How to Determine Road Maintenance Priorities? An Analysis Using a
Probabilistic Approach
Melchior Bria, Priska G. Nahak and Anastasia H. Muda
Department of Civil Engineering, State Polytechnic of Kupang, Adisucipto Street, Kupang, NTT, Indonesia
Keywords: Road Maintenance, Probabilistic Approach, Binary Logistic.
Abstract: The provision of limited maintenance budget for basic infrastructures often lead to unrepair and dilapidation
of the untargeted roads, and this tends to be wasteful. Presently, treatment priority is one of the alternatives
that have been carried out. This study aims to discuss determining road maintenance priorities based on a
probabilistic approach, namely the probability of a road section being proposed in the maintenance program.
The samples used were obtained from the Bakunase-Oenesu and Soe-Fatumnasi roads. The variables utilized
were the pavement structure, average daily traffic, accessibility, economic benefits, time travel, and asphalt
recycling technology. The binary logistic analysis showed that the condition of the pavement structure and
the average daily traffic were variables that had significant effect on road maintenance. There is a greater
chance for the Soe-Fatumnasi road section than the Bakunase-Oenesu being drafted for maintenance. A
probabilistic approach can be applied in determining roads for maintenance programs. This method will be
more appropriate if applied to planning at the first level. In making road maintenance decisions, policymakers
need to pay attention to various factors to make their decisions right on target.
1 INTRODUCTION
In road maintenance planning, obstacles are often
encountered while producing a predictive model for
it conditions, mainly due to errors in the visual survey
and lack of knowledge about essential factors
affecting degradation (Zhao B, Soga K and Silva E
2019). When this continues, it affects the road's
serviceability until it reaches its economical age and
this increases costs for users (Salih J, Edum-Fotwe F
and Price A 2016) (Mikolaj J, Remek L and
Margorinova M 2019), also impacting the residents
and the environment around the area negatively
(Freitas E, Silva L and Vuye C 2019). Furthermore,
highway repair is closely related to budget
availability, and it is necessary to carry out an
appropriate assessment (Muda A, Dumin L and
Nahak P 2019), which include the deterioration of this
area and its sections, such as inspection of the
pavement, shoulders, drainage channels, and
subgrade (Bhuva C R, Patel B and Kanam M 2019)
(Tchemou G, Minsili L S, Mokotemapa A M, Eko R
M and Manguelle J H 2011). This implies that there
is need for right decision making when planning road
maintenance (Siswanto H, Supriyanto B, Pranoto,
Putra Y A M and Huda A S 2018). Consequently,
various efforts have been made by government both
technically and managerially, these involved carrying
out routine checks periodically, and conducting a
priority analysis for sustainability. However, the
sustenance budget is often not followed by useful
methods, therefore, it does not produce a good result
(Susanna A, Crispino M, Giustozzi F and Toraldo E,
2017). Maintenance is an essential part of traffic
management (Ma J, Cheng L and Li D, 2018), and
when planning, it is necessary to consider various
aspects before making decisions, including the
performance and strength of the pavement structure
(Bazlamit S M, Ahmad H S and Al-Suleiman T I,
2017), traffic load, age, road level, and non-technical
aspects, such as resources, disposition, and
bureaucratic formation (Li H, Ni F, Dong Q and Zhu
Y 2018) (Hayat E and Amaratunga D 2014). Besides,
due to limited natural and financial resources,
handling critical highways needs to be performed
with a good strategy by relying on renewable
materials and choosing the right construction method
(Susanna A, Crispino M, Giustozzi F and Toraldo E
2017).
One of the techniques is planning a proper
maintenance program. Studies on organizing have
been carried out, with a priority approach in making
1018
Bria, M., Nahak, P. and Muda, A.
How to Determine Road Maintenance Priorities? An Analysis Using a Probabilistic Approach.
DOI: 10.5220/0012028700003575
In Proceedings of the 5th International Conference on Applied Science and Technology on Engineering Science (iCAST-ES 2022), pages 1018-1022
ISBN: 978-989-758-619-4; ISSN: 2975-8246
Copyright © 2023 by SCITEPRESS Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
decisions using multicriteria analysis (Ahmed S,
Marcelino P, Liem F, 2019), based on road structural
data (Bazlamit S M, Ahmad H S and Al-Suleiman T
I 2017). However, the use of a probabilistic approach
in planning maintenance programs has never been
carried out. To fill this gap, this paper aims to discuss
determining road maintenance priorities based on the
probability approach, namely the chance of a road
section being included in the maintenance program.
