Sunteți pe pagina 1din 14

Available online at www.sciencedirect.

com

ScienceDirect
International Journal of Pavement Research and Technology 10 (2017) 158–170
www.elsevier.com/locate/IJPRT

Prioritization of pavement maintenance sections using objective


based Analytic Hierarchy Process
Sarfaraz Ahmed ⇑, P. Vedagiri 1, K.V. Krishna Rao 1
Department of Civil Engineering, Indian Institute of Technology Bombay, Powai, Mumbai 400076, India

Received 5 October 2016; received in revised form 31 December 2016; accepted 3 January 2017
Available online 12 January 2017

Abstract

The application of Analytic Hierarchy Process (AHP) method for the prioritization of pavement maintenance sections is widespread
now-a-days. Although the evaluation of pavement maintenance section through AHP method is simple, where the relative importance
(on Saaty’s scale) assigned to each parameter in the hierarchy varies between the experts (transportation professionals) consulted, which
leads to discrepancies in the final rankings of the sections’, due to the subjectivity in the process. Further, experts base their decisions
solely on their experience while consideration is not given to the actual quantitative physical condition of the roads. To overcome these
difficulties an objective based AHP method is proposed in this study, where pairwise comparison values are assigned based on the col-
lected field data from a road network in Mumbai city, consisting of 28 road sections. The final ranking list of candidate sections takes
into consideration the priority weight of alternatives, which reflect the road conditions. The solution of priority ratings of AHP method is
compared with the corresponding solution of road condition index method, a traditional pavement maintenance procedure. The findings
of the present study suggest that objective based AHP method is more suitable for the prioritization of pavement maintenance of roads.
Ó 2017 Chinese Society of Pavement Engineering. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Prioritization; Analytic Hierarchy Process; Road condition index; Objective method; Rating and ranking

1. Introduction system [1]. Prioritization is executed in a sequential manner,


beginning with enlisting of the pavement maintenance
Excessive road deterioration, due in part to improper projects required to be implemented. Based on their
and irregular maintenance, results in increased Vehicle relative perceived urgency of need for repair, engineers and
Operating Costs (VOC), increased number of accidents managers are able to prioritize and schedule the maintenance
and in general reduced reliability of transport services. of pavement sections. To prioritize pavement maintenance
The function of pavement maintenance is to diminish activities, a number of decision making methods have been
pavement deterioration and extend the life of a pavement. introduced and implemented under Pavement Manage-
Pavement maintenance, if not done at appropriate times in ment System (PMS) study. These methods vary from sim-
a preplanned manner, negatively impacts the transport ple ranking to complex optimization [2]. The main
objective of the PMS is to avoid the bias derived from judg-
⇑ Corresponding author. Fax: +91 22 2576 7302. ment and help in the decision-making by using objective
E-mail addresses: 114040011@iitb.ac.in, sarfaraz6215@gmail.com information based on pavement distress and other objec-
(S. Ahmed), vedagiri@civil.iitb.ac.in (P. Vedagiri), kvkrao@civil.iitb.ac.in tive measures. Most of the highway agencies have adopted
(K.V. Krishna Rao). a practice of expressing the pavement maintenance priority
1
Fax: +91 22 2576 7302. in the form of priority index, which is computed by means
Peer review under responsibility of Chinese Society of Pavement
Engineering.
of empirical expression. Although using a mathematical

http://dx.doi.org/10.1016/j.ijprt.2017.01.001
1996-6814/Ó 2017 Chinese Society of Pavement Engineering. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 159

