See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329337718

Developing of Pavement Maintenance and Management System for the Urban

Road Network by Calibrating the HDM-4 Distress Model

Conference Paper · November 2018

CITATIONS READS

0 29

2 authors:

Aditya Singh Tanuj Chopra

CSIR Central Road Research Institute Thapar University
2 PUBLICATIONS   0 CITATIONS    8 PUBLICATIONS   8 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Research paper Scopous indexed journal View project

All content following this page was uploaded by Aditya Singh on 01 December 2018.

The user has requested enhancement of the downloaded file.

 International Conference on Pavements and Computational Approaches (ICOPAC- 2018)

DEVELOPMENT OF PAVEMENT MAINTENANCE

MANAGEMENT SYSTEM FOR THE URBAN ROAD NETWROK
BY CALIBRATING THE HDM-4 DISTRESS MODELS

Author A Aditya, Singh Author B Dr. Tanuj, Chopra

Student Assistant Professor
Thapar Institute of Engineering and Technology Thapar Institute of Engineering and Technology
Patiala, India Patiala, India
adityasingh.singh44@gmail.com tchopra@thapar.edu

ABSTRACT

Increasing traffic load and insufficient funds for maintenance are the key factors which are responsible for
the continuously deteriorating condition of the Indian Road Network. This issue has led to a wide-scale
research in the area of pavement maintenance and management strategies for which tools like Highway
Development and Maintenance (HDM-4) have been developed. HDM-4 is a software package designed
by the World Bank which acts as a powerful system for analysis of road management and investment
alternatives and hence, is very useful in determining the right road network strategies. This tool has been
designed for use over a wide range of environmental conditions and in order to enable an HDM-4 model
to accurately predict the pavement performance for a specific location, it needs to be calibrated. This
study aims to calibrate an HDM-4 model to predict pavement deterioration due to structural cracking and
ravelling in a selected road section in Patiala (Punjab, India). The steps involved in calibration include
modelling the past patterns of pavement deterioration and comparing them with the present-day
measurement in order to make adjustments to the model. Data from 15 selected road sections of Patiala
was collected for three successive years, starting from the year 2012 to the end of the year 2014, to create
the model. Root Mean Square error and Coefficient of Determination were used for evaluating the
deviation between the observed and calculated values of pavement deterioration parameters. The
calibrated values were further validated, and it was deduced that the HDM-4 models predict the pavement
distress with high accuracy and can be further used for developing effective management and maintenance
strategies for the Indian urban road network.

Keywords: HDM-4, structural cracking, ravelling, pavement, urban road

1. INTRODUCTION

With increase in developmental activities in India, the road length has increased from 3.99 lakh kilometers
(as on 31st March, 1951) to 46.98 lakh kilometers (as on 31st March, 2011) [1]. Subsequently, the
accelerated increase in traffic load and the poor condition of the urban road network have caused a large-
scale deterioration in urban road pavements [2]. The amount of allocated funding resources is insufficient
given the increase in maintenance demands which is also responsible for deterioration in pavement
condition. This complex problem of matching resources, time, material, labor, equipment, design and
decision making can be taken care by a pavement maintenance management system (PMMS) [3] which is

