vehicle-pedestrian collision. Intuitively, the pe destrian is almost always killed or severely injured when a vehicle runs a red light an d collides with them. As a result, intersections with high pedestrian volumes and/ or traffic volumes need to be closely examined. 2. Intersection Degree of Saturation With higher degrees of saturations at intersections the headway gaps between vehicles are smaller. Consequentially, there is a greater probab ly of a vehicle running a red light whether intentionally or unintentionally. The diffe rence between these two intentions needs to be recognized and quantified. 3. Perceived Benefit to Cost As with all engineering countermeasures, the costs associated with the installation of a system should be evaluated. These costs do not only include the costs of construction and equipment, but also the beneficial costs received f rom the reduction in crashes due to red

PAGE 80

61 light violators. (Qu, 2003) From analysis of the data, there are intersections in Rhode Island where RLR is a problem. A model is developed to prioritize interse ctions based on a Composite Intersection Index (CII), where the highest score i ndicates the most problematic intersection. The CII is based on a comprehensive s et of variables including the following: (1) the entering average daily traffic ( ADT) (in 10,000s of vehicles) per number of lanes entering the intersection; (2) the rate of RLR violations occurring after 1 second; (3) the number of phases; and the (4) avera ge approach speed (based on approach speed limits). (Hunter, 2003)

PAGE 81

62 CHAPTER IV METHODOLOGY Introduction Although red light cameras are widely used to disco urage red light running, relatively few studies have been done on how to bes t deploy these cameras to maximize their efficacy. Due to their high price tags, many jurisdictions will purchase a relatively smaller number of cameras and rotate them among a l arger number of camera-ready intersections. This methodology chapter is divided into two sections. The first section is a brief description of the reasons behind choosing th ese cities and a representation of the data required to complete the study. The second sec tion covers the analytical-based methodology, which will be used to determine the nu mber of cameras needed to effectively enforce locations within a certain city limit. The study will use data from the cities of Colorado Springs, Fort Collins, and Denve r to be able to implement the methodology and derive the findings. Why These Locations as Case Studies? There are several reasons for choosing these locati ons, which I divided into general and specific reasons: 1) General Reasons There is an ongoing controversy in the state of Col orado regarding the effectiveness of red light cameras, which generated two bills to the State Senate during the years of 2012 and 2013. Areas like RLC is main objective, types of crashes it is causing, and the involvement in public privacy are some of concerns citizens hold on RLC, and therefore studies like this will clarify s ome argumentative points.

PAGE 82

63 The case studies represent three cities with differ ent characteristics, which is another concern associated with implementing RLC sy stems, and therefore results should indicate whether the system can or cannot be impact ed by different city characteristics. Additionally, it is obvious that such a topic is hi ghly based on the data collection stage (as mentioned earlier), which may require sev eral calls, field investigations, and physical visits to the data sources, besides the ne ed to physically commute to these cities several times to collect and update some of the col lected data. Therefore, data accessibility is another reason why these cities we re chosen. 2) Specific Reasons (effective and ineffective prog rams) In Colorado Springs, the RLC program started back i n 2010 and was shut down one year after that for ineffective results. It se ems obvious that there was something implemented or managed differently than in the Fort Collins and Denver cases that have been running their programs for more years (over 10 years already). Consequently, the study can make a very solid comparison between curr ent actual RLC locations and the ones suggested by the study. Finally, availability of data for a good length of time in the city of Colorado Springs, Fort Collins, and Denver is a plus for cho osing the cities as case studies, especially with RLC site selection studies that req uire at least a period of 3 years data to show meaningful results. Data Required and Field Investigation In order to implement the criterion of RLC sites se lection, specific data are required. Most of the data were obtained from the traffic engineering office, police

PAGE 83

64 department, and from some field investigation of ea ch city, while other types of data were not available and therefore not included in the ana lysis chapter. The following data were collected for 82 signalized intersections in Colorado Springs (from 2007-2009), 106 signalized intersecti ons in Fort Collins (from 2010-2012), and 309 signalized intersections in Denver (from 20 10-2012). (Note: all data are for a 3year period). Traffic volume/ approach/ intersection. Crash Types which were categorized into front to si de, rear-end crashes and other. Crash severity that is divided into fatal crashes, injury crashes, and property damage crashes. Approach or direction of the “At-fault” vehicle. Intersection characteristics. Overall final locations of RLC. Table 10 illustrates the type of data, its source, and time period where data was collected.

PAGE 84

65 +r *rrB0.11 Data Name Data Types Source Time Period Traffic Volume / Approach / Intersection EB, WB, NB SB City Traffic Engineering Office. 3 Years Crash Types Front to side, Rear-end City Police Dep t. 3 Years Crash Severity Fatal, Injury, PDO City Police Dept. 3 Years Vehicle Types Commercial, Pass, Trucks City Police Dept. 3 Years Crash History Date of Crashes City Police Dept. 3 Y ears Economic Evaluation Cost per Crash types, Cost of RLC. City Traffic Engineering Office, Insurance agencies Current Intersection Characteristics Intersection Layout, Approach Speed, City Traffic Engineering office and Field Investiga tion Current Social Structure, and Traffic Signal Timing Direction of At-Fault Vehicle EB, NB, WB, SB City P olice Dept. 3 Years RLC Selected Locations Overall RLC locations Field Investigation Current

PAGE 85

66 Methodology There are several possible criteria and procedures that could be used for RLC site selection, but the availability of data must be tak en into account in developing the criteria and procedures. In the study, there are two main p hases, each having specific criteria that could be followed in order to identify candidate RL C sites. All signalized intersections with available data in each city will be screened and tested using the criteria in Phase I. Before movin g to Phase II, candidate RLC sites from each test will be scaled by weighting factors deter mined by the city stakeholders and decision makers to make the final ranking list. Fi nally, qualified intersections will be evaluated by conducting a comprehensive field inves tigation in Phase II. The following provides a detailed description of th e two phases and tests under each of them. Phase I “Includes Four Criteria” 1Criterion of Crashes Severity Although crash frequency has often been the primary consideration in the implementation of RLC, crashes differ in severity. There are several levels of crash severity which should be considered when choosing R LC location. From the literature review, crash severity is mostly divided into three different levels: crashes resulting in fatalities, injuries, and property damage only. Eac h of these crash severities was given a relative weight representing its impact level. 100 for crashes resulting in fatalities. 10 for crashes resulting in injuries.

PAGE 86

1 for crashes resulting in property damage only. all intersections within the city limit by dividing the total number of categorized by severity level the following formula: NCSL 9Brnr< 1r2. Where: N-CSL = Normalized Crashes F = Crashes Resulting in Fatalities. I = Crashes Resulting in Injuries. PD = Crashes Resulting in Property Damage. TC = Total Crashes. It is important to note that high crash severity but few severity level without normalizing the equation by following equ ation, using the following CSL 9Br 1r2. 1 for crashes resulting in property damage only. Crash Severity could be ranked for all intersections within the city limit by dividing the total number of crashes categorized by severity level and then normalized by the total number of CSL = 1r2. !1r)+r9@" # Crashes Severity Level Fatalities. Injuries. Property Damage. It is important to note that the crash severity level equation may identify sites with severity but few crashes. Thus, it is also preferred to calculate the severity level without normalizing the equation by the total number of crashes ation, using the following equation: CSL = 1r2. !1r)+r9@" # 67 Severity could be ranked for crashes that are crashes using !1r)+r9@" # equation may identify sites with it is also preferred to calculate the crash crashes using the !1r)+r9@" #

PAGE 87

68 The higher crash severity level is for a specific i ntersection, the higher that intersection is ranked among the list of the cityÂ’s intersections. (This applies to both equations). Where: CSL = Crash Severity Level 2Potential For Improvement (PFI) The term potential for improvement (PFI) is used as a measure for identification of the criteria crash rate and crash frequency for all locations included in a city, based on a certain parameter or benchmark. Figure (19) shows a graphical illustration of the concept of PFI, and why some values are below the l ine (negative). In the graphic, sites 1 and 2 have a positive value for the PFI, as they ar e above the blue line. Conversely, site 3 has a negative value for the PFI, as it is below th e line. In each of the case studies, potential for improvem ent will be measured in crash rate and crash frequency. Since average rate was us ed as a parameter, the locations will be divided into positive and negative values. Negat ive values mean that there is no potential for improvement. In fact, these locations are performing better than normal / average. The negative crash rate (or frequency) val ue means that this is the number of crashes below an average crash rate (or frequency) level and since it is below, there is no PFI. In the other hand, the positive crash rate an d frequency value means that this is the number of crashes above an average crash rate (freq uency) level and since it is above, there is a PFI. Normally, collision prediction model (CPM) is used as a parameter to measure potential for improvement, however, in this study t he normal average rate will be used instead. (See equation 5)

PAGE 88

,r)r From three years crashes thankfully provided, potential for improvement follows: PFI in Crash Rate It is a model that has become a standard method for measure performance and es pecially for RLC candidate site calculated by subtracting of each site. As shown in equation 3, this criterio n crash rate for each intersection expressed as follows: PFI in Annual Crash ,r)r ) crashes and volume per approach data that each of the cities potential for improvement can be obtained from the Rate (Crash/Movement) It is a model that has become a standard method for measure ment of road safety pecially for RLC candidate site s selection. This criterion calculated by subtracting estimated crash rate (the parameter) from annual crash rate of each site. As shown in equation 3, this criterio n will provide PFI in for each intersection based on three years data and it can be PFI in Crash Rate (Crash / Movement) per intersection Crash Rate per intersection – Estimated Crash Rate per 9Br$>,rr!;.+" $# 69 each of the cities has two criteria as of road safety This criterion can be annual crash rate in relation to and it can be mathematically = per intersection

PAGE 89

Where Annual Crash calculated using the following equations: Annual Crash where, = Total Crashes per intersection. = Annual Average Daily Traffic per intersection. And, Estimated Where: PFI in Crash F requency ( While PFI in crash rate of each intersection is RLC candidate sites, PFI in when making the candidate list Rate per intersection and Estimated Crash Rate per intersection calculated using the following equations: Crash Rate per intersection/million vehicles = 9Br%;rrr = Total Crashes per intersection. = Annual Average Daily Traffic per intersection. Estimated Crash Rate per intersection/million vehicles/ year = 9Br&9rrr Obtained by performing regression analysis. requency ( Crash/ Year) rate of each intersection is an important step towards selecting PFI in crash frequency is another important criterion when making the candidate list s. 70 intersection are Obtained by performing regression analysis. important step towards selecting criterion to consider

PAGE 90

Once PFI in collision frequency can also be cal estimated annual crashes 9Br !6-$nn89: As a result, both PFI in two lists of candidate RLC sites that the higher crash rate among the list of the candidate candidate list. 3Criterion of Crashes One of the major argument points with crashes that usually decrease or increase due to the implem entation of the RLC. As indicated in the literature review chapter, the front to side type of crashes wherever an RLC is implemented. This is an importa phase one especially with within the city limit, crash high proportion of rearend collision rate is calculated for each intersection, PFI in frequency can also be cal culated by multiplying annual crashes per intersection by estimated annual crashes It can be expressed as follows: PFI in Crash Frequency (Crash/ Year) = 9Br '>,1rB2!;.+" $# PFI in crash rate and PFI in crash frequency will make another RLC sites based on the potential for improvement criterion. rate for a specific intersection, the higher that inters ection is ranked candidate intersections. Similarly, this applies to the Crashes Types major argument points with regard to RLC programs is the type of that usually decrease or increase due to the implem entation of the RLC. As indicated in the literature review chapter, most studies have shown an overall decrease on crashes and contrarily an increase on rearend type wherever an RLC is implemented. This is an importa nt criterion that could be used as part of RLC sites selection especially with the availability of such data. From all signalized intersections crash types are available and the idea here is to avoid l ocations with end crashes and targ et those with high proportion front to side 71 PFI in crash per intersection by frequency will make another criterion. Note for a specific intersection, the higher that inters ection is ranked the crash frequency regard to RLC programs is the type of that usually decrease or increase due to the implem entation of the RLC. As studies have shown an overall decrease on end type crashes sites selection signalized intersections types are available and the idea here is to avoid l ocations with et those with high proportion front to side

PAGE 91

type of crashes. The following formula is used to calculate the fron t to side type of crashes rate: where 4Weighting Scale for It is very important to note that top 10 lists of candidate intersections projects in the same field, this is This step requires a subjective judgment on the par t of the group making the evaluation which is in this case would the city eng ineering office represented by their city engineers (See tables 11,12,and 13) importance and the n by using a formula of proportionality to obtain r elative ratio weights (Nicholas J. Garber, 2015) 9Br(>) r2rrA Where The following formula is used to calculate the fron t to side type of 1rrr2)r!;.+" $# for RLC Intersection Candidates It is very important to note that all criteria from phase I could result in different of candidate intersections and according to many previous studies and projects in the same field, this is a normal scenario. This step requires a subjective judgment on the par t of the group making the evaluation which is in this case would the city eng ineering office represented by their city (See tables 11,12,and 13) Next, each intersection will be ranked in order o n by using a formula of proportionality to obtain r elative ratio weights (Nicholas J. Garber, 2015) r2rrA !nrC8r" &# 72 The following formula is used to calculate the fron t to side type of could result in different and according to many previous studies and This step requires a subjective judgment on the par t of the group making the evaluation which is in this case would the city eng ineering office represented by their city each intersection will be ranked in order o f n by using a formula of proportionality to obtain r elative ratio weights !nrC8r" &#

PAGE 92

73 Next, weighted scores from all top 10 locations fro m each of the criteria will be added up to make the final top 10 that combined all the criteria in phase I. The final top 10 should be processed and moved on t owards Phase II for further analysis to conclude the final RLC candidate list f or each city. +r?)r> r,1r)!1r)+r 9@" # ()"* +'#.$((!*& $((!*&n $((/n $((,0%&n $((1&) +r?)r> r,1!1+r9 @" # ()"* +'#.$((!*&n $((!*& $((/ $((,0%& $((1&) +r$?)r> r,*!*+r9@ # ()"* +'#.$((!*& $((!*& $((/ $((,0%&rn $((1&)

PAGE 93

Phase II “Includes Seven 1Fluctuation of Crashes For the final top 10 RLC candidate can be used to see if crash this criterion over the final are abnormally fluctuated and excessive variability This can be calculated using the coefficie 9Br Where: S = T he standard deviation. = Sample mean (Annual collision). As known, the sample mean Where: x = Sample data (Annual collision). n = Sample size (Number of years of data). And, the standard deviation Seven Criteria” Crashes 10 RLC candidate locations some traditional statistical measures crash frequencies are historically fluctuated or stable this criterion over the final top 10 list from phase I will help eliminate sites where data are abnormally fluctuated and excessive variability in crashes and violations are can be calculated using the coefficie nt of variation (V) formula: 100 rr !.-7" # he standard deviation. mean (Annual collision). sample mean () can be calculated by using the following formula: x = 9Brr)r = Sample data (Annual collision). n = Sample size (Number of years of data). deviation (s) can be calculated using the following formula: 74 some traditional statistical measures fluctuated or stable Performing will help eliminate sites where data in crashes and violations are found. !.-7" # by using the following formula: can be calculated using the following formula:

PAGE 94

9Br 2Type of Vehicles From the final top 10 candidate RLC sites, type of vehicles atypically involved in crashes installation of the RLC. RLC benefits might be limited at these locations be cause they license plates of tractor unit of multiunit vehicl license plates of very large trucks. photographing multiunit vehicles are recommended at these locations. Type of vehicles can be screened and analyzed using the shown in the following formula: 9Br +2) Where vehicle type in reference population and s = rr 2rrrrr From the final top 10 candidate RLC sites, type of vehicles that have been crashes or violations can be screened to consider during the RLC benefits might be limited at these locations be cause they cannot of tractor unit of multiunit vehicl es or are not capable of photographing license plates of very large trucks. As a result, cameras with features that are capable of photographing multiunit vehicles are recommended at these locations. Type of vehicles can be screened and analyzed using the chisquare test shown in the following formula: +2) 2rr1Br !.-7" # where is the proportion of a vehicle type in reference population and f is the total vehicle types at a sit e 75 2rrrrr that have been consider during the cannot photograph es or are not capable of photographing As a result, cameras with features that are capable of square test () !.-7" # is the proportion of a

PAGE 95

76 3Economic Evaluation As discussed in the literature review chapter, a bi g chunk of the argument related to the RLC systems is based on the opinion that it is more about increasing citiesÂ’ revenue rather than improving the safety level. Th erefore, considering this element while selecting RLC locations is a plus toward the succes s of the program in any community. An economic evaluation of each of the final top 10 RLC candidates will evaluate these locations after installation in an annual bas is. Cost of RLC, revenue generated from the system, and safety benefits are the three eleme nts involved to execute this criterion. Total Cost of RLC per year + Total Revenue of RLC per year < Total Safety Benefits per year 9Br0.19rr!;.+" $ # Where: Total Cost of RLC = Overall cost of RLC installatio n, operation, maintenance in a given location / Year. (This can be obtained from RLC providers). Total Revenue of RLC = Overall revenue generated by RLC in a given location / Year. (Total of RLC tickets value) Ave Safety Benefits = Average cost of a crash producing PDO and injuries. (Determined by major auto insurance companies) According to the city of Fort Collins around 80-100 tickets/month are generated by RLC which means a revenue of $120,000/year. If quick as sumptions are made, a RLC costs $40,000 which means $480,000/year, that includes th e process of installation, operation, and maintenance. This will total up to $600,000 (if we assume city and operator are lobbying together as claimed by parties standing ag ainst the system).

PAGE 96

77 A study published by the National safety council es timated motor-vehicle crashes as (National Safety Council, 2000): Death = $1M Injuries= $35,500 PDO= $6,500 The estimates were based on wage and productivity l osses, medical expenses, administrative expenses, motor vehicle damaged, and employer costs. If estimates are used as average safety estimate, then there is no d oubt that average safety benefits will exceed total cost and revenue of RLC, because if RL C prevents 2 crashes resulting in injuries/ month = $71,000 (which is more than $850, 000 per year. Injuries level of severity alone will exceed RLC cost and revenue com bined. By the end of each year, locations where safety ben efits exceed total RLC cost and revenue are proofing their economic effectivene ss and should remain under operation. In contrast, locations where total RLC cost and rev enue are more than its safety benefits are not considered economically effective and shoul d be eliminated. 4Intersection Characteristics During sites selection criterion “Phase II”, more v ariables are used to eliminate these qualified intersections even further. One of the steps is to visit the sites and evaluate their characteristics and suitability for RLC based on (PASS and FAIL) scoring system. Locations with high PASS score indicate a n eed for further action such as a RLC system, while locations with low PASS score indicat e a need to fix these characteristics before installing a RLC.

PAGE 97

78 There are four characteristics that should be inclu ded as part of the field evaluation, these are: Intersection Layout: It includes four guidelines: Lane width: Lane widths are commonly narrower on low volume roa ds and wider on higher volume roads. According to ITE, 12foot lanes are desirable, al though widths as narrow as 10 feet have been used in sever ely constrained situations unless large trucks and buses are using the lane. T herefore, the range from 10-12 will be sat as a parameter for all final top 10 sit es. Lighting: Statistics indicate that the non-daylight accident rate is higher than that during daylight hours. This fact, to a large degree may be attributed to impaired visibility. In urban and suburban areas where there are concentrations of pedestrians and roadside and intersectional interfe rences, fixed-source lighting tends to reduce crashes (American Association of St ate Highway and Transportation Officials (AASHTO), 2001). Clear Signage: Roadway signs in the United States i ncreasingly use symbols rather than words to convey their message. Symbols provide instant communication with roadway users, overcome language barriers, and are becoming standard for traffic control devices throu ghout the world. Familiarity with symbols on traffic signs is important for ever y road user in order to maintain the safety and efficiency of our transportation fac ilities. Proper and clear signs associated with the nature of the intersection desi gn is a must for each of the final top 10 sites to pass the field investigation.

PAGE 98

79 Channelization: One of the most effective and efficient methods of controlling the traffic on a highway is the adoption of high in tersection geometric design standards. Channelization is an integral part of at grade intersections and is used to separate turning movements from through movement s where this is considered advisable and hence helps reduce the intensity and frequency of loss of life and property due to accidents to a large extent. Proper Channelization increases capacity, improves safety, provides maximum conveni ence, and instills driver confidence. Improper Channelization has the opposit e effect and may be worse than none at all. Over Channelization should be avo ided because it could create confusion and worsen operations. Channelization is defined as the separation or regulation of conflicting traffic movements into de lineated paths of travel by traffic islands or pavement marking to facilitate t he safe and orderly movements of vehicles, bicycles, and pedestrians. All guidelines should meet CDOT requirements for si gnalized intersections ( Colorado Department of Transportation, 2000). Approach Speed: Speed limits are set by each state or territory. S peed limits are always posted in increments of five miles per hour. Some states have lower limits for trucks and at night, and occasionally there are minimum speed limits. Most speed limits are set by state or local statute, alt hough each state allows various agencies to set a different, generally lower, limit However in this study, the pass parameter is that to have the average speed limit l ess or equal to the posted speed limit in order to pass the field investigation.

PAGE 99

80 Yellow Phase Change Interval: As discussed earlier, excessively short or long yellow change intervals may encourage driver disres pect and unsafe operating practices. Therefore, it is important to confirm th at all top 10 RLC candidates are within the suggested ITE yellow interval values (Se e table 14). +r%>rr2Arrr ) &$+$," #"% -"."/'(0 #% n n rn rr nr nr Social Structure: Another worthwhile evaluation characteristic here is to make sure the area where RLC will be installed is suitab le and has no excessive vandalism that would target the camera. This can b e measured by reviewing criminal history and income level of the area, whic h is normally provided by the city police department.

PAGE 100

81 ;rr<0.1>/;
PAGE 101

82 6Approaches Determination From the literature review, implementing RLC does n ot apply to all approaches because of many reasons like the cost of the equipm ent itself, which exceeds $40,000 to lease per month. Therefore, it is highly important to define the approach that the vehicle at fault was using as part of the sites selection p rocesses. This step could be processed after specifying the intersections from Phase I and by eliminating the number of intersections to approaches only. (See Table 16) +r'+rDr rErr))r!;.+" $# +%'2%#*#(3)) Intersection # Intersection EB WB NB SB TC 1 Name 7Red Light Camera Locations As noted in the literature review chapter, placing RLCs close to each other or distributing them along the same corridor might lim it their safety impact. It is also important to place the cameras in a manner where re sidents of the city feel the equity and do not have the feeling that they are targeted from other parts of the city. Therefore, two major factors can be the guidance in this regard; those are distance and direction. When determining RLC final location s, there should not be any cameras that are located in the same traffic travel directi on unless it is located in a distance of 3 miles or more (ALTurki, 2013). Table 17 presents a conclusion for all criteria, fo rmulas, and expected findings from phases I and II of the methodology chapter.

PAGE 102

83 +r(rrr & ()" (+! #)"1$% (+! &2 "" 3". 4 4 ,5#$(/6$(3*'74 33$ / $##( !*&## 5(( '#.2& #%'2 (( / $##( !*&## 5(( / ((

PAGE 103

84 & ()" ( 4! 5* 3..2 3". "%# $( / )( 8 9('"%#$(( )( ,5#$(,0%&6$(37 = 6'2&(#(7 33$ )( ,5 2( ##( / (( ,52( ##( 0%& 1: ( ,#)n #( !(;

PAGE 104

85 & ()" ( 4!2 6" 6. Total Cost 0 #% 3". nn Total Cost of RLC per year + Total Revenue of RLC per year < Total Safety Benefits per year 5(<&% "))!) 33$ -((* *2#& (( *#( $'( )) '%#% *#( Safety $')( 2;(& 2( /<$( ))(

PAGE 105

86 & ()" 3". 33$ ( ##;( 5*# !#!%% 2( (( 7" =>",%#*#(? ")) > 5$,(" =+('*#%#((' '#(;& '/<$? 5((& ') 0%&#*# ((

PAGE 106

87 Expected Findings This part of the study will provide an analytical-b ased methodology for RLC sites selection that could be used by any city that decid es to launch a RLC program within a certain jurisdiction. This analytical-based method ology was also supported by a field investigation that ensures comprehensive analysis a nd more accurate final RLC candidates. The advantage of this methodology is the fact that it is based on two main phases that require accessible and available data in many cities. This allows any city to analyze and implement the program faster than similar proje cts normally take while maintaining comprehensiveness, as it was shown in table 17.

