|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 41
CHAPTER 4. RESULTS OF SMOOTHNESS DATA ANALYSIS
Introduction
This chapter presents the results of several analyses conducted on the effect of
initial pavement smoothness and on the effect of smoothness specifications. The first
part of the chapter summarizes the analyses of the effect of initial pavement
smoothness on the future smoothness and life of the pavement. This is followed by a
section discussing the effect of smoothness specifications on ~rutial pavement
smoothness. The final part of the chapter describes a procedure for evaluating the
cost effectiveness of initial smoothness levels and presents the results of an evaluation
of the cost effectiveness of smoothness specifications.
As discussed In chapter 3, the data for these analyses were assembled from
several different sources, including existing data bases, literature surveys, and SHAs.
The data from SHAs are believed to be the most representative and form the
cornerstone for most of the analyses that were conducted.
Analysis of the Effect of Initial Smoothness on Future Pavement Smoothness
Introduction
Very lithe research has been conducted on the effect of Crucial smoothness on the
future smoothness of the pavement. Although He design equations developed from
the AASHO Road Test (AASHO 1962) imply that initially smoother pavements
maintain higher levels of rifle quality, little information is available outside of that
work. However, one study reported by Janoff (1991) analyzed historical roughness
data spanning 10 years and representing about 400 sections of roadway from Arizona
and Pennsylvania. It was found that nutial smoothness is related to long-term
roughness, as illustrated in figure 16 danoff 1991~. The smoothness values shown in
figure 16 are in terms of the Mays Meter roughness Index, and it is observed that the
range of values are rather small, indicating that most of the pavements used in the
evaluation had not appreciably deteriorated. lanoff (1991) also determined that the
annual maintenance costs were lower for those sections constructed with a higher
initial smoothness, as shown in figure 17. While these findings are indications of He
beneficial effects of initial pavement smoothness on the ride quality of He pavement,
they are based on Innited data over a Ignited range of roughness values.
Under this project, pavement projects from a wide range of data sources were
used to further evaluate the effect of initial pavement smoothness on the future
smoothness of the pavement. The projects used in the analyses include a myriad of
pavement designs, subject to various climatic and traffic factors, and characterized by
different roughness indices. A description of the data sources used in the analysis is
provided In the following section.
41
OCR for page 42
40
JO
20
10
o
Long-Term Roughness (in/mi)
-
.
-
1 . ' ' _
-
-
-
-
-
-
-
-
0 10 20 30 40
Initial Pavement Smoothness (in/mi)
Figure 16. Filial pavement smoothness versus long-term pavement roughness
(Mays Meter roughness index values) Janoff 1991).
1500
1000
500
lo
Average Annual Maintenance Costs
($/lane mile)
/
-
-
, .. . .
-
I . . . I . . , 1
10
20
Initial Pavement Smoothness finimi)
30
40
Figure 17. Initial pavement smoothness versus average annual maintenance
costs Janoff 1991~.
42
1
OCR for page 43
Overview of Data Sources
Pavement projects from several of the data sources presented in chapter 3 were
used to evaluate the effect of initial pavement smoothness on future smoothness.
However, the vast majority of the projects analyzed toward this objective were those
contributed by SHAs. Roughness data from the Alabama Pavement Roughness
Study were also included with the SHA data for use as the centerpiece of the
analysis.
Roughness data from the various road tests and pavement performance studies
were generally much less revealing than the State-furIiished data. This was largely
attributed to the time period in which the studies were conducted (1950s and 1960s
for most of the road tests), the inadequacy of the experimental designs (factors other
than roughness [e.g., pavement design] were usually the main focus), and the lack of
standard, reliable smoothness-measur~ng equipment and procedures used throughout
the study. Consequently, outside of the SHA data, only data from the AASHO Road
Test and the LTPP program were deemed worthy of detailed analysis and reporting.
The evaluation results from these two sources are presented as a supplement to the
State data evaluation results. Figure iS depicts the States contributing data used in
the analysis of the effect of initial smoothness on future smoothness.
In performing the analyses of roughness or ride quality over time, it was
considered important that each pavement project consist of two or more adjacent
"replicate" sections along a highway, as illustrated in figure 19. In this way, the
effects of major factors, such as traffic and climate, are eliminated, and direct
comparisons could be macle between sections with significantly different initial
smoothness values.
Table 9 provides a summary of the data sources used In the analysis, and
indicates the number of replicate sections within each project. At He AASHO Road
Test, for instance, a specific pavement design, subjected to a standard load (e.~.
6,000-Ib [26.7-kN] single-axIe truck load), was duplicated to form two replicate
sections. Longer pavement projects, however, such as those from the States, provided
numerous replicate sections, as they were broken down into short, uniform segments
(typically 0.1 or ~ mi [0.16 or I.6 km]) defined by the available roughness testing
interval.
The preliminary analysis consisted of preparing time-series (or, in some instances
load-series) roughness trends for each of the pavement sections from each data
source. This was done by plotting the series of initial and extended roughness
measurements for each replicate section within a project. The number of time-series
data points varied by data source, as some sources provided sets of routine
roughness measurements (e.g., Arizona, Kentucky), whereas others only provided
two sets of roughness measurements (e.g., Iowa, RPPR).
43
OCR for page 44
Description of Data Bases
Agency Survey Data Bases
Work on the agency survey data bases was initiated with the establishment of the
NCHRP 20-7 task 53 State survey data base. The tables In this data base were
structured such that States are listec! In the first column and the survey questions are
listed, in order, in the column headings to the right. The State responses were then
manually entered into the appropriate cells.
Once He NCHRP I-31 State survey form was completed, the data base for this
survey was fully structured using the same format as the NCHRP 20-7 survey data
base. As State responses became available, the results were again manually entered
into the appropriate cells.