equation is convenient, often they do not have a clear phys- The main aim of the study was to identify an approach
ical meaning and cannot accurately combine different fac- that can reflect the engineering judgment of highway
tors into a single equation. This inevitably leads to agencies and engineers more closely. They examined three
overlooking of various contributing effects of actual char- forms of AHP, namely, the distributive-mode relative
acteristics of distress [3]. Furthermore, not all the factors AHP, the ideal-mode relative AHP and the absolute
and parameters involved can be expressed quantitatively AHP. The study concluded that absolute AHP is most
and measured in compatible units. In view of these short- suitable for the pavement maintenance process. Farhan
comings and constraints, Analytic Hierarchy Process and Fwa [10] continued the same study and applied
(AHP) is the most suitable choice for the prioritization of AHP method for multiple distresses in each pavement seg-
pavement sections for maintenance at network level. ment. In the same study the solution of priority ratings by
Although the evaluation of the pavement maintenance sec- AHP method was compared with the corresponding solu-
tion through AHP method is simple, but the relative tion by the widely adopted PAVER pavement mainte-
importance (on Saaty’s scale) assigned to each parameter nance procedure. Shah et al. [11] proposed two methods
in the hierarchy varies between the experts (transportation for priority ranking of road maintenance; viz-a-viz (a)
professionals) consulted, which leads to discrepancies in subjective rating based rank (b) economic indicator based
the final rankings of the sections. Hence the process can rank. Maintenance Priority Index (MPI), a function of
be termed subjective. Further, experts base their decisions road condition index, traffic volume factor, special factor
solely on their experience while due consideration is not and drainage factor was used to obtain subjective rank-
given to the actual quantitative physical condition of the ing. On the other hand NPV/Cost ratio was calculated
roads. To overcome these difficulties an objective based for each pavement section using HDM-4 software to
AHP method is proposed in this study, where pairwise obtain economic based rank. Prakasan et al. [12] devel-
comparison values are assigned based on the collected field oped priority ranking model for the maintenance of urban
data from a road network in Mumbai city, consisting of 28 roads, using AHP method. They also developed priority
road sections. Furthermore, the evaluated results of prior- ranking model using direct assessment method and com-
ity ratings of objective based AHP method are compared pared the results with AHP model.
with the results of subjective based Road Condition Index It is observed from the existing studies that pavement
(RCI) method for the validation purpose. maintenance prioritization using AHP technique is subjec-
tive in nature. In AHP, experts assign the pairwise compar-
2. Review of existing studies ison to criteria and subcriteria considered in the
prioritization. The criteria of a particular section does
Researchers in the field of pavement management sys- not change during the evaluation of pavement section,
tem have developed various methods for the prioritization but the relative importance (on Saaty’s scale) assigned to
of pavement sections for maintenance, and a few of them each parameter in the hierarchy varies between the experts
are discussed here. Reddy and Veeraragavan [4] developed (transportation professionals) consulted, which leads to
a methodology of priority ranking which involves assign- discrepancies in the final rankings of the sections. Hence
ing a priority index to different sections based on their the process can be termed subjective. To overcome this
overall distress index model and traffic adjustment factors. problem the present study proposes an objective based
AHP is a powerful and established prioritizing tool, a AHP method for the evaluation of prioritization of pave-
mathematical technique, which is used for multi-criteria ment sections for maintenance. In this method, each sec-
decision making to help the decision maker in the selec- tion has been evaluated based on the collected field data.
tion of the best alternatives [5]. Ramadhan et al. [6] The objective of the present study is to assess the effective-
applied AHP technique in order to determine the priority ness of objective based AHP method in determining the
weights of pavement-maintenance factors such as road pavement maintenance prioritization for the selected pave-
type, pavement condition, traffic volume, riding quality ment sections, consisting of multiple distresses. Further-
of pavement, safety, maintenance cost and overall impor- more, the solution of priority ratings by the proposed
tance of the road for the community. AHP has been used AHP is compared with the corresponding solution by the
in Tehran city for 131 sections to determine the priority traditional pavement maintenance procedure, the Road
rating of pavement maintenance while considering model- Condition Index (RCI) method.
ing parameters like road condition index, traffic volume
and road type [7]. Almeida et al. [8] conducted a study 3. Methodology of the study
on prioritization of maintenance of unpaved roads in
the northeast region of Brazil based on AHP method, This study is an attempt to apply objective based AHP
and considered a group of variables that are related to method to establish the relative maintenance priorities of
traffic, climatic, physical, management and social aspects, the pavement sections and builds up on the existing AHP
as modeling parameters. Farhan and Fwa [9] adopted methodologies to study PMS. There is a need to compare
AHP method for the prioritization of pavement activities the effectiveness of this method with current methods. For
of pavement segments having single pavement distress. a pavement evaluation study, the Pavement Condition
160 S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170

Index (PCI) method is considered the most precise index 3.1. Selection of study area
method; it incorporates data from different types of dis-
tresses as well as their severity and quantity [13]. Like- In order to establish Pavement Management System
wise, a condition index called Road Condition Index (PMS), the first step is to identify a network for which
(RCI) has been developed for Indian conditions during the PMS needs to be developed. For this, visual condition
PMS study of Noida city [11]. Hence, in the present study surveys were conducted in different regions of Mumbai city
RCI method is adopted as the reference. The flow chart of in India and a small road network of 73 km is selected. The
study methodology is shown in Fig. 1 and each of these selected road network consists of flexible as well as rigid
components is discussed in detail under its respective pavement. For the present study 28 sections of flexible
section. pavement were selected. These sections were selected after

Selection of Road Network

Surface Distress Data Collection Using Terrestrial Laser Scanner Instrument

Data Extraction in Cyclone Software

Analysis of Prioritization of Pavement Maintenance Activities Using Different Methods

Analytic Hierarchy Process (AHP) Method Road Condition Index (RCI) Method

Assign Judgmental Preference Value of 1 to 9 for Assign Respective Degree and Extent Value of
1 to 5 for each Distress
each Attributes based on collected data

Calculation of Urgency Index for Individual


Calculation of Priority Weight for Criteria
Distress

Assign Weightage Factor Value of 1 to 5 for each


Calculation of Priority Weight for Subcriteria Distress

Determination of Priority Weight of Alternatives Calculation of RCI Value

Priority Ranking of Pavement Sections for


Maintenance

Comparison of Priority Ranking Values


between AHP and RCI Methods

Fig. 1. Framework of study methodology.


S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 161

observing surface failures of flexible pavement during the 3.2.2. Pavement condition data
survey. The length of each test section considered is Pavement condition data collected in this study includes
approximately 500 m. The detailed road network map of different types of distresses. The selected sections were
the selected study area is shown in Fig. 2. found to be prone to a number of surface distresses such
as alligator cracking, patching, rutting, potholes and ravel-
3.2. Pavement field surveys ling. These distresses are considered as the main variables
in this study. A 2.0 m straight edge was used to measure
The different types of data collected from the selected the rutting of a section along the wheel path. Remaining
study area are presented in the following sections. distresses i.e. alligator cracking, patching, potholes and
ravelling were measured using a laser based instrument
3.2.1. Pavement inventory data called Terrestrial Laser Scanner (TLS). TLS is a high-
The pavement inventory data collected in this study definition surveying instrument that works on the principle
includes the following details about the road sections: road of laser scanning. One of the major benefits of using TLS
name, road functional class (expressway/major road), sur- instrument for collecting pavement condition data is user
face type (flexible/rigid), carriageway width, length of the safety. It could prove very dangerous to collect data man-
road, number of lanes (divided/undivided) etc. ually from the roadway, whereas it can be collected safely,

Fig. 2. Pavement sections selected at Study Area.


