Citation
Pavement management for the city and county of Denver, Colorado

Material Information

Title:
Pavement management for the city and county of Denver, Colorado the impact of extreme weather
Creator:
Yan, Bo
Publication Date:
Language:
English
Physical Description:
viii, 103 leaves : illustrations ; 28 cm

Subjects

Subjects / Keywords:
Pavements -- Maintenance and repair -- Colorado -- Denver ( lcsh )
Roads -- Maintenance and repair -- Colorado -- Denver ( lcsh )
Severe storms ( lcsh )
Pavements -- Maintenance and repair ( fast )
Roads -- Maintenance and repair ( fast )
Severe storms ( fast )
Colorado -- Denver ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 102-103).
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Bo Yan.

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:
436864361 ( OCLC )
ocn436864361
Classification:
LD1193.E53 2009m Y36 ( lcc )

Downloads

This item has the following downloads:


Full Text
PAVEMENT MANAGEMENT FOR THE CITY AND COUNTY
OF DENVER COLORADO THE IMPACT OF EXTREME
WEATHER
By
Bo Yan
B.S., Air Defense Institute of Technology, 2002
M.S., East China Normal University, 2006
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Master of Engineering
Civil Engineering
2009


This thesis for the Master of Engineering
Degree by
Bo Yan
has been approved
by
Lynn E. Johnson
Cs
o
r
Date


Bo, Yan (M.Eng., GIS)
Pavement Management for the City and County of Denver Colorado the
impact of Extreme Weather
Thesis directed by Professor Kevin L. Rens
Abstract
Pavement management systems (PMS) are a critical tool used to manage and
maintain the transportation infrastructure around the world. The City and
County of Denver (CCD) has been managing its street network systematically
since 2000. Pavement management depends on accurate pavement inventory
and assessment related data. It depends on analysis models to prioritize
rehabilitation efforts and provide upper level management and politicians with
the knowledge they need to set future budgets. In 2006/2007 the Denver
metropolitan area experienced an exceptionally harsh winter period during a 3
week span. The city was hit with a series of intense storms resulting in
significant and accelerated deterioration to the CCD roadway network. The
actual measured deterioration was significantly greater than that predicted by
the analysis models in the existing PMS. Street Maintenance personnel of the
CCD were aware that the pavement network had accelerated deterioration and


needed to quickly quantify the damage and assess the long term impact on
their work programs and future budgets. To quickly quantify these effects, a
windshield assessment method was developed. Using this method, a sample of
600 miles (966 km) of arterial and collector roadways were inspected in two
weeks at an estimated internal cost on the order of $20,000. In order to re-
establish reliable detailed condition data for the PMS and to quantify the
effectiveness of the windshield inspection method, the same 1,572 miles
(2,530 km) of street networks inspected during 2002-2006 were again
inspected and rated during a 15 month period from 2007 to 2008 using a
standard assessment method. This 1,572 mile street network included 277
miles (446 km) of collector roads and 1,295 miles (2,084 km) of local roads.
The cost to manually inspect the network during the 15 month post storm
period was approximately $120 per mile. This was similar to the cost of to
manually inspecting the pre-storm network which totaled was approximately
$100 per mile in 2008 dollars. This thesis details the methodology used to
complete the pavement condition inspection and assessment. A comparison of
pre-storm and post-storm pavement conditions was completed for the CCD
road network. The results indicate that the 2006/2007 extreme winter events
accelerated aging of the CCD road networks. The overall pavement condition
index (PCI) value decreased by an average of 7 percentage points. Normal
winter months age the CCD network 2 or 3 points in PCI values. Comparison
of the windshield inspection technique to the standard inspection method
revealed that there is a high degree of correlation between the two.


This abstract accurately represents the content of the candidates thesis. I
recommend its publication.


ACKNOWLEDGEMENT
My utmost gratitude goes to my thesis advisor, Dr. Rens for his contribution
and support of my research, for his expertise, kindness, continual
encouragement and most of all, for his patience. I believe that one of the main
gains of this 3-year program was working with Dr. Rens and gaining his trust
and friendship. Thanks and appreciation goes to my thesis committee
members, Dr. Johnson and Dr. Durham, for their valuable participation and
insight. I want to thank Mr. Patrick Kennedy and Ms Angie Hager from the
City and County of Denver for their support and advice. I especially want to
thank all the students who were involved in the data collection without the
students, this research would be impossible. I have immensely enjoyed
working with all of the undergraduate students Above all, I am greatly
indebted to my husband and his family, who stood beside me and encouraged
me every day. I am thankful to all my friends for giving me happiness and joy
and continuous support and interest in what I do. Finally, I would like to thank
my parents and my sister whose love is boundless. Without their support
education in the united states would not have been possible.


Contents
Figures.........................................................iii
Tables..........................................................vii
1. Introduction...................................................1
1.1 Background.............................................. 1
1.2 Pavement Management Systems.............................12
1.3 Geographic Information Systems (GIS)....................15
1.4 Pavement Condition Index................................18
1.5 Objectives..............................................20
2. CCD Pavement data Management..................................21
2.1 CCD Pavement Database...................................21
2.2 CCD Pavement Data Management Process....................28
3. Pavement Inspection and Rating methodology....................31
3.1 Windshield Method.......................................31
3.1.1 Data Collection for the Windshield Method.........32
3.1.2 WCI Calculation...................................36
3.1.3 WCI Transformation to PCI.........................40
3.2 Standard Method.........................................41
3.2.1 Data Collection for the Standard Method...........42
3.2.2 PCI Calculation for the Standard Method...........52
3.2.3 Data Quality Analysis for the Standard Method.....69
4. Pavement Condition Analysis...................................73
l


4.1 Pre-storm and Post-storm Road Condition Comparison Analysis
using the Standard Method................................74
4.1.1 Overall Condition Comparison......................74
4.1.2 Individual Distress Comparison....................80
4.2 Correlation Analysis between the Windshield Method Data and
Standard Method Data using an Identical Road Dataset.....88
4.2.1 The Procedure of Identifying the Same Roads from the
Windshield Method Post-storm Data and the Standard
Method Post-storm Data...............................88
4.2.2 The Correlation Analysis of the Same Roads between the
Windshield Method and the Standard Method............94
5. Conclusion and Future Recommendation..........................97
5.1 Conclusion...............................................97
5.2 Discussion and Recommendation for the Future Work.......100
Bibliography....................................................102
ii


Figures
Figure
1.1 Pocket computer.............................................3
1.2 Denver downtown during the 2006/2007 snowstorm..............4
1.3 Daily snow fall during 12-15 to 1-5 for the winters of 2005,2006,
and 2007....................................................6
1.4 Daily low temperature during 12-15 to 1-15 for the winters of
2005, 2006, and 2007........................................7
1.5 Daily high temperature during 12-15 to 1-15 for the winter of
2005, 2006, and 2007........................................8
1.6 A 22 block arterial section on 18th Avenue from Broadway to
York shown in a red line...................................10
1.7 Dynamic segmentation.......................................18
2.1 GIS Road networks for the entire CCD area..................23
2.2 Arterial roads are in the red line, Collector roads are in the blue
line and Local roads are in the green line.................23
2.3 An example of inventory data in the PMS database...........25
2.4 An example of pavement condition data in the PMS database ... 27
2.5 Pavement data management process in the CCD................28
2.6 Field data collection using ArcPad.........................29
3.1 Potholes and patches damage Level 1 (high severity)......33
3.2 Alligator cracking -Level l(high severity).................33
iii


3.3 General cracking Level l(high severity).....................34
3.4 Overall condition Level 3 (moderate severity)...............34
3.5 Pavement performance curve used in the CCD road network.... 36
3.6 Windshield analysis deduct value curve.......................37
3.7 WCI/PCI conversion curve.....................................40
3.8 Transverse crack in the low severity level...................44
3.9 In-wheel path longitudinal crack in the low severity level...45
3.10 Non-wheel path longitudinal crack in the low severity level.46
3.11 Alligator crack in the moderate severity level..............47
3.12 Patch in the moderate severity level........................48
3.13 Pothole in the high severity level..........................49
3.14 Example of Rutting in the high severity level................50
3.15 Example of Roughness in the high severity level..............51
3.16 PCI factors and PCI scale...................................53
3.17 Transverse crack deduct value curve at each severity level..54
3.18 In-wheel path longitudinal crack deduct value curve
at each severity level.......................................55
3.19 Non-wheel path longitudinal crack deduct value curve
at each severity level.......................................55
3.20 Alligator crack deduct value curve at each severity level...56
3.21 Patch deduct value curve....................................56
3.22 Pothole deduct value curve..................................57
3.23 Rutting deduct value curve..................................57
IV


3.24 Roughness deduct value curve
58
3.25 PCI calculation from individual condition indexes............59
3.26 Testing area for the data quality evaluation.................69
4.1 Pre storm overall roads PCI distribution using the standard
method........................................................76
4.2 Post storm overall roads PCI distribution using the standard
method........................................................76
4.3 Pre storm overall condition PCI distribution for collector roads
using the standard method.....................................78
4.4 Post storm overall condition PCI distribution for collector roads
using the standard method.....................................78
4.5 Pre storm overall condition PCI distribution for local roads
using the standard method.....................................79
4.6 Post storm overall condition PCI distribution for local roads
using the standard method.....................................79
4.7 Pre storm PCI distribution with alligator crack..............81
4.8 Post storm PCI distribution with alligator crack...............81
4.9 Pre storm PCI distribution with transverse crack..............82
4.10 Post storm PCI distribution with transverse crack............82
4.11 Pre storm PCI distribution with Non-wheel path longitudinal
crack.........................................................83
4.12 Post storm PCI distribution with Non-wheel path longitudinal
crack.........................................................83
v


4.13 Pre storm PCI distribution with In-wheel path longitudinal
crack...........................................................84
4.14 Post storm PCI distribution with In-wheel path longitudinal
crack...........................................................84
4.15 Pre storm PCI distribution with patch..........................85
4.16 Post storm PCI distribution with patch.........................85
4.17 Pre storm PCI distribution with roughness......................86
4.18 Post storm PCI distribution with roughness.....................86
4.19 Pre storm PCI distribution with rutting........................87
4.20 Post storm PCI distribution with rutting.......................87
4.21 Local sector and Super segment definition ( Deighton, 2009)... 89
4.22 the Procedure of identifying 180 miles of same collector roads
from the windshield method data and the standard method data
.....................................................................93
4.23 Distribution of PCI difference between the standard method PCI
data and the windshield method PCI data
95


T ables
Table
1.1 Top 15 severe winter weather events from 1898 to 2008 (NOAA) 9
2.1 Geographic properties for the CCD road network in GIS.....24
3.1 An example of the inspection sheet used for the windshield
method................................................35
3.2 Distress extent on Bryant Street...........................61
3.3 Distress density for each distress type at each severity level on
Bryant Street.........................................63
3.4 Deduct value for each distress at each severity level on Bryant
Street................................................65
3.5 Condition index for each distress on Bryant Street.........66
3.6 PCI calculation of all the distresses on Bryant Street.....68
3.7 PCI value for roads in the testing area....................71
3.8 Comparison of average PCI value for testing roads..........72
4.1 Pre storm and Post storm road condition data using the
windshield method and the standard method.............74
4.2 Overall roads PCI comparison using the standard method....76
4.3 Overall condition PCI comparison for collector roads using the
standard method.......................................78
4.4 Overall condition PCI comparison for local roads using the
standard method.......................................79
Vll


4.5 Condition comparison for Alligator crack
81
4.6 Condition comparison for Transverse crack.................82
4.7 Condition comparison for Non-wheel path longitudinal crack... 83
4.8 Condition comparison for In-wheel path longitudinal crack.84
4.9 Condition comparison for patch............................85
4.10 Condition comparison for roughness.....................86
4.11 Condition comparison for rutting.......................87
4.12 An example of super road segments and their corresponding
block street segments.............................90
4.13 An example of difference of Weighted_Mean_PCI values of
standard method segments and WCI PCI values of
windshield method segments........................94
viii


1. Introduction
1.1 Background
Denver, Colorado is located in the South Platte River Valley on the High
Plains just east of the front range of the southern Rocky Mountains. Within
the 154 square mile area, Denver Public Works maintains roughly 1,900
centerline miles (3,058 km) of streets of which there are 600 miles (966 km) of
arterial & collector roads and 1,300 miles (2,092 km) of local roads. The City
and County of Denver (CCD) began systematic pavement management in
2000. The City realized that maintaining and repairing pavement within the
CCD road networks in an efficient way involves complex decisions. Hence,
the decision to move toward analytical pavement management systems
brought new thinking and techniques in the CCD.
The Pavement Management System (PMS) used by the CCD consists of three
major components: data collection, database management, and analytical
modeling. Data collection can be divided into arterial and collector roads and
local roads. A company called Roadware has been contracted since 1999 for
the arterial and collector road condition data collection. The data was
collected by driving on a road and collecting information at a constant speed.
l


This approach can be cost-effective for major busy roads such as highways or
arterial roads, but it can be costly when used in roads with low volumes of
traffic such as local streets or neighborhood collector streets. Therefore, the
CCD contracted the University of Colorado Denver (UCD) for the local street
and collector streets inspection since 2003. The equipment used by students
from UCD consists of a hand-held pocket computer running a custom ArcPad
application as shown in Figure 1.1. This simplified data collection allowed the
UCD team to collect inventory and network condition data based on map
information and customizable forms. The UCD research team manually
collects data on foot throughout the street network and records the appropriate
data which are ultimately stored on a secure digital server media.
Subsequently, all field data is downloaded into the PMS database on a weekly
basis to help support the pavement condition analysis and preservation
strategies. The detailed description of the data collection process using the
pocket computer will be discussed in Chapter 3.
2


Figure 1.1 Pocket computer
The PMS analysis models were configured using various factors. Normally,
the model can track the condition and deterioration of pavement and can
subsequently predict the condition of roads in the years between actual field
assessments. Occasionally, in situations where the actual deterioration of a
roadway does not match the analytical deterioration it can be difficult for city
planners to appropriately allocate budgets. This thesis details one such case
where unexpected extreme winter weather resulted in unpredicted accelerated
damage to the roadway network.
Denver's winter can vary from warm to mild to cold. In 2006/2007, the CCD
experienced an exceptionally harsh winter. For several weeks, the city was hit
with a series of intense storms which resulted in significant and accelerated
3


deterioration of the CCD road network. Figure 1.2 shows a photograph of
downtown Denver during one evening of one of the 2006/2007 snowstorms.
Figure 1.2 Denver downtown during the 2006/2007 snowstorm
Below is an excerpt quoted from ABC Denver Channel 7 about the Denver
holiday snowstorms of 2006/2007 winter:
"December 2006 brought one of the most historic winter weather events
in the state's history. Back-to-back blizzards struck the foothills and
eastern plains of Colorado during one of the busiest 10 days of the year...
By noon on Wednesday, Dec. 20, snow began to fly in the Denver area. It
began falling as early as 8 a.m. in the foothills west of town. Snow fell
along with gusty north winds for the next 24 to 36 hours, leaving much of
the Front Range under 1 to 3 feet of snow. Heavy snow even fell across
the eastern plains during the event, with as much as a foot in Washington,
Logan, and Phillips Counties. Southeastern Colorado saw amounts
generally between 6 and 12 inches. Gusty winds in excess of 30 mph
drifted the snow between 4 and 8 feet deep in exposed areas just to the
east and south of Denver. The storm made national headlines as it closed
Denver International Airport for two days, canceling some 2,000flights
4


and ruining the holiday travel plans for thousands of travelers connecting
through, flying to, or flying from the Mile High City.
As the storm exited Colorado and the recovery process began, forecasters
were busy tracking a second storm system following almost the same path
as the first. Historically, two snowstorms, each having the capability to
paralyze and stop a major city like Denver, were virtually absent from the
weather record. The only documented event that could compare was the
great storm in December 1913...
By Dec. 27, 2006, a new round of winter storm watches was in effect for
places still digging out from the first storm. By noon on Thursday, Dec.
28, the snow was flying across eastern Colorado, Denver, and the
foothills. At first the snow rates were light, with rain mixed in across
parts of the northern Denver Metro area, from Longmont to Firestone
and extending into northeastern Colorado around Greeley. But in time,
the snow filled in and the intensity picked up. By 7 p.m. on Dec, 28,
numerous phone calls and e-mails flooded 7NEWS with reports of
thunder and lightning in the northwest Denver Metro area and northern
foothills. Thunder-snow is simply a thunderstorm that drops snow instead
of rain. It is a sign of a very unstable atmosphere and usually indicates
heavy rates of snowfall, on the order of 2 to 4 inches per hour in many
cases. That is exactly what happened with the second of the twin holiday
blizzards. Communities from Lakewood to Golden and Evergreen to Estes
Park picked up 2-4 inches of snow per hour for several hours. By Friday
morning, the snow tapered off to showers in the foothills and the Denver
Metro area as the energy shifted onto the southeastern plains, but not
before dropping another 1 to 3 inches on the area...
While Denver and the foothills began the recovery process, ... Residents
found themselves buried alive in their homes. Drifts as high as the
rooftops blanketed homes and farm buildings; 12 to 36 inches of snow
had fallen during the storm, heaviest across the southeast counties. Drifts
were measured at lOto 15 feet deep, and up to 18 feet deep east of
Sheridan Lake in Kiowa County. Thousands of head of cattle were
stranded in the deep snow, and ranchers lost many of their herds, right in
the midst of calving season. Despite valiant efforts by ranchers and the
Colorado National Guard, hay dropping from military helicopters was
not sufficient to save many of the lost cattle. Many longtime ranchers and
farmers said that the late December storms of2006 were worse than the
October 1997 blizzard and as bad as any storm in memory. In the
mountains and foothills of southern Colorado, 30 to 48 inches of snow
were measured from the storm. The storm closed all major roads for days,
and smaller secondary roads for weeks. Food supplies ran low at stores
once citizens could get out of their homes, and merchants were quite
5


distressed at the timing of the storms, right in the heart of the big retail
season. (7 News, 2006)
Figure 1.3 shows the daily snow fall from Dec 15 to Jan 15 for the 2005/2006
(blue line), 2006/2007 (magenta line), and the 2007/2008 (yellow line) winters
(NOAA). The comparison of the snowfall for the three winters indicates that
three intense snow storms hit Denver in one month in the 2006/2007 winter,
while the 2005/2006 and 2007/2008 winters were comparatively dry.
IOCOt-^I-I^-OCVJIOOOt-tJ-
t- t- CNI CM CM CO i i-t-T-
OJ CM CM CM CM CM -A -c^
2005/ 2006
Wnter_Snow
*-2006/2007
Wnter_Show
2007/2008
Writer Stow
C&ys
Figure 1.3 Daily snow fall during 12-15 to 1-5 for the winters of 2005,
2006, and 2007
Figure 1.4 demonstrates a comparison of the daily low temperature from Dec
15 to Jan 15 for the 2005/2006 (blue line), 2006/2007 (magenta line) and
2007/2008 (yellow line) winters (NOAA). Comparing the data, the daily low
6


temperatures for the three winters were below the freezing point in the most of
evening time. During this period of time, pavements were subjected to
frequent freeze-thaw cycles
LI
C
c
£
£
<
-
2005/ 2006 Wnter_Low_Tenp
2006/2007 Wnter.LowTmp
2007/2008 Wntfif InwTrnn
C&ys
Figure 1.4 Daily low temperature during 12-15 to 1-15 for the winters
of 2005, 2006, and 2007
Figure 1.5 shows the daily high temperature from Dec 15 to Jan 15 for the
2005/2006 (blue line), 2006/2007 (magenta line) and the 2007/2008 (yellow
line) winters (NOAA). Comparing the data, the 2005/2006 winter was
comparatively warm. The 2006/2007 and 2007/2008 winters were similarly
cold. During the first and second storm period in 2006/2007 winter, the
temperature went up shortly after each storm so that helped to melt the snow
7


on the road. However, after the third snow happened on January 5 and January
6 (Figure 1.3), a sharp temperature decline in January below the average
prolonged the freezing and thaw cycles, severely impacting road conditions
(Figure 1.5).
2005/2006
Wndter_H ghTenp
-*-2006/2007
Writer _H gh_Tenp
2007/2008
Wnter Hah Term
^ ^ ^ ^ ^ ^ \T n? v* ^
O' o o O O' O O O O v v
C&ys
Figure 1.5 Daily high temperature during 12-15 to 1-15 for the winter
of 2005, 2006, and 2007
Table 1.1 shows the top 15 severe weather events that occurred from 1898 to
2008. Snow fall used in Table 1.1 refers to the monthly mean snow fall in
inches. Temperature in Table 1.1 refers to the monthly mean temperature in F
degree. December 2006 was ranked as the second snowiest month for
December in Denver historical weather data. January 2007 was ranked as the
8


fourth snowiest month for January, and ranked as the 14th coldest month for
January in historical weather data for Denver.
Table 1.1 Top 15 severe winter weather events from 1898 to 2008
(NO A A)
Snowiest December (1898-2008) Snowiest January (1898-2008) Coldest January (1898-2008)
Year Snow fall (inch) Year Snow fall (inch) Year Temperature(F)
1913 52.5 1948 35.0 1930 16.5
2006 45.5 1996 29.1 1937 18.0
1988 31.5 1949 28.2 1949 21.9
1967 31.4 2007 27.5 1979 22.6
1909 31.0 1962 25.1 1940 23.5
2007 30.0 1940 24.2 1963 23.5
1987 27.5 1987 21.4 1962 25.3
1979 23.5 1946 21.2 1916 25.4
1982 23.5 1945 20.3 1918 26.2
1989 21.6 1978 19.6 1922 26.4
2008 20.9 1997 19.0 1929 26.5
1926 20.8 1973 18.5 1985 26.8
1978 19.5 2002 18.5 1978 27.2
1972 19.5 1947 17.0 2007 27.2
1915 19.4 1991 17.0 1957 27.6
The 2006/2007 intense storms resulted in observed damage to the pavement
which was significantly greater than pavement conditions predicted by PMS
analytical model for a normal winter. Quickly quantifying the damage on the
road was needed to support the observed conclusion.
A few months before the onset of the winter, the CCD street maintenance
department assessed a twenty-two block arterial section on 18th Avenue from
Broadway to York using the standard assessment technique. Hence, this
twenty-two block arterial section was chosen as a test section to confirm the
9


deterioration in streets because it was inventoried just prior to the onset of the
2006/2007 storms. The 22 block section is shown in a red line in Figure 1.6.
Figure 1.6 A 22 block arterial section on 18th Avenue from
Broadway to York shown in a red line
Over the course of 2 days after the 2006/2007 winter had ended this roadway
section was re-assessed using the standard manual assessment technique. The
data before and after the storm were compared and a noticeable decrease in
the actual Pavement Condition Index (PCI) was observed, which indicated that
the road was damaged significantly by the snow storm. In other words, the
actual PCI was significantly lower than the PMS predicted PCI. Subsequently
the CCD decided to assess the damage on the entirety of its arterial and
collector roads in order to assess the long term impact on their work programs
10


and future budgets. Time constraints and safety issues prohibited the use of
the standard manual assessment method. Therefore, to quickly quantify these
effects, the windshield inspection method was developed by the CCD
engineering staff. Using this method, around 600 miles (966 km) of arterial
and collector roadways were quickly inspected in three weeks at an estimated
internal cost of $20,000. In order to re-establish reliable detailed condition
data for the PMS and to quantify the effectiveness of the windshield inspection
method, around 1,600 miles (2,575 km) of street networks inspected during
2002-2006 were again inspected and rated during a 15 month period from
2007 to 2008 using a standard assessment method. This 1,600 miles (2,575
km) of street network included 277 miles (446 km) of collector roads and
1,323 miles (2,129 km) of local roads. The cost to manually inspect the
network during the 15 month post storm period was approximately $120 per
mile in 2008 dollars. This was similar to the cost of manually inspecting the
pre-storm network which was approximately $100 per mile in 2008 dollars.
This thesis details the methodology used to complete the pavement condition
assessment and the analysis of pre-storm and post-storm pavement conditions
for the CCD road network.
The content of this thesis is organized and presented in a chapter wise manner
as follows:
li