As part of the agency surveys, selected paving contractors were contacted in order
to obtain the contractor's perspective on pavement roughness measuring procedures
and practices. Both AC and PCC paving contractors were contacted. The same
procedure for the clevelopment of the agency survey data base was followed for the
contractor data base.
Project Analysis Data Bases
Work on the project analysis ciata bases began with the development of a list of
anticipated data base elements. This list, shown in table 7, was prepared following
the Strategic Highway Research Program (SHRP) LTPP Data Collection Guide (FHWA
1993), and contains the necessary data elements for conducting later analyses to
address the following key project objectives:
i. Determine the effect of initial smoothness on the ride quality of the pavement
over its life and on pavement service life.
2. Determine the effect of existing smoothness specifications on He initial as-
constructed smoothness.
3. Determine He cost-effectiveness of smoothness specifications, including
incentives and disincentives.
Sources of Data for Project Analysis Data Base
Past Road Tests
AASHO Road Test
The AASHO Road Test represents one of the most comprehensive studies of
pavement performance that has ever been conducted. Over 800 AC and PCC
pavement study sections were constructed and evaluated from 1958 to 1960.
30
OCR for page 45
Paving Project
(Same design, traffic, subgrade, etc.)
. . .
Sect. Sect. 2 . Sect. 3 . Sect. 4
. ,
1 , . .
Initial
Smoothness S1 S2 s3 s4
Time-Series T T T T
Smoothness 1 2 3 4
Figure 19. Illustration of Replicate sections along a highway pavement project.
As seen in table 9, some pavement projects were tested crucially win one
particular roughness device and then tested at a later date with a different device.
For those projects in which strong correlations existed between the two testing
devices, conversions were made so that a single roughness index could be used to
better reflect Me roughness trends. Specific examples of this occurrence include Me
three {owe PCC projects (California profilograph used initially for construction
acceptance, South Dakota-type profiler used later for pavement management
purposes), and the five Minnesota PCC projects (GM-type profilometer used initially
for construction acceptance, South Dakota-type profiler used later for pavement
management).
To some extent, Me ~rutial smoothness data collected represented tests performed
within the first few weeks of construction. With several sources, however, such
Initial data were not available, either because the construction records containing the
test results had been discarded or because the ~rutial pavement smoothness was not
being monitored by the SHA at that time. In those instances, pavement management
smoothness measurements made within 2 to 3 years of construction were used as
Initial smoothness values. Projects that fit this category include those from Illinois,
Michigan, and Washington, as well as Minnesota AC and AC overlay projects.
45
OCR for page 46
Table 9. Summary of data used to analyze the effect of ~rutial pavement smoothness
on future smoothness.
Number of | Basic Level of I
Data Source
AASHO Road
Test
WASHO Road
Test
San Diego Test
Road _
Minnesota _
Investigation 183
Alabama
Roughness Study
RPPR
LTPP GPS
Arizona
Georgia
Illinois
Iowa
Kentucky
Michigan
Minnesota
South Dakota
Washington
Wisconsin
1 ft = 0.305 m
a For AASHO Road Test, WASHO Road Test, San Diego Road Test, Minnesota Investigation 183, RPPR,
and LTPP GPS, a project is equated to a specific pavement design.
Full-depth AC.
c Time lapse between construction and initial measurement greater than 1 year but less than 3 years.
_ Roughness Measuring Device(s)
Pavement Replication
a
Projects Section No. of Pavement
Examined Length, ft Sections Types Included Initial Future
32 flexible 1 1 on to '4n 1 2
36 rigid
10 flexible
120 to 240
300
_
4 flexible 250
5 flexible
16 flexible
26 rigid
18 rigid
2 flexible
1 composite
3 rigid
5 rigid
3 composite
12 flexible
3 rigid
6 composite
9 flexible
37 rigid
4 composite
2 rigid
11 composite
7 flexible
6 rigid
9 flexible
5 rigid
8 flexible
AC, JPC, JRC
AC
FDACD
700 or
1200
2,640
500 to
1,000
500
6 to 8
2 to 6
3 to 18
2 to 6
2 to 8
5250 5 to 14
5,280
528
- 5 _
_
5,280 5 to 10
~_
6tol3 _
5250 3 to l0
5230 5 to 24 _
5,280 4 to 9
AC
AC, PCC
JPC, JRC, CRC
AC, FDAC,
JPC, CRC
PSI
(Chloe profiler)
BPR Roughometer
PSI
(Chloe profiler)
PSI
(BPR Roughometer)
. BPR Roughometer
Various
KJ Law
Pro fit o me terC
PSI
(Chloe profiler)
BPR Roughometer
PSI
(Chloe profiler)
PSI
(BPR
Roughometer)
BPR Roughometer
Soup Dakota
type Profiler
KJ Law
Profilometer
JPC, AC,
AC/JPC,
AC/AC
JPC, AC,
AC/JPC,
AC/AC
CRC,
AC/CRC
AC/JPC
JPC
JPC,
AC/JPC
JRC, AC
AC,
_ I I I PCC-doweled
13 rigid 528 3 to 11 PCC
PCC, AC,
AC/AC
JPC, JRC, AC
CRC, AC/AC,
AC/PCC
6 rigid
12 flexible
25 rigid
16 composite
30 flexible
_ 5,280 3 to 9
e S;ttlO (typ) 3 to 23
Mays Meter
Mays Meter
BPR Roughometer
California
Profilograph
PI
(Mays Meter)
GM ProfilometerC
GM Profilometer
(PCC)
South Dakota-type
Profiler (AC)C
California
Profilograph
.