PAGE 107

In this chapter, three of the RLC site selection Section two presents the field investigation II); and finally the final findi presented from all three cases. Section I: Analyses of RLC Sites Phase I: From phase I of first criterion. The following equation based on crash severity level using the following e quation: CSL Tables 18 and 19 show the top 10 RLC candidate locations based on (CSL) and the Normalized the city. CHAPTER V ANALYSES AND FINDINGS three sections are presented; section one shows complete analyses RLC site selection from Colorado Springs, Fort Collins, and Denver the field investigation to eliminate intersection candidates (Phase the final findi ngs, conclusion, and recommendations of the study cases. RLC Sites Selection for Colorado Springs Colorado Springs of the methodology chapter, test of crash severity comes as The following equation will be used to rank 82 signalized intersections based on crash severity level using the following e quation: CSL = the top 10 RLC candidate locations based on crash and the Normalized -crash Severity Level (N-CSL) from all 82 locations 88 complete analyses from Colorado Springs, Fort Collins, and Denver (Phase I). to eliminate intersection candidates (Phase of the study are severity comes as the signalized intersections crash Severity Level from all 82 locations studied in

PAGE 108

89 +r0r) 0.1rr1r )rr9@<3A5$B9/!>/nn >C@<5+@<3+C+5D+@9+D,1:9ED>!/>rnn r>/9++"+/>3!"$">9@9"Arnnr "5/D/1/>3!"$">9@C@<5+@<3+D9/!@31C11@ +r0r) 0.1rr1r )rr2 580)"(+,*$ "5/D/1/>3!"$">9@9@<3A5$B9/!>/n >C@<5+@<3+C+5D+@C@<5+@<3+D9/!@9@9+D,1:9ED>!/>rnnn "5G9<"+>/>3+"$">9@/9++"+/>3!"$">9@

PAGE 109

90 Tables 20 and 21 show the top 10 RLC candidate loca tions in the city of Colorado Springs based on their potential for improvement in relation to crash rate and crash frequency. +r 0r) 0.1rr1r )r)r) rrr &30 58 0)"(5 *$ @/5"/E"193+D9/!@3+D9/!@9+/>35nnr r"5/D/1/>3!"$">9@/>3+"$">9@3!D9/!@ +r0r) 0.1rr1r )r)r) rrB2 &30 58 0)"(39 *$ 9DD>9+/>35nn @"/+9!/>3+D9/!@3!"$">9@9+D,1:9ED>!/>n +D9/!@<3!191!D+:5<9@

PAGE 110

91 1r))rr rrrrr2r After running a regression analysis with an R-squar e = .47 and p-value that is well below 0.05, A graph was drawn as shown in figure 21 which reports how much the data of 3 years of crashes varies around the fitted blue curv e. Table 22 shows top 10 RLC candidates from all 82 si gnalized intersections studied in the city of Colorado Springs that combin e both of a high proportion of front to side crashes and low proportion of rear end crashes +r0r) 0.1rr1r )rr2) 5803+$5*$ 9<"119"A3+"$">9@9@<3"<9/"/B@/nnn +D9/!@<3D<>/"+$:/>nnn /5+19/!3!"/B!5>9>/nnn >C@<5+@<3+C+5D+@
PAGE 111

92 Before moving to phase II of the red light camera s ite selection in Colorado Springs and after calculating the top 10 candidate intersections from each of the criteria show earlier, it is essential to present the final top 10 RLC candidate intersections from all criteria combined. Using the weighting formula described in the methodology chapter, the final top 10 candidates (highlighted in grey) c ome as shown in table 23:

PAGE 112

93 +r$r) 0.1rr1r) rr)r, 580:((+,)(+,&30(3 9&30(5(4!4 @/5"/E"193+D9/!@< n nn nn nnn nn "5/D/1/>3!"$">9@< nnn nn n nnnrn 9DD>9+/>35nnnnn nnn nnnrn r 9<"119"A3+"$">9@< nnnn nnnn nnnr @"/+9!/>3+D9/!@9@<3A5$B9/!>/ n nnnnnnnnrnn +D9/!@<3!191!D+:5</>3+"$">9@< nnrnnnn nnnnnn n >C@<5+@<3+C+5D+@< n nnrnnnnnn nnnn +$"/9,/99$/3+D9/!@< nnnnn nnnnnnnn 53E"/>9+D,1:9ED>!/>nnnn nnnnrnnn >C@<5+@<3+D9/!@< nn nnnnrnnrnnnr r"C!15+@3!D9/!@9@<3"<9/"/B@31C11@/"+$:/>nnrnnrnnnnn nnn >/9++"+/>3!"$">9@< nn nnnnnnnnrnr /5+19/!3!"/B!5>9>/nnnnnnnnnnn nnn +$:9!1+C1!13E"/>9+D,1:9ED>!/> nnn nnnnrnnnnrrn

PAGE 113

94 580:((+,)(+,&30(3 9&30(5(4!4 r9C5+1":!13+$"!$">9"Annn nnrnnnnnnnrrn 53C5+1":!1nnnnnnnnn nnnnn 53!$5/$<9>/nnnnnnnnn nnnn +"$">9@<3!:/5>9//>nnr nnrnnnnnnnnr ;

PAGE 114

Section II: Further Analysis and Colorado Springs >r,, #n $nn. r$> Fluctuation of Not applicable due to the unavailability of the typ e of vehicles data. Type of Vehicles Not applicable due to the unavailability of the typ e 99rr Total Cost of RLC per year Not applicable due to the unavailability of the ave rage safety benefits data from the state of Colorado. Further Analysis and Field Investigation of Top 10 RLC C Colorado Springs 4#n+@nn 76n r$> Fluctuation of Crashes 100 x = s = Not applicable due to the unavailability of the typ e of vehicles data. Type of Vehicles Not applicable due to the unavailability of the typ e of vehicles data. 99rr per year + Total Revenue of RLC per year < Total Safety Benefits Not applicable due to the unavailability of the ave rage safety benefits data from the state 95 C andidates in 76n Total Safety Benefits per year Not applicable due to the unavailability of the ave rage safety benefits data from the state

PAGE 115

96 ,1rr +r%,rr1r )) 0.1rr Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) BRIARGATE PY/N POWERS BL Intersection Layout Lane Width P 6 Lightening P Channelization F Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P AIRPORT RD/S ACADEMY BL Intersection Layout Lane Width F 3 Lightening F Channelization F Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level F E WOODMEN RD/I-25 Intersection Layout Lane Width P 6 Lightening P Channelization F Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P E PLATTE AV/N ACADEMY BL Intersection Layout Lane Width F 4 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P BARNES RD/N POWERS BL Intersection Layout Lane Width P 5 Lightening P Channelization P

PAGE 116

97Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P E PLATTE AV/N UNION BL Intersection Layout Lane Width P 4 Lightening P Channelization F Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P N ACADEMY BL/VICKERS DR Intersection Layout Lane Width P 6 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P N POWERS BL/STETSON HILLS BL Intersection Layout Lane Width P 5 Lightening P Channelization F Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P MAIZELAND RD/N ACADEMY BL Intersection Layout Lane Width F 4 Lightening F Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P DUBLIN BL/N UNION BL Intersection Layout Lane Width P 4 Lightening F Channelization P

PAGE 117

98Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level F :rr>2-n>A:!87r)#

PAGE 118

99 $;)0-;r2:!87r)# %9?06,&!87r)#

PAGE 119

100 &9>r;-n;r2:!87r)# ':r0-n>A:

PAGE 120

101 (9>r;-n: N Academy Blvd & Vickers Dr (Google Maps)

PAGE 121

102 n>A:-4: $ 7r
PAGE 122

103 $*:-n:!87r)# As a conclusion of the intersection characteristics field investigation presented by table 24, 3 out of 10 intersections that have the h ighest score will be qualified to get RLC installed since they passed most of the intersectio n characteristics, but still have red light related crashes. Those intersections are: 1) E Woodmen Rd & I-25. 2) Briargate Py & N Powers Blvd. 3) N Academy Blvd & Vickers Dr.

PAGE 123

104 Approach Determination. +r&nrr)r))r! 1r)# )." 22.: 0; 0 6< *< )< +< 4( E Woodmen Rd/I-25 r Briargate PY/N Powers Blvd r nn N Academy Blvd/Vickers Dr Considering the fact that RLC is normally installed at one approach of the intersection, it was recommended that RLC should be installed at the northbound approach of E Woodmen Rd/I-25, the southbound appro ach of Briargate PY/N Powers Blvd, the northbound approach of N Academy Blvd/Vic kers Dr. This was determined given the history of at fault vehicles crashes per approach of each of the intersections for the period of three years. (Table 25) RLC Locations Below is a map with final RLC locations noting that they cannot be located within 3 miles of each other unless they are located in di fferent directions. The intersection of Briargate PY/N Powers Blvd is located within 3 mile s of E Woodmen Rd/I-25. However, their RLCsÂ’ are recommended in different directions therefore, they will be installed in both locations. Additional RLC is recommended at t he northbound of N Academy Blvd/Vickers Dr. Final RLC locations in the city of Colorado Springs are shown in the following map.

PAGE 124

105 $r0.1r!1r)#

PAGE 125

Section I: Analyse s of RLC Sites Phase I: From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 106 signalized intersections based on crash severity level using the following equatio n: CSL Tables 26 and 27 show the top 10 RLC candidate locations based on Level (CSL) and the Normalized studied in the city. s of RLC Sites Selection for Fort Collins. Fort Collins From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 106 signalized intersections on crash severity level using the following equatio n: CSL = show the top 10 RLC candidate locations based on Crash Level (CSL) and the Normalized -Crash Severity Level (NCSL) from all 106 From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 106 signalized intersections Crash Severity CSL) from all 106 locations

PAGE 126

107 +r'0r) 0.1rr1 rr!!1/"5+1/99Fn r!:59<>!!1/D<<"+>DD/9Fn !:59<>!!1E9Fn D/1:5+E1D+>/"B9Fn 1/">515D+:D/!91DD1:/>n !:59<>!!1!"<"/:"/D+/>nn ;<4=7rn n"> +r(0r) 0.1rr1 rr2 58 0)"(+, *$ $D<<9E9"AD+/D9Fr n 15@9/<5+9/>:D/!91DD1:/>F n $D<<9E9"A:D/!91DD1:/>Fn nr r<9":"/D+/>F nr $D<<9E9"A/D!9$1/>Frr n @D"/>"/:"/D+/>r n !:59<>!!1!!1/D!9$1/>Frr n $D<<9E9"A >/"B9/>F rn nn n 15@9/<5+9/> :"/D+/>F r nn ;

PAGE 127

108 Tables 28 and 29 show the top 10 RLC candidate loca tions in the city of Fort Collins based on their potential for improvement in relation to crash rate and crash frequency. +r0r) 0.1rr1 r)r) rrr &30 58 0)"(5 *$ G59E<9//D$B$/99BFnn $D<<9E9"AD+/D9nnn !:59<>!!1C<@9//!1nn r$D<<9E9"A1/5<@/>nn <9":"/D+/>nn 15@9/<5+9/>:D/!91DD1:/>nn 15@9/<5+9/>>/"B9/>nnn !:59<>!!19<5G"@91:!1nnr !:59<>!!1Fnn ;<4=7rn n(> +r0r) 0.1rr1 r)r) rrB2 &30 58 0)"(39 *$ $D<<9E9"AD+/D920.0 0.250 <9":"/D+/>18.0 0.225 15@9/<5+9/>:D/!91DD1:/>15.8 0.198 r$D<<9E9"A1/5<@/>14.6 0.183 $D<<9E9"A:D/!91DD1:/>F13.9 0.174 15@9/<5+9/>>/"B9/>12.5 0.157 !:59<>!!19<5G"@91:!112.2 0.152 !:59<>!!1C<@9//!112.2 0.152 $D<<9E9"AC<@9//!1F10.5 0.131 n!:59<>!!1

PAGE 128

109 $$1)rr rrrrr2r After running a regression analysis with an R-squar e = .60 and p-value that is below 0.05, A graph was drawn as shown in figure 32 which repor ts how much the data of 3 years of crashes varies around the fitted blue curve. Table 30 shows top 10 RLC candidates from all 106 s ignalized intersections studied in the city of Fort Collins that combine bo th of a high proportion of front to side crashes and low proportion of rear end crashes. +r$ 0r) 0.1rr1 rr2) 58 03+$5*$ $D<<9E9"A1/5<@/>nnnn !:59<>!!1/D!9$1/>nnnn 15@9/<5+9/>:D/!91DD1:/>nnnr r$D<<9E9"A:D/!91DD1:/>nnnr $D<<9E9"AD+/D9nnnr $D<<9E9"A@D"/>">/"B9/>nrnnrn $D<<9E9"A!"</"B9/>nnn

PAGE 129

110 Before moving to phase II of the red light camera s ite selection in Fort Collins and after calculating the top 10 candidate intersec tions from each of the criteria show earlier, it is essential to present the final top 1 0 RLC candidate intersections from all criteria combined. Using the weighting formula desc ribed in the methodology chapter, the final top 10 candidates (highlighted in grey) c ome as shown in table 31. +r$r) 0.1rr1 rr)r, Rank Intersection / Criteria CSL N-CSL PFI-Crash Freq PF I-Crash Rate Crash Types Total $##H nn nn nn nn nnr nr 1'2#H:(/ nrnn n n nnr n <'&H:'&/ nrnn n nnn n r $##H12#& nnnn n n nnn nr $##H:(/ nrnnr nrnn nnr nn !#(H#%' nn n nn nnn n 1'2#H>/ nnnn n nn nnrn nr !#(H%#2&! nnrnn n nnn n !#(!H9#.2! nnnn n nrnn nr n G#H/$ nnnnnnn nnnnn nr $##"*H%#2&! nnnnr nnnnn nr @;#H:'& n nnnnnnnn nr $2H:'&/ nnnnn nnnn nrn r<'&H>/ nnrnnrnnn nn nr !#H() nnnnnnnn nnn n !#(H!;##; nn nnnnnnrnnnr nr $##H()/ nnnnnnnnnnn n $&)H9#.2 nn nnnnnnnnnn nn <'&H!% nnn nnnnnnnnnn nn n$##H@;# nnnnnnnnrr nnr nn $##H!;##; nnrnnrnnnnn nnrn nn $##H1%' nnnnrnnnnr nnrn nn H> nn nn nnnnrnnn nn r!#(H/ nnr nnnrnnnnn nn 1'2#/H:'&/ nnrnnrnnnnnn nn 1H:%( nnn nn nnrnnnnn nnn <'&H/2( nnn nnn nnnrrnnnn nnn !#H/## nn nnnnnnnnn nnr $##H> nnnnrnrnrnnr nn ;

PAGE 130

AA< #$-nn#Anr Phase II: From table 30 (Final top 10 RLC candidat used to analyze these intersections further. Fluctuation of Not applicable due to the unavailability of the typ e of vehicles data. Type of Vehicles Not applicable due to the unavailability 99rr Total Cost of RLC per year Not applicable due to the unavailability of the ave rage safety benefits data from the state of Colorado. #$-nn#Anr 9+@ nn# Fort Collins From table 30 (Final top 10 RLC candidat e intersections), six criteria to analyze these intersections further. Fluctuation of Crashes 100 x = s = Not applicable due to the unavailability of the typ e of vehicles data. Type of Vehicles Not applicable due to the unavailability of the type of vehicles data. 99rr per year + Total Revenue of RLC per year < Total Safety Benefits Not applicable due to the unavailability of the ave rage safety benefits data from the state 111 nn# e intersections), six criteria will be Total Safety Benefits per year Not applicable due to the unavailability of the ave rage safety benefits data from the state

PAGE 131

112 ,1rr +r$,rrr!1 # Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) College Ave & Monroe Intersection Layout Lane Width P 6 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P Timberline & Horsetooth Rd Intersection Layout Lane Width P 7 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P Lemay & Harmony Rd Intersection Layout Lane Width P 5 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level F College & Tribly Intersection Layout Lane Width P 6 Lightening P Channelization P Signage F Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P College Ave & Horsetooth Rd Intersection Layout Lane Width F 4 Lightening P Channelization F Signage P Yellow Change Meet ITE guidelines P

PAGE 132

113Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Interval Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level F Shields St & Plum Intersection Layout Lane Width F 5 Lightening F Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P Timberline & Drake Rd Intersection Layout Lane Width P 6 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P Shields St & Mulberry St Intersection Layout Lane Width F 4 Lightening P Channelization F Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level F Shields St & Elizabeth St Intersection Layout Lane Width P 5 Lightening F Channelization P Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P Ziegler & Rock Creek Intersection Layout Lane Width P 5 Lightening P Channelization P Signage P

PAGE 133

114Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P $%1;-7!87r)#

PAGE 134

115 $& Timberline Rd & Horsetooth Rd. (Google Maps) $' Lemay & Harmony. (Google Maps)

PAGE 135

116 $( College Ave & Tribly Rd. (Google Maps) Figure 38 College Ave & Horsetooth Rd. (Google Maps)

PAGE 136

117 $ S Shields St & W Plum St. (Google Maps) % Timberline Rd & Drake Rd. (Google Maps)

PAGE 137

118 % Shields St & Mulberry St. (Google Maps) Figure 42 Shields St & Elizabeth St. (Google Maps)

PAGE 138

119 %$ Ziegler Rd & Rock Creek Dr. (Google Maps) As a conclusion of the intersection characteristics field investigation (See table 32), 4 out of 10 intersections that have the highes t score will be qualified to get RLC installed since they passed most of the intersectio n characteristics, but still have red light related crashes. Those intersections are: 1) S College Ave & W Monroe Dr. 2) S Timberline Rd & E Horsetooth Rd. 3) S Timberline Rd & E Drake Rd. 4) S College Ave & W Tribly Rd.

PAGE 139

120 Approach Determination. +r$$nrr)r))r! 1# )." 22.: 0; 0 6< *< )< +< 4( 1 S College Ave & W Monroe Dr 16 13 47 35 111 2 S Timberline Rd & E Horsetooth Rd 11 31 17 41 100 3 S Timberline Rd & E Drake Rd 16 9 34 19 78 4 S College Ave & W Tribly Rd 7 12 46 17 82 Considering the fact that RLC is normally installed in one approach of the intersection, it was recommended that RLC should be installed at Northbound of S College Ave & W Monroe Dr, S College Ave & E Harmon y Rd, S College Ave & W Tribly Rd, and S Timberline Rd & E Drake Rd. An ad ditional RLC to be installed at the Southbound of S Timberline Rd & E Horsetooth Rd. Th is was determined given the history of at fault vehicle crashes per approach of each of the intersections for the period of three years. (Table 33) RLC Location Below is a map with final RLC locations noting that they cannot be located within 3 miles of each other unless they are located in di fferent directions. Although S Timberline Rd & E Horsetooth Rd and S Timberline Rd & E Drake Rd are located within 3 miles distance, however, RLCsÂ’ are recommended in two different directions, therefore, two more RLCsÂ’ are to be installed at both intersec tions. The intersections of S College Ave & W Monroe Dr and S College Ave & W Tribly Rd should be installed with no restrictions since they meet the installation condi tions. Final RLC locations in the city of Fort Collins are shown in the following map.

PAGE 140

121 %%r0.1r!1#

PAGE 141

Section I: Analyses of RLC Sites Phase I: From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 309 signalized intersections based on crash severity level using the CSL Tables 34 and 35 Severity Level (CSL) and the Normalized locations studied in the city. RLC Sites Selection for Denver Denver From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 309 signalized intersections based on crash severity level using the following equation: CSL = show the top 10 RLC candidate locations based on Severity Level (CSL) and the Normalized -Crash Severity Level (NCSL) from all 309 locations studied in the city. 122 From phase I of the methodology chapter, test of cr ash severity comes as the first criterion. The following equation will be use d to rank 309 signalized intersections show the top 10 RLC candidate locations based on Crash CSL) from all 309

PAGE 142

123 +r$%0r) 0.1rr* rr +r$&0r) 0.1rr* rr!I%2!Fr nnr !!<(#>Fr nnr r(((())"*!#/*>F r nnr 9"* +<#!F r nnr !,#@#*"#'"*Fr nnrn +$#@#*9$#-"*Fr nnrn !C*(&@#*9("*F nn !,#@#*,#"*Fr nn n+$#@#*9r"*F nn ;

PAGE 143

124 Tables 36 and 37 show the top 10 RLC candidate loca tions in the city of Denver based on their potential for improvement in relatio n to crash rate and crash frequency. +r$'0r) 0.1rr1 r)r) rrr &30 580)"(5*$ 9"#'"*<(#>rrnn 9"*+,#!Fnn <(#>!D!Fnn r$#-"*+B#'!nnn +$#@#*9"*Fnnr +I%2!9"*Frrnn <(#>!I%2!Frnnnr 9$#-"*+Fnn n9*("*!!@#*F nn ;<4=7rn n"> +r$(0r) 0.1rr1 r)r) rrr &30 580)"(39*$ $#-"*+B#'!nrnn !!<(#>Fn <(#>!I%2!n r 9"* +<#!F n 9"#'"*!!nrn +C*(&@#*99*("*Fn 9"#'"*<(#>n (((())"*!#/*>F n +$#@#*9$#-"*Fn n9"*+$#@#*F nn ;
PAGE 144

125 %&*F)rr rrrrr2r After running a regression analysis with an R-squar e = .2 and p-value that is well below 0.05, A graph was drawn as shown in figure 44 which reports how much the data of 3 years of crashes varies around the fitted blue curv e. Table 38 shows top 10 RLC candidates from all 309 s ignalized intersections studied in the city of Denver that combine both of a high proportion of front to side crashes and low proportion of rear end crashes. +r$0r) 0.1rr* rr2) 58 0)"3+$5*$ (((())"*!#/*>nnn +$#@#*9$#-"*nnnnr $#-"*+B#'!nnr r9"#'"*!!nn +!9r"*nn !,#@#*"#'"*nn !$#@#*9<%("*nn !!<(#>nn !C*(&@#*9("*nn n +$#@#*9"* nnr

PAGE 145

126 Before moving to phase II of the red light camera s ite selection in Denver and after calculating the top 10 candidate intersection s from each of the criteria show earlier, it is essential to present the final top 10 RLC can didate intersections from all criteria combined. Using the weighting formula described in the methodology chapter, the final top 10 candidates (highlighted in grey) come to be as shown in table 39.

PAGE 146

127 +r$r) 0.1rr* rr)r, 58 Intersection CSL N-CSL PFI-Crash Freq PFI-Crash Ra te Crash Types Total 9"#'"* <(#>nnnn n nnnnnrr $#-"* +B#'! nnnnn nrnn nn nnrnnn <(#> !I%2! nnrnn n nnrnnn r !! <(#> nnrnn nnn nnn 9"* +<#! nnrnnn nnnnnn (((())"* !#/*> nnrnnr nnn nnnn +$#@#* 9$#-"* nnrnnn nnnn nnrnr !,#@#* "#'"* nnrnnnnnn nnn 9"#'"* !!nnnn nrnnn nnn n !C*(&@#* 99*("*nnnn nnnnnnnn 9"*+$#@#*nnnn nnnnnnn +C*(&@#*9("* nnnnnnnn nnn 9:')"*!1'>nnnnn nnnnn r+$#@#*9"*nnnnnnn nnrnn +!9r"*nnnnnnnn nnn <(#>!D!nnnnnr nnnnn !$#@#*9<%("*nnnnn nnn nnn 9"*+,#!nnnnnnn nnnnnnr +$#@#*9r"* nnnnrnnnnnnnn n9*("*!!@#*nnrnnnr nnnnn +$#@#*9"*nnnnnn nnrnnnn !,#@#*,#"* nnnnnnnnnnnn 9$#-"*+
PAGE 147

128 58 Intersection CSL N-CSL PFI-Crash Freq PFI-Crash Ra te Crash Types Total rrr"*+<;##@#*nn nnnnnnnnnnn +I%2!9"*nnnnnnn nnnnnn 9"*+!nnn nnnnnnnnnnn "*+5*!nn nnnnnnnnnnnn "*+5*!nn nnnnnnnnnnnnnnn 9r"*+$#&!nnn nnnnnnnnnnnnn n"*1'#nnn nnnnnnnnnnnnnn $#!!nnn nnnnnnnnnnnn +,#@#*("*nn nnnnnnnrnnnnnn +I%2!9"*nnn nnnnnnnnnnnn r+!@#*r"*nnn nnnnnnnnnnnn ;
PAGE 148

Section II: Further Analysis and Field Investigation of Denver Phase II: From table 39 (Final top 10 RLC candidate intersection used to analyze these intersections further. Fluctuation of Not applicable due to the unavailability of the typ e of Type of Vehicles Not applicable due to the unavailability of the typ e of vehicles data. Economic Evaluation Total Cost of RLC per year Not applicable due to the of Colorado. Further Analysis and Field Investigation of Top 10 RLC Candidates in Denver (Final top 10 RLC candidate intersection s), six criterions will be to analyze these intersections further. Fluctuation of Crashes 100 x = s = Not applicable due to the unavailability of the typ e of vehicles data. Type of Vehicles Not applicable due to the unavailability of the typ e of vehicles data. Economic Evaluation per year + Total Revenue of RLC per year < Total Safety Benefits Not applicable due to the unavailability of the average safety benefits data from in the state 129 Candidates in criterions will be Total Safety Benefits per year unavailability of the average safety benefits data from in the state

PAGE 149

130 Intersection Characteristics +r% ,rrr!*# Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) E Alameda Ave & Leetsdale Dr Intersection Layout Lane Width F 5 Lightening F Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P W Colfax Ave & N Kalamath St Intersection Layout Lane Width F 5 Lightening F Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P Leetsdale Dr & S Quebec St Intersection Layout Lane Width P 4 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level F S Monaco St & Leetsdale Dr Intersection Layout Lane Width P 7 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P E 6th Ave & N Lincoln St Intersection Layout Lane Width P 5 Lightening P Channelization P Signage P Yellow Change Meet ITE guidelines F

PAGE 150

131Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Interval Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P W Mississippi Ave & S Platte River Dr Intersection Layout Lane Width F 4 Lightening F Channelization P Signage F Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P N Colorado Blvd & E Colfax Ave Intersection Layout Lane Width P 6 Lightening P Channelization P Signage F Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P S Federal Blvd & W Alameda Ave Intersection Layout Lane Width P 5 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines P Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level F E Alameda Ave & S Monaco St Intersection Layout Lane Width P 5 Lightening P Channelization P Signage P Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed F Social Structure Criminal history and Income level P S University Blvd & E Evans Ave Intersection Layout Lane Width P 6 Lightening P Channelization P Signage P

PAGE 151

132Intersection Characteristic Name Evaluation Points Score (P/F) Overall Score (out of 7) Yellow Change Interval Meet ITE guidelines F Approach Speed Ave Speed Posted Speed P Social Structure Criminal history and Income level P Figure 46 E Alameda Ave & Leetsdale Dr. (Google Maps)

PAGE 152

133 Figure 47 W Colfax Ave & N Kalamath St. (Google Maps) Figure 48 Leetsdale Dr & Quebec St. (Google Maps)

PAGE 153

134 % S Monaco St & Leetsdale Dr. (Google Maps) Figure 50 E 6th Ave & N Lincoln Blvd. (Google Maps)

PAGE 154

135 Figure 51 W Mississippi Ave & S Platte River Dr. (Google Maps ) Figure 52 N Colorado Blvd & E Colfax Ave. (Google Maps)

PAGE 155

136 Figure 53 S Federal Blvd & W Alameda Ave. (Google Maps) Figure 54 E Alameda Ave & S Monoco St (Google Maps)

PAGE 156

137 && S University Blvd & E Evans Ave. (Google Maps) As a conclusion of the intersection characteristics field investigation (See table 40), 3 out of 10 intersections that have the highest sco re will be qualified to get RLC installed since they passed most of the intersection characte ristics, but still have red light related crashes. Those intersections are: 1) S Monaco & Leetsdale Dr. 2) N Colorado Blvd & E Colfax Ave. 7 S University Blvd & E Evans Ave.