Chapter One introduces the background, Pavement
Management Systems, Geographic Information
Systems, Pavement Condition Index and Objectives.
Chapter Two discusses the CCD pavement data management
(database and data management).
Chapter Three introduces pavement inspection and rating
methodology (windshield method, standard method,
and data quality assurance analysis).
Chapter Four presents the pavement condition data analysis
(overall condition PCI analysis and individual distress
PCI analyses).
Chapter Five discusses the overall conclusions and findings, and
possible future recommendations.
1.2 Pavement Management Systems
A Pavement management system is defined as: a procedure which provides a
systematic and consistent method for selecting maintenance and
12


rehabilitation needs and determining priorities, and the optimal time for
repair by predicting future pavement conditions (Muntasir, 2006).
It is hard to find the exact answer of when methodically managing pavement
networks first began. In many aspects, pavement management systems started
with AASHO Road Test in 1956. The method developed for the road test was
based on a pavements serviceability (Finn, 1998). Regarding as to how to
estimate the serviceability of the pavement, Carey and Irick proposed a
system where a group of people rode over selected sections of pavement and
their opinion was recorded regarding the quality of the ride. Then, physical
measurements were taken for each of the pavement sections and correlated
with the subjective responses (Carey and Irick, 1960).
In the 1970s, the Highway Research Board sponsored a workshop that
discussed the structural design of asphalt concrete pavement systems. The
participants discussed the positives and negatives of pavement management.
Dr. Karl Pister, a professor of civil engineering at the University of California
at Berkeley, emphasized to the participants that system engineering was one
way to tackle the complicated problems faced in pavement management and
design. After that, there was considerable controversy over the use of the
word system as it applies to pavement management. From 1968 to 1980, a
number of engineers and scientists pointed out that a pavement management
system was a good idea and that they were willing to utilize such techniques.
13


The first generation of pavement management systems consisted mainly of a
database, a condition index, and a ranking system that assisted in developing a
prioritized list of projects (Finn, 1998). To quote Finn, The ranking system
subjectively weighed factors, such as roughness, cracking of various kinds,
raveling, rutting and spalling, and produced a combined score or index. Some
condition surveys included as many as 15 categories to be evaluated and
recorded. Indices of this kind are still used by many agencies as a way to
summarize pavement conditions within a specific network. The U.S. Army
Corps of Engineers developed a somewhat more rational way of calculating
an index, but in the final analysis, it too was based largely on engineering
judgment (Finn, 1998).
Since 1980, new ideas evolved regarding the development of PMSs where
pavement management systems developed into network level management that
here to fore had focused primarily on individual projects. The best
recommendation at the network level and individual project levels may not be
the same, however. This determination depends largely only whether the
decision is to allocate resources over a wide distribution of a network
improving the overall network to a satisfactory level, or if resources are to be
focused on individual projects achieving a higher level of conditions but on a
more individual basis. The solution largely depends on the overall resources
14


and funding available, as well as prioritization of network projects in a given
area (Finn, 1998).
Meanwhile, pavement management has developed globally with a growing
consortium of nations that have contributed to its refinement, developing it
into a complete system that can address virtually any road network conditions.
Today, PMSs play an integral role in maintaining transportation
infrastructures around the world. PMSs have developed into a complex
analytical approach using a host of computerized mapping, GIS, and analysis
tools. PMSs pull from a wide range of disciplines including civil engineering,
business, finance, and computer science to create an overall systemic approach
that addresses the multifactoral issues that arise when evaluating an overall
pavement network system (Kulkami, 2003).
1.3 Geographic Information Systems (GIS)
Geographic Information System is defined as Geographic Information
System (GIS) is an integrated collection of computer software and data used
to view and manage information about geographic places, analyze spatial
relationships, and model spatial processes. A GIS provides a framework for
gathering and organizing spatial data and related information so that it can
be displayed and analyzed (ESRI dictionary, 2006).
15


Spatial information compared with other information entails a richness of
complexity as it gains information from two attributes: where and what. For
thousands of years, people developed maps for navigation through unfamiliar
terrain and seas. Its historical use emphasized representing the accurate
location of physical features on the earth. Hence forward, analysis of mapped
data has become an important part of understanding and managing geographic
space. This new perspective drove forward the use of spatial information from
one of emphasizing physical locations on earth to one of systemically
analyzing mapped data and spatially characterizing data into a conjoined geo
spatial relationship that characterizes a GIS (Berry, 2006).
Since 1960s, the decision-making process has hailed in the use of
mathematical models as quantitative approaches have become the convention.
Before the advent of computerized mapping, spatial analyses were constrained
to manual procedures. The computer has streamlined the processing of both
spatial and analog data. GIS development includes the early computer
mapping in 70s, the spatial database management systems of the 80s, and
map analysis modeling in the 90s. Computer mapping reigned in a new era of
computer cartography that utilized a digital computer mapping system from
the paper inking methods of mapping in the past. Spatial databases were
developed on a relational database system that joined spatial and tabular data
into a complete system, creating a geographic information system. As GIS
evolved, it turned to raster and vector forms of mapping analysis and
16


modeling. GIS today has developed various powerful spatial analysis tools,
such as surface analysis, distance analysis, statistical analysis, map algebra,
and graphical modeling. GIS has evolved from an emerging science to a
cornerstone of any geo spatial related project that helps solve problems,
increase productivity, and stay competitive (Berry, 2006).
A host of spatially integrated datasets are an essential component to pavement
management decision making. GIS technology, with its spatial analysis
capabilities is shown to be the most appropriate tool to enhance pavement
management operations. The PMS process can be build on a well-developed
GIS on which a set of functions can be provided, including thematic mapping,
a flexible database editor, linear referencing systems, dynamic segmentation,
statistics, charting, network generations and integration with external programs.
The key to effectively linking a pavement management system to GIS is
Dynamic Segmentation. Dynamic segmentation is the process of
transforming linearly referenced data (also known as events) that have been
stored in a table into features that can be displayed and analyzed on a map
(Cadkin, ESRI 2002). For Example, as shown in Figure 1.7, the pavement
management system requires segmenting streets dynamically according to the
condition of the road. Attribute information describing the road condition
characteristics specific to each road segment can then be maintained without
splitting the road network. The dynamic segmentation process necessitates
17


two requirements of the data.
Each event in an event table must include a unique identifier and position
along a linear feature. Each linear feature must have a unique identifier and
measurement system (Cadkin, ESRI 2002).
Good
Fair
L
Poor
&
2 4 6
.v J A
11111
a
Figure 1.7 Dynamic segmentation
1.4 Pavement Condition Index
The pavement condition index was developed by the U.S. Army Corps of
Engineers Construction Engineering Research laboratory. It was a
maintenance tool for the US Air Force in the 1970s developed primarily to
18


assist in the allocation of funds for Air Force bases that was necessary to
maintain their structures (McNemey, 2008).
The PCI is a numerical index between 0 and 100 and is used to indicate the
condition of a roadway condition with 100 representing the best possible
condition and 0 being the worst possible condition. Through a visual survey of
the pavement, PCI values can be obtained. This method can be used on both
asphalt surfaces as well as jointed portland cement concrete (PCC) pavements.
The following general procedure is used to determine the PCI value of the
pavement (Headquarters, U.S. Army Corps of Engineers. (1989).
1. ) Divide pavements into features.
2. ) Divide pavement feature into sample units.
3. ) Inspect sample units, determine distress types and severity levels and
measure density.
4. ) Determine deduct values
5. ) Compute total deduct value
6. ) Adjust total deduct value
7. ) Compute pavement condition index
8. ) Compute PCI of entire feature (average PCIS sample units)
The PCI can be used to trigger the maintenance when the pavements
condition reaches a certain level. It can also be used to determine the extent
and cost of repair, determine a network condition index by combining the PCI
19


score for each individual road segment, and it allows for equal comparison of
different pavements. Since a pavement condition score accounts for all types
of pavement performance measures it can therefore then be used to compare
two or more pavements with different problems equally (Deighton, 1998).
1.5 Objectives
This thesis addresses the following goals:
Presentation of the procedures and details of the windshield condition
assessment method. In addition, inspection of around 600 miles (966
km) of the CCD arterial and collector roads using this method will be
discussed which will quickly quantify the storm impact on the
pavement
Presentation of the procedures and details of the standard manual
pavement assessment method. In addition, inspection of around 1600
miles (2,575 km) of the CCD local streets and collector streets using
this method will discussed which obtained reliable detailed condition
data for the PMS
A detailed study of the effects of the storm on the pavement condition
was completed
Explore the correlation between the windshield and standard methods
20


2. CCD Pavement data Management
2.1 CCD Pavement Database
At the heart of any pavement management system is the database, which is
used as a archive of historical and descriptive data regarding the road network.
The pavement database provides useful input for accurately reporting on and
evaluating current network conditions, forecasting life cycle costs for different
maintenance and repair treatments, and developing annual and long range
budgets and repair plans.
The database used by the CCD PMS possesses several features, including:
static segmentation, dynamic segmentation, concurrent transformation,
multiple location reference methods, ad hoc queries, a large capacity, user
friendly access, and flexibility for future expansion. The concurrent
transformation feature allows for the simultaneous operation of merging any
static segmentation with attributes. The database also supports multi-location
reference methods which include mile point/kilometer point, mile
post/kilometer post, reference point, reference post, and reference section. The
database includes the entire CCD street network. The CCD uses this database
to make multiple and complex calculations quickly and efficiently. The data
21


collected and stored in the database can be divided into three categories: road
networks in GIS, inventory data, and pavement condition data. Each of these
data categories are described below:
GIS Road networks
The GIS road network is comprised of static road segmentations which is the
fundamental part of the CCD pavement database which is internally developed
and maintained by DenverGIS. DenverGIS is basically the CCD GIS
department that manages the development, maintenance, and distribution of
Denver's comprehensive spatial GIS information and related databases. The
static road segment was defined based on a block by block basis. The CCD
road network is divided into three functional classifications including arterial,
collector and local streets. The arterial streets are approximately 280 miles
(450 km) which carry heavy volumes of traffic. Collector streets are designed
to collect traffic from neighborhoods and distribute it to arterial streets. The
CCD has approximately 320 miles (515 km) of collector roads. Local roads in
the CCD are approximately 1300 miles (2,092 km) which carry local and light
traffic. Figure 2.1 shows the entire CCD road network using GIS. Figure 2.2
shows the three types of streets with red line representing the arterial roads,
blue line representing the collector roads, and green line representing the local
roads. Table 2.1 illustrates the geographic properties of the CCD road network
using GIS.
22


Figure 2.1 GIS Road networks for the entire CCD area
Legend
Streets_a rc
VOLCLASS
...- ARTERIAL
--- COLLECTOR
LOCAL
Figure 2.2 Arterial roads are in the red line, Collector roads are in
the blue line and Local roads are in the green line
23


Table 2.1 Geographic properties for the CCD road network in GIS
Projected Coordinate System NAD_1983_HARN_StatePlane_Colorado_Central_FIPS_0502_Feet
Projection Lambert_Conformal_Conic
False_Easting 3000000.00031608
False_Northing 999999.99999600
Central_Meridian -105.50000000
StandardParalleM 38.45000000
Standard_Parallel_2 39.75000000
Latitude_Of_Origin 37.83333333
Linear Unit Foot_US
Geographic Coordinate System GCS_North_American_1983_HARN
Datum D_North_American_1983_HARN
Prime Meridian Greenwich
Angular Unit Degree
Inventory data:
Inventory data is a collection of the physical characteristics of the pavement,
and it usually does not change between maintenance activities. The most basic
information about the road is included to reference the pavement, such as the
road name, location (referencing system), number of lanes, width and length of
the road, and pavement type. Figure 2.3 shows an example of the inventory
data in the PMS database.
24


P Q R S Y AA
1 FULLNAME FROMNAME TONAME LEN Ml VOLCLASS TRAVLANES IANEWI
2 FILLMORE SIB E43RDAVE E44THAVE 0.08700 LOCAL 200000
3 W25TH AV3 NHOOKER ST N IRVING ST 0.08400 LOCAL 200000
4 IRVING ST3 W25THAVE W26THAVE 0.08300 COLLECTOR 200000
5 W43RD AV4 N CLAY ST N DECATUR ST 0.08200 LOCAL 200000
6 WOLFF ST1 W33RDAVE W34THAVE 0.06200 LOCAL 200000
7 GROVE S103 WSCOTT PL W46THAVE 008400 LOCAL 200000
8 GAYLORD SI1 E STEAVENSON PL/NMCHG E47THAVE 0.06500 LOCAL 200000
9 W47TH AV3 NALCOTT ST NBEACH CT 0.06200 LOCAL 200000
to QUIVAS S12 W46THAVE W47THAVE 0.08300 LOCAL 200000
11 49TH AV02 N CLARKSON ST N EMERSON STlNMCHG 0.05600 LOCAL 200000
12 COOK ST2 E44THAVE E45THAVE 0.06700 LOCAL 200000
13 36TH AVI N HUMBOLDT ST N FRANKLIN ST 0.06500 LOCAL 200000
14 14TH AV NPENNSYLVANIA ST NPEARL ST 0.06300 ARTERIAL 300000
15 16TH ST1 CENTRAL ST BOULDER ST 0.09200 LOCAL 200000
16 W45TH AV4 N WYANDOT ST N WYANDOT ST 0.02100 LOCAL 200000
17 38TH AV4 N STEELE ST NADAMSST 0.06300 LOCAL 200000
18 EMERSON ST2 E48THAVBNMCHG E49THAVEMHG 0.10100 LOCAL 200000
19 37TH AVI N MARION ST N LAFAYETTE ST 0.06600 LOCAL 200000
20 37TH AVI N DOWNING ST N MARION ST 0.06600 LOCAL 200000
21 PERRY ST2 W37THAVE W38THAVE 0.12600 LOCAL 200000
22 TEJON ST2 W43RDAVE W44THAVE 0.08400 COLLECTOR 200000
23 JACKSON STS E30THAVE E 31 ST AVE 0.11800 LOCAL 200000
24 BOTH AV5 N GARFIELD ST N RICHARD ALLEN CT 0.05100 LOCAL 200000
25 W46TH AV N QUIVAS ST NSHOSHONE ST 0.08300 COLLECTOR 200000
26 VINE ST3 E17THAVE E1BTHAVE 0.10900 LOCAL 200000
27 W12TH AV5 N PECOS ST7RRX RRX 0.03000 LOCAL 200000
28 W35TH AV2 NSHOSHONE ST NTEJONST 0.08500 LOCAL 200000
29 STEELE ST5 E11THAVE E12THAVE 0.10100 LOCAL 200000
30 W26TM AV N CLAY ST N DECATUR ST 0.08200 ARTERIAL 200000
31 BROADWAY! BLAKE ST RROVPASS 0.15400 ARTERIAL 200000
32 2BTH ST WELTONST CALIFORNIA ST 0.06400 LOCAL 200000
33 W39TH AVI N WINONA CT N WOLFF ST 0.06200 LOCAL 200000
34 INTERSTATE 25 AURARIA PKWY/STR mm 0.01400 ARTERIAL 400000
35 DOWNING ST1 E23RDAVE E24THAVE 0.08700 ARTERIAL 200000
36 CENTRAL ST 17THST TOST 0.09100 COLLECTOR 200000
37 SQUITMANST1 W CEDAR AVE WBYERS PL 0.10900 LOCAL 200000
38 25TH AVI N HUMBOLDT ST N FRANKUN ST 0.06600 LOCAL 200000
39 ALCOTT ST3 W40THAVE W41 ST AVE 0.08300 LOCAL 200000
40 GRANT SI5 E16THAVE E17THAVE 0.11000 ARTERIAL 300000
41 CLAY ST3 W25THAVE W26THAVE 0.08400 COLLECTOR 200000
42 WILLIAMS ST2 E28THAVE E29THAVE 0.08700 LOCAL 200000
43 W26TH AV N JULIAN ST NK1NGST 0.04300 ARTERIAL 200000
44 MEADE STD7 W COLFAX AVE W CONEJOS PL 0.08100 LOCAL 200000
45 9TH AV3 NADAMSST N COOK ST 0.05900 LOCAL 200000
46 17TH AVI DENVER ZOO RD/N STEELE ST NADAMSST 0.06500 ARTERIAL 400000
47 W49TH AV2 NKWGST N LOWELL BLVD 0.06400 LOCAL 200000
48 ADAMS STB E7TH AVENUE PKWY E8THAVE 0.12300 LOCAL 200000
il JL. / - NRACF ST NVINFST JUmiDCAL , , imnm
Figure 2.3 An example of inventory data in the PMS database
25


Condition data
Condition data refers to information about the past and present surface
condition of a section of pavement. Accurate historical pavement condition
information is absolutely essential for the operation of the pavement
management system because all system recommendations are ultimately based
on past and present condition data. Figure 2.4 shows an example of the
pavement condition data in the PMS database.
26


2 F11M0RE STB
CL
CP CQ
CS CT
IE LONG WP L LONG WP M LONG WP H LONG L LONG M
!3 W25IH AV3 0 oo 60 OO 0 OO 1 OO 1 OO -1.00000 0
4 N43RD AV4 18 oo 120 OO 0 oo 1 OO 1 OO LOB 42
:5 m sn 6 oo 24 oo 0 oo 12 oo 0 oo 0.00000 32
; 6 GROVE SI 0 oo 75 oo 0 oo 0 oo 32 oo 0.00 0
:7 GAYLORD ST1 0 oo 44 oo 16 OB 0 oo 0 OB 7500 1
18 m AV3 1 oo 1 oo 1 OO 9 oo 0 OO 0.00 0
7 QUIVAS ST2 43 oo 0 oo 0 oo 0 oo 0 OO 1500 0
7 41 AV02 45 oo 0 oo 0 oo 120 oo 0 oo 0.00000 180
:ii COOK S12 1 oo 1 oo 1 oo 1 oo 1 oo LOO 0
*12 £114 AVI 42 oo 0 oo 0 oo 1 oo 1 DO LOO 0
;13 11 S11 0 oo 12 oo 0 oo 0 oo 3 OO 000000 0
7 WH AV4 0 oo 0 oo 66 oo 1 oo 1 OO 1.00 0
I £1H AV4 0 oo 22 oo 0 oo 0 oo 50 oo 000000 0
7 EMERSON ST2 1 oo 1 DO 1 OB 1 oo 1 oo LOO 1
;17 31 AVI 1 oo 1 oo 1 oo 1 oo 1 oo LOO 0
7 31 AVI 0 oo 0 oo 0 oo 0 03 24 OB moo 0
:is PERRY ST2 3 oo 87 oo 0 oo 1 oo 1 OO 1 00000 0
7 JACKSONS! 313 oo 475 oo 0 oo 156 oo 0 oo 000000 188
121 m AV5 0 oo 171 oo 0 oo 1 oo 1 oo 1 00000 0
:22 Vi SB 0 oo 0 oo 0 oo 1 oo 1 oo LOO 1
;23 Wll AV5 0 oo 0 oo 316 oo 0 oo 170 oo 000000 0
7 m AV2 0 oo 0 oo 671 oo 0 00000 0 oo 108 00000 0
7 STEELE ST5 0 oo 100 oo 0 oo 1 oo 1 oo LOO 60
7 m sr 1 oo 1 oo 1 oo 0 oo 99 oo 000000 0
;27 m Avi 50 oo 0 oo 0 oo 30 oo 0 oo 000000 1
7 SQUBMANST1 145 oo 0 oo 0 oo 0 oo 80 oo 000000 0
H 21 AVI 0 oo 210 oo 51 oo 1 03 1 oo LOB £
30 ALC0T1 SB 118 oo 0 oo 0 oo 0 oo 6 oo 000000 0
7 MSB 0 oo 0 oo 0 oo 0 oo 180 oo 0.00 0
7 MEADE SI 1 oo 1 OB 1 oo 87 oo 90 oo 000000 0
7 1 AV3 0 oo 127 oo 8 oo 0 oo 54 oo 9400 0
7 m AV2 0 oo 212 oo 27 oo 0 oo 21 oo 27.00000 1
7 ADAMS SB 0 oo 11 oo 0 oo 0 oo 1 oo 0.00000 210
LONG H
3000000
l irans m
1.00000
L E
00000 12400000 0 00000
aims 3400000
0 00000 45 00000 405
120
00000 050 00000 250
31.00000 I
30
72
14400000 108
36 00000 432
100000
00000 24600 36
240
Figure 2.4 An example of pavement condition data in the PMS database
27


2.2 CCD Pavement Data Management Process
The basic procedures required to manage and operate the CCD pavement data
include data collection, data synchronization, input and data preparation as
shown in Figure 2.5.
Data
Preparation
Data collection
and inputting
into database
Figure 2.5 Pavement data management process in the CCD
28


Data collection
A custom ArcPad application was used for the data collection including map
information and customizable forms that allowed the data collection teams to
collect the data using the Pocket PC. The ArcPad software was linked to the
database. The data collected from the field was stored on the secure digital
media for later synchronization and inputting into the database at the office as
shown in Figure 2.6.