PCA Roadmeter
PSI
(Mays Meter)
Mays Meter
Mays Meter
South Dakota
type Profiler
South Dakota
type Profiler
RI
(Mays Meter)
GM Profilometer
South Dakota
type Profiler
California
Profilograph
PCA Roadmeter
PSI
(Mays Meter)
46
OCR for page 47
Some data sources, such as the AASHO Road Test, Kentucky, and Wisconsin,
furnished serviceability-type data. These data took the form of serviceability or
rideability indices, derived mostly from roughness measurements. The resulting
time-series trends are a reciprocal of the roughness trends of other data sources, as
lower index values represent rougher surfaces.
Provided in appendix C are pavement project summary tables and time-series
roughness plots of each of the sections evaluated in this study. The tables and plots
are arranged according to data source. The summary tables include useful
information about each pavement project, including project identification, location,
design, construction year, evaluation period, and traffic estimates. The plots
represent the roughness trends of the various fixed-interval sections that comprise a
given pavement project.
It should be noted Hat not all of the projects for which data were collected are
included In the summary tables and roughness plots. Several projects were identified
as having extended smoothness measurements, but no initial measurements (within 3
years of construction). This was the case for the majority of the Michigan projects,
where only 15 of the 47 subject sections could be analyzed because of the lack of
Initial measurements. Another reason for the exclusion of some projects was the
inability to closely align (by milepost, reference point, or station) ~rutial smoothness
measurements with extended smoothness measurements. This problem was
encountered with some of the Minnesota and Washington projects.
Overall Evaluation of Effect of Initial Smoothness Using SHA Data
This section describes the results of the overall evaluation of He SHA data. The
data sources described in the previous section are analyzed over He entire period for
which smoothness data are available to determine whether He ~rutial smoothness of
the pavement influences the future smoothness of He pavement.
Evaluation Approach
An examination of the t~me-series smoothness plots (contained In appendix C)
indicate the following:
1. Pavement sections built smoother generally remain smoother over time (all
over Wings being equal).
2. The performance curves of two pavements constructed to different initial
smoothness levels but of otherwise similar design roughly "parallel" each
other, with the rougher section perhaps deteriorating more rapidly due to
greater dynamic loadings and/or greater variability in construction.
An example chart illustrating the "parallel curve" concept is shown in figure 20.
Sections I, 2, and 3 all roughly parallel one another, with section 3, constructed
47
OCR for page 48
In
a)
s
o
Section ~ ~
-
Section 2
Time or Traffic
Figure 20. Example illustration of pavement performance curves for sections along a
given construction project.
Initially smoother, remairung smoother over time (or traffic). Section I, constructed
rougher than Me other two sections, shows perhaps a slightly greater rate of
deterioration.
As discussed previously, pavement projects were solicited from SHAs containing
multiple sections constructed to different Crucial smoothness values. Because the
pavement design, cross section, subgracle support, traffic loadings, age, and climatic
forces are approx~nately We same for each of the sections within Me project, the
effect of these variables on the performance of all of Me sections is assumed to be
constant and Me effect of nutial pavement smoothness can therefore be isolated.
For each project containing multiple sections, a statistical regression analysis was
conducted relating the smoothness at any time to Me initial smoothness and to the
age, as expressed in the following general form:
St = aO + a~Si + a2t
where:
SO = Pavement smoothness at time t.
aO, al, a2 = Regression coefficients (at is Me regression coefficient on Si).
Si = Initial pavement smoothness.
t = Verne (age), years since construction or overlay to time of smoothness
testing So.
48
OCR for page 49
For the purposes of this evaluation, a multiple linear regression analysis was
conducted based on observations of the trends shown in the plots. A linear mode]
was selected in order to provide an Indication of the significance of the initial
smoothness variable. In many cases, nonlinear regression provided a better "fit" of
the data, and these improved moclels were used in the analysis on the effect of initial
pavement smoothness on pavement life (described later).
The output of the linear regression provides the regression coefficients and
information on the statistical significance of the independent variables (in this case,
initial smoothness and age) on the dependent variable (smoothness at time t). The
regression coefficient al is the slope of the So versus Si regression line. If this constant
is approximately 1.0, this indicates that a strong one-to-one relation exists between
the initial smoothness and the future smoothness. This means that if one section is,
say, 10 in/ml (0.16 m/km) smoother at initial construction Man another, then the
future smoothness of that section will remain 10 Mimi (0.16 m/km) smoother than
the other section over time. If the regression coefficient al is zero, this indicates that
future smoothness is not dependent on initial smoothness. This concept is illustrated
In figure 21.
-
T
._
_,
Ct
to
In
s
o
~ alma /- ,/
al= 1
al=0
, . ~.
_ _ _ ~
1 , 1 , 1 ,
Initial Roughness
Figure 21. Effect of a, regression coefficient In the relationship between Crucial
smoothness and future smoothness.
49
OCR for page 50
Another way to look at the meaning of Me a, coefficient is to consider its effect In
the time-series plots of pavement roughness. Figure 22 illustrates the physical
significance of the al coefficient for four different cases, assuming four sections of a
pavement project with different ~rutial pavement smoothness values. In the first case,
when the al coefficient is zero, the Initially smooth and initially rough sections
converge at some point within the data range, and no relationship exists between the
initial smoothness and the future smoothness. (It is ~rnportant to note that the slopes
of the lines shown in figure 22 are indicative of the relationship between time bagel
and future smoothness, and not between initial anc! future smoothness.)
In case 2 of figure 22, when the al coefficient is between O and I, the initial
smoothness does have an effect on the future smoothness; however, the initially
smooth and initially rough sections tend toward convergence, but not within the
available data range. In other words, the initially smooth and ~rutially rough sections
will exhibit the same level of smoothness at some point In the future, but that future
point could be 5 years or it could be 25 years.
