1

CHAPTER 1 1.0 INTRODUCTION

1.1 GENERAL

Development of road infrastructure is currently being given high priority by the government

of India to (i) meet the requirement of growing travel demand and (ii) help the growth of

economic activity at a faster rate. Construction of divided four and six-lane highways under

the National Highway Development Programme (NHDP), presently in progress, aims at

connecting (a) four metropolitan cities namely Delhi, Kolkata, Chennai and Mumbai forming

the golden quadrilateral and (b) Srinagar to Kanyakumari and Silchar to Porbandar,

constituting the North-South and East-West corridors. The Pradhana Mantri Gram Sadak

Yogana (PMGSY), launched recently aims at providing all-weather connectivity to the

villages of India by 2007. It is obvious that the highway infrastructure that has been created

at a great cost needs to be evaluated on a regular basis to assess the requirement of

rehabilitation measures. It is important to adopt a rational approach for the evaluation of

the pavements so that more efficient use of materials can be made to improve the

pavement performance and lower the life cycle cost.

1.2 BACKGROUND OF THE RESEARCH PROJECT

It is during the last three decades that the approach to design of flexible pavements has

begun to undergo transformation from empirical method to mechanistic method because of

the improved understanding of the behaviour of materials and the availability of analytical

tools for the analysis of pavements. Examples of some of the popular analytical design

methods currently used in different countries are those developed by SHELL [1978], Asphalt

Institute [1981] and Austroads [1992]. The AASHTO [1993] guideline for design of

Pavement structures is being replaced by a new guideline [Development of 2002 Guide,

2003], which uses a mechanistic approach for design and rehabilitation of pavement

structures. In India also, a massive pavement performance study was undertaken during

1983 to 1993 [Research Schemes R-6, 1995; R-19 and R-56, 1999] as a result of which a

new standard for design of flexible pavements was published by Indian Roads Congress

[IRC: 37, 2001]. The new design method uses a mechanistic approach for the

determination of design pavement thicknesses.

Properties of different pavement layers are essential inputs for mechanistic pavement

design. These properties can be obtained by conducting laboratory tests under near-field

2

conditions on representative samples of pavement materials. The properties of these

materials can also be estimated from empirical relationships developed from field evaluation.

Alternatively, realistic values of layer moduli can be obtained from the structural evaluation

of in-service pavements. For this purpose, nondestructive testing [NDT] techniques are

being popularly used all over the world.

The Indian Roads Congress [IRC: 37,2001] guidelines for design of flexible pavements

recommend the use of different models for estimating the moduli of subgrade and granular

layers. Typical moduli obtained from extensive laboratory investigations [Road Research

Scheme R-56, 1999] for various types of bituminous mixes are frequently used for analysis

of pavements in India. The main concern among the researchers in India in using the

empirical relationships recommended by IRC is that there has not been any validation of the

relationships for the specifications and construction practices adopted in India. Thus, it is

essential to have adequate data for selection of realistic layer moduli appropriate for the

conditions prevailing in India. Hence, a rational approach for predicting the pavement layer

moduli is desirable for the analytical design of pavements and overlays.

As far as the structural evaluation of in-service pavements is concerned, since its

development in 1953 [Zube and Forsyth, 1966], Benkelman Beam became a standard tool

used by several agencies for nondestructive testing of pavements. Indian Roads Congress

(IRC) recommends the evaluation of in-service pavements using the Benkelman beam for

design of flexible overlays. In the IRC design method [IRC: 81, 1997], the measured

pavement deflections, corrected for standard temperature and moisture, are used to

determine the required overlay thickness. As only one surface deflection is measured using

this equipment, it is not possible to get sufficient information regarding the structural

condition of different layers of the pavement. Thus, this method does not permit a reliable

prediction of the performance of pavements. With the advances made in the mechanistic

approach, some attempts were made in India [Reddy and Pandey, 1994; Road Research

Scheme R-56, 1999] to incorporate mechanistic principles in overlay design procedure.

For the mechanistic design of an overlay, the properties of the existing pavement layers can

be evaluated in the laboratory by taking cores from the field. The remaining life of the

pavement and the requirement of overlay thickness can be determined using mechanistic

approach. A more rational approach is to carryout structural evaluation of in-service

pavements by nondestructive testing of pavements, which is quick and causes least the

disruption to the traffic.

Chapter 1 Introduction

3

A number of NDT equipments have been developed during the last three decades for

evaluating in-service pavements. Among them, FWD is considered to be the most

appropriate since it simulates the short duration loading of a moving wheel. Since six or

more deflections are measured by the FWD, it is possible to explain the structural behaviour

of pavements more accurately. The deflections measured by the FWD can be used for

backcalculating the pavement layer moduli, which in turn, can be used for the analysis and

estimation of the remaining life of the pavement and for determination of the requirement

for overlay.

Realistic data for moduli of different layers of highway pavements in India are not available

currently and hardly any investigation was made on the variation of layer moduli with

season. With the adoption of analytical approach for design of flexible pavements in India

[IRC: 37-2001], it has become necessary to develop a proper pavement evaluation system

for estimating pavement layer moduli based on field evaluation of Indian Highways. The

models adopted in the Indian Roads Congress guidelines for design of flexible pavements

for the estimation of elastic moduli are based on pavement performance studies during

1985 and 1993 and must be re-examined because of use of better specifications in the

construction of Highway pavements in India.