] QHWolfl b i
jstreets ^ p B pc, t£ & [it]
$ ArcPad 0 -4-10:11
Individual Segnient Details
Road |QJRT1S ST
Fran
15THST
16THST
303592
Length (if) Width
I490j5 I ET~
Last Inventory [
Corritm Assessn
Ruttng
QBIodiOk
[] Patches
Q Transverse Gac
[] Long Cracking
[] Long Crackrg
Rutting Seventy
Q Low Severity
(Less than l)2 depth)
[] Moderate Severity
(Between 1/2' and 1" depth)
QHtfi severity
(Greater than 1,5'depth)
Q Bctreme Severity
(Between 1" and 1,5" depth)
fjj ArcPad g i-10:37
Figure 2.6 Field data collection using ArcPad
29


Data synchronization
Data Synchronization is the process of merging the newly collected data to the
existing database. The field data was collected and stored on the different
secure digital media. In order to download the data from multiple data files
into the database at the same time, the data has to be synchronized first to
create one single file which contains all the data from different sources. After
synchronization, the data was imported into the database. At this point, the
data can be queried or used by other data analysis functions.
Data preparation
Data Preparation, on the other hand, essentially takes the information output
from the database and creates the files that are used on the Pocket PC during
the data collection.
30


3. Pavement Inspection and Rating methodology
This chapter presents two pavement inspection and rating methodologies used
for this study. One is the windshield method and the other one is the standard
method. The windshield method was developed by the CCD engineers for the
purpose of rapid assessment of the road condition. The standard method was
developed by Deighton company as a supportive approach to the CCD road
maintenance strategy. The detailed description about the two methods is
shown below.
3.1 Windshield Method
The windshield inspection method was developed by the CCD due to the need
to quickly quantify the effect of 2006/2007 intensive winter on the pavement
network. It is a customized visual rating system using vehicles and six
engineers to quickly collect the condition data of the pavement. The condition
data then was calculated as the windshield condition index (WCI). The WCI
was less detailed than the standard inspections method. Therefore calibration
to the standard pavement condition index was completed.
31


3.1.1 Data Collection for the Windshield Method
The windshield inspection about 595 miles (958 km) of arterial and collector
roads was completed during a two week period in March 2007 using a
shortened list of distresses. A cursory ride through the city revealed that roads
that previously had average to moderate distress levels had failed. This was
around 6 weeks after the intense storm period and was prompted by knowledge
obtained from assessing the twenty-two block arterial roads on 18th avenue as
shown in Figure 1.6 that indicated the roads had distressed greatly. Arterial
and collector road segments were driven and four distress criteria were
inspected and assigned a 1-5 rating which indicated the condition of the
roadway with 5 representing the best possible condition and 1 being the worst
possible condition. The four criteria includes patches and potholes as shown
in Figure 3.1, alligator cracking as shown in Figure 3.2, overall crack
development (longitudinal and transverse) as shown in Figure 3.3, and general
condition as shown in Figure 3.4 (rutting, roughness, etc.).
32


Figure 3.1 Potholes and patches damage Level l(high severity)
Figure 3.2 Alligator cracking -Level l(high severity)
33


Figure 3.3 General cracking Level l(high severity)
Figure 3.4 Overall condition Level 3 (moderate severity)
The windshield inspection required some subjective analysis and teams of two
were mixed on a daily basis in order to try to avoid bias. Each team would
34


drive the segment at approximately 25 miles per hour and agree upon a rating
for each of the four distress types for each segment. The team recorded their
inspection results on a prepared form inspection sheet that had the windshield
inspection classification levels. Table 3.1 shows an example of the inspection
sheet used for the windshield method.
Table 3.1 An example of the inspection sheet used for the windshield
method
Street From To Patches Alligator Cracks General
Monaco Pkwy. S Montview 26th Ave. 3 1 3 2
Quebec 8th Ave. Colfax Ave. 3 1 3 2
10th Ave. Federal Blvd. Knox Ct. 3 1 2 1
10th Ave. Knox Ct. Perry St. 3 1 2 1
13th Ave. Holly St. Monaco 2 2 3 3
13th Ave. Dahlia St. Holly St. 3 2 3 3
13th Ave. Colorado Blvd. Dahlia St. 3 3 2 3
17th Ave. Holly St. Monaco 4 4 3 3
17th Ave. 18th Ave. Colorado Blvd 2 3 1 1
35


3.1.2 WCI Calculation
WCI calculation description
The windshield condition index (WCI) was created based on the existing
pavement performance curve in the CCDs PMS developed by Deighton
(Deighton, 2009), a consultant company who provides software solutions for
infrastructure asset management as shown in Figure 3.5
Pavement Performance Curve
Figure 3.5 Pavement performance curve used in the CCD road
network
The pavement performance curve is a function of age and PCI, where the older
the age of the road, the lower PCI value. The idea was to use this curve to
create the equation that could be used to calculate the deduct value for each
individual distress. In order to create the curve shape as shown in Figure 3.5,
the 5 level rating system for the windshield method was applied to X axis
36


representing the condition levels in the range of 1 to 5, and the Y axis
represents the deduct value in the range of 0 to 10 as shown in Figure 3.6. The
WCI deduct value curve was formulated using a best fit model that
approximated the pavement performance curve. Therefore, condition level 1
was assigned a deduct value of 10 and condition 5 had a deduct value of 0 as
the curve boundaries. Then by approximating the trend in the pavement
performance curve, condition level 2 was assigned a deduct value of 8 and
condition level 4 was assigned a deduct value of 1 and condition level 3 was
assigned a deduct value of 4.
Figure 3.6 Windshield analysis deduct value curve
The equation developed from the deduct value curve is shown as Equation 3.1.
37


DVj= Sj(-0.833x4+1.333x3-6.917x2+10.667x+5)
(3.1)
where: i = distress condition (patches/potholes, alligator cracks, general cracks,
overall condition)
DV = deduct value
S = Severity factor (2.5 for patches and alligator cracks, 1.5 for cracks
and 1.0 for general)
x = distress score (1 -5)
Each of the four distress categories had varying significance in the overall
condition of the pavement. Therefore, a severity factor was applied to each of
the measured distress categories so as to weigh the distresses appropriately.
The severity factor was determined based on the CCD engineers knowledge
and experience on street condition performance in the Denver local area.
Patches and potholes were deemed the most serious of distresses while general
conditions had smaller weight factors. Three Weighting factors were assigned
with 1 for general conditions, 1.5 for cracks, and 2.5 for patches/potholes and
alligator cracks.
The WCI for each road segment is calculated by subtracting the total deduct
value from each of the four distresses from a starting value of 100 as shown in
Equation 3.2.
WCI = 100-EDVi (3.2)
where: i = distress category (patches/potholes, alligator cracks, general cracks,
overall condition)
DV = deduct value
38


An Example of WCI calculation
1.) Field data collection
Two engineers from the CCD evaluated 10th Ave from Speer Blvd to
Broadway using the windshield inspection method. Four distresses were
found with pothole/patch in condition level 4, alligator cracks with the
condition level 4, general cracks with condition level 3, and the overall
condition with condition level 5.
2.) Calculate the deduct value for each distresses using equation 3.1
Patch/pothole deduct value = Si(-0.833x4+1.333x3-6.917x2+10.667x+5)
=2.5(-0.833*(4)4+1.333*(4)3-6.917*(4)2
+10.667*(4) + 5
=2.5
Alligator deduct value = Si(-0.833x4+1.333x3-6.917x2+10.667x+5)
=2.5(-0.833*(4)4+1.333*(4)3-6.917*(4)2
+10.667*(4) +5
=2.5
^ General crack deduct value = Sj(-0.833x4+1.333x3-6.917x2+10.667x+5)
=1.5(-0.833*(3)4+1.333*(3)3-6.917*(3)2
+10.667*(3) +5
=6
-v* Patch/pothole deduct value = Sj(-0.833x4+1.333x3-6.917x2+10.667x+5)
=l(-0.833*(5)4+1.333*(5)3-6.917*(5)2
+10.667*(5) +5
=0
3.) WCI Calculation using equation 3.2
WCI = 100 L DVi = 100- (2.5 + 2.5 + 6 + 0) = 89
39


3.1.3 WCI Transformation to PCI
WCI transformation to PCI description
The pavement condition data was stored in the PMS as the standard PCI.
Therefore, calibration of WCI to the standard manual PCI was necessary. A
sampling of the twenty-three arterial road segments inspected using the
windshield method was selected for the standard condition analysis. By rating
select segments by both methods, a scatter plot was created to explore the
correlation of these two datasets as shown in Figure 3.7.
WCI PCI Conversion
Figure 3.7 WCI/PCI conversion curve
40


A second order equation 3.3 was fitted to the data with a correlation coefficient
of 0.9265 and can be used to convert WCI values to PCI values. The equation
3.3 allowed for a direct quantification of the impacts of the 2006/2007 winter
season and will be discussed in the chapter 4.
PCI = 0.007(WCI)2 + 0.259(WCI) -2.3577 (3.3)
An Example of WCI transformation to PCI
The WCI value of 89 calculated from the WCI calculation example was used
for the PCI calculation using equation 3.3.
PCI = 0.007(WCI)2 + 0.259(WCI) -2.3577
= 0.007(89)2 + 0.259(89) -2.3577
= 76
3.2 Standard Method
The standard manual inspection method was developed by the CCD and
Deighton, a consulting company who provides software solutions for
infrastructure asset management. This method is mainly used for local and
collector streets with a low average daily traffic because the inspection is
completed on foot. It is comprised of data collection and PCI calculation. The
41


PCI calculation is used to plan for in-house annual preventative maintenance
requirements and to define major rehabilitation projects. The CCD has
effectively used the standard assessment method to develop and prioritize
successful, high impact projects to rehabilitate its aged roads.
3.2.1 Data Collection for the Standard Method
Every five years, 1,573 miles (2,531 km) of local and collector roads are
inspected using the standard manual procedure. Prior to the 2006/2007 winter
season, the UCD research team inspected the CCD road networks during a 4
year period from 2002 to 2006. The next inspection was planned to begin in
2010 based on the five years cycle of data collection. Due to the 2006/2007
intense snow impact on the CCD road network, the inspection was completed
two years in advance. This time the inspection was completed in a ten month
period from November 2007 to August 2008. The post-storm inspection began
around 10 months after the 2006/2007 storm period and was prompted by the
knowledge from the windshield road condition analysis that roads had
distressed greatly. The research team manually walked over each individual
road segment to record and measure distress types and severity using the hand
held computer. Eight distress types were analyzed including transverse crack,
In-wheel path longitudinal crack, Non-wheel path longitudinal crack, alligator
42


crack, pothole, patches, rutting and roughness and are described below. Three
subjective severity levels were introduced as low, medium, and high for each
distress. In addition, the approximate extent of each distress (e.g., surface area
of cracks) needed to be estimated. The definition used for each distress was
referenced by the distress identification manual developed by Federal
Highway Administration (FHWA, 2009).
43


Transverse Crack
Description
Transverse cracking is cracks that are predominantly perpendicular to
pavement centerline as shown in Figure 3.8. It may be caused by
shrinkage of the asphalt surface due to low temperatures or hardening
of the asphalt or caused by cracks beneath the surface course.
Transverse cracking is measured in linear feet. If the crack does not
have the same severity level along its entire length, each portion should
be recorded separately.
Figure 3.8 Transverse crack in the low severity level
Severity Levels
Low
A transverse crack with a mean width V" or less.
Moderate
A transverse with a mean width V" and
High
A transverse crack with a mean width >
44


In-wheel Path Longitudinal Crack
Description
Longitudinal Cracks are predominantly parallel to the pavement
centerline and are contained within the defined wheel paths of traffic as
shown in Figure 3.9. They may be caused by a poorly constructed
paving lane joint or shrinkage of the asphalt surface due to low
temperatures. They may also be the result of hardening of the asphalt or
poor substructure. In Wheel Path Longitudinal Cracking is measured in
linear feet. If the crack does not have the same severity level along its
entire length, each portion is to be recorded separately.
f
V
Figure 3.9 In-wheel path longitudinal crack in the low severity level
Severity Levels
Low
Any light spalling crack with a mean width V or less.
MODERATE
Any moderate spalling crack with a mean width Vi and V.
HIGH
Any severe spalling crack with a mean width > V".
45


Non-wheel Path Longitudinal Crack
Description
Mon-Wheel path longitudinal crack is the same as In-wheel path
longitudinal crack with the exception that the crack is not contained
within the defined wheel paths as shown in Figure 3.10. The cause for
this type of distress is as the same as the in wheel path longitudinal crack.
Non-Wheel Path longitudinal crack is measured in linear feet. If the
crack does not have the same severity level along its entire length, each
portion should be recorded separately.
Figure 3.10 Non-wheel path longitudinal crack in the low severity
level
Severity Levels
Low
A non-wheel path longitudinal crack with a mean width Vi'.
Moderate
A non-wheel path longitudinal crack with a mean width V and
High
A non-wheel path longitudinal crack with a mean width >
46


Alligator Crack
Description
Alligator cracking is a series of interconnected cracks in the early stages
of development, which eventually develops into many sided, sharp-
angled pieces characteristically in an alligator pattern in later stages as
shown in Figure 3.11. Alligator crack is measured in square feet of
surface area. The major difficulty in measuring alligator cracking is that
more than one level of severity may exist within one distressed area. If
these portions can be easily distinguished from each other, measurement
should be recorded separately. However, if the different levels of
severity cannot be easily divided, the entire area should be rated at the
highest, most severe level present.
Figure 3.11 Alligator crack in the moderate severity level
Severity Levels
Low
An area of alligator crack which has no or a few interconnecting cracks.
Moderate
An area of alligator crack which has interconnected cracks forming a
complete pattern.
High
An area of alligator crack which has pattern cracking progressed so that
pieces are well defined and spalled at the edges.
47
Try


Patch
Description
A patch is the portion of pavement surface that has been removed and
replaced with asphalt filler or additional materials after original
construction as shown in Figure 3.12. A patch is measured in square
feet of surface area. If a single patch has areas of different severity
levels, these areas are measured and recorded separately. Any other
distress found in a patched area will not be recorded; however, its
effects on the patch need to be considered when determining the patch
severity level.
Figure 3.12 Patch in the moderate severity level
Severity Levels
Low
A low severity level patch is in good condition and is performing
satisfactorily.
Moderate
A moderate level patch has moderate severity distress of any type and
affects riding quality to some extent.
High
A high severity level patch is badly deteriorated and affects riding
quality significantly.
48


Pothole
Description
Potholes are bowl-shaped holes of various sizes in the pavement surface
as shown in Figure 3.13. They generally have sharp edges and vertical
sides near the top of the hole. Generally, potholes are the end result of
alligator cracking. As alligator cracking becomes severe, the
interconnected cracks create small chunks of pavement, which can be
dislodged as vehicles drive over them.
Figure 3.13 Pothole in the high severity level
Severity Levels
Low
A pothole is less than 1 deep;
Moderate
A pothole is between 1 and 2 deep;
High
A pothole is greater than 2 deep;
49


Rutting
Description
A rut is a longitudinal surface depression in the wheel path as shown in
Figure 3.13. It is usually caused by consolidation or lateral movement of
the materials due to traffic loads. Rutting is measured by visual
subjective judgment. Precise measurement of the rut depth was not
completed in this study.
Figure 3.14 Example of Rutting in the high severity level
Severity Levels
Low
A rut in which its mean depth is less than 'A.
Moderate
A rut in which its mean depth is between V" and 1.
High
A rut in which its mean depth is between 1 and 1.5.
Extreme
A rut in which its mean depth is greater than 1.5.
50


Roughness
Description
Roughness is shown in Figure 3.14 and is defined as the wearing away
of the pavement surface caused by the dislodging of aggregate particles
and loss of asphalt. Rutting is measured by visual judgment.
Figure 3.15 Example of Roughness in the high severity level
Severity Levels
Low
Low severity roughness is pavement where the aggregate has started to
wear away.
Moderate
Moderate severity roughness is pavement where the aggregate has worn
away. The surface texture is moderately rough.
High
High severity roughness is pavement where the aggregate has worn
away. The surface texture is severely rough.
51


3.2.2 PCI Calculation for the Standard Method
The raw pavement distress data collected from the field needs to be converted
into PCI before it can be used to rank project importance or make decisions
about rehabilitation (Aaniewski and Hudson, 1987). As shown in Figure 3.15,
the PCI is a function of distress type, extent and severity. A six level PCI scale
ranging from (0-100) is used to describe the pavement condition. This scale
has a range from very good (91-100), good (76-90), fair (66-75), fair to poor
(56-65), poor (41-55) and very poor (0-40) (NCPP, 1998). The PCI of a road
segment is automatically computed when the distress data from the field is
synched to the main database. The following steps describe the procedure to
compute the PCI of a road segment which was referenced by the Deighton
configuration report used in the CCD:
52


Figure 3.16 PCI factors and PCI scale
Determine distress extent: the total of each distress type at each severity
level is added up.
Determine distress density: the total quantity of each distress type at
each severity level is divided by the total area of the sample unit and
multiplied by 100 to obtain the percent of density for each distress type
and severity.
Determine deduct values: the deduct value for each distress type
(Figure3.17 through 3.24) and each severity level is determined by using
the deduct value curve for that particular distress type. The deduct value
for patch and pothole at different condition levels were all calculated only
53


based on a high deduct value curve because these two types of distresses
were identified as an indicator that the roads were in severe condition.
Figures 3.17, 3.18, 3.19, 3.20, 3.21, 3.22, 3.23 and 3.24 illustrate the
deduct values based on the distress density respectively for each individual
distress.
Figure 3.17 Transverse crack deduct value curve at each severity
level
54


Figure 3.18 In-wheel path longitudinal crack deduct value curve
at each severity level
Figure 3.19 Non-wheel path longitudinal crack deduct value curve
at each severity level
55


Alligator Crack Deduct Values
100.00 q-
DeckEt 10.00
1.00-


IFt
Lav Deduct
Mxl Deduct
Hcji Deduct
1.00
10.00
Extent (%)
100.00
Figure 3.20 Alligator crack deduct value curve at each severity level
100.00 , p at< ch D e du ct Values I I -t-M-H-l-
t "
rwtirt mnn - --

High Deduct
L^QUCl IU.UU a

1 nn -


I.UU 1. 00 E> 1C te .00 nt (%) 1 "h 00.00
Figure 3.21 Patch deduct value curve
56


Pothole Deduct Values
100.00
Deduct 10.00
1.00
h-
High Deduct
1.00
10.00
Extnt (%)
100.00
Figure 3.22 Pothole deduct value curve
Figure 3.23 Rutting deduct value curve
57


Figure 3.24 Roughness deduct value curve
Determine the condition index (Cl) for each distress: the condition
index for each individual distress as shown in Equation 3.4 is determined
by summarizing the deduct value of all severity levels and subtracting the
sum value from 100. However, the condition index value for each distress
has to be adjusted to 0 if the result value is less than 0. The condition
index for rutting and roughness assigned the extreme high severity level a
value of 0, the high severity level a value of 40, a moderate severity level
with a value of 60 and the low severity level a value of 80.
CIm = 100-£(DVi)m (3.4)
Where: i= each severity level
m= distress type
58


PCI
Once the individual condition indexes are calculated, the PCI can be calculated
The PCI is an average of all individual condition indexes minus one standard
deviation to account for the overall condition of the road segment as shown in
Figure 3.25. It is important to note as well that the In-wheel path longitudinal
crack condition index, Non-wheel path longitudinal crack condition index and
the transverse crack condition index are averaged together first, prior to being
included within the PCI calculation.
ci
(Transverse
crack)
CI
(In-wheel path
long crack)
CI
(Non-wheel path
long crack
Figure 3.25 PCI calculation from individual condition indexes
59


An Example of PCI calculation for the standard method
Bryant Street from west 38th Ave to west 39th Ave is presented here as an
example of the PCI calculation. Bryant Street has the length of 449 feet and
the width of 30 feet whose area is 13,470 square feet.
1.) Filed data collection
The eight types of distress were quantified at each severity level. The
measurement unit used for the field data collection is feet or square feet.
2.) Distress extent for each distress type at each severity level
The quantity of each distress at each severity level was summarized as shown
in table 3.2. Rutting and roughness have no quantity value and will be
assigned individual distress condition indexes later. Therefore, any quantity
calculation process was not applied to these two distresses.
60


Table 3.2 Distress extent on Bryant Street
Alligator (Ft2) Long WP (Ft) Long Non WP (Ft) Transverse (Ft) Patch (Ft2) Pothole (Ft2) Rutting Roughness
Low 1000 100 200 50 200 200 L L
Moderate 2500 100 200 100 400 400
High 1000 100 200 150 600 600


3.) Distress density for each distress type at each severity level as shown in
table 3.3. The density value for alligator crack, patch, and pothole was
calculated based on the percentage of the street area. The density value for in-
wheel path longitudinal crack, Non-wheel path longitudinal crack and
transverse were based on the percentage of the street length.
62


Table 3.3 Distress density for each distress type at each severity level on Bryant Street
Alligator Long WP Long Non WP Transverse Patch Pothole Rutting Roughness
Low 7.42% 11.13% 14.84% 5.57% L L
Moderate 18.56% 11.13% 14.84% 11.14%
High 7.42% 11.13% 14.84% 16.70% 8.91% 8.91%


4.) Deduct value calculation for each distress type at each severity level
The deduct values are calculated based on the deduct curve for each distress as
shown from Figure 3.17 to Figure 3.24. The table 3.4 shows the deduct value
for each distress at each severity level calculated based on the deduct curve.
64


Table 3.4 Deduct value for each distress at each severity level on Bryant Street
Alligator Long WP Long Non WP Transverse Patch Pothole Rutting Roughness
Low 5.18 11.04 2.03 1.16 L L
Moderate 29.29 31.52 7.89 10.09
High 40.46 41.46 20.01 28.90 42.38 42.38


5.) The condition index for each distress
The condition index for each distress is calculated by subtracting the total of
deduct values at each severity level from 100. Below is an example of the
alligator condition index calculated by the deduct values from Table 3.4 using
equation 3.4:
The alligator condition index = 100 E(DVi)m
= 100-(5.18+29.29+40.16)
= 25.07
Table 3.5 shows the condition index value for each distress. The condition
index for rutting and roughness use the same condition index classification
system with the low level having a condition index of 80, the moderate level
with the condition index of 60, the high level with the condition index of 40,
and the extremely high level with a condition index of 0. In this example,
rutting and roughness were rated as L which represents the low severity level
having the condition index of 80.
Table 3.5 Condition index for each distress on Bryant Street
Alligator Long WP Long Non WP Transverse Patch Pothole Rutting Roughness
Cl 25.07 15.98 70.07 59.85 57.62 57.62 80.00 80.00
66


6.) PCI calculation
The PCI value is calculated based on each individual distress condition index
as shown in the Table 3.5 using the procedure as shown in Figure 3.25. First
all, the average condition index of In-wheel path long crack, Non-wheel path
long crack and transverse crack need to be calculated which is called as
AVG l. Then the PCI is an average of all condition indexes minus one
standard deviation. Table 3.6 shows the value of PCI calculation. Below is the
PCI calculation procedure:
4- AVG_1= (15.98 + 70.07 + 59.85) /3 = 48.63
+ Average of all index = (25.07+48.63+57.62++57.62+80+80)/6 = 58.16
Standard Deviation of all index
={ [(25.07-58.16)2 + (48.63-58.16)2 + (57.62-58.16) 2+(57.62-58.16) 2+(80-
58.16)2
+ (80-58.16)2]/6}05
= 20.69
$* PCI = Average of all index Standard Deviation of all index
= 37.47
67


Table 3.6 PCI calculation of all the distresses on Bryant Street
Alligator Long WP Long Non WP Transverse Patch Pothole Rutting Roughness
Cl 25.07 15.98 70.07 59.85 57.62 57.62 80.00 80.00
Avg _1 25.07 48.63 57.62 57.62 80.00 80.00
PCI 37.47


3.2.3 Data Quality Analysis for the Standard Method
In order to control the quality of data, a comparison of data collected by
different inspectors on a given road segments were analyzed for similarities
and differences. Twenty-two segments of CCD roads were used for the data
quality evaluation as shown in Figure 3.26.
'3
' Pi a
Aepgren
Pal
J. .
tP
Vandertt 4
p* $
Vande$. -I
Part Pond yg
WMasofpAK
V-.'f
ill
M
cm
i : K-.H1 . :
Pappas -~
Pak

- to nan
-P k
J*
'0W
> Lai! Pi -

fRuldy,:
HiiPak
i OiMiand fi
;;.Goif Coast i
W Evans A*
(gl
OvePand
23CCSJ?
m
Harvard-
GtfchPtfi
OUmatory
Park
y>* f5K5MsjC"CS,TAjW-@ if
Figure 3.26 Testing area for the data quality evaluation
L K
69


Eight teams of two people from the UCD research team were utilized to assess
the same target area independently and without knowing the purpose of the
task in advance. In addition to the team, two CCD supervisors also assessed
the same area. Table 3.7 shows the PCI value of the overall condition for
twenty-two road segments inspected by the UCD research team and the two
supervisors from the CCD. T1 to T8 represents the 8 teams from UCD. SI and
S2 represent the two supervisors from the CCD.
70