162 S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170

and without interrupting the traffic, with the TLS Likewise, the scanned images of pothole, alligator crack-
instrument. ing and ravelling for each location were captured with the
It is also observed that greater objectivity is obtained TLS and transferred to the Cyclone software. Alike, the sur-
through automated means as the data accuracy and con- face area of remaining distresses were measured as measured
sistency improve, along with having a permanent record for patching. The measuring unit of the area of these dis-
of pavement condition data. Furthermore, it is important tresses is Sq.m. The surface area of a section is the represen-
to take advantage of appropriate innovative technologies tation of cumulative sum of surface area of each subsection
for reducing (improving efficiency) the project duration of individual distress. Further, the area of the distresses was
[14]. converted to percentage of the total area of the individual
Each of these 28 test sections are 500 m long, divided section, and is represented in Table 1. A 2.0 m straight edge
into a subsections of 100 m each. Once the section is was used to measure the rutting of the surface along the
selected then the instrument is located near the distress wheel path, at an interval of 50 m of 500 m long section.
position at a certain known distance in such a way that The measuring unit of rutting is ‘mm’ and was converted
TLS will not get disturbed throughout the scanning pro- to percentage of the total length of the section. The data from
cess. Stepwise scanning procedure has been described in each section is represented in Table 1.
Leica ScanStation C10 user manual [15], and the same There is a significant variation in the extracted distress
guidelines are followed in this study. data, as shown in Table 1, and it is found that all sections
do not have all the distresses, with the exception of sections
2 and 27. In the study sections, compared to other dis-
3.3. Data extraction tresses, ravelling is much more significant, hence the overall
impact of ravelling is much greater than other distresses.
The captured scan images of the distresses of each sub- The maximum share of each type of sections with different
section were exported to the Cyclone software which is distresses can be seen in the present study: (i) section 20 has
built within the instrument. The procedure to measure maximum patching and cracking, (ii) section 23 has maxi-
the surface area of a patching in a typical section is summa- mum pothole and rutting and (iii) section 27 has maximum
rized in Fig. 3. ravelling. It can also be concluded from Table 1 that sec-
tion 27 is the most deteriorated section whereas section 6
 As an illustration, the scan image of a patched surface is is the least deteriorated section.
shown in Fig. 3a.
 A number of points were selected throughout the patch-
ing surface to generate a polyline, as shown in Fig. 3b. 3.4. Pavement maintenance prioritization using AHP
 The generated polyline is filled with colour as shown in technique
Fig. 3c.
 Finally, the surface area is measured using the com- 3.4.1. Basic principle of Analytic hierarchy process (AHP)
mands provided in the Cyclone software, as shown in AHP was developed by Saaty in the 1970s for dealing
Fig. 3d. The surface area is measured in Sq.m. with complex problems of technological, economical and

Fig. 3. Steps to calculate the patching surface area of a typical section.


S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 163

Table 1
Evaluated distress data.
Section No. Road Type % Patch % Pothole % Ravelling % Cracking % Rutting
1 MR 0.1911 0.02864 – – 0.014
2 MR 0.2893 0.01477 1.19098 0.02157 0.106
3 EW – 0.01132 0.43582 0.00101 0.041
4 EW 9.3893 – – – 0.052
5 EW – – 1.59933 – 0.051
6 EW – 0.05541 – 0.01268 0.021
7 EW 0.0955 – – – 0.043
8 EW – 0.02947 0.76058 – –
9 EW 0.2924 – 0.24531 0.00505 –
10 EW – – 11.8889 – 0.039
11 EW 1.3143 0.02947 0.65889 – 0.041
12 MR 0.0592 0.0241 0.4583 – –
13 MR 0.4067 0.19905 1.23179 0.02103 –
14 MR 11.454 – 4.15973 2.43448 0.291
15 MR – – 1.9092 0.26114 0.213
16 EW 4.1020 – 1.37019 1.92522 0.041
17 EW 3.9913 0.03723 – – –
18 EW 7.0159 0.28736 3.51748 – 0.055
19 EW – 0.07324 5.48353 0.22670 0.065
20 MR 18.125 – 30.6115 2.70634 0.241
21 EW 2.0653 – 7.65221 0.22446 0.040
22 MR 0.4189 1.30099 13.2144 – 0.364
23 MR – 5.60633 36.522 – 1.333
24 EW 2.2732 0.10062 8.16105 – 0.060
25 MR – 2.30353 30.9171 – 0.147
26 MR – 0.14908 13.3844 1.34267 0.148
27 MR 5.5714 1.32147 47.2443 0.52338 0.882
28 EW 1.6169 0.49682 8.47501 – 0.051
Note: (–) indicates no distress found, MR: Major Road, EW: Expressway.

sociological. AHP aims to quantify relative priorities for a Rule 2. If Ci is judged to be of equal relative importance
given set of alternatives on a ratio scale [6]. AHP is a math- as Cj, then aij = aji = 1. Obviously aii = 1 for all i.
ematical technique, and is used for multi-criteria decision 2 3
aii aij ::: ain
making to help the decision maker to select the best alter- 6 1=a
native [3]. In this method, the complex structure of the 6 ji ajj ::: ajn 7
7
A¼6 7 ð2Þ
problem is reduced by handling this complexity at different 4 : : ::: : 5
levels. Each level consists of a group of parameters possess- 1=ani 1=anj ::: 1=ann
ing similar characteristics. In this method, an overall goal is
at the top or first level followed by a set of criteria at mid- A positive reciprocal matrix which consists of a different
level, followed by a set of alternatives to reach the overall set of pairwise comparison is represented in Eq. (2).where i,
goal. Usually, the criteria are further divided into subcrite- j 6 n, n indicates the number of alternatives being com-
ria, sub-subcriteria and so on, depending on the complexity pared within one set of pairwise comparisons, aij denotes
of the problem. A nine-point scale is suggested for AHP to the importance of alternative i over alternative j.
compute the relative importance of all elements, compared The judgmental values to each element in matrix A are
pairwise. The relative importance of each numeric value in assigned, and the priority vector w is determined. Saaty’s
Saaty’s scale is: 1 if criteria i and j are of equal importance, Eigen vector method (EM) is often applied to derive the
3 if criterion i is little more important than criterion j, 5 if priorities of the alternatives and compute the value of w0 ,
criterion i is strongly more important than criterion j, 7 if the principal Eigen vector, the vector corresponding to
criterion i is very strongly more important than criterion the largest Eigen value, kmax of the matrix A as shown in
j, 9 if criterion i is absolutely more important than criterion Eq. (3) [8].
j, and 2, 4, 6, 8 for intermediate importance [16]. The judg- Aw0 ¼ kmax w0 ð3Þ
mental value for pairs of attributes Ci and Cj are presented 0
by an n-by-n matrix as shown in Eq. (1). where, kmax P n, w = [w1, w2, w3, . . ., wn] , and the super-
T