Case 3 of figure 22 shows the effect when the al coefficient is equal to 1. In this
case, the smoothness of Individual sections parallels one another, with initially
smooth sections remaining smoother over dine.
Finally, case 4 of figure 22 illustrates that when the a, coefficient is greater than I,
the initially smooth and initially rough sections tend to diverge over the available
data range. This would suggest that dynamic loading effects are causing the
roughness of the initially rougher sections to increase at a more rapid rate.
Tests for the statistical significance of the regression coefficients of We
independent variables (Si and t) are also outputs of the regression analysis. The
statistical significance of Me independent variables is evaluated using the p-value,
which indicates the probability that the significance of the effect of the independent
variable fin this case, initial pavement smoothness) on Me dependent variable (future
pavement smoothness) is due to chance alone. Obviously, the smaller the p-value,
the stronger the indication that the independent variable (initial smoothness) has a
truly significant effect on the dependent variable (future smoothness). For this
evaluation, a significance level of 10 percent was selected, meaning that if the p-value
of the nutial smoothness regression coefficient balk is less than 10 percent, the results
are considered significant.
An example plot showing regression results is provided in figure 23 for a PCC
construction project in Georgia (~-575 NB, Cherokee and Pickens Counties). This
figure shows the pavement roughness at year ~ as a function of the initial roughness.
The results of the regression indicate a good correlation (R2 = 0.82) for the linear
model, and Me a, value of 0.89 demonstrating a strong relationship between the
initial pavement roughness at construction and Me pavement roughness at year 8.
For this project, each point on the graph is a I-mi (~.6-km) section along the length of
the construction project.
50
OCR for page 51
~a)
o
Act
° 3=
O G 0
~ can
Cat t~ m
O='
U:) ~
0
0
E
cn
~ cad co ~
~ t ~ ~ ~ ~ ~
~ ~ ~ ~ .
-.,\
T~
8
= ~ ·~
~°
.4pt, .
~ A\
., ., ., ., ., .\ A\
Iw/ul 'xapul sseu46noj
_ ~ ~ ~
~ At t; ~6 ~6
lime V..
Ace
bile
E
an
lo
In
E
o
lo
C~
E
F
U.
' ' ' ' :~o
!UlIU! 'xapul ssauq6no~
~ c ~co ~
au O t 0 O ~ O '
~., ~ 1 O:
a) u, u' cn cn
._
cn
0 _
A ~ ~
_ s ~
o Ct
~ E tt
cn ~ ~
C~ C,0 O
o
-
cn
·` \
. ~ ·' \
~ ; \
he ~ -~N
. e~ ;`
'~ ~ ·.
a~ ~ ·.
'5 ~ ·` ~
· \; V
'"~e \~\
'.,
C~
U)
o
._.
\
\
o
C~
E
u,
a
O
·~
o
.~
a~
._
~ 0 + 03 ' (,0 1 (0 ~
11~=,~ ~ ~ 0
.. ~ ~ ~:,~\ 0 ~
~ e tV ~ ,=
E 0
cn ~w/u! 'xapul ssau46no~
'eD
'~0 *~\
., ., ., ., . ,. ,` ;,
~u'IUl 'xapul ssau46no~
. 1 .
51
OCR for page 152
Table 38. Comparison of current and hypothetical pay adjustments for State 4 PCC.
.
Pay Factor Corresponding to:
Maximum Incentive ~ | | Maximum Disincentive
PI=0 inlmi PI=7 in/ml ~ PI=20 inlmi
=
1.05
Pavement Family Full-Pay Level
State 4 PCC
Current LeKI
PI=7 inimi
New Level Option 1
PI=5 in/ml
1.10
New Level Option 2
PI=3 in/ml
1.00 0.90
0.95 j 0.33
0.89 0.28
l '
1.05
1 in/ml = 0.016 m/km
As discussed earlier, the SHA should be willing to pay the cost of building a
pavement to a nominal smoothness plus any benefits of building it to an even
smoother level. Although significantly greater pay adjustments, like those shown in
figures 83 through 85, might seemingly entice many contractors to build to O in/ml (O
m/km), contractors must take into consideration their cost of Paining additional
smoothness. Hence, the level to which Hey build will be dictated by the difference
between the anticipated incentive and the ASC. Contractors will build only to the
level where He incentive still exceeds the ASC, for as one contractor pointed out In
his response to the first contractor survey, "No contractor will pay more to get less."
By subtracting the ASC in He numerator of He pay factor equation (equation 9),
as shown below, the contractor's perspective on what smoothness level to build at
can be better understood.
PFaC = [BPa~ + (Lccad. ~ LCCaC)-ASC]/BPad (9)
where:
PFac = Pay factor for as-constructed smoothness level.
BPad = Bid price corresponding to as-designed, or target, smoothness
level, $.
LCCad = PW life-cycle cost of maintenance and rehabilitation corresponding to
as-designed smoothness level, $.
LCCac - PW life-cycle cost of maintenance and rehabilitation corresponding to
as-constructed smoothness level, $.
ASC = Additional smoothness cost, $.
The resulting pay factor curve indicates two things to the contractor: the break-
even smoothness level and the most profitable smoothness level. These levels are
illustrated in figure 87. At the break-even smoothness level, the cost of additional
smoothness equals the incentive payment, and the resulting pay factor is 1.0. At the
most profitable smoothness level, the difference between He incentive payment and
the ASC is greatest, and the resulting pay factor is represented by the maximum
152
OCR for page 153
· Con~actor's view of pay factor curve (with |
consideration of AS C)
· Pay factor curve
1.05 - _....................................
1.00
0.95
0.90
0.8=
, , ... Hi,,
_ -
. ~
I'
_ :
Contractor s
cost of
............ . . . ...... .
additional
smoothness
............ . . . ...... . .
(ASC) .
