|
Items in cart [1] |
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 89
CHAPTER 5. INVESTIGATION OF HIGHWAY
MAINTENANCE QA AND LONG-TERM
PERFORMANCE OF HIGHWAY FACILITIES
Introduction
Establishing and operating a maintenance QA program can be a substantial
inves~anent for a highway agency. As such, Me QA program must show a lasting
beneficial effect on the overall condition of highway facilities and We perception of Cat
condition by He traveling public. The program must result in a higher quality
highway system Bat in He long term provides benefits greater Han He costs of He
program. Such benefits will only occur if substantial improvements in long-term
performance of highway facilities are made.
The long-term performance of highway facilities can be judged In terms of He
following four characteristics, which, in essence, are the four main considerations of
maintenance:
Safety.
Comfort and convenience.
· Aesthetics.
· Preservation of investment (service life).
The maintenance QA program developed under NCE]RP Project ItI2 was
specifically designed to evaluate He characteristics most apparent to He traveling
public: safety, comfort and convenience, and aesthetics. An implementing agency,
however, has He additional responsibility of optimizing He preservation of Heir
investments, be Hey pavements, bridges, or over major highway features. This aspect
of facility performance, which is not immediately discernable to He vast majority of
highway users, is a major focus in every transportation agency, as evidenced by the
development of BMS's and PMS's. Highway agencies, therefore, are interested in
implementing a QA program Hat will not only satisfy He user's needs by achieving
and maintaining He desired LOS, but one Cat shows a positive influence on He
preservation of ~nveshnent. In over words, Heir desire is for a QA program that has
He dual effect of ensuring quality to He traveling public (~rough safe, comfortable,
and visually pleasing roadway facilities) and to He highway agency (through longer
lasting and more cost~ffective roadway facilities).
How an agency can determine whether He QA program effectively and
economically preserves key transportation inves~anents is not an easy task, as there are
numerous factors Hat can confound He {ong-term performance of highway facilities.
Perhaps He most significant factor in long-term performance is the aDocation of
highway funds among He various transportation infrastructure components (design,
89
OCR for page 90
construction, maintenance, rehabilitation). Over major factors include roadway design
(type of materials, Sicknesses, and so on), construction (quality of materials and
workmanship, conditions during construction, and so on), maintenance, Me amount of
traffic and heavy loads carried by We roadway facility, and We climatic conditions in
which We facility functions. These factors affect We rate of deterioration of a roadway
facility, and the identification of a method mat substantively links We quality of
maintenance with long-term performance would be a valuable tool to maintenance
managers.
This chapter discusses We investigative work effort undertaken to identify a process
for tracking We relationship between maintenance LOS ratings and We long-term
performance of highway features. Several tracking methods were examined and are
briefly described In We section below, titled "Work Approach." Two methods were
considered to have We most potential for relating maintenance quality and facility
performance and were, therefore, evaluated in greater detail using actual maintenance
LOS rating data and highway performance data collected from two States. These two
methods are formally presented in We QA program Implementation Manual, but We
results of We detailed evaluations regarding their effectiveness and usefulness are
provided In this chapter. The last section in this chapter provides a discourse on We
merits and ramifications of Me ideas brought to light in Me investigation of QA
Output-long-term performance tracking relationships.
Work Approach
In the search for an appropriate method for tracking Me relationship between
maintenance LOS ratings and Me long-term performance of highway features, a
number of objectives were used to define a successful methodology. First, it was
desired Mat Me methodology make use of readily available performance data
contained in existing agency management information systems (PMS's, BMS's, SMS's).
Second, Me chosen method was to minimize Me effects of confounding factors
(funding, design, traffic) on performance. Third, Me chosen method would be
developed by focusing solely on pavement facilities. The decision to focus on
pavements was based on Me following factors:
Pavement performance is most difficult to assess. That is, Me effects of
pavement maintenance on pavement performance are not as easy to assess as
Me effects of maintenance on the performance on some of Me other highway
features, such as pavement striping, signs, or guardrails.
· A large amount of Me maintenance budget goes Into maintaining pavements.
· Highway users are greatly affecter! by Me maintenance applied to pavements.
· PMS's are an excellent external source for performance clata.
Even Cough Me melons presented herein relate specifically to pavement facilities, Me
concepts are believed to be applicable to over maintenance features. This is because
90
OCR for page 91
We concepts are based on deterioration, which is a phenomenon experienced by all
man-made items.
SK different methods for evaluating the effectiveness of the maintenance QA
program were identified for investigation. These methods covered a wide array of
approaches and complexities meant to accommodate the diverse natures, goals, and
management styles of many highway agencies. The six methods are listed below, with
a brief description given for each method.