PAGE 157

138 Approach Determination. +r%nrr)r))r! *# )." 22.: 0; 06<*<)<+<4( !!<(#>r rn154 +$#@#*9$#-"*rr 206 !C*(&@#*99*(@#* 99 Considering the fact that RLC is normally installed in one approach of the intersection, it was recommended that RLC should be installed at the southbound approach of S Monaco & Leetsdale Dr, Northbound approach of N Colorado Blvd & E Colfax Ave, and Eastbound approach of S University Blvd & E Eva ns Blvd. This was determined given the history of at fault crashes per approach of each of the intersections for the period of three years. (Table 41) RLC Location Below is a map with final RLC locations noting that they cannot be located within 3 miles of each other unless they are located in di fferent directions. According to intersection characteristics and number of crashes per approach, RLCsÂ’ are recommended to be installed in the following locations: The in tersections of the southbound approach of S Monaco & Leetsdale Dr, Northbound approach of N C olorado Blvd & E Colfax Ave, and Westbound approach of S University Blvd & E Eva ns Blvd are located in a distance of more than 3 miles and their RLC locations are recom mended in different directions and therefore they still can be installed in all three locations. Final RLC locations in the city of Denver are shown in the following map:

PAGE 158

139 &'r0.1r!*#

PAGE 159

140 Analysis of DenverÂ’s RLC Systems Before and After I nstallation rr+@nnnn!nnnn rrn$6! >nn!n!n nn$+@ nnnnn!n$n 49nn6 #n6nnn7>9 nn!nr$ $+@nn$+@rr nn $&>4'67 +r%+rr22r0.1r 0:4(nn nn nnr nn nn#5,(+$% nn nnn 6r=),+ n r r 6r7 )>. +=6r 67 *=)+<$ n r 67nn )?"+=*r rn n n n 67 ; &(+rrrr 2r0.1rrrr< 0nnnn $rB nr"4nr+@n n76)n n!nnnrnC0&-5; n$Cn>4n5#$7

PAGE 160

141 +r%$2)r22r 0.1r* 0:3$nn nn nnr nn nn#5,(+$% nn nnn 6r=),+ r n n 67 )>. +=6r n 67 *=)+<$ 6r7rn )?"+=*r r 6r7 ; &+2)r rr2r0.1rrr r<* 0nrrn $rB nr"4nr+@n n76$!nn rn4!n$ 7nrnnr+@nnr$ n!!$ r+@$nn$ rrrn>49n 5#$"7 +r%%0r2)r22r 0.1r* 0:5$nn nn nnr nn nn#5,(+$% nn nnn 6r=),+ r )>. +=6r r r *=)+<$ r n n )?"+=*r r r ;

PAGE 161

142 &+r2)r rr2r0.1rrr r< 'n6!nnr n$ rBnr"4n r+@nn76$ !nnnn4! n$7nrnn r+@nnn! !$r+@$ nn$rrrn> 49n5#$&7 From past research and the results of the evaluati ons conducted in this dissertation, the installation of RLCs generally is normally asso ciates with intersections with high collision rate or traffic violations, while there a re several criteria that should be considered before selection is made. This research used criter ia that represent both comprehensiveness and accessibility, and were divided into two phases to ensure the quality of the final selections. In phase I, all signalized intersection s in a city were included and tested using three of the major criteria which are usually assoc iated with RLC studies, those are crash severity (normalized and non-normalized), potential for improvement based on crash rate and crash frequency, and finally crash types. All were weighted using proportionality to obtain relative weights equation (which was determi ne by the city engineers). The results of the statistical analysis (top 10 from phase I) w ere moved to phase II for further analysis

PAGE 162

143 that contain both statistical criteria (the economi c evaluation, type of vehicles, and fluctuation of crashes), besides three other field investigation criteria (intersection characteristics, approach determination, and final red light camera locations). The study took a practical path when it implemented the methodology on three of the major cities in the state of Colorado, those ar e Colorado Springs, Fort Collins, and Denver and concluded findings of RLC candidate site s that are problematic and require a safety countermeasure according to each city engine ers. More data for Denver was available to conduct more analysis that could enhance the final research findings. Data includes total c rashes, and crash types (front to side, and rear end) reported before four years of RLC install ation (which was in summer of 2008). Results were consistent with findings from most RLC studies in all three compared points. Recommendations and Conclusions Generally, when reviewing the study, several recomm endations and conclusions can be derived in the following bullets: This study grouped most of the criteria that were k nown to be effective when selecting RLC locations as well as additional crite ria in two phases of analysis processes. Besides using several criteria, the study kept one of its main objectives, which is basing its criteria on accessible data. The study methodology was tested in three different cities with different characteristics, so it ensures diversity.

PAGE 163

144 The cases studies were chosen to target cities with and without RLC systems to compare final results (RLC candidates) to existing RLC locations. Benefit of performing relative weight at the end of phase I is quite noticeable when looking at intersections that are among top 10 in a certain criterion, but are not in the final top 10 candidates. Additionally, total weighted score at some intersec tions was reduced by negative PFI values (These locations reporting nega tive PFI are indicating no potential for improvement in relation to crash rate or crash frequency). Thus, it provides more accurate results. It is noticeable that some intersections were among top 10 in a specific criterion, but it was far from the final top 10 for all criter ia combined. This is basically due to the original weight determined by the city i n the first place. The level of importance the city considers for each criterion in phase I by weighting each of them have a significant impact on the final top 10. This is a plus because it is always believed that the city ha s a chance to participate and provide inputs as oppose to leave it all to the con sultant or the operator. Colorado Springs was one of the cities that has no RLC system under operation, however, the final locations determined by the stud y were examined by the city engineer Andy Richter who quoted: “Yes, the locations you have listed have been prob lematic for us for many years. We are starting to install flashing yellow beacons wit h a sign stating that the signal is about to change to red when flashing. Basically the vehic les will not make it on green. We are trying this as an alternative to see if we can reduce red light running. We will see if it works. CDOT has done the very same thing on stat e highways where the speeds are much higher. Take care Mansour and thank you for th e results of the locations”.

PAGE 164

145 This was a perfect indication of the criteria and p rocesses chosen when selecting RLC locations. In Fort Collins, RLC candidate locations were all r ecommended in the same corridor of the current RLC locations, which indica tes no spillover effect of current RLC locations. It might also be the learni ng curve that drivers have become familiar with the area, especially when know ing that the current two locations were installed back in 1997 and 2006. In Denver, it was surprisingly derived that E 6th Ave & N Lincoln St is among the final top 10 of candidate locations concluded f rom phase I despite the fact that it is one of the current RLC locations and has been under operation since the summer of 2008. However, it was not recommended eventually to receive a RLC due to its failure to pass the field investigat ion (Failed in average speed & yellow phase). This pretty much explains the reaso n why E 6th Ave & N Lincoln St has not had very successful results comp ared to the other three current RLC locations. During one of my frequent visits to Anderson Academ ic Commons at University of Denver, I noticed police and ambulanc e vehicles forcing traffic into certain directions and arranging pedestrians c rossing the intersections of S University Blvd & E Evans Ave. After I asked one of the police officers around I was told that a red light runner hit a bicyclist who unfortunately passed away at the same moment. Later on during my analysis of Denver intersections, the intersection of S University Blvd & E Evans Ave cam e to be one of the top 10 candidates concluded from phase I, and it passed al l the filed investigation

PAGE 165

146 conducted in phase II except that it 1.5s less than the minimum yellow phase timing required by ITE. Eventually, S University B lvd & E Evans Ave was recommended as one of the RLC candidates. 12*Arr 2r2r))r S University Blvd & E Evans Ave. Data was a major obstacle especially when consideri ng the nature of the subject and therefore, some of phase II criteria were not i ncluded in the analysis and thus not included in the final results. As indicated earlier, all intersections of the city were included with exception to those with no data available. Case studies like Fo rt Collins and Denver where RLC system currently exists; yellow highlighting wa s used in all of the analysis tables to identify these current locations and make comparison to other locations easily.

PAGE 166

147 Although RLC’s before and after analysis in Denver shows overall consistency with findings from other RLC studies, there are sti ll some points that worth further investigation. For instance, intersections of W N Qubic St & E 36th Ave and N Kalamath St & W 6th Ave are not shown sig nificant decrease after RLC installation in terms of number of crashes (See table 44), and therefore a question would be why the city chose to make a big investment (up to 40K a month) and install RLC at these locations. When referring to table 44, it was very interesting to see that number of rear end crashes were in fact reduced in the year of 200 9 (just one year of installation), however the number increased again a s most studies concluded in the years after. This might be interpreted by the term “learning curve” where drivers gain more awareness of these RLC locations as time passed and get to slam on their breaks as they approach these sites. In Fort Collins, before RLC installation data was n ot available for the two RLC intersections S College Ave & E Drake Rd (installed back in 1997) and S Timberline Rd & Harmony Dr (Installed back in 2006) however, RLC top 10 candidates recommended by this study did not includ e any of these two locations. In fact, none of them were recommended under any of the criteria studied and there were always ranked down the list of the 106 signalized intersections studied. This is an indication of an effective RLC system experience in the city of Fort Collins that worth c onsidering as an ideal example for any future studies.

PAGE 167

148 This last bullet concluded that Fort Collins RLC sy stem experience is good example according to this study results, however, i t is also important to know that the city did not experience that immediately a t the intersection of S College Ave & E Drake Rd. A media report published back in 2005 confirmed that the city has increased the yellow time interval for one second, and the intersection has experience dramatic drops since then. In few mo nths after the change and for two months of comparison, crashes were reduced by 58% while citations dropped by 63%. (BENSON, 2005) From the literature review conducted in this study which mostly based on studies and researches available from Transportatio n Research Board (TRB), it is obvious that majority of the findings agree that RLC systems decrease total red light related crashes, front to side crashes, a nd increase rear end ones. In the other hand, there are still some different find ings that are usually biased by the level of analysis when comparing to the ones av ailable at TRB. For instance, a speaker from New Jersey in a report pub lished by NBC news says front to side crashes increased 400%, which is a fi nding that does not comply with any of the studies I have reviewed. It can on ly be possible if the speaker was referring a particular intersection(s) where ot her traffic signal changes were made and there were very few front to side crashes there previously so an increase from 1 to 4 is 400%, or there was somethin g wrong with the selecting process of these RLC locations at the first place ( NBC news, 2014)

PAGE 168

Finally, below is a summary of the steps needed to perform the red light camera sites selection criteria methodology chapter of this dissertation: Phase I (includes all intersections in a given juri sdiction): 1) Normalized c 2) C rash severity level 3) Potential for improvement in relation to .(5 4) Potential for improvement in relation to crash freq uency 5) Crash types 6) Weighting for all criteria Final outcome is a list that contains final top 10 RLC candidates. Phase II (includes the final top 10 RLC candidates) 1) Fluctuation of crashes Finally, below is a summary of the steps needed to perform the red light camera sites selection criteria including the equations indicated in details via methodology chapter of this dissertation: Phase I (includes all intersections in a given juri sdiction): Normalized c rash severity level rash severity level Potential for improvement in relation to crash rate ,5#$(/6$(3*'74 .(5 @6"$.( Potential for improvement in relation to crash freq uency ,5#$(,0%&6$(37 = Crash types Weighting for all criteria Final outcome is a list that contains final top 10 RLC candidates. Phase II (includes the final top 10 RLC candidates) Fluctuation of crashes nn 149 Finally, below is a summary of the steps needed to perform the red light camera details via the ,5#$(/6$(3*'74

PAGE 169

2) Type of vehicles 3) Economic Total Cost of RLC 4) Intersection characteristics Analysis of four major characteristics as follows: 5) Approach determination 6) Red light camera locations + ('*#%#((' Type of vehicles Economic evaluation per year + Total Revenue of RLC per year < Total Safety Benefits Intersection characteristics Analysis of four major characteristics as follows: 5(<&% "))!) ##;(5*# !#!%% Approach determination >",%#*#( Red light camera locations ('*#%#((' '#(;& '/<$ 150 Total Safety Benefits per year '/<$

PAGE 170

151?@0G1,+9* nnr9nn>47>'9 >nn1nB,5 nn6>060>>47>06>4"7> nrnrnr rrn r n/D> 02n'n.n>47> rrn 02'>En6>E>65;nF60>3>47> "r#$%r&rrn n'(!#$)'n rr ?;nn 9nnG+nnAn 9 -nG;nn2!n9nrrnrn> -n6#>2>47> "*'+'%"$,"-.-*#'&/$01"2'#"// '3',%',22"#+ -@9$)6,>47> &rn#$rr $nr >-n-nrn2!nn9n n'rrn4--29'7> 47>-13r2!nn > $6@>6$6H>65;!n6>47> 'r!rnrn%nr44 5rrnrr ,n/> n6->4((&7> rr6#r7 #2-> *;';6,>47> 2r#6*8r#$r r 9 nn> 6 >650n69>47> 'n#$rr A9*H$n> )69>46H/;*7>+nnn $nrI+nI ..)'0 ,. > n9nrr*'rr>47 > r *>47>9nnrn >

PAGE 171

152*6;>65+n!n6*>4&7> !#r#/$rr'n '9!r8%!:;8!:8<==;8 ><8>:/>? A9*H$n> *)6->4(7> #/nrr6"r/rr n/ nrrnr n)6+>->66 >>66+>,>65@ n69>4&7> 'r!r 1!&r"!r#$rr) r 2$> @$61>65,6,>47> n)nr2 &nrrr7 !@r8rrr H$nr 9nn+n n>9nrr*'rr>47> nrr > #n6@>46H$7>+r1/ 6#B1r9nrr@ n@9/+@nn#+ $;1$nr< <::!!!>!>:n:1B/B#B 1rBB@B#nB B&">#n2!n-n>47> )r!r! #2-> #n2!n-n>47> r'r!r#$rr #n2!n-nJA9n nH1n 'rr>47> "!r%n&nr!rnn rr >+r <::!!!>r!n>>::!:>r#.6*>H>62nn)6>@>6,n69>H >6'nn6,>6,n$n6>65 1n!6,>4&7> ,rnr'r!rr 'nrA"!r#$#! )r A9*& 9nrn*>#.6*>H>62nn)6>@>6'nn6,>6 ,n69>65,n$n6>4(7> r%r#!nrn "rn"!r# $#!'nrr8r8, r/nr"r H$nr9nn*>#)>H6> >4(((7> rrB!# n r*rnn n#n>r1n)n9nrr6#> #9nrr*'rr>47> )rr

PAGE 172

1533n6;>H>6,6H>>6-6+>*>6*n 6>65D$)n6>D>4"7> rn#$rrC)/#'nD r%r ?n'9>3n61>*>47> rnn!7r"!n #/ $#!)'n2r /r,n:;!*n /9nn>3n6H>46&"7> #$rr6"'rn +r 3n'9+n<<::Fn>!> ::&:":BB nnBnBnB:3n>47> 1 3,n>4>>7> ,"r<=E< >+r<::n>>: 3n2n$>4(&7> rn)n! >+r <::!!!>>nBn>!n>$:r:F$n:r n> A$nA$r2!nnr>47> #$rr A$n A$r2!nnr>A$nA$r2!nnr>47> #$rr A, *+3*+6D>9>47> 0,&'$.'*-#'.0/,)".&-)'#"-.-* 0."$'&.'#'-.1#'&$01"2'#" 6-$nn<+'-;9+-##A-/92'+A9E4+9-7>2$6>47> #'&$01#,...0.#1-&'$".& D2$6H>*>65$n6>,>47> 'r!r'n#$ #!rr'n#r. H$nr9nnnr> 2nn6->6,n6 >6$6H>H>65+6 H>4"7> nrn )#$rrr 6"nr0rF 'r'(!rr'nr7r H$nr9nn+n n>2nn)6>6'nn6,>6#.6*>H>6, n69>65,n$n6>47> # $#!r6"!r'n)r 'r!r7rr "r 9nn+n n> 2n62>66>2>652n%$6,>,>4&7> !rn rrrnnrrn6"7r rrnnrrr

PAGE 173

154260>6+.)n6H>656#>4((7> "'r!r#$rr 2)n6->65En$n6;>47> 'r!r,r#$rr )r*rr5!8%" 2)nn6>4((7> 'r!rr)nn" nr 1r-r*r# nrn1nn/> Dn6->4(((7> 2-)'-)$',#%'+'&*'"##'&/$01#,..'# 9nrr nr4n7>D6H>n>4((7> rn&n1r@B8n' #2->/>r 9nn>D6@>->4(("7> 2/rrr2!&r r )n ,-> D6@>->4(((7> *r#62@!rnr! nr *-rr/%<6*--@ n 8-rFr8&8 #!8n!8r#nr /rnrn> D6,>3>47> 'nrnr#$ rr 9n> @!6>47> r2rr4r-r1r@B @'-9'+62>9>4>>7> r2r >+r n>!n>;nn/r9nrr@n!n' n>47> %n ;/9@'> ;nnnr$>47>;>;nnnr$>4(7> >+r <::!!!>>:nrKn:nr:1n: >n ; !>46,n&7> .7 >+,n&66r <::!!!>!>:B!:BB nrrBnnB& ;!n6*>*>47> +r/$$rr#$ rr)r A9*-$n,n*>

PAGE 174

155;nH>3n6@>->47> rn1r' nr */r?n>10>;2+1>47> rn#$rr'nr'5 n ,nn.62> nnrrr ,n6->4>>7> 7!r >+r !!!>nnn>:!:n:$::$,n.6D>@>6516 >*>47> rrnF#$#!* rr)'n)r ,n6->9>47> "!r#$rr'n rr)r ,n6->9>652$60>47> 'n#$rr'n# $%r"!8%r A$nA$r2!nnr> ,362>0>65*6D>->47> 1'2)"-*#'&/$01"2'#" '.*-#'2'.-.#"1'3)'#'.' A9*H$n> '6>;>467>LBn nrnn> M$69>47> .%'0"-.".&-)2G"-.-*2-7$'-,#''2 -. 2-&'$.02'1-&.6-2)'.&,26)")'#-."&%".' &,#*"' #".)-#"-.+'2 9n9nnA$G$!+ /9nn>$nnn6+>4(7> 2%nrnrrr$rr! &r ,r >+r<::!!!>>n>>:1$: &>r $6H>,>6,n6 >65 $)63>47> nrr$nr# $rr -$nrr9nn-n rnnn> 6D>650n6>1>4"7> rnnrrr "Fr #rrnB >47>+r <::>r>:::nnB B)> +63>2>47> 77nB+! 4((&7> #$#!8&H'!nr

PAGE 175

156+6+>->47> #'&,-...9,#+#"1'"-"'&1#'& $01"2'#"'.*-#'2'..-3."#&8"$*-#." 0n< 9nn+n n>+6+>->650n6->#>4((7> 1"#"'#-*#'&$01%-$"-#6 #',$-*"*'$&.%'0"-. H$nrnr+n> +nnr>47>+r<::!!!>$ >:n:+nB nrB">9$)6;>4(6;7>490n17+ rn nnn$nBn) !n>:!B::n:(:::-+(""> 9n)6->1>65+6;>47> '%"$,"-.-*"*'+'.*-#'2'.-. 1".0.0&#%'#7'1"%-#/#,.-.#'& 1$$/:Ann nr9nn:#2->9/>>nr9nn>467> "!r1r )"*''"/$, >+r$>< <::!!!>r!n>>:nrn$:$n>9/>>nr9nn>4((&7> rr'(!"n +r<::!!!>r!n>>:n:$ > 9;nnn2!n+n1n> 47> 9;nn9nn@n>4((7> r!rnrr 'nn"nE??EI!r >+r<::>>:':>

PAGE 176

157 ;>>9n*,H +r%&;r21r) rr2rr9+/>35nrnn :3<"B9"Anrr r9<"119"A3+"$">9@9+D,1:9ED>!/>nn 53!$5/$<9>/nrnn 53!+9A">""Ann "C!15+@""Arnr @"/+9!/>31C11@9@3D/D@<"+$D>/rnr n+D9/!@<3D<>/"+$:/>nrrrr @"/+9!/>3+D9/!@/>3+"$">9@3!"$">9@9>/n

PAGE 177

158 0; 0)" nn 4 nn 0A.! nn 3 nn &7B)(+,(+, 9<"119"A3+C+5D+@/3+"$">9@/rrnnnn 9DD>9+/>3<9K5+E1D+>/rn >C@<5+@<3+D9/!@/9++"+/>3!"$">9@3!D9/!@>/nr "!1/DGD+@<3!"$">9@9@9"Arr r+C//"@<3"<9/"/B@rnrr r9<"119"A3+":!"1$:"An rrE"<<9/>3+"$">9@9@9@<3A5$B9/!>/ r$D+!151C15D+"A3+"$">9@9+/>rnr n>C@<5+@<3+"$">9@@3!C//"@9@<3"<9/"/B@
PAGE 178

159 0; 0)" nn 4 nn 0A.! nn 3 nn &7B)(+,(+, r+D9/!@<3!$"/9,/99$/rr 9<"119"A3+C//"@C@<5+@9@3!$5/$<9>/rn +"$">9@<3+$"/9,/99$/n n!+9A">""A3!DC1:E"19/>rn 9,DC+1"5+@<3!C//"@9+D,1:9ED>!/>rr rr $D+!151C15D+"A3+D9/!@9@<3+C+5D+@rr >C@<5+@<3+C+5D+@9+/>3/"+E9DD>>/r "C!15+@D<"+>@9@<3!$:9<1D+/>rnrn "$">9"/B<3!"$">9@3+"$">9@9+/>3+"$">9@9@9@<3!:/5>9//>r E"<<9/>3+D9/!@9@9@E9>/3+"$">9@D<"+>@<3+"$">9@
PAGE 179

160 0; 0)" nn 4 nn 0A.! nn 3 nn &7B)(+,(+, 9,DC+1"5+@<3!D9/!@
PAGE 180

161 +r%'1r)rr r9@<3A5$B9/!>/nn >C@<5+@<3+C+5D+@9+D,1:9ED>!/>rnn r>/9++"+/>3!"$">9@9"Arnnr "5/D/1/>3!"$">9@C@<5+@<3+D9/!@31C11@9@<3!:/5>9//>nnr 9DD>9+/>3/"+E9DD>>/nnrn 9">D<"+>@<3+"$">9@/>3+"$">9@/"+$:/>rrnnr 9,DC+1"5+@<3!D9/!@9@9+/>rnn @"/+9!/>3D/D@<"+$D>/rnnn 5<1D+9/D@3!D9/!@9@9@rnn r+$"/9,/99$/3+D9/!@nn +D9/!@<3!$"/9,/99$/nn "C!15+@9@9"/B<3!"$">9@3!D9/!@9@9+/>3+"$">9@""A3!DC1:E"19/>nnn E"<<9/>3+D9/!@C@<5+@<3+"$">9@9@<3!$:9<1D+/>rnn 9<"119"A3+C//"@
PAGE 181

162 580)")(+,*$ rn9<"119"A3+C+5D+@3+"$">9@""Ann rr"C!15+@@9@9+/>3<9K5+E1D+>/nn rrnnJ"+519<3!$5/$<9>/nn n9<"119"A3+":!"1$:"Ann "C!15+@D<"+>@3+D9/!@9@/3+"$">9@9@>/nn n"C!15+@C@<5+@/nn +"$">9@<3"<9/"/B@9+D,1:9ED>!/>nn 53!$5/$<9>/nn 53,5<E9>/3+"$">9@9>/nnn E"<<9/>3+"$">9@9@""Annn +"$">9@<3+$"/9,/99$/nnn +"$">9@<3+C+5D+@9+/>35nnn "C!15+@
PAGE 182