For this reason, the considerations that will be used in
the analysis are not only on the structural aspects of
roads but also considering the economic value of
roads and the application of green construction
principles.
2 METHOD
The research location was in West Timor, East Nusa
Tenggara Province. The samples were the roads of
Bakunase-Oenesu in Kupang Regency and Soe-
Fatumnasi in South Central Timor. The main concept
of this study was to identify the chances for a section
to be included in the road maintenance program,
because in one budget year, the available repair funds
were very limited, below what was needed.
Therefore, to strengthen the related technical
agencies' proposals, the planning approach with this
probability concept was used as a guide in proposing
maintenance for a particular road section. The
considerations used to measure how many
opportunity was accepted or not were based on these
main criteria, namely the condition of the pavement
structure (X1), average daily traffic (X2),
accessibility (X3), economic benefits (X4), time
travel (X5) and use of asphalt recycling technology
(X6). The dependent variable was the sustenance of
the Bakunase-Oenesu and Soe-Fatumnasi roads.
The approach used to obtain the value of the
formulas above was quantitative subjective; that is,
the respondent made an assessment based on the
information they had on the distributed
questionnaires. Therefore, they understood the
problems involved in road maintenance activities and
were responsible for the problems, including
budgeting. Summarily, they are the policymakers in
relations to construction works. In this study, the
number of respondents were 150 from technical
agencies at the provincial to district levels.
The analytical method used to ascertain the
probability of road maintenance was binary logistic,
and because the questionnaire questions were
qualitative, each response gave a certain nominal
scale, as in Table 1.
Table 1: Variables and scales used.
No. Variable Res
p
onse
1 The Pavement
Structure
(0) slightly damaged,
(1) moderate, (2)
heavil
y
dama
g
e
d
2 Average daily traffic
(0) low, (1) moderate,
(2) hi
g
h
3 Accessbility
(0) low, (1) moderate,
(2) hi
g
h
4 Economic Benefits
(0) low, (1) moderate,
(2) hi
g
h
5 Travel Time
(0) slow, (1) moderate,
(2) normal
6 Application of
recycling technology
(0) is not urgent, (1)
important and ur
g
en
t
7 Maintenance of roads
(0) not yet feasible, (1)
feasible
Based on the variables above, the following model
was formed:
The basic form of binary logistics:
𝑝=
exp(𝛽
+𝛽
𝑋)
1+exp (𝛽
+𝛽
𝑥)
=
𝑒
 
1+ 𝑒

From this basic form, then used as a binary logistic
equation model tested in this study are:
𝐿𝑜𝑔
𝑃/
(
1−𝑃
)
= 𝛽
+ 𝛽
(
𝑋
)
+𝛽
(
𝑋
)
+𝛽
(
𝑋
)
+𝛽
(
𝑋
)
+𝛽
(
𝑋
)
+𝛽
(𝑋
)
β
0
is a constant; β
i
is the regression coefficient of each
variable; Xi is the independent variable. P is the
probability that the road will be included in the
maintenance program; (1-P) is the chance that the
road is not included in the maintenance program.
The model evaluation included the following
tests: (1) goodness of fit, concerning the level of
significance 0.05; (2) simultaneous variable effect
based on the chi-square significance value; (3)
determination using the Negelkerke R-square criteria,
and (4) the significance of the variables in the
equation. Because two dependent quantities were
compared, each section's analysis was carried out
separately for the Bakunase-Oenesu and Soe-
Fatumnasi roads.
3 RESULT AND DISCUSSION
In modeling the probability between two alternative
options, using a discrete approach, the model needed
to meet the test conditions. Table 2 describes the
results of the model test in the form of the goodness
How to Determine Road Maintenance Priorities? An Analysis Using a Probabilistic Approach
1019
of fit test, simultaneous and determination test. The
goodness of fit test and the simultaneous test are
based on a significance value of α = 0.05, while the
determination test shows the magnitude of the
influence of the independent variables in the model.
Table 2: Result of model test.
Model
The goodness of
fit test
Model
simultaneous
test
Determination
test
Y
1
:
Maintenance
of Bakunase-
Oenesu road
1. The chi-square
significance
value 0.211 >
α = 0.05
The chi-square
significance
value on the
omnimbus test is
0.000 < α = 0.05
The
Negelkerke R-
square value is
0.406
2. The percentage
on the
classification
test is 80.2%.