Table 3.7 PCI value for roads in the testing area
Road Name T1 T2 T3 T4 T5 T6 T7 T8 S1 S2
SEMERSON St 65.5 67.5 55.8 79.2 72.3 79.7 60.3 66.3 71.9 68.2
IOWA AV1 77.4 71.0 53.2 73.7 75.4 66.9 71.9 73.6 75.6 73.5
SOGDEN ST1 68.5 66.3 59.1 51.5 71.1 74.6 53.4 69.2 64.6 64.1
IOWA AV1 57.2 77.2 59.0 73.9 73.3 75.6 69.9 77.4 62.0 64.2
IOWA AV2 47.8 42.4 9.7 15.2 37.5 69.9 12.9 41.0 42.0 47.5
SCORONA ST1 65.3 56.8 58.8 57.8 55.5 55.9 62.3 67.2 63.0 63.0
IOWA AV3 76.4 83.1 69.1 79.3 80.3 85.5 77.3 85.0 79.2 79.6
SOGDEN ST2 62.3 62.7 55.7 57.3 61.2 59.9 45.7 69.8 57.2 54.2
MEXICO AV1 76.3 77.3 75.6 76.4 80.3 78.7 72.0 83.9 81.2 78.8
SCORONA ST4 67.2 56.1 57.7 54.8 56.2 52.5 43.1 68.3 62.3 62.5
MEXICO AV1 78.8 73.7 78.5 79.5 84.8 84.2 74.6 89.7 90.2 85.0
SEMERSON ST1 68.3 66.8 54.6 48.4 62.8 62.0 56.9 74.1 59.4 62.5
MEXICO AV2 75.1 69.4 66.4 79.7 77.4 90.7 70.3 76.5 77.5 73.5
MEXICO AV3 76.9 69.3 67.5 71.3 73.4 82.5 71.9 76.6 76.6 68.9
COLORADO AV1 71.7 65.2 60.5 41.9 53.4 60.2 54.4 73.9 47.4 57.9
SCORONA ST2 59.6 55.9 20.6 21.1 62.6 54.0 60.8 60.0 53.3 52.6
COLORADO AV2 60.6 63.4 57.9 62.8 50.1 48.6 58.4 69.9 50.3 50.3
SOGDEN ST2 70.5 61.9 63.3 79.1 74.1 67.6 61.6 73.9 70.8 69.6
COLORADO AV3 61.5 63.4 51.5 49.3 62.3 50.6 54.0 60.9 60.0 65.9
SEMERSON ST2 73.5 68.3 65.0 75.2 72.8 66.2 62.8 77.5 76.0 76.0
COLORADO AV4 58.5 55.1 35.2 15.7 32.3 51.5 12.7 60.4 43.0 43.0
SCORONA ST3 65.4 57.6 19.4 21.2 56.1 68.3 53.4 51.2 62.7 60.6
SOGDEN ST3 67.8 61.0 62.2 73.0 70.9 67.8 61.2 77.5 61.6 61.6
SEMERSON ST3 66.7 73.2 65.0 54.1 73.9 87.6 60.2 83.4 70.9 70.9
As shown in table 3.8, both CCD supervisors concluded that the average PCI
value of roads in the testing area was around 65. The average PCI range
obtained from the UCD research team was from 54 to 70 in which 6 out of 8
71


had average PCI values in the range of 60 to 70. The results of the data
comparison indicate a high level of data collection and rating consistency.
Table 3.8 Comparison of average PCI value for testing roads
Inspector PCI AVG Condition Level
Teaml 67 Fair
Team2 65 Fair to poor
Team3 54 Poor
Team4 60 Fair to Poor
Team5 66 Fair
Team6 70 Fair
Team7 56 Fair to Poor
Team8 70 Fair
Supervisor 1 65 Fair to Poor
Supervisor2 65 Fair to Poor
72


4. Pavement Condition Analysis
For a 10 month period from November, 2007 to August, 2008 after the
2006/2007 winter storm, the UCD research team collected a total of 1,573
miles (2,531 km) of road condition data using the standard method. The 1,573
miles (2,531 km) consisted of 277 miles (446 km) of neighborhood collector
roads and 1,296 miles (2,086 km) of local roads as shown in table 4.1. For a 3
week period in March, 2007 after the storm, the CCD engineers collected 595
miles (958 km) of road condition data using the windshield method. The 595
miles (958 km) consisted of 272 miles (438 km) of neighborhood collector
roads and 323 miles (520 km) of arterial roads as shown in table 4.1. In order
to complete the pre-storm and post-storm comparison of the road conditions,
pre storm data collected from 2002 to 2006 was needed. All pre-storm data
was collected using the standard method. 318 miles (512 km) of arterial roads
and 277 miles (446km) of collector roads were found in the pre-storm database
as shown in Table 4.1. However, for a number of reasons, only 939 miles
(1,511 km) of local roads existed in the pre-storm database with valid
condition data as shown in Table 4.1. In order to complete an accurate
comparison between the pre-storm and post-storm data using the standard
method, the exact 1,216 miles (1,957 km) of the same roads were compared.
The overall PCI and the individual distress PCIs were studied to compare pre
and post storm data collected by the standard method. Then the correlation of
73


the windshield method data and standard method data was studied. In order to
validate the accuracy of the windshield method, the same 180 miles (290 km)
of collector roads were compared between the standard method post storm data
and the windshield method post storm data.
Table 4.1 Pre storm and Post storm road condition data using the
windshield method and the standard method
Pre Storm Data (mile) Post Storm Data ( mi es)
Arterial Collector Local Arterial Collector Local
Windshield 0 0 0 323 272 0
Standard 318 277 939 0 277 1296
4.1 Pre-storm and Post-storm Road Condition Comparison
Analysis using the Standard Method
4.1.1 Overall Condition Comparison
Using the standard method of analysis, 1,216 miles (1,957 km) of roads were
compared between the pre and post storm database. Of these 1,216 miles
(1,957 km) of roads, 939 miles (1,511 km) are local roads and 277 miles (446
km) are neighborhood collector roads. The average Overall roads condition
74


was 77 for pre storm, with a PCI range from a low of 0 to a high of 100. After
the storm, the average overall condition of the roads was reduced to 70, with
PCI ranging from 0 to 100. In other words, the PCI values decrease an average
of 7 points due to the winter condition of 2006/2007. Table 4.2 demonstrates
the distribution of miles for overall roads using the standard condition PCI data.
Comparing the data, the post storm data indicates that there is an increase in
the very poor rating of the roads, 6% poor rating after the storm, from 2%
before storm, a 4% increase. Also, there is a 6% increase in post storm poor
pavement ratings comparing the post storm and pre storm conditions. The
very good rating of pavement condition also goes down to 15% from 26% in
2003-2005, an 11% decrease in the quality of road conditions. Figure 4.1
shows the pre storm overall roads PCI distribution with the standard analysis
and Figure 4.2 shows the post storm overall roads PCI distribution with the
standard analysis.
75


Table 4.2 Overall roads PCI comparison using the standard method
Condition Score Standard Overall Roads Condition Analysis
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-00 312.7 25.7% 183.7 15.1%
Good 76-90 364.7 30.0% 345.2 28.5%
Fair 66-75 218.2 17.9% 212.6 17.5%
Fair to poor 56-65 174.3 14.3% 205.2 16.9%
Poor 41-55 116.9 9.6% 189.2 15.6%
Very Poor 0-40 29.7 2.4% 77.3 6.4%
Figure 4.1 Pre storm overall roads PCI distribution using the
standard method
Figure 4.2 Post storm overall roads PCI distribution using the
standard method
76


The overall road condition comparison in terms of the road type was also
completed due to the fact that the road deterioration varies depending on the
road type. Table 4.3 demonstrates the distribution of miles based on the six
levels PCI scale for collector roads. Figure 4.3 shows the pre storm PCI
distribution for collector roads while Figure 4.4 shows the post storm PCI
distribution for collector roads. Similarly, compared to the overall road
conditions, the very good and good rating of roads had a substantial drop
compared to the pre storm condition data. As well, there was an increase in the
fair, fair to poor, poor and very poor conditions post storm. Table 4.4
demonstrates the distribution of miles based on the six levels PCI scale for
local roads. Figure 4.5 shows the pre storm PCI distribution for local roads
while Figure 4.6 shows the post storm PCI distribution for local roads. The
local roads also exhibited a similar decline in road conditions in the post storm
conditions, with a decrease in the very good rating in the post storm condition
data.
77


Table 4.3 Overall condition PCI comparison for collector roads using
the standard method
Condition Score Standard Overall Condition Analysis for Collector Roads
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-00 70.0 36.4% 33.2 17.1%
Good 76-90 79.4 41.3% 66.2 34.2%
Fair 66-75 22.7 11.8% 40.7 21.0%
Fair to poor 56-65 11.4 5.9% 30.2 15.6%
Poor 41-55 7.8 4.1% 16.9 8.8%
Very Poor 0-40 1.0 0.5% 6.2 3.2%
Q\fery good
Gbod
n Fai r
DFai r to poor
Fbor
a\ferv noor___
Figure 4.3 Pre storm overall condition PCI distribution for collector
roads using the standard method
Figure 4.4 Post storm overall condition PCI distribution for collector
roads using the standard method
78


Table 4.4 Overall condition PCI comparison for local roads using
the standard method
Condition Score Standard Overall Condition Analysis for Local Roads
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-00 242.8 23.7% 150.5 14.8%
Good 76-90 285.3 27.9% 279.1 27.4%
Fair 66-75 195.5 19.1% 171.9 16.9%
Fair to poor 56-65 162.1 15.8% 175.0 17.2%
Poor 41-55 109.9 10.7% 172.3 16.9%
Very Poor 0-40 28.6 2.8% 71.0 7.0%
Figure 4.5 Pre storm overall condition PCI distribution for local roads
using the standard method
Figure 4.6 Post storm overall condition PCI distribution for local roads
using the standard method
79


4.1.2 Individual Distress Comparison
Individual distresses analysis evaluated the 1,216 miles (1956 km) including
939 miles (1,511 km) local roads and 277 miles (446 km) neighborhood
collector roads between the pre-storm and post-storm database using the
standard method.
Comparing the data in the tables for the 2006/2007 winter storm season
indicates that the alligator crack (Table 4.5) and patch (Table 4.9) dramatically
increased due to the severe storm, while transverse cracking (Table 4.6), long
cracking not in wheel path (Table 4.7), long cracking in wheel path (Table 4.8),
roughness (Table 4.10), and rutting (Table 4.11) basically stayed at the same
level as they were before the storm. Figures 4.7, 4.8, 4.9, 4.10, 4.11, 4.12,
4.13, 4.14, 4.15, 4.16, 4.17, 4.18, 4.19 and 4.20 illustrate the data respectively
for each of the individual condition distresses.
80


Table 4.5 Condition comparison for Alligator crack
Condition Score Alligator crack
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-00 814.1 66.9% 684.3 56.4%
Good 76-90 208.7 17.2% 196.5 16.2%
Fair 66-75 59.0 4.9% 103.6 8.5%
Fair to poor 56-65 47.5 3.9% 77.8 6.4%
Poor 41-55 55.5 4.6% 84.1 6.9%
Very Poor 0-40 22.6 1.9% 67.0 5.5%
Vfery good
Gbod
Fai r
Fai r to poor
Fbor
\terv Fhnr
Figure 4.7 Pre storm PCI distribution with alligator crack
Figure 4.8 Post storm PCI distribution with alligator crack
81


Table 4.6 Condition comparison for Transverse crack
Condition Score Transverse crack
Pre-storm Pre-storm
miles percentage miles percentage
Very good 91-00 268.6 22.1% 248.5 20.7%
Good 76-90 457.4 37.6% 457.1 37.1 %
Fair 66-75 220.2 18.1% 211.9 17.7%
Fair to poor 56-65 148.9 12.2% 145.1 12.1%
Poor 41-55 102.4 8.4% 114.7 9.6%
Very Poor 0-40 19.0 1.6% 22.7 1.9%
n\fery good
Gbod
1=1 Fai r
1=1 Fai r to poor
Fbor
a\ferv Fhor____
Figure 4.9 Pre storm PCI distribution with transverse crack
Figure 4.10 Post storm PCI distribution with transverse crack
82


Table 4.7 Condition comparison for Non-wheel path longitudinal
crack
Condition Score Non-Wheel Path Longitudinal crack
Pre-storm Pre-storm
miles miles miles miles
Very good 91-00 299.4 24.6% 322.9 26.6%
Good 76-90 887.3 72.9% 861.7 71.0%
Fair 66-75 8.7 0.7% 8.1 0.7%
Fair to poor 56-65 0.5 0.04% 0.5 0.04%
Poor 41-55 20.7 1.7% 20.0 1.6%
Very Poor 0-40 0 0% 0 0%
Figure 4.11 Pre storm PCI distribution with Non-wheel path
longitudinal crack
Figure 4.12 Post storm PCI distribution with Non-wheel path
longitudinal crack
83


Table 4.8 Condition comparison for In-wheel path longitudinal crack
Condition Score In-Wheel Path Longitudinal crack
Pre-storm Pre-storm
miles percentage miles percentage
Very good 91-00 267.0 21.9% 291.6 24.0%
Good 76-90 600.5 49.4% 591.0 48.7%
Fair 66-75 162.9 13.4% 155.1 12.8%
Fair to poor 56-65 103.4 8.5% 96.7 8.0%
Poor 41-55 79.0 6.5% 75.3 6.2%
Very Poor 0-40 3.7 0.3% 3.5 0.3%
7 A 0 A
n Vfery good
Gbod
n Fai r
1=1 Fai r to poor
Fbor
a\ferv Fhor
49/
Figure 4.13 Pre storm PCI distribution with In-wheel path longitudinal
crack
Figure 4.14 Post storm PCI distribution with In-wheel path longitudinal
crack
84


Table 4.9 Condition comparison for patch
Condition Score Patch
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-100 577.2 47.4% 393.0 32.4%
Good 76-90 248.5 20.4% 296.2 24.4%
Fair 66-75 196.9 16.2% 238.1 19.6%
Fair to poor 56-65 109.6 9.0% 140.0 11.5%
Poor 41-55 71.8 5.9% 119.0 9.8%
Very Poor 0-40 12.5 1.0% 27.1 2.2%
Figure 4.15 Pre storm PCI distribution with patch
Figure 4.16 Post storm PCI distribution with patch
85


Table 4.10 Condition comparison for roughness
Condition Score Roughness
Pre-storm Post-storm
miles percentage miles percentage
Very good 91-100 986.7 81.1% 940.7 77.5%
Good 76-90 101.3 8.3% 93.4 7.7%
Fair 66-75 0 0% 0 0%
Fair to poor 56-65 95.4 7.8% 92.2 7.6%
Poor 41-55 0 0% 0 0%
Very Poor 0-40 33.2 2.7% 86.9 7.2%
Figure 4.17 Pre storm PCI distribution with roughness
OW
TA
77 DVfery good
Gbod
a Fai r
n Fai r to poor
Boor
a\ferv Root____
Figure 4.18 Post storm PCI distribution with roughness
86


Full Text

PAGE 1

PAVEMENT MANAGEMENT FOR THE CITY AND COUNTY OF DENVER COLORADOTHE IMPACT OF EXTREME WEATHER By BoYan B.S., Air Defense Institute of Technology, 2002 M.S., East China Normal University, 2006 A thesis submitted to the University of Colorado Denver in partial fulfillment of the requirements for the degree of Master of Engineering Civil Engineering 2009

PAGE 2

This thesis for the Master of Engineering Degree by BoYan has been approved by Kevin L. Rens

PAGE 3

Bo, Yan (M.Eng., GIS) Pavement Management for the City and County of Denver C olorado the impact of Extreme Weather Thesis directed by Professor Kevin L. Rens Abstract Pavement management systems (PMS) are a critical tool used to manage and maintain the transportation infrastructure around the world The City and County of Denver (CCD) has been managing its street network systematically since 2000 Pavement management depends on accurate pavement inventory and assessment related data. It depends on analysis models to prioriti z e rehabilitation efforts and provide upper level management and politicians with the knowledge they need to set future budgets. In 2006 / 2007 the Denver metropolitan area experienced an e x ceptionally har s h winter period during a 3 week span. The city was hit with a series of intense storms resulting in significant and accelerated deterioration to the CCD roadway network. The actual measured deterioration was significantly greater than that predicted by the analysis models in the existing PMS. Street Maintenance personnel of the CCD were aware that the pavement network had accelerated deterioration and

PAGE 4

needed to quickly quantify the damage and assess the long term impact on their work programs and future budgets To quickly quantify these effects, a windshield assessment method was developed. Using this method, a sample of 600 miles (966 km) of arterial and collector roadways were inspected in two weeks at an estimated internal cost on the order of $20,000 In order to re establish reliable detailed condition data for the PMS and to quantify the effectiveness of the windshield inspection method, the same I ,572 miles (2 530 km) of street networks inspected during 2002-2006 were again inspected and rated during a 15 month period from 2007 to 2008 using a standard assessment method This 1,572 mile street network included 277 miles ( 446 km) of collector roads and I ,295 miles (2 084 km) of local roads. The cost to manually inspect the network during the 15 month post storm period was approximately $120 per mile. This was similar to the cost of to manually inspecting the pre-storm network which totaled was approximately $100 per mile in 2008 dollars This thesis details the methodology used to complete the pavement condition inspection and assessment. A comparison of pre-storm and post-storm pavement conditions was completed for the CCD road network. The results indicate that the 2006/2007 extreme winter events accelerated aging of the CCD road networks. The overall pavement condition index (PCI) value decreased by an average of 7 percentage points Normal winter months age the CCD network 2 or 3 points in PCI values. Comparison of the windshield inspection technique to the standard inspection method revealed that there is a high degree of correlation between the two.

PAGE 5

This abstract accurately represents the content of the candidate's thesis. I recommend its publication. Kevin L. Rens

PAGE 6

ACKNOWLEDGEMENT My utmost gratitude goes to my thesis advisor, Dr. Rens for his contribution and support of my research, for his expertise, kindness continual encouragement and most of all for his patience. I believe that one of the main gains of this 3-year program was working with Dr Rens and gaining his trust and friendship. Thanks and appreciation goes to my thesis committee members Dr. Johnson and Dr. Durham, for their valuable participation and insight. I want to thank Mr. Patrick Kennedy and Ms Angie Hager from the C ity and County of Denver for their support and advice. I especially want to thank all the students who were involved in the data collection without the students, this research would be impossible. I have immensely enjoyed working with all of the undergraduate students Above all, I am greatly indebted to my husband and his family, who stood beside me and encouraged me every day I am thankful to all my friends for giving me happiness and joy and continuous support and interest in what I do. Finally, I would like to thank my parents and my sister whose love is boundless Without their support education in the united states would not have been possible

PAGE 7

Contents Figures ................................................................................... iii Tables ................................................................................... vii 1. Introduction .................................................................................................. 1 1.1 Background ........................................................................................ 1 1.2 Pavement Management Systems .................................................... 12 1.3 Geographic Information Systems (GIS) ........................................ 15 1.4 Pavement Condition Index .............................................................. 18 1.5 Objectives .......................................................................................... 20 2. CCD Pavement data Management ........................................................... 21 2.1 CCD Pavement Database ................................................................ 21 2.2 CCD Pavement Data Management Process .................................. 28 3. Pavement Inspection and Rating methodology ....................................... 31 3.1 Windshield Method .......................................................................... 31 3.1.1 Data Collection for the Windshield Method ....................... 32 3.1.2 WCI Calculation ................................................................... 36 3.1.3 WCI Transformation to PCI ................................................ 40 3.2 Standard Method ............................................................................. 41 3.2.1 Data Collection for the Standard Method .......................... 42 3.2.2 PCI Calculation for the Standard Method ......................... 52 3.2.3 Data Quality Analysis for the Standard Method ............... 69 4. Pavement Condition Analysis ................................................................... 73

PAGE 8

4.1 Pre-storm and Post-storm Road Condition Comparison Analysis using the Standard Method ............................................................. 74 4.1.1 Overall Condition Comparison ........................................... 74 4.1.2 Individual Distress Comparison .......................................... 80 4.2 Correlation Analysis between the Windshield Method Data and Standard Method Data using an Identical Road Dataset ............. 88 4.2.1 The Procedure of Identifying the Same Roads from the Windshield Method Post-storm Data and the Standard Method Post-storm Data ........................................................ 88 4.2.2 The Correlation Analysis of the Same Roads between the Windshield Method and the Standard Method ................... 94 5. Conclusion and Future Recommendation ............................................... 97 5.1 Conclusion ........................................................................................ 97 5.2 Discussion and Recommendation for the Future Work ............. 100 Bibliography ................................................................................................. 102 11

PAGE 9

Figures Fi2ure 1.1 Pocket computer ................................................................................. 3 1.2 Denver downtown during the 2006/2007 snowstorm ...................... 4 1.3 Daily snow fall during 12-15 to 1-5 for the winters of 2005,2006, and 2007 ............................................................................................ 6 1.4 Daily low temperature during 12-15 to 1-15 for the winters of 2005, 2006, and 2007 ........................................................................ 7 1.5 Daily high temperature during 12-15 to 1-15 for the winter of 2005, 2006, and 2007 ........................................................................ 8 1.6 A 22 block arterial section on 18th Avenue from Broadway to York shown in a red line ............................................................... 10 1.7 Dynamic segmentation .................................. .................................. 18 2.1 GIS Road networks for the entire CCD area ................................ 23 2.2 Arterial roads are in the red line, Collector roads are in the blue line and Local roads are in the green line .................................... 23 2.3 An example of inventory data in the PMS database ..................... 25 2.4 An example of pavement condition data in the PMS database ... 27 2.5 Pavement data management process in the CCD ......................... 28 2.6 Field data collection using ArcPad ................................................. 29 3.1 Potholes and patches damage-Levell(high severity) ................. 33 3.2 Alligator cracking -Levell(high severity) .................................... 33 lll

PAGE 10

3.3 General cracking-Level l(high severity) ...................................... 34 3.4 Overall condition-Level3 (moderate severity) ............................ 34 3.5 Pavement performance curve used in the CCD road network .... 36 3.6 Windshield analysis deduct value curve ........................................ 37 3.7 WCI/PCI conversion curve ............................................................. 40 3.8 Transverse crack in the low severity level. .................................... 44 3.9 In-wheel path longitudinal crack in the low severity level ........... 45 3.10 Non-wheel path longitudinal crack in the low severity level. ..... 46 3.11 Alligator crack in the moderate severity level. ............................ 47 3.12 Patch in the moderate severity level. ............................................ 48 3.13 Pothole in the high severity level .................................................. 49 3.14 Example of Rutting in the high severity level. ............................. 50 3.15 Example of Roughness in the high severity level ........................ 51 3.16 PCI factors and PCI scale ............................................................. 53 3.17 Transverse crack deduct value curve at each severity level ...... 54 3.18 In-wheel path longitudinal crack deduct value curve at each severity level ...................................................................... 55 3.19 Non-wheel path longitudinal crack deduct value curve at each severity level ...................................................................... 55 3.20 Alligator crack deduct value curve at each severity level.. ........ 56 3.21 Patch deduct value curve ............................................................... 56 3.22 Pothole deduct value curve ........................................................... 57 3.23 Rutting deduct value curve ........................................................... 57 lV