script T denotes transpose of a matrix. The priority vector


A ¼ ðaij Þ ði; j ¼ 1; 2; 3; . . . ; nÞ; ð1Þ w is obtained by normalizing the principal Eigen vector w0 ,
where aij is defined by the following entry rules: and is called the normalized principal Eigen vector of the
Rule 1. If aij = a, then aji = 1/a, a – 0. pairwise comparison matrix. It is established for each crite-
rion, sub-criterion as well as the alternatives under each
164 S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170

sub-criterion. The overall priority weight of alternatives is objectively. These judgmental values were assigned based
computed using Eq. (4). on the extracted field data, instead of basing on the opinion
X of experts, which is the usual practice. To understand the
Vi ¼ W j X ij ð4Þ concept of objective evaluation, the procedure to assign
j
judgmental values for Expressway (EW) is illustrated here:
where Vi = overall priority weight of alternative i, Wj =
weight assigned to criterion j, and Xij = weight of alterna-  Referring to Table 1 the mean value of each distress of
tive i given criterion j. EW was computed (AC = 0.399; PT = 2.728;
Literature shows that AHP allows 10% inconsistency in pH = 0.188; RV = 4.475; RT = 0.059), where, EW =
human judgments [8,12]. To check for consistency in judg- expressway, AC = Alligator cracking, PT = Patching,
ments of a decision maker, the consistency ratio (CR) PH = Pothole, RV = Ravelling and RT = Rutting.
which is defined as the proportion of the consistency index These mean values for pairs of distresses were arranged
(CI) and the random index (RI) is used to examine the in matrix form using Eq. (1).
entire matrix A using Eq. (5).  Among these mean values of distress, the difference of
CI minimum (0.059) and maximum (4.475) value were
CR ¼ ð5Þ found out. This difference value (4.416) is further
RI
divided by 8 to obtain the interval value (0.552).
where CI is as given in Eq. (6). Define RI and CI  Saaty’s scale contains 8 intervals between 9 integer val-
kmax  n ues, from 1 to 9. Starting from 1 for equal importance
CI ¼ ð6Þ of distress (pairwise comparison between the same dis-
n1
tresses), each successive rating is obtained by adding
where n is the size of the matrix. The values of the random the interval value (0.552) to the distress value of the pre-
index for quantities of attribute to the different size of the ceding rating (2 = 0.552; 3 = 1.104; 4 = 1.656;
matrix were adopted. Also, a matrix is considered consis- 5 = 2.208; 6 = 2.760; 7 = 3.312; 8 = 3.864; 9 = 4.416).
tent only if CR 6 0.1 [16].  Pairwise comparison values of distress lying between the
distress values representing the rating values are treated
3.4.2. Hierarchy structure for AHP method as having the lower rating value if they are less than the
In this study a hierarchy is developed by considering a average distress value of the boundary distress values,
number of parameters at different levels of the structure. and higher otherwise. The outcome of each set of pair-
Road type and pavement surface distresses are considered wise comparison is expressed as a positive reciprocal
as the modelling parameters of criteria and subcriteria level matrix as follows:
respectively. Road type is an important criterion in pave-
ment maintenance prioritization. The parameters Express-
way (EW) and Major Road (MR) were in the criteria
2 3
level for road type. The major difference between these EW AC PT PH RV RT
parameters is that an expressway has higher traffic volume 6 AC 7
6 1 1=5 2 1=8 2 7
than a major road. Therefore, surface distresses are more 6 7
6 PT 5 1 6 1=4 6 7
prone to occur in expressway than in major road. A pave- A¼6
6 PH
7
ment surface distress plays a vital role and is a very impor- 6 1=2 1=6 1 1=9 2 7
7
6 7
tant criterion considered herein. Distresses such as alligator 4 RV 8 4 9 1 9 5
cracking, patching, pothole, ravelling and rutting were trea- RT 1=2 1=6 1=2 1=9 1
ted as subcriteria in pavement surface distress type for AHP
modelling. These distresses occur due to environmental fac-
tors, seasonal factors and excessive traffic loading etc. The stepwise calculation of priority weight and consis-
The overall objective, to prioritize pavement sections for tency ratio of expressway (EW) is explained here. For
maintenance is placed at the top i.e. level 1 in the hierarchy instance, assigned judgment indicated that the importance
structure. Road functional class is taken as the main crite- of PT subcriteria over AC was 5 (strongly preferred). Next,
rion at level 2, followed by distress type as the subcriterion this normalized matrix having the division of each i-th ele-
at level 3. All 28 candidate sections for pavement mainte- ment by the sum of the elements of i-th column is illus-
nance are considered as the alternatives. These sections trated here.
are placed at the bottom, the fourth level in the hierarchy 2 3
structure. This hierarchy structure of AHP analysis is 0:067 0:036 0:108 0:078 0:100
shown in Fig. 4. 6 7
6 0:333 0:180 0:324 0:157 0:300 7
6 7
A¼6
6 0:033 0:030 0:054 0:070 0:100 7
7
3.4.3. Objective based evaluation of AHP method 6 7
4 0:533 0:722 0:486 0:631 0:450 5
In order to obtain more realistic and reliable compar-
ison matrices, each attribute is assigned a judgmental value 0:033 0:030 0:027 0:070 0:050
S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 165

Fig. 4. Hierarchy structure for AHP analysis.

