Journal of Civil Engineering and Architecture 11 (2017) 701-714

doi: 10.17265/1934-7359/2017.07.007
D DAVID PUBLISHING

Effect of Sample Unit Size on Visually Examining

Pavement Condition for Asphalt-Surfaced Roads

Bishnu Prasad Devkota

Nepal Engineering College, Pokhara University, Kathmandu 44600, Nepal

Abstract: Road surface condition evaluation involves the collection of data over pavement surface for different types of distresses. The
exercise consumes a lot of resources if the whole road section length is surveyed and may be prone to errors as a result of surveyors’
fatigue. It is therefore important to develop a representative sample to be used when evaluating road condition manually. This study
aimed at determining an adequate sample size for section level as well as a way forward for network level condition evaluation of
highways in Nepal. Again the study was conducted to quantify the effects of altering the sample unit size for performing a distress
survey according to the PCI (pavement condition index) and SDI (surface distress index) method separately for asphalt surfaced roads.
The effect of reducing/increasing sample unit size was investigated adopting visual examination through field survey by eight teams in
July, 2015, along the section of Banepa-Bardibas highway. The PCI was then calculated for each sample unit using standard deduct
curves and PCI calculation methodology as per SHRP (Strategic Highway Research Program) recommendations and the computation
of SDI was done as per DoR (Department of Roads) guidelines. The results show that 13% sample unit are needed for SDI and 21% for
PCI computation, however, the results are out of the significant level. This is higher than DoR and SHRP guidelines. Again no strong
relationship is observed between SDI and PCI values.

Key words: Pavement condition evaluation, PCI, SDI, sample size, policy implications.

1. Introduction assessment, it is important to ensure that the data

collected and used is accurate enough to provide the
1.1 Background
desired level of support. However, since the collection
Pavement condition evaluation is one of the of condition data is the most expensive portion of
important components of pavement design, maintaining the PMS, the cost must be matched to the
rehabilitation and management which include resources and needs of the adopting agency [2].
evaluation of distress, roughness, friction and structure. Several new nondestructive technologies have been
Most of the cost effective M&R (maintenance and developed and applied in collecting raw condition data
rehabilitation) strategies developed using PMS and processing them to produce useful condition input
(pavement management system) is due to accurate to infrastructure IM&R (inspection, maintenance, and
pavement evaluation [1]. Pavement condition rehabilitation) decision making aimed at minimizing
information is used to evaluate the current condition, expected total life-cycle cost [3]. Such advances
determine rate of deterioration, project future condition, initially motivated the quantification of condition
determine M&R needs, and determine the costs to measurement uncertainty and the incorporation of this
repair pavement segments. It is also used to establish uncertainty in decision making. Following this
M&R strategies and is often used to help prioritize development, the spatial variation of condition has
M&R fund expenditures [2]. Since so many decisions been quantified and has led to the recent extension of
supported by the PMS are based on the condition decision-making methods to take into account
sampling uncertainty and determine the optimal sample
Corresponding author: Bishnu Prasad Devkota, M.Sc.,
research fields: traffic and transportation engineering. E-mail:
size, along with the other IM&R activities [3].
[email protected].
702 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