FWD is the most suitable equipment for pavement evaluation as indicated earlier. Though

different types of FWDs are available commercially [Irwin, 2002], the high cost of the

imported FWDs is making it difficult for most of the agencies in India to use them. Hence,

the present study is aimed at the development of an FWD at a low cost and evaluating

some in-service and new pavements. It is also necessary to develop software for estimating

the effective pavement layer moduli from measured deflections and to suggest overlay

design procedure by incorporating mechanistic principles. Using the field data, it is proposed

to develop models for estimating the moduli of different layers of the pavement.

In the light of the discussion presented in the preceding paragraphs, the objectives of the

research scheme R-81 are identified as given below.

1.3 OBJECTIVES OF THE RESEARCH SCHEME

In view of the demand for adoption of mechanistic approach in the pavement design and

evaluation, Transportation Engineering Section of Civil Engineering Department, Indian

Institute of Technology, Kharagpur has taken up research scheme (R-81) Structural

Evaluation of Pavements in Eastern India using Falling Weight Deflectometer sponsored by

the Ministry of Road Transport & Highways, Government of India at a cost of Rs.22.07

lakhs. Following were the terms of reference.

4

i. To modify the existing Falling Weight Deflectometer (FWD) developed in-house in the

Transportation Engineering laboratory, Civil engineering Department, Indian Institute of

Technology, Kharagpur.

ii. To review available literature on the structural evaluation of pavements using FWD,

various procedures available for the backcalculation of pavement layer moduli and

different pavement design procedures in vogue.

iii. To evaluate the structural condition of selected Highways in Eastern India using

modified FWD.

iv. To develop a computer program for backcalculating the effective pavement layer moduli

using the FWD evaluation

v. To develop methodology for design of overlays using FWD evaluation.

1.4 EXTENSION OF THE SCOPE OF THE OBJECTIVES

Besides the objectives mentioned in the terms of reference of the research scheme, it was

necessary to extend the scope of the work for better understanding the material behaviour

under varying climatic conditions. The extended objectives of the research scheme are as

follows.

As a part National Highway Development Programme (NHDP), National Highway NH-6 was taken up for widening and strengthening to have a four-lane divided carriageway.

The four-lane pavement consists of the existing carriageway strengthened to have a

two-lane carriageway and a new two-lane carriageway. Some of the newly constructed

pavement sections on NH-6 were selected for FWD evaluation during different stages of

construction.

Bituminous layer modulus changes with temperature. No studies are conducted to

findout the effect of temperature on layer moduli. Keeping this in view, some of the

newly pavement sections on NH-6 were considered for evaluation under different

temperatures.

Recycling of bituminous layers is a recent practice in Indian paving industry. In one of

the projects being implemented on National Highway-6, which is very near to

Kharagpur, the top portion of the damaged bituminous surfacing was being milled and

recycled in some stretches. Considering that there is hardly any experience in India with

recycled pavement layers, the pavement stretch was selected for evaluation with FWD

before and after recycling.

Chapter 1 Introduction

5

1.5 IMPLEMENTATION OF THE RESEARCH SCHEME

Transportation Engineering Section of Civil Engineering Department, IIT Kharagpur was the

organization, which has executed the research scheme R-81 sponsored by Ministry of Road

Transport & Highways (MORT&H). The scope of the research project was extended at

various stages to bring the completeness to the present research scheme.

Mr. M. Amaranatha Reddy, a full time Junior Project Officer, was appointed for the

research scheme R-81 to carryout research. Improvements to the existing FWD, data

collection, analysis, reports preparation of the research scheme have been carried out by

the project officer under the guidance of the Principal and Co-Principal Investigators of the

research scheme (R-81) at the Transportation Engineering Section of Civil Engineering

Department, IIT Kharagpur.

1.6 PREVIOUS TECHNICAL REPORTS AND RESEARCH DIGEST

In the first technical report submitted to the MORH&H [Technical Report-I, December 1999]

review of various back-calculation procedures was presented. Details of the Falling Weight

Deflectometer developed in the Transportation Engineering Section through a number of in-

house projects, were also given. Deflection data collected in September 1999 using the

semi-automated FWD on a number of pavement sections situated close to Kharagpur were

presented.

The second technical report [Technical Report-II, June 2000] contains the details of the test

sections selected on different highways in the states of Orissa, West Bengal and Bihar.

Deflection data collected using the semi-automated FWD during March 2000 was presented.

Detailed drawings of the proposed automated in-house FWD were also given.

Salient features of the Genetic Algorithm (GA) based backcalculation program (BACKGA)

developed for the backcalculation of pavement layer moduli were presented in the third

technical report [Technical Report-III, February 2001]. Also, the salient features of the new

In-vehicle automated Falling Weight Deflectometer developed by the Transportation Section

of Civil Engineering Department, IIT, Kharagpur were discussed. Some photographs of the

equipment were given.

The fourth technical report [Technical Report IV, April, 2002] has the research work carried

out for the selection of Genetic Algorithm parameters for backcalculation of pavement layer

moduli using Genetic Algorithms. Also presented in that report was the data collected from

the structural evaluation of some of the pavement sections on National Highways, state

6

Highways and major district roads using modified FWD designed and developed by

Transportation Engineering Department, IIT, Kharagpur. The deflection data collected

during winter (2001) was also presented. Soft copy of the BACKGA program was also

enclosed (in a floppy) along with this technical report for analyzing three layer pavement

systems.