The weights of each attribute obtained by the arithmetic has the least priority rank with maximum priority rating
mean of the elements of the i-th line of the normalized value, whereas section 6 is found to be highest priority rank
matrix are: w1 = 0.077, w2 = 0.259, w3 = 0.057, with minimum priority rating value, and the same is
w4 = 0.563 and w5 = 0.042. Further, Eigen values (k) of observed in Table 1 discussed in Section 3.3.
each attribute were determined (k1 = 5.12, k2 = 5.35,
k3 = 4.98, k4= 5.53, k5 = 5.12) from Eigen vector method 3.5. Pavement maintenance prioritization using RCI method
using Eq. (3). The mean of Eigen value was computed as
kmax = 5.22. The consistency in data was checked by using 3.5.1. Concept of road condition index method
Eq. (6), CI = 0.055. RI value is chosen as 1.12 correspond- Traditional pavement priority rating is obtained by con-
ing to n = 5, and CR value was calculated using Eq. (5). verting pavement distress data into a single condition
For this experiment, the calculated CR value is 0.049 index. The representation of all the distresses of a pave-
(4.9%), less than 10%. Likewise, the priority weight of each ment section into a single equation in numerical form pro-
attribute of criteria and subcriteria were calculated and dis- vides the condition index.
played in the Table 2. Consistency ratio of each attribute For pavement evaluation studies, Pavement Condition
was found to be less than 0.1. Index (PCI) method is considered as one of the most pre-
where, W1, W2 . . . W28 indicates the priority weight of cise indices that incorporate data from different types of
each section with respect to individual distress type. Pair- distresses as well as their severity and quantity [13]. Like-
wise comparison was made between the sections based on wise, a condition index called Road Condition Index
the available distresses on those sections and the same is (RCI) has been developed for Indian conditions during
represented in the Table 2. The priority weight of each sec- PMS study of Noida city [11]. In the present study, the
tion was computed using Eq. (4) and the same is repre- road condition index method is used to validate the priority
sented in the Table 3. rating and ranking of selected sections.
These priority weights range from 0 (perfect condition A road condition index is the weighted average of all
pavement) to 1 (deteriorated condition pavement). Sections urgency indexes, product of degree and extent of distress.
with more priority weight should be given higher priority RCI is represented in terms of number that indicates the
for maintenance. These weights were further transformed overall performance of the study area which consists of
into rating scale called priority rating. On the basis of pri- number of distresses such as: patching, rutting, ravelling,
ority rating value each section was assigned a priority rank- potholes and cracks. The value of RCI varies between 1
ing as represented in the Table 3. In this study, section 27 for a new pavement with no distress and 25 for a failed
166
Table 2
Priority weight of criteria and subcriteria (Consistency ratio < 0.1).

S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170
Major Road Expressway
0.167 0.833
Priority weight of subcriteria with respect to Major Road
Alligator Cracking Patching Potholes Ravelling Rutting
0.060 0.155 0.060 0.664 0.060
Priority weight of subcriteria with respect to Expressway
Alligator Cracking Patching Potholes Ravelling Rutting
0.077 0.259 0.057 0.563 0.042
Priority weight of sections with respect to Alligator Cracking
W2 = 0.019, W3 = 0.018, W6 = 0.019, W9 = 0.019, W13 = 0.020, W14 = 0.227, W15 = 0.054, W16 = 0.157, W19 = 0.031, W20 = 0.269, W21 = 0.031, W26 = 0.081, W27 = 0.050
Priority weight of sections with respect to Patching
W1 = 0.018, W2 = 0.018, W4 = 0.099, W7 = 0.017, W9 = 0.018, W11 = 0.029, W12 = 0.017, W13 = 0.018, W14 = 0.143, W16 = 0.063, W17 = 0.030, W18 = 0.077, W20 = 0.270, W21 = 0.030, W22 = 0.018,
W24 = 0.029, W27 = 0.065, W28 = 0.032
Priority weight of sections with respect to Pothole
W1 = 0.024, W2 = 0.024, W3 = 0.024, W6 = 0.025, W8 = 0.024, W11 = 0.024, W12 = 0.024, W13 = 0.040, W17 = 0.024, W18 = 0.044, W19 = 0.025, W22 = 0.055, W23 = 0.316, W24 = 0.027, W25 = 0.119,
W26 = 0.036, W27 = 0.082, W28 = 0.050
Priority weight of sections with respect to Ravelling
W2 = 0.014, W3 = 0.014, W5 = 0.014, W8 = 0.014, W9 = 0.014, W10 = 0.042, W11 = 0.014, W12 = 0.014, W13 = 0.014, W14 = 0.021, W15 = 0.015, W16 = 0.014, W18 = 0.021, W19 = 0.027, W20 = 0.112,
W21 = 0.029, W22 = 0.029, W23 = 0.139, W24 = 0.029, W25 = 0.108, W26 = 0.046, W27 = 0.211, W28 = 0.032
Priority weight of sections with respect to Rutting
W1 = 0.021, W2 = 0.024, W3 = 0.021, W4 = 0.023, W5 = 0.023, W6 = 0.021, W7 = 0.021, W10 = 0.022, W11 = 0.021, W14 = 0.062, W15 = 0.045, W16 = 0.021, W18 = 0.023, W19 = 0.021, W20 = 0.041,
W21 = 0.021, W22 = 0.043, W23 = 0.235, W24 = 0.023, W25 = 0.034, W26 = 0.047, W27 = 0.152, W28 = 0.023
S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 167