The evaluation of the contributions of the condition  To develop the relationship between SDI and PCI
sampling-related advances to improved decision indices;
making is presented by Mishalani and Gong [4]. The  To compare the maintenance strategy as
results of the application of this evaluation recommended by SDI and PCI.
methodology indicate that the magnitudes of the value
2. Literature Review
of the condition-sampling advances of interest are
found to be appreciable in both expected total life-cycle For the time being, prioritizing in the selection of the
cost and IM&R agency cost. roads (or resealing) will be limited to the consideration
The PCI (pavement condition index) and SDI of the four parameters namely road age, visual survey
(surface distress index) are the numerical indicator that ratings, traffic and strategic importance [8].
rates the surface condition of the pavement through It is recommended that a section sample of 20 m
visual examination. These indicators provide a long from the beginning of a 100 m section be used in
measure of the present condition of the pavement based evaluating the pavement surface condition of such
on the distress observed on the surface of the pavement roads. This will result in a reasonably accurate
which also indicates the structural integrity and surface representation of the condition of the whole section
operational condition (localized roughness and safety). with huge savings in resources [9]. Comparisons were
However, these indicators cannot measure the made between PCI values calculated using standard
structural capacity; neither they provide direct PCI procedures (19 distress types) and PCI values
measurement of skid resistance or roughness. They calculated using modified distress identification
provide an objective and rational basis for determining procedures developed by the Metropolitan
maintenance and repair needs and priorities [5, 6]. Transportation Commission (seven distress types) [10]
Continuous monitoring of the PCI is used to establish The study area in Ref. [11] consists of 10 urban road
the rate of pavement deterioration, which permits early sections constituting 29.92 km of Noida city. The
identifying of major rehabilitation needs. The PCI methodology includes identification of urban road
provides feedback on pavement performance for sections, pavement distress data collection,
validation or improvement of current pavement design development of individual distress index and finally
and maintenance procedures [5]. Surface distress developing a combined OPCI (Overall Pavement
surveys of the strategic network have been undertaken Condition Index) for the network. The four
annually by planning branch since fiscal year performance indices, namely, pavement condition
1992-1993 [7]. The survey was interrupted for some distress index (PCI distress), pavement condition
year and it is continued form fiscal year 2012-2013. roughness index (PCI roughness), pavement condition
structural capacity index (PCI structure) and pavement
1.2 Objectives and Scope
condition skid resistance index (PCI skid) are
The main objective of this study is to assess the developed individually. Then all these indices are
effect of sample unit size on pavement condition index combined together to form an OPCI giving importance
for asphalt-surfaced roads. Specifically following of each indicator. The proposed index is expected to be
objectives are set out: a good indicative of pavement condition and
 To evaluate the pavement section for SDI and PCI performance. The developed OPCI was used to select
values; the maintenance strategy for the pavement section [11].
 To recommend the suitable sample size for the Pavement condition has been known as a key factor
evaluation of pavement; related to ride quality, but it is less clear how pavement
 Effec
ct of Sample Unit Size on Visually
V Exam
mining Pavem
ment Conditio
on for Aspha
alt-Surfaced Roads
R 7033

conditions are
a related to traffic crashees. The resultts in indiicate that by not adddressing th he samplingg
Ref. [12] suggested
s that poor pavvement condiition unccertainty andd decisions,, the optim mum IM&R R
scores and ratings
r were associated
a wiith proportionnally decisions wouldd not be achhieved, and consequently,
c ,
more severre crashes, but very poor pavem ment marrked unnecesssary overspennding could taake place [4]..
conditions were
w actuallyy associated with less severe The
T effect off sample sizze on PCI accuracy a wass
crashes. Verry good pavem ment conditioons might indduce inveestigated for asphalt roaddways [10] by b employingg
speeding beehaviors and therefore coould have cauused the 35-mm film m automatedd distress data collectionn
more severre crashes, especially on non-freeeway tech
hnique. Tweenty four aspphalt pavem ment sectionss
arterials annd during faavorable drivving conditiions. werre surveyed. Fig. 1 show ws a plot betw ween relativee
These resullts provide insights on how pavem ment sammple unit sizee for regular ssize and expeected amountt
m have conttributed to craashes, which may
conditions may of error
e in the PCCI. Fig. 2 shoows a comparison betweenn
be valuable for safety im mprovement during pavem ment PCII calculated using
u 10% reggular sample size and PCII
design and maintenancee. Although the study foound calcculated using a full road seection. As lonng as the sizee
statistically significant effects
e of pavvement variaables is within
w 40% from the regulaar size, from Figs. 1 and 2
on crash seeverity, the effects
e were rather minoor in the error is limited to about 22%.
reality as sugggested by frrequency anallyses [12]. The
T evaluatioon of the conttributions of the t conditionn
Infrastruccture manageement is the process throough sammpling-relatedd advances to improv ved decisionn
which IM&R R decisions are
a made to minimize
m the total
t mak king is presennted by Mishaalani and Gon ng [4]. In thiss
life-cycle coost. Measurem ment, forecassting, and spatial papper, the methodology iis based on n comparingg
sampling aree three main sources of errors
e introduucing decision-makingg frameworkss that reflect the advancess
uncertainty into
i the proceess. The first two uncertainnties of in
nterest with those
t that do nnot. The basicc idea behindd
are captureed in the infrastructuure managem ment commparing any two framew works is to use each too
literature. However,
H thhe third onee has not been b produce optimall IM&R policcies that are based on thee
recognized and quantiified. Ref. [4] presentts a specific assumpttions they refflect and then n to simulatee
methodologyy where the spatial
s sampliing uncertaintty in thesse optimal poolicies withinn the framewo ork reflectingg
question is captured
c and the
t sample sizze is incorporrated the truth with regard
r to cappturing the most
m realisticc
as a decision variable inn an optimizaation. The ressults assuumptions. Thhe results off the applicaation of thiss
Change in PCI