Y
2
:
Maintenance
of
Soe-
Fatumnasi
road
3. The chi-square
significance
value 0.912 >
α = 0.05
The chi-square
significance
value on the
omnimbus test is
0.000 < α =
0.05.
Nilai
Negelkerke R-
square sebesar
0.564
4. The percentage
of the
classification
test is 81.7%.
First, to assess the goodness of fit of the standard,
the terms used were the chi-square significance value.
From Table 2 , it was observed that the two models
met the requirements where the chi-square value of
the Hosmer and Lemeshow test was more than 0.05.
These indicated that the model was used to predict the
observations made.
Secondly, the simultaneous results based on the
omnimbus test (Table 2) show that the chi-square
significance level was less than 0.05 in both. These
proved that all the variables that make up the model
(X
1
- X
6
) had a simultaneous influence on road
maintenance. In other words, at least one of the
quantities had a significant effect on road
maintenance.
Thirdly, from Table 2, it was also observed that
the negelkerke R-square value of the two was far
above 0.05. These results proved that the
independent explain the dependent variable's
instability by 40.6% in the Y
1
model and 56.4% in the
Y
2
model.
Fourth is a partial effect test to see the effect of
each variable. In Table 3, there are constant values for
each variable (β) and significance (Sig) as well as the
odd ratio value (Exp (β)) of the Y
1
model. The results
showed that in the Bakunase-Oenesu road
maintenance model, there were two quantities that
partially have a significant effect, namely X
1
and X
2
because the value was <0.05. Variable X
1
, which was
positive, indicated that the greater the pavement
structure's damage, the greater the chance for the
Bakunase-Oenesu road section to be included in the
road maintenance program. The magnitude of this
opportunity, observed from the odds ratio, was 8,425
times higher than if the damage does not increase.
Likewise, with average daily traffic, it has 1,923
chances of being included in the program when it
increases than when constant or depreciates.
Table 3: Result of Y
1
Model: maintenance of the Bakunase-
Oenesu road.
Variable
β
S.E. Wald Df Sig.
Exp(β)
X
1
2.131 .561 14.430 1 .000 8.425
X
2
.654 .290 5.097 1 .024 1.923
X
3
.150 .309 .235 1 .628 1.162
X
4
.126 .304 .172 1 .678 1.134
X
5
.096 .407 .056 1 .814 1.101
X
6
.439 .449 .955 1 .328 1.551
Constant -4.385 1.354 10.496 1 .001 .012
Similar to Table 3, Table 4 also describes the
constants of each variable in the Y2 model, the
significance value and the odd ratio value of each
variable. In contrast to the Y
1
model, it was observed
that the magnitude of the influence of the X
1
on the
Y
2
was positive by 20,980, greater than if the road
section does not experience additional damage. For
X
2
, it also had a positive effect with an odds ratio of
2.475.
Table 4: Result of Y
2
model: maintenance of Soe-
Fatumnasi road.
Variable B S.E. Wal
d
df Si
g
.
Exp (β)
X
1
3.044 .702 18.811 1 .000 20.980
X
2
.906 .340 7.093 1 .008 2.475
X
3
.291 .353 .680 1 .410 1.338
X
4
.597 .385 2.407 1 .121 1.817
X
5
.797 .489 2.663 1 .103 2.220
X
6
-.442 .457 .934 1 .334 .643
Constant
-6.161 1.714 12.917 1 .000 .002
The analysis results showed that in the Y
1
and Y
2
models, the variables that had significant effect are
the condition of the pavement structure and average
daily traffic. These results confirmed the general
trend that had occurred, where pavement conditions
are the major basis for making road maintenance
iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science
1020
decisions (Bazlamit S M, Ahmad H S and Al-
Suleiman T I 2017, Li H, Ni F, Dong Q and Zhu Y
2018). It means that policymakers have not fully
developed various criteria in assessing the feasibility
of a road section being included in the repair program.
Economic benefits and accessibility have not yet
received attention. It becomes a serious problem
when faced with the choice where many roads are in
the same condition, and such decision are not easy to
make. In this study, two roads were assessed since
they have a strategic role as a route to a tourist area.
On the Soe-Fatumnasi road, apart from this function,
it is also the only access from the interior to urban
areas. However, this study showed that the large
probability of this road segment compared to
Bakunase-Oenesu is only affected by pavement
damage and average daily traffic. Therefore,
policymakers need to determine the direction of the
maintenance program, in order to pay attention to
various criteria, and to obtain the best decision from
the several alternatives. Besides, sustainability issues
need to be considered, for example, by rewiring
asphalt pavements (Liem F, Nenobais O, Bria M and
Mata A 2019) in maintenance work.