· Method ~-QA/COS! InJeX. This method is a relatively simple procedure Mat
allows an agency to make yearly comparisons of its LOS ratings and Me
associated maintenance costs. It allows an agency to first determine whether its
maintenance activities are actuary providing Me desired LOS and then evaluate
the annual cost of performing Me maintenance activities. The relationship is
expressed as a QA/Cost Index and is a function of QA {LOS) rating, maintenance
. ~·. ~
cost expenditures, and the total lane mileage tor a chosen set of highway
pavement sections (e.g., interstate concrete pavements In district 1, AC overlays
on primary routes agency wide). The QA/Cost Index is calculated using Me
following equation:
QA/Cost Index = (QA Rating x Total Mileage)/(F~rst Year $)
where: First Year $
Current Year $
n
=
Eq. 8
Current year's maintenance dollars converted to
present-worth dollars corresponding to Me QA
program's first year.
(Current Year $)/~1 ~ i)
Current year's maintenance dollars.
Number of years since Me start of the OA
do. .
program (current year - HA program first years.
i = Discount rate. 4 percent fi=0.04)is
recommended, based on historical data.
QA Rating = QA program rating data representing Me chosen
set of pavement projects (i.e., the chosen analysis
group).
Total Mileage = Total lane mileage of the highway sections
· ~ ~ ~ · .~ ~ ~ -
Included in the chosen analysts group.
The Inclusion of Total Mileage has a normalizing effect on Me index that allows
for year-to-year comparisons for Me same analysis group.
Plotted on a yearly basis (as illustrated in figure 6), Me QAICost Index could be
examined for meaningful trends. For instance, if Me index increased over time,
Me Intuitive interpretation would be Mat Me OA program is becoming more
efficient at providing Me desired LOS.
91
_ , ~
OCR for page 92
x
-
us
o
¢
1
·
1
~ -
.
-
First Year of QA
Program
1 i 1 1 1 1 1 1 1 >
Time
Figure 6. Example plot of QA/Cost Index versus time.
Method 2 Backlog Analysis. A common practice in pavement management is to
define a critical pavement condition threshold Cat is used to signal We need for
rehabilitation. The critical threshold is generally expressed in terms of
roughness (e.g., IRI, MRN), serviceability (e.g., PST, PSR), or visual condition
(e.g., PCI), and Me group of pavements win conditions below Me critical
threshold are commonly referred to as Me pavement backlog. Since a continuing
goal of any agency is to niininiize, or possibly eliminate, its backlog, Me Backlog
Analysis method was designed to track not only Me percentage of pavements In
Me backlog, but also Me rate Mat pavements deteriorate from one condition
range, or performance category, to Me next (i.e., Me accrual rate).
The Backlog Analysis method attempts to identify the relationship between
maintenance quality and pavement performance by focusing on pavement
backlog-related trends. Because higher LOS ratings signify more effective and
efficient maintenance, it is logical Mat an increase in pavement LOS should be
accompanied by a reduction in the pavement deterioration rate (i.e., Increased
pavement performance), all over factors being equal. A reduction in the rate of
pavement deterioration would subsequently lessen Me backlog accrual rate, as
well as the rates at which pavements move from a higher performance category
to a lower one (e.g., Very Good to Good, Good to Fair).
· Me~od 3-Change in Condition Indicator vs. Time. This method compares Me
annual pavement ICES rating for a chosen analysis group to Me average annual
change in pavement condition indicator (e.g., IRI, PST, PCI) representative of that
group. The memos first entails plotting Me 2- or 3-year condition trends of all
pavements in a chosen analysis group, and then developing a best-fit linear
trend by computing the average slope of deterioration for Mat group. This is
92
OCR for page 93
illustrated in figure 7, where the change in condition indicator fin this case,
condition rating survey [CRS]) for many projects is plotted versus age. The
resulting change In CRS exhibited by the linear trend line over a I-year penod
(also referred to as Me yearly condition deduct value) is Men cletermined, as
shown in figure 8.
With deduct values established for multiple years, Me annual pavement LOS
ratings for the corresponding years are linked together, so Cat Me resulting
trend between pavement maintenance quality and pavement performance can be
examined. Figure 9 provides a simple illustration of this concept using
performance curves associated win two levels of routine maintenance good
and poor. All over factors being equal, a well maintained pavement (one win a
high LOS) is expected to have a flatter deterioration slope Man a poorly
maintained pavement (one with a low LOS). Subsequently, given a critical
condition threshold for rehabilitating pavements, Me flatter slope would result
In a longer pavement service life.
9
~ . . . .
, 1~
r3 ~ \ \
3
z i
~ ~3
Al
to 20 3e
Age
Figure 7. illustration of average rate of deterioration for a group of similar
highway pavement sections (ERES, 1995~.
93
OCR for page 94
l ~
~/
I
Yearly CRS deduct ,
O' ~:
/ CRS Value from survey
-
Future
Figure 8. illustration of yearly condition deduct (ERES, 1995).