Relative sample uniit size (%)

Fig. 1 Relatiive sample uniit size and expeected amount of

o error in PCII calculation [110].
704
PCI, 10% sample unit size Effec
ct of Sample Unit Size on Visually
V Exam
mining Pavem
ment Conditio
on for Aspha
alt-Surfaced Roads
R

PCI, rregular samplee unit size

parison of the effect

Fig. 2 Comp e of samplle unit size on PCI
P accuracy [10].

evaluation methodology
m indicate thatt the magnituudes  Distress typpe—identifyinng each type of distress;
of the valuee of the conndition-samplling advancees of  Distress seeverity—the llevel of seveerity of eachh
interest are found to be appreciable in both expeected disttress present showing the degree of detterioration off
total life-cyccle cost and IM&R agencyy cost [4]. the pavement;
 Distress exxtent—relativve area affeccted by eachh
3. Data Coollection an
nd Computtation for SDI
S com
mbination of distress
d type aand severity.
and PCI
3.2 SDI (Surfacee Distress Inddex) Survey
3.1 Criteria for
f Site Selecction
There
T are vaarious methoods of colleccting surfacee
The projeect sites were chosen meetiing the follow
wing disttress data annd these incrrease in com mplexity andd
criteria: “Rooads with low
w to medium traffic volum
me in sophistication acccording to tthe quality off informationn
order to avooid accidents and convenieently collect data requuired. The method
m adopteed in this research is thee
and the roaads with as many types of distresses as metthod adopted by DoR (Deepartment of Roads) R and iss
possible [133]. Banepa Bardibas
B (BP
P) highway was a simplified proocedure recoommended by y the Worldd
selected for the study. Thhe total lengthh of the highhway Bannk which hass been modiffied to suit the t particularr
is about 2066 km and thee carriagewayy width is 5.55 m, connditions in Neepal and the needs of DoR. The SDI is a
with unpavved shoulderrs. Traffic is a hazardd as six--level rating inndex from 0 tto 5. The ratin
ng 0 indicatess
inspectors may
m walk on the pavemennt to perform
m the a pavement
p surrface withoutt any defectss, whereas a
condition suurvey. Accuraate, consistennt, and repeattable ratinng of 5 inndicates for the maximu um possiblee
distress evalluation surveyys can be perrformed by using
u deteerioration. A shoulderr condition survey iss
the Distress Identificatioon Manual for the Long-T
Term recoommended too carry out at tthe same timee using ratingg
Pavement Peerformance Project
P [13]. Eight
E groups with
w in thhe range 0 too 4. However,, the shoulderr condition iss
two trained engineer in each
e have connducted condiition out of the scoppe of this reesearch. Thee three mostt
surveys for the determiination follow
wing categorrical preddominant typpes of defectt present in each samplee
distresses: secttion are recorrded.
 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads 705

