Development of a Pavement Maintenance System : A Case Study

Ashoke K. Sarkar, Ph.D Professor of Civil Engineering Birla Institute of Technology and Science, Pilani
and Amarendra K. Sandra, Ph.D. GMR Group, Hyderabad

BITS Pilani
Pilani Campus

Need for the study

Functional condition continuously due to of highway pavements deteriorates

o repeated loads of vehicular traffic o environmental factors such as temperature, snow and frost

Number of countries have failed to maintain the roads properly not only due to lack of funds also due to absence of simple maintenance management system to use the funds effectively Though number of maintenance management systems are available in the form of comprehensive packages, but most of them require huge database.
o History of pavement, which might not be easily available in India o Collection of data at frequent intervals

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 There is a need to develop a system which follows the existing methods used for maintaining the roads by the Public Works Departments (PWD) in various states. It is also essential to develop a simple and user friendly software for maintaining the flexible pavement stretches by incorporating the prioritization, grouping techniques and roughness model.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Objectives of the Study

To establish the relationship between road roughness with the causative distress parameters To prioritize the pavement stretches using appropriate prioritization technique To identify the homogenous road stretches through a clustering technique. To develop a practical network level pavement management strategy using the outcome of the above stages and develop a user friendly computer interface
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Methodology
Methodology of the present study has been explained in the following steps Step 1: Selection and Identification of Pavement Stretches Step 2 : Pavement Condition and Pavement Roughness Surveys Step 3: Development of Relationship between Roughness and Pavement Distresses
Determination of natural or Initial Roughness Development of Model between Roughness and Distresses

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Step 4: Prioritization and Grouping of Pavement Stretches Prioritizing using Fuzzy Approach Classifying the stretches into Manageable Number of groups Step 5: Development of a decision-making Software Finding Rank, Predicting the Roughness and Grouping Maintenance of Stretches Under Budgetary Constraints Summary of Budget, Final Roughness and PSI Values
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Case study in Rajasthan

In total 858 stretches of 50m each were chosen from NH, SH and MDR in Rajasthan Distresses considered: cracking, pothole, ravelling, patching, rutting and edge failure Each distress was divided into three categories low, medium and high depending on severity.

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Weight on the distresses: Expert Opinion Survey

Distress parameters which affect the condition of the pavement might not have same impact; hence, an expert opinion survey was conducted to ascertain the weights of the pavement distress parameters As it is difficult to express the weights in quantifiable terms experts were asked to send the responses in linguistic variables such negligible, low, moderate, high and very high Questionnaire along with photographic clues and cues has been prepared and send to number of experts having field and academic background across India

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Roughness model for predicting the roughness from pavement distress parameters
The following steps involved in developing model Standardization of roughness values obtained at different operating speeds Minimum UI value on a virgin road surface An average initial UI value of 1750 mm/km on NHs, 2025 mm/km on SHs and 2225 mm/km on MDRs Converting the UI into International Roughness Index (IRI) IRI = 0.0032 (BI) 0.89 Where, IRI = International Roughness Index (m/km) BI =UI obtained from BI (mm/km)
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Normalization of the Distress Parameters

PAD = (AD *100/ A) Where, PAD = Percentage Area of Distress parameters AD= Area of Distress parameter, A = Total area of the pavement considered Only rutting is expressed in terms of meter per km as it follows a linear path. Model development Out of 858 data points 770 (aprox 90%) used for model development and remaining 10% used for validation
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 IRID (m/km) = 0.0148*RL+ 0.0219*RM+ 0.0343*RH+ 0.0307*PAL+ 0.0425*PAM + 0.0437*PAH+ 0.1089*PL+ 0.1392*PM+ 0.1631*PH +0.0113*CL +0.0179*CM+ 0.0310*CH+ 0.0017*RUL+ 0.0024*RUM +0.0034*RUH (r = 0.954) IRI (m/km) =A+ IRID (m/km) A = 2.4 for National Highways = 2.8 for State Highways = 3.0 for Major District Roads

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Model validation
Absolute Percentage of Deviation= Mean Absolute Percentage Deviation (MAPD) was calculated The MAPD between observed and predicted IRI value was 8.08, which is quite acceptable considering the wide range of data collected form different stretches of roads 7
6

Predicted IRI (m/km)

5 4 3 2 1 1 2 3 4 5 6 7 Measured IRI (m/km)

Plot between Measured and Predicted Unevenness Index values

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Prioritization and Grouping of Pavement Stretches for Maintenance

In most of their studies weights of various distress parameters are directly expressed in quantifiable terms there by neglecting the uncertainty Uncertainty was quantified by researches through Approach and in the present study also it was used Fuzzy

For a large network instead of individual prioritization, grouping is more useful Grouping was done by using Clustering technique after determining the number of optimum clusters.
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Ranking of Pavement Stretches in Network of Roads