In the, fifth technical report, being submitted to MORT& H [Technical Report-I, July 2002]

contains the deflection data collected during summer reason on selected pavement sections

on different highways using the modified FWD. Also analysis of the data for the estimation

of backcalculation of pavement layer moduli using BACKGA program was also presented.

Various models developed to predict the layer moduli were also included along with some

conclusions.

Research digest [Research Digest, August 2002] contains the salient features of a FWD

system consisting of a Falling Weight Deflectometer, Data Acquisition system and analysis

carried out using backcalculation software developed by the Transportation Engineering

section of IIT, Kharagpur for evaluation of highways in India. Various models developed for

estimating layer modulus values from different pavement parameters was also reported in

the research digest.

1.7 ORGANISATION OF THE REPORT

The various chapters of the report have been organised in the following manner.

The first chapter gives an introduction and objectives of the research project. The second chapter deals with the review of relevant literature related mostly to various

methods of nondestructive testing of pavements, backcalculation techniques, pavement

material characterization including models available for the estimation of pavement layer

moduli and FWD based overlay design procedures.

In the third chapter, the details of the improvements made to the existing Falling Weight Deflectometer are presented.

Chapter four gives the details of the structural evaluation carried out on in-service pavement (old and new) sections.

Salient features of BACKGA, a genetic algorithm program for backcalculation of effective moduli of pavement layers, are discussed in chapter five. The method adopted for

selection of the GA parameters is also presented in this chapter.

Chapter six contains the details of backcalculation analysis for the deflection data collected on both in-service and new pavements. Various models developed for the

Chapter 1 Introduction

7

estimation pavement layer moduli from different parameters are included in this

chapter.

Seventh chapter contains the proposed flexible pavement overlay design methodology based on FWD evaluation.

Conclusions drawn from the present investigation and scope for further research is presented are given in Chapter eight.

In addition to the above, an user friendly executable program BACKGA for the

estimation of effective layer moduli of the pavement system is also included in the

report on a floppy.

8

CHAPTER 2 2.0 LITERATURE REVIEW 2.1 INTRODUCTION

A Large sum of money is being invested in India for the construction of expressways, and

highways. These facilities need to be evaluated periodically in terms of their functional and

structural performance to assess the requirement for maintenance and rehabilitation

measures. The methodology to be adopted, especially for the structural evaluation of

pavements, should have a rational basis and also be compatible with the current design

trends and practices. Since the present research is aimed at the development of a method

for structural evaluation of pavements, relevant literature on various commonly used

pavement evaluation techniques has been reviewed with emphasis on impulse loading

equipment such as the Falling Weight Deflectometer (FWD). As a large portion of the

construction activity on highways in India involves flexible pavements, the review has been

confined to the work relevant to flexible pavements. The review covers various models used

for the analysis and interpretation of the data obtained from structural evaluation of

pavements. Different models available for the selection of properties of pavement layers,

including those developed from the evaluation of in-service pavements, have been

examined. Some mechanistic methods currently in use for the design of new flexible

pavements and overlays have also been reviewed. The following sections of this chapter

present an overview of different structural evaluation methods, backcalculation techniques,

selection of appropriate material properties for the analysis of pavements and some

mechanistic methods of design of flexible pavements and overlays to know the current

practices adopted around the world.

2.2 STRUCTURAL EVALUATION OF PAVEMENTS

Structural evaluation of pavements commonly involves applying a standard load to the

pavement and measuring its response. The response measured can be stress, strain or

deflection. The most commonly measured response is deflection. Benkelman has been

among the earliest equipment used for structural evaluation of pavements. Since its

development in 1953, the Benkelman Beam has become a standard tool used by several

agencies for nondestructive testing of pavements. Significant developments have taken

place since then in the equipment used and the analytical tools adopted for the evaluation

of pavements. The following paragraphs deal with some Nondestructive Testing (NDT)

6

equipment used for pavement evaluation. Depending on the duration of the load applied,

these equipment are broadly classified under two categories- a) static and b) dynamic.

2.2.1 Static / Creep Loading Equipment

In this category, either a static or a slow-moving load is applied to the pavement surface

and the resulting deflections are measured at one or more locations. Plate load testing,

deflection measurement using equipment such as Benkelman beam, Double Benkelman

beam, Multiple Benkelman beam, Modified Benkelman beam, Lacroix deflectograph,

Traveling deflectometer, etc., can be considered under this category.

Benkelman Beam [Zube and Forsyth, 1966] is a 3.66 m long, portable instrument used to

measure surface deflection of the pavement loaded by the rear axle of a standard truck.

The main disadvantage with this equipment is that the support legs of the beam often lie

within the deflection basin, which affects the measured deflections. Also, a single deflection

does not give adequate information about the condition of various layers of the pavement.

Double, multiple and modified Benkelman beams have been used to measure deflections at

different radial distances under static loading condition. The Lacroix Deflectograph

[Nondestructive Testing- Lacroix Deflectograph, 2003] is essentially a truck-mounted

Benkelman Beam, which moves forward with the vehicle. Testing with this equipment is

faster compared to Benkelman beam. The Traveling Deflectometer [Zube and Forsyth,

1966] developed by the California division of highways has dual probes to simultaneously

measure the deflections between each set of dual wheels. CEBTP Curviameter [Paquet,

1978] is another device that operates on the principle of Benkelman beam and measures

not only the pavement surface deflections, but also the radius of curvature of the pavement

deflection bowl, which is more useful for evaluating the pavement strength.