3.2.1 Field Work Plan and Sampling presented in Table 1.

Pavement distresses surveys are carried out These values are based on conditions in Nepal.
manually by the trained engineers using drive and walk Planned maintenance can be carried out on roads in
survey. Surface distress comprises cracking, good/fair condition and rehabilitation or reconstruction
disintegration (potholes), deformation, textural is generally needed for roads in poor condition to bring
efficiency, pavement edge defects and maintenance them to a maintainable state. Similarly, an indicator for
works (patching). These faults are visually assessed different types of pavement remedial action is given by
using a 10% sampling procedure and recorded the percentage of the number of sample section with
using a cumulative index called an SDI. The distress the given SDI values of a particular link as shown in
elements are divided into two groups: major and minor Table 2.
defects. Among the different defect types, cracking,
3.3 PCI (Pavement Condition Index) Survey
raveling and potholes are generally characterized by
extent and severity, while rut depth being continuous in The information was processed to obtain PCI values
nature, only the severity of the deformation is noted. for each road section sample and for whole section. The
The defect types and therefore resulting score are pavement is divided into branches that are divided into
different for bitumen and gravel roads, which are sections. Each section is divided into sample units. The
separately presented in Ref. [6]. type and severity of pavement distress is assessed by
The 10% sampling procedure comprises a walk-over visual inspection of the pavement sample units [5].
survey generally covering the last 100 meter section in Information about date, location, branch, section,
each kilometer of the road on which the SDI is to be sample unit size, slab number and size, distress types,
determined [6, 7]. The full width of pavement is to be severity levels, quantities, and names of surveyors,
evaluated for each sample of length 100 m. were recorded on data sheets [14]. Again the
3.2.2 SDI Data Use instruments used are measuring tape (30.0 m and 3.0 m
The SDI is averaged over each road link or section length with 2 mm and 1 mm least count) straightedge,
under consideration. The results can be used to provide scale (300 mm).
the objective assessment of the pavement condition and 3.3.1 Field Work Plan and Sampling
to indicate the need for periodic maintenance, First, divide the pavement sections into sample units.
rehabilitation or reconstruction. For assessing Individual sample units to be inspected should be
pavement condition, the terms “Good”, “Fair” and marked or identified in a manner to allow inspectors
“Poor” are used based on averaged values of SDI as and quality control personnel to easily locate them on

Table 1 SDI chart for pavement condition in Nepal [6].

SDI values Condition
0~1.7 Good
1.8~3.0 Fair
3.1~5.0 Poor

Table 2 Pavement remedial action based on SDI values in Nepal [6].

Percentage SDI values Action
20%, SDI = 5 Reconstruction
10%~30%, SDI = 4 Rehabilitation
20%~30%, SDI = 3 Resealing with local patching
20%~30%, SDI = 2 Resealing only
706 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

Fig. 3 PCI rating system [15].

the pavement surface. The minimum number of sample unit to another within the section;
units (n) that must be surveyed within a given section N = total number of sample units in the section.
to obtain a statistically adequate estimate (95% If obtaining the 95 % confidence level is critical, the
confidence) of the PCI of the section is calculated using adequacy of the number of sample units surveyed must
Eq. (1), the following formula and rounding n to the be confirmed. The number of sample units was
next highest whole number [5]: estimated based on an assumed standard deviation. The
n = Ns2/[(e2/4)(N – 1) + s2] (1) actual standard deviation (s) can be calculated as
where: follows (Eq. (2)):
e = acceptable error in estimating the section PCI;
S =  i =1 ( PCI i − PCI s ) 2 / (n − 1)1/2
n
(2)
commonly, e = 65 PCI points;
s = standard deviation of the PCI from one sample where:
 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads 707