PSI)o
i -
cr,
x
(PSI)14
._
-
a)
·O
:>
h
US
CO
(1) Pavement Performance Deduct Curve
Win Good Routine Maintenance.
(2) Pavement Performance Deduct Curve
Win Poor Routine Maintenance.
-
- (2)
Time, t
Figure 9. Relationship between pavement performance and
routine maintenance (Fwa and Sinha, 1986).
94
OCR for page 95
· Method 4 Chance in Condition Indicator vs. LOS. A slight variation of method 3,
~ ~ _
this method entails plotting Me average annual change in condition indicator
(Acondition index) versus Me measured pavement LOS rating for a sequence of
time periods (see figure 10). Ideally, as the LOS raking increases, the /`condition
index would decrease, suggesting that a higher maintenance LOS corresponds
win increased pavement performance. At some point along the LOS rating
scale, Me rate of reduction in the Acondition index will decrease win an increase
in LOS rating. Hence, a plot of Acondition index versus LOS for a range of LOS
ratings may allow an agency to select Me LOS raking that bow meets the desired
LOS and optimizes pavement performance. If Were is no relationship found
between the Acondition index and the corresponding LOS raUng, then an agency
would need to carefully review Me quality of Me EMS data and reevaluate its
LOS rating program.
Method 5 Condition Indicator vs. LOS. This method Involves plotting condition
indictor values win corresponding pavement LOS ratings for individual
sections within a chosen analysis group (figure 11). Such plots can help define
Me uncertain relationships between maintenance quality ratings and pavement
condition, Hereby allowing an agency to estimate He level of maintenance
required to bring He average condition of highway pavements to a specified
level.
Because He "conditions" evaluated as part of He LOS rating program are often
different from He "conditions" measured and recorded In a PMS, and because
He LOS rating program is attribute-based whereas He PMS is variable-based, a
large amount of variability can be expected in He relationship between
pavement LOS and He selected condition indicator. Such variability is depicted
In figures 12 and 13.
· Method WRegression Model. This complex method involves conducting
simultaneous pavement condition and maintenance condition field evaluations
of several "like" pavement sections (i.e., pavements of similar type and located
in similar clunatic regions), and then performing a statistical regression analysis
to determine He influence of maintenance quality on pavement condition. In
this approach, a sufficient number of random sample units within each
pavement section are identified and then surveyed, first for pavement condition
(PCI) and Hen for maintenance condition (LOS). Using He PC] and I~OS ratings
from each sample unit, an average PC] rating and an average l.OS rating for He
pavement section are determined. The age of each pavement section at the time
of the evaluation is also required. Traffic data (average daily traffic [ADT],
equivalent singl~axle loads [ESALs]) may also be used if a wide range in traffic
levels is known to exist among He various "like" sections.
95
OCR for page 96
-
o
·
·~
o
- -
1 1 1 1 1 1 1 1 1
0 50 100
LOS Rating
Figure 10. Example plot of Acondition index versus LOS raking.
100
x
I_
~ 50
._
o
r
X:
X
AX
x
x
X /X
X X/
Xp
X)
1 1 1
0 50 100
LOS Rating
Figure Il. Conceptual illustration of pavement condition-LOS
rating relationship.
96
OCR for page 97
200
a
~ 100
._
~_
x x x x
x x
~ x x x x x x~-
x x x x
y
x x
x x
O
o
x x x x
x x
x x x
x
x x x x~
x x
x x xx x x x x
x x
x x x
x
v
x ^
x
x x x
x x
x x
X A ~
X X
100
LOS Rating
Figure 12. Conceptual iDustration of variability in roughness-LOS
rating relationship.
U 50-
P~
o
/ - x x x x x x
x x
x
~x x x x x
~x x x x x x
~v Y
~ A V V
v A A
~SC X ~X X X X
°A' X
X
X
X
OLOS Rahng
100
Figure 13. Conceptual illustration of variability ~n condition-LOS
rating relationship.
97
OCR for page 98
One year later, Me same random sample units are again evaluated for pavement
condition and maintenance condition. The change In PCT for each section is
calculated by subtracting We current year's PC] from me previous year's PCI. A
mean LOS rating is computed for each section by averaging Me current year's
LOS rating and Me previous year's LOS rating.
Table 13 gives an example of Me data used in this type of analysis. These data
represent 30 individual projects of a given pavement type (e.g., jointed plain
concrete DPC] pavement) and located bra one geographical area (e.~. District Il.
~O .
Using this type of data set, a statistical regression analysis is then conducted Mat
relates Me maintenance LOS win the change in PCT, as expressed in Me
following general form:
~lPC] = aO + AL (~+ a:Age + a3Tra~c
where: Am
as, al, as, as =
L(W
Age
Traffic =
Eq. 9
Annual change in PCI.