3284-1
Development of Pavement Maintenance Management System

considered as an important strategy for minimizing road deterioration rates, elongating pavement life and
increasing efficiency in utilizing resources in proper way [4].
The development of pavement maintenance and management system requires modelling the pavement
deterioration pattern using the data collected over years which is done using highway development and
management tools like HDM-4. The tool attempts to build a regression model for the complex interaction
between the pavement structure, traffic loading and the environmental conditions for predicting the
various kinds of distress developed in pavements over time. Since the distress levels and the weather
conditions of the area surrounding the pavement network are highly correlated, these pavement distress
models need to be calibrated with respect to a specific location before they can be used for inferencing. [4].
Pavement engineers and researchers over the world have attained successful results in developing
pavement deterioration models using HDM, particularly for highway road networks [5]. In [6], De
Solminihac et al. highlight the novel “windows” methodology that was employed in their work that
involved calibrating various distress models for a road network in Chile using HDM-4. They also
presented a comprehensive evaluation of the models developed using HDM-4 and those developed using
HDM-III to clearly highlight the advantages of the former. Li et al. [7] calibrated the HDM-4 model for
developing a Pavement Management Study (PMS) for long term performance of the pavement for the
state of Washington. Deori et al. [8] depicted the advantages of using modified flexible pavements
containing crumb rubber and polymer modifier by calibrating and validating the HDM-4 model developed
for the same. In India, most of the development is constricted to highway road networks. One such novel
work has been presented by Chakarabati and Rawat [9] who has calibrated various distress models in
HDM-III for the Indian Highway in the state of Gujarat.
The brief summary of existing pavement deterioration models reveals that there are only a few pavement
deterioration models for urban road networks [10] and even lesser in case of developing countries like
India. Therefore, in the present activity we focus on developing pavement deterioration models for 15 road
sections in the city of Patiala (Punjab) in terms of cracking and ravelling, using the Highway Development
and Management tool (HDM-4).

2. ROAD NETWORK

In this study 15 of the 52 road sections in Patiala were selected for developing the pavement distress
models. The details of the 15 road sections selected for this study are given in Table 1. Each road section
has been assigned a Section ID (UR01, UR02, etc.) and includes the length of the road, its carriageway
width and its drainage conditions. The traffic volume data in terms of AADT (Average Annual Daily
Traffic) is given in Table 2. The 15 road sections are further classified into 4 Groups as shown in Table 3
based on pavement age and commercial traffic. Each Group consists of minimum 3 sections and a
maximum of 4 sections. Although Group 1 and 3, and Group 2 and 4 exhibit similar pavement age and
commercial traffic, they differ in values of modified structure number (MSN). The MSN of road sections
in Group1 is less than the MSN of road sections in Group3. Similarly, Group2 exhibited different
modified structure number as compared to Group 4. Therefore, Table 3 consists of four categories instead
of two.
Table 1: Inventory Data of Selected Road Sections

Section ID Name of Length Width Of Drainage Classification

Road in km Carriageway Condition
UR01 Thapar Univ- 0.80 6.80 Fair Other Road
Bhadson Road
UR02 Thapar Univ- 1.05 7.30 Fair Sub-Arterial
Bhupindra Road
UR03 Thapar Univ- 2.50 7.50 Good Other Road
Gurudwara Sahib
 International Conference on Pavements and Computational Approaches (ICOPAC- 2018)

(Table 1 continued)
UR04 Passey Road-Civil 1.00 7.20 Very Poor Other Road
Line
UR05 Gurudwara Sahib 2.25 7.00 Good Sub-Arterial
Chowk-Sirhind
Road
UR06 Leela Bhawan 0.70 11.50 Good Sub-Arterial
Chowk- Cantonment
UR07 Gurudwara Sahib 0.90 7.50 Poor Sub-Arterial
Chowk-Bus stand
Road
UR08 Thikriwala Chowk- 1.00 7.50 Good Sub-Arterial
Sangrur Road
UR09 Thikriwala Chowk- 0.80 11.80 Fair Other Road
Badungar Road
UR10 Bus Stand Chowk- 2.10 6.0 Poor Other Road
Gurbax Colony
UR11 Fountain Chowk- 0.70 12.5 Good Sub-Arterial
Leela Bhavan
UR12 Fountain Chowk- 2.25 7.5 Fair Other Road
Lower Mall
UR13 Thapar Univ- 2.25 7.30 Fair Sub-Arterial
Gurudwara Sahib
UR14 Leela Bhawan 2.10 7.50 Good Sub-Arterial
Chowk- 22 no
bridge
UR15 Leela Bhavan- 1.46 10.0 Fair Sub-Arterial
Gurudwara Sahib

Table 2: Traffic Volume Data

Section ID Motorised Non-Motorised Section Motorised Non-
AADT AADT ID AADT Motorised Year
AADT
UR01 12000.00 2365.00 UR09 13500.00 1430.00 2012