PCIi = PCI of surveyed sample units i; number of occurrences, depending on the distress type.
PCIs = PCI of section (mean PCI of surveyed sample Divide the total quantity of each distress type at each
units); severity level by the total area of the sample unit and
n = total number of sample units surveyed. multiply by 100 to obtain the percent density of each
The revised minimum number of sample units distress type and severity. Determine the DV (deduct
should be calculated (Eq. (1)) which is to be surveyed value) for each distress type and severity level
using the calculated standard deviation (Eq. (2)). If the combination from the distress deduct value curves.
revised number of sample units to be surveyed is Determine the maximum CDV (corrected deduct value).
greater than the number of sample units already The following procedure must be used to determine the
surveyed, select and survey additional random sample maximum CDV [5]:
units. These sample units should be spaced evenly  If none or only one individual deduct value is
across the section. Repeat the process of checking the greater than two, the total value is used in place of the
revised number of sample units and surveying maximum CDV in determining the PCI; otherwise,
additional random sample units until the total number maximum CDV must be determined;
of sample units surveyed equals or exceeds the  List the individual deduct values in descending
minimum required sample units (n) in Eq. (1), using the order. Determine the allowable number of deducts, m,
actual total sample standard deviation. Once the using the formula (Eq. (4)):
number of sample units to be inspected has been m = 1 + (9/98)(100 – HDV) ≤ 10 (4)
determined, compute the spacing interval of the units where:
using systematic random sampling. Samples are spaced m = allowable number of deducts including fractions
equally throughout the section with the first sample (≤ 10); and
selected at random. The spacing interval (i) of the units HDV = highest individual deduct value;
to be sampled is calculated by the following formula  The number of individual deduct values is
(Eq. (3)) rounded to the next lowest whole number: reduced to the m largest deduct values, including the
i = N/n (3) fractional part. For:
where: (1) Determine maximum CDV iteratively;
N = total number of sample units in the section; and (2) Determine total deduct value by summing
n = number of sample units to be inspected. individual deduct values. The total deduct value is
The first sample unit to be inspected is selected from obtained by adding the individual deduct values;
sample units 1. The sample units within a section that (3) Determine q as the number of deducts with a
are successive increments of the interval i after the first value greater than 2.0;
selected unit also are inspected. (4) Determine the CDV from total deduct value and q
Additional sample units only are to be inspected by looking up the appropriate correction curve for AC
when non-representative distresses are observed. These pavements;
sample units are selected by the user. (5) Reduce the smallest individual deduct value
3.3.2 Computation of PCI greater than 2.0 and repeat above steps until q = 1;
The total quantities of each distress type are added (6) Maximum CDV is the largest of the CDVs.
for at each severity level, and recorded in the “Total  PCI calculation by subtracting the maximum CDV
Severities”. The units for the quantities may be either in from 100 (Eq. (5)):
square feet (square meters), linear feet (meters), or PCI = 100 − maxCDV (5)
708 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

 ( PCI
m n
ri * Ari ) PCI r ( A −  Ari ) + PCI a ( (* Aai )
PCI S = PCI r = i =1
n
(6) PCI s = i =1 i =1
(8)
A
A
i =1
ri
where;
where: A = area of section;
m = number of additional sample units surveyed; and
PCI r = area weighted PCI of randomly surveyed
PCIs = area weighted PCI of the pavement section.
sample units; The overall condition rating of the section should be
PCIri = PCI of random sample unit i; determined by using the section PCI and the condition
Ari = area of random sample unit i; rating [5].
n = number of random sample units surveyed.
4. Results and Discussions
If there is no additional sample, then PCI of the
section is given by Eq 6 but if additional sample units Following are the summary of the data that are
are surveyed, the area weighted PCI of the surveyed collected from the field.
additional units (PCIa) is calculated using Eq. (7). The Table 3 shows for SDI values whether Table 4
PCI of the pavement section is calculated using Eq. (8). shows for PCI values. As per DoR guidelines, the SDI
n value of the section based on Table 3 is 2.3. Similarly,
 ( PCI ai * Aai ) the PCI value from Table 4 based on SHRP (Strategic
PCI a = i =1
n (7) Highway Research Program) guideline is computed as
A ai
63.4. Based on Table 1, and the SDI value 2.3, the
i =1 pavement condition is “Fair”. Again based on PCI
where: value of 63.4 and from Fig. 3, the pavement condition
PCIa = area weighted PCI of additional sample units; is “Fair”. Hence, it is concluded that the findings of
PCIai = PCI of additional sample unit i; both the system (SDI and PCI) are same, i.e., pavement
Aai = area of additional sample unit i. is in “Fair Condition”.