163 580)")(+,*$ n+C//"@<3"<9/"/B@3!C//"@9@
PAGE 183

164 +r%(1r)rr r2 58 0)"(+, *$ "5/D/1/>3!"$">9@9@<3A5$B9/!>/n >C@<5+@<3+C+5D+@C@<5+@<3+D9/!@9@9+D,1:9ED>!/>rnnn "5G9<"+>/>3+"$">9@/9++"+/>3!"$">9@9@9@9@9"Arnnn 9DD>9+/>35nn +D9/!@<3!191!D+:5<9+D,1:9ED>!/>nnn "5/D/1/>3!D9/!@""Annr +"$">9@<3!:/5>9//>nnr 5<1D+9/D@3!D9/!@nnrr +D9/!@<3D<>/"+$:/>rnnr @"/+9!/>3+D9/!@D<"+>@<3+"$">9@9+/>nnr n9DD>9+/>3/"+E9DD>>/nnr @"/+9!/>31C11@9+/>3+"$">9@rnn @"/+9!/>3D/D@<"+$D>/rnn 9<"119"A3+C+5D+@
PAGE 184

165 58 0)"(+, *$ rn$D+!151C15D+"A3+D9/!@C@<5+@<3+"$">9@""A3!DC1:E"19/>nn rE"<<9/>3+D9/!@9@/nnn r!"$">9@<3!$:9<1D+/>nnn r+"$">9@<3"<9/"/B@/nnnn "C!15+@9@9@9"/B<3!"$">9@9@/3+"$">9@9@<3+C+5D+@3+"$">9@9@9@<3+$"/9,/99$/nn 9DD>9+/>3<9K5+E1D+>/nn rnnJ"+519<3!$5/$<9>/nn ,<5+1/5>E9>/3+"$">9@""Annr :r@35nnr n9,DC+1"5+@<3!C//"@@>/rnn "C!15+@C@<5+@3+"$">9@D<"+>@9@
PAGE 185

166 58 0)"(+, *$ n/5+19/!3!"/B!5>9>/nnn +C//"@<3"<9/"/B@3!C//"@
PAGE 186

167 +r%;r21r), r)r)r rrrrB2 6C$ 4. .6C$.&30&30 0; 0)"74#nn% ( ( Crash Rate Crash Rate Crashes Crash Rate Crash Freq 53$5"//D+!1nnnnnnn n 9DD>9+/>35rrnnnrnnnrn :3<"B9"Annnnrn r r 9<"119"A3+"$">9@9+D,1:9ED>!/>nnnrnnn n 53!$5/$<9>/rnnnnnnn 53!+9A">""Annnnnnnnnn rn "C!15+@""Arnnrnnnnrn r @"/+9!/>31C11@9@3D/D@<"+$D>/nnnn n n +D9/!@<3D<>/"+$:/>rnnnnnn nnnn @"/+9!/>3+D9/!@/>3+"$">9@3!"$">9@9>/nnnn nn

PAGE 187

168 6C$ 4. .6C$.&30&30 0; 0)"74#nn% ( ( Crash Rate Crash Rate Crashes Crash Rate Crash Freq 9<"119"A3+C+5D+@/3+"$">9@/nnrnn n 9DD>9+/>3<9K5+E1D+>/rrrnnrnnn nn >C@<5+@<3+D9/!@/9++"+/>3!"$">9@3!D9/!@>/nnnnn nnn "!1/DGD+@<3!"$">9@9@9"Arrnnnrn n r +C//"@<3"<9/"/B@nnrnn nnn r 9<"119"A3+":!"1$:"Arnnnnr nnrn rr E"<<9/>3+"$">9@9@9@<3A5$B9/!>/nnnnn n r $D+!151C15D+"A3+"$">9@9+/>nnnrnnn rnn n >C@<5+@<3+"$">9@@
PAGE 188

169 6C$ 4. .6C$.&30&30 0; 0)"74#nn% ( ( Crash Rate Crash Rate Crashes Crash Rate Crash Freq "5/D/1/>3!C//"@9@<3"<9/"/B@C@<5+@9@3!$5/$<9>/nnnrn nnn +"$">9@<3+$"/9,/99$/nnnn nnrn n !+9A">""A3!DC1:E"19/>nnnrnnnr rnnnr 9,DC+1"5+@<3!C//"@9+D,1:9ED>!/>nnnn nrnrn $D+!151C15D+"A3+D9/!@9@<3+C+5D+@nnnnnr nn >C@<5+@<3+C+5D+@9+/>3/"+E9DD>>/nnnnrn rnr "C!15+@D<"+>@9@<3!$:9<1D+/>nnrrnnn nnrn "$">9"/B<3!"$">9@3+"$">9@9+/>3+"$">9@9@9@<3!:/5>9//>nnrnrnr nnr E"<<9/>3+D9/!@
PAGE 189

170 6C$ 4. .6C$.&30&30 0; 0)"74#nn% ( ( Crash Rate Crash Rate Crashes Crash Rate Crash Freq 9!"+5EC99@9@E9>/3+"$">9@D<"+>@<3+"$">9@
PAGE 190

171 +r%1r)rr )r)rr r &30 58 0)"(5 *$ @/5"/E"193+D9/!@3+D9/!@9+/>35nnr r"5/D/1/>3!"$">9@/>3+"$">9@3!D9/!@9@3D/D@<"+$D>/nnn r95B9!9"B"A3+"$">9@9@9+D,1:9ED>!/>nnn >C@<5+@<3+D9/!@9@<3"<9/"/B@9@9@9>/nnn 9<"119"A3+$5/$<9>/nnn r$D+!151C15D+"A3+"$">9@>/nnnn 53!+9A">""Annnn n+"$">9@<3+C+5D+@31C11@nnnn +D9/!@<3!$"/9,/99$/nnnn r"C!15+@C@<5+@/"+$:/>nnnn +D9/!@<3+C+5D+@
PAGE 191

172 &30 58 0)"(5 *$ +"$">9@<3+$"/9,/99$/nnrnn 9D/1"<>/3+"$">9@9@C@<5+@<3+C+5D+@nnnnn r!+9A">""A3!DC1:E"19/>nnnnn r"5/D/1/>3!C//"@9@<3!$:9<1D+/>nnrnn nE"<<9/>3+"$">9@/nnnn "C!15+@9+/>nnnn r9DD>9+/>3+"$">9@C@<5+@<3+"$">9@9@9@9+/>3<9K5+E1D+>/nnnn ,<5+1/5>E9>/3+"$">9@/9++"+/>3!"$">9@3+D9/!@9@<3A5$B9/!>/nnn 53!19JD+!1nnn r95B9!9"B"A3+C+5D+@9"Annn rnnJ"+519<3!$5/$<9>/nnnr +"$">9@<3!:/5>9//>nnnnr 9DD>9+/>3/"+E9DD>>/nnnr @D<"+>@<3+"$">9@9@""Annn r"$">9"/B<3!"$">9@9+D,1:9ED>!/>nnn !!1!13$5"//D+!1nnn

PAGE 192

173 &30 58 0)"(5 *$ 9,DC+1"5+@<3!D9/!@@D<"+>@3+"$">9@
PAGE 193

174 +r& 1r)rr )r)rr B2 &30 58 0)"(39 *$ 9DD>9+/>35nn @"/+9!/>3+D9/!@3!"$">9@9+D,1:9ED>!/>n +D9/!@<3!191!D+:5<9@/>3+"$">9@3!D9/!@9@9@9@9@<3"<9/"/B@C@<5+@<3+D9/!@9@""Ann 9<"119"A3+$5/$<9>/nn r$D+!151C15D+"A3+D9/!@9@3D/D@<"+$D>/nn +"$">9@<3+C+5D+@nn n+C//"@<3"<9/"/B@9>/nrnn "C!15+@>/nnn r+D9/!@<3D<>/"+$:/>nnnn +"$">9@<3+$"/9,/99$/nnnn @"/+9!/>31C11@C@<5+@
PAGE 194

175 &30 58 0)"(39 *$ 9D/1"<>/3+"$">9@9@C@<5+@<3+C+5D+@nnnn r"5/D/1/>3!C//"@""A3!DC1:E"19/>nrnnn r!"$">9@<3!$:9<1D+/>nnn rE"<<9/>3+"$">9@9+/>nn 53@5JDC!1rrnn 9DD>9+/>3<9K5+E1D+>/nn 9!"+5EC99@C@<5+@<3+"$">9@9@E9>/3+"$">9@9+/>3+"$">9@/9++"+/>3!"$">9@9"Ann 53!$5/$<9>/nn 95B9!9"B"A3+C+5D+@3!$5/$<9>/nn +"$">9@<3A5$B9/!>/nn E"<<9/>3+D9/!@9+/>3/"+E9DD>>/rnn +"$">9@<3!:/5>9//>rnn n"C!15+@D<"+>@9"/B<3!"$">9@@D<"+>@<3+"$">9@9@
PAGE 195

176 &30 58 0)"(39 *$ +$:9!1+C1!13E"/>9+D,1:9ED>!/>nnn :3<"B9"Ann 9,DC+1"5+@<3!D9/!@3+"$">9@""Ann :r@35n

PAGE 196

177 +r&;r21r) rr2) 0;03$55$ 5B4(4." 53$5"//D+!1nnnrnnnn 9DD>9+/>35nrnrnrrrnn :3<"B9"Anrnnnn r9<"119"A3+"$">9@9+D,1:9ED>!/>nnnrnr nnn 53!$5/$<9>/nnnnrnn 53!+9A">""Anrnnnnn "C!15+@""Anrnrrnn @"/+9!/>31C11@9@3D/D@<"+$D>/nnrnn nn n+D9/!@<3D<>/"+$:/>nnnnrn nn @"/+9!/>3+D9/!@/>3+"$">9@3!"$">9@9>/nnn nnn 9<"119"A3+C+5D+@
PAGE 197

178 0;03$55$ 5B4(4." 9D/1"<>/3+"$">9@/nnnrrr nn 9DD>9+/>3<9K5+E1D+>/nnnrrr rnn >C@<5+@<3+D9/!@/9++"+/>3!"$">9@3!D9/!@>/nrnn nn "!1/DGD+@<3!"$">9@9@9"Annr rnn r+C//"@<3"<9/"/B@nrnnr nn r9<"119"A3+":!"1$:"Annn rnn rrE"<<9/>3+"$">9@9@9@<3A5$B9/!>/nnnn nn r$D+!151C15D+"A3+"$">9@9+/>nnn rnnnn n>C@<5+@<3+"$">9@@3!C//"@9@<3"<9/"/B@C@<5+@
PAGE 198

179 0;03$55$ 5B4(4." 95B9!9"B"A3+"$">9@3!$5/$<9>/nnn rnnn +"$">9@<3+$"/9,/99$/nnn nn n!+9A">""A3!DC1:E"19/>nnnr nnn 9,DC+1"5+@<3!C//"@9+D,1:9ED>!/>n nnrnnn $D+!151C15D+"A3+D9/!@9@<3+C+5D+@nnnnr nn >C@<5+@<3+C+5D+@9+/>3/"+E9DD>>/nnn nn "C!15+@D<"+>@9@<3!$:9<1D+/>nnnr nn "$">9"/B<3!"$">9@3+"$">9@9+/>3+"$">9@9@9@<3!:/5>9//>nnnrn n E"<<9/>3+D9/!@9@9@E9>/3+"$">9@D<"+>@<3+"$">9@
PAGE 199

180 +r&1r)rr r 5803+$5*$ 9<"119"A3+"$">9@9@<3"<9/"/B@/nnn +D9/!@<3D<>/"+$:/>nnn /5+19/!3!"/B!5>9>/nnn >C@<5+@<3+C+5D+@9+/>3/"+E9DD>>/nnnn 53@5JDC!1nnn r@"/+9!/>31C11@9@9"/B<3!"$">9@/>3+"$">9@3D/D@<"+$D>/nnnn !+9A">""A3!DC1:E"19/>nnnn n53!19JD+!1nnn 9<"119"A3+$5/$<9>/nnn "!1/DGD+@<3!"$">9@3!D9/!@C@<5+@9@<3+C+5D+@9@E9>/3+"$">9@9+/>35nrnnr 53!+9A">""Anrnnr r53++9A">""Anrnnr "5/D/1/>3!"$">9@/9++"+/>3!"$">9@>/nrnnr 53,5<
PAGE 200

181 5803+$5*$ rn95B9!9"B"A3+C+5D+@9@9"Annnrr r+C//"@<3"<9/"/B@nnnrr r>C@<5+@<3+"$">9@9+D,1:9ED>!/>nnnrr r!1:!13$5"//D+!1nnnrr r9DD>9+/>3<9K5+E1D+>/nnnnrn r+"$">9@<3A5$B9/!>/nnnnrn n$D+!151C15D+"A3+"$">9@nnnnrn "C!15+@D<"+>@9+D,1:9ED>!/>nnn 9<"119"A3+":!"1$:"Annn +"$">9@<3+$"/9,/99$/nnn !"$">9@<3!$:9<1D+/>nnn 9">D<"+>@<3+"$">9@3!C//"@3+"$">9@9+/>nnn "C!15+@@9+/>3+"$">9@/3+"$">9@9@3+"$">9@9@<3!:/5>9//>nnn rE"<<9/>3+D9/!@C@<5+@<3+D9/!@9@3!$5/$<9>/nnn

PAGE 201

182 5803+$5*$ n@/5"/E"19@<3+"$">9@3+D9/!@9@
PAGE 202

183 +r&$;r21r r2rrn 15@9/<5+9/>:D/!91DD1:/>rnnr $D<<9E9"A1/5<@/>rn r@D"/>"/:"/D+/>rr !:59<>!!1>/"B9/>rrnrn !:59<>!!19<5G"@91:!1nn !:59<>!!1!"</D!9$1/>rn n !:59<>!!1 :"/D+/> nr !:59<>!!1C<@9//!1nnrn !:59<>!!1/D!9$1/>rr 1",1:5<:D/!91DD1:/>rn r$51"/B9<5G"@91:!1nnnn 1",1:5</D!9$1/>rnrn $D<<9E9"ADC+1"5+rrnrn !:59<>!!1:D/!91DD1:/>n "!D+!1:"/D+/>nrn $D<<9E9"A:"/D+/>rrn n$C//:"/D+/>n <9">/"B9/>rnr 1",1:5<>/"B9/>rrrr 1",1:5<C<@9//!1nrn rG59E<9//D$B$/99Brnr !:59<>!!1>/"B9/>nn J,B@D"/>"!!1/D<<"+>DD/9n 15@9/<5+9/>+"+$E/"nn $D<<9E9"A,D!!5<$/99Brnrn n<9"C<@9//!1nnnrn $D<<9E9"AC<@9//!1nr J,B:"/D+/>rn

PAGE 203

184 0 ; 0)" 3! 07 n 4 n 0A.! n 3 n &7B )(+, (+, <9"D"B/5>E9rnr r$D+!151C15D+9<5G"@91:!1nrrr J,B1/DC1"+rn 9<5G"@91:!1n <9"!1C"/1nrn 9<>/CC<@9//!1rnr rn<9"!DC1:/5>E9rn rD/1:5+E1D+>/"B9rn r15@9/<5+9/>$"/5@DCrnnrn r1",1:5<A"<<9,D/E9rnr rr/5A9/!5>9"ADC+1"5+nrnn r!:59<>!!1/D$BDC+1"5+n rD/B!:5/9>/"B9nnnnn r15@9/<5+9/>A9/D+1n r<9",D!!5<$/99Bnn r<9"9++D$Bnn n/95+E1D+C<@9//!1nn <9"@D"/>"n 1/">515D+:D/!91DD1:/>rn !1"/,nrrr 15@9/<5+9/>@"11<9$/99Bnn G59E<9/$DC+$5<1/99rrn :9>@99C<@9//!1rn n<9"/D!9$1/>nrr <9"/5A9/!5>9rnrnnn $9+1/9/D!9$1n !1DA9/>/"B9nr r1",1:5<:"/D+/>nr /95+E1D+/D!9$1nrnr $D<<9E9"AD<5A9rnr

PAGE 204

185 0 ; 0)" 3! 07 n 4 n 0A.! n 3 n &7B )(+, (+, $D<<9E9"A!B"nn $D<<9E9"A$:9//nn $D<<9E9"A<"D/19nr n<9"9<5G"@91:!1nn /5A9/!5>9"AC<@9//!1nnrnn $D<<9E9"A"E+D<5"nr $D<<9E9"A5<9+J9,,9/!D+nn <9">D$1D/!<+nn /9!9"/$:39">D<"/B>/"B9nr !1"+,D/>:D/!91DD1:/>n 15@9/<5+9/>15@9/DD>nn $D<<9E9"A"<93J9,,9/!D+nr n!:59<>!!1/"5+1/99n "!D+!1:D/!91DD1:/>nnn J,B:D/!91DD1:/>nrrnr /5A9/!5>9"A/D!9$1/>nn r15@9/<5+9/>$C!19/rnnn $D<<9E9"AD+/D9rrr $D<<9E9"A >/"B9/> n r r rn 15@9/<5+9/> :"/D+/> r r $D<<9E9"A:D/!91DD1:/>rn $D<<9E9"A,DD1:5<rnnnr $D<<9E9"A<"C/9nrrn $D<<9E9"A@D$B"+nrrnr r!+D9!":"/D+/>nr G59E<9/:"/D+/>n $D<<9E9"AB9+!5+E1D+nn $D<<9E9"A@D"/>"
PAGE 205

186 0 ; 0)" 3! 07 n 4 n 0A.! n 3 n &7B )(+, (+, n<">DD+:"/D+/>nr n$D<<9E9"A1/DC1"+rnrr n$D<<9E9"A51B5+n nr$D<<9E9"A!/5+E"/Bnnnnn n$D<<9E9"A/D!9$1/>rrrr n$D<<9E9"A!1C"/1nn ;

PAGE 206

187 +r&%1rr r!!1/"5+1/99n r!:59<>!!1/D<<"+>DD/9n !:59<>!!1E9n D/1:5+E1D+>/"B9n 1/">515D+:D/!91DD1:/>n !:59<>!!1!"<"/:"/D+/>nn 1",1:5<A"<<9,D/E9rnn $D<<9E9"A$D!!1 :"/D+/>nn r15@9/<5+9/>:D/!91DD1:/>nn <9"9++D$Bnnn 15@9/<5+9/>A9/D+1nn <9"9<5G"@91:!1nn !:59<>!!1/D!9$1/>nn $D<<9E9"ADC+1"5+nn n$D<<9E9"AD+/D9nn $D<<9E9"A@D"/>"!!1:D/!91DD1:/>nn J,B1/DC1"+nn r!:59<>!!1/D$BDC+1"5+nn :9>@99C<@9//!1nn $D<<9E9"A<"C/9nnn <9"D"B/5>E9rnn n$D+!151C15D+9<5G"@91:!1rnn 9<>/CC<@9//!1rnn !1"/,rnn 1",1:5<C<@9//!1rnn r<9"!1C"/1rnn $D<<9E9"A >/"B9/> r nn J,B:D/!91DD1:/>rnn $D<<9E9"A@D$B"+rnn 1",1:5<>/"B9/>rrnn $D<<9E9"A,D!!5<$/99Brnn rn15@9/<5+9/>$"/5@DCrnn r<">DD+:"/D+/>rnn r$D<<9E9"A:"/A"/>rrnn r15@9/<5+9/>>/"B9/>rnn rr$D<<9E9"A/D!9$1/>rnn r$D<<9E9"A1/5<@/>nn r 15@9/<5+9/> :"/D+/> nn r$D<<9E9"A:D/!91DD1:/>rnn r"!D+!1:"/D+/>nn

PAGE 207

188 Rank 0)")(+,Weighted r15@9/<5+9/>+"+$E/"nn n/5A9/!5>9"ADC+1"5+nn <9">D$1D/!<+nn $D<<9E9"AB9+!5+E1D+nn 15@9/<5+9/>15@9/DD>nn r$D<<9E9"A51B5+nn $D<<9E9"A!"</D!9$1/>nn J,B:"/D+/>nn 15@9/<5+9/>/D!9$1/>nn 1",1:5<:D/!91DD1:/>nn n$D<<9E9"A,DD1:5<":D/!91DD1:/>nnr "+:"11"+:D/!91DD1:/>nnr <9"/D!9$1/>rnnr !:59<>!!1>/"B9/>nnr n1",1:5<9<5G"@91:!1nnr G59E<9/:"/D+/>nnr <9">/"B9/>nnr <9"C<@9//!1nnnr rnnr n!:59<>!!19<5G"@91:!1nnr !:59<>!!1C<@9//!1rnnr !+D9!":"/D+/>rnnr $C//:"/D+/>nnr rG59E<9//D$B$/99Bnnr /95+E1D+C<@9//!1nnr !1DA9/>/"B9nnr /9!9"/$:39">D<"/B>/"B9nnr $D<<9E9"A/C1E9/!rnnr $D/@911:"/D+/>nnr n$D<<9E9"A<"D/19nn $D<<9E9"AC<@9//!1rnn 1",1:5<:"/D+/>nn <9"/5A9/!5>9nn r/5A9/!5>9"AC<@9//!1nnn $D<<9E9"A5<9"A/D!9$1/>nn $D<<9E9"A$:9//nn "!D+!1:D/!91DD1:/>nn

PAGE 208

189 Rank 0)")(+,Weighted <9",D!!5<$/99Bnn nn<9"@D"/>"@"11<9$/99Bnn n<5+>9+J9,,9/!D+nn n15@9/<5+9/>$C!19/nn nr$D<<9E9"A!/5+E"/Bnn n$D<<9E9"A!1C"/1nn nD/B!:5/9>/"B9nn ;

PAGE 209

190 +r&&1rr r2 Rank 0)"(+,Weighted $D<<9E9"AD+/D9rn 15@9/<5+9/>:D/!91DD1:/>n $D<<9E9"A:D/!91DD1:/>nnr r<9":"/D+/>nr $D<<9E9"A/D!9$1/>rrn @D"/>"/:"/D+/>rn !:59<>!!1!!1/D!9$1/>rrn $D<<9E9"A >/"B9/> rn nn n 15@9/<5+9/> :"/D+/> r nn $D<<9E9"A:"/D+/>nn 15@9/<5+9/>>/"B9/>nnn $D<<9E9"A1/5<@/>nn r$D<<9E9"A<"C/9!!1:D/!91DD1:/>nnn !:59<>!!1 :"/D+/>nn $D<<9E9"A@D"/>"/D!9$1/>nn !:59<>!!1!"<>/"B9/>nn $D<<9E9"ADC+1"5+rnnn <9"C<@9//!1rnnn !:59<>!!1>/"B9/>nn r$D<<9E9"A,DD1:5<!!19<5G"@91:!1nn <9">/"B9/>nn <9"/D!9$1/>nn !:59<>!!1/"5+1/99nn n$D<<9E9"A!"<nnr $C//:"/D+/>nnr !:59<>!!1C<@9//!1nnnr rG59E<9/:"/D+/>nnr $D<<9E9"A1/DC1"+rnnr 1",1:5</D!9$1/>nnr J,B:"/D+/>nnr $D/@911:"/D+/>nn $D<<9E9"A$D9<5G"@91:!1nn r1",1:5<:D/!91DD1:/>nn r$D<<9E9"AB9+!5+E1D+nn r$D<<9E9"A:"/A"/>nnn rr$D<<9E9"A,D!!5<$/99Bnnn r/5A9/!5>9"AC<@9//!1nnn rJ,B:D/!91DD1:/>nrnn r$D<<9E9"A@D$B"+nrnn r1",1:5<C<@9//!1nnn

PAGE 210

191 Rank 0)"(+,Weighted r<9"!1C"/1nnn n<9"/5A9/!5>9nnnn <9"9++D$Bnn $51"/B9<5G"@91:!1nnn $D<<9E9"A/C1E9/!nn r/95+E1D+C<@9//!1nn !+D9!":"/D+/>nn "+:"11"+:D/!91DD1:/>nn J,B@D"/>"15@9/DD>nn n$D<<9E9"A51B5+nn 15@9/<5+9/>$"/5@DCnnn <9"D"B/5>E9nn 9<>/CC<@9//!1nn r:51$D@/D!9$1nn 15@9/<5+9/>A9/D+1nn <9"9<5G"@91:!1nn D/1:5+E1D+>/"B9nn !:59<>!!1/D<<"+>DD/9nn "!D+!1:D/!91DD1:/>nnn n$D<<9E9"A$:9//nn /5A9/!5>9"A/D!9$1/>nn $D<<9E9"A5<:"/D+/>rnn r$D<<9E9"A<"D/19rnn G59E<9//D$B$/99Brnn !1DA9/>/"B9rnn /9!9"/$:39">D<"/B>/"B9rnn $D<<9E9"A"E+D<5"rnn $D<<9E9"A"<93J9,,9/!D+rnn n/95+E1D+/D!9$1rnn $D<<9E9"AD<5A9rnn :D/!91DD1:/>nn 15@9/<5+9/>+"+$E/"nn /5A9/!5>9"ADC+1"5+nn <9">D$1D/!<+nn <">DD+:"/D+/>nn n$D+!151C15D+9<5G"@91:!1rnn !1"/,rnn J,B1/DC1"+nn !:59<>!!1/D$BDC+1"5+nn r:9>@99C<@9//!1nn 1",1:5<A"<<9,D/E9nn <9"!DC1:/5>E9nn 1/">515D+:D/!91DD1:/>nn <9"/D@9/1!D+nn

PAGE 211

192 Rank 0)"(+,Weighted $D<<9E9"A!1C"/1nnn nn$D<<9E9"A!/5+E"/Bnnnn n15@9/<5+9/>@"11<9$/99Bnnn n15@9/<5+9/>$C!19/nnn n<9",D!!5<$/99Bnnn nr<9"@D"/>"9+J9,,9/!D+nnn nD/B!:5/9>/"B9nn ;