UR02 9000.00 1254.00 UR10 7800.00 989.00 2012

UR03 8000.00 980.00 UR11 14100.00 1350.00 2012

UR04 4300.00 890.00 UR12 13500.00 650.00 2012

UR05 14800.00 1650.00 UR13 15700.00 690.00 2012

UR06 11500.00 1106.00 UR14 11800.00 786.00 2012

UR07 7200.00 980.00 UR15 18500.00 1190.00 2012

UR08 12500.00 955.00

3284-3
Development of Pavement Maintenance Management System

Table 3: Homogenous Section Group

Homogenous Group Section Name Pavement Age Commercial Traffic

GROUP 1 UR08, UR10, UR15 0-6 Years Less than 7%

GROUP 2 UR02, UR04, UR06, UR12 6-12 Years More than 7%

GROUP 3 UR01, UR03, UR05, UR11 0-6 Years Less than 7%

GROUP 4 UR07, UR09, UR13, UR14 6-12 Years More than 7%

3. METHODOLOGY

The first step involved collecting enough required data on pavement deterioration for which extensive
field work was carried out for three successive years starting from 2012 to the end of the year 2014. The
field work involved identification of roads and collection of information like the road geometry details,
structure evaluation data (to check load carrying ability of the structure) and the functional evaluation data
(for checking pavement condition, etc.). Traffic data volume was also collected from the Municipal
Corporation, Patiala. This collected data were used to calibrate the HDM-4 model.

3.1 Structure Evaluation

Structure evaluation of the pavement was done to assess its load carrying ability. The magnitude of
deflection in pavement was measured through rebound deflection. The higher the rebound deflection, the
poorer is the structural capacity and performance of the pavement. The rebound deflection of the
pavement was measured by the Benkelman Beam method which has been elaborated in IRC 81[11]. The
procedure mentions that the deflection measurements should be taken at 35 and temperature corrections
should be applied when there is variation in the temperature. Benkelman Beam readings are also
influenced by the seasonal variation in climate and temperature, therefore corrections were also made to
the soil subgrade readings to make up for the deflection in these values. After applying corrections, the
characteristic deflection value was obtained as the sum of mean and standard deviation for corrected
rebound deflection for a particular section. The MSN (Modified Structure Number) was also calculated
from the BBD (Benkelman Beam Deflection) readings as: -

MSN = 3.2 ×BB𝐷-0.63 (1)

3.2 Functional Evaluation of Pavement

3.2.1 Cracking

Cracking is a form of distress which usually develops in flexible pavement. For measuring cracking, test
sections of 100m each were selected for each pavement and the area covered under the cracking distress
was marked in the form of a rectangular box with chalk and the marked area was measured with a
tape. The cracked area was measured as the percentage of the total area of the pavement. The cracked
developed in the road section UR04 is shown in Figure 1.
 International Conference on Pavements and Computational Approaches (ICOPAC- 2018)

Figure 1: Cracks on road section UR04

3.2.2 Ravelling

Ravelling is defined as loss of material (both binder and aggregate) due to wearing of the surface of the
pavement. Ravelling results in loose detritus of the pavement which leads to loss of skid resistance. The
ravelled area is measured in the form of geometric shapes and expressed as a percentage of the total area
of the pavement. Ravelled surface for section UR13 is shown in Figure 2.

Figure 2: Ravelling on section UR13

3.3 Calibration of HDM-4

For calibrating the HDM-4 model, the tool was run for all the selected pavement sections with traffic
volume and pavement condition data as the input. The calibration factor was obtained by determining
minimum RMSE and maximum R2 calculated by using two equations given below: -

(2)

(3)

where RMSE = root mean square error, R2 =coefficient of correlation, Ob=observed value of distress,
Pd=predicted value of distress by HDM-4 model, Oavg= average observed value of distress, N= no of
observations
A two-stage process was used for calibrating the model. The calibration factor obtained after the first stage
was used with an increment of 0.01 during the second stage of training and calibrating the prediction

3284-5
Development of Pavement Maintenance Management System

model. The pavement data obtained for the year 2012 were used as input dataset for calibrating the HDM-
4 model and the prediction made by the model for the year 2013 was compared with the actual measured
value of the selected pavement section. For different groups consisting of homogeneous sections, different
calibration factors were obtained due to varying traffic and different pavement structure. The calibration
of cracking distress model of Group 4 is shown as an example in Table 4.