Table 3 SDI of each of the sample.

Chainage (m) Chainage (km)
SDI for
Sample No. from To 9+ 10+ 11+ 12+ 13+ 14+ 15+ 16+ 17+
sample unit
Area ==> 550 550 550 550 550 550 550 550 550
1 0 100 2 3 3 2 2 3 3 2 2 2.444
2 100 200 3 2 2 2 2 2 2 2 2 2.111
3 200 300 3 2 2 3 3 3 3 2 2 2.556
4 300 400 2 3 3 2 2 3 3 1 1 2.222
5 400 500 3 3 2 2 3 2 3 2 1 2.333
6 500 600 2 2 2 2 2 2 2 2 2 2.000
7 600 700 3 2 3 2 3 3 2 2 2 2.444
8 700 800 1 3 3 2 3 3 2 2 3 2.444
9 800 900 3 3 2 2 2 2 2 2 2 2.222
10 900 1,000 5 2 1 2 2 2 2 2 2 2.222
Average SDI of the section 2.300
 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads 709

Table 4 PCI of each sample unit.

Chainage (m) Chainage (km)
PCI for
Sample No. from To 9+ 10+ 11+ 12+ 13+ 14+ 15+ 16+ 17+
sample unit
Area ==> 550 550 550 550 550 550 550 550 550
1 0 100 76 50 81 58 87 37 2 58 65 57.111
2 100 200 64 60 56 39 70 78 69 33 42 56.778
3 200 300 71 71 86 74 86 35 53 56 58 65.556
4 300 400 78 56.1 78 92 92 45.9 57.5 45.9 69.5 68.322
5 400 500 60 60 44 40 91 65 16 74 100 61.111
6 500 600 27 88 89 26 94 30 25 62 32 52.556
7 600 700 56 79 59 70 88 35 78 92 69 69.556
8 700 800 85 62 79 67 71 60 51 85 63 69.222
9 800 900 60 60 91 68 96 58 80 74 45 70.222
10 900 1,000 22 62 90 77 73 84 59 60 62 65.444
Average PCI of the section 63.381

Cumulative Mean of SDI Values

2.60
2.55
2.50
2.45
SDI value

2.40
2.35
2.30
2.25
2.20
0 10 20 30 40 50 60 70 80 90 100
Number of sample units

Fig. 4 Cumulative mean value of SDI over number of sample units.

pavement condition data was found best fit on

4.1 Effect of Sample Size on Error and Precision
logarithmic equation R2 value = 72% (Fig. 6 and 7). For
According to SHRP recommendations, the PCI linear equation R2 value = 55%, for polynomial R2
values should lie within ±5%, however, based on DoR value = 67%, power and exponential are not feasible
recommendations, there is no criteria for the validation for these trendline. In other words, the error decreases
of the data observed from field for SDI based on DoR exponentially at lower sample size and the rate of error
recommendations. So based on ±5% permissible error reduction will decline for higher sample size. So, it can
and from Figs. 4 and 5, the recommended sample size be said that the taking 100% sample for the pavement
for SDI is 13% and that for PCI is 21%. condition evaluation is the loss of time and money. So
The effect of sample size over the accuracy on the selection of accurate sample size is very much important
710 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

Cumulative mean of PCI values

70.00
69.00
68.00
67.00
PCI value

66.00
65.00
64.00
63.00
62.00
61.00
60.00
0 10 20 30 40 50 60 70 80 90 100
Number of sample units

Fig. 5 Cumulative mean value of SDI over number of sample units.