PAGE 212

193 +r&';r2)r) r1r rrrB2 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 <9":"/D+/>nnnn n 15@9/<5+9/>:D/!91DD1:/>nnnnnn n $D<<9E9"A1/5<@/>rnn nr r@D"/>"/:"/D+/>nrnn n !:59<>!!1>/"B9/>rrnnn n !:59<>!!19<5G"@91:!1rrnn n !:59<>!!1!"</D!9$1/>nrn nn n!:59<>!!1:"/D+/>rnrnnn nr !:59<>!!1C<@9//!1nnn n !:59<>!!1/D!9$1/>nnn nr 1",1:5<:D/!91DD1:/>nnrn rrn r$51"/B9<5G"@91:!1nnnn nnn 1",1:5</D!9$1/>nrrn nnr $D<<9E9"ADC+1"5+rrrrnn n !:59<>!!1:D/!91DD1:/>rnnrn rn "!D+!1:"/D+/>nrnnn nrn $D<<9E9"A:"/D+/>nrnrn nrn n$C//:"/D+/>rnnn nr <9">/"B9/>rrnnrnn 1",1:5<>/"B9/>rnnnnnr n 1",1:5<C<@9//!1nnnrnr rn rG59E<9//D$B$/99Bnnnnn r !:59<>!!1>/"B9/>nnn nn

PAGE 213

194 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 J,B@D"/>"!!1/D<<"+>DD/9nnnnn nr 15@9/<5+9/>+"+$E/"nnnnnnn nr $D<<9E9"A,D!!5<$/99Bnnnrnnnn nnnrn n<9"C<@9//!1rnnnrnn $D<<9E9"AC<@9//!1nn nnn J,B:"/D+/>rnrrnnn <9"D"B/5>E9nnrrnnrn n r$D+!151C15D+9<5G"@91:!1nnnrnn nnrn J,B1/DC1"+rnnnnnnn 9<5G"@91:!1rnnnn nr <9"!1C"/1rnnnnnnn 9<>/CC<@9//!1rnnrrnnrnn rn<9"!DC1:/5>E9rnrnnnnr n rD/1:5+E1D+>/"B9nnnnrrrn r15@9/<5+9/>$"/5@DCrnnnrn nr r1",1:5<A"<<9,D/E9nnn nn rr/5A9/!5>9"ADC+1"5+nnnrn nnn r!:59<>!!1/D$BDC+1"5+nnnn nr rD/B!:5/9>/"B9rnnnnnn n r15@9/<5+9/>A9/D+1nnnnnrr nr r<9",D!!5<$/99Brnnnnnn nrn r<9"9++D$Bnnnnrnrnn n n/95+E1D+C<@9//!1nnnnnr nrr <9"@D"/>"
PAGE 214

195 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 :51$D@/D!9$1nrnrnrrn nn <9"/D@9/1!D+nnnnnnnrnr r r"+:"11"+:D/!91DD1:/>nnnn n 1/">515D+:D/!91DD1:/>nnrnrnrr nn !1"/,nnnnn nnn 15@9/<5+9/>@"11<9$/99Bnnnnrn rnnn G59E<9/$DC+$5<1/99nnnrnnrn nnn :9>@99C<@9//!1nnnnnrrn n<9"/D!9$1/>nnnnn r <9"/5A9/!5>9nnrnnn n $9+1/9/D!9$1nnrnnn n !1DA9/>/"B9nnnrrrnnn r1",1:5<:"/D+/>nnnnnrnrrr nn /95+E1D+/D!9$1nnrrnrnrrn nn $D<<9E9"AD<5A9nnrrnrnrrnn n $D<<9E9"A!B"rnnrnnnn $D<<9E9"A$:9//rnnnnnnnnn n $D<<9E9"A<"D/19rnnnnnrrn n<9"9<5G"@91:!1nnnn /5A9/!5>9"AC<@9//!1nnn nr $D<<9E9"A"E+D<5"nnnnrnn nnn $D<<9E9"A5<9+J9,,9/!D+nnnnnrnrnn <9">D$1D/!<+nnnnnnnn /9!9"/$:39">D<"/B>/"B9nnnrn nnn !1"+,D/>:D/!91DD1:/>nnrnnrn n

PAGE 215

196 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 15@9/<5+9/>15@9/DD>nnnnnn nnnr $D<<9E9"A"<93J9,,9/!D+nrnnrn nnn n!:59<>!!1/"5+1/99rnnnrnr n "!D+!1:D/!91DD1:/>rnnnn n J,B:D/!91DD1:/>nnnnn /5A9/!5>9"A/D!9$1/>rnnn nn r15@9/<5+9/>$C!19/nnnnn n $D<<9E9"AD+/D9rnnnnrn nnn $D<<9E9"A >/"B9/> nn n r nr r 15@9/<5+9/> :"/D+/> n n n n n n $D<<9E9"A:D/!91DD1:/>nnr n $D<<9E9"A,DD1:5<rrnnnn nn $D<<9E9"A<"C/9rnnnnnr n $D<<9E9"A@D$B"+rnnrrnn nr r!+D9!":"/D+/>nnnnn n G59E<9/:"/D+/>rnnnn nr $D<<9E9"AB9+!5+E1D+rnnnnnrn n $D<<9E9"A@D"/>"DD+:"/D+/>nnnn nrr n$D<<9E9"A1/DC1"+nrnnrnn rn n$D<<9E9"A51B5+rnnnrn nr

PAGE 216

197 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 nr$D<<9E9"A!/5+E"/Brrnnnnnn rn n$D<<9E9"A/D!9$1/>rrr nnn n$D<<9E9"A!1C"/1rnnnnnn n 1#nn "*n ;

PAGE 217

198 +r&(1rr )r)rrr &30 580)"(5*$ G59E<9//D$B$/99Bnn $D<<9E9"AD+/D9nnn !:59<>!!1C<@9//!1nn r$D<<9E9"A1/5<@/>nn <9":"/D+/>nn 15@9/<5+9/>:D/!91DD1:/>nn 15@9/<5+9/>>/"B9/>nnn !:59<>!!19<5G"@91:!1nnr !:59<>!!1nn <9"/5A9/!5>9nn 1",1:5<:D/!91DD1:/>nnr $D<<9E9"ADC+1"5+nnr r1",1:5</D!9$1/>nnr J,B@D"/>">/"B9/>nn /5A9/!5>9"AC<@9//!1nrnr $D<<9E9"AC<@9//!1nnn $D<<9E9"A:D/!91DD1:/>nnn n<9">/"B9/>nn !:59<>!!1:"/D+/>nrn /95+E1D+C<@9//!1nrn "!D+!1:"/D+/>nrnn r1",1:5<9<5G"@91:!1nn $D<<9E9"A,DD1:5<"/:"/D+/>nnn 15@9/<5+9/>/D!9$1/>nnn 1",1:5<:"/D+/>nnnn G59E<9/:"/D+/>nnn n1",1:5<C<@9//!1nnn $C//:"/D+/>nnn <9"C<@9//!1nnnn <9"D"B/5>E9nnnn r$D<<9E9"A5<!!1/D!9$1/>nnn "!D+!1:D/!91DD1:/>nnn /5A9/!5>9"A/D!9$1/>nnnn

PAGE 218

199 &30 580)"(5*$ $D<<9E9"A$:9//nnnnn J,B1/DC1"+nnnn rn$D<<9E9"A<"C/9nnnr r!:59<>!!1>/"B9/>nnnrr r<9"/D!9$1/>nrnnr rr!:59<>!!1!"<nnn r$D<<9E9"A<"D/19nnn r!:59<>!!1:D/!91DD1:/>nnn r9<>/CC<@9//!1nnn r"+:"11"+:D/!91DD1:/>nnn nnnn !1DA9/>/"B9nnnnn G59E<9/$DC+$5<1/99nnnn r15@9/<5+9/>@"11<9$/99Bnnnn $9+1/9/D!9$1nnnn /5A9/!5>9"ADC+1"5+nnnn $D+!151C15D+9<5G"@91:!1nnrnn 15@9/<5+9/>15@9/DD>nnnnn <9"!1C"/1nnnnn n$D<<9E9"AD<5A9nnnnn $D<<9E9"A/D!9$1/>nnnnn $51"/B9<5G"@91:!1nnnnn $D<<9E9"A"E+D<5"nnnnn r/95+E1D+/D!9$1nnnnn :51$D@/D!9$1nnnnn <9"!DC1:/5>E9nnrnn <9",D!!5<$/99Bnnrnn /9!9"/$:39">D<"/B>/"B9nnnn <9"9++D$Bnnnnn n$D<<9E9"A"<93J9,,9/!D+nnnn !1"/,nnnn D/B!:5/9>/"B9nnnn $D<<9E9"A1/DC1"+nnnr r<9"@D"/>"$"/5@DCnnn J,B:D/!91DD1:/>nnn

PAGE 219

200 &30 580)"(5*$ $D<<9E9"A!"<"/"B9nnnr 1",1:5<A"<<9,D/E9nnnr 15@9/<5+9/> :"/D+/> n nn r15@9/<5+9/>+"+$E/"nnn !1"+,D/>:D/!91DD1:/>nnn <9">D$1D/!<+nnn $D<<9E9"A/C1E9/!nrnn <9"9<5G"@91:!1nnnr !:59<>!!1/"5+1/99nnn n$D<<9E9"A!B"nnn :9>@99C<@9//!1nnnn 1/">515D+:D/!91DD1:/>nnn <5+>9+J9,,9/!D+nnnnn r$D<<9E9"A51B5+nrnn $D<<9E9"A:"/A"/>nnn !:59<>!!1/D<<"+>DD/9nnn 15@9/<5+9/>A9/D+1nnn $D<<9E9"A >/"B9/> nr nr !:59<>!!1/D$BDC+1"5+nrnn nn<">DD+:"/D+/>nrrn n$D<<9E9"A@D$B"+nrn n<9"/D@9/1!D+nrnrr n$D<<9E9"A$D$C!19/nn n$D<<9E9"A!1C"/1nn n$D<<9E9"A!/5+E"/Bnnnn ;
PAGE 220

201 +r&1rr )r)rr B2 &30 58 0)"(39 *$ $D<<9E9"AD+/D9nnnn <9":"/D+/>nn 15@9/<5+9/>:D/!91DD1:/>n r$D<<9E9"A1/5<@/>rn $D<<9E9"A:D/!91DD1:/>nr 15@9/<5+9/>>/"B9/>n !:59<>!!19<5G"@91:!1n !:59<>!!1C<@9//!1n $D<<9E9"AC<@9//!1nn n!:59<>!!1nn 1",1:5<>/"B9/>nn <9">/"B9/>nn r<9"/5A9/!5>9nnnn /5A9/!5>9"AC<@9//!1nnr 1",1:5</D!9$1/>rnn $D<<9E9"ADC+1"5+nn $D<<9E9"A,DD1:5<:D/!91DD1:/>nn n15@9/<5+9/>/D!9$1/>nn @D"/>"/:"/D+/>nn !:59<>!!1:"/D+/>nn "!D+!1:"/D+/>nn r<9"C<@9//!1nn $C//:"/D+/>rnn 1",1:5<9<5G"@91:!1rnnn J,B@D"/>"rnn !:59<>!!1/D!9$1/>rnn $D<<9E9"A:"/D+/>rnnn !:59<>!!1>/"B9/>nnn r$D<<9E9"A<"C/9nn 1",1:5<:"/D+/>nn

PAGE 221

202 &30 58 0)"(39 *$ 1",1:5<C<@9//!1nnr <9"/D!9$1/>nn rn$D<<9E9"A$:9//nn r/5A9/!5>9"A/D!9$1/>nn r!:59<>!!1:D/!91DD1:/>nn rJ,B:"/D+/>nn rr<9"D"B/5>E9nnn r!:59<>!!1!"<nn r!+D9!":"/D+/>nn r$D<<9E9"A<"D/19nn r9<>/CC<@9//!1nnr n/"B9nnn J,B1/DC1"+nnn $9+1/9/D!9$1nnn rG59E<9/$DC+$5<1/99nnnn 15@9/<5+9/>@"11<9$/99Bnnnn /5A9/!5>9"ADC+1"5+nnnn 15@9/<5+9/>15@9/DD>nrnnn $D+!151C15D+9<5G"@91:!1nnnn <9"!1C"/1nnnn n$D<<9E9"AD<5A9nnnn $51"/B9<5G"@91:!1nnnn <9"!DC1:/5>E9nnnn <9",D!!5<$/99Bnnnn r$D<<9E9"A/D!9$1/>nnnn $D<<9E9"A"E+D<5"nnnn /95+E1D+/D!9$1nnnnr :51$D@/D!9$1nnnnr D/B!:5/9>/"B9nnnn /9!9"/$:39">D<"/B>/"B9nnnn n!1"/,nnnn <9"@D"/>"A"<<9,D/E9nn 15@9/<5+9/>$"/5@DCrnn D/1:5+E1D+>/"B9nn

PAGE 222

203 &30 58 0)"(39 *$ J,B:D/!91DD1:/>nn $D<<9E9"A,D!!5<$/99Bnnn $D<<9E9"A1/DC1"+nn n<5+>9+J9,,9/!D+nn $D<<9E9"AB9+!5+E1D+nn 15@9/<5+9/>+"+$E/"rnnn $D<<9E9"A!"<D$1D/!<+nn !1"+,D/>:D/!91DD1:/>nn :9>@99C<@9//!1nn 1/">515D+:D/!91DD1:/>nnr $D<<9E9"A@D"/>"!!1/"5+1/99nnr $D<<9E9"A/C1E9/!rnn !:59<>!!1/D<<"+>DD/9rnn r15@9/<5+9/>A9/D+1rnnn 15@9/<5+9/> :"/D+/> n nn !:59<>!!1/D$BDC+1"5+nn $D<<9E9"A51B5+nn <9"/D@9/1!D+rnn $D<<9E9"A:"/A"/>nn nn<">DD+:"/D+/>nn n15@9/<5+9/>$C!19/nnr n$D<<9E9"A@D$B"+nn n$D<<9E9"A$D/"B9/> r nr ;
PAGE 223

204 +r&;r21r r2) 0;03$55$ 5B4$4." -A+-+ <9":"/D+/>nnnn 15@9/<5+9/>:D/!91DD1:/>nrnn nnn $D<<9E9"A1/5<@/>nnr r@D"/>"/:"/D+/>nrrn !:59<>!!1>/"B9/>nrrrrrn !:59<>!!19<5G"@91:!1nrr n !:59<>!!1!"</D!9$1/>n n n !:59<>!!1 :"/D+/>nnrrn !:59<>!!1C<@9//!1nnn n !:59<>!!1/D!9$1/>nrnn n 1",1:5<:D/!91DD1:/>nnnr nn r$51"/B9<5G"@91:!1rnnrnn 1",1:5</D!9$1/>nnnr n $D<<9E9"ADC+1"5+nnrrr !:59<>!!1:D/!91DD1:/>nnr n "!D+!1:"/D+/>nnnrn $D<<9E9"A:"/D+/>nnn n$C//:"/D+/>nnrn <9">/"B9/>nnnrrrnn 1",1:5<>/"B9/>nnrnrnnn 1",1:5<C<@9//!1nn nn rG59E<9//D$B$/99Bnnnnnnn !:59<>!!1>/"B9/>nrrnnn J,B@D"/>"!!1/D<<"+>DD/9nnnrnnn 15@9/<5+9/>+"+$E/"nnnnnnn n $D<<9E9"A,D!!5<$/99Bnnnrnn n n<9"C<@9//!1nrrn $D<<9E9"AC<@9//!1nr J,B:"/D+/>nnrrn <9"D"B/5>E9nnnrnn

PAGE 224

205 0;03$55$ 5B4$4." -A+-+ r$D+!151C15D+9<5G"@91:!1nnnnnnnn n J,B1/DC1"+rnnnrnn 9<5G"@91:!1nnnr rn <9"!1C"/1nnrn 9<>/CC<@9//!1nnrrrnn rn<9"!DC1:/5>E9nnnnnrrn rD/1:5+E1D+>/"B9nnnrnnn r15@9/<5+9/>$"/5@DCnrnrnn r1",1:5<A"<<9,D/E9nnnnn n rr/5A9/!5>9"ADC+1"5+nnnnn r!:59<>!!1/D$BDC+1"5+nnnrn n rD/B!:5/9>/"B9nnnnnnrn r15@9/<5+9/>A9/D+1nrnrnnn r<9",D!!5<$/99Bnnnnrnn r<9"9++D$Bnnnnn n/95+E1D+C<@9//!1nnnnn <9"@D"/>"nrnn n 1/">515D+:D/!91DD1:/>nnnnnrnn !1"/,nnnrnnn 15@9/<5+9/>@"11<9$/99Bnnnnnnn n G59E<9/$DC+$5<1/99nnnnnn :9>@99C<@9//!1nnrnnn n<9"/D!9$1/>nnnn <9"/5A9/!5>9nrnrnn $9+1/9/D!9$1nnnrn !1DA9/>/"B9nrnn r1",1:5<:"/D+/>nnnrnn n /95+E1D+/D!9$1nnrrnn $D<<9E9"AD<5A9nnnnrnn $D<<9E9"A!B"rnrnrrnn

PAGE 225

206 0;03$55$ 5B4$4." -A+-+ $D<<9E9"A$:9//nnnnnrn $D<<9E9"A<"D/19nnnnrn n<9"9<5G"@91:!1rnnrn /5A9/!5>9"AC<@9//!1nr n $D<<9E9"A"E+D<5"nnnnn $D<<9E9"A5<9+J9,,9/!D+nnnnnnnnn <9">D$1D/!<+nnnnnnnnn /9!9"/$:39">D<"/B>/"B9nnnn nn !1"+,D/>:D/!91DD1:/>rnnnn 15@9/<5+9/>15@9/DD>nnn nn $D<<9E9"A"<93J9,,9/!D+rnnnrn r n!:59<>!!1/"5+1/99rnnrrnn "!D+!1:D/!91DD1:/>nrnrn J,B:D/!91DD1:/>nnnrnn /5A9/!5>9"A/D!9$1/>nnnr rn r15@9/<5+9/>$C!19/nnnnnnn $D<<9E9"AD+/D9rnrrrn $D<<9E9"A >/"B9/> r n nn 15@9/<5+9/> :"/D+/> nn r n $D<<9E9"A:D/!91DD1:/>nn n $D<<9E9"A,DD1:5<nrnrrnn $D<<9E9"A<"C/9nnnrrnn $D<<9E9"A@D$B"+rnnrrn r!+D9!":"/D+/>nnnnn G59E<9/:"/D+/>nnnrn $D<<9E9"AB9+!5+E1D+nnrnrn $D<<9E9"A@D"/>"DD+:"/D+/>nnnnnnn

PAGE 226

207 0;03$55$ 5B4$4." -A+-+ n$D<<9E9"A1/DC1"+nrnrrnrnn n$D<<9E9"A51B5+nnrrn nr$D<<9E9"A!/5+E"/Brnnnnrrnn n$D<<9E9"A/D!9$1/>nrn n$D<<9E9"A!1C"/1nnnnrnnn 1#n "*n ;
PAGE 227

208 +r' 1rr r 5803+$5*$ $D<<9E9"A1/5<@/>nnnn !:59<>!!1/D!9$1/>nnnn 15@9/<5+9/>:D/!91DD1:/>nnnr r$D<<9E9"A:D/!91DD1:/>nnnr $D<<9E9"AD+/D9nnnr $D<<9E9"A@D"/>">/"B9/>nrnnrn $D<<9E9"A!"</"B9/>nnn $D<<9E9"AC<@9//!1nnn @D"/>"/:"/D+/>nnn !:59<>!!1:D/!91DD1:/>nnn r<9"/D!9$1/>nnn /95+E1D+C<@9//!1nnn !:59<>!!1C<@9//!1nnnn 1",1:5<>/"B9/>nnnn $D<<9E9"A/D!9$1/>nrnn <9"/5A9/!5>9nrnn n!:59<>!!1>/"B9/>nrnn $D<<9E9"A >/"B9/> n nnr !:59<>!!19<5G"@91:!1nnn $D<<9E9"A:"/D+/>nnn r<9"C<@9//!1nnn $D<<9E9"AB9+!5+E1D+nnn $D<<9E9"A/C1E9/!nnn $D<<9E9"A51B5+nnn !:59<>!!1 :"/D+/>nnn 1",1:5<:D/!91DD1:/>nnn n1",1:5<C<@9//!1nnn !:59<>!!1/D!9$1/>nnn 1",1:5<9<5G"@91:!1nnn r<9":"/D+/>nnn 15@9/<5+9/>/D!9$1/>nnn 15@9/<5+9/>15@9/DD>nnn !:59<>!!1!"<"E9nnnr rn<9"!1C"/1nnnr r9<>/CC<@9//!1nnnr r/95+E1D+/D!9$1nnnr r$D<<9E9"A<"D/19nnnr rr$D<<9E9"A,DD1:5<nnnn r$C//:"/D+/>nnnn r$D+!151C15D+9<5G"@91:!1nnnn r$9+1/9/D!9$1nnnn

PAGE 228

209 5803+$5*$ r$D<<9E9"A$:9//nnnn nJ,B:D/!91DD1:/>nnnn 15@9/<5+9/> :"/D+/> nn nn $D<<9E9"A<"C/9nnnn r!+D9!":"/D+/>nnnn G59E<9/:"/D+/>nnnn $D<<9E9"ADC+1"5+nnnn J,B:"/D+/>nnnn <9"9++D$Bnnnn /5A9/!5>9"AC<@9//!1nnnn n$D<<9E9"A5<nnnn $D/@911:"/D+/>nnnn 15@9/<5+9/>$"/5@DCnrnnn r15@9/<5+9/>A9/D+1nrnnn :51$D@/D!9$1nrnnn "+:"11"+:D/!91DD1:/>nrnnn !1DA9/>/"B9nrnnn $51"/B9<5G"@91:!1nnnn J,B1/DC1"+nnnn n$D<<9E9"A!B"nnnn <9"9<5G"@91:!1nnnn !1"+,D/>:D/!91DD1:/>nnnn $D<<9E9"A"<93J9,,9/!D+nnnn r!:59<>!!1/"5+1/99nnnn $D<<9E9"A@D$B"+nnnn $D<<9E9"A!/5+E"/Bnnnn G59E<9//D$B$/99Bnnnnn E9nnnnn nD/1:5+E1D+>/"B9nnnnn /5A9/!5>9"ADC+1"5+nnnnn G59E<9/$DC+$5<1/99nnnnn 1",1:5<:"/D+/>nnnnn r$D<<9E9"AD<5A9nnnnn <9">D$1D/!<+nnnnn /9!9"/$:39">D<"/B>/"B9nnnnn /5A9/!5>9"A/D!9$1/>nnnnn $D<<9E9"A$D!!1/D<<"+>DD/9nnnnn 1",1:5<A"<<9,D/E9nnnnn 1/">515D+:D/!91DD1:/>nnnnn !1"/,nnnnn r:9>@99C<@9//!1nnnnn $D<<9E9"A"E+D<5"nnnnn 15@9/<5+9/>+"+$E/"nnnnn $D<<9E9"A,D!!5<$/99Bnnnnn !:59<>!!1/D$BDC+1"5+nnnnn

PAGE 229

210 5803+$5*$ <9",D!!5<$/99Bnnnnn nn<5+>9+J9,,9/!D+nnnnn nD/B!:5/9>/"B9nnnnnnn n<9"@D"/>"@"11<9$/99Bnnnnnnn n15@9/<5+9/>$C!19/nnnnnnn n<">DD+:"/D+/>nnnnnnn ;
PAGE 230

211 +r';r2*r r2rrrnr n n+,#@#*5(nrnn r EA##&/@#*+:'#&/ nn +A(0%.@#*9r"*nn +$#@#*9r"*rnrn rr"*+G%!rnrnrnnr r"*+(!nnnnrnr r +!@#*5(nnnnr r +!@#*r"*nn r +!9r"*rrnn +:*!9r"*rrnrn r

PAGE 231

212 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, n+(!5(nn 9r"*+>##(!nnnr +I%2!+!$/nn r +A(0%.@#*+!#!nn n r+(!9r"*nnrnnr +$#@#*5(nrn r +>#!9!)#+>rn n 9r"*+J()!rrr 9r"*+!#!r 9r"*+$#&!nn rn9r"*+!rrnnnrr r+,#@#*r"*rnr rr"*+(!nnnnr nn rr"*+G%!rrnrrrnn r rr+(!5(nnnnn r rr"*+<;##@#*nnr r r+>#!9!)#!%>nr r+1&(!r"*nnnr r r+!@#*r"*nn r+I%2!5(nnn rnr n+:*!9r"*nnrrn +$#@#*5(nn nnnnn +(!5(nn r +!#!9r"*nrrnnn r9!)#+>+!n +:##&!9!)#+>n +!9r"*rnnnnr nr +(!9r"*nnnnn nr +!9!)#!%>n nrr 9!)#!%>+:##&!n n+I%2!5(nrnn

PAGE 232

213 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, +:*!5(nnn r +$'2(/9r"*rrrn rr rr"*+<;##@#*r rrr"*+5*!rnn +,#@#*rr"*nr n +1&(!rr"*rrrn r +(!/(2&$rn r +!@#*rr"*rnr +1;/9r"*nrrrn n n+!5(nnnrnr rnrn !"($rnr +(!r"*nrnn 9rn"*+$'2(/rnr nn r+!5(nrn +:*!9rn"*nrn n 9!'/+!nrnnrr +$#@#*9rn"*nnn rn +!#!9rn"*nnn rrn 9rn"*+!rnnnn n n+@@#*!rrn +,#@#*r("*nrn +!@#*r("*rn +I%2!9!'/rnr r+!9"*rnrnr r #%!!n "*+<;##@#*rrnn n +>;!#%!nnnnnrr "*+5*!rrr n +<)!"*rrnn n"*+&!nnrnr r "*++*L!n