·-
, ~._ ~
e.~. \'
At' 2
...................... ;,,.. 1
a'_
Frost profitable "Break-even"
smoothness levet .. smoothness level
, _ .
: :
4 4.5 5
Initial RI
Figure 87. Illustration of contractor's perspective on pay adjustments (based on
analysis of State 3 PCC parkway pavements).
point on the contractor pay factor curve. It is exactly at this point that the rate at
which additional incentive money is obtained becomes less Man the rate at which
additional smoothness costs are incurred.
Inclusion of User Costs in [CCA
The inclusion of user costs in a LCCA is a controversial discussion topic in the
transportation community. There is a general consensus that user costs must be
included when conducting a LCCA; however, there is much controversy on what
user costs should be included, how they should be calculated, and how they should
be used. Although most States feel the LCCA should include user costs, there is a
need for guidelines on how to accomplish this {ask.
The LCCA performed up to now have only considered initial construction and
future rehabilitation (overlay only) costs. However, because many people believe
that user costs should be included in a comprehensive LCCA, an attempt was made
in this study to include the effects of user costs as a function of initial serviceability.
Although any estimate of user costs is a crude approximation, the magnitudes of the
estimated costs presented in this report are believed to be representative. A
discussion of this user cost investigation is provided below.
153
OCR for page 154
User costs were estimated based on results reported by McFarland (1972~. Four
different types of user costs were considered in that investigation, consisting of lime
(delay), vehicle operating, accident, and discomfort costs. All four were evaluated In
terms of cents per vehicle mile as a function of serviceability (PSI) and the following
six highway types:
I. Rural 2 lane.
2. Rural- ~ lane, undivided.
3. Rural ~ or more lanes, divided.
4. Urban 2 lane.
5. Urban ~ lane, undivided.
6. Urban ~ or more lanes, divided.
The cost tables provided In McFarIand's 1972 report were updated to 1996 costs using
an annual Inflation rate of 5 percent.
Because user costs are based on the number of vehicles, traffic data were required
for the analysis. Average daily traffic (ADT) data was plotted versus time for a
particular project. A polynomial regression equation was fit through the ADT versus
time data In order to establish a reasonable relationship to serve as a means of
extrapolating future traffic over the analysis period.
The developed best-fit mode} (a function of Crucial smoothness and limed was used
to predict the future roughness on a year-by-year basis for different Initial
smoothness levels. Rehabilitation (for simplicity just the placement of an AC overlay)
was applied when Me serviceability reached a level of 2.75 (as determined from the
predicted-life-versus-initial-smoo~ness curse developed for each pavement type in
each State). Overlays were assumed to last 10 years and deteriorate linearly from the
target smoothness level to We trigger smoothness level over time.
The appropriate cost per vehicle mile is obtained from Me updated cost tables by
knowing the projected pavement smoothness and identified pavement type. Yearly
user costs are calculated by multiplying the appropriate cost per vehicle mile tunes
the projected number of vehicles traveling on the project for a given year (365 days
times the projected ADT). For a given project, the cumulative user costs for different
initial smoothness levels can be plotted versus time.
For this Investigation, two specific Interstate pavement projects were evaluated.
These projects consisted of one AC and one PCC project each located In Me same
State and built In the mid 1970s.
~- , - - ,
Figures 88 and 89 show the cumulative PW user costs versus time for these two
projects, respectively. Both graphs include all four types of user costs (i.e., time,
vehicle operating, accident, and discomfort). Because of the magnitude of these costs,
it is difficult to get a clear idea of the cumulative PW dollars saved over time due to
154
OCR for page 155
80,OOO,oOo
70,000,000
-
-
~60,000,000
c
lo,
' 50,000,000
40,000,000
30,000,000
1 O,000,000
60,000,000
a,, 50,000,000
~40,000,000
o
U
u,
30,000,000
20,000,000
-
-
U 10,000,000
o
Included User Costs: Time, VOC, Accident, Discomfort.
Trigger of 17o in/mi, and a target of 60 in/ml were used.
. ~
.-~~ T
,~
l
I'>
it,
A'
A'
it,'
O- , 1 1
0 5
1 0 15 20 25 30 35 40 45 50
Age (years)
Figure SS. Cumulative PW user costs versus time for
selected PCC pavement project.
Included User Costs: Time, VOC, Accident, Discomfort.
Trigger of 175 ~n/mi, and a target of 50 in/ml were used.
4,¢7#
0 5 10 15 20 25 30 35 40 45 50
Age (years)
Figure 89. Cumulative PW user costs versus hme for
selected AC pavement project.
155
O in/ml
60 in/ml (target)
---90 in/ml
O in/ml
50 in/ml (target)
---70 inlmi
OCR for page 156
changes in initial smoothness. However, an estimate of the savings associated with a
given initial smoothness level as compared to a target smoothness level can be
determined using the following equation:
SAV~GSX~ = CPWUCTARGEr~-CPWUCX(Y) (10)
where:
SAV~GSxc$'
=
Savings in cumulative PW user cost for a given initial
smoothness level (X) at a particular year (Y), $.
Cumulative PW user cost calculated for the target ~rutial
smoothness level at a particular year (Y), $.
CPWUCx = Cumulative PW user cost calculated for the given initial
smoothness level (X) at a particular year (Y), $.