Table 3 Table 4
Section rating and ranking value using AHP technique. Detailed description of degree of distress [11].
Section number Priority weight of Priority Priority Degree Description
sections rating ranking (D)
1 0.0416 4.16 26 1 Very small: Difficult to detect the defects, less significance.
2 0.1069 10.69 18 Maintenance is not needed
3 0.0787 7.87 22 2 Small: Easy to detect the defects, but of less significance.
4 0.1563 15.63 14 Immediate maintenance is not needed
5 0.0624 6.24 24 3 Moderate: Notable defects, but acceptable. Maintenance is
6 0.0235 2.35 28 needed
7 0.0328 3.28 27 4 Severe: Significant amount of defects, undesirable.
8 0.0654 6.54 23 Maintenance is needed regularly
9 0.0916 9.16 20 5 Very severe: Insignificant amount of defects, unacceptable.
10 0.1702 17.02 13 Immediate maintenance is needed.
11 0.1153 11.53 16
12 0.0913 9.13 21
13 0.1082 10.82 17
14 0.4163 41.63 5
Table 5
15 0.0949 9.49 19
Detailed description of Extent of distress [11].
16 0.2282 22.82 7
17 0.0545 5.45 25 Extent Description
18 0.2188 21.88 8 (E)
19 0.1359 13.59 15 1 Very few isolated occurrences, less than 5% of the affected
20 0.9661 96.61 2 pavement
21 0.1779 17.79 10 2 Discontinuous occurrence, range between 5% and 15% the
22 0.1703 17.03 12 affected pavement
23 0.7079 70.79 3 3 Regular occurrence, range between 15% and 30% of the
24 0.1723 17.23 11 affected pavement
25 0.4662 46.62 4 4 Extensive regular occurrence, range between 30% and 60% of
26 0.2387 23.87 6 the affected pavement
27 1 100 1 5 Throughout, Extensive regular occurrence, greater than 60%
28 0.196 19.6 9 of the affected pavement

Table 6
pavement. Pavements with higher RCI should be given Weight assigned to various distress [11].
higher priority for maintenance. RCI can be calculated Sl. no. Distress type Relative weight Weight (Wi)
using Eq. (7). (as per expert’s opinion)
P
ðUI i  W i Þ 1 Patching 0.080 1
RCI ¼ P ð7Þ 2 Rutting 0.120 2
Wi 3 Ravelling 0.183 3
where UIi and Wi indicates urgency index and weight of a 4 Potholes 0.262 4
5 Cracks 0.355 5
distress corresponding to distress parameter i respectively.
Urgency index indicates the road condition; higher UI
value indicates largely deteriorated road which requires
urgent attention. UI is calculated by multiplying degree 3.5.2. Priority ratings and ranking using RCI method
and extent values for a particular road, as shown in Eq. (8). In this study, Tables 4 and 5 were used, in terms of
Urgency Index ¼ Degree  Extent ð8Þ degree and extent respectively, to classify the collected data
for all the sections. The degree and extent values were
Urgency index plays an important role in the evaluation assigned for each section and urgency index values for
of road condition. In a functional evaluation of pavement, the individual section were calculated using Eq. (8). Fur-
individual distress was assigned a particular value of degree ther, RCI values for each section were computed using
and extent, ranges from 1 to 5. Degree is indicated as the Eq. (7), as shown in Table 7.
level to which the road is affected and Extent is defined The RCI values of the study vary from 1 to 8.67, which
as the frequently occurrence of a particular distress. The is normalized to a range of 1–25 by using standard normal-
degree and extent value were assessed for each section ization procedure. Further, this RCI rating value is nor-
based on their field measured distresses by using Tables 4 malized to the same scale of AHP i.e. 1–100 and shown
and 5 of detailed description of degree and extent respec- in Table 7. The priority rankings derived from the priority
tively as per existing literature [11]. Weightage to each dis- ratings is also shown in the Table 7. The priority rank is
tress are assigned based on experts’ opinion survey, which given to sections considering the higher RCI value as hav-
range from 1 to 5, as shown in Table 6 [11]. ing higher priority. It is found that section 23 has the least
168 S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170

Table 7
Section rating and ranking value using RCI method.
Section number RCI value Priority rating RCI based priority rank AHP based priority rank
(subjective rating) (objective rating)
1 1.00 11.6 20 26
2 1.13 13.08 19 18
3 1.00 11.6 20 22
4 2.00 23.08 8 14
5 1.00 11.6 20 24
6 1.00 11.6 20 28
7 1.00 11.6 20 27
8 1.00 11.6 20 23
9 1.00 11.6 20 20
10 4.00 46.12 5 13
11 1.00 11.6 20 16
12 1.00 11.6 20 21
13 1.30 15.08 17 17
14 3.90 45.08 6 5
15 1.40 16.16 16 19
16 2.00 23.08 8 7
17 1.20 13.84 18 25
18 2.00 23.08 8 8
19 1.64 18.96 14 15
20 3.36 38.8 7 2
21 1.90 22 12 10
22 4.50 51.88 3 12
23 8.67 100 1 3
24 1.90 21.88 13 11
25 4.22 48.68 4 4
26 1.57 18.12 15 6
27 7.20 83.04 2 1
28 2.00 23.04 8 9