PAGE 11

3.24 Roughness deduct value curve ...................................................... 58 3.25 PCI calculation from individual condition indexes .................... 59 3.26 Testing area for the data quality evaluation ................................ 69 4.1 Pre storm overall roads PCI distribution using the standard method ............................................................................................. 76 4.2 Post storm overall roads PCI distribution using the standard method ............................................................................................. 76 4.3 Pre storm overall condition PCI distribution for collector roads using the standard method ............................................................ 78 4.4 Post storm overall condition PC I distribution for collector roads using the standard method ............................................................ 78 4.5 Pre storm overall condition PCI distribution for local roads using the standard method ............................................................ 79 4.6 Post storm overall condition PCI distribution for local roads using the standard method ............................................................ 79 4.7 Pre storm PCI distribution with alligator crack ........................... 81 4.8 Post storm PCI distribution with alligator crack .......................... 81 4.9 Pre storm PCI distribution with transverse crack ........................ 82 4.10 Post storm PCI distribution with transverse crack .................... 82 4.11 Pre storm PCI distribution with Non-wheel path longitudinal crack ................................................................................................ 83 4.12 Post storm PCI distribution with Non-wheel path longitudinal crack ................................................................................................ 83 v

PAGE 12

4.13 Pre storm PCI distribution with In-wheel path longitudinal crack ................................................................................................ 84 4.14 Post storm PCI distribution with In-wheel path longitudinal crack ................................................................................................ 84 4.15 Pre storm PCI distribution with patch ........................................ 85 4.16 Post storm PCI distribution with patch ....................................... 85 4.17 Pre storm PCI distribution with roughness ................................ 86 4.18 Post storm PCI distribution with roughness ............................... 86 4.19 Pre storm PCI distribution with rutting ...................................... 87 4.20 Post storm PCI distribution with rutting ..................................... 87 4.21 Local sector and Super segment definition (Deighton, 2009) ... 89 4.22 the Procedure of identifying 180 miles of same collector roads from the windshield method data and the standard method data ......................................................................................................... 93 4.23 Distribution of PCI difference between the standard method PCI data and the windshield method PCI data .................................. 95 Vl

PAGE 13

Tables Table 1.1 Top 15 severe winter weather events from 1898 to 2008 (NOAA) 9 2.1 Geographic properties for the CCD road network in GIS .......... 24 3.1 An example of the inspection sheet used for the windshield method ......................................... ........................................... 35 3.2 Distress extent on Bryant Street .......... ....... .................................. 61 3.3 Distress density for each distress type at each severity level on Bryant Street .......................................................................... 63 3.4 Deduct value for each distress at each severity level on Bryant Street ........................................................ .............................. 65 3.5 Condition index for each distress on Bryant Street. ..................... 66 3.6 PCI calculation of all the distresses on Bryant Street .................. 68 3.7 PCI value for roads in the testing area .......................................... 71 3.8 Comparison of average PCI value for testing roads ..................... 72 4.1 Pre storm and Post storm road condition data using the windshield method and the standard method ..................... 74 4.2 Overall roads PCI comparison using the standard method ......... 76 4.3 Overall condition PCI comparison for collector roads using the standard method .................................................................... 78 4.4 Overall condition PCI comparison for local roads using the standard method .................................................................... 79 Vll

PAGE 14

4.5 Condition comparison for Alligator crack .................................... 81 4.6 Condition comparison for Transverse crack ................................. 82 4.7 Condition comparison for Non-wheel path longitudinal crack ... 83 4.8 Condition comparison for In-wheel path longitudinal crack ...... 84 4.9 Condition comparison for patch ..................................................... 85 4.10 Condition comparison for roughness ........................................... 86 4.11 Condition comparison for rutting ................................................ 87 4.12 An example of super road segments and their corresponding block street segments ........................................................... 90 4.13 An example of difference of Weighted_Mean_PCI values of standard method segments and WCI_PCI values of windshield method segments ............................................... 94 Vlll

PAGE 15

1. Introduction 1.1 Background Denver Colorado is located in the South Platte River Valley on the High Plains just east of the front range of the southern Rocky Mountains. Within the 154 square mile area Denver Public Works maintains roughly 1 900 centerline miles (3,058 km) of streets of which there are 600 miles (966 km) of arterial & collector roads and 1,300 miles (2,092 km) oflocal roads. The City and County of Denver (CCD) began systematic pavement management in 2000. The City realized that maintaining and repairing pavement within the CCD road networks in an efficient way involves complex decisions Hence the decision to move toward analytical pavement management systems brought new thinking and techniques in the ceo. The Pavement Management System (PMS) used by the CCD consists of three major components: data collection database management, and analytical modeling. Data collection can be divided into arterial and collector roads and local roads. A company called Roadware has been contracted since 1999 for the arterial and collector road condition data collection The data was collected by driving on a road and collecting information at a constant speed.

PAGE 16

This approach can be cost-effective for major busy roads such as highways or arterial roads, but it can be costly when used in roads with low volumes of traffic such as local streets or neighborhood collector streets Therefore, the CCD contracted the University of Colorado Denver (UCD) for the local street and collector streets inspection since 2003. The equipment used by students from UCD consists of a hand-held pocket computer running a custom ArcPad application as shown in Figure 1 1 This simplified data collection allowed the UCD team to collect inventory and network condition data based on map information and customizable forms The UCD research team manually collects data on foot throughout the street network and records the appropriate data which are ultimately stored on a secure digital server media Subsequently, all field data is downloaded into the PMS database on a weekly basis to help support the pavement condition analysis and preservation strategies The detailed description of the data collection process using the pocket computer will be discussed in Chapter 3. 2

PAGE 17

Figure 1.1 Pocket computer The PMS analysis models were configured using various factors. Normally the model can track the condition and deterioration of pavement and can subsequently predict the condition of roads in the years between actual field assessments Occasionally, in situations where the actual deterioration of a roadway does not match the analytical deterioration it can be difficult for city planners to appropriately allocate budgets. This thesis details one such case where unexpected extreme winter weather resulted in unpredicted accelerated damage to the roadway network. Denver's winter can vary from warm to mild to cold. In 2006 / 2007 the CCD experienced an exceptionally harsh winter. For several weeks the city was hit with a series of intense storms which resulted in significant and accelerated 3

PAGE 18

deterioration of the CCD road network. Figure 1.2 shows a photograph of downtown Denver during one evening of one of the 2006/2007 snowstorms. Figure 1.2 Denver downtown during the 2006/2007 snowstorm Below is an excerpt quoted from ABC Denver Channel 7 about the Denver holiday snowstorms of2006/2007 winter: "December 2006 brought one of the most historic winter weather events in the state's history Back -to-hack blizzards struck the foothills and eastern plains of Colorado during one of the busiest 10 days of the year ... By noon on Wednesday, Dec. 20, snow began to fly in the Denver area. It began falling as early as 8 a.m. in the foothills west of town Snow fell along with gusty north winds for the next 24 to 36 hours leaving much of the Front Range under I to 3 feet of snow. Heavy snow even fell across the eastern plains during the event, with as much as afoot in Washington, Logan, and Phillips Counties. Southeastern Colorado saw amounts generally between 6 and 12 inches. Gusty winds in excess of 30 mph drifted the snow between 4 and 8 feet deep in exposed areas just to the east and south of Denver The storm made national headlines as it closed Denver International Airport for two days, canceling some 2 000 flights 4

PAGE 19

and ruining the holiday travel plans for thousands of travelers connecting through, flying to, or flying from the Mile High City. As the storm exited Colorado and the recovery process began, forecasters were busy tracking a second storm system following almost the same path as the first Historically two snowstorms, each having the capability to paralyze and stop a major city like Denver, were virtually absent from the weather record. The only documented event that could compare was the great storm in December 1913 ... By Dec. 27, 2006, a new round of winter storm watches was in effect for places still digging out from the first storm. By noon on Thursday Dec. 28, the snow was flying across eastern Colorado Denver and the foothills. At first the snow rates were light with rain mixed in across parts of the northern Denver Metro area, from Longmont to Firestone and extending into northeastern Colorado around Greeley. But in time, the snow filled in and the intensity picked up. By 7 p m on Dec, 28, numerous phone calls and e-mailsflooded 7NEWS with reports of thunder and lightning in the northwest Denver Metro area and northern foothills. Thunder -s now is simply a thunderstorm that drops snow instead of rain. It is a sign of a very unstable atmosphere and usually indicates heavy rates of snowfall, on the order of 2 to 4 inches per hour in many cases. That is exactly what happened with the second of the twin holiday blizzards Communities from Lakewood to Golden and Evergreen to Estes Park picked up 2-4 inches of snow per hour for several hours. By Friday morning the snow tapered off to showers in the foothills and the Denver Metro area as the energy shifted onto the southeastern plains, but not before dropping another 1 to 3 inches on the area ... While Denver and the foothills began the recovery process, . Residents found themselves buried alive in their homes. Drifts as high as the rooftops blanketed homes and farm buildings; 12 to 36 inches of snow had fallen during the storm, heaviest across the southeast counties. Drifts were measured at 1 Oto 15 feet deep and up to 18 feet deep east of Sheridan Lake in Kiowa County. Thousands of head of cattle were stranded in the deep snow, and ranchers lost many of their herds, right in the midst of calving season. Despite valiant efforts by ranchers and the Colorado National Guard hay dropping from military helicopters was not sufficient to save many of the lost cattle Many longtime ranchers and farmers said that the late December storms of 2006 were worse than the October 1997 bli zz ard and as bad as any storm in memory. In the mountains andfoothills of southern Colorado, 30 to 48 inches of snow were measured from the storm. The storm closed all major roads for days, and smaller secondary roads for weeks. Food supplies ran low at stores once citizens could get out of their homes, and merchants were quite 5

PAGE 20

distr e ss e d at the timing of the s torms right in the h e art of the big r e tail se ason." (7 News, 2006) Figure 1.3 shows the daily snow fall from Dec 15 to Jan 15 for the 2005 / 2006 (blue line) 2006 / 2007 (magenta line) and the 2007 /2 008 (yellow line) wint e rs (NOAA). The comparison of the snowfall for the three winters indicates that three intense snow storms hit Denver in one month in the 2006 / 2007 winter while the 2005 / 2006 and 2007 / 2008 winters were comparatively dry. 25 20 -..s:::: g15 '$ 5 0 1.()
PAGE 21

temperatures for the three winters were below the freezing point in the most of evening time. During this period of time, pavements were subjected to frequent freeze-thaw cycles 50 40 ;::; 30 20 ( 10 ------C"..C"...C"'C""'oo.oC"..C"""> 10 I I I I I I I I I -c-..,c-..,c-..,c-..,c-..,c-..,c-..,c-..,c-.., -20 Cays -+-2005/2COO W nt er _LowJ errp _._ '2ftJ3) 2007 Wnt er _LowJenp 2007/2 W t Lrm TP111l Figure 1.4 Daily low temperature during 12-15 to 1-15 for the winters of 2005, 2006, and 2007 Figure 1. 5 shows the daily high temperature from Dec 15 to Jan 15 for the 2005/2006 (blue line), 2006 / 2007 (magenta line) and the 2007/2008 (yellow line) winters (NOAA) Comparing the data, the 2005/2006 winter was comparatively warm The 2006 / 2007 and 2007/2008 winters were similarly cold During the first and second storm period in 2006 / 2007 winter, the temperature went up shortly after each storm so that helped to melt the snow 7

PAGE 22

on the road. However after the third snow happened on January 5 and January 6 (Figure 1 3), a sharp temperature decline in January below the average prolonged the freezing and thaw cycles, severely impacting road conditions (Figure 1 .5). -u (j = -(' (j c E
PAGE 23

fourth snowiest month for January and ranked as the 141h coldest month for January in historical weather data for Denver Table 1.1 Top 15 severe winter weather events from 1898 to 2008 (NOAA) Snowiest December Snowiest January Coldest January (1898-2008) (1898-2 0 08) (1898-2008) Year Snow fall Year Snow fall Year Temperature(F) (inch) (inch) 1913 52.5 1948 35.0 1930 16.5 2006 45.5 1996 29.1 1937 18. 0 1988 31.5 1949 28.2 1949 21. 9 1967 31.4 2007 27.5 1979 22.6 1909 31.0 1962 25.1 1940 23.5 2007 30.0 1940 24.2 1963 23 5 1987 27.5 1987 21.4 1962 25.3 1979 23.5 1946 21.2 1916 25.4 1982 23.5 1945 20.3 1918 26 2 1989 21.6 1978 19.6 1922 26.4 2008 20.9 1997 19.0 1929 26. 5 1926 20.8 1973 18.5 1985 26.8 1978 19.5 2002 18.5 1978 27. 2 1972 19.5 1947 17.0 2007 27. 2 1915 19.4 1991 17.0 1957 27.6 The 2006/2007 intense storms resulted in observed damage to the pavement which was significantly greater than pavement conditions predicted by PMS analytical model for a normal winter. Quickly quantifying the damage on the road was needed to support the observed conclusion A few months before the onset of the winter the CCD street maintenance department assessed a twenty-two block arterial section on 18th A venue from Broadway to York using the standard assessment technique. Hence this twenty-two block arterial section was chosen as a test section to confirm the 9

PAGE 24

deterioration in streets because it was inventoried just prior to the onset of the 2006 / 2007 storms. The 22 block section is shown in a red line in Figure 1 .6. Figure 1.6 A 22 block arterial section on 18th Avenue from Broadway to York shown in a red line Over the course of 2 days after the 2006 / 2007 winter had ended this roadway section was re-assessed using the standard manual assessment technique The data before and after the storm were compared and a noticeable decrease in the actual Pavement Condition Index (PCI) was observed which indicated that the road was damaged significantly by the snow storm In other words, the actual PCI was significantly lower than the PMS predicted PCI. Subsequently the CCD decided to assess the damage on the entirety of its arterial and collector roads in order to assess the long term impact on their work programs 10

PAGE 25

and future budgets Time constraints and safety issues prohibited the use of the standard manual assessment method. Therefore, to quickly quantify these effects, the windshie ld inspection method was developed by the CCD engineering staff. Using this method, around 600 miles (966 km) of arterial and collector roadways were quickly inspected in three weeks at an estimated internal cost of $20,000. In order to re-establish reliable detailed condition data for the PMS and to quantify the effectiveness of the windshield inspection method, around 1,600 miles (2,575 km) of street networks inspected during 2002-2006 were again inspected and rated during a 15 month period from 2007 to 2008 using a standard assessment method This 1 600 miles (2 575 km) of street network included 277 miles ( 446 km) of collector roads and 1 ,323 miles (2, 129 km) of local roads. The cost to manually inspect the network during the 15 month post storm period was approximately $120 per mile in 2008 dollars. This was similar to the cost of manually inspecting the pre-storm network which was approximately $100 per mile in 2008 dollars. This thesis details the methodology used to complete the pavement condition assessment and the analysis of pre-storm and post-storm pavement conditions for the CCD road network. The content of this thesis is organized and presented in a chapter wise manner as follows: II

PAGE 26

Chapter One introduces the background, Pavement Management Systems, Geographic Information Systems, Pavement Condition Index and Objectives Chapter Two discusses the CCD pavement data management (database and data management) Chapter Three introduces pavement inspection and rating methodology (windshield method standard method and data quality assurance analysis). Chapter Four presents the pavement condition data analysis (overall condition PCI analysis and individual distress PCI analyses) Chapter Five discusses the overall conclusions and findings, and possible future recommendations 1.2 Pavement Management Systems A Pavement management system is defined as : ua procedure which provides a systematic and consistent method for selecting maintenance and 12

PAGE 27

rehabilitation needs and determining priorities, and the optimal time for repair by predicting future pavement conditions" (Muntasir 2006). It is hard to find the exact answer of when methodically managing pavement networks first began In many a s pects pavement management systems started with AASHO Road Test in 1956 The method developed for the road test was ba s ed on a pavement s serviceability (Finn 1998) Regarding as to how to estimate the serviceability of the pavement Carey and Irick proposed a s ystem wher e a group of people rode over selected sections of pavement and their opinion was recorded regarding the quality of the ride. Then, phy s ical measurements were taken for each of the pavement s ections and correlated with the subjective res ponses (Carey and Irick 1960) In the 1970 's, the Highway Research Board sponsored a workshop that discussed the structural design of asphalt concrete pavement systems The participants discussed the positives and negatives of pavement management. Dr. Karl Pister a professor of civil engineering at the University of California at Berkeley emphasi z ed to the participants that system engineering was one way to tackle the complicated problems faced in pavement management and design After that there was considerable controversy over the use of the word "system as it applies to pavement management. From 1968 to 1980, a number of engineers and scientists pointed out that a pavement management system was a good idea and that they were willing to utili z e such techniques 13

PAGE 28

The first generation of pavement management systems consisted mainly of a database, a condition index, and a ranking system that assisted in developing a prioritized list of projects (Finn, 1998) To quote Finn, "The ranking system subjectively weighed factors, such as roughness, cracking of various kinds, raveling, rutting and spa/ling, and produced a combined score or index. Some condition surveys included as many as 15 categories to be evaluated and recorded. Indices of this kind are still used by many agencies as a way to summarize pavement conditions within a specific network. The US. Army Corps of Engineers developed a somewhat more rational way of calculating an index but in the final analysis, it too was based largely on engineering judgment" (Finn, 1998). Since 1980, new ideas evolved regarding the development of PMS 's where pavement management systems developed into network level management that here to fore had focused primarily on individual projects. The best recommendation at the network level and individual project levels may not be the same, however. This determination depends largely only whether the decision is to allocate resources over a wide distribution of a network improving the overall network to a satisfactory level, or if resources are to be focused on individual projects achieving a higher level of conditions but on a more individual basis. The solution largely depends on the overall resources 14

PAGE 29

and funding available, as well as prioritization of network projects in a given area (Finn 1998). Meanwhile pavement management has developed globally with a growing consortium of nations that have contributed to its refinement developing it into a complete system that can address virtually any road network conditions Today, PMS's play an integral role in maintaining transportation infrastructures around the world PMS's have developed into a complex analytical approach using a host of computerized mapping, GIS, and analysis tools. PMS's pull from a wide range of disciplines including civil engineering business, finance and computer science to create an overall systemic approach that addresses the multifactoral issues that arise when evaluating an overall pavement network system (Kulkarni 2003) 1.3 Geographic Information Systems (GIS) Geographic Information System is defined as "Geographic Information System (GIS) is an integrated collection of computer software and data used to view and manage informatioll about geographic places, analyze spatial relationships, and model spatial processes. A GIS provides a framework for gathering and organizing spatial data and related information so that it can be displayed and analyzed'' (ESRI dictionary, 2006) 15

PAGE 30

Spatial information compared with other information entails a richness of complexity as it gains information from two attributes : where and what. For thousands of years, people developed maps for navigation through unfamiliar terrain and seas. Its historical use emphasized representing the accurate location of physical features on the earth. Hence forward, analysis of mapped data has become an important part of understanding and managing geographic space This new perspective drove forward the use of spatial information from one of emphasizing physical locations on earth to one of systemically analyzing mapped data and spatially characterizing data into a conjoined geo spatial relationship that characterizes a GIS (Berry, 2006) Since 1960's, the decision-making process has hailed in the use of mathematical models as quantitative approaches have become the convention. Before the advent of computerized mapping spatial analyses were constrained to manual procedures The computer has streamlined the processing of both spatial and analog data GIS development includes the early computer mapping in 70's, the spatial database management systems of the 80's, and map analysis modeling in the 90's. Computer mapping reigned in a new era of computer cartography that utilized a digital computer mapping system from the paper inking methods of mapping in the past. Spatial databases were developed on a relational database system that joined spatial and tabular data into a complete system, creating a geographic information system. As GIS evolved, it turned to raster and vector forms of mapping analysis and 16

PAGE 31

modeling. GIS today has developed various powerful spatial analysis tools, such as surface analysis distance analysis, statistical analysis, map algebra, and graphical modeling GIS has evolved from an emerging science to a cornerstone of any geo spatial related project that helps solve problems, increase productivity, and stay competitive (Berry, 2006). A host of spatially integrated datasets are an essential component to pavement management decision making. GIS technology, with its spatial analysis capabilities is shown to be the most appropriate tool to enhance pavement management operations The PMS process can be build on a well-developed GIS on which a set of functions can be provided including thematic mapping, a flexible database editor linear referencing systems dynamic segmentation statistics, charting network generations and integration with external programs. The key to effectively linking a pavement management system to GIS is Dynamic Segmentation "Dynamic segmentation is the process of transforming linearly referenced data (also known as events) that have been stored in a table into features that can be displayed and analyzed on a map" (Cadkin, ESRI 2002). For Example, as shown in Figure 1.7, the pavement management system requires segmenting streets dynamically according to the condition of the road. Attribute information describing the road condition characteristics specific to each road segment can then be maintained without splitting the road network The dynamic segmentation process necessitates 17

PAGE 32

two requirements of the data Each event in an event table must include a unique identifier and position along a linear feature. Each linear feature must have a unique identifier and measurement system (Cadkin ESRI 2002). Good 4 r "' v Fair Poor Figure 1.7 Dynamic segmentation 1.4 Pavement Condition Index The pavement condition index was developed by the U.S. Army Corps of Engineers Construction Engineering Research laboratory It was a maintenance tool for the US Air Force in the 1970 s developed primarily to 1 8

PAGE 33

assist in the allocation of funds for Air Force bases that was necessary to maintain their structures (McNerney 2008) The PCI is a numerical index between 0 and 100 and is used to indicate the condition of a roadway condition with 100 representing the best possible condition and 0 being the worst possible condition Through a visual survey of the pavement PCI values can be obtained. This method can be used on both asphalt surfaces as well as jointed portland cement concrete (PCC) pavements. The following general procedure is used to determine the PCI value of the pavement (Headquarters, U.S Army Corps of Engineers ( 1989) 1.) Divide pavements into features 2.) Divide pavement feature into sample units. 3 ) Inspect sample units, determine distress types and severity levels and measure density 4 ) Determine deduct values 5.) Compute total deduct value 6.) Adjust total deduct value 7.) Compute pavement condition index 8.) Compute PCI of entire feature (average PCI S sample units) The PCI can be used to trigger the maintenance when the pavement s condition reaches a certain level. It can also be used to determine the extent and cost of repair, determine a network condition index by combining the PCI 19

PAGE 34

score for each individual road segment, and it allows for equal comparison of different pavements Since a pavement condition score accounts for all types of pavement performance measures it can therefore then be used to compare two or more pavements with different problems equally (Deighton, 1998). 1.5 Objectives This thesis addresses the following goals: Presentation of the procedures and details of the windshield condition assessment method. In addition inspection of around 600 miles (966 km) of the CCD arterial and collector roads using this method will be discussed which will quickly quantify the storm impact on the pavement Presentation of the procedures and details of the standard manual pavement assessment method In addition, inspection of around 1600 miles (2,575 km) of the CCD local streets and collector streets using this method will discussed which obtained reliable detailed condition data for the PMS A detailed study of the effects of the storm on the pavement condition was completed Explore the correlation between the windshield and standard methods 2 0

PAGE 35

2. CCD Pavement data Management 2.1 CCD Pavement Database At the heart of any pavement management system is the database, which is used as a archive of historical and descriptive data regarding the road network. The pavement database provides useful input for accurately reporting on and evaluating current network conditions, forecasting life cycle costs for different maintenance and repair treatments, and developing annual and long range budgets and repair plans The database used by the CCD PMS possesses several features including: static segmentation, dynamic segmentation, concurrent transformation, multiple location reference methods ad hoc queries, a large capacity, user friendly access, and flexibility for future expansion. The concurrent transformation feature allows for the simultaneous operation of merging any static segmentation with attributes The database also supports multi-location reference methods which include mile point/kilometer point, mile post/kilometer post, reference point, reference post, and reference section. The database includes the entire CCD street network. The CCD uses this database to make multiple and complex calculations quickly and efficiently. The data 21