In the case of the PCC project, table 39 contains values of actual yearly user costs,
PW yearly user costs, and cumulative PW user costs for initial smoothness levels of 0,
60 (target), and 90 in/ml (0, 0.95, and 1.42 m/km). These three levels correspond
roughly to California PI values of 0, 7 (target), and 15 in/ml (0, 0.11, and 0.24
m/km). Figure 90 illustrates the savings in cumulative PW user costs over time for
the three different initial smoothness levels, as compared to the yearly values at the
target initial smoothness level. For example, at year 15, the cumulative user costs
(taken from table 39) for the three initial smoothness levels are as follows:
Level 1: 0 in/ml (0 m/km) = $22,994,ooo
Level 2: 60 in/ml (0.95 m/km) = $24,883J000 (target level)
Level 3: 90 in/ml (1.42 m/km) = $26J143,OOO
Therefore, the respective savings in cumulative PW user costs at year 15 (compared
to the estimated value at the target of 60 in/ml (0.95 m/km)) are calculated as
follows:
Savings at O in/ml (0 m/km) => $24,883,000 - $22,994,000 = $1,889,000.
Savings at 60 in/ml (0.95 m/km) => $24,883,000 -$24,883,000 = $0.
Savings at 90 in/ml (1.42 m/km) => $24,883,000 - $26,143,000 = -$1,260,000.
These three respective savings are plotted at year 15 in figure 90. It should be noted
that the irregularities seen in figure 90 are due to the applications of overlays at
different years for different initial smoothness levels. These different overlay
schedules result in different rates of increasing cumulative PW user costs (see table 31
for further explanation of specific values and their interactions).
By incorporating into the LCCA the cumulative PW user costs associated with
different initial smoothness values and corresponding to a specified analysis period, a
comprehensive graph of "total life-cycle costs plus user costs" versus initial
smoothness can be developed. Figures 91 and 92 illustrate such graphs for the two
subject projects. As can be seen in each of these figures, three different versions of
156
OCR for page 157
Age
(years)
1
2
-3
4
5
6
8
10
11
2
13
14
~15
16
17
is
19
20
2
22
23
24
25
26
27--
28
. 29
30
31
32
33
34
35
36 .
37
38
39
40
41
42
43
.
44
45
=46~_
_47 _
48
49
50 .
Table 39. Actual yearly, PW yearly, and cumulative PW user costs
for different Crucial smoothness levels ($ thousands).
Yearly User
Cost Per
Mile
1,554
1,636
1,718
1,800
1 882
1964
2,046
2,128
2,210
2,292
2,374
2456
2,538
2,620
2 702
2,784
2,866
2,948
3,030
3,112
3194
3,276
3,358
3,440
3,672
3,811
3,982
4,150
4,315
4491
4,664
4,832
4,995
5,150
4527
4679
4,868
5,053
5,234
5,427
5,617
5800
,
5,975
6,142
5,381
5,546
5,754
5,956
6,153
6,364
Yearly PW
User Cost
Per Mile
1,494
1,512
1,527
1,539
1 547
1 552
1,555
1,555
1,553
1,548
1,542
1,534
-
1,524
1,513
1500
1,486
1 471
1,455
1,438
1,420
1401
1,382
1,362
1342
1,377
1,375
1,381
1,384
1,384
1 385
1 383
1,377
1,369
1,357
1 147
1,140
1,141
1,138
1,134
1,130
1,125
1 117
1,106
1,093
921
913
911
907
900
895
~ Represents year of overlay.
Cum. PW
User Cost
Per Mile
1,494
3,007
4,534
6,072
7619
9171
10,726
12,281
13,833
15,381
16,923
18 457
19,981
21,494
22994
24,481
25 952
27,407
28,845
30f265
-
31 666
33,049
34,411
35,753
37,130
38,505
39,886
41,270
42,654
44038
45,421
46,798
48,167
49J5254
50 672
,
51,812
- 5-2,952
54,091
55,~5
56,355
-57,48 - 0
58,597
59703
,
60,797
61,718
_ -62,63
63 542
. ,
64,448
65,349
66,244
~ Yearly User
| Cost Per
L Mile
1,619
1,711
1,805
1,900
2,002
2108
. ~
2~16
2,326
2,438
2,551
2,666
2,788
2,911
2,731
2,858
3 008
. ~
1 3157
. ,
3,304
3,461
3,616
3 768
. ~
1 3 916
. ,
4,060
3,586
3,725
3,893
4,060
4~3
1 4,397
1 4 569
. ,
4,735
4,897
5,051
4,441
4,592
4,779
4,963
5,142
5,334
5,522
5,703
5877
, ,
6,042
5296
5,459
5,665
5,866
6,061
6,270
6,474
157
.
Yearly PW
User Cost
Per Mile
1, 5 5 7
1,582
1,604
1,624
1,645
1,666
1,684
1 700
1,713
1,723
1,732
1,741
1 748
1,577
1,587
1 606
1621
1 631
1,643
1,650
1,654
-
1,653
1,647
1,399
1,397
1,404
1,408
1,408
1,410
1 409
1,404
1,396
1,384
1,170
1,164
1,165
1,163
1,158
1,155
1,150
1,142
1132
1,119
943
935
933
928
923
918
911
_
Cum. PW
User Cost
Per Mile
1,557
3,138
4,743
T 6,367
8,012
9,678
T 11362
, ~
1 13061
, ~
1 14,774
1 16 498
, ,
18,229
9,971
21719
, ,
1 23,296
24,883
~26,489
28 109
, ,
1 29740
, ,
1 31,383
T 33,033
1 34 687
, ,
1 36 339
, ,
37,986
39,386
4~0,783
42,187
43,595
1 45,003
r 46413
, ,
1 47 822
, ,
49,226
50,622
52,0~06
53,177
54,340
55,505
56,668
57,826
58,982
-
60,132
61,274
62,406
63,525
64,467
65,402
66 335
, ,
67,263
68,186
69,103
70,014
~ Yearly User| Yearly PW | Cum. PW
| Cost Per | User Cost | User Cost
| Mile ~ Per Mile | Per Mile
1~774 1 1,706 1 1,706
_1,878 1 1,737 1_ 3,443
1,985 1 1,765 1 5,207
2,094~ -1,790-r- 6,998
2,205 ~1,812 T 8,810
2,318 11,832 ~10,642
2 432 11 848 ~12 490
, . .. . .