priority rank with maximum priority rating value which method evaluates the pavement sections for maintenance
needs immediate attention. On the other hand nine sections in an objective manner rather than subjective manner, the
have the same rank. Though each section has a different priority list obtained through it can be considered as more
type of distress with different magnitude as seen in Table 1, accurate for implementation.
but ranks of these nine sections are same, whereas the same
sections as evaluated using objective based AHP technique 3.7. Evaluation of priority ratings and priority rankings
have different rank, as seen in the Table 3. This is the draw-
back of the RCI method. The present study overcomes this It is observed from Tables 3 and 7 that there is only a
difficulty, where each section evaluated using AHP method slight variation in the priority rating values obtained from
in objective manner has its own priority rank. the AHP and RCI methods. This hints at a linear relation-
ship between the ratings. Hence, to find the best fit, assess-
3.6. Comparison of AHP and RCI ranking ment of the priority rating score is performed with the
Pearson correlation coefficients, which reflects the degree
The priority rankings of the road sections obtained by of linear relationship between the results of AHP and
two methods, objective rating (AHP) and subjective rating RCI methods. Fig. 5a presents the plot of the priority rat-
(RCI), are shown in the Table 7. It is seen that for few sec- ing scores obtained by the AHP method against RCI
tions, the priority rank differs significantly while for other method. Pearson correlation coefficient of 0.79 is observed
pavement sections the rankings are close to each other. in a scatter plot for these two methods as shown in Fig. 5a.
It is observed that the RCI method has assigned the A strong correlation coefficient of 0.79 is evident from
same rank to 9 sections. Though each section has a differ- Fig. 5a; however, there appears to be some discrepancies
ent type of distress with different severity, but the rank of that are reflective of the differences in the basic approach
nine sections is same. The same sections when evaluated and rating scales used by the AHP method and RCI
using objective based AHP technique had different ranks. method.
This drawback of the RCI method, where RCI values, Basically, in the AHP method pairwise values are
dependent on degree and extent of distress, were computed assigned quantitatively on a ratio scale, whereas in the
in subjective manner, is overcome by the present study by RCI method the assigned values of degree and extent are
incorporating an objective evaluation. Since the AHP qualitative. Since it is a relative magnitude of the ratings
S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170 169

120 30
110 r= 0.79
27 ρ =0.85
100
24
90
80 21
AHP Rating

AHP Ranking
70 18
60 15
50
12
40
9
30
20 6

10 3
0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 0 3 6 9 12 15 18 21 24 27 30
RCI Rating RCI Ranking
(a) (b)
Fig. 5. AHP and RCI method correlations between (a) priority ratings and (b) priority rankings.

n > 10 the significance of Spearman correlation can be


Table 8 tested by student’s statistics t as defined in Eq. (10) [10].
Results of statistical hypothesis testing of rank correlation. q
tn2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð10Þ
Statistic AHP versus RCI method ð1  q2 Þ=ðn  2Þ
Observations 28
Degrees of freedom 26
The results of the hypothesis test are summarized in the
Confidence level tested 95% Table 8. The priority rankings obtained between AHP and
Pearson Correlation 0.85 RCI methods are statistically consistent at the 95% confi-
Student’s t-test (tn-2 for n > 10) 8.13 dence level.
Critical one-sided T-value (ta, n-2) 1.70 Since tn-2 < ta, n-2, we reject the null hypothesis, and con-
Result tn-2 > ta, n-2
Conclusion Accept H1: q > 0
clude that there is a non-zero correlation between AHP and
RCI method. However, the results of priority rankings
of different maintenance activities rather than the absolute from AHP method are statistically consistent with the
differences in their rating scores that matter in pavement rankings from RCI method. In the comparison of the
maintenance planning, an evaluation based on the relative applicability of the RCI method and the proposed objec-
rankings of the maintenance work is appropriate. The tive based AHP method, the ease of application of the
assessment of priority ranking is performed through the RCI procedure was recognized as a major advantage.
use of nonparametric Spearman rank correlation coeffi- However, the degree and extent values in RCI method
cient, q, expressed in the following Eq. (9) [10]. are fixed and do not reflect the variation in the distress
Pn data. It is not fitting that the urgency index values thus
6 i¼1 d 2i obtained will provide the best results for highway agencies
q¼1 ð9Þ
nðn2  1Þ to base their maintenance decisions on. The proposed
objective based AHP method can overcome this limitation.
where di is the difference between the ranks of pavement
section i by the AHP and RCI methods being evaluated 4. Conclusions
and n is the number of the alternatives involved in the test.
AHP and RCI method show a positive rank correlation A simple, suitable and refined decision-making method
of 0.85 between the priority rankings of two methods as is required for the road, a maintenance program which
shown in Fig. 5b. This shows that AHP was able to pro- consists of databank availability and profitability. Avail-
duce the priority ranking of pavement maintenance sec- able research in India for the prioritization of pavement
tions in excellent consistence with the RCI method. maintenance activities for PMS study is empirical. Pave-
ment condition data such as pothole, alligator cracking,
3.8. Statistical hypothesis testing for rank correlation patching and ravelling were measured using terrestrial laser
scanner instrument. Because prioritization is a decision
The concept of the nonparametric rank correlation test making process, statistical models are not very responsive.
can again be used for testing the hypothesis checking the The Analytic Hierarchy Process is found to be one of the
correlation relationship of obtained q value which is 0.85. simplest and most useful methods in multi-criteria decision
The test was performed with null hypothesis H0: q = 0, making process. This method is based on pairwise compar-
against the alternative hypothesis H1: q > 0 [10]. When isons which facilitate calculations and judgments.
170 S. Ahmed et al. / International Journal of Pavement Research and Technology 10 (2017) 158–170