Regression coefficients.
Average LOS rating.
Age of pavement at time of second survey, years.
Estimated traffic, vehicles/day or equivalent s~ngle-
axIe loads.
For maintenance LOS to have a significant effect on Me change in PCT, Me
resuming probability value (p-value) corresponding to Me a, regression
coefficient would have to be less Man the chosen significance level (e.g., 5
percent or 10 percent). If LOS is found to have a significant effect on APCI, then
Me type and magnitude of Me effect can be determined by examining me
resulting al value. IdeaDy, as, as, and as should be negative, whereas al should be
positive. ~ Ads way, increased age and Increased traffic result In larger
decreases In PCI, and increased LOS results In smaller decreases in PCI.
Preliminary Assessment of Tracking Methods
Many factors were taken into consideration when determining Me worthiness of Me
six me~ods for tracking Me relationship between maintenance LOS ratings and Me
Tong-term performance of highway features. Key factors in this initial evaluation
Included Me complexity of Me memos, the availability of required data, and Me
associatecl confounding variables (e.g., traffic, climate, design, rehabilitation).
98
OCR for page 99
Table 13. Regression mode! data example.
Site
No. Locabon
1
2
3
4
5
6
7
.
8
9
10
11
12
13
14
15
-
16
17
18
19
20
21
22
23
24
25
26
27
28
29
l
30
Rt S3 hDP 0-7.6
Rt 53 hDP 7.6-11.2
Rt 53 0P 22.5-25.6
Rt 53 hDP 25.6-32.2
Rt 53 hDP 51.3-60.0
I-17~BB hDP 14.7-20.4
I-17~IB hDP 20.4-24.8
I-17bJB hDP 24.8-30.5
I-17~IB hDP 30.5-40.6
I-17SB MP 14.7-20.4
I-17SB MP 20.4-24.8
I-17SB MP 24.8-30.5
I-17SB hDP 30.5-36.2
Rt 3 hDP 94.7-98.3
Rt 3 hDP 104.7-110.0
Rt 3 hDP 110.0-113.6
Rt 26 hIP 0-3.4
Rt 26 hDP 3.4~6.4
Rt 26 MP 24.3-27.9
Rt 26 hDP 27.9-33.4
Rt 26 MP 40.3~42.3
Rt 26 0P 42.3-46.8
Rt 26 MP 46.8-50.5
US 64EB hLP 174.3-180.3
US 64EB 0P 180.3-188.1
US 64EB 0P 204.5-209.8
US 64VVB hIP 168.6-174.3
US 64VVB hIP 174.3-180.3
US 64VVB hIP 180.3-188.1
US 64VVB MP 204.~-209.8
PCI RaUng LOS Radng
1996 1 1997 APCI 1996 ~ 1997
76 73 -3 88
66 62 80
98 96 -2 100
94 93 -1 100
87 87 0 96
89 87 -2 86
- 67 63 -3 78
50 43 -7 56
46 38 -8 50
69 64 -5 78
74 68 -6 79
83 81 -2 93
80 77 -3 94
.
94 ~ 94 0 100
100 98 -2 100
66 62 -4 79
77 73 -4 84
93 90 -3 94
90 88 -2 100
76 73 -3 86
73 69 ~4 84
74 70 -4 83
45 36 9 54
48 41 -7 58
58 51 -7 65
42 35 -7 52
45 37 -8 58
65 60 -5 73
71 66 -5 80
62 56 -6 80
85
75
97
98
-
92
82
72
51
46
75
l
72
90
88
100
100
74
78
90
92
87
78
l
81
46
54
.
60
.
54
53
67
72
75
Mean | Ag~ | ADT
LOS 1 1997 1 1997
~:~
77.S 1 10
1
~L:
94 1 6
1
84 1 6
.
l ~
53.5 1 19
l
48 1 19
..
~76.5 1 15
. 75.5
.
91.5 1 7
1
91 1 7
~_
100 1 3
1
76.5 1 13
~L~
92T 5 1
1 1
96 1 5 1
1 1
82 1 9 1
1 1
50 1 17 1
.
~LN
62.5 1 17
53 1 18
1
55.5 1 18
70 1 15
.
76 1 15
1
77.5 1 15
4,490
4,760
4,050
1
4,050
3,870
12,540
11
112,5401
11
12,310
12,070
r 116601
11
11,6601
1
11,380
11,240
2,740
2~430
2,430
-
6,880
6,880
7,240
7,240
.
7,310
_ .
7,310
7,260
8~040
-
8,040
7,920
8,180
1
8,180
8,180
99
OCR for page 106
Table 18. 1991 pavement condition range probability matrix.