PAGE 233

214 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, "*+(!nnnn nn "*+1L!rrnrnr rr r"*+G%!nrn "*+,-!rn +,#@#*"*rnrrn "*+1&(!nn rr "*+$#&!rrnr +!@#*"*n rn nn+!9"*nnnrr r n +I%2! 9"* nn nnn rn n rn n n"*5(n n+I%2!9"*nrn r nr+$#@#*9"*rrrn r n"*+E#2*##/n n+,#@#*"*nnn nrr n+,#@#*"*nnrn rrr n9<%B@#*+I%2!nrn nnnn n9<%B@#*+!nrr rrn n+$#@#*9<%B@#*r nrnn "*+,#@#*nnn "*+!@#*nn rn 9("*+!nrrn rrr r+,#@#*+!)@#*rrnn rn +@;&@#!nnnrnn r +$#@#*9"*nnn rnrnn +<;##@#*"*rnn +,#@#*"*rn n +!)@#*"*rrnnn rr n"*+5*!rn +!@#*"*nnrn rrr !$#!nnrr r

PAGE 234

215 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, 9"*+!nnn n r"*@#!rnrnrr +@;&<'!rnnrn !#!nr r @#!!rnn +I%2!9"*rrnn +!9"*n n+$#@#*9"*rnrn r !<'!rrnrrnnn +,#@#*"*nrn rrr "*+5*!rrrn nn r9"*+>;!rrnn +!9"*nrrrrnr n!@#!nnrrnr !<;!nrnr rr n!!nrrnrnrnr !")!rrn rn!@#!rrn r r!+@;&nrrnr r!@#!rnrnnrr r+!)@#*9#$nrnnr rrn!<;!rnnnr n r!!rnr n r+@;&$')!n r+I%2!9"*nnr r+$#@#*9"*rnn rr r!$%(!nnrn n"*1'#rnn ")!!nnnr r +!)@#*@#!nn n!#!rrnrr rr

PAGE 235

216 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, r9*;@#*+I%2!rnrn rrr +!)@#*"%;&n rrr +$#@#*9*;@#*nn nrrn +@;&#!rnrnn r "*9"*nrn rn +<#!9"*nrnr n+@;&9"*rrnnnr !$')!rnr !#!rn $#!!rrn r9"*+,#!nrr +$#(!9"*nn !!%!nnr +('!9"*rnr +!9"*rnnr 1'#!nr n+$#@#*9"*rnn r +,#@#*"*nn r +!@#*"*nnn r #!!nnnrr r+@;&9"*rnr 9"*+>;!rrnr "*9"*rnnr +<#!9"*nn !1'#nr E#'#r!nrrr n9"*+!n 9$#-"*+I%2!rrnr nrnr 9$#-"*+!rrrn +$#@#*9$#-"*rnn nnr r9$#-"*+9#.2!nr

PAGE 236

217 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, $#-"*+5*!nrrn $#-"*+B#'!n $#-"*#!rnnnrr r 9$#-"*+!rrrnr +!@#*$#-"*rnrr r n9$#-"*+(!rrnr nrrr 9$#-"*+;!rnr n9r"*+#!rnrn n9r"*+(!n n9r"*+;!nnn 9"*+(!n n r+$#@#*9"*nn r +
PAGE 237

218 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, 9"*+E!rnn +<#!9"*nr +$#@#*9"*rrnrn +<#!9"*rnr rrnr n+,#@#*n"*nnnn r n"*+B-$rnn 9"*+>;!nnnnn +!)@#*+@!rn nrrrn r+I%2!9"*nrrrr r +!9"*rrnrr r +$#@#*9"*rrn 9"*+$!nn 9"*+$#(!nrrn "*+@;&n n"*+!,>n r +B#'!"*rnr 9"*+!rnnr r 9"*+$#@#*nnn r 9"* +<#! n r "*+@;&nnr 9"*+$!nnr +$#@#*9"*nn n +@;&"*nn 9!)@#*+$!rnrr rn+@;&("*nn r+C*(&@#*9("*rrr r r9("*+!%#!nnr r+$#@#*9("*n rr+,#@#*("*nrn r+!@#*("*rnn r9("*+!#!rnnr nn

PAGE 238

219 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, r!!#!9@&%"*rnn rnr r!$#@#*9"#'"*rr nnr r9"#'"*9,'%>n n9"#'"*!I%2!nnn 9"#'"*!:*!nnn rr 9"#'"*<(#>nnrrnr 9"#'"*!!rrn nrnr r!C*(&@#*9"#'"*rrn nn 9"#'"*!>;!nr n 9"#'"*!<#!nn "#'"*!B#'!nrnnrn r !@;&"#'"*rnr 9"#'"*!(!rnrnn rrr n"#'"*!#/*>nr rr "#'"*!!@#*rn r "#'"*!%'!nn rn "#'"*!&!rnr n r"#'"*!B-$rnr n !,#@#*"#'"*nrrnn nrrr <(#>!:##&!nrn n !$#@#*9$&$+>nn nrnr !,#@#*A"*n r !!<(#>rrn rr n!$#@#*9D"*rn !@;&9D"*nnr n !@;&9D"*nn <(#>!D!nnn r r!,#@#*B%&"*nrr r !@;&B%&"*nr n (/B%&"*nnn n <(#>!I%2!r rr

PAGE 239

220 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, 9(((())"*!/nnr nr 9(((())"*!$#@#*n nnnn n!!,>(((())"*rn r (((())"*!#/*>rrn rnrnrr !@;&(((())"*rnrrr !,#@#*(((())"*rnr nnnr r!$#@#*9<%("*nnrr nr !$#@#*9"(("*nn n 9,#"*!:##&!rrn r !!,>,#"*nnr rnn !,#@#*,#"*rnrn r ,#"*!5*!rn n!$#@#*95;"*nn !!,>5;"*nrr rr !$#@#*9-"*nr n !,#@#*J;##"*nrrn r!!@#*J;##"*rrrnn nr 9*("*!!@#*rnn rrr !$#@#*99*("*rrn 99*("*!>;!nrn r 99*("*!:!rnn !C*(&@#*99*("* nnn nn!@;&99*("*rn n99*("*!I%2!nnr n!$#@#*9#"*nn r n9:')"*!>&!nnrnnnn rr nr9:')"*!('! r n9:')"*!1'>nn r n9:')"*!!rnn rn n9:')"*!<%(!rnrnn rr

PAGE 240

221 0;0)"-A+-+ 74#n%n3n0n&7B4()(+, (+, n "* +!)@#* n n n +B#'! "* r n 1#nnr ;
PAGE 241

222 +r'*rrr ;!nnr !@;&(((())"*nnr 9r"*+J()!nnr r"*+G%!nn +,#@#*rr"*nn n+(!/(2&$nn +(!r"*nn "*+$#&!nn +('!9"*nn r9"*+$!nn 9"#'"*!(!rnn +!9"*nn +A(0%.@#*9"*nn +:*!9("*nn 9r"*+!#!nn rn9r"*+
PAGE 242

223 580)")(+,*$ rE#'#r!nn rr9"#'"*9,'%>nn rr"*+G%!nn r!,#@#*B%&"*nn r+@;&("*nnr r+!@#*"*nnr r"*+5*!nnr n+@;&<'!nnr 9r"*+#!nnr +$#@#*9"*nnr +E!9r"*nn r+1;/@#*nn 9$#-"*+(!rnn 9("*+!rnn "*5(nn !@#!nn !,#@#*A"*nn n!,#@#*,#"*nn +!)@#*9#$nn ,#"*!5*!nn $#-"*+5*!nn r+!)@#*"*nn +A(0%.@#*+!#!nnn +$#@#*5(nnnn +!@#*r("*nnn "*+(!nnn 9"*+>;!nnn n"#'"*!B-$nnn !@;&9D"*nnn 99*("*!:!nnn 9:')"*!('!rnn r!!rnn +$#@#*9*;@#*rnn 9"*+!&%(!rnn +!@#*$#-"*rnn +,#@#*"*rnn n"*+,#@#*rnn n"#'"*!#/*>rnn !$#@#*9#"*rnn +I%2!9"*rnn 9:')"*!>&!rrnn r"*9"*rnn 99*("*!I%2!rnn

PAGE 243

224 580)")(+,*$ +,#@#*"*rrnn 9*("*!!@#*rrnn +!@#*"*rrnn !<;!rnn n1'#!rnn +$#@#*9"*rnn !!,>,#"*rnn +!9!)#!%>rrnn r!@#!rnn 9"*+J()!rnn 9!)@#*+$!rnn +,#@#*5(nrnn r"*+<;##@#*rnn +1&(!r"*rnn nn+1&(!rr"*rnn n+>;!#%!rnn n"*+&!rnn n!$#!rnn nr9"*+>;!rnn n9"*+$!rnn n!C*(&@#*9("*rnn nr"*+(!rnn n"#'"*!!@#*rnn n+!)@#*r"*rnn n+<#!9"*rnn 9("*+!rrnn !,#@#*"#'"*rrnn +!)@#*+@!rnn r9("*+!%#!rnn 9:')"*!1'>rnn 9:')"*!<%(!rnn "*+!@#*rnnnr !$%(!rnnnr 9$#-"*+I%2!rnnnr n(((())"*!#/*>rnrnnr !"($rnnr 9"*+!rnnr !!,>5;"*rnnr r!$#@#*9$&$+>n nr +@;&"*nnr +$#@#*9r"*nnr +$#@#*9rn"*rnnr 9:')"*!!rnnr

PAGE 244

225 580)")(+,*$ +!9"*nnr n9$#-"*+9#.2!nnr +,#@#*n"*nnr (/B%&"*nnr 9,#"*!:##&!nnr r+!@#*"*nnr !")!nn +:*!9r"*rnn +$#@#*9r"*nn +$#@#*9$#-"*nn +<)!"*nn rn"*+,#@#*nn r+<;##@#*"*nn r+!9"*nn r+I%2!9"*nn rr9"#'"*!>;!nn r+,#@#*"*nn r$#-"*+B#'!nn r+A(0%.@#*9r"*nn r+>#!9!)#+>nn r+!@#*rr"*nn n+<#!9"*nn +!9r"*nn +@@#*!nn +!9";(>nn r!$#@#*95;"*nn n!!nn 9r"*+>;!nn "*+,-!nn 9"#'"*!I%2!nn +!5(nrnn n+!#!9rn"*rnn +,#@#*r"*nn !<'!nn +$#(!9"*nn r!!#!9@&%"*nn 9r"*+!nn 9!'/+!nn 9<%B@#*+!nn "*+<;##@#*nn +<#!9r"*rnn n"#'"*!B#'!rnn +I%2!9!'/nn

PAGE 245

226 580)")(+,*$ +:*!9r"*nn "*+G%!nn r+@;&@#!nn 9"*+!nn !@;&"#'"*nn n!#!rrnn 9*;@#*+I%2!rnn $#-"*!rnn n+I%2!9"*rnn +,#@#*+!)@#*rnnn +!)@#*"%;&rnnn 9"*@#*rnnn r"*+1L!nnn +,#@#*"*nnn #!!nnn 9r"*+!nnn +$#@#*9"*rnnn +1;/9"*nnn n+(!5(nnnn "*++*L!nnn "*+1&(!nnn n!@#!nnn r!!,>(((())"*nnn 9<%B@#*+I%2!nnn 9"* +<#! nnn +I%2!+!$/rnn +(!9r"*nn !1'#nn nn<(#>!I%2!nn n"*+E#2*##/nn nn"*+<;##@#*nn n+$#@#*9"*nn nr!!%!nn n!,#@#*J;##"*nn n!$#@#*99*("*nn n+$#@#*9"*rnn n+,#@#*r("*nn n+!@#*"*nn n"*@#!nn n!<;!nn "* +!)@#* nn 9rn"*+$'2(/nnn r9("*+!#!nnn

PAGE 246

227 580)")(+,*$ !$#@#*9"(("*nnn !C*(&@#*9"#'"*nnn +:*!9rn"*nnn +!9r"*nrnn +(!9r"*nrnn nr"*+(!nnn +$#@#*9"*nnn 9(((())"*!$#@#*nnn +:##&!9!)#+>nn r!#!nn 9$#-"*+!nn +!@#*("*nn !!<(#>nn 9(((())"*!/nn +@;&9"*nn n9$#-"*+!nn +,#@#*r"*rnn !,#@#*(((())"*rnn +,#@#*"*nn r+!)@#*@#!nn 9"#'"*!<#!nn +!9"*nn +!9"*nn +1;/9r"*nn +$#@#*9("*nn rn+@;&#!rnn r99*("*!>;!rnn r9"#'"*<(#>nn r!!@#*J;##"*nn rr+!@#*r"*nn r+>#!9!)#!%>nn r+$#@#*9"*nn r+$#@#*9"*nn r"*+@;&nn r!$')!nn n$#-"*#!nn !$#@#*9-"*nn 9!)#+>+!nn +!5(nnn r+!9"*nn #%!!nn +$#@#*9<%B@#*n n 9"*+,#!nn

PAGE 247

228 580)")(+,*$ +C*(&@#*99*("*nn ++:##&!nn r9rn"*+!nn 9r"*+(!nn +I%2! 9"* nn 9"*+E!nn 9$#-"*+!:##&!nn +,#@#*"*nn r9"*+>;!nn +B#'!"*nn <(#>!D!nn +I%2!9r"*nn +@;&9r"*nn !+@;&nn n$#-"*+)(!nn +$#@#*9"*rnn +(!5(nnn "*+!,>nn r9"#'"*!!nn +I%2!5(nrnn +@;&9"*nnr +,#@#*"*nnr +@;&$')!nnr +<#!9"*nnr n@#!!nnr +!@#*r"*nnr +B#'! "* nnr !#!nnr r+<#!9"*nn !@;&B%&"*nn +!@#*5(nnn +$#@#*5(nnn +!#!9r"*nn +I%2!5(nnn nn+:*!5(nnn

PAGE 248

229 580)")(+,*$ nrr"*+5*!nn n+I%2!9#nn n9r"*+>##(!nn nr9"*+:*!nnn n9"*+I%2!rnnn n+(!9("*nnn n+(!9n"*nnn n9"*+!nnn n9"*+$'2(/nnn ;
PAGE 249

230 +r'$*rrr 2 580)"(+,*$ $#-"*+B#'!nnn <(#>!I%2!rnnr !!<(#>rnnr r(((())"*!#/*>rnnr 9"* +<#! r nnr !,#@#*"#'"*rnnrn +$#@#*9$#-"*rnnrn !C*(&@#*99*("*nnn !,#@#*,#"*rnn n+$#@#*9r"*nn +!9r"*nn 9"#'"*!I%2!nn 9:')"*!1'>nn r+!@#*$#-"*nn "* +!)@#* nn +!)@#*r"*nn +$#@#*9"*nn 9"*+$#@#*nn !$#@#*9$&$+>nn n!$#@#*99*("*nn +C*(&@#*9("*nn "#'"*!B#'!nn +$#@#*9"*nn r+$#@#*9("*nn 9"#'"*!!rnn 9:')"*!('!nn !!,>(((())"*nnr 9$#-"*+I%2!rnnr 9:')"*!<%(!rnnr n9*("*!!@#*rnnr 9("*+!rnnr !$#@#*9"#'"*rnnr !$#@#*9<%("*rnn r$#-"*!nn 9<%B@#*+I%2!nn 9"#'"*9,'%>nn rr"*+<;##@#*nn 9r"*+!#!nn +!9!)#!%>nn rn!@;&(((())"*nn r!$#@#*95;"*nn r9$#-"*+!nn

PAGE 250

231 580)"(+,*$ r9"#'"*<(#>nn rr"#'"*!&!nnnn r"#'"*!B-$nnnn r"#'"*!%'!nnnn r+$#@#*9rn"*nnn r+$#@#*9*;@#*nnn r9:')"*!!nnn n!!,>,#"*nnnn +,#@#*"*nn +$#@#*9<%B@#*nn !,#@#*(((())"*nn r!,#@#*J;##"*nn +,#@#*rr"*nn !@;&9D"*nn !$#@#*9-"*nn "*+5*!nn !@;&"#'"*nn n,#"*!5*!nn +I%2!+!$/nn +$#@#*9"*nn "*+@;&nn r9<%B@#*+!nn "*+@;&nn 9"#'"*!(!rnn !C*(&@#*9"#'"*nn !@;&99*("*nn !$#@#*9"(("*nn nn"*+,#@#*nn 9:')"*!>&!nn 9r"*+J()!nn 9r"*+J()!nn r!$#@#*9#"*nn +!)@#*"%;&rnn "*5(nn +!9"*nn +I%2!9!'/nn +,#@#*n"*nn n!,#@#*B%&"*rnn 99*("*!I%2!nn "*+5*!nnn 9(((())"*!$#@#*nnn r+$#@#*9"*nn "#'"*!!@#*nn

PAGE 251

232 580)"(+,*$ +!@#*"*nn !!@#*J;##"*nn 9*;@#*+I%2!nn "*+(!nnn nn!<;!nnn <(#>!:##&!nnn !<;!rnnr +!)@#*"*rnnr r+$'2(/9r"*rnnr +,#@#*"*rnnr +<#!9r"*rnnr !!,>5;"*rrnnr !!#!9@&%"*rnnr +!9r"*rnnr nn+,#@#*r"*rnnr n+!)@#*+@!rnnnr n"*+,-!nn n+@;&9"*nn nr9(((())"*!/nn n9r"*+!rnn n"*+,#@#*nn n+!9"*nn n!$#@#*9D"*nn n!,#@#*A"*nn n+E!9r"*nn +$#@#*9"*nn +@;&"*nn +!@#*"*nn r!+@;&nn +@;&("*nn +,#@#*("*nn +:*!9r"*nn 9("*+!%#!nn +1;/9"*nn n"*+E#2*##/nn <(#>!D!nn +@@#*!nn $#-"*+5*!nn r!@#!nn 9"*+J()!nn +B#'!"*nn 9!)@#*+$!nn +!9"*rnn

PAGE 252

233 580)"(+,*$ 9$#-"*+(!rnn n+!9"*rnn EA##&/@#*+:'#&/nn +>#!9!)#+>nn !")!nn r9$#-"*+!nn "#'"*!#/*>nn (/B%&"*nnn +,#@#*+!)@#*nnn 9$#-"*+nnnn r"*+<;##@#*nnnn rn"*+B-$nnnn r9rn"*+$'2(/nnnn r9("*+!#!nnnn n!@;&B%&"*nnnn +$#@#*9"*nnnnn n!!nnn 9r"*+>;!nnn r+I%2!5(nnnn +I%2!9"*nnn !$')!nnn 9"#'"*!:*!rnnn +<#!9"*nnn n!@#!nnn n+@;&#!nnn 99*("*!>;!nnn +1;/@#*nnn 9r"*+$#&!nnn r#!!nnn 9("*+!nnn 9"*+$#(!nnn n!#!nnn +!)@#*@#!nnn 9"*+$!nnn n9"#'"*!<#!nnn $#-"*+)(!rnnn

PAGE 253

234 580)"(+,*$ 1'#!nnn 9r"*+!nnn r+(!9r"*nnn 9!'/+!nnn +@;&9r"*nnn +:##&!9!)#+>nnn !#!nnn "*9"*nnn n+!@#*("*nnn +!5(nnnnn +I%2!9r"*nnnn 9r"*+;!nnn 9"*+>;!nnn 9"#'"*!>;!nnn "*@#!nnn r@#!!nnn +!@#*r"*nnn 99*("*!:!nnn +$#@#*9"*nnn !!%!nnn +,#@#*"*nnn nn+I%2!5(nnnn n9"*+E!nnn n9"*+!nnn n+(!9r"*nnn nr!1'#nnn n9rn"*+!nnnn n+<#!9"*nnn n+A(0%.@#*9"*nnn n+:*!9("*nnn n+(!5(nnnn nr"*+<;##@#*nnn +!@#*r"*nnn +
PAGE 254

235 580)"(+,*$ +I%2!9"*nnn 9"*+,#!nnn +$#@#*9r"*nnn +I%2!9"*nnn 9"*+$!nnn n$#-"*#!rnnn +I%2!9"*rnnn "*+G%!nnn 9"*+!nnn r+!@#*r"*nnn +>#!9!)#!%>nnn +$#@#*9"*nnn 9"*+!&%(!nnn !<'!nnn +@;&$')!nnn n+$#(!9"*nnn +,#@#*"*nnn +!)@#*9#$nnn +<;##@#*"*nnn r"*1'#nnn +A(0%.@#*9r"*nnn 9r"*+#!nnn +$#@#*9"*nnn +@;&9"*nnn +!9"*nnn rn9$#-"*+9#.2!nnn r9,#"*!:##&!nnn r!#!rnnn rE#'#r!rnnn rr"*+!@#*nnn r!$%(!nnn r+:*!9rn"*nnn r"*+$#&!nnn r+('!9"*nnn r9"*@#*nnn n+!@#*"*nnn +,#@#*5(nnnn +1&(!rr"*nnn 9"*+>;!nnn r9"*+!nnnn "*9"*nnnn +,#@#*"*rnnn "*+1&(!rnnnr

PAGE 255

236 580)"(+,*$ ")!!rnnnr +$#@#*5(nrnnnr n"*+1L!rrnnnr +,#@#*"*rrnnnr +I%2!9#rnnnr +(!5(nrnnnr r!@#!rnnnr "*+!,>rnnnr +@;&@#!rnnnr r"*+G%!rnnnr +!#!9rn"*rnnnnr n"*+<;##@#*nnnr n+<)!"*nnnr !"($nnn 9!)#!%>+:##&!nnn 9r"*+(!nnn r!$#!rnnn 9!)#+>+!nnn +!5(nnnn #%!!nnn $#!!nnn +A(0%.@#*+!#!nnnn n9"*+>;!nnnn +1;/9r"*nnn +:*!5(nnnn "*+5*!nnn r+@;&<'!nnn r"*+G%!nnn +!#!9r"*nnn +(!/(2&$nnn +,#@#*r("*nnn +$!9r"*rnnn n9"*+:*!nnn +(!5(nnnn "*++*L!nnn +:*!9r"*nnn r+$#@#*5(nnnnn +!9"*nnn 9"*+I%2!nnn +1&(!r"*nnn +>;!#%!nnn "*+&!nnn nnrr"*+5*!nnn

PAGE 256

237 580)"(+,*$ n+!@#*r("*nnn n"*+(!rnnn n+(!r"*nnn nr+(!9n"*nnn n9r"*+>##(!nnn n9"*+$'2(/nnn n9"*+!nnnn n+(!9("*rnnnn n+!@#*5(nrnnnn ;
PAGE 257

238 +r'%;r2)r) *rrrrB 2 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 +1;/@#*rrrnnn nr +1;/9"* 16066 nnr 9"*@#*nnr n r9"*+:*!nnr nn 9"*+!nrr r 9"*+I%2!rnn nr +I%2!9#n nn 9"*+$'2(/nnrn +A(0%.@#*9"*nnn nrrn n"*+(!rnnnn r +,#@#*"*rnnn nnr +!@#*"*nrrnn nnr +:*!9("* nr r9("*+!rnn nn +(!9("*rnnn nn +(!9n"*rnrnn nnn n"*+,#@#*nrn rnn n"*+<;##@#*rrnnr n +!9";(>rnn nnn n+,#@#*5(nnrnnrn n EA##&/@#*+:'#&/nnnrr rr +A(0%.@#*9r"*nnnn nr +$#@#*9r"*rnn nrrn rr"*+G%!rnrnrn r"*+(!nrnr nr

PAGE 258

239 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 +!@#*5(nnnnr rn +!@#*r"*nnn nnr +!9r"*nn n +:*!9r"*rnnr nrn n+(!5(nnr nr 9r"*+>##(!nrnrn r +I%2!+!$/rn n +A(0%.@#*+!#!nnn nr r+(!9r"*rnr nnnnr +$#@#*5(nrnn nr +>#!9!)#+>rnnn nrn 9r"*+J()!r n 9r"*+!#!n n 9r"*+$#&!nnn nn rn9r"*+!nn nr r+,#@#*r"*rnn n rr"*+(!rrn n rr"*+G%!rrrnrrrnn rr+(!5(nnnn n rr"*+<;##@#*nnnr nnr r+>#!9!)#!%>r nn r+1&(!r"*nnrn nr r+!@#*r"*nnn nnnr r+I%2!5(nrnnn nn n+:*!9r"*rnnrrrr nnr +$#@#*5(nnnn nn +(!5(nn nn +!#!9r"*rrnrnr n

PAGE 259

240 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 r9!)#+>+!rn rrnnr +:##&!9!)#+>nn nnnnn +!9r"*nnn nnn +(!9r"*nn nn +!9!)#!%>n nn 9!)#!%>+:##&!rn nrr n+I%2!5(nrnrrn n +:*!5(nrrnn nn +$'2(/9r"*rrrrnr nn rr"*+<;##@#*nnr nn rrr"*+5*!rr n +,#@#*rr"*nnnnn nn +1&(!rr"*rn nn +(!/(2&$rrn r +!@#*rr"*rnnn rrnrr +1;/9r"*rrnnrn nr n+!5(nnrnn rnr !"($rnrn +(!r"*nnnnr nr 9rn"*+$'2(/nn nrr r+!5(nrrnnn n +:*!9rn"*rn nnr 9!'/+!nrrr +$#@#*9rn"*nnnn n +!#!9rn"*n n 9rn"*+!nrnr nn n+@@#*!nn nr +,#@#*r("*rnn nn