PAGE 36

collected and stored in the database can be divided into three categories: road networks in GIS inventory data and pavement condition data. Each of these data categories are described below : GIS Road networks The GIS road network is comprised of static road segmentations which is the fundamental part of the CCD pavement database which is internally developed and maintained by DenverGIS. DenverGIS is basically the CCD GIS department that manages the development, maintenance, and distribution of Denver's comprehensive spatial GIS information and related databases The static road segment was defined based on a block by block basis. The CCD road network is divided into three functional classifications including arterial, collector and local streets The arterial streets are approximately 280 miles (450 km) which carry heavy volumes of traffic. Collector streets are designed to collect traffic from neighborhoods and distribute it to arterial streets. The CCD has approximately 320 miles (515 km) of collector roads. Local roads in the CCD are approximately 1300 miles (2,092 km) which carry local and light traffic Figure 2 1 shows the entire CCD road network using GIS. Figure 2.2 shows the three types of streets with red line representing the arterial roads, blue line representing the collector roads, and green line representing the local roads Table 2 1 illustrates the geographic properties of the CCD road network using GIS 22

PAGE 37

Figure 2.1 GIS Road networks for the entire CCD area N T y ylr--H----t1-+--t '-1 H _1 ] f ( f-Legend Str-eets_ar-c VC>LCLASS --ARTERIAL ---COLLECTOR LOCAL Figure 2.2 Arterial roads are in the red line Collector roads are in the blue line and Local roads are in the green line 23

PAGE 38

Table 2.1 Geograph i c pro perti es for the CCD road network in GIS P r o je cted Coordinate NAD 1983 HARN StatePiane Colorado Centrai FIPS 0502 Feet System Project i on Lambert Conforma l Conic False Easting 3000000 00031608 False Northing 999999 99999600 Central Merid i an 105 50000000 Standard Parallel -1 38.45000000 Standard Parallel 2 39 75000000 Lat i tude Of_ O rig i n 37 83333333 Linear U n it Foot U S Geographic Coordinate GCS N orth Ameri can -1983 HARN System Datum D North Amer i can 1983 HARN Prime Meridian Gree n wich Angula r Uni t Degree I n vent o ry d ata: In ve ntory data i s a c oll ec tion of th e physica l charact eris tic s of the pa ve m e nt and it u s ually doe s not c hang e b e tw ee n maint e nanc e a ctivitie s Th e mo s t b a sic information about th e road i s includ e d to refe r ence the pav e m e nt such as the road name location (refe rencin g sy s t e m) numb e r o f lanes width and length of th e road, and pav e ment type F i g ure 2 3 s hows an ex ample of the in v ento ry data in th e PMS databa se. 24

PAGE 39

Q R FROt.WAME TOIWtiE U3ROAVE EMTHAVE N HOMR Sf N IMMJ S T W25THAVE WJiTHAVE N ClAY ST N DECATUR Sf W33RO AVE W34TH AVE WSCOTIPI. W4STHAVE E STEAVENSON Pl.JNMCHG E 47TH AVE N N..COTI ST N BEACH CT W4STHAVE W47THAVE N ClARKSON ST N EMERSON ST/NMCHG E 44TH AVE E 45TH AVE N HW.OOlllT ST N FRANIQW Sf N ST N PEARL ST CENTIW. ST BOULDER ST N WYMOOT ST N WYANOOT ST N STEELE ST N ADM1S Sf E 48TH AVE/NMCH G E 49TH AVE!NMCHG N MAAION ST N lAFAYETIE ST N DOWNING Sf N MARION ST W37THAVE WmHAVE W 43ROAVE W MTHAVE EllTHAVE E31ST AVE N GARFIELD ST N Rk:HARO N..LEN CT N QUWAS ST N SOOSHONE Sf E 17TH AVE E 18TH AVE N PECOS STIRRX RRX N SHOSHONE ST N TE.KlN ST E liTH AVE E 12TH AVE N ClAY ST N DECATUR Sf BLAKE ST RROVPASS WELTON Sf ST N v.1NOW, CT N WUFF ST PKWY /STR RAMP/STR E2lROAVE E24THAVE 17TH s 18TH Sf W CEDAR AVE W BYERS PL N HUMBOLDT ST N ST W40THAVE W41STAVE E 16TH AVE E 17TH AVE W25THAVE WJiTHAVE E EHAVE E 29TH AVE WOOJAXAVE WCONEJOS PL N ADM1S Sf N COO< Sf ZOO RO/N STEELE Sf N ADM1S Sf N Sf N LOWCLL BL VO E 7TH AVENU E PKWY E 8TH AVE LEN_MI VQClASS 0 .00700 LOCN.. oro LOCN.. 0 llllil CCUECTOR O .IE200 LOCN.. om L OCN.. o m LOCAL omLoCN.. O .IE200 LOCN.. O llllll LOCN.. O(ffi]] LOCN.. OJIJIOO LOCN.. O[ffi]LOC.AJ.. OJ6lll ARTERW. O .!mOO LOCN.. 0.02100 LOCN.. O m LOCN.. 0 .10100 LOCN.. Olffill L OCN.. O lffill LOCN.. om CCUECTOR 0 .11800 L OCN.. O .lliiOO LOC.AI. O llllll CCUECTOR O .IIBIJ LOCN.. O .Ollll LOCN.. 0 00 LOCN.. 0 .10100 LOCN.. O .IE200 ARTERIN.. 0 .15400 ARTERIN.. o .mLOCN.. o m LOCN.. 0 .01400 ARTERIN.. 0.111700 ARTERIN.. O.IBIOO CCUECTOR O .IIBIJ LOCN.. O lffill LOCN.. O llllll LOCN.. 0 .11000 ARTERIN.. om CCUECTOR 0 .00700 LOCN.. 0 .04300 ARTERW. 0 .00100 LOCN.. OWLOCN.. omARTERW. OOOLOCN.. 0 .12lll LOCN.. M lANEWIDTH 2 00000 2 .00000 2 .00000 2 00000 200000 2 .00000 2 .00000 2 .00000 2 .00000 2 00000 2 00000 2 .00000 100000 2 .00000 2 .00000 2 00000 2 .00000 2 00000 2 .00000 2 .00000 2 .00000 2 .00000 2 .00000 2 00000 2 00000 2 00000 2 00000 2 .00000 2 .00000 2 .00000 2 00000 2 .00000 4 00000 2 .00000 2 .00000 2 .00000 2 .00000 200000 100000 2 00000 2 .00000 2 .00000 2 .00000 200000 4.00000 200000 2 00000 ,---JL!Uil.L..>.L ___ -"'-'""""''--------'LU.la'LL.UJL."'--1.1 I ?mJll Figure 2.3 An example of inventory data in the PMS database 25

PAGE 40

Condition data Condition data refers to information about the past and present surface condition of a section of pavement. Accurate historical pavement condition information is absolutely essential for the operation of the pavement management system because all system recommendations are ultimately based on past and present condition data Figure 2.4 shows an example of the pavement condition data in the PMS database 26

PAGE 41

CL i CM I CN I co CP I CQ CR I cs I cr cu I LONG I'IP L LONG I'IP M LONG I'IP H LONG L LONG M LONG H ML IRAAS M TPMS H BLOCK L t 10.[0] o m 0 .[0] ij][(J] lJ.[(J] ilO.[(J] 0[0] mm -1.[0] o m IJJ.[(J] 0 .[0] .J[(J] 1[0] -1.[0] o m s.I()[(J] o .m -1.[0] 18[0] 1JJ[(J] o m -1.00000 .[0] -1.[0] 1f6[(J] mm -1.[0] 6 .[0] 0 .[0] 12[0] O[(J] o m 32[0] mm 0[0] .[0] 0[0] o m o m 32.[0] 0 .[0] 0 .[0] 2.U[(J] o m -1.[0] 0[0] M .[(J] 16[0] 0 .[0] 0[0] -1.[0] .[0] .[0] -1.[0] -1.[0] .[0] .[0] 9111Ill 0 .[0] 0 .[0] 0[0] [(J] 100.[0] -1.[0] U[(J] o m o m 0 .[0] 0[0] 15.[0] 0 .[0] o m .[0] o m o m mm 0[0] 0[0] 100[0] 0[0] 0[0] -1.[0] -1[0] .[0] .[0] .[0] -1.[0] .[0] 0[0] !lli [(J] 1 .[0] o m 0[0] -1.[0] -1.[0] -1.[0] o m 1JJ[(J] o m .[0] o m 1L.[(J] o m o m 1[0] 0 .[0] O[(J] 192[0] O[(J] -1.[0] o m o m ffi.[(J] -1.[0] -1.[0] -1.[0] 0[0] mm mm -1.[0] o m 22.[0] o m o m ffi[(J] 0 .[0] 0 .[0] 100.[0] 1JJ.[(J] .[0] -1.[0] 1 .[0] -1[0] .[0] -1.[0] -1.[0] .J[(J] .[0] .[0] 1 .[0] -1.[0] -1[0] .[0] .[0] 1 .[0] 1 .[0] o m 210.[0] o m .[0] 0[0] 0[0] 0 .[0] o m o m 0[0] mm 0[0] -1[0] 1[0] 81[0] o m .J[(J] -1.[0] -1.[0] 0 .[0] 15[0] 9 .[0] 1 .[0] 313.[0] mm 0[0] 1ffi.[(J] 0[0] o m 1ffi[(J] f8l[(J] lJJ[(J] .[0] o m 111.[0] o m 1 .[0] .J[(J] 1 .[0] o m 191[0] 10.[0] .[0] o m o m 0[0] .[0] -1.[0] -1.[0] 6ll.[(J] 100[0] m m -1.[0] o m o m 316.[0] 0 [(J] 110 [(J] o m 0[0] ffi[(J] o m .[0] 0 .[0] 0 .[0] 611.[0] 0.[0] 0 .[0] 100 [(J] 0[0] 31 [(J] -1[0] o m 100[0] o m -1.[0] -1.[0] .[0] mm mm lJ.[(J] 1 .[0] 1 .[0] -1.[0] -1.[0] 0[0] 93[0] 0[0] o m 100[0] nm -1[0] ffi.[(J] 0[0] 0 .[0] lJ[(J] 0[0] 0 .[0] 100.[0] O[(J] 0[0] .[0] o m o m o.m 00.[0] o m o m 1M. m 1oo. m 1 .[0] o m 110[0] .J[(J] -1.[0] -1[0] ])[(J] lim m m -1[0] 118[0] 0[0] o m o m 6[0] o m o m 3]][0] 0[0] .J[(J] O[(J] O[(J] 0[0] 0 .[0] 100[0] o m 0[0] Lffi[(J] o m -1.[0] .J[(J] -1.[0] -1.[0] 81.[0] !JJ[(J] 0[0] o m 2.ID[(J] o m .[0] 0[0] 1L7[(J] B[(J] 0[0] 5([0] 9t[(J] o m 1mm 1ffi. m 1 .[0] o m mm 21.[0] o m 21.[0] 21.[0] E m 2.uJ[(J] :[0] .[0] 0[0] 1lJO[(J] 0[0] o m mm o m 210.[0] 6]][0] i.m [(J] .[0] I Figure 2.4 An example of pavement condition data in the PMS database 27

PAGE 42

2.2 CCD Pavement Data Management Process The basic procedures required to manage and operate the CCD pavement data include data collection, data synchronization, input and data preparation as shown in Figure 2.5. Data Preparation Data collection Data synchronization and inputting into database Figure 2.5 Pavement data management process in the CCD 28

PAGE 43

Data collection A custom ArcPad application was used for the data collection including map information and customizable forms that allowed the data collection teams to collect the data using the Pocket PC The ArcPad software was linked to the database. The data collected from the field was stored on the secure digital media for later synchronization and inputting into the database at the office as shown in Figure 2 6 lli0 ( 314Jmj.116972a2.i I z fJ AtcPad ll!l 10:11 0 Rut till)-Sevlfity 0 Low Sevrrity (lesl thll 1/'l ():) 1]2' nJ 1 ((pth) 0 ftti Semity ( G re
PAGE 44

Data synchronization Data Synchronization is the process of merging the newly collected data to the existing database The field data was collected and stored on the different secure digital media In order to download the data from multiple data files into the database at the same time the data has to be synchronized first to create one single file which contains all the data from different sources. After synchronization, the data was imported into the database. At this point the data can be queried or used by other data analysis functions. Data preparation Data Preparation on the other hand essentially takes the information output from the database and creates the files that are used on the Pocket PC during the data collection 30

PAGE 45

3. Pavement Inspection and Rating methodology This chapter presents two pavement inspection and rating methodologies used for this study. One is the windshield method and the other one is the standard method The windshield method was developed by the CCD engineers for the purpose of rapid assessment of the road condition. The standard method was developed by Deighton company as a supportive approach to the CCD road maintenance strategy. The detailed description about the two methods is shown below 3.1 Windshield Method The windshield inspection method was developed by the CCD due to the need to quickly quantify the effect of 2006/2007 intensive winter on the pavement network. It is a customized visual rating system using vehicles and six engineers to quickly collect the condition data of the pavement. The condition data then was calculated as the windshield condition index (WCI) The WCI was less detailed than the standard inspections method. Therefore calibration to the standard pavement condition index was completed. 31

PAGE 46

3.1.1 Data Collection for the Windshield Method The windshield inspection about 595 miles (958 km) of arterial and collector roads was completed during a two week period in March 2007 using a shortened list of distresses. A cursory ride through the city revealed that roads that previously had average to moderate distress levels had failed. This was around 6 weeks after the intense storm period and was prompted by knowledge obtained from assessing the twenty-two block arterial roads on 181 h avenue as shown in Figure 1.6 that indicated the roads had distressed greatly. Arterial and collector road segments were driven and four distress criteria were inspected and assigned a l-5 rating which indicated the condition of the roadway with 5 representing the best possible condition and 1 being the worst possible condition. The four criteria includes patches and potholes as shown in Figure 3.1, alligator cracking as shown in Figure 3.2, overall crack development (longitudinal and transverse) as shown in Figure 3.3, and general condition as shown in Figure 3.4 (rutting, roughness, etc.). 32

PAGE 47

Figure 3.1 Potholes and patches damage-Levell(high severity) Figure 3.2 Alligator cracking -Levell(high severity) 33

PAGE 48

Figure 3.3 General cracking-Levell(high severity) Figure 3.4 Overall condition Level 3 (moderate severity) The windshield inspection required some subjective analysis and teams of two were mixed on a daily basis in order to try to avoid bias. Each team would 34

PAGE 49

drive the segment at approximately 25 miles per hour and agree upon a rating for each of the four distress types for each segment. The team recorded their inspection results on a prepared form inspection sheet that had the windshield inspection classification levels Table 3.1 shows an example of the inspection sheet used for the windshield method. Table 3.1 An example of the inspection sheet used for the windshield method Street From To Patches Alligator Cracks General Monaco Pkwy. S Montview 26th Ave 3 I 3 2 Quebec 8th Ave. Colfa x A v e. 3 I 3 2 lOth Ave Federal Blvd Knox Ct. 3 I 2 I lOth Ave. Knox Ct. Perry St. 3 I 2 I 13th Ave Holly St. Monaco 2 2 3 3 13th Ave. Dahlia St. Holly St. 3 2 3 3 Colorado 13th Ave. Blvd. Dahlia St. 3 3 2 3 17th Ave Holly St. Monaco 4 4 3 3 Colorado 17th Ave. 18th Ave. Blvd 2 3 I I 35

PAGE 50

3.1.2 WCI Calculation WCI calculation description The windshield condition index (WCI) was creat e d based on the existing pavement performanc e curve in th e CCD s PMS d e veloped by Deighton (Deighton 2009), a consultant company who provides software solutions for infrastructure asset management as shown i n Figure 3.5 Pavement Perfonnance Curve 120.00.,--------------------------, M 80.00 40.00t----------------=-....... ---....___;o::--,---------1 20.00+--------------------------1 0 .00 0 .00 5 .00 10.00 1 5 .00 20.00 Age 25.00 30.00 35.00 40.00 Figure 3.5 Pavement performance curve used in the CCD road network The pavement performance curve is a function of age and PCI where the older the age of the road, the lower PCI value. The idea was to use this curve to create the equation that could be used to calculate the deduct value for each individual distress In order to create the curve shape as shown in Figure 3 .5, the 5 level rating system for the windshield method was applied to X axis 3 6

PAGE 51

representing the condition levels in the range of 1 to 5, and the Y axis represents the deduct value in the range ofO to 10 as shown in Figure 3.6 The WCI deduct value curve was formulated using a best fit model that approximated the pavement performance curve Therefore, condition level 1 was assigned a deduct value of 10 and condition 5 had a deduct value of 0 as the curve boundaries Then by approximating the trend in the pavement performance curve, condition level 2 was assigned a deduct value of 8 and condition level 4 was assigned a deduct value of 1 and condition level 3 was assigned a deduct value of 4. WCI Deduct Values 10 4 3 2 y = -0.0833x + 1.3333x -6.9167x + 10.667x + 5 ... 8 = .., > 6 ... "' .., 4 ... = "CI ... 2 0 1 2 3 4 5 Distress Score Figure 3 .6 Windshield analysis deduct value curve The equation developed from the deduct value curve is shown as Equation 3 .1. 37

PAGE 52

DVi = Si(-0.833x4+ 1 333x3 -6.917x2+ 10. 667x + 5) (3.1) where : i =distress condition (patches / potholes, alligator cracks general cracks, overall condition) DV = deduct value S = Severity factor (2.5 for patches and alligator cracks, 1.5 for cracks and 1.0 for general) x = distress score (1-5) Each of the four distress categories had varying significance in the overall condition of the pavement. Therefore a severity factor was applied to each of the measured distress categories so as to weigh the distresses appropriately. The severity factor was determined based on the CCD engineer s knowledge and experience on street condition performance in the Denver local area. Patches and potholes were deemed the most serious of distresses while general conditions had smaller weight factors Three Weighting factors were assigned with 1 for general conditions, 1.5 for cracks, and 2 5 for patches/potholes and alligator cracks The WCI for each road segment is calculated by subtracting the total deduct value from each of the four distresses from a starting value of 100 as shown in Equation 3.2 WCI= 100 -EDVi (3. 2) where: i = distress category (patches/potholes, alligator cracks, general cracks, overall condition) DV = deduct value 38

PAGE 53

An Example of WCI calculation 1.) Field data collection Two engineers from the CCD evaluated 101h Ave from Speer Blvd to Broadway using the windshield inspection method. Four distresses were found with pothole / patch in condition level 4 alligator cracks with the condition leve14 general cracks with condition level 3, and the overall condition with condition level 5 2 ) Calculate the deduct value for each distresses using equation 3.I --> Patch/pothole deduct value = Si( -0 833x4+ I 333x3 -6.9I7x2+ I 0.667x +5) = 2.5( -0.833*( 4)4+ 1.333*(4)3 -6 917 *(4i +I0. 667*(4) + 5 = 2 5 --> Alligator deduct value = Si( -0.833x4+ I 333x3 -6.917x2+ I 0.667x+5) = 2.5(-0 833*(4)4+ I 333*(4)3 -6.9I7*(4i + I0.667*(4) + 5 = 2.5 --> General crack deduct value = Si( -0.833x4+ 1.333x3 -6.9I7x2+ I 0.667x+5) = I.5(-0 833*(3)4+ 1 333*(3)3 -6 .9I7*(3i + I 0 667*(3) + 5 = 6 --> Patch/pothole deduct value = Si( -0.833x4+ 1.333x3 -6.9I7x2+ 1 0.667x + 5) = I( -0.833*(5)4+ 1 333*(5)3 -6 9I7*(5)2 +I0.667*(5) + 5 = 0 3.) WCI Calculation using equation 3.2 WCI = IOOE DVi = 100(2 5 + 2.5 + 6 + 0) = 89 39

PAGE 54

3 1.3 WCI Transformatio n t o PCI WCI transformation to PCI description The pavement condition data was stored in the PMS as the standard PCI. Therefore, calibration of WCI to the standard manual PCI was necessary A sampling of the twenty-three arterial road segments inspected using the windshield method was selected for the standard condition analysis By rating select segments by both methods a scatter plot was created to explore the correlation of these two datasets as shown in Figure 3.7 WCI PCI Conversion 100.00 -r-----------------------'-' 80.00 00.00 y = 0 .007r: + 8.1S9:d 3577 R=O.mo '*l.OO 10.00 0 .00 0 1 0 40 JCI 70 so F i gure 3.7 WCIJPCI c on versi on curve 40 90 100

PAGE 55

A second order equation 3 3 was fitted to the data with a correlation coefficient of0.9265 and can be used to convert WCI values to PCI values. The equation 3 3 allowed for a direct quantification of the impacts of the 2006 / 2007 winter season and will be discussed in the chapter 4. PCI = 0 .007(WCii + 0.259(WCI) -2.3577 (3. 3) An Example ofWCI transformation to PCI The WCI value of 89 calculated from the WCI calc ulation example was used for the PCI calculation using equation 3.3 PCI = 0 007(WCI)2 + 0.259(WCI) -2.3577 = 0 007(89)2 + 0.259(89) -2.3577 = 76 3.2 Standard Method The standard manual inspection method was developed by the CCD and Deighton, a consulting company who provides software solutions for infrastructure asset management. This method is mainly used for local and collector streets with a low average daily traffic because the inspection is completed on foot. It is comprised of data collection and PCI calculation The 41

PAGE 56

PCI calculation is used to plan for in-house annual preventative maintenance requirements and to define major rehabilitation projects. The CCD has effectively used the standard assessment method to develop and prioritize successful, high impact projects to rehabilitate its aged roads. 3.2.1 Data Collection for the Standard Method Every five years, 1,573 miles (2,531 km) of local and collector roads are inspected using the standard manual procedure Prior to the 2006 / 2007 winter season, the UCD research team inspected the CCD road networks during a 4 year period from 2002 to 2006. The next inspection was planned to begin in 2010 based on the five years cycle of data collection. Due to the 2006 / 2007 intense snow impact on the CCD road network, the inspection was completed two years in advance. This time the inspection was completed in a ten month period from November 2007 to August 2008. The post-storm inspection began around 10 months after the 2006/2007 storm period and was prompted by the knowledge from the windshield road condition analysis that roads had distressed greatly. The research team manually walked over each individual road segment to record and measure distress types and severity using the hand held computer. Eight distress types were analyzed including transverse crack, In-wheel path longitudinal crack, Non-wheel path longitudinal crack, alligator 42

PAGE 57

crack, pothole patches mtting and roughness and are described below. Three subjective severity levels were introduced as low, medium, and high for each distress. In addition, the approximate extent of each distress (e g surface area of cracks) needed to be estimated The definition used for each distress was referenced by the distress identification manual developed by Federal Highway Administration (FHW A, 2009) 43

PAGE 58

Transverse Crack Description Transverse cracking is cracks that are predominantly perpendicular to pavement centerline as shown in Figure 3 8 It may be caused by shrinkage of the asphalt surface due to low temperatures or hardening of the a s phalt or caused by cracks beneath the surface course. Transverse cracking is measured in linear feet. If the crack does not have the same severity level along its entire length, each portion should be recorded separately. Figure 3.8 Transverse crack in the low severity level Severity Levels Low A transverse crack with a mean width W or less Moderate A transverse with a mean width W' and %". High A transverse crack with a mean width > %". 44