1991 Distribution 1992 Year Distribution
| Miles of Very ~ Very
Condition Pavements Good Good Fair PoorPoor
Very Good 3.099 0.00 1.00 0.00 0.00
_
Good 147.043 0.10 0.31 0.43 0.09
Fair 96.590 0.00 0.13 0.53 0.34
Poor 23.904 0.00 0.00 0.10 0.45
Very Poor 77.993 ~0.00 0.00 0.00 0.02
Total ~348.629 ~0.041 0.181 0.33' 0.171 0.25
0.00
0.00
0.00
0.45
0.98
Table 19. 1995 pavement condition range probability matrix.
Rehabilitated
°.oo
0.07
o.oo
o.oo
o.oo
0.03
1995 Distribution 1996 Year Distribution
| Miles of ~Very ~ Very
Condition Pavements Good Good Fair PoorPoor
Very Good 54.594 0.53 0.45 0.00 0.02
Good! 278.609 0.07 0.75 0.12 0.00
Fair ~26.494 ~0.00 ~0.00 ~0. ~ 7 ~0.00
Poor 10.872 0.00 0.00 0.02 0.59
Very Poor ~ 17P`O~] r ooo ! °°°! Cal! 026! ~
| Total 1 388.3691 0.131 0.601 0.121 0.031 0.07
Rehabilitated
0.00
0.00
0.01
0.05
0.00
0.00
0.74
Table 20. Average LOS ratings by year.
~..
| 11 Mean LOS
I Year || Rating
.
11
1 1991 11 70
. ·.
995 Il -90
106
0.53
0.39
l
0.00
°.°51
OCR for page 107
1.o l
QGj
It
Q7
~ QB
Z 05
o
Q.
-
~ 0.3
C]
no
Q1
~0
0.1
QO .
~ Q7
8 os rams ~ - - 1
Mean = 7.59
e,Q. S~Dev=~.16
`~ Q3 Jilt,,,/
02 :"//
. ;~
Very Good
l 1991-92, QA = 70
c . Mean = 8.00
, Std Dev= 0
1 1
1995~96~ ~ =
. Mean = &52
Std Den= Q84
/ '
/ ~
; Far ; Poor ; Very Poor , Rehabilitated
New Category
Figure 14. Probability plot for pavements win a previous
condition of Very Good.
, ; 1991-92, QA = 70
. - 7~ Mean = 7.47
\j _r/ Std Dev= 0~74 ,
IT -a m ~
0 ~' Fair , Poor ' Very Poor Rehabilitated
New Category
Figure 15. Probability plot for pavements win a previous
condition of Good.
107
OCR for page 108
1.0,
Q.
0.8
0.7
QB
0 05
g
_~ 0.~
~2
`~, Q3
C]
Q2
Q1
1.0
Qua
Q8
0~7
~-
O ~
I:
3 Q4
~ Q3
C]
02 .
Q1 .
An .
. ~
~1 1
# 1 ~
~1 1
1 ~1
# 8
8
1
,' ~199~96, QA = 90
; ' I / Mean = 660
~ I / Std D3v= Q38
', /
`' tat
1
1991-92, QA- 70
~; ~Mean = &63
~j~/ Std Dev= 0.72
t\ ~I
t\ · 1
\\ ~1
~'
r
_
Ve y Good I Good I Fair I Poor , Very Poor
New Category
Figure 16. Probabilibr plot for pavements w~ a previous
condition of Fair.
VeJy Good ' Good I Fair I
New Category
,,~,.
, Rehabilitated
199~;96, QA- 90
Mean = ~12
~/ Std Dev= 0.66
.
\ 1991-92, QA = 70
~ Mean = ~50
/ Std D~= 1.35
I '
~ '
Poor ' V~v Poor , Rehabditated
Figure 17. Probability plot for pavements wi~ a previous
condition of Poor.
108
OCR for page 109
1.0,
0.~
oft
~ ~7
EN o.e
Cal
o 05
I:
o
3 0.4
O 3
-
02
Q1
TO '
Very Good; Good
199~961 QA = 90
Mean = 0.94
Std Dev= 1.61 ;
1
L'.~' 111~
Poor Vey Poor
New Categoly
Figure IS. Probability plot for pavements win a previous
condition of Very Poor.
.
.
\ 199i-92, QA= 70
~ Mean = 3.20
/ Std Dev= 1.37
Rehabilitated
Figure 14 (previous condition Very Good)- This figure shows Cat Me 1995-96
pavements are actually increasing in condition, whereas the 1991-92
pavements are remaining We same. Although logic and experience tell us
Cat pavements do not Increase In condition win dine, proactive maintenance
practices can have a positive effect on certain pavement management
condition indicators In the higher pavement condition categories. For
example, a sealed crack is often not rated as severe as an unsealed crack, or a
surface treahnent may hide or mask over small distresses.