Table 4: Calibration for cracking distress model for Group 4

Section UR 07 UR09 UR13 UR14

Calibration Obsv Pred Obsv Pred Obsv Pred Obsv Pred
Factor
0.20 9 9.18 12 10.46 11 9.16 6.0 4.58
0.30 9 9.77 12 10.91 11 9.75 6.0 4.95
0.35 9 10.07 12 11.41 11 10.05 6.0 5.13
0.36 9 10.13 12 11.48 11 10.11 6.0 5.17
0.37 9 10.18 12 11.54 11 10.16 6.0 5.20
0.38 9 10.24 12 11.61 11 10.22 6.0 5.24
0.39 9 10.30 12 11.67 11 10.28 6.0 5.27
0.40 9 10.36 12 11.73 11 10.34 6.0 5.31
0.41 9 10.42 12 11.80 11 10.40 6.0 5.35
0.42 9 10.48 12 11.86 11 10.46 6.0 5.38

The calibrated factors for the cracking distress model, as shown for Group 4 in Table 3, were obtained by
minimizing the RMSE and maximizing R2. The minimum RMSE (0.837) and maximum R2 (0.883) were
achieved at calibration factor of 0.40. Similarly, for ravelling distress model the calibrated factor was
obtained as 0.16 by keeping minimum RMSE value of 1.03 and maximum R2 value of 0.889. In this way
calibrated factors for all groups were determined and are shown in Table 5.

Table 5: Final Calibrated Value for all Road Sections

Distress Model Group 1 Group 2 Group 3 Group 4

Crack Progression 0.5 0.4 0.46 0.4
Ravelling Progression 0.09 0.14 0.13 0.16

4. RESULTS AND DISCUSSIONS

The results obtained by HDM-4 distress model after calibration were validated using the data from year
2014 for different group sections to check the reliability and efficacy of the calibrated HDM-4 distress
model. The validations were done using regression analysis and percentage variability. Validations were
done for all groups consisting of homogeneous sections and validation of Group 4 consisting of 4 sections
is shown as an example:

4.1 Validation of the Cracking Distress Model (Group 4)

For validation of the model, the measured value of the cracking progression was compared with the value
obtained by the distress model for year 2014 and the variability in the percentage was determined, as is
shown in Table 6. These values were also plotted for determining correlation between them as shown in
Figure3. The variability in percentage obtained between the two values (measured and predicted) of
distress lies between 5 to 22% which is quite reasonable, and the coefficient of relation obtained was close
 International Conference on Pavements and Computational Approaches (ICOPAC- 2018)

to 1 which tells the efficacy and dependability of the calibrated HDM-4 distress model.

4.2 Validation of the Ravelling Distress Model (Group4)

For validation of the model, the measured value of the ravelling progression was compared with the value
obtained by the distress model for year 2014 and the variability in the percentage was determined which is
shown in Table 7. These values were also plotted for determining correlation between them as shown in
Figure 4. The variability in percentage obtained between the two value (measured and predicted) of
distress lies between 13 to 26% which is quite reasonable, and the coefficient of relation obtained was
0.9736 which is close to 1 which tells the efficacy and dependability of the calibrated HDM-4 distress
model.
Table 6: Cracking validation for Group 4 Table 7: Ravelling validation for Group 4
Section Observed Predicted Variability R2 Section Observed Predicted Variability R2
UR07 11 11.55 5% 0.997 UR07 11 12.63 14.81% 0.973
UR09 14 15.48 10.57% UR09 9 11.32 25.77%
UR13 13 13.92 7.07% UR13 14 17.38 24.14%
UR15 7 5.5 21.42% UR15 8 9.04 13%

Figure 3: Graph for Cracking validation model Figure 4: Graph for Ravelling validation model

4.3 Pearson Chi Square Test

Pearson chi square test was performed for determining the goodness of fit which establishes how well the
distress values, as determined by the calibrated HDM-4 model, fit the actual or measured values. The
formula used for finding the goodness of fit is: -

(4)

n= no of observations Ob= measured distress value Ed= predicted value by calibrated HDM-4 model
The value of X2 calculated for the distress model for all the groups were compared with the X2 critical
value with level of significance 5% and degree of freedom (n-1). The test was performed for all the groups
and the Group 4 is shown as an example in Table 8.