PAGE 59

In-wheel Path Longitudinal Crack Description Longitudinal Cracks are predominantly parallel to the pavement centerline and are contained within the defined wheel paths of traffic as shown in Figure 3.9 They may be caused by a poorly constructed paving lane joint or shrinkage of the asphalt surface due to low temperatures They may also be the result ofhardening of the asphalt or poor substructure In Wheel Path Longitudinal Cracking is measured in linear feet. If the crack does not have the same severity level along its entire length each portion is to be recorded separately Figure 3.9 In-wheel path longitudinal crack in the low severity level Severity Levels Low Any light spalling crack with a mean width \12'' or less. MODERATE Any moderate spalling crack with a mean width \12'' and % HIGH Any severe spalling crack with a mean width>%". 45

PAGE 60

Non-wheel Path Longitudinal Crack Description MonWheel path longitudinal crack is the same as In-wheel path longitudinal crack with the exception that the crack is not contained within the defined wheel paths as shown in Figure 3.1 0. The cause for this type of distress is as the same as the in wheel path longitudinal crack. NonWheel Path longitudinal crack is measured in linear feet. If the crack does not have the same severity level along its entire length, each portion should be recorded separate l y Figure 3.10 Non-wheel path longitudinal crack in the low severity level Severity Levels Low A non-wheel path longitudinal crack with a mean width W' Moderate A non-wheel path longitudinal crack with a mean width Y2" and 14'' High A non-wheel path longitudinal crack with a mean width>%" 46

PAGE 61

Alligator Crack Description Alligator cracking is a series of interconnected cracks in the early stages of development, which eventually develops into many sided, sharpangled pieces characteristically in an alligator pattern in later stages as shown in Figure 3.11. Alligator crack is measured in square feet of surface area. The major difficulty in measuring alligator cracking is that more than one level of severity may exist within one distressed area. If these portions can be easily distinguished from each other, measurement should be recorded separately However, if the different levels of severity cannot be easily divided, the entire area should be rated at the highest, most severe level present. Figure 3.11 Alligator crack in the moderate severity level Severity Levels Low An area of alligator crack which has no or a few interconnecting cracks Moderate An area of alligator crack which has interconnected cracks forming a complete pattern High An area of alligator crack which has pattern cracking progressed so that pieces are well defined and spalled at the edges. 47

PAGE 62

Patch Description A patch is th e portion of pavement surface that ha s been remo v ed and replaced with asphalt filler or additional materials after original construction as shown in Figure 3 .12. A patch is measured in square feet of surface area. If a single patch has areas o f different severity levels these areas are measured and recorded separately. Any other distress found in a patched area will not be recorded ; however its effects on the patch need to be considered when determining the patch severity level. Figure 3.12 Patch in the moderate severity level Severity Levels Low A low severity level patch is in good condition and is performing satisfactorily. Moderate A moderate level patch has moderate severity distress of any type and affects riding quality to some extent. High A high severity level patch is badly deteriorated and affects riding quality significantly 48

PAGE 63

Pothole Description Potholes are bowl -s haped holes of various sizes in the pavement surface as shown in Figure 3.13 They generally have sharp edges and vertical sides near the top of the hole Generally, potholes are the end result of alligator cracking. As alligator cracking becomes seve re, the interconnected cracks create small chunks of pavement, which can be dislodged as vehicles drive over them. Figure 3.13 Pothole in the high severity level Severity Levels Low A pothole is less than 1 deep; Moderate A pothole is between 1" and 2" deep ; High A pothole is greater than 2" deep ; 49

PAGE 64

Rutting Description A rut is a longitudinal surface depression in the wheel path as shown in Figure 3.13. It is usually caused by consolidation or lateral movement of the materials due to traffic loads. Rutting is measured by visual subjective judgment. Precise measurement of the rut depth was not completed in this study. Figure 3.14 Example of Rutting in the high severity level Severity Levels Low A rut in which its mean depth is less than 'h" Moderate A rut in which its mean depth is between 'h" and 1 ". High A rut in which its mean depth is between 1" and 1.5'' Extreme A rut in which its mean depth is greater than 1.5". 50

PAGE 65

Roughness Description Roughness is shown in Figure 3.14 and is defined as the wearing away of the pavement surface caused by the dislodging of aggregate particles and loss of asphalt. Rutting is measured by visual judgment. Figure 3.15 Example of Roughness in the high severity level Severity Levels Low Low severity roughness is pavement where the aggregate has started to wear away Moderate Moderate severity roughness is pavement where the aggregate has worn away. The surface texture is moderately rough High High severity roughness is pavement where the aggregate has worn away. The surface texture is severely rough 51

PAGE 66

3.2.2 PCI Calculation for the Standard Method The raw pavement distress data collected from the field needs to be converted into PCI before it can be used to rank project importance or make decisions about rehabilitation (Aaniewski and Hudson, 1987). As shown in Figure 3 15, the PCI is a function of distress type extent and severity. A six level PCI scale ranging from (0-1 00) is used to describe the pavement condition This scale has a range from very good (91-100), good (76-90), fair (66-75), fair to poor (56-65) poor (41-55) and very poor (0-40) (NCPP, 1998). The PCI of a road segment is automatically computed when the distress data from the field is synched to the main database. The following steps describe the procedure to compute the PCI of a road segment which was referenced by the Deighton configuration report used in the CCD: 52

PAGE 67

Distress Type Distress extent Distress Severity Figure 3.16 PCI factors and PCI scale Determine distress extent: the total of each distress type at each severity level is added up etermine distress density: the total quantity of each distress type at each severity level is divided by the total area of the sample unit and multiplied by 100 to obtain the percent of density for each distress type and severity. Determine deduct values: the deduct value for each distress type (Figure3.17 through 3 24) and each severity level is determined by using the deduct value curve for that particular distress type. The deduct value for patch and pothole at different condition levels were all calculated only 53

PAGE 68

based on a high deduct value curve because these two types of distresses were identified as an indicator that the roads were in severe condition. Figures 3.17, 3.18 3.19, 3.20, 3.21, 3.22, 3 .23 and 3.24 illustrate the deduct values based on the distress density respectively for each individual distress Transverse Crack Deduct Values 100.00 .. -0 _.j .... / -+-L
PAGE 69

In wheel Path Longitudinal Crack Deduct Values 100 .00 --. ----...,. ..... .---+-Low Deduct Oeck.lct 10 .00 ----Mod Deduct Hgh Deduct 100 1 .00 10. 00 100 .00 E xtent(% ) Figure 3.18 In-wheel path longitudinal crack deduct value curve at each severity level Non-Wheel Path Longitudinal Crack Deduct Values 100.00 -. 1111 -.... v r--+-Law [educt Dtcllct 1 0.00 -aMod Deduct Hgh cecJuct v / 1.00 v 1.00 10.00 100.00 E x ttnt (%) Figure 3.19 Non-wheel path longitudinal crack deduct value curve at each severity level 55

PAGE 70

Alligator Crack Deduct Values 100.00 ... --. .. -. .. uiJI ..... ........ LoN Ced.d: Mr.t 10.00 Hq,Ced.d: 1.00 1.00 10.00 100.00 Ext ett (0kj Figure 3.20 Alligator crack deduct value curve at each severity level Patch Deduct Values 1.00 +--+--+-IH-+-H-+t--+--+-+-+++++1 1.00 1000 Extent(%) 100.00 Figure 3.21 Patch deduct value curve 56 High Deduct I

PAGE 71

Pothole Deduct Values 1.00 +---+--+---+-H-H+-+----+--+-+-1-Hf-H-.l 1.00 100 90 80 70 60 lledld 50 40 30 20 10 0 10.00 E x tent{%) 100.00 Figure 3.22 Pothole deduct value curve Rutting Deduct Values I I I / L / / / / ./ N:ne Low ModS"ite Hgh &trEn1! JlihsuD!d Sl!wlity Figure 3.23 Rutting deduct value curve 57 High Ceduct I --+-RuttiJcor Jndn

PAGE 72

Roughness Deduct Values 100 I QO I 80 I 70 / 60 / DeWd 50 R.usJuwss _/ Ind.@x 40 / 30 _/ 20 / 10 / 0 1\b'le Low Modente Hgh &trent! Figure 3.24 Roughness deduct value curve Determine the condition index (CI) for each distress: the condition index for each individual distress as shown in Equation 3.4 is determined by summarizing the deduct value of all severity levels and subtracting the sum value from 100 However, the condition index value for each distress has to be adjusted to 0 if the result value is less than 0. The condition index for rutting and roughness assigned the extreme high severity level a value ofO, the high severity level a value of 40, a moderate severity level with a value of 60 and the low severity level a value of 80. Clm = 100 -E(DVi)m Where: i= each severity level m= distress type 58 (3.4)

PAGE 73

e PCI Once the individual condition indexes are calculated, the PCI can be calculated. The PCI is an average of all individual condition indexes minus one standard deviation to account for the overall condition of the road segment as shown in Figure 3.25. It is important to note as well that the In-wheel path longitudinal crack condition index, Non-wheel path longitudinal crack condition index and the transverse crack condition index are averaged together first prior to being included within the PCI calculation. CI (Transverse crack) C I (Patch& Pothole) CI (Roughness) CI (In-wheel path long crack) AYG I Cl (Non-whee l path long crack C I (Alligator) Cl (Rutting) Figure 3.25 PCI calculation from individual condition indexes 59

PAGE 74

An Example of PCI calculation for the standard method Bryant Street from west 38th Ave to west 39th Ave is presented here as an example of the PCI calculation Bryant Street has the length of 449 feet and the width of30 feet whose area is 13, 470 square feet. 1.) Filed data collection The eight types of distress were quantified at each severity level. The measurement unit used for the field data collection is feet or square feet. 2.) Distress extent for each distress type at each severity level The quantity of each distress at each severity level was summarized as shown in table 3 .2. Rutting and roughness have no quantity value and will be assigned individual distress condition indexes later. Therefore, any quantity calculation process was not applied to these two distresses 60

PAGE 75

Table 3.2 Distres s e x t e n t on Bryant Stre et Alligator LongWP Long Non WP Transverse Patch Pothole (Ff) (Ft) (Ft) (Ft) (Ff) (Ff) Rutt i ng Roughness Low 1000 100 200 50 200 200 L L Moderate 2500 100 200 100 400 400 High 1 000 100 200 1 50 600 600

PAGE 76

3.) Distress density for each distress type at each severity level as shown in table 3.3. The density value for alligator crack, patch, and pothole was calculated based on the percentage of the street area. The density value for in wheel path longitudinal crack, Non-wheel path longitudinal crack and transverse were based on the percentage of the street length 62

PAGE 77

a..., Table 3.3 Distress density for each distress type at each severity level on Bryant Street Long Non Alligator Long WP WP Transverse Patch Pothole Rutting Roughness I Low 7.42% 11. 13% 14 84% 5 57% L L I Moderate 18.56% 11. 13% 14.84% 11. 14% I High 7.42% 11. 13% 14 84% 16 70% 8 91% 8 .91%

PAGE 78

4.) Deduct value calculation for each distress type at each severity level The deduct values are calculated based on the deduct curve for each distress as shown from Figure 3 .17 to Figure 3.24 The table 3.4 shows the deduct value for each distress at each severity level calculated based on the deduct curve. 64

PAGE 79

aVI Table 3.4 Deduct v alu e for e ach distress at each severity leve l on Bryant Street Alligator Long WP Long Non WP Transverse Patch Pothole Rutting Low 5 18 11. 04 2.03 1 16 L Moderate 29 29 3 1 .52 7 89 1 0 09 High 40 46 4 1 .46 20 0 1 28.90 42 38 42.38 Roughness L

PAGE 80

5 ) The condition index for each distress The condition index for each distress is calculated by subtracting the total of deduct values at each severity level from 100. Below is an example of the alligator condition index calculated by the deduct values from Table 3.4 using equation 3.4: The alligator condition index = 100 E(DVi)m = 100 (5.18+29 29+40.16) = 25.07 Table 3.5 shows the condition index value for each distress The condition index for rutting and roughness use the same condition index classification system with the low level having a condition index of 80, the moderate level with the condition index of 60, the high level with the condition index of 40, and the extremely high level with a condition index ofO. In this example, rutting and roughness were rated as L which represents the low severity level having the condition index of 80 Table 3.5 Condition index for each distress on Bryant Street Long Alligator Long Non WP WP Transverse Patch Pothole Rutting Roughness Cl 25 .07 15. 98 70.07 59. 85 57. 62 57.62 80.00 80. 00 66

PAGE 81

6 ) PCI calculation The PCI value is calculated based on each individual distress condition index as shown in the Table 3.5 using the procedure as shown in Figure 3.25. First all, the average condition index ofln-wheel path long crack, Non-wheel path long crack and transverse crack need to be calculated which is called as A VG _1. Then the PCI is an average of all condition indexes minus one standard deviation. Table 3.6 shows the value of PCI calculation Below is the PCI calculation procedure: -> AVG 1= (15.98 + 70 07 + 59 85) /3 = 48.63 -> Average of all index = (25.07+48.63+57.62++57 .62+80+80)/6 = 58.16 -> Standard Deviation of all index = { [(25.07-58.16)2 + (48.63-58.16) 2 + (57.62-58 .16) 2+(57.62-58 16) 2+(8058.16) 2 + (80-58.16f] /6}0 5 = 20.69 -> PCI = Average of all index Standard Deviation of all index = 37.47 67

PAGE 82

0\ 00 Cl Avg _1 PCI Table 3.6 PCI calculation of all the distresses on Bryant Street Alligator Long Non Long WP WP Transverse Patch Pothole Rutting Roughness 25 07 15 98 70. 07 59. 85 57.62 57 62 80. 00 80. 00 25 07 48.63 57 62 57.62 80. 00 80.00 37. 47

PAGE 83

3.2.3 Data Quality Analysis for the Standard Method In order to control the quality of data, a comparison of data collected by different inspectors on a given road segments were analyzed for similarities and differences Twenty-two segments of CCD roads were used for the data quality evaluation as shown in Figure 3 26. = (/) 10 Overland \ Pa1H\est Univenity 1 of Denver E EvnAve Clb&eM1cry CD 0 u 0 C/1 I Park e\e.A!II! '' Figure 3.26 Testing area for the data quality evaluation 69

PAGE 84

Eight teams of two people from the UCD research team were utilized to assess the same target area independently and without knowing the purpose of the task in advance. In addition to the team, two CCD supervisors also assessed the same area Table 3.7 shows the PCI value of the overall condition for twenty-two road segments inspected by the UCD research team and the two supervisors from the CCD T1 to T8 represents the 8 teams from UCD. S I and S2 represent the two supervisors from the CCD. 70

PAGE 85

Table 3 7 PCI value for roads in the test i ng area Road Name T1 T2 T3 T4 T5 T6 T7 T8 S1 S2 SEM ERSO N St 65 5 67 .5 55 8 79 2 72.3 79 7 60 3 66 3 71.9 68 2 I OWA AV1 77.4 71.0 53 2 73 7 75.4 66 9 71.9 73 6 75.6 73.5 SOGDEN ST1 68 .5 66.3 59 1 51. 5 71. 1 74 6 53.4 69 2 64 6 64 1 IOWA AV1 57 2 77.2 59 0 73 9 73.3 75 6 69 9 77.4 62.0 64 2 IOWA AV2 47 8 42.4 9 7 15 2 37 .5 69 .9 12 9 41.0 42 .0 47. 5 SCOR ONA ST1 65 .3 56 8 58 8 57 8 55 .5 55 .9 62 3 67 2 63.0 63 0 IOWA AV3 7 6.4 83 1 69 1 79 3 80 3 85 5 77 .3 85 0 79 2 79 6 SOGDEN ST2 62 .3 62 7 55 7 57 3 6 1 2 59 .9 45 7 69 8 57 2 54. 2 MEXICO AV1 76 .3 77.3 75 6 76.4 80 3 78 7 72.0 83 9 81. 2 78 8 SCORON A ST4 67 2 56 1 57.7 54 8 56 2 52 .5 43 1 68 .3 62 .3 62 5 MEXICO AV1 78 8 73 7 78 .5 79 .5 84.8 84 2 74.6 89 7 90 2 85.0 SEMERSO N ST1 68 .3 66 8 54 6 48.4 62 8 62 0 56 9 74 1 59.4 62 .5 MEXICO AV2 75 1 69.4 66.4 79 7 77.4 90 7 70 3 76 5 77 .5 73 5 MEXICO AV3 76 .9 69 3 67.5 71.3 73.4 82 .5 71.9 76 6 76 6 68 9 COL ORADO AV1 71. 7 65 2 60 5 41. 9 53.4 60 2 54.4 73 9 47.4 57 .9 SCORONA ST2 59 6 55 .9 20 6 21. 1 62 6 54 0 60.8 60 0 53.3 52.6 COL ORADO AV2 60 6 63.4 57 9 62 8 50 1 48 6 58.4 69 9 50 3 50 3 SOGDEN ST2 70 .5 61.9 63 3 79 1 74 1 67 6 61. 6 73 .9 70 8 69.6 COL ORADO AV3 61.5 63.4 51.5 49 .3 62 3 50 6 54 0 60 .9 60 0 65 9 SEMERSON S T 2 73 .5 68 3 65 0 7 5 2 72.8 66 2 62 8 77 5 76 0 76 0 COLORADO AV4 58 .5 55 1 35 2 15 7 32 3 5 1 .5 12 7 60.4 43.0 43 0 SCOR ONA ST3 65.4 57. 6 19.4 21. 2 56 1 68 .3 53.4 5 1 2 62 7 60.6 SOGDEN ST3 67 8 61.0 62 2 73 0 70 .9 67 8 61. 2 77.5 61. 6 61. 6 S EMERSON ST3 66 7 73 2 65 0 54 1 73 9 87. 6 60 2 83.4 70 9 70 .9 A s sho w n in tabl e 3 .8, both CC D s up e rvi s ors concluded that th e a v erage P C I v alue of roads in the t es ting area was around 65. The a ve rage P C I range obtained from th e UC D res e arch t e am wa s from 54 to 70 in which 6 out of 8 7 1

PAGE 86

had average PCI values in the range of 60 to 70. The results of the data comparison indicate a high level of data collection and rating consistency Table 3.8 Comparison of average PCI value for testing roads Inspector PCI AVG Condition Level Teaml 67 Fair Team2 65 Fair to poor Team3 54 Poor Team4 60 Fair to Poor TeamS 66 Fair Team6 70 Fair Team? 56 Fair to Poor TeamS 70 Fair Supervisor 1 65 Fair to Poor Supervisor2 65 Fair to Poor 72

PAGE 87

4. Pavement Condition Analysis For a 10 month period from November, 2007 to August, 2008 after the 2006 / 2007 winter storm, the UCD research team collected a total of 1 573 miles (2,531 km) of road condition data using the standard method. The 1,573 miles (2,531 km) consisted of277 miles (446 km) of neighborhood collector roads and 1,296 miles (2,086 km) of local roads as shown in table 4.1. For a 3 week period in March, 2007 after the storm, the CCD engineers collected 595 miles (958 km) of road condition data using the windshield method. The 595 miles (958 km) consisted of272 miles (438 km) of neighborhood collector roads and 323 miles (520 km) of arterial roads as shown in table 4 .1. In order to complete the pre-storm and post-storm comparison of the road conditions, pre storm data collected from 2002 to 2006 was needed. All pre-storm data was collected using the standard method. 318 miles (512 km) of arterial roads and 277 miles (446km) of collector roads were found in the pre-storm database as shown in Table 4.1. However, for a number of reasons, only 939 miles ( 1,511 km) oflocal roads existed in the pre-storm database with valid condition data as shown in Table 4.1. In order to complete an accurate comparison between the pre-storm and post-storm data using the standard method, the exact 1,216 miles (1,957 km) of the same roads were compared The overall PCI and the individual distress PCI's were studied to compare pre and post storm data collected by the standard method. Then the correlation of 73

PAGE 88

the windshield method data and standard method data was studied. In order to validate the accuracy of the windshield method, the same 180 miles (290 km) of collector roads were compared between the standard method post storm data and the windshield method post storm data Table 4.1 Pre storm and Post storm road condition data using the windshield method and the standard method Pre Storm Data (mile) Post Storm Data ( miles) Arterial Collector Local Arterial Collector Local Windshield 0 0 0 323 272 0 Standard 318 277 939 0 277 1296 4.1 Pre-storm and Post-storm Road Condition Comparison Analysis using the Standard Method 4.1.1 Overall Condition Comparison Using the standard method of analysis, 1,216 miles (1, 957 km) of roads were compared between the pre and post storm database Of these 1,216 miles (1,957 km) of roads, 939 miles (1,511 km) are local roads and 277 miles (446 km) are neighborhood collector roads The average Overall roads condition 74

PAGE 89

was 77 for pre storm, with a PCI range from a low of 0 to a high of 100. After the storm, the average overall condition of the roads was reduced to 70, with PCI ranging from 0 to 100 In other words, the PCI values decrease an average of7 points due to the winter condition of2006/ 2007. Table 4 2 demonstrates the distribution of miles for overall roads using the standard condition PCI data. Comparing the data, the post storm data indicates that there is an increase in the very poor rating of the roads, 6 % poor rating after the storm, from 2 % before storm a 4 % increase Also, there is a 6% increase in post storm poor pavement ratings comparing the post storm and pre storm conditions. The very good rating of pavement condition also goes down to 15% from 26 % in 2003-2005 an 11% decrease in the quality of road conditions. Figure 4.1 shows the pre storm overall roads PCI distribution with the standard analysis and Figure 4 2 shows the post storm overall roads PCI distribution with the standard analysis 75

PAGE 90

Table 4.2 Overall roads PCI comparison using the standard method C ondition Score Standard Overall Roads Condition Analysis Pre-storm Post-storm miles percenta2e miles percenta2e V e r y good 9 1-00 312 7 25 7 % 183.7 15. 1 % Good 76-90 364 7 30.0 % 345 2 28 5 % Fair 66-75 218 2 17.9 % 212.6 17. 5 % Fair to poor 56-65 174 3 14. 3 % 205 2 16. 9 % Poor 41-55 116.9 9 6 % 189 2 15. 6 % Very Poo r 0-40 29 7 2.4 % 77 3 6.4 % D \kry good Gxxl 0 Fair 0 Fai r t o poor Fbor 0 Figure 4.1 Pre storm overall roads PCI distribution using the standard method 18"/c 0\kfy good d 0 Fair 0 Fair to poor Fbor 0 Figure 4.2 Post storm overall roads PCI distribution using the standard method 76

PAGE 91

The overall road condition comparison in terms of the road type was also completed due to the fact that the road deterioration varies depending on the road type Table 4 3 demonstrates the distribution of miles based on the six levels' PCI scale for collector roads. Figure 4.3 shows the pre storm PCI distribution for collector roads while Figure 4.4 shows the post storm PCI distribution for collector roads Similarly, compared to the overall road conditions, the very good and good rating of roads had a substantial drop compared to the pre storm condition data. As well, there was an increase in the fair, fair to poor, poor and very poor conditions post storm. Table 4.4 demonstrates the distribution of miles based on the six levels' PCI scale for local roads Figure 4 .5 shows the pre storm PCI distribution for local roads while Figure 4 6 shows the post storm PCI distribution for local roads. The local roads also exhibited a similar decline in road conditions in the post storm conditions, with a decrease in the very good rating in the post storm condition data 77