No comparison of Me foDow~ng year's pavement distribution can be made
for this category. The 1991-92 category contains only one pavement section
of 3.! mi (5.0 km). Therefore, Me distribution for the following year exactly
mirrors Cat section's performance. More pavements in this category are
required to support Me results.
Figure 15 (previous condition Good) This figure shows Cat Me 1995-96
pavements have a higher probability of remaining in the Good category, and
Me 1991-92 pavements have a higher probability of moving to a lower
category. This is further supported by Me distribution analysis. The
distribution of Me 1995-96 pavements shows a higher mean condition rating
value and a larger area between the two curves in the higher pavement
category, although Me difference in the area between Me distribution curves
i09
OCR for page 110
is rawer small. A more complete comparison could be made if Me actual
conditions of the rehabilitated pavements were known prior to We
application of the rehabilitation activity. Even Cough the rehabilitated
pavements represent only a small percentage (7 percent) in this category,
Hey may still affect He results of He analysis.
Figure 16 (previous condition Fair) The results of this figure are
inconclusive. Over half (53 percent) of He 1995-96 pavements were
rehabilitatecl. When such a significant percentage of an individual category
Is rehabilitated In He same year, it is Impossible to determine which
condition category He majority of He pavements would progress to if He
rehabilitation was not appliecl. The only way to make a complete
comparison would be to record the actual condition of He rehabilitated
pavements prior to the rehabilitation activity.
· Figure 17 (previous condition Poor) Even Cough a substantial percentage
(39 percent) of He 1995-96 pavements were rehabilitated, He pavements win
He higher LOS rating have a higher probability of remaining In He same
category and not deteriorating to a lower category, even if ah of He
rehabilitated pavements fell to He Very Poor category. However, just like He
Fair category, the only way to make a complete and accurate comparison
would be to know He actual condition of the rehabilitated pavements prior
to He application of He rehabilitation activity.
· Figure 18 (previous condition Very Poor - This figure seems to show He
greatest difference of all of the figures. Ahnost one-third (29 percent)
pavements win the higher LOS raking moved from He Very Poor category to
He Poor category, compared to 2 percent of the pavements win the lower
LOS. This shows that a higher LOS can keen pavements in He backlog at a
~ . ~
higher condition until Hey can be rehabilitated. This figure also shows He
effects of reactive maintenance activities performed on the pavement sections
in this example.
Discussion of Resulls
Upon reviewing the results of He Florida example, it was difficult to determine if
He difference In He 1991 and 1995 LOS ratings substantially affected He rate at which
He pavements in He analysis group progressed toward a lower pavement condition
and ultimately became part of He backlog. A number of factors confounded He
analysis, He most obvious and important of which are described below.
Analvsis croup size The analysis croup used In He example represented only a
~~~~~~~ o--~r ~ a--- Cal - r ~ ~ - -a --
.. ~ . ~ . ,^ ~^ · r-,r, ~ ~ ~ ~~~ ~ ^^r~ ~ r,m, ~ ~
small sunset or pavements t~ m1 LocU Km~ In Hi and J~o nu Lozo KmJ In
1995~. When a small analysis group is used, the amount of rehabilitation and He
110
OCR for page 111
total amount of pavement in each individual category can affect Me analysis
results.
In the example, the amount of rehabilitation greatly affected the analysis. A
substantial percentage of me pavements in certain pavement categories were
rehabilitated in 1995. When a substantial percentage of pavements in a category
are rehabilitated, it becomes virtually impossible to analyze the pavement
distribution of that category for the next year, unless the conditions of these
pavements are known immediately before rehabilitation.
Likewise, when Me analysis group is small, Me amount of pavements In an
individual category can be very small, or a given category may not even contain
pavements. In this example, certain categories contained a very small amount of
pavement miles. In these cases, unusual performance of one pavement can skew
Me results of Me entire category analysis.
Level of maintenance during the years between 1991 and 1995-The LOS ratings
used n this example were for 1991 and 1995. These years were chosen because
Hey exhibited a large difference in Me LOS ratings. However, Me pavement
condition and He backlog accrual rate is not only affected by He current year's
maintenance, but the maintenance during the past years as well. By selecting
non-consecutive years, no consideration was given to He level of maintenance
during He years between 1991 and 1995. To asses He Hue accuracy of He
backlog analysis, a year-to-year review of the LOS ratings should be performed
and incorporated into He results.
Missing or inaccurate pavement condition data The analysis group for this
example was not only chosen because of He large difference In LOS, but also
because almost ad highway segments in He EMS database contained pavement
condition data. However, some sections did not have pavement condition data
for 1991 or 1995, or He pavement condition ratings were not updated to reflect
recent rehabilitation projects. In an attempt to mistune He effects of these
factors, sections having these characteristics were eliminated from the analysis.