Though deflection measurement under static load is simple, it does not simulate the loading

conditions produced by a moving vehicle in pavements. The evaluation of pavements by

such methods is, in general, slow.

2.2.2 Dynamic Loading Equipment

Two types of devices are, in general, considered in this category. While vibratory loading is

produced in one category of equipment, the other category consists of impulse loading

equipment. Dynaflect, Heavy Vibrator and Road Rater are some of the vibratory equipment

used for pavement evaluation. Falling Weight Deflectometer (FWD), Loadman Portable FWD

and Rolling Weight Deflectometer (RWD) fall into the category of impulse equipment.

Chapter 1 Introduction

7

Dynaflect pavement testing device [Scrivner et al, 1966] produces sinusoidal vibration at a

frequency of 8 Hz. It is fitted with five velocity transducers (geophones), each spaced 305

mm apart. The output from the transducers is integrated to measure pavement deflection.

The use of the Shell heavy vibrator for pavement evaluation was reported by Heukelom and

Foster [1960], Heukelom and Klomp (1962), Nijboer and Metcalf [1962], and Jones et al

[1967]. In this method, the modulus of elasticity of each layer can be computed from the

wave velocity and wavelength for a spectrum of frequencies of oscillation. In Road Rater

[Hoffman and Thompson, 1982], a dynamic force is applied by a steel mass accelerated by

a servo-controlled hydraulic actuator. Deflections are measured using four or more

transducers. Load magnitudes vary for different models. Road Rater is available as trailer

mounted and in-vehicle models.

Though the vibratory equipment are useful for structural evaluation of pavements, they are

not very popular because of certain limitations. In the case of Dynaflect, the maximum

peak-to-peak force that can be applied is 1000lb. Magnitude and frequency of load cannot

be varied. The main drawback of Heavy vibrators is that they can operate only at slow

frequency rates. Heavy static (or seating) loads are required. The technical limitations of

Road Rater device are: - a) limited load level for some models and b) high static pre-load for

heavier models which changes the stiffness of the material and produces deflections that

are not representative of a moving wheel load.

The development of an impulse loading equipment, which closely simulates the timing and

amplitude of a rolling wheel load, began in the sixties. Isada [1966] reported the use of a

falling mass device to study the seasonal changes in the strength of flexible pavements.

Bonitzer and Leger [1967] and Bohn et al [1972] discussed about the evaluation of

pavements using Falling Weight Deflectometer (FWD). This equipment has undergone

several improvements over the last three decades. Some of the current FWDs have

sophisticated features such as electronic distance measurement, and Global Positioning

System (GPS) hardware to make the equipment more versatile.

Major applications of the FWD are in the following areas.

Evaluation of structural capacity of in-service flexible, semi-rigid and rigid pavements.

Quality control of subgrade and granular layers of pavements during the construction stage.

Assessment of the need for and design of thickness of overlays. Determination of the rate of deterioration of pavement structures. Evaluation of the degree of bonding between pavement layers.

8

Assessment of equivalent moduli of concrete blocks in block pavements. Evaluation of the load transfer capacity in the joints of concrete pavements. Detection of voids under rigid pavements.

The operating principle of FWD and the salient features of a few commercially available

models of FWD are discussed in the following paragraphs.

2.3 OPERATING PRINCIPLE OF FWD

The basic working principle of the impulse loading equipment is to drop a mass on the

pavement to produce an impulse load and measure the surface deflections. The mass is

dropped on a spring system, which in turn transmits the load to the pavement through a

loading plate. The resulting deflection bowl characteristics are observed and used in the

backcalculation of pavement material properties. The principle is illustrated in Figure 2.1.

Figure 2.1 Working Principle of FWD

2.4 SOME COMMERCIALLY AVAILABLE FWD MODELS

Some of the commercially available FWD models are:

Dynatest (with manufacturing facilities in Denmark and the United States)

KUAB (Sweden)

JILS, Foundation Mechanics, Inc. (United States) and

Carl Bro (Denmark)

In addition to the above-mentioned models, Komatsu company of Japan also manufactures

FWDs [Irwin, 2002]. There are a few other models of FWD that were developed in small

numbers by individual entrepreneurs and academic institutions. The two models developed

at IIT Kharagpur in India [Kumar et al, 2001; Reddy et al, 2002a] can be listed in this

category.

Falling Mass

Spring (Rubber Pads)

Deflection SensorsLoad Cell

Deflected Surface

Chapter 1 Introduction

9

Dynatest FWD

Dynatest company manufactures two FWD models 8000 and 8081 [Dynatest FWD/ HWD

Test Systems, 2003] used for evaluation of road and airport pavements respectively. These

two models are complete with back-up battery and all other accessories for evaluation of

pavements. Both the models are trailer-mounted and have the capability to apply loads in

the ranges of 7 to 120 kN and 30 to 240 kN respectively. The microcomputer based

software system, ELMOD, is used for the analysis of flexible as well as rigid pavements.

Elastic moduli, residual life and overlay requirement are the main outputs from the analysis.

WINPCN program, which computes Pavement Classification Number (PCN) values is used for

the analysis of airfield pavements.