PAGE 92

Table 4.3 Overall condition PCI comparison for collector roads using the standard method Condition Score Very goo d 91-00 G o od 76-90 Fair 66 -75 Fair to po or 56 -65 P oo r 41-55 Very P oor 04 0 Standard Overall Condition A nal y sis for Collector Roads Pre-storm miles percentage 70. 0 36.4 % 79.4 41.3% 22.7 11.8 % 11.4 5. 9 % 7 8 4 1 % 1.0 0 5 % miles 33. 2 66 2 40 7 30 2 1 6.9 6 2 Post-storm percentage 1 7 1 % 34 2 % 21.0 % 1 5 6 % 8 8 % 3 2 % OVery good
PAGE 93

Table 4.4 Overall condition PCI comparison for loc al roads using the standard method C ondition S core V e ry good 91-00 Good 76-90 Fair 66-75 Fair to poor 56-65 Poor 41-55 Very Poor 0-40 1 Standard Overall Condition A nalysis for Local Roads Pre-storm miles percentage 242.8 23. 7 % 285.3 27 9 % 195. 5 19. 1 % 162 1 15. 8 % 109 9 10. 7 % 28 6 2 8 % miles 150.5 279 1 171.9 175. 0 172.3 71.0 Post-storm percentage 14.8 % 27.4 % 16.9 % 17. 2 % 16.9 % 7 0 % Very good .Qxxj O Fai r o Fai r t o poor AJar 0 Figure 4.5 Pre storm overall condition PCI distribution for local roads using the standard method 27"/. 17"/. Very good QX>d O Fair o Fair to poor A:x:>r 0 Figure 4.6 Post storm overall condition PCI distribution for local roads using the standard method 79

PAGE 94

4.1.2 Individual Distress Comparison Individual distresses analysis evaluated the 1,216 miles ( 1956 km) including 939 miles (1,511 km) local roads and 277 miles (446 km) neighborhood collector roads between the pre-storm and post-storm database using the standard method. Comparing the data in the tables for the 2006/2007 winter storm season indicates that the alligator crack (Table 4.5) and patch (Table 4.9) dramatically increased due to the severe storm while transverse cracking (Table 4.6) long cracking not in wheel path (Table 4.7) long cracking in wheel path (Table 4 8), roughness (Table 4.1 0), and rutting (Table 4.11) basically stayed at the same level as they were before the storm. Figures 4.7, 4 8, 4 .9, 4.10, 4 11, 4.12 4 13, 4 14, 4 15, 4 16, 4.17, 4 18, 4 .19 and 4.20 illustrate the data respectively for each of the individual condition distresses. 80

PAGE 95

Condition V e ry good Good Fair Fair to poor Poor Very Poor Table 4.5 Condition comparison for Alligator crack Score Alligator crack Prestorm Poststorm miles percentage mile s percentage 9 1 -00 814. 1 66 9 % 684 3 56.4 % 76-90 208 7 17. 2 % 196. 5 16. 2 % 66-75 59 0 4.9 % 103. 6 8.5% 56-65 47 5 3 9 % 77 8 6.4 % 41-55 55. 5 4 6 % 84. 1 6 9 % 0-40 22 6 1.9% 67 0 5.5 % D 'very good G:xxl 1 O Fair O Fair t 0 poor Rxlr D r Figure 4. 7 Pre storm PCI distribution with alligator crack 56o/c good GxxJ 0 Fair 0 Fai r t o poor R:>or 0 Figure 4.8 Post storm PCI distribution with alligator crack 81

PAGE 96

Table 4.6 Condition comparison for Transverse crack Condition Very good Good Fair Fair to p oor Poor Very Poor S core 91-00 76-90 66-75 56-65 41-55 0-40 Transverse crack Pre-storm miles 268.6 457.4 220 2 148.9 102.4 19.0 .. percentage 22. 1 % 37.6% 18. 1 % 12. 2% 8.4% 1 6% .. ,. . Prestorm miles 248.5 457 1 211.9 145. 1 114. 7 22.7 percentage 20.7% 37.1% 17.7% 12.1% 9 6% 1 9% D Very good Gxxl 0 Fair 0 Fai r t o poor Fbor 0 Figure 4.9 Pre storm PCI distribution with transverse crack '..; ,r D Very good Qxxj 0 Fair 0 Fai r t o poor Rx>r 0 Figure 4.10 Post storm PCI distribution with transverse crack 82

PAGE 97

Table 4.7 Condition comparison for Non-wheel path longitudinal crack Condition Score miles Very good 91-00 299.4 Good 76-90 887.3 Fair 66-75 8.7 Fa ir to poor 56-65 0 5 Poor 41-55 20.7 Very Poor 0-40 0 Non-Wheel Path Longitudinal crack Pre-storm miles 24.6% 72.9% 0 7% 0 04 % 1.7% 0 % Pre-storm miles 322.9 86 1 7 8 1 0.5 20.0 0 miles 26.6% 71.0% 0.7% 0.04 % 1.6% 0% \ry good Gxx:l o Fair o Fai r t o poor Rx>r 0 Figure 4.11 Pre storm PCI distribution with Non-wheel path longitudinal crack o 'kry good Gxx:l 0 Fair 0 Fai r t o poor Rx>r 0 Figure 4.12 Post storm PCI distribution with Non-wheel path longitudinal crack 83

PAGE 98

Table 4.8 Condition comparison for In-wheel path longitudinal crack C ondition S core Very good 9 1 -00 Good 76 90 Fair 66-75 Fair to p oor 56-65 P o o r 4 1 -55 Ver y P oo r 0-40 In Wheel Path Longitudinal crack Pre-storm miles percentage 26 7 0 21.9 % 6 00.5 49.4 % 1 62 9 13.4 % 103.4 8.5 % 7 9 0 6 5 % 3 7 0 3 % 49o/c Pre-storm miles 291.6 591.0 155. 1 96 .7 75.3 3 5 percentage 24 0 % 48 .7 % 12.8 % 8 0 % 6 2 % 0 3 % D \kry good GxxJ 0 Fair 0 Fai r t o poor R:x>r D Figure 4.13 Pre storm PCI distribution with Inwheel path longitudinal crack I I. A' .1\ 49o/c \kry good GxxJ o Fair 0 Fair to poor R:x>r 0 Figure 4.14 Post storm PCI distribution with In-wheel path longitudinal crack 84

PAGE 99

C ondition Very goo d Goo d F a ir Fair to poor P oor Ve r y P oo r Table 4.9 Condition comparison for patch S cor e Patch Pre-storm mile s percentage mile s 91100 577.2 47.4 % 393 0 76-90 2 4 8.5 20 .4 % 296 2 66-75 1 96 9 1 6 2 % 238 1 56 65 109.6 9 0 % 1 40 0 4 1 55 71.8 5 9 % 1 1 9 0 0-40 1 2 5 1 0 % 2 7 1 . .. r a r'f ... . : -' .. . 20o/c Post-storm percentag e 32.4 % 24.4 % 1 9.6 % 11.5 % 9 8 % 2 2 % good Gx>d 0 Fair 0 Fai r to poor Fbor 0 Figure 4.15 Pre storm PCI distribution with patch 24o/c good Gx>d o Fair 0 Fai r t o poor Fbor 0 Figure 4.16 Post storm PCI distribution with patch 85

PAGE 100

Condition Very go od Goo d Fair Fai r t o po o r P oo r Very P o or Table 4.10 Condition comparison for roughness S core Roughne ss Pre-storm mile s percentage 9 1 1 0 0 98 6 7 81.1% 76-90 1 0 1 3 8 3 % 66 -75 0 0 % 56-65 95.4 7 8 % 41-55 0 0 % 0-40 33. 2 2.7 % Post-storm mile s 940 7 93.4 0 92 2 0 86 9 percentage 77. 5 % 7.7 % 0 % 7 6 % 0 % 7 2 % D Very good od D Fair D Fai r t o poor Fbor D Figure 4.17 Pre storm PCI distribution with roughness Very good GxxJ D Fair D Fair to poor Ax>r D Figure 4.18 Post storm PCI distribution with roughness 86

PAGE 101

Condition Very good Good Fa ir Fair to poor Poor Very Poor Table 4.11 Condition comparison for rutting Score Rutting Pre-storm miles percentage miles 91-100 1050.3 86.3% 1049.6 76-90 60.4 5.0% 91.1 66-75 0 0 % 0 56-65 86.0 7.1% 49 5 41-55 0 0 % 0 0-40 19.7 1.6 % 23 0 Post-storm percentage 86.5% 7.5% 0% 4.1% 0% 1.9 % D \kry good Qxxj 0 Fair 0 Fai r t o poor R:lor 0 Figure 4.19 Pre storm PCI distribution with rutting 86o/c Very good G:xxl D Fair D Fai r t o poor .Fbor D Figure 4.20 Post storm PCI distribution with rutting 87

PAGE 102

4.2 Correlation Analysis between the Windshield Method Data and Standard Method Data using an Identical Road Dataset 4.2.1 The Procedure of Identifying the Same Roads from the Windshield Method Post-storm Data and the Standard Method Post-storm Data In order to determine the correlating data from the standard method and the windshield method GIS had to be implemented to find corresponding streets in which both methods had been used The standard method was utilized for most of the Denver streets, including local arterial, and collector streets. The windshield method was only performed on the arterial and collector streets after the 2006/2007 s now storm. Therefore, the only matching streets from both sets of data had to be found in the collector streets dataset. Unfortunately the entirety of all the collector streets was not analyzed by the standard method after the 2006/2007 snow storm. Likewise, only some of the collector streets were analyzed using the windshield method. As a result the two sets of data, even looking at only the collector streets, were not identical, some collector streets were done in the windshield method some were done in the standard method, and some in both 88

PAGE 103

The challenge was to find those that were truly done using both methods. To complicate the identification of streets that were completed in both methods the segment lengths of analysis done in each method were different. The street segments in the standard method data were based on individual block by block segments that corresponded to a local sector ID or super segment ID. The street segments in the windshield method data were formed by longer street segments that also corresponded to a local sector ID or super segment ID that were composed of multiple block segments. The CCD defined 272 local sectors and 532 super segments. The local sectors are made up of multiple block road segments sharing the same geographic area boundary by the arterial and collector road segments that form super segments. Figure 4.21 displays local sectors using different shades while the super segments form a boundary between local sectors and are shown in red with their termini marked with an X (Deighton,2009). Table 4.12 shows a spreadsheet example of super segments with their corresponding block street segments i ............ I I ......... ..,. ...... .. ... ... l --! I I I ,. --- I I 1 -""f'L j i W-T"-I I I i f \ I .... ... ... --: --'I' i I . -=1 .. -! -I --Figure 4.21 Local sect o r and Super segment definition ( Deighton, 2009) 89

PAGE 104

Table 4.12 An example of super road segments and their corresponding block street segments Super Segment From To Block street From To segment ID segment N Dillion St NE085 E 47th E Boiling Dr N Dillon St E 47th Ave E ELK PI Ave N Dillion St E ELK PI E 48th Ave N Deeph N Dillion St E 48th Ave Ave N Deeph N Dillion St Ave E Boiling Dr E 55th Ave NE086 N Peoria N Scranton E 55th Ave N Peoria St N Quari St St St N Quentin E 55th Ave N Quari St St N Quentin E 55th Ave St N Racine St N Revere E 55th Ave N Racine St St N Revere E 55th Ave St N Salem St N Scranton E 55th Ave N Salem St St After analyzing the data structure from the two datasets it was very clear that the street name and local sector ID or super segment ID for road segments would play a key role in linking the two datasets The strategy to find the identical streets performed by both methods first involved the following datasets : l) Standard Method All PCI data for the entire CCD road networks available in shapefile format 90

PAGE 105

2) Windshield Method PCI complete data available only in spreadsheet format Initially, the collector streets assessed by the standard method after the 2006/2007 snow storm were selected from the Standard Method All PCI data shapefile. A new shapefile called Standard Method Collector Streets was created which represented a group of collector streets that could have potentially been covered by both the standard method and by the windshield method as well. Next, grouping the thousands of block by block street segments into longer street segments based on the street name and local sector ID or super segment ID was completed using the dissolve function in GIS. It classified 2,840 records down into 389 records that all had the same matching street name and local sector ID or super segment ID. The weighted Mean PCI value for each common group of segments was also calculated automatically from using the dissolve function. With this completed list, a complete list of streets had been created that could be used to compare the data of long street segments in the windshield methods master list. Last, with the complete list of grouped collector street segments now available, the work began manually with the master windshield list to actually find out from the potential list, which of the collector streets had been analyzed from 91

PAGE 106

the potential windshield dataset. Manually, the same collector streets then finally were identified from both methods of data With the full list of collector streets that were completed in both the standard method and the windshield method, the Standard method Collector Grouped Segments shapefile was then edited to exclude any segments that were not found in the windshield data master spreadsheet. As well, the PCI values from the windshield method segments that corresponded with the standard method segments also were added to a new field in the standard methods segments shapefile called WCI PCI that was then populated by the corresponding PCI value from the windshield method. Overall, 180 miles (290 km) of roads were able to fall into this exact level of spatially corresponded referencing that allowed for a true comparison of data collected by these two methods. The flowchart as shown in Figure 4 .18 demonstrates the procedure of identifying the 180 miles (290 km) of same collector roads from the windshield method data and standard method data 92

PAGE 107

Ready to see the correlatiOn between two methods of data Standard Method CoiiPctor Roads Ld< ,,,, C oiiPctor Po,Jd 'J' Figure 4.22 the Procedure of identifying 180 miles of same collector roads from the windshield method data and the standard method data 93

PAGE 108

4.2.2 The Correlation Analysis of the Same Roads between the Windshield Method and the Standard Method In order to validate the accuracy of the windshield data, the difference betwe en the weighted mean standard PCI value and the windshield converted P CI value was calculated for each corresponding street segment which can be used to compare the difference of P CI values between these two methods As shown in Table 4.13, the WCI PCI column has been subtracted from the Weighted Mean PCI column resulting the Diff_ PCI column which is the difference between the weighted mean standard PCI value and the windshield converted PC I. Table 4.13 An example of difference of Weighted_Mean_PCI values of standard method segments and WCI_PCI values of windshield method segments Street Name Weighted Mean PCI WCI PCI Diff PCI Length E 11TH AVENE024 82. 6 82. 0 0 6 0.7 E 11TH AVENE032 79. 5 33 0 46 5 0 1 E 11TH AVENW012 95 0 79. 0 16 0 0.1 E 12TH AVENE024 61. 8 72.0 1 0 0 8 E 12TH AVENE025 71. 6 55 0 16.6 0 5 E 12TH AVENE026 71.3 86.0 -1.5 0.5 E 12TH AVENE027 75. 3 75.0 0.3 0 3 E 12TH AVENW012 97 0 52.0 45.0 0 1 E 1ST AVENE001 67.7 55.0 12.7 0.7 E1ST AVENE005 56.3 61.0 -4.7 0 5 E 29TH AVENE051 97.8 89.0 8.8 0 5 E29TH AVENE052 72 8 85 0 -1.2 0.4 94

PAGE 109

Figure 4.23 demonstrates the distribution of the differences between the PCI values for the windshield method and the standard methods based on corresponding street segments. The X axis displays the values from the Diff_PCI column as shown in Table 4.13. TheY axis represents the overall percentage of roads that fall into each range of the PCI difference values in the Diff PCI column as shown in Table 4 .13. The standard deviation of the PCI differences is 17.92 the mean value of the PCI difference is 1.97 and the median value of the PCI difference is -0 99 and the range ofthe PCI difference is from 52.69 to 57. PCI Difference between Standard PCI and Windshield Converted PCI 14 12 1.17 -17.92 10 lledion -G. n i u 8 l 57.00 6 4 2 -45 -30 -15 0 15 30 45 -ofPCI_IIIe ___ llleWI-Ielll Figure 4.23 Distribution of PCI difference between the standard method PCI data and the windshield method PCI data 95

PAGE 110

The breakdown of the histogram bins is as follows beginning with the bin directly above 0 on the X axis near the median of -0. 99 on the histogram as shown in Figure 4.23 The percentage of road segments of the 389 total roads that fall in that bin is the first number, followed by the range ofPCI differences for that bin: {>Bin above zero on x axis, 12.40 % (-2 .5 2 5) {>To the Right of the median bin values are as follows: 8 97% (2.5, 7.5), 8.44% (7.5, 12.5), 6 60% (12 .5, 17. 5), 4 75% (17.5, 22.5) {>To the Left of the median bin values are as follows : 14.78 % (-7.5, -2.5), 11.08 % (-12 .5, -7.5), 9.50 % (-17 5 -12 5) 4.75% (-22.5, -17.5) This distribution of differences in PCI values indicates that approximately 46% of the data is within 10 points difference between the two assessment methods and around 65% of the data is approximately 15 points difference or less between the two assessment methods The distribution of the PCI value differences indicates that the windshield assessment method, while not perfect, bears out as a fairly strong evaluation method in comparison to the standard method of pavement assessment, each with its own merits, as will be discussed in the conclusion in the next chapter. 96

PAGE 111

5. Conclusion and Future Recommendation 5.1 Conclusion As set out in Chapter 1 there were four goals to be accomplished in this study. The first of these major goals was the presentation of the procedures and details of the windshield condition assessment method and the standard manual pavement assessment method. The detailed description of these two methodologies was presented in Chapter 3. The next major goal was the inspection of around 600 miles (966 km) of the ceo arterial and collector roads using the windshield method and the inspection of around 1600 miles (2,575 km) of the ceo local streets and collector streets using the standard method. Through March, 2007 around 600 miles (966 km) of the ceo arterial and collector roads were inspected and evaluated using the windshield method. It took three weeks at an estimated internal cost on the order of $20,000 From November of 2007 to August of 2008 1 573 miles (2,531 km) of the ceo local and collector roads were inspected and assessed by the research team from UCO using the standard method. The cost of obtaining this data in 2008 dollars was approximately $120 per mile which is approximately $190,000 total cost. 9 7

PAGE 112

The third major goal was the detailed study of the effect of the 2006 / 2007 winter snow storm on the pavement condition Overall condition comparison and individual distress comparisons between pre storm and post storm data using the standard method was presented in Chapter 4. Comparing the pre storm and post-storm data collected by the standard method, the analysis results indicated an average decline of7 points ofPCI in the CCD road network. In the normal winter, the road PCI value is expected to have a 2 to 3 point drop due to natural deterioration of the road. The significant decline in PCI points for the 2006 / 2007 winter snow storm indicated that the storm had a severe impact on the CCD road network. The analysis result enabled the CCD street maintenance department to prioritize pavement maintenance and secure an additional $20 million dollars in supplemental budget allocations that will be used to bring those streets back to a good condition. The last goal was the exploration of the correlation between the windshield method data and the standard method data which is the key study that the author tried to explore The detailed correlation analysis between the two methods' dataset was demonstrated in Chapter 4 Comparing the 180 miles (290 km) of same collector roads assessed by both the windshield method and the standard method, the results indicate that approximately 46% of the data is within 10 PCI points difference between the two assessment methods around 65% of the data is approximately within 15 points difference or less between the two assessment methods and around 80% of the data is within 20 PCI 98

PAGE 113

points difference between the two assessment methods. The distribution of the PCI value differences indicates that the windshield assessment method performs fairly strongly as an evaluation method in comparison to the standard method of pavement assessment. Overall the windshield analysis can be used as a rapid assessment method of the overall road conditions. It is not a detail oriented pavement condition assessment method but more of a rough general assessment method for the pavement overall condition Under the unexpected circumstances such as unusual weather, it is an efficient way to obtain a feeling about the overall condition of the roads. The windshield assessment method enabled engineers to compute a fair estimate of the accelerated damage that occurred due to the unusual winter weather. The windshield method was derived from solid engineering principals the study of the dynamics of pavement deterioration and was designed to complement the CCD existing pavement management system. Therefore it can be applied to various extreme and unexpected conditions not only limited to winter environments. 99

PAGE 114

5.2 Discussion and Recommendation for the Future Work The windshield method analysis in this study gave a strong assessment of the pavement conditions for the CCD collector streets, and the study indicates that there is a strong correlation between the windshield and standard methodologies. Future recommendations would be the inspection of the local roads in addition to the collector roads using the windshield method and compare it with the data collected by the standard method. If it shows a strong correlation between those two datasets, then the effectiveness of the windshield method would be further borne out. There should be several studies conducted in order to fully verify the effectiveness of the windshield method before any evolution of the pavement condition assessment in the CCD is to begin Overall, this study has been greatly enlightening and insightful in the author s overall research work involved the pavement assessment project for the CCD. Initially, the author was involved in the data field data collection for the pavement conditions in this project and became so intertwined in the work that the author's role developed into manager of the pavement assessment field data collection team for the CCD. The opportunity evolved into work that involved QA/QC of a vast inventory of data, incorporation of GIS analysis to assess the overall project work, and finally statistical verification of the work conducted. The author s work in this project was helpful in creating a 100

PAGE 115

substantive base from which the PMS will evolve into higher levels of prediction, accuracy, and faster turnover time from the assessment of conditions, to the management of the data, and verification of quality standards, that translates into greater efficiency in assessment and overall faster response times for engineering planning that maintains the integrity of a vast road network. 101

PAGE 116

Bibliography Aaniewski, John P., Hudson, Stuart W., and Hudson, Ronald W. (1987). Pavement Condition Data Anaysis. J. Transp. Engrg. ASCE. 113(8), 413 421. ABC Channel 7 news, http://www .thedenverchannel.com/weather/14750017 /detail.htm l Berry, Joseph K(2006). GIS Technology in Environmental Management: a Brief History, Trends and Probable Future, GeoPiace http://www .geoplace.com!ME2/Default.asp Cadkin, Jennifer (2002). Understanding Dynamic Segmentation -Working With Events in ArcGIS 8.2. ESRI, http://www .esri.com/news/arcuser/1 002/files/dynseg 2.pdf Carey, W.N and Irick. P.E ( 1960), The Pavement Serviceability Performance Concept, Highway Research Bulletin 250, Highway Research Board, Washington, D.C., 40-58 Deighton (1998). http://www.deighton.com/ Deighton (2009), Configuration Report. Street Maintenance Department in the City and County of Denver FHW A(2009). Distress Identification Mannual for the Long-Term Pavement Performance Program(Fourth Revised Edition. Federal Highway Administration(FHW A). http://www.tfhrc.gov/pavement/ltpp/reports/03031/index.htm Finn, Fred. (1998). Pavement Management Systems-Past,Present, and Future. Public Roads,Vol62 No1 GIS definition(2006), http://support.esri.com/index.cfm?fa=knowledgebase.gisDictionar y.search&search=true&searchTerm=GIS Headquarters, U.S. Army Corps of Engineers. (1989). Procedures for U.S. Army and U.S. Air Force Airfield Pavement Condition Surveys, TM 5-826-6/AFR 93-5, Washington, D.C. 102

PAGE 117

Kulkarni, R B., and MiUer, R W. (2003). Pavement Management Systems: Past, Present, and Future. Transportation Research Record No. 1853 Transportation Research Board,Washington, D.C., 65-71 MeN erney, Michael T. (2008). Remaining Service Life Analysis of Concrete Airfield Pavements at Denver International Airport Using the FACS Method. Proceedings of the 2008 Airfield and Highway Pavements Conference, ASCE. Muntasir, A. (2006). Summary and Comparison of Airfield Pavement Inventory, Condition and Performance Across The United States. Airfield and Highway Pavements 191,82. Pavement Condition Rating Procedure (1998). The National Center for Pavement Preservation (NCPP). http://www.pavementpreservation.org/publications/PCR %20MAN UAL/3pcrproc.pdf Temperature Data, National Oceanic And Atmospheric Administration (NOAA) http://www.crh.noaa.gov/bou/?n=climo 103