This resulted in a decrease in the size of the analysis group and, consequently,
magnified He problems Cat result from using a small analysis group.
· Reactive or proactive maintenance approach The maintenance approach
selected by an agency will greatly affect the categories in which an increase in
pavement performance characteristics may occur. If an agency performs
proactive maintenance and is performing He maintenance activities before
substantial visual distress occurs, the results may be seen in aD condition
categories, but He most noticeable affects win be in He higher pavement
· ~. -
~ AL
condition categories. lt an agency performs reactive maintenance, the results
win not be seen unto He middle to lower pavement condition categories, where
visual distress is evident.
OCR for page 112
Change in Condition Indicator vs. Time
The following section describes the data collection and analysis work conducted to
evaluate We proficiency and usefulness of this method as a way of tracking Me
relationship between QA program outputs and long-term performance of pavement
facilities. The formal methodology for the Change in Condition Indicator vs. Time method
is presented in appendix C of Me QA program Implementation Manual.
Data Collection and Analysis
To evaluate Me Change in Condition Indicator vs. Time method, comprehensive
pavement condition data and maintenance quality retina data were obtained from the
Iowa DOT. The pavement condition data were downloaded from Me Iowa EMS
database and included annual PCI ratings and surface roughness measurements for all
interstate and primary route pavement sections for Me years 1988 through 1993. The
LOS data were obtained from Me 1993 maintenance quality evaluation report (Iowa
DOT, 1993) and consisted of annual pavement surface element LOS ratings for all
interstate and primary highways in Iowa for fiscal years 1983 through 1993.
The Change in Condition Indicator vs. Time method was tested using the data collected
from Iowa and the seven procedural steps described in the Implementation Manual. The
application of each step and its results are given below.
Step 1. Determine the Analysis Group.
The analysis group for this example was limited to all primary roads and
nterstates, of all pavement types, In Me State of Iowa.
Step 2. Compile Table of Yearly Condition Indicator Data Representing the Chosen Analysis
Group.
Yearly PCI data (1988 Trough 1993) were obtained from Me State's PMS
database. The data were matched (year-to-year) based on knowing each section's
pavement type, cons~uchon year, and section length.
Step 3. Delennine the Yearly Deduct Values.
Yearly deduct values were obtained for Me analysis group using Me PMlool
computer software. PMTool is a proprietary software Cat first calculates yearly
deduct values for individual sections based on consecutive years of data, Men
determines the overall deduct value (representing Me data) as an average of Me
individual section deduct values.
~2
OCR for page 113
Both the two-point and three-point analysis approaches (described in appendix
C of QA program Implementation Manual) were used to calculate yearly deduct
values. The results of both approaches are su~runar~zed in table 21.
Step 4. Plot of Condition Indicator Data Versus Age (optional).
A plot of condition indicator vs. age was not constructed for this example
because Me analysis group contained many different pavement types.
Step 5. Summarize LOS Rating Data Representing the Chosen Analysis Group.
The annual roadway element LOS rating data for all of We highway sections
included in Me analysis group are summarized In table 22.
Step 6. Plots of Yearly Deduct Curves for Different Average LOS Rahugs (optional).
The step 6 option was not exercised in this example.
Step 7. Plots of LOS Rating and Yearly Deduct Values Versus Time.
Figure 19 contains the resulting plots of LOS ratings and yearly deduct values
vs. time for Me Iowa highway sections. As can be seen, Me trend of LOS over time
generally mimics the trend of PCI deduct values over time. Beginning in 1989 and
ending in 1992, maintenance ratings dropped from 82.2 to 69.8, which corresponded
with steady increases in the PCI deduct values (i.e., higher negative values). Then, a
slight increase in maintenance quality between 1992 and 1993 was accompanied by
a much sharper decrease In Me deduct values (i.e., lower negative values).
Table 21. Yearly deduct values calculated using the two- and three-point analyses.
Analysis Type
Tw+Point Analysis 1989
1990
Three-Point Analysis
1992
1993
1990
1991
1992
1993
Years
Included in Calculated Deduct
~Villa
1991
1989-1990
1990-1991
-1.51
~991-1992
1992-1993
1988-1990
1989-1991
1990-1992
1991-1993
OCR for page 114
Table 22. Yearly maintenance LOS ratings for Iowa pavements.
1989
1990
90.0
80.0
70.0
m.0
· -
50.0-
cn
o
40.0
30~0
20.0
10.0
1989 1990 1991
Year
Year ! Roadway LOS Rating
100.0 1 '
. ~...
·. t ~
i · ~
t s
; -___
. ~
· ~
. . .
i j ~
. , ~
. .
. .. .
. .
·
_-,.
. ~
i a
t .
., ,
. . :
r
., .