2,547 11,861 L 14,351
2,304 11619 1 15970
, .. . .
1 2424 11638 1 17607
, , I, , ,
2,565 11,666 1 19,273
2,705 11,690 1 20,963
r ~2,845 T 1,7081-- 22,671
2,992 1 1,728 1 24,399
1 3,140 1 1,743 1 26,143
L_ 3,284 T 1,754 1 27,896
r ~ 3,426 1 1,759 1 29,655
3,564 1 1,759 1 31,415
1 ~ 3,159 1 1,499 1 32,914
r 3,?91T 1,5-02~ --34,416
rim 3,441~ ~1,514 1~_~35,930
r36081 15231 37453
,. . . . .
13,764 1 1,527 1 38,980 1
.- .
3~929 1 1,533 1 40,513
4,092 1 1,535 1 42,048
14,2521 1,534 r 435811
4,407 1 1,528 1 45,110
14,555 1 1,519 1 46,6296
I~ 4014 1 1,287 1 479161
I~ 1 1 ~
14,158 1 1,282 1 49,198 1
.
i4,336 1 1~286 1 50,483
--4,511 ~1,286 1 51,769
4,683 ~1,284 1 53,053
,866 t 1~2821 54,335
15,045 1 1,279 1 55,614
i5,219 1 1,272 1 56,886
5,387 1 1,262 1 58,1480
5,547 1 1,250 1 59,397
4,869 1 1,055 1 60,452
5,026 1 1,047 1 61,499
1 ~ Owl ~:~
1 5,414 r 1,043 1 63,587 1
. . ,
1 5,602 1 1,037 1 64,625 1
1 5,802 r-10331 656-581
5,99~1__~ 1,027 l 66,684:
6 187 T 1 018 1 67 703
, . . . . .
6,367 1 1,008 1 68,711
6,53-8 T 9951 69,706
*5,723 T 838 1 70,543
5,893 1 829 1 71,372
OCR for page 158
.,, 3,000,000
._
OJ
c~
v 2,000,000
U
v'
1 ,000 ,000
._
-
U
u,
._
U. -1 ,000,000
-2,000,000
~.
~ _
~.. .... ,\..
_ 4;0 ~ ; 2
~Nf,
v
I T I I I
Included User Costs: Time, VOC, Accident, Discomfort.
, . . ~ . ., ~
r ~ ~ ~ ~ r
) 2 5
~,
Age (years)
, ,
~0 s e
Figure 90. DIustration of sav~ngs ~n cumulative PW user costs over fime.
so,ooo,ooo - . - . . . . .
1 1 1 1 1
70~000tO00 ~- ~ - ; . .. ................................
60,OOO,OOO
6~
-
U 50J000~000
-
U 40,000,000
30,000,000
20,000,000
10,000,000
+O in/ml
~60 in/ml (target)
~90 in/ml
1
_ _
_ ~
0 20 40 60 80 100 120 140
Initial Roughness (MRN), in/ml
Total LCC
· Total LCC + VOC
~Total LCC + Time Delay +
VOC
Total LCC + Time Delay +
VOC + Accident +
Discomfort
Figure 91. Graph of total LCC (including user costs) vs. ~nitial roughness
for selected PCC pavement project.
158
OCR for page 159
4D,000,000 1
40,000,000 ~ '_ -A
3~,000,000
30,000,000
25,000,000
20,000,000
15,000,000
10,000,000
5,000,000
0 20 40 60 80 100 120 140
Initial Roughness (MRN), in/ml
| Total LCC
| · TotalLCC +VOC
Total LCC + Time Delay +
VOC
. Total LCC + Time Delay +
VOC ~ Accident +
Discomfort
Figure 92. Graph of total LCC (including user costs) vs. iriitial roughness
for selected AC pavement project.
user costs were depicted; the top curve representing all four user cost components,
the curve below that representing two user cost components twine and vehicle
operating), and the one below Cat representing only vehicle operating costs. The
total life-cycle cost curve, cor~sist~ng of initial construction cost and future overlay
costs, is located close to the x-axis.
The effect of including user costs In the LCCA is obvious in both of these figures.
User costs so overwhelmingly dominate the project life-cycle costs that the least
overall cost occurs at the smoothest possible level, which is a MRN of O m/ml (O
m/km). Although other projects were not analyzed, variations in savings from
project to project would be expected, primarily due to factors such as the
smoothness-life relationship and traffic. Nevertheless, it is strongly believed that the
magnitude of savings associated with smoother levels for other projects would stay
relatively uniform, such Cat Me smoothest level is always identified as Me most cost-
effective level.
Summary
This section outlined a procedure for evaluating the cost-effectiveness of initial
smoothness levels and presented a detailed evaluation of several current pavement
smoothness specifications using that cost-effectiveness procedure. The evaluation
procedure introduced is based on LCCA techniques and is contingent on the
159
OCR for page 160
inclusion of relationships between initial smoothness and initial construction cost and
between initial smoothness and pavement life.
The evaluation of current smoothness specifications focused on two aspects: the
optimum cost-effective smoothness level and the appropriateness of
~ncentive/dis~ncentive payment amounts. Asphalt and concrete specifications from
five different States were evaluated using LCCA techniques, contractor estimates on
the cost of obtaining additional smoothness, and the smoothness-life relationships
developed earlier for various pavement families (i.e., combinations of State and
pavement type).
Some of the key findings of the smoothness specification cost-effectiveness
evaluation are as follows:
.
Seven of rune concrete pavement families showed Me optimum cost-
effectiveness (Pl) range, excluding consideration of user costs, as being
between O and 5.5 in/ml (O and 0.09 m/km).