Relative sample unit size and expected amount of error

on SDI calculatioin
0.16
y = -0.05ln(x) - 0.004
0.14
R² = 0.720
Absolute amount of error

0.12
0.10
0.08
0.06
0.04
0.02
0.00
-0.02 0% 20% 40% 60% 80% 100% 120% 140%
Sample size adopted

Fig. 6 Relative sample unit size and expected amount of error in SDI calculation.
 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads 711

Relative sample unit size and expected amount of

error on PCI calculatioin
10.00
y = -2.94ln(x) + 0.368
R² = 0.837
Absolute amount of error

8.00

6.00

4.00

2.00

0.00
0% 20% 40% 60% 80% 100% 120% 140%
-2.00
Sample size adopted

Fig. 7 Relative sample unit size and expected amount of error in PCI calculation.

Comparison of effect of sample unit size on SDI

accuracy
5
SDI, 10% sample unit size

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
SDI, all population samples

Fig. 8 Comparison of effect of sample unit size on SDI accuracy.

712 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

Comparison of effect of sample unit size on PCI

accuracy
100
90
PCI, 10% sample unit size

80
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90
PCI, regular sample unit size

Fig. 9 Comparison of effect of sample unit size on PCI accuracy.

PCI and SDI relationship

100
90
y = -9.477x + 85.38
80
R² = 0.084
70
PCI Values

60
50
40
30
20
10
0
0 1 2 3 4 5
SDI Values

Fig. 10 SDI & PCI relationship in linear pattern.

that it should be representative of the road section. 4.2 Relationship between SDI and PCI
These results are also supported by Fig. 1 [10]. It is tried to develop the relationship between SDI
Figs. 8 and 9 show the comparison between 10% and PCI based on the same data. The single and
sample data and the whole population. The results multiple regressions is analyzed for the development of
show that the 10% sample will not be the best fit model. The power, exponential and logarithmic
representative of the population. Similar results are models are also being tested along with linear and
also supported by Fig. 2 [10]. curvilinear pattern of the database system. Fig. 10
 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads 713