KUAB 2m-FWD

The KUAB 2m-FWD [KUAB Falling Weight Deflectometer, 2003] is a trailer mounted dynamic

impulse loading device, which can be towed by any suitable towing vehicle. The equipment

is completely enclosed by a metal housing for protection against harmful elements. Testing

can be done with all the protective features in place. Bay doors in the bottom of the housing

open automatically during testing, eliminating the need for the FWD operator to leave the

tow vehicle. In this equipment, a two-mass configuration is used for the production of a

load pulse that simulates the actual effects of a moving vehicle. The loading plate is

segmented to ensure uniform pressure distribution over the full area of the plate. The three

most widely used models of the KUAB 2m-FWD vary primarily in terms of their loading

capacity. The KUAB 50 model is a light and versatile testing system suitable for a broad

range of highway, street and parking lot pavements, with a loading range of 13.3 to 62.2

kN. The largest KUAB 2m-FWD available, Model 150, is capable of generating a dynamic

load of 290 kN.

JILS FWD

JILS-20-FWD model [JILS Falling Weight Deflectometer, 2003] is mounted in a two-axle

trailer that can be towed by a van or pick-up. The machine is operated by the driver from

his seat in the tow vehicle. The loading capability ranges from 9 kN to 120 kN. The loading

plate used is a 300 mm diameter rigid steel disc with an 8 mm thick heavy duty, neoprene

pad attached to it for uniform distribution of the applied loading. Upto nine sensors can be

used for measuring deflections and there is facility to record the pavement temperature.

Carl Bro FWD

Phonix FWD [Carlo Bro Falling Weight Deflectometer, 2003] is the commercial name of Carl

Bros FWD. The latest FWD model is PRI 2100, which is modular in design, i.e., the loading

10

capability and sensor configuration can be varied as per requirement. The equipment is

available as trailer and vehicle built-in models.

FWDs are extensively used in many countries because of the following features.

cost effective and highly accurate- for many models, only one operator is required. wide acceptance as it is possible to simulate traffic loading closely compared to

other available equipment.

efficiency - a typical test sequence can be completed in a short time. mobility - highly maneuverable in traffic. multi-purpose pavement applications-evaluation of different types of facilities

ranging from unpaved roads to airfields.

wide loading range. repeatability of results.

2.5 OTHER FWD MODELS

In addition to the commercially available FWD models, a few indigenously developed FWD

models are also available. Details of a few such models are given in the following

paragraphs.

Nagaoka FWD

Nagaoka FWD is a modified KUAB FWD model-50 developed by Himeno et al [1989] for the

evaluation of local highways in Nagaoka city area in Japan. This is a trailer model enclosed

by a metal housing and towed by a truck. The magnitude of the impulse load is sensed by

pressure gauge placed on the loading plate. Deflections are measured using LVDTs mounted

on the reference frame. The surface temperature of the pavement and the distance

travelled are also recorded. The software used for operating the FWD consists of three

modules - system, measurement and data processing. System module is used for

conditioning of the pavement and calibrating the FWD. Measurement module is used to

monitor deflections where as the data processing module is for processing the data using

LMBS (Layer Moduli Backcalculation System) software, which uses ELSA (Elastic Layer

System Analysis) as a subroutine.

IITKGP_FWD1

The first indigenous FWD model in India was developed [Kumar et al, 2001] by the

Transportation Engineering Section of the Department of Civil Engineering, Indian Institute

of Technology, Kharagpur, India. A view of FWD1 is shown in Photograph 2.1. This model

is mounted in a trailer, which can be towed with the help of a jeep. With this model, it is

possible to apply a load of magnitude ranging from 20 kN to 65 kN with a loading time of

Chapter 1 Introduction

11

about 20 to 30 milli-seconds. This loading time is similar to that produced by a vehicle

moving at 50 to 60 km/h. Rubber pads of suitable stiffness were used as spring system to

obtain these loading times. Six surface deflections can be measured at radial distances of

0, 300, 600, 900, 1200 and 1500mm with the help of geophones.

Photograph 2.1 IITKGP_FWD1

A chain and pulley arrangement is used for lifting and lowering the mass whereas a chuck

arrangement is made for holding the mass at any desired height. One load cell and six

geophones are used to measure the magnitudes of load and deflections respectively. The

load and deflection signals are recorded in the computer with the help of a data acquisition

system.

Extensive field studies were conducted using this equipment and the data collected during

field investigations was used to backcalculate the pavement layer moduli [Kumar, 2001].

This low-cost equipment is quite suitable for developing countries like India. Some of the

shortcomings of this model are: - a) many of the operations such as pulling of chain for

lifting the mass, placing the geophones on the pavement surface and releasing the mass,

are done manually and hence longer time is needed for data collection. Also, maneuvering

the equipment on heavily trafficked two-lane two-way highways in India was found to be

difficult.

2.6 VARIATIONS OF FWD

Besides the standard models of FWD discussed in the previous sections, some variations of

the equipment are also available. LOADMAN [Livneh et al, 1995] is a portable FWD,

available in two models (light and heavy weight) and is currently used by many

12

organizations. The heavier version of this model is mounted inside a vehicle [Loadman,

2003]. It is used for compaction control of bound and unbound layers and for measuring the

bearing capacity of the pavement.

Rolling Wheel Deflectometer (RWD) is the most recent and advanced NDT equipment for

evaluating pavements. RWD device measures pavement deflections under an 18-kip rolling

wheel load using a laser sensor. Designed to operate at 35 mph, the RWD can travel at

highway speeds and cover greater distances than a standard FWD. It gathers real-time

deflection data as it travels [Bay and Stokoe II, 1998; Rolling Wheel Deflectometer, 2003].

There is no risk to workers and no decrease in the traffic-carrying capacity of the highway

while deflection measurements are taken.