Four of five asphalt pavement families showed the optimum cost-effectiveness
(Pl) range as being between O and 3.5 in/ml (0 and 0.06 m/km).
Eleven of 13 asphalt overlay families showed Me optimum cost-effectiveness
(Pl) range as being between O and 2 in/ml (0 and 0.03 m/km).
In comparison with actual current pay adjustment curves, the theoretical pay
acljustment curves, on the whole, showed that greater incentive and
disincentive amounts are warranted in terms of the benefits/disbenefits
obtained from various initial smoothness levels.
When shifted to coincide with full-pay (Pl) smoothness levels of 5 and 3 in/ml
(0.08 and 0.05 m/km) for PCC and AC pavements, respectively, the
recalculated theoretical pay adjustment curves still showed greater magnum
Incentive amounts and more punitive disincentive amounts in comparison with
the current pay adjushnent curves.
The inclusion of user costs in a comprehensive LCCA has a profound
effect on the determination of the most cost-effective smoothness level.
For two selected pavement projects, a comparison of cumulative PW
user costs associated with Tree distinct crucial smoothness levels (MRNs
of 0, 60, and 90 in/ml [0, 0.95, and 1.42 m/km]) showed significant cost
savings for Me pavement constructed to the smoothest level. For both
projects, Me addition of user costs to Me project costs (construction cost
plus future overlay costs) clearly dominated the total LCC such that the
most cost-effective smoothness level was found to be 0 ~n/mi (0 m/km).
Overall Chapter Summary
This chapter presented the results of several analyses conducted on the effect of
Initial pavement smoothness and of pavement smoothness specifications. The results
of the analysis of the effect of initial pavement smoothness on the future smoothness
of the pavement clearly show that nutial pavement smoothness has a significant effect
160
OCR for page 161
on the future smoothness of the pavement. This analysis was conducted for different
pavement types and for different age ranges, with the results being that initial
pavement smoothness is significant to the future smoothness of the pavement. The
effect was stronger for new pavement construction than for overlay pavement
construction, suggesting that the performance of overlays is governed more by other
factors (e.g., reflection cracking). These results were conducted on many sets of
independent pavement performance data from across the US and over different
pavement types.
Analyses were also conducted to investigate the effect of Crucial pavement
smoothness on pavement life. The results of the analysis strongly indicate that initial
pavement smoothness has a significant effect on pavement life, using both roughness
mode! and pavement failure analysis techniques. The analyses show that added
pavement life is obtained by achieving a higher level of initial smoothness over the
range of initial smoothness values that were available for analysis. The rate at which
additional life is achieved is dependent upon, among other things, pavement type,
facility type, and location.
Sensitivity analyses, in which the percentage change in life as a function of
percentage change in smoothness was determined, showed sizable increases In life for
most pavement families, corresponding to nominal increases in smoothness. At least
a 9 percent increase In life corresponding to a 25 percent increase In smoothness
(from a target profile index of 7 in/ml [0.~l m/km] for concrete and 5 in/ml [0.08
m/km] for asphall) was observed for the vast majority of the pavement families. A
50 percent increase in smoothness from these target levels was found to Increase life
by at least 15 percent ~ many cases. Me results of Ads analysis are based on the
assumption that roughness is a primary factor influencing the decision to rehabilitate
a pavement.
An analysis of the effect of pavement smoothness specifications on the resulting
levels of initial pavement smoothness was also conducted. This analysis indicated
that pavement smoothness specifications do have a beneficial effect on the resulting
initial smoothness of the pavement. Comparison of pavement smoothness
distributions before and after the implementation of a smoothness specification
indicate decreased initial roughness values and decreased variability for pavements
constructed under the smoothness specification.
A procedure for evaluating the cost-effectiveness of initial smoothness levels was
developed and presented. The procedure illustrates how the life-cycle costs of a
pavement constructed to different smoothness levels (under various
incentive/disincentive scenarios) can be determined and Men compared to see which
level is most cost-effective. Such an evaluation was conducted on several pavement
families, and the results showed Mat Me most cost-effective smoothness (Pl) levels
are considerably higher than what is typically used as the current target (PI between
5 and 10 in/ml (0.08 and 0.16 m/km. Seven of nine concrete pavement families
showed the optimum cost-effectiveness range as being between O and 5.5 ~n/mi (0
161
OCR for page 162
and 0.09 m/km), whereas four of five asphalt pavement families showed the
optimum range as being between O and 3.5 Mimi (0 and 0.06 m/km).
In comparison with actual current pay adjustment curves, the theoretical pay
adjustment curves developed in this study showed, on the whole, much greater
incentive amounts and much more punitive disincentive amounts. When shifted to
coincide with full-pay (Pl) smoothness levels of 5 and 3 in/ml (0.08 and 0.05 m/km)
for PCC and AC pavements, respectively, recalculated theoretical pay adjustment
curves stiD showed greater maximum incentive amounts and more punitive
disincentive amounts than current pay adjustment curves.
The inclusion of user costs in a comprehensive LCCA has a profound effect on the
determination of the most cost-effective smoothness level. For two selected State
pavement projects, me addition of user costs to total life-cycle costs resulted In O
n/mi (O m/km) as being the most cost-effective smoothness level.
Taken together, the implications of these findings are very significant. Initial
smoothness has been shown to provide value in the performance and service life
of pavements. This serves to justify the importance of achieving high initial
smoothness from a pavement structural standpoint, which is in addition to the
importance of smoothness to the nding comfort of the user. And, this strongly
supports the benefits of employing smoothness specifications, which have been
shown to be an effective means of achieving higher levels of initial smoothness.
162
Representative terms from entire chapter:
pavement smoothness