However, in this study, the assumption of
respondents was that they fully understand the
maintenance and condition of the two sample roads.
This certainly has weaknesses because respondents
did not fully understand these problems. Moreover,
data collection was carried out by distributing
questionnaires and there was no dialogue. Therefore,
the data obtained did not completely reflect the actual
conditions.
4 CONCLUSION
Determining road maintenance priorities based on a
probability approach using the logistic regression
model, results in a different assessment form
compared to the priority determination based on
ranking or weight. The logistic regression showed
how possible it is for a road section to be included in
the maintenance program. The benchmark is the
amount of opportunity from an alternative, and it is
more appropriate when applied to planning at the first
level.
ACKNOWLEDGMENTS
The authors express their gratitude to the State
Polytechnic of Kupang, for funding this research.
REFERENCES
Zhao B, Soga K and Silva E 2019 Pavement degradation :
a city-scale model for Proceeding of the Institution of
Civil Engineers-Smart Infrastructure and Construction
vol 171 pp 93–109
Salih J, Edum-Fotwe F and Price A 2016 Investigating the
road maintenance performance in World Acad. Sci. Eng.
Technol. Int. J. civil, Environ. structral, Constr. Archit.
Eng. 10 395–9
Mikolaj J, Remek L and Margorinova M 2019 Road user
effects related to pavement degradation based on the
highway development and management tools Transp.
Res. Procedia 40 1141–9
Freitas E, Silva L and Vuye C 2019 The influence of
pavement degradation on population exposure to road
traffic noise Coatings 9
Muda A, Dumin L and Nahak P 2019 Analysis of the
Technocratic and Participatory Approach Models on
District Road Maintenance Proceedings of the 1st
International Conference on Engineering, Science, and
Commerce
Bhuva C R, Patel B and Kanam M 2019 Review on
pavement maintenance and prioritization by using
analytical hierarchy process Int. J. Tech. Innov. Mod.
Eng. Sci. 5 423–5
Tchemou G, Minsili L S, Mokotemapa A M, Eko R M and
Manguelle J H 2011 Prediction of flexible pavement
degradation: application to rutting in cameroonian
highways Electron. J. Geotech. Eng. 16 O 1301–19
Siswanto H, Supriyanto B, Pranoto, Putra Y A M and Huda
A S 2018 Evaluation of road maintenance priority using
PCI and road note 1 for Indonesian district roads AIP
Conference Proceedings vol 1977
Susanna A, Crispino M, Giustozzi F and Toraldo E 2017
Deterioration trends of asphalt pavement friction and
roughness from medium-term surveys on major Italian
roads Int. J. Pavement Res. Technol. 10 421–33
Ma J, Cheng L and Li D 2018 Road Maintenance
Optimization Model Based on Dynamic Programming
in Urban Traffic Network J. Adv. Transp. 2018
Bazlamit S M, Ahmad H S and Al-Suleiman T I 2017
Pavement Maintenance Applications Using Geographic
Information Systems Procedia Eng. 182 83–90
Li H, Ni F, Dong Q and Zhu Y 2018 Application of analytic
hierarchy process in network level pavement
maintenance decision-making Int. J. Pavement Res.
Technol. 11 345–54
Hayat E and Amaratunga D 2014 The Impact of the Local
Political and Socio-Economic Condition to the
Capacity of the Local Governments in the Maintenance
of Post-Disaster Road Infrastructure Reconstruction
Assets Procedia Econ. Financ. 18 718–26
Ahmed S, Vedagiri P and Krishna Rao K V. 2017
Prioritization of pavement maintenance sections using
objective based Analytic Hierarchy Process Int. J.
Pavement Res. Technol. 10 158–70
Marcelino P, Antunes M de L, Fortunato E and Gomes M
C 2019 Development of a multi criteria decision
analysis model for pavement maintenance at the
How to Determine Road Maintenance Priorities? An Analysis Using a Probabilistic Approach
1021
network level: Application of the MACBETH approach
Front. Built Environ. 5 1–10
Liem F, Nenobais O, Bria M and Mata A 2019
Development of The Use of Reclaimed Asphalt
Pavement and Plastic Waste in Hot Mix Asphalt for
Road Maintenance Proceedings of the 1st International
Conference on Engineering, Science, and Commerce.
iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science
1022