In this study, objective evaluation process is introduced ficult for an evaluator to make pairwise comparison man-
to assign judgmental value for pairwise comparison in ually in consistent manner, which may lead to rank
AHP technique. This objective evaluation is done based reversal problem. In such cases the priority weight can be
on the collected field data. The objective method avoids calculated using Expert Choice or Super Decision software.
the problems faced in subjective methods, where experts
assess the pavement section based on their experience Acknowledgment
alone. These experts opinion for a particular section vary
significantly. An attempt has been made to apply an objec- We would like to thank Prof. RAAJ Ramsankaran and
tive AHP technique for the prioritization of pavement Mr. Shivraj Patil, Geodesy and Remote Sensing group, IIT
maintenance activities to 28 sections in Mumbai city. The Bombay for their help and constant support during data
priority weights of criteria, subcriteria and alternatives collection.
were calculated using the concepts of AHP technique.
The priority weight of alternatives which varies from 0 to References
1 was converted to priority rating scores. These sections
were then ranked based on the priority rating scores. [1] NCHRP, Synthesis of Highway Practice 523: Optimal Timing of
The results of the objective AHP method were assessed Pavement Preventive Maintenance Treatment Applications, Trans-
portation Research Board of the National Academies, Washington,
with the results of the road condition index method, which D.C., 2004.
is an index based method developed for Indian condition. [2] E. Sharaf, Ranking versus simple optimization in setting pavement
In the RCI method, weighted average of urgency indices maintenance priorities: a case study from Egypt, Transport. Res.
of distresses is represented in a single equation in numerical Record: J. Transport. Res. Board 1397 (1993) 34–38.
[3] T.F. Fwa, R. Shanmugam, Fuzzy Logic Technique for Pavement
form. It was observed in the study that slightly different
Condition Rating and Maintenance Needs Assessment, in: Proc.
priority rating values were obtained. However, the objec- Fourth International Conference on Managing Pavements, May,
tive based AHP generated priority ratings were positively Durban, South Africa, Vol. 1 (1998) 465–476.
correlated with those obtained by the RCI method. Simi- [4] B.B. Reddy, A. Veeraragavan, Priority ranking model for managing
larly, priority ranking evaluation by objective based AHP flexible pavements at network level, J. Indian Road Congr. 62 (3)
and RCI methods were found to be strongly correlated. (2001) 379–394.
[5] T.L. Saaty, The Analytic Hierarchy Process, McGraw-Hill, New
This priority ranking correlation was further checked York, 1980.
through statistical hypothesis testing. The statistical test [6] R.H. Ramadhan, H.I. Wahhab, S.O. Duffuaa, The use of an
which was performed at 95% confidence level confirmed analytical hierarchy process in pavement maintenance priority rank-
that the priority rankings between AHP and RCI methods ing, J. Qual. Maint. Eng. 5 (1) (1999) 25–39.
are strongly correlated. [7] D. Moazami, R. Muniandy, H. Hamid, Z. Yusoff, The use of
analytical hierarchy process in priority rating of pavement mainte-
This study also shows the comparison between RCI and nance, Sci. Res. Essay 6 (12) (2011) 2447–2456.
proposed objective AHP methods from their application [8] R.V.O. Almeida, E.F.N. Júnior, B.A. Prata, Prioritization of Earth
point of view, the RCI method were found to be more ben- roads maintenance based on analytic hierarchy process, Int. J.
eficial than the proposed method with the ease of applica- Pavement Res. Technol. 5 (3) (2012) 187–195.
tion of the RCI procedure. In the RCI method, the [9] J. Farhan, T.F. Fwa, Pavement maintenance prioritization using
analytical hierarchy process, Transport. Res. Record: J. Transport.
generated urgency index values were fixed for a given set Res. Board 2093 (2009) 12–24.
of distress data and it is very difficult to find out the exact [10] J. Farhan, T.F. Fwa, Use of analytic hierarchy process to prioritize
maintenance strategy and preferences for highway agen- network-level maintenance of pavement segments with multiple
cies. This limitation is overcome by using the proposed distresses, Transport. Res. Record: J. Transport. Res. Board 2225
objective AHP method. The findings of this study suggest (2011) 11–20.
[11] Y.U. Shah, S.S. Jain, M. Parida, Evaluation of prioritization methods
that AHP approach is suitable for the purpose of pavement for effective pavement maintenance of urban roads, Int. J. Pavement
maintenance prioritization. This study has introduced an Eng. (2012) 1–13
objective manner of evaluation of pavement maintenance [12] A.C. Prakasan, D. Tiwari, Y.U. Shah, M. Parida, Pavement
sections for prioritization using the AHP technique. Using maintenance prioritization of urban roads using analytical hierarchy
process, Int. J. Pavement Res. Technol. 8 (2) (2015) 112–122.
the proposed method, transportation agencies would be
[13] M.Y. Shahin, Pavement Condition Survey and Rating Procedure,
able to prioritize the maintenance of impaired sections, Pavement Management for Airport, Roads, and Parking Lots, Kluwer
and the priority order will be reflective of the condition Academic Publishers, New York, USA, 2002.
of the sections. In this study, number of alternatives for [14] C.W.H. Johnson, A.M. Johnson, Operational considerations for
the maintenance of pavement sections, for a road network terrestrial laser scanner use in highway construction applications, J.
is considered as n = 28, but practically it would be much Survey. Eng. 138 (4) (2012) 214–222.
[15] Leica Geosystem. Leica ScanStation C10/C5 User Manual, Version
greater than this. In addition, the distress types will also 5.0.
be more than the five types considered in the study. Fur- [16] K.C. Sinha, S. Labi, Evaluation of Transportation Projects and
thermore, the size of the problem can increase significantly Programs Using Multiple Criteria, Transportation Decision Making
with the addition of more factors in different levels during Principles of Project Evaluation and Programming, John Wiley & Sons
the assessment process, so the pairwise comparison Inc, Hoboken, New Jersey, 2007.
increases rapidly, and it becomes time consuming and dif-
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

S-ar putea să vă placă și