. ·,
. .
.
.
. . ~
. . .
.................. - ~_ _ ~_ ~. ~. ~.............
I I ~ Tw>Point Analysis |
- ~ Three Point Analysis
. ~
-* - LOB Ratings
_.
-
i
1992 1993
6.00
. 400
In
0.00 ~
en
~5
- 400
- ~.00
Figure 19. Plot of LOS rating and PCI deduct values vs. time for Iowa
Interstate and primary roads, ad pavement herpes.
~4
OCR for page 115
Discussion of Resulls
Although the maintenance quality and PCI deduct value trends are fairly similar,
Me effects of over factors, such as highway funding allocations, traffic, pavement
design, and materials, can be sensed in this evaluation. For instance, in 1990,
maintenance quality increased slightly, whereas the PCI deduct increased considerably.
Also, in 1993, a small increase in maintenance quality was accompanied by a large
decrease in the PCI deduct. Although engineers largely involved win Me performance
of pavements in Iowa during this period will have a much better understanding for the
noted occurrences, one possible scenario is that a large number of pavements reached
Me steepest part of Me pavement deterioration curve from 1989 to 1992, and Mat
several pavements were rehabilitated in 1992 as a result of increased funding or
reallocation of funds. Maintenance may have only been able to keep quality up until
1990, at which time We same level of maintenance funding, or even slightly increased
funding, was insufficient to keep conditions up. Maintenance quality may have then
rebounded In 1993 due to Me infusion of "new" pavements into Me network.
Interpretation, Appraisal, and Applications
Presented in this chapter were sac possible me~ods for tracking the relationship
between maintenance LOS and the long-term performance of highway facilities. Two
of these six me~ods, Me Backlog Analysis memos and the Change in Condition Indicator
vs. Time method, were selected for detailed evaluation to determine Weir overall
effectiveness and practicality as tracking melons. The evaluations were performed in
accordance with the procedures set forth for each memos, using actual maintenance
LOS and pavement condition data collected from two SHAs.
The evaluation results were, at best, fairly supportive of Me effectiveness of each
methodology. In bow cases, it was quite clear Mat not being able to eliminate
. ~. -
confounding factors Is a serious drawback. Significant changes or differences in
~ v
funding, design, materials, tragic, climate, and rehabilitation criteria can cloud the
relationship between maintenance quality and long-term performance. Perhaps more
importantly, these factors can offset one another, leading one to wrongly interpret the
maintenance quality-performance relationship.
From a practicality standpoint, bow methods were relatively simple and
straightforward. It is believed that infrastructure management personnel at both State
and local highway agencies would be very capable of applying Me Backlog Analysis
method to the condition data Hey have collected and stored. Likewise, it Is believed
Cat infrastructure management personnel at more advanced State and local agencies
would have very few problems following He Change in Condition Indicator vs. Time
methodology. Many pavement management groups at He State level have developed
and routinely update pavement performance models, and are therefore very familiar
win He rates of pavement deterioration unique to Heir network.
~5
1
OCR for page 116
In light of Me above appraisals, it is recomanended that bow We Backlog Analysis
and Change in Condition Indicator vs. Time me~ods be viewed as precursory evaluation
me~ods. Each is capable of providing an Axial indication of the maintenance
quality-performance relationship, but each should be supported by a more reliable
methodology (one similar to the Regression Mode! approach discussed earlier in this
chapters that accurately accounts for Me effects of funding allocation, design variables,
traffic, climate, materials, and over factors.
Interested agencies must be advised to recogruze me differences between Me
purposes for and Me inferences to be derived from information collected for PMS and
LOS rating systems. These differences are as follows:
· PMS data generally tends to quantify the remaining life of specific pavement
sections or projects. The impacts of age, design parameters, materials, traffic,
weaver, geography, construction and maintenance techniques each contribute to
Me total life expectancy of pavements. Although Me purpose of most
maintenance treatments is to prolong Me life or overwise slow deterioration, it
cannot cost-effectively hak the eventual need for rehabilitation.
LOS rating clata generally provide an indication of how well the maintenance
operation within an agency is performing those activities under its span of
control. Although it Is reasonable to assume a good maintenance program can
indeed slow the deterioration of a properly designed and constructed pavement,
factors beyond Me control of most maintenance orgaruzations Innit the ultimate
time before a pavement reaches Me point of requiring rehabilitation.
~ closing, comparisons of PMS and LOS data were conducted in this study with Me
idea of identifying possible linkages between highway maintenance quality programs
and over management information systems. Each agency, depending on its
circumstances, should carefully consider Me value to be received when attempting to
use over management information system data in lieu of a program specifically
designed to identify Me LOS being provided by its maintenance operation.
T%% ran ~T ~ ~ ~,
116
Representative terms from entire chapter:
condition indicator