PAGE 260

241 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 +!@#*r("*nnnn nn +I%2!9!'/rr rr r+!9"*rrnnnr n #%!!rnnrrn r "*+<;##@#*n nn +>;!#%!nnrnr "*+5*!n nnn +<)!"*rrnn nn n"*+&!rrnn n "*++*L!nnr nr "*+(!nnnn nr "*+1L!nrr n r"*+G%!rnnnnn rnn "*+,-! nr +,#@#*"*rrrn nn "*+1&(!nrrnr nrr "*+$#&!nnn +!@#*"*rnnn n nn+!9"*nnnr rn n +I%2! 9"* rn rn n nn n nn r n"*5(n nrr n+I%2!9"*nrrn n nr+$#@#*9"*rnrrn nn n"*+E#2*##/nnn nnn n+,#@#*"*nnn nn n+,#@#*"*rrnn n n 9<%B@#*+I%2!nnn n9<%B+!rr nrr

PAGE 261

242 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 @#* n+$#@#*9<%B@#*n rnnr "*+,#@#*nn nnn "*+!@#*rnrn n 9("*+!rrnn n r+,#@#*+!)@#*nn nnn +@;&@#!rnrnnn nn +$#@#*9"*nnrn rn +<;##@#*"*nn n +,#@#*"*rnnnnn nnnn +!)@#*"*nrn rn n"*+5*!nn +!@#*"*nnrnnn nr !$#!nrnn r 9"*+!nn r"*@#!rrnrr n +@;&<'!nrnrr rr !#!nn n @#!!nrr nr +I%2!9"*nnnn n +!9"*n nn n+$#@#*9"*rnn rn !<'!nn n +,#@#*"*rnnn nnnn "*+5*!n n r9"*+>;!nn nnnn +!9"*rnnnr nnrnr n!@#!rrn rn

PAGE 262

243 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 !<;!nrr n n!!rnrnnn n !")!rr n rn!@#!nn r!+@;&r r!@#!rnnnrnn r+!)@#*9#$nnn rnnn rrn!<;!rnrnnn rn r!!nnn n r+@;&$')!n r r+I%2!9"*n rrn r+$#@#*9"*rr rr r!$%(!rn n n"*1'#nnrr nn ")!!nnnr nn +!)@#*@#!nnn nnn n!#!nnnr n r9*;@#*+I%2!nrrrrrn rr +!)@#*"%;&nnn nr +$#@#*9*;@#*nrrn nn +@;&#!nnnrn nnrn "*9"*n n +<#!9"*nn nnn n+@;&9"*nnnn n !$')!rrn !#!rnnn r $#!!rrnnrn n r9"*+,#!rnr

PAGE 263

244 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 +$#(!9"*nnr nn !!%!nrrn +('!9"*rrr nnr +!9"*rnrrn nr 1'#!nr n n+$#@#*9"*rnnnn rr +,#@#*"*nnn n +!@#*"*nnn n #!!nnr r+@;&9"*rrn nnr 9"*+>;!rrrnr nnn "*9"*nnn nnrn +<#!9"*nnnr rnr !1'#rn nn E#'#r!nnn nn n9"*+!nrr nnnr 9$#-"*+I%2!nn nrn 9$#-"*+!rrnn rn +$#@#*9$#-"*nnn nrn r9$#-"*+9#.2!rrn rnr $#-"*+5*!rnn nn $#-"*+B#'!nr n $#-"*#!nnrn nnn 9$#-"*+!rrn nrnn +!@#*$#-"*rrnn nrn n9$#-"*+(!rnnn nnnn 9$#-"*+
PAGE 264

245 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 $#-"*+)(!rnnn n r+I%2!9r"*nnr n +!9r"*rnr n 9r"*+J()!r nr 9r"*+!nrrn nr +$#@#*9r"*rnrn n +$!9r"*nn rr nn9r"*+>;!rnr n n9r"*+#!rnn n n9r"*+(!rr rnrnr n9r"*+;!nnnnrn r 9"*+(!rn rnn r+$#@#*9"*nr rrnn +
PAGE 265

246 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 n"*+B-$rr nnn 9"*+>;!nnnnrn n +!)@#*+@!rrnn rnnn r+I%2!9"*rrnr +!9"*nr n +$#@#*9"*rnnn rn 9"*+$!nn n 9"*+$#(!rrrnnr n "*+@;&nn nn n"*+!,>n n +B#'!"*rrnnr rn 9"*+!nnnn n 9"*+$#@#*nnn n r 9"* +<#! n n n "*+@;&nn n 9"*+$!nnn nnr +$#@#*9"*rn nn +@;&"*nnnn nnrn 9!)@#*+$!rnnn nnnnn rn+@;&("*nnn n r+C*(&@#*9("*rrrn nr r9("*+!%#!nnnnnn nnn r+$#@#*9("*nnn nnnn rr+,#@#*("*rrnn n r+!@#*("*nnnrn nr r9("*+!#!rnnnn nn r!!#!9@&%"*rnrrr n r!$#@#*9"#'"*nnn n

PAGE 266

247 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 r9"#'"*9,'%>nn nnnr n9"#'"*!I%2!nnn rr 9"#'"*!:*!nnn nn 9"#'"*<(#>rrrr nrr 9"#'"*!!nnn n r!C*(&@#*9"#'"*rn nnn 9"#'"*!>;!nn nn 9"#'"*!<#!nnrn nnn "#'"*!B#'!nrn nr !@;&"#'"*rrnrn nnn 9"#'"*!(!n nrr n"#'"*!#/*>rn nr "#'"*!!@#*r rnn "#'"*!%'!n rnnrn "#'"*!&!rnn nnnnn r"#'"*!B-$rrrn nn !,#@#*"#'"*nnn nrn <(#>!:##&!nrn !$#@#*9$&$+>rn nr !,#@#*A"*rn rn !!<(#>rrn r n!$#@#*9D"*rn nn !@;&9D"*nrrr !@;&9D"*nnn nn <(#>!D!rn n r!,#@#*B%&"*rn nnr !@;&B%&"*r (/B%&"*nr rnn

PAGE 267

248 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 <(#>!I%2!rrnn rn 9(((())"*!/nrrn n 9(((())"*!$#@#* rnnrr n!!,>(((())"*r r (((())"*!#/*>rrn nr !@;&(((())"*rrn n !,#@#*(((())"*nn rnnnr r!$#@#*9<%("*rr n !$#@#*9"(("* nr 9,#"*!:##&!rrr n !!,>,#"*rn nr !,#@#*,#"*rr nnn ,#"*!5*!rnn nr n!$#@#*95;"* n !!,>5;"*nnrn rnnr !$#@#*9-"*nn nn !,#@#*J;##"*n nnnnn r!!@#*J;##"*rrnrnn n 9*("*!!@#*nr n !$#@#*99*("*rr n 99*("*!>;!rnnrn nnnn 99*("*!:!rnnr nrn !C*(&@#*99*("*n nnnr nn!@;&99*("*nnr n n99*("*!I%2!nrrn n n!$#@#*9#"*r nnnrr n9:')"*!>&!nnnnnrn rnnnn nr9:')"*!('! nnr

PAGE 268

249 6C$ 4..6C$.&30&30 0;0)"74#n%(( (5(5((5( 39 n9:')"*!1'>rn rn n9:')"*!!nnr n n9:')"*!<%(!rnnnn nnnr n "* +!)@#* n nn n n n +B#'! "* r n n n nn n r nn ;
PAGE 269

250 +r'&*rr) r)rrr &30 580)"(5*$ 9"#'"*<(#>rrnn 9"*+,#!nn <(#>!D!nn r$#-"*+B#'!nnn +$#@#*9"*nnr +I%2!9"*rrnn <(#>!I%2!rnnnr 9$#-"*+nn n9*("*!!@#*nn 9:')"*!!nnr !$#@#*9"#'"*nn !!#!9@&%"*nn r<(#>!:##&!nn +C*(&@#*99*("*rnn !!,>(((())"*rnn !!<(#>rnn 9<%B@#*+I%2!nnn !@;&B%&"*nn n+$#@#*9"*nn !$#@#*9$&$+>nn 9"* +<#! nn +I%2!9!'/rnn r+$#@#*9*;@#*nn !@;&(((())"*nnn +!9"*nnn !!@#*J;##"*nn 9"*+$#@#*nn "#'"*!B#'!nn n9*;@#*+I%2!nn !$#@#*99*("*nn !@;&9D"*nn 9!'/+!nn r+I%2!9"*nn !$#@#*9"(("*rnn 9"#'"*!!nn !$#@#*95;"*nn

PAGE 270

251 &30 580)"(5*$ n!<;!rnn +$#@#*9("*nnr rn "* +!)@#* nnr r+$#@#*9<%B@#*n nr r9"*+>;!nnr r+$#@#*9"*nn rr!,#@#*"#'"*rnn r9"#'"*!(!nn r(((())"*!#/*>rnn r9<%B@#*+!rnn r+@;&$')!rnn r9$#-"*+!nn n#!!nn +$#@#*9$#-"*rnnn +$#@#*9"*rnnn 9"#'"*!I%2!rnnn r+!9r"*nnn !,#@#*(((())"*nnn !@;&99*("*nnn !+@;&nnn !C*(&@#*9("*nnn !$#@#*9<%("*nnn n+$#@#*9"*nnn !!,>,#"*nnn 99*("*!I%2!nnnn +$#@#*9r"*nnnn r!,#@#*J;##"*nnnn @#!!nnnn "#'"*!!@#*nnnn 9r"*+!#!nnnn "#'"*!B-$nnn "*+<;##@#*nnnn n"*+@;&nnnn 9r"*+>;!nnnn 9(((())"*!/nnnn !#!nnnn r+!@#*r"*nnnn +:*!9r"*nnnn 9$#-"*+I%2!nnnn

PAGE 271

252 &30 580)"(5*$ "*+@;&nnnn +<#!9"*nrnnn +1;/9"*nnnn n(/B%&"*nnnn 9:')"*!<%(!nnnn +('!9"*nnnn ,#"*!5*!nnnn r!,#@#*,#"*nnnn 9("*+!nnnn +@;&9"*nnnn +,#@#*"*nnnn "*+,-!nnnn +>#!9!)#!%>nnnn n+!9!)#!%>nnnn 9r"*+!nnnn +>#!9!)#+>nnnn +I%2!+!$/nnnn r9("*+!nnnn +nrnnnr n9r"*+!nrnnn "*+G%!nrnnn !<;!nnnn +I%2!5(nnnnn r9r"*+J()!nnnn r"*+(!nnnn

PAGE 272

253 &30 580)"(5*$ !,#@#*A"*nnnn 9r"*+J()!nrnnn n!@#!nrnnn +!)@#*r"*nnnn n!$#@#*9-"*nnnn +I%2!9r"*nnnn "*@#!nnnn +(!9r"*nnnn r!!%!nnnn +<#!9r"*nnnn +@;&9r"*nnnn +,#@#*r"*nnnn +B#'!"*nnnn !@;&9D"*nnnn n!")!nnnn +,#@#*n"*nnnn 9"*+J()!nnnn 9"*+:*!nnnn r+I%2!5(nnnnn !,#@#*B%&"*nnnn 9$#-"*+!nnnn +$#@#*9rn"*nnnn +!)@#*"%;&nnnn +!9"*nnnn rn+$'2(/9r"*nnnn r+!@#*"*nnnn r9rn"*+!nnnn rr"*+<;##@#*nnnn rr9rn"*+$'2(/nrnnn r"*+(!nrnnn r+!9"*nnnn r+,#@#*+!)@#*nnnn r!<'!nnnn r9"#'"*9,'%>nnnn nn"*+,#@#*nnnn !1'#nnnn +I%2! 9"* nn nnn +B#'! "* nn nnn r"*+E#2*##/nnnnn

PAGE 273

254 &30 580)"(5*$ +(!9r"*nnnnn 9"*+>;!nnnnn "*+,#@#*nnnnnn "*+5*!nnnnnn +@;&#!nnrnnnn n"#'"*!%'!nnrnnnn 9("*+!%#!nnnnnn +,#@#*"*nnnnnn "#'"*!&!nnnnnn r+!)@#*@#!nnnnnn 9"#'"*!<#!nnnnnn +E!9r"*nnnnnn 99*("*!>;!nnnnnn "*9"*nnrnnnn +!9"*nnrnnnn n+@;&"*nnrnnnn +!9";(>nnnnnn 9:')"*!>&!nnnnnn +!)@#*+@!nnnnn rE#'#r!nnnnn 9("*+!#!nnnnn !!,>5;"*nnnnn +:##&!9!)#+>nnnnn +,#@#*"*nnnnn 9"*+!nnnnn n9"*+E!nnnnn 9!)@#*+$!nnnnn +,#@#*rr"*nnnnn n"*+B-$nnnn r+!9r"*nnnn +I%2!9#nnnn $#-"*+)(!nnnn +<#!9"*nnnn n!#!nnnn +!@#*"*nnnn n9r"*+$#&!nnnn !#!nnnn 9"#'"*!>;!nnnn n!!nnnn

PAGE 274

255 &30 580)"(5*$ r!!nnnn 9!)#+>+!nnnn 1'#!nnnn +@;&9"*nnnn +,#@#*"*nnnn !@#!nnnn nn+!)@#*"*nnnn n+!@#*("*nnnn n$#-"*#!nnnn n+:*!9rn"*nnnn nr$#-"*+5*!nnnn n+<;##@#*"*nnnn n9"*@#*nnnn n+!9r"*nnnn n+!@#*r"*nnnn n+<#!9"*nnnn n!$%(!nnnn 9"*+!nnnn +,#@#*r"*nnnn +@;&("*nnnn r!$#@#*9D"*nnnn +$#@#*9"*nnnn "*+!,>nnnn 9$#-"*+(!nnnn +$#@#*9"*nnnn +$#(!9"*nnnn n+!@#*rr"*nnnn +$#@#*9"*nnnn +!5(nnnnn +,#@#*"*nnnn r"*+1L!nnnn 9,#"*!:##&!nnnn +I%2!9"*nnnn +(!5(nnnnn 9r"*+(!nrnnn r"*+(!nrnnn n+$#@#*9r"*nrnnn +A(0%.@#*9r"*nrnnnr +1;/@#*nrnnnr

PAGE 275

256 &30 580)"(5*$ "*+1&(!nrrnnnr rEA##&/@#*+:'#&/nrrn nnr +!9"*nrnnnr 9$#-"*+9#.2!nrnnnr +!@#*"*nrnnnr 9"*+(!nnnnr +1&(!rr"*nnnnr rn9r"*++:##&!nrnnn #%!!nnnn 9"#'"*!:*!nnnn r+@;&@#!nnnn n"*+<;##@#*nnnn 9"*+$#(!nnnn +,#@#*5(nnnnn 9"*+I%2!nnnn 9"*+>;!nnnn n+,#@#*"*nnnn +A(0%.@#*9"*nrnnn 9"*+$!nnnn +!#!9rn"*nnnn r+$#@#*5(nnnnn +,#@#*r("*nnnn +1;/9r"*nnnn +$#@#*9"*nnnn rr"*+<;##@#*nnnn +<)!"*nnnn n+!@#*r"*nnnn 9r"*+#!nnnn

PAGE 276

257 &30 580)"(5*$ !"($nnnn +I%2!9"*nnnn r+$#@#*5(nnnnn +(!5(nnnnn +!@#*r("*nnnn rr"*+5*!nnnn !$#!nnnn "*9"*nnnn n!@#!nnnn "*+5*!nnnn "*+$#&!nnnn "*1'#nnnn r"*+&!nnnn "*++*L!nnnn +(!5(nnnnn +A(0%.@#*+!#!nrnnn r"*+G%!nnnn +(!9n"*nnnn n")!!nnnn +:*!9r"*nnn +!9"*nnn +(!r"*nrnnn r9"*+>;!nnnn +@;&<'!nnn +(!9("*nnn +1&(!r"*nnn "*+5*!nn +!@#*5(nnn nn9r"*+>##(!nn n9"*+$'2(/nn n+>;!#%!nn n+$!9r"*rnn nr+(!/(2&$nn nr"*+G%!nn n9"*+!rnn n$#!!nn n"*+(!rnnn n9"*+!nn ;
PAGE 277

258 +r''*rr) r)rr B2 &30 580)"(39*$ $#-"*+B#'!nrnn !!<(#>n <(#>!I%2!n r 9"* +<#! n 9"#'"*!!nrn +C*(&@#*99*("*n 9"#'"*<(#>n (((())"*!#/*>n +$#@#*9$#-"*n n9"*+$#@#*nn !,#@#*"#'"*n !$#@#*9"#'"*n +$#@#*9("*nnn r "* +!)@#* n !$#@#*99*("*n +$#@#*9"*n !C*(&@#*9("*n 9:')"*!1'>n +!9r"*n n!!,>(((())"*n !$#@#*9<%("*n +$#@#*9"*rn 9"#'"*!I%2!rn r"#'"*!B#'!rn 9<%B@#*+I%2!nn +$#@#*9r"*nn +$#@#*9<%B@#*rn r !$#@#*9$&$+>rn r 9$#-"*+!nrn n!@;&99*("*n !,#@#*(((())"*rn <(#>!:##&!nr <(#>!D!nr r9*("*!!@#*nr "*+@;&n +$#@#*9"*nn !$#@#*95;"*n

PAGE 278

259 &30 580)"(39*$ !$#@#*9"(("*n +$#@#*9"*n rn"*+@;&nnn r9:')"*!!nnn r!!@#*J;##"*nnn r!,#@#*J;##"*nnn rr+,#@#*"*nnn r+$#@#*9"*nnn r!,#@#*,#"*nnn r9$#-"*+I%2!nnn r$#-"*!nn r9<%B@#*+!rnn n9:')"*!<%(!rnn +I%2!9!'/nnr !@;&B%&"*nn n!<;!nn r9$#-"*+,#"*rnn +!)@#*r"*nnr +!9"*nn !@;&"#'"*nn 9(((())"*!/nn +!9!)#!%>nn n+@;&9"*rnn 9(((())"*!$#@#*rnn "#'"*!B-$nnr 9"*+,#!nn r9:')"*!('!nn 9("*+!nn !@;&(((())"*nn

PAGE 279

260 &30 580)"(39*$ "#'"*!!@#*nnr 99*("*!I%2!nn 9r"*+!#!nn n+I%2!9"*nn +1;/9"*rnnr 9r"*+!rnnr !#!rnnr r9"*+>;!rnnr "*+,-!rnnr +<#!9"*rnnr @#!!rnnr ,#"*!5*!rnnr 9"#'"*!(!rrnnrr n+!@#*r"*rnnr +:*!9r"*rnnrn +,#@#*r"*rnnrn "*+<;##@#*nn r+@;&$')!nn !$')!nn !$#@#*9#"*nn #!!nn +<#!9"*nn +B#'!"*nn nn9!'/+!nn n+$#@#*9rn"*nn n+@@#*!rnnr n+!)@#*"%;&rnnr nr+<#!9r"*nn n+>#!9!)#+>nn n9r"*+>;!nn n(/B%&"*nn n+I%2!5(nnn n+!9"*nn nr"*+(!nn "*5(nn !<;!nn +,#@#*n"*nn r!@;&9D"*nn +I%2!5(nnn

PAGE 280

261 &30 580)"(39*$ +#!9!)#!%>nnn rn!@#!nnn 9r"*+J()!nn "#'"*!#/*>nn +(!9r"*nn !")!nn +I%2!9"*nn n9r"*+!nn +!9"*nn "*@#!nn !,#@#*A"*nn r9rn"*+$'2(/rnnr +I%2! 9"* r nnr 9"*+J()!rnnr 9"*+$!rnnr 9"#'"*9,'%>rnnr !,#@#*B%&"*rnnr rnn"*+,#@#*nn r!!%!nn r"*+(!nn r+,#@#*+!)@#*nn rr9rn"*+!nnn r"*+G%!nnn r"*+E#2*##/nnn r+$'2(/9r"*nnnn r99*("*!:!nnnn r +B#'! "* n nnn n"*+,#@#*nnnn 9"*+:*!nnnn +@;&#!nnnn !1'#nnnn r+(!9r"*nrnnnr

PAGE 281

262 &30 580)"(39*$ r"*+<;##@#*nrnnnr +('!9"*nrnnnr !<'!nnnn 9("*+!%#!nnnn "*+5*!nnnn n"#'"*!%'!nnnn +,#@#*"*nnnn 9"*+>;!nnnn "#'"*!&!nnnnn rE#'#r!nnnn n"*+B-$nnnn 9r"*+$#&!nnnn +!)@#*@#!nnnn "*9"*nnnn +E!9r"*nnnn n9"#'"*!<#!nnnn 99*("*!>;!nnnn 9!)#+>+!nrnnnr +!9"*nrnnnr r9"*+!nrnnnr +!9";(>nnnn +@;&"*nnnn "*1'#nnnn +:##&!9!)#+>nnnn !$%(!nnnn n+I%2!9#nnnn !#!nnnn 9"*+E!nnnn +!9r"*nnn r+,#@#*"*nnn +<#!9"*nnn +!)@#*+@!nnn 9!)@#*+$!nnn 9:')"*!>&!nnn 9"*@#*nn n9"*+$'2(/nn +,#@#*rr"*nn +<;##@#*"*nn 9("*+!#!nn

PAGE 282

263 &30 580)"(39*$ r"*+1L!nn 9"#'"*!>;!nn 9"*+!nn 1'#!nn +:*!9("*rnnr +:*!9rn"*rnnr nn9r"*+(!rnnr n!!,>5;"*rnnr nrr"*+<;##@#*nn n+!9r"*nn nr!$#@#*9D"*nn n9,#"*!:##&!nn nr"*+G%!nn nrr"*+5*!nn n!"($nn n9"*+(!nn n+1&(!rr"*nn !!nn "*+!,>nn 9"*+$!nn r9"*+!nn +$#(!9"*nnn 9"*+!nn n!#!nn "*+1&(!nn +@;&9"*nn n"*+(!nn 9r"*+
PAGE 283

264 &30 580)"(39*$ n!!nn r+(!/(2&$nn +<)!"*nn +,#@#*"*nn +@;&@#!nnn $#!!nnn $#-"*#!nnn rnn"*+<;##@#*nn rr"*+(!nn r9"*+$#(!nn r+(!5(nnn rr"*+5*!nn r$#-"*+5*!nn r+!@#*r"*rnnr r9!)#!%>+:##&!rnnr r!@#!nn r+>;!#%!nn n!@#!nn 9$#-"*+(!nn +,#@#*r"*nn 9r"*+#!nn r+@;&("*nn +1;/@#*nn +!#!9rn"*nn 9"*+>;!nn +A(0%.@#*9"*rnnrn +$#@#*9r"*rnnrn n+$#@#*5(nrnnr !$#!rnnr 9"*+I%2!rnnr +(!5(nrnnr r9r"*+>##(!rnnr +$!9r"*rnnr "*++*L!rnnr +!@#*("*rnnr EA##&/@#*+:'#&/rrnn rr +!@#*rr"*rrnnrr n+@;&<'!rrnnrr +,#@#*"*rnnr

PAGE 284

265 &30 580)"(39*$ +:*!9r"*rnnr +1&(!r"*rnnr r+(!9n"*nn +!)@#*"*nn +$#@#*9"*nn +A(0%.@#*+!#!nnr +1;/9r"*rnn +$#@#*9"*rnn n+I%2!9"*nn "*9"*nn +,#@#*5(nnn +!9"*nn r+<#!9"*nn +(!5(nnn +(!9("*nn +!@#*5(nnn ")!!nn +$#@#*5(nnn n+!@#*"*nn 9"*+>;!nnr +!@#*"*rnn +:*!5(nnn r"*+!@#*nn +$#@#*9"*nn +$#@#*9"*nnn +,#@#*"*nn +!5(nrnn "*+&!nn nn+(!r"*nn n+,#@#*("*nn n+,#@#*"*nnn n9"#'"*!:*!nn nr+!5(nnn n+,#@#*r("*nn n+!@#*r"*rnn n+I%2!9"*nn n+!@#*r("*nnn n+!)@#*9#$nnn ;
PAGE 285

266 +r'(;r2*r r2) 0;0)"74#n%3$ 535B4( +1;/@#*rrrnrn +1;/9"*nnn 9"*@#*nnnn r9"*+:*!nrnn 9"*+!nrnnnnr 9"*+I%2!nrnnr +I%2!9#nrn 9"*+$'2(/nnnnn +A(0%.@#*9"*nnrnnnr n"*+(!rnnnnnnn +,#@#*"*rnnnr +!@#*"*nrrnnr r +:*!9("*nnn r9("*+!rnrnr +(!9("*rnnnnnnnn +(!9n"*rnrnnnn n"*+,#@#*nn n"*+<;##@#*rnrnn +!9";(>rrnrnn n+,#@#*5(nnrnn r EA##&/@#*+:'#&/n rnn +A(0%.@#*9r"*nnnr +$#@#*9r"*rnnr rr"*+G%!rnnnnnnnnr r"*+(!nrnn +!@#*5(nnnnnnnn n +!@#*r"*rnrnrn r +!9r"*nnr +:*!9r"*rnnr