shows the linear relation between SDI and PCI value Acknowledgments
with R2 value of 8%, which means that the model is
The author would like to express sincere thanks to
only 8% reliable. The polynomial equation of second
Assoc. Prof. Dr. Thusitha Chandani Shahi for
degree is found same level of goodness. So it is
continuous motivation and guidance during
concluded that the not strong relationship is found
preparation of this paper. The author also would like to
between SDI and PCI.
thank the following research groups for collecting data
5. Conclusion and Recommendations from various samples of the highway section: Aashish
Kandel, Ajay Kumar Gupta, Anga Lal Rokaya,
The value of SDI and PCI is computed for the
Bhaskar Chataut, Bhim Bahadur Maharjan, Bidur
section of Banepa-Bardibas highway and the SDI value
Chaulagain, Bijaya Rana, Gopal Gautam, Indra
as per DoR guidelines is computed as 2.3 and that for
Tamang, Madhav Prasad Adhikari, Pradeep Prasad
PCI based on SHRP guideline is computed as 63.4.
Bhatt, Pushkar Poudel, Rabi Pokhrel, Shailendra Malla,
Based on both SDI and PCI, the recommended
Subash Pyakurel, Suddhumna Hamal, Sunil Kumar
condition of the pavement condition is the same which
Tilak, and Ujwal Shrestha.
is in “Fair Condition”. For SDI = 2.3, resealing with
local patching is recommended as the M&R technique, References
however, as per PCI, the recommended techniques are [1] Huang, Y. H. 1993. Pavement Analysis and Design. New
based on distress types and the probable causes of Jersy, USA: Prentice-Hall, Inc.
distresses. The comparison between sample data and [2] FHWA (Federal Highway Administration). 1995.
Pavement and Road Surface Management for Local
the population data shows that the 10% sample will not
Agencies. Washington D.C.: Federal Highway
be the representative of the population. Again, the Administration.
effect of sample size over the accuracy on the [3] Mishalani, R., and Gong, L. 2011. “Evaluating Impact of
pavement condition data was found best fit on Pavement Condition Sampling Advances on Life-Cycle
Management.” Journal of Trasnportation Research
logarithmic equation with R2 value = 72% for SDI and Borard. doi:10.3141/2068-01.
that for PCI is 84%. Hence, it is concluded that the [4] Mishalani, R. G., and Gong, L. 2009. “Optimal
pavement condition evaluation survey for the whole Infrastructure Conidition Samling over Space and Time
for Maintenance Decision-Making under Uncertainty.”
population is the loss of time, labour and money, and
Transport Research (Part B) 43: 311-24.
selection of accurate sample size is very much doi:10.1016/j.trb.2008.07.003.
important such that the sample will be the [5] ASTM (American Society for Testing and Materials).
representative of the population within the permissible 2007. Standard Practice for Roads and Parking Lots
Pavement Condition Index Surveys. ASTM.
precision. Based on ±5% permissible error, the
[6] DoR. 1995. Road Pavement Management. Kathmandu:
recommended sample size for SDI is 13% and that for Maintenance and Rehabilitation Cordination Unit
PCI is 21%. It is concluded that the 10% sample size as (MRCU), Department of Roads, Nepal.
recommended by Ref. [6] seems insufficient, so a [7] DoR. 1995. “Visual Pavement Condition Survey.”
Highway Management Information System (September):
rigorous analysis with higher sample size is
1-3.
recommended in order to revise the national guidelines [8] DoR. 1995. The DoR Strategy. Departmental Policy
for the pavement condition evaluation for SDI survey. Document-Policy Options and Key Measures. p. 11.
Finally, the no strong relation is found between SDI [9] Mfinanga, D. A. 2007. “Sampling Procedure for Pavement
Condition Evaluation of Local Collectors and Access
and PCI. As the samples are limited with similar type Roads.” Tanzania Journal of Engneering and Technology
of failures, a rigorous analysis with higher sample size (TJET) 1 (2): 99-109.
is recommended for further studies. [10] Shahinq, M. Y., Stock, C., Crovetti, M., and Beckberger, L.
714 Effect of Sample Unit Size on Visually Examining Pavement Condition for Asphalt-Surfaced Roads

1995. “Effect of Sample Unit Size and Number of Surveyed Prevention 59: 399-406. doi:10.1016/j.aap.2013.06.028.
Distress Types on Pavement Condition Index for [13] SHRP. 1993. Distress Identification Manual for the
Asphalt Surfaced Roads.” Transportation Research Board, Long-Term Pavement Performance Project. Washington,
60-71. DC: Strategic Highway Research Program.
[11] Shah, Y. U., Jain, S. S., Tiwari, D., and Jain, M. K. 2013. [14] Shahin, M. Y. 2005. Pavement Management for Airports,
“Development of Overall Pavement Condition Index for Roads, and Parking Lots. 2nd ed. New York, 233 Spring
Urban Road Network.” Social and Behavioral Science 104: Street, United States of America: Springer
332-41. doi:10.1016/j.sbspro.2013.11.126. Science+Business Media, LLC.
[12] Li, Y., Liu, C., and Ding, L. 2013. “Impact of Pavement [15] ODoT. 2010. Pavement Distress Survey Manual. Oregon:
Conditions on Crash Severity.” Accident Analysis and Deartment of Transportation.