2.7 BACKCALCULATION OF PAVEMENT LAYER MODULI

The response measured with the FWD is the surface deflection of the pavement at different

distances from the centre of the load. The measured deflections along with other relevant

information are used as inputs either to backcalculate the effective pavement layer moduli

for use in analytical evaluation methods or to estimate the overlay requirement from

empirical relationships. Salient features of some existing backcalculation procedures are

presented in the following sections.

Determination of Youngs modulus of elasticity for pavement materials using measured

surface deflections by working backwards is generally called Backcalculation. More

specifically, it is the process of selection of layer moduli using a suitable technique (iteration,

database searching, closed-form solution, optimization) so that the deflections computed

using the layer moduli are close to the measured deflections.

Scrivner, et al. [1973] developed the first closed-form solution for two-layer pavement

system based on Burmisters [1945] layer theory. The first closed-form solution for

backcalculating layer moduli for multi-layer pavements was developed by Yih Hou [1977]

using least squares method. The first graphical method for determining the moduli of two-

layer pavements was developed by Swift [1973]. Odemarks [1949] equivalent layer concept

was used in some backcalculation models to simplify the pavement systems and thereby

facilitate the use of Boussinesq's theory for the analysis of pavements. The backcalculation

method developed by Ullidtz [1987] is based on this concept and reportedly gives

reasonable layer modulus values for pavements in which the layer stiffness decreases with

depth. Lytton and Michalak [1979] used a more general form of Odemarks assumptions to

convert a multi-layered pavement into a single layer placed above a rigid base. With

Chapter 1 Introduction

13

advances in the computational facility, a number of computer based backcalculation

programs are available now.

The computer based backcalculation procedures are typically associated with (i) a suitable

theory selected for the analysis of layered pavement systems (ii) an optimization techniques

for selection of a set of layer moduli that produce computed responses (deflections) similar

to the observed responses and (iii) an objective function which reflects the differences

between the measured and computed responses. The backcalculation procedures differ

from one another in terms of the following features.

a) Pavement system considered

i) Number of layers

ii) Type of interface (rough, smooth)

iii) Presence of rigid layer (bed rock)

iv) Depth of rigid layer

b) Theory used for the analysis of the pavement

i) Linear or non-linear material behaviour

ii) Elastic or visco-elastic

iii) Static or dynamic analysis / Layered or FEM analysis

c) Requirement of seed moduli and range of moduli

d) Backcalculated parameters: - Pavement layer moduli and/or thicknesses

e) Number of loads used (corresponding to the FWD system used) and the type

of contact area

f) Responses measured, sensor configuration

g) Convergence criteria

h) Backcalculation Technique: - Regression models, ANN models, traditional

optimization techniques and Genetic Algorithm models.

Salient features of some backcalculation programs are presented briefly in Table 2.1.

14

Table 2.1 Salient Features of Some Backcalculation Programs [Irwin, 1977], [Ullidtz and Coetzee, 1995], [SHRP, 1993], [Fwa et al, 1997], [Rwebangira et al, 1987]

Program Forward

Calculation Backcalculat

-ion Non

linear Analysis

Theoretical Model

Seed Moduli

Convergence Scheme

Number of

Layers ELMOD WES5 Iterative Yes Method of

Equivalent Thickness

No Relative error on 5 sensors

Five

EVERCALC WESLEA Iterative Yes Multilayer elastic

Generated Sum of absolute error

Five

MODULUS WESLEA Data base No Multilayer elastic

Yes Sum of relative squared error

Four

MODCOMP3 CHEVRON Iterative Yes Multilayer elastic

Yes Relative deflection error

at sensors

Four

BOUSDEF MET Iterative Yes Method of Equivalent Thickness

Yes Sum of percent errors

Four

BISDEF BISAR Iterative No Multilayer elastic

Yes Sum of squares of absolute error

Best for

three CHEVDEF CHEVRON Iterative No Multilayer

elastic Yes Sum of squares

of absolute error Best for

three ELSDEF ELSYM5 Iterative No Multilayer

elastic Yes Sum of squares

of absolute error Best for

ThreeWESDEF WESLEA Iterative Yes Multilayer

elastic Yes Sum of squares

of absolute error Four

COMDEF DELTA Database No Multilayer elastic

Yes Various schemes Five

DBCONPAS FEACONS Database Yes Finite Element

No N.A N.A

MICHBACK SAPIV Iterative Yes Multilayer elastic

Yes Sum of relative squared error

Four

PADAL ILLIPAVE Iterative Yes Multilayer elastic

Yes Sum of relative squared error

Three

FPEDD1 BASINPT Iterative Yes Multilayer elastic

Generated N.A N.A

UMPED PAVRAN Iterative No Multilayer elastic

No N.A N.A

ISSEM4 ELSYM5 Iterative Yes Multilayer elastic

Yes Relative deflection error

N.A

DIPLOBACK DIPLOMAT ANN No Multilayer elastic

N.A N.A Three

NUSGABACK CHEVRON GA No Multilayer elastic

Yes Root mean squared

difference

Four

BKGREEN GREEN Iterative No Multilayer elastic

No N.A Four

N.A: Information not available

15

SID, SIDMOD, FEDPAN, BACKLAY, DAPS, FEAD, PEDD, MFPDS, CARE, CANUV, LMBS,

PROBE, LMBS, DEFMET, RPEDD1, PHONIX, PEACH, FALMAN, CLEVERCALC, EPLOPT, OAF,

SEARCH, EFROMD [Ullidtz and Coetzee, 2003] are some other backcalculation programs

available for estimating the layer moduli.