PAGE 286

267 0;0)"74#n%3$ 535B4( n+(!5(nnnnr 9r"*+>##(!nnnnnr +I%2!+!$/nn r +A(0%.@#*+!#!nnnn r+(!9r"*rnnnr +$#@#*5(nrrnn n +>#!9!)#+>rnnn nn 9r"*+J()!rnnn 9r"*+!#!nnnr 9r"*+$#&!nnnnr rn9r"*+!nnnnr r+,#@#*r"*rrnrnnr rr"*+(!nrn rr"*+G%!rrrnnnnn rr+(!5(nnnnrn n rr"*+<;##@#*nnrnn r+>#!9!)#!%>n nn r+1&(!r"*nnnnnnnnn r r+!@#*r"*nnnnn r+I%2!5(nnnr rn n+:*!9r"*rnnnn +$#@#*5(nnrnn nn +(!5(nnrn r +!#!9r"*rrnnrrn r9!)#+>+!nr nn +:##&!9!)#+>nr nrr +!9r"*nnrnr +(!9r"*nnrnn +!9!)#!%>nn nrn 9!)#!%>+:##&!nr nr n+I%2!5(nrnnn

PAGE 287

268 0;0)"74#n%3$ 535B4( +:*!5(nnnnr r +$'2(/9r"*rrrrnnnnnn nr rr"*+<;##@#*nnnnr rrr"*+5*!rnrnn +,#@#*rr"*nrnn nn +1&(!rr"*nnn +(!/(2&$rnnnn nr +!@#*rr"*rnrnn n +1;/9r"*rrnrnnn n+!5(nnrnnrn !"($rrnnnnnn +(!r"*nnnnnn 9rn"*+$'2(/nrnr r+!5(nrnnnr +:*!9rn"*rnnn 9!'/+!nnnnnnnn +$#@#*9rn"*nnnr r +!#!9rn"*nnnn 9rn"*+!nnnnn n+@@#*!nrnn +,#@#*r("*rnnnn r +!@#*r("*nnnn n +I%2!9!'/nnnn r r+!9"*rnn r #%!!nrnr "*+<;##@#*nnnr +>;!#%!nnnnnnr "*+5*!nnnnn +<)!"*rrnnrnrn n"*+&!rnnnnr "*++*L!nnnr

PAGE 288

269 0;0)"74#n%3$ 535B4( "*+(!nnn n "*+1L!nrnnnn r"*+G%!rnrnrr "*+,-!nn +,#@#*"*rrrnr "*+1&(!nnrn r "*+$#&!nnnr +!@#*"*nnn rn nn+!9"*nrnrn n +I%2! 9"* rn n nr rn n"*5(rnrnrnr nr n+I%2!9"*nnr nr+$#@#*9"*rnrn n"*+E#2*##/rnrnn n+,#@#*"*nnnn n n+,#@#*"*rnrn n 9<%B @#*+I%2!nnrnnn n 9<%B @#*+!rrrnrrn n+$#@#* 9<%B @#*nnn "*+,#@#*nrnnnn nr "*+!@#*rnn 9("*+!rnnn r r+,#@#*+!)@#*nnn r +@;&@#!rnnrnr +$#@#*9"*nnn r +<;##@#*"*nnrn n +,#@#*"*rnnn rn +!)@#*"*nrnr n"*+5*!nnnn

PAGE 289

270 0;0)"74#n%3$ 535B4( +!@#*"*nnrnnn !$#!rnnn 9"*+!nnnnnnn r"*@#!rnnnnnnnr +@;&<'!nrnnrnn !#!nnn @#!!nnr +I%2!9"*nnnn +!9"*nnnn n+$#@#*9"*rnn r !<'!nnrrn +,#@#*"*rnn "*+5*!nnnrn r9"*+>;!nrnn +!9"*rnrnn n!@#!rnrnrn !<;!nrnnn r n!!rnrnnr !")!rnnr rn!@#!nnnr r!+@;&nnnr r!@#!rnnrrn r+!)@#*9#$nnrnnr rrn!<;!rnnnn r r!!nrn r+@;&$')!nrnr r+I%2!9"*nnrnr r+$#@#*9"*nnn r r!$%(!rnnr n"*1'#nnnnn ")!!nnnnnn

PAGE 290

271 0;0)"74#n%3$ 535B4( +!)@#*@#!nnrrnrn n!#!nn r9*;@#*+I%2!nrrnrnn rrr +!)@#*"%;&nrn +$#@#*9*;@#*nn nr +@;&#!nrnrn "*9"*rnnn +<#!9"*nrn n+@;&9"*nnrnn !$')!nrnr !#!rnnn $#!!rrnnnnnnn r9"*+,#!nrn r +$#(!9"*nnrnr !!%!nnn +('!9"*rnnnnr +!9"*rnnnnnnn r 1'#!nnr n+$#@#*9"*rnr nn +,#@#*"*nnn r +!@#*"*nn #!!nnnn r+@;&9"*rnnr r 9"*+>;!rrnrnnr "*9"*nnrnrr +<#!9"*rnrnrnrn n !1'#nnr E#'#r!nnrn n9"*+!nrn 9$#-"*+I%2!nrn n 9$#-"*+!rnrnr r

PAGE 291

272 0;0)"74#n%3$ 535B4( +$#@#*9$#-"*nn nrn r9$#-"*+9#.2!nnn rr $#-"*+5*!rnn $#-"*+B#'!nr n $#-"*#!nnnn 9$#-"*+!rrnn +!@#*$#-"*rrn nn n9$#-"*+(!rnrn r 9$#-"*+;!rnrn n n9r"*+#!rrnnnr n n9r"*+(!nrrn n9r"*+;!nnnr n 9"*+(!nnnr

PAGE 292

273 0;0)"74#n%3$ 535B4( r+$#@#*9"*nn n +;!nnnnnn +!)@#*+@!rnn r r+I%2!9"*nrnr +!9"*nn +$#@#*9"*rnnn 9"*+$!nrnnn 9"*+$#(!rrrnrnr "*+@;&nn n"*+!,>nnn r +B#'!"*rnrn rr 9"*+!nnnrn 9"*+$#@#*rrrr nnn r 9"* +<#! r n "*+@;&nnr n 9"*+$!nnrr n +$#@#*9"*rn +@;&"*nnn 9!)@#*+$!rnrn r rn+@;&("*nnnr r+C*(&@#*9("*rrrn r9("*+!%#!nnn rn r+$#@#*9("*n rr+,#@#*("*rrnnr

PAGE 293

274 0;0)"74#n%3$ 535B4( r+!@#*("*nnnr r9("*+!#!rnrnrnnr r!!#!9@&%"*nrn rn r!$#@#*9"#'"*n nnn r9"#'"*9,'%>nnrr nrn n9"#'"*!I%2!nnr n 9"#'"*!:*!nnrrn 9"#'"*<(#>rrnrrn n 9"#'"*!!nn n r!C*(&@#*9"#'"*rnrnrr n 9"#'"*!>;!nnrn rn 9"#'"*!<#!nnn "#'"*!B#'!nrnr nrr !@;&"#'"*rnn r 9"#'"*!(!nnn r n"#'"*!#/*>nr nrr "#'"*!!@#*rnnn "#'"*!%'!nnrn "#'"*!&!nn nr r"#'"*!B-$rnn r !,#@#*"#'"*nnrr nn <(#>!:##&!nn !$#@#*9$&$+>rn nnrr !,#@#*A"*nnn nrr !!<(#>rn rr n!$#@#*9D"*rrnnn n !@;&9D"*nnn r !@;&9D"*nrnrnn <(#>!D!nnr rr r!,#@#*B%&"*nnnr nr !@;&B%&"*rnrnrn r

PAGE 294

275 0;0)"74#n%3$ 535B4( (/B%&"*nrnn <(#>!I%2!rrnn r 9(((())"*!/nrrnrn nr 9(((())"*!$#@#*nr n n!!,>(((())"*rn n (((())"*!#/*>rr rrnn !@;&(((())"*rrnn !,#@#*(((())"*nn nr r!$#@#*9<%("*rrr nr !$#@#*9"(("*n n 9,#"*!:##&!rrnnr rr !!,>,#"*nrn r !,#@#*,#"*rn n ,#"*!5*!rrnn n!$#@#*95;"*nn !!,>5;"*nrn !$#@#*9-"*n n !,#@#*J;##"*nn r!!@#*J;##"*rrnnr nrnr 9*("*!!@#*nrnn !$#@#*99*("*rrn 99*("*!>;!rnn 99*("*!:!rnnrnn !C*(&@#*99*("*rrn r nn!@;&99*("*nrrn n99*("*!I%2!nnr n!$#@#*9#"*nn n9:')"*!>&!nnnnrn nr9:')"*!('!nrn n9:')"*!1'>rnn n9:')"*!!nnn

PAGE 295

276 0;0)"74#n%3$ 535B4( n9:')"*!<%(!rnn n n "* +!)@#* r n n n n +B#'! "* r n nr 1#nn "* ;
PAGE 296

277 +r'*rr r 580)"3+$5*$ W Mississippi Ave S Platte River Dr 2.33 nnn N Colorado Blvd E Colfax Ave 2.00 nnr W Colfax Ave N Kalamath St 1.93 nnr r E Alameda Ave S Monaco St 1.72 nn N Peoria St E 47th Ave 1.69 nn S Federal Blvd W Alameda Ave 1.66 nn S Colorado Blvd E Louisiana Ave 1.66 nn S Monaco St Leetsdale Dr 1.63 nn S University Blvd E 1st Ave 1.63 nn n N Colorado Blvd E 3rd Ave 1.57 nnr N University Blvd E Evans Ave 1.38 nnn E 6th Ave N Colorado Blvd 1.35 nn Leetsdale Dr S Quebec St 1.35 nn r W Evans Ave S Sheridan Blvd 1.32 nn E Hampden Ave S Tamarac Dr 1.32 nn E Martin Luther King Blvd N Quebec St 1.29 nn "* +!)@#* 1.29 nn N Federal Blvd W 38th Ave 1.26 nn N Colorado Blvd E 17th Ave 1.26 nn n W Colfax Ave 7th St 1.26 nn N Colorado Blvd E Martin Luther King Blvd 1.17 nn E 6th Ave N Lincoln St 1.17 nn S Federal Blvd W Mississippi Ave 1.17 nn r N Colorado Blvd E 1st Ave 1.14 nnr N Sheridan Blvd W Colfax Ave 1.11 nnr W Alameda Ave S Kalamath St 1.11 nnr S Broadway W Alameda Ave 1.11 nnr S Colorado Blvd E Cherry Creek North Dr 1.11 nnr S Colorado Blvd E Alameda Ave 1.07 nn n S Colorado Blvd E Mexico Ave 1.07 nn S Federal Blvd W Jewell Ave 1.07 nn S Sheridan Blvd W Jewell Ave 1.07 nn E Martin Luther King Blvd N Monaco St 1.04 nn r E Alameda Ave Leetsdale Dr 1.04 nn S University Blvd E Alameda Ave 1.04 nn N Colorado Blvd E 13th Ave 1.01 nn Leetsdale Dr S Holly St 1.01 nn E Hampden Ave S Locust St 0.98 nn W 8th Ave N Broadway 0.95 nnn rn S Colorado Blvd E Iowa Ave 0.95 nnn r N Sheridan Blvd W 38th Ave 0.89 nn

PAGE 297

278 580)"3+$5*$ r E Alameda Ave S Quebec St 0.89 nn r E 31st Ave N York St 0.83 nn rr W 6th Ave N Broadway 0.83 nn r N Peoria St E 39th Ave 0.80 nn r E Colfax Ave N Monaco St 0.80 nn r Leetsdale Dr S Oneida St 0.80 nn r S Colorado Blvd E Arkansas Ave 0.80 nn r S Federal Blvd W Florida Ave 0.80 nn n N Quebec St E Smith Rd 0.77 nn 20th St Lawrence St 0.77 nn E Colfax Ave N York St 0.77 nn E Evans Ave S Quebec St 0.77 nn r E Hampden Ave S Monaco St 0.77 nn W Florida Ave S Irving St 0.74 nn N Monaco St E Stapleton South Dr 0.71 nn 17th St Welton St 0.71 nn N Colorado Blvd E 8th Ave 0.71 nn N Broadway E 17th Ave 0.68 nnr n N Colorado Blvd E 14th Ave 0.68 nnr N Sheridan Blvd W 14th Ave 0.68 nnr S Broadway E Ohio Ave 0.68 nnr W 46th Ave N Pecos St 0.64 nnr r W 32nd Ave N Federal Blvd 0.64 nnr W Colfax Ave N Mariposa St 0.64 nnr N Broadway E 14th Ave 0.64 nnr N Speer Blvd W 14th Ave 0.64 nnr N Federal Blvd W 10th Ave 0.64 nnr S Broadway E Evans Ave 0.64 nnr n E Alameda Ave E Fairmount Dr 0.61 nn W 50th Ave N Federal Blvd 0.58 nn N Havana St E 47th Ave 0.58 nn E 46th Ave N Steele St 0.58 nn r N Peoria St E 37th Ave 0.58 nn N Tower Rd E 56th Ave 0.55 nn N Quebec St N Sand Creek Rd 0.52 nn N Quebec St E 36th Ave 0.52 nn Blake St 22nd St 0.52 nn 22nd St N Broadway 0.52 nn n E 14th Ave N York St 0.52 nn N Lincoln St E 14th Ave 0.52 nn E Evans Ave S Downing St 0.52 nn S Colorado Blvd E Yale Ave 0.52 nn r E Hampden Ave S Yosemite St 0.52 nn

PAGE 298

279 580)"3+$5*$ N Peoria St E 45th Ave 0.49 nn N Washington St E 45th Ave 0.49 nn N Speer Blvd W 29th Ave 0.49 nn E Speer Blvd N Corona St 0.49 nn S Santa Fe Dr W Iowa Ave 0.49 nn n N Holly St E Stapleton North Dr 0.46 nnn N York St E 26th Ave 0.46 nnn 18th St Blake St 0.46 nnn N Quebec St E 23rd Ave 0.46 nnn r N Speer Blvd Auraria Pkwy 0.46 nnn E 18th Ave N Franklin St 0.46 nnn E Colfax Ave N Quebec St 0.46 nnn E 14th Ave N Downing St 0.46 nnn N Kalamath St W 7th Ave 0.46 nnn N Peoria St E Andrews Dr 0.43 nnn nn N Sheridan Blvd W 48th Ave 0.43 nnn n N Federal Blvd W 46th Ave 0.43 nnn n Park Ave W Interstate 25 0.43 nnn n Park Ave W N Globeville Rd 0.43 nnn nr E Montview Blvd N Quebec St 0.43 nnn n N Lincoln St E 17th Ave 0.43 nnn n E 13th Ave N Grant St 0.43 nnn n E 1st Ave N Steele St 0.43 nnn n S Broadway E Ohio Ave 0.43 nnn n S Broadway W Kentucky Ave 0.43 nnn n E Mississippi Ave S Parker Rd 0.43 nnn N Quebec St Interstate 70 0.40 nnn E 40th Ave N York St 0.40 nnn N Federal Blvd W 29th Ave 0.40 nnn r 22nd St Lawrence St 0.40 nnn 20th St Market St 0.40 nnn W Colfax Ave Welton St 0.40 nnn E 14th Ave N Josephine St 0.40 nnn N Grant St E 14th Ave 0.40 nnn N Lincoln St E 13th Ave 0.40 nnn n E 1st Ave N Saint Paul St 0.40 nnn N Sheridan Blvd W 1st Ave 0.40 nnn S Steele St E Bayaud Ave 0.40 nnn S Santa Fe Dr W Mississippi Ave 0.40 nnn r E 46th Ave N Josephine St 0.37 nnn N Dahlia St E Stapleton South Dr 0.37 nnn N Federal Blvd W 26th Ave 0.37 nnn 15th St Champa St 0.37 nnn

PAGE 299

280 580)"3+$5*$ N Federal Blvd W 17th Ave 0.37 nnn 15th St Tremont Pl 0.37 nnn n N Federal Blvd W 14th Ave 0.37 nnn E 13th Ave N Downing St 0.37 nnn W Alameda Ave S Perry St 0.37 nnn S Colorado Blvd E Evans Ave 0.37 nnn r N Quebec St E 53rd Pl 0.34 nnn N Federal Blvd W 52nd Ave 0.34 nnn N Washington St E 46th Ave 0.34 nnn E 46th Ave N York St 0.34 nnn N Havana St E 40th Ave 0.34 nnn W 38th Ave N Lowell Blvd 0.34 nnn rn W 38th Ave N Pecos St 0.34 nnn r W 38th Ave N Fox St 0.34 nnn r N Federal Blvd N Speer Blvd 0.34 nnn r 20th St Blake St 0.34 nnn rr 22nd St Arapahoe St 0.34 nnn r Market St 18th St 0.34 nnn r 20th St Welton St 0.34 nnn r N Broadway Welton St 0.34 nnn r Tremont Pl 17th St 0.34 nnn r N Sheridan Blvd W 17th Ave 0.34 nnn n Welton St 15th St 0.34 nnn N Speer Blvd N Bannock St 0.34 nnn W 7th Ave N Santa Fe Dr 0.34 nnn E 6th Ave N Corona St 0.34 nnn r S Broadway W Mississippi Ave 0.34 nnn N Dahlia St E Stapleton North Dr 0.31 nnn W 38th Ave N Irving St 0.31 nnn S Federal Blvd W Virginia Ave 0.31 nnn S Federal Blvd W Kentucky Ave 0.31 nnn N Vasquez Blvd E 48th Ave 0.28 nnn n N Washington St Interstate 70 0.28 nnn N Colorado Blvd E 29th Ave 0.28 nnn 15th St Stout St 0.28 nnn W Colfax Ave N Irving St 0.28 nnn r E Colfax Ave N Logan St 0.28 nnn N Logan St E 13th Ave 0.28 nnn N Monaco St E 8th Ave 0.28 nnn N Broadway W 3rd Ave 0.28 nnn N Steele St E 45th Ave 0.25 nnn N Federal Blvd W 44th Ave 0.25 nnn n N Sheridan Blvd W 44th Ave 0.25 nnn

PAGE 300

281 580)"3+$5*$ E 40th Ave N Chambers Rd 0.25 nnn W 38th Ave N Zuni St 0.25 nnn N Colorado Blvd E 35th Ave 0.25 nnn r N Broadway Champa St 0.25 nnn E Colfax Ave N Washington St 0.25 nnn W Alameda Ave S Platte River Dr 0.25 nnn S Santa Fe Dr W Florida Ave 0.25 nnn N Colorado Blvd E 48th Ave 0.21 nnn W 48th Ave N Pecos St 0.21 nnn n N Tennyson St W 44th Ave 0.21 nnn N Quebec St E 35th Ave 0.21 nnn N Federal Blvd W 33rd Ave 0.21 nnn N Sheridan Blvd W 29th Ave 0.21 nnn r 15th St Platte St 0.21 nnn N Colorado Blvd E 26th Ave 0.21 nnn N Colorado Blvd E 23rd Ave 0.21 nnn N Colorado Blvd E Montview Blvd 0.21 nnn E 16th Ave N York St 0.21 nnn N Monaco St E 14th Ave 0.21 nnn n E 13th Ave N Josephine St 0.21 nnn N Lincoln St E 12th Ave 0.21 nnn N Quebec St E 8th Ave 0.21 nnn N Broadway W 1st Ave 0.21 nnn r E Alameda Ave S Lincoln St 0.21 nnn E Alameda Ave S Washington St 0.21 nnn W Alameda Ave S Knox Ct 0.21 nnn +B#'! "* 0.21 nnn Green Valley Ranch Blvd N Himalaya Rd 0.18 nnnr W 46th Ave N Lowell Blvd 0.18 nnnr nn E Stapleton North Dr N Monaco St 0.18 nnnr n N Brighton Blvd 38th St 0.18 nnnr n W 38th Ave N Tennyson St 0.18 nnnr n N Lowell Blvd W 29th Ave 0.18 nnnr nr 22nd St Larimer St 0.18 nnnr n N Speer Blvd Blake St 0.18 nnnr n N Lincoln St E 19th Ave 0.18 nnnr n N Clarkson St E 18th Ave 0.18 nnnr n E 17th Ave N Downing St 0.18 nnnr n Park Ave E 17th Ave 0.18 nnnr n E 14th Ave N Washington St 0.18 nnnr E 13th Ave N Syracuse St 0.18 nnnr E 8th Ave N Corona St 0.18 nnnr W Alameda Ave S Yuma St 0.18 nnnr

PAGE 301

282 580)"3+$5*$ r E Mississippi Ave S Colorado Blvd 0.18 nnnr E Hampden Ave S Dayton St 0.18 nnnr N Tower Rd Pena Blvd 0.15 nnn E 56th Ave N Havana St 0.15 nnn E 56th Ave N Quebec St 0.15 nnn E 51st Ave N Peoria St 0.15 nnn n W 50th Ave N Lowell Blvd 0.15 nnn N Washington St Interstate 70 0.15 nnn E Stapleton South Dr N Holly St 0.15 nnn W 44th Ave N Irving St 0.15 nnn r N Tower Rd E 43rd Ave 0.15 nnn Walnut St 38th St 0.15 nnn N Broadway Blake St 0.15 nnn E 26th Ave N Downing St 0.15 nnn N Speer Blvd Elitch Cir 0.15 nnn E 14th Ave N Logan St 0.15 nnn n W 10th Ave N Knox Ct 0.15 nnn E Alameda Ave S Havana St 0.15 nnn E Evans Ave S High St 0.15 nnn N Vasquez Blvd E 52nd Ave 0.12 nnn r N Sheridan Blvd W 52nd Ave 0.12 nnn N Federal Blvd Interstate 70 0.12 nnn N Colorado Blvd Interstate 70 0.12 nnn N Colorado Blvd Interstate 70 0.12 nnn N Chambers Rd E 46th Ave 0.12 nnn 38th St Arkins Ct 0.12 nnn rn W 32nd Ave N Sheridan Blvd 0.12 nnn r 15th St Central St 0.12 nnn r 19th St Curtis St 0.12 nnn r Park Ave E 19th Ave 0.12 nnn rr N Broadway E 19th Ave 0.12 nnn r E 14th Ave N Pearl St 0.12 nnn r E 8th Ave N Clarkson St 0.12 nnn r N Federal Blvd W 1st Ave 0.12 nnn r S Colorado Blvd E Ohio Ave 0.12 nnn r Morrison Rd W Kentucky Ave 0.12 nnn n E Florida Ave S Holly St 0.12 nnn E 56th Ave Pena Blvd 0.09 nnn N Havana St E 51st Ave 0.09 nnn E 46th Ave N Clayton St 0.09 nnn r N Havana St E 45th Ave 0.09 nnn N Havana St Interstate 70 0.09 nnn W 44th Ave N Lowell Blvd 0.09 nnn

PAGE 302

283 580)"3+$5*$ N Peoria St Interstate 70 0.09 nnn N Federal Blvd W 41st Ave 0.09 nnn W 38th Ave N Tejon St 0.09 nnn n N Federal Blvd W 35th Ave 0.09 nnn W 26th Ave N Irving St 0.09 nnn Park Ave W Tremont Pl 0.09 nnn N Yosemite St E 17th Ave 0.09 nnn r E Colfax Ave N Elizabeth St 0.09 nnn N Quebec St E 14th Ave 0.09 nnn E 13th Ave N Washington St 0.09 nnn N Colorado Blvd E 12th Ave 0.09 nnn E 6th Ave N Monaco St 0.09 nnn W Alameda Ave S Sheridan Blvd 0.09 nnn n N Washington St E 50th Ave 0.06 nnn N Vasquez Blvd N Steele St 0.06 nnn W 46th Ave N Zuni St 0.06 nnn N Sheridan Blvd W 46th Ave 0.06 nnn r N Washington St Ringsby Ct 0.06 nnn N Peoria St Interstate 70 0.06 nnn E Smith Rd N Monaco St 0.06 nnn N Colorado Blvd E 40th Ave 0.06 nnn N Lipan St W 38th Ave 0.06 nnn W 38th Ave N Clay St 0.06 nnn n W 29th Ave N Irving St 0.06 nnn N Quebec St E 26th Ave 0.06 nnn N Monaco St E 26th Ave 0.06 nnn Glenarm Pl 14th St 0.06 nnn r E Alameda Ave S Downing St 0.06 nnn E 56th Ave N Peoria St 0.03 nnn E 53rd Ave N Chambers Rd 0.03 nnn W 52nd Ave N Pecos St 0.03 nnn N Washington St E 51st Ave 0.03 nnn W 48th Ave N Zuni St 0.03 nnn n N Pecos St Interstate 70 0.03 nnn E 47th Ave N Dallas St 0.03 nnn N Tennyson St W 46th Ave 0.03 nnn N Quebec St Interstate 70 0.03 nnn r N Pecos St W 42nd Ave 0.03 nnn N Steele St E 40th Ave 0.03 nnn N Sheridan Blvd W 41st Ave 0.03 nnn N Downing St Walnut St 0.03 nnn W 38th Ave N Perry St 0.03 nnn W 38th Ave N Navajo St 0.03 nnn

PAGE 303

284 580)"3+$5*$ nn N Broadway Larimer St 0.03 nnn n 19th St Blake St 0.03 nnn n Arapahoe St 18th St 0.03 nnn n California St 16th St 0.03 nnn nr N Corona St E 14th Ave 0.03 nnn n E 9th Ave N Downing St 0.03 nnn n N Sheridan Blvd Interstate 70 0.00 nnnn n E 28th Ave N York St 0.00 nnnn n Park Ave W Blake St 0.00 nnnn n N Monaco St E 17th Ave 0.00 nnnn ;