In case of network level maintenance, the stretches would be from different functional class of highway. Hence, high preference should be given to the highways based on their importance and traffic Accordingly, the fuzzy prioritization index would be multiplied with the prioritization factor as presented in table (next slide)

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Functional Class

Level

No. of Commerci al Vehicles >5000 3000-5000 <3000 >3000 1500-3000 <1500 >1500 500-1500 <500

Prioritization Factor (F) 1 0.9 0.8 0.8 0.75 0.7 0.75 0.7 0.6

Code

High Express way/ National Highway Medium Low High State Highway Medium Low High Other Roads Medium Low

1 2 3 4 5 6 7 8 9

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Development of Software for Pavement Maintenance at Network Level

An user friendly software package was developed in visual basic environment for maintenance of the flexible pavements particularly for the Public Works Department (PWD). The user has the option of taking policy decisions Application of software has been explained with an example by taking 45 stretches from different functional class of highways

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Demonstration of Software With an Example As an example, a total of 45 stretches comprising 15 each from three functional classes of highways such as Major District Roads, State Highways and National Highways were selected 15 uniform sections of 100 m length each. Pavement distress parameters such as cracking, ravelling, potholes, patching, rutting and edge failure data in three severity levels were collected on all these sections Along with the pavement condition, traffic census data was also collected

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Grouping the Stretches

All the stretches were grouped into optimum number of groups They were grouped into 5 clusters

Stretches In Various Groups after Iteration 1

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Different types of treatments are available for repairing the pavement distresses. In the present study the treatments suggested in the Guidelines for Maintenance of Flexible Pavements (IRC: 82-1982) were chosen
S.No Type of distress Cracking Severity Level Low 1 Medium High Low 2 Raveling Medium High Low 3 Potholes Medium High Treatment Bitumen binder having low viscosity Bitumen binder /Slurry seal / Fog seal Slurry seal/Bituminous premix patching More quantity of binder Cutback bitumen covered with coarse sand / slurry seal Premix Carpet Premix Open Graded / dense /penetration macadam patching graded patching

Premix Open Graded or dense graded patching/ macadam patching Premix Open Graded or dense graded patching / macadam patching
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Unit Costs for Various Types of Treatments as Per Basic Schedule Rates (BSR)
S.No Type of Failure Severity Low 1 Cracking Medium High Low 2 Raveling Medium High Low 3 Potholes Medium High Treatment Selected Spraying Bitumen Binder or Tack Coat Fog Seal Sand Bituminous Premix Patching Fog Seal Sand Bituminous Premix Patching Slurry Seal Premix Dense Graded Patching of 20 mm thick with seal coat Premix Dense Graded Patching of 50 mm thick with seal coat Penetration Macadam of 75 mm thick with seal coat Unit Sq.m Sq.m Sq.m Sq.m Sq.m Sq.m Sq.m Sq.m Sq.m Cost Rs. 6.5 9 12 9 12 18 85 145 205

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 Maintenance of pavement stretches

Budget was selected between 1 increment of 1 million. to 5 million Rupees with a uniform

Initially, a budget of Rupees 1 million was considered and different maintenance options were selected and the corresponding change in roughness levels were studied. As the potholes have very high contribution on roughness as well as they cause discomfort to the road users, it was decided to repair the potholes of all the three severity levels on all three stretches. The fund required were Rupees 0.12 million. Since, funds were still available, the steps were repeated until it is exhausted.
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 This process would be continued until the budget gets exhausted. In this example available budget of Rupees 1 million was used up after seven iterations After each iteration, different types of maintenance options were implemented. The above discussion was based on an assumption that the available budget was rupees 1 million. To study the change in pavement roughness in each stretch in a network when the allocated fund was varied between rupees 1 to 5 million with an uniform increment of rupees 1million, predicted roughness in all the stretches with different budgets were determined
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Change in Roughness Levels under Different Budget Scenario

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

 It might be noted from the graph that when the budget was rupees 3 million thousand the roughness was less than 4 m/km in all the stretches except in one stretch. When the budget was raised to Rupees 4 million, the roughness was equal to or less than 3 m/km in all the stretches. However, to bring the roughness to its initial roughness of 3 m/km on stretches of MDR, 2.8 m/km on stretches of SH and 2.4 m/km on stretches of NH, there was a need to spend about rupees 4.2 million for all the 45 stretches

BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Conclusions
The initial roughness values on freshly laid pavements vary depending on the kind of highway. On the National Highway it was 2.4 m/km, on State Highway it was IRI=2.8 m/km and on Major District Roads it was IRI=3.0 m/km The relationships developed between the roughness and pavement distresses in this study could be used to predict the current roughness on any pavement having similar distress patterns The wheel paths are quite predominant on NHs whereas it is difficult to identify on MDRs and few SHs The software developed in the study can be used to take the maintenance decisions so as to distribute the available funds in a logical manner. It also provides the user to select the most appropriate policy decision for maintaining the pavement network.
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956

Thank you
BITS Pilani, Deemed to be University under Section 3 of UGC Act, 1956