It can be observed that almost all the backcalculation programs use linear multi-layer elastic

theory. Most of the methods follow an iterative approach in which an initial set of layer

moduli is assumed and the moduli are then used to compute surface deflections. The

computed deflections are compared to the measured deflections. The moduli are adjusted

suitably to reduce the differences between the measured and computed deflections. The

process is repeated until the calculated deflections match with the measured deflections

within some specified tolerance value. Seed moduli are required for many backcalculation

programs.

Backcalculation models can be used for the estimation of the effective properties of

pavement materials for use in the analysis of in-service pavements. Discussion of various

other approaches usually followed for the selection of layer moduli is presented in the

following section. These approaches include laboratory testing of representative samples

and use of empirical relationships obtained from the evaluation of in-service pavements.

2.8 SELECTION OF LAYER MODULI FOR ANALYSIS OF FLEXIBLE PAVEMENTS

Selection of appropriate layer moduli for analysis is a key element in the mechanistic design

of new pavements and overlays. Various agencies use different methods for the selection of

the moduli values. For new pavements, these properties are determined by conducting

laboratory tests on representative samples of materials or by using empirical relationships

that estimate layer moduli from material properties which can be obtained with relative

ease. While assessing the condition of the in-service pavements also, laboratory tests can

be conducted on samples collected from the pavements. But most of the current overlay

design procedures require structural evaluation of pavements besides using some laboratory

based material properties. Results of the structural evaluation are used either for designing

the overlay directly or backcalculating the material properties. The backcalculated

properties, in turn, can be used in the design of the overlay.

The following paragraphs present some of the approaches commonly adopted for

determination or selection of material properties for analysis of pavements.

16

2.8.1 Elastic Modulus of Subgrade

Laboratory Evaluation

Elastic modulus of subgrade soils is normally determined by conducting repeated triaxial test

in the laboratory simulating the triaxial stress condition expected in highway pavements.

The elastic modulus, frequently termed as Resilient Modulus (MR), is computed using the

following expression.

)1.2(/)(M ra31R = where ra = recoverable axial strain; 1, 3 = principal stresses

Hveem [1955] introduced the term resilient deformation to represent the elastic

component of the total deformation.

Shifley and Monismith [1968] represented the non-linear behaviour of fine-grained soils

using the following bi-linear equations.

MR = k2 + k3 (k1 - (1 - 3)) when k1 > (1 - 3) (2.2) MR = k2 + k4 ((1 - 3) - k1) when k1 < (1 - 3) (2.3)

where k1, k2, k3 and k4 = Material constants.

Another relationship used for estimating the stress dependent resilient modulus of fine-

grained soils is

)4.2()/)((kM 2k3311R=

where k1 and k2 = material constants

A more involved relationship [Uzan, 1985] correlating the resilient modulus with the state of

stress is given as

MR= k1 pa (/ pa) k2 (oct/ pa) k3 (2.5)

where oct = octahedral shear stress; Pa = atmospheric Pressure; = bulk stress; k1, k2, k3 = material Parameters

Estimation

A number of empirical relationships are available for estimating the subgrade modulus. The

most common parameter used to estimate elastic modulus of subgrade soil is the California

Bearing Ratio (CBR). Equation 2.6 gives a generalized relationship used for the estimation of

elastic modulus from CBR. The values of k suggested by different investigators/agencies

are given in Table 2.2.

Subgrade Modulus= k x (CBR) (2.6)

17

Table 2.2 Values of k suggested by Different Investigators/ Agencies

Design method/ Researchers K (MR in MPa) IRC: 37 [2001] Shell [1978], AASHTO [1993]

10 for all soils

Dauzats and Linder [1982] 5 for CBR

18

where MR = backcalculated subgrade resilient modulus (psi); P = applied load

(pounds); dr = deflection at a distance r from the center of the load (inches); r =

distance from the center of load (inches); Poisson ratio assumed as 0.5.

The subgrade modulus value estimated from Equation 2.9 has to be adjusted for using it for

pavement design in AASHTO method to be consistent with the modulus values used to

represent the AASHTO road test soils.

Garg and Thompson [1998] proposed regression equations for estimating the subgrade

modulus from FWD test using pavement deflection, D3 in mils (0.001 inch) measured at

1097 mm radial distance from the centre of the loading plate. The equations are:

For conventional pavements:

Log ERi = 1.51-0.19 D3 +0.27 log (D3) (2.10 a)

For full depth AC pavements:

Log ERi = 24.7-5.41 D3 +0.31 (D3)2 (2.10 b)

where ERi = subgrade modulus ( ksi)

Roque et al, [1998] presented the following equation for the estimation of subgrade

modulus based on the deflections measured using a dual loading FWD system.

)11.2()60/D(334.36)ksi(M 015.1xR=

where Dx/60 = FWD deflection (mils) measured at 60 inches radial distance

from the center of the dual plates.

Molenaar and Van Gurp [1982] developed the following equation to predict subgrade soil

modulus from the FWD deflections.

)12.2(dx10x614.6)MPa(E 00915.123

sub=

where d2 = FWD deflection (in metres) measured at a radial distance of 2000 mm.

Wimsatt [1999] developed a regression model given as

)13.2()8.1828xW(

Px24.0)MPa(E

7Sub =

where W7 = FWD deflection (mm) measured at a distance of 1828.8 mm from the

center of the load plate; P= FWD load level (N)

Choubane and McNamara [2000] proposed the following empirical equation for predicting

embankment subgrade modulus from FWD deflection data.