Table 8: Pearson Chi Square Test

Distress Model Linear Regression X2calculated X2critical DOF R2
Cracking Progression y = 1.368x-1.866 0.6375 7.815 3 0.997
Ravelling Progression y = 0.804x+2.55 1.4628 7.815 3 0.973

3284-7
 Development of Pavement Maintenance Management System

It was found that the value of chi square was less than the critical values, therefore null hypothesis was
accepted, and a notable relationship is developed between the distress values having no statiscal
difference.

5. CONCLUSION

• The two HDM-4 distress models that were calibrated for cracking and ravelling obtained a good
correlation coefficient between the observed and predicted value of distress. This validates their
use as standard models for the purpose of pavement maintenance, recovery and overall
management in cities like Patiala that have homogenous traffic and climatic conditions prevailing
in most regions.
• The percentage of variability for cracking distress model ranges from 2 to 18% and for ravelling
distress model ranges from 2 to 10% which is quite reasonable considering the variability in
pavement age, traffic and climatic conditions.
• The calibration factor of distress model ranges from 0.4 to 0.5 for cracking distress model, and
0.09 to 0.16 for ravelling distress model. The ranges distress model is quite rational for flexible
pavement which are flexible in urban city road network of Patiala, Punjab, India.

REFERENCES

1. MoRT&H, Road Transport Year Book 2010/2011. Transport Research Wing, Ministry of Road
Transport and Highways, Government of India, New Delhi, 2012.
2. Chopra, T., Parida, M., Kwatra, N. and Chopra, P. (2018). Development of Pavement Distress
Deterioration Prediction Models for Urban Road Network Using Genetic Programming. Advances
in Civil Engineering, 2018, pp.1-15.
3. Chopra, T., Parida, M., Kwatra, N. and Mandhani, J. (2017). Development of Pavement
Maintenance Management System (PMMS) of Urban Road Network Using HDM-4
Model. International Journal of Engineering & Applied Sciences, 9(1), pp.14-14.
4. Bannour, A., El Omari, M., Lakhal, E., Afechkar, M., Benamar, A. and Joubert, P. (2017).
Optimization of the maintenance strategies of roads in Morocco: calibration study of the
degradations models of the highway development and management (HDM-4) for flexible
pavements. International Journal of Pavement Engineering, pp.1-10.
5. Jain, S.S., Aggarwal, S. and Parida, M., 2005. HDM-4 pavement deterioration models for Indian
National Highway Network. Journal of Transportation Engineering, 131 (8), 623–631.
6. De Solminihac, H., Hidalgo, P. and Salgado, M. (2003). Calibration of Performance Models for
Surface Treatment to Chilean Conditions The HDM-4 Case. Transportation Research Record:
Journal of the Transportation Research Board, 1819, pp.285-293.
7. Li, Jianhua, et al., 2005. The Highway Development and Management System in Washington
State: Calibration and application for the Department of Transportation Road Network.
Transportation Research Record, Journal of the Transport Research Board, 1933, 53–61.
8. Deori, S., Choudhary, R., Tiwari, D. and Gangopadhyay, S. (2016). Calibration of HDM-4 models
for Indian conditions of flexible pavement having modified bitumen in wearing
course. International Journal of Pavement Engineering, pp.1-14.
9. Chakrabarti, S. and Rawat, M.S., 1995. Highway design & maintenance standard model (HDM):
calibration and adaptation to Indian conditions. Journal of Indian Roads Congress, 56 (1), 75–101.
10. Shah, Y., Jain, S., Tiwari, D. and Jain, M. (2013). Development of Overall Pavement Condition
Index for Urban Road Network. Procedia - Social and Behavioral Sciences, 104, pp.332-341.
11. Guidelines for strengthening of flexible road pavements using Benkelman beam deflection
technique (First Revision) Indian Road Congress, New Delhi, India, 1997.

View publication stats