19

ESFWD =0.03764 (P/dr) 0.898 (2.14)

where ESFWD = predicted embankment modulus based on FWD data (psi);

P= applied load (lbs); dr = Deflection measured at a radial distance of 1097 mm.

Alexander et al, (1989) proposed an equation for evaluating subgrade modulus from the

deflection (mils) measured at a radial distance of 1830 mm (D72) from the centre of the

loading plate for an applied load of 111206 N.

Es (psi) =59304.82 (D72)-0.98737 (2.15)

Kim et al (2000) established a relationship between the Base Damage Index (BDI) and

Shape Factor F2 for different subgrade moduli, where BDI is (1- 2) and F2 is ((1- 3)/

2) and 1, 2, 3 are the deflections measured at 305, 610 and 914 mm distances

respectively from the centre of the FWD load. Figure 2.2 shows the relationship.

Figure 2.2 Relationship between BDI, Shape factor and Subgrade Modulus Dai et al [1998] found that the subgrade modulus backcalculated using EVERCALC 5.0

matched with the laboratory results obtained for low deviator stress levels. Tests on soils

having higher plasticity index resulted in lower subgrade modulus values.

Subgrade modulus can also be determined [Harr, 1966] from the average deflection value

measured during the third, fourth and fifth drops of the load in a Portable Falling Weight

Deflectometer using Equation 2.16.

20

Es (MPa)= 2 P (1-2) r a/ A/d (2.16)

where P = dynamic load (kN); = Poisson ratio; r = Plate radius (m); a =

plate shape and rigidity factor (0.79 for rigid, 1 for flexible); A= plate area

(m2); d= deflection (mm).

2.8.2 GRANULAR LAYER MODULUS

Laboratory Tests

Granular materials constitute a major portion in the thickness of flexible pavements. Moving

wheel load induces rotational principal stresses in the unbound pavement layers. The

estimation of representative resilient modulus value of granular layers has always been a

difficult task due to the high degree of stress dependency of the modulus value. Since 1960,

numerous efforts were made to characterise the resilient behavior of granular materials.

Repeated triaxial test is generally conducted for determining the modulus value of granular

materials in the laboratory.

Considering the nonlinear behavior of the granular material, Dunlap [1966] and Monismith

et al, [1967] suggested that the resilient modulus of a coarse soil increases with

confinement stress and is less affected by the deviator stress and proposed a relationship

between the resilient modulus and the confinement stress given by the following equation.

MR = k1(3) k2 (2.17)

where 3 = confinement stress; k1, k2 = regression constants,

Several earlier studies [Monismith et al 1967; Hicks, 1970; Smith and Nair, 1973; Uzan,

1985; Sweeere, 1990) indicated that the resilient modulus of untreated granular material

has a high degree of dependence on the confining pressure and the sum of the principal

stresses. Monismith et al [1967] reported an increase as large as 500 % in the MR value for

a change in confining pressure from 20 to 200 kPa.

Some other researchers (Pezo, 1993), Garg and Thompson (1997) found it necessary to

include deviator stress in the expression for estimation of resilient modulus.

MR = N1qN2 3 N3 (2.18)

where N1, N2, N3 = constants, q = deviator stress (1- 3)

Tam and Brown (1988) expressed MR as a simple function of stress ratio

MR = k1 (p/q) k2 (2.19)

where k1= constant; P= mean normal stress (1+2+3)/3

21

Johnson et al [1986] showed that MR is dependent on both the first invariant of stress and

the stress ratio and suggested the following model.

MR = k1 (J2/oct) k2 (2.20)

where J2= first stress invariant =12 +23+31 ; oct = shear stress Seed et al [1965], Brown and Pell [1967] and Hicks and Monismith [1971] suggested that

resilient modulus is a function of the sum of principal stresses or bulk stress as expressed by

Equation 2.21.

MR = k1 ()k2 (2.21) where = bulk stress (sum of principal stresses), k1, k2 = regression constants.

Equation 2.21, popularly known as K- model, was adopted by several researchers and organizations [Allen and Thompson, 1974; Boyce et al, 1976]. Table 2.4 gives the

regression constant values obtained by different researchers for the K- model for unbound materials.

Table 2.4 Regression Coefficients for K- model for Different Unbound Granular Materials

Material studied k1 k2 Crushed gravel and stone [Hicks, 1970] 1600-5000 0.57-0.73 Unbound base materials [Hicks and Monismith, 1971] 2100-5400 0.61 In-service base and subbase materials [Zhou et al, 1992] 2900-7750 0.46-0.65 Crushed stone [Zhou et al, 1992] 4000-9000 0.46-0.64 Crushed gravel and stone [Allen, 1973] 1800-8000 0.32-0.70 Well graded crushed aggregate [Boyce et al, 1976] 8000 0.67 Crushed aggregate (saturated) [Zhou et al, 1992] Crushed aggregate (at O.M.C)

1300-2000 2000-2600

0.69-0.78 0.70-0.73

Well graded crushed lime stone [Brown and Papin, 1981], kPa 8634 0.69 Uniformly graded crushed lime stone [Brown and Papin, 1981] 19455 0.5 Dense graded crushed stone base material [Thompson, 1989] 9000 0.33

Winter 3250 0.55 Summer 3850 0.55 Spring 3900 0.55

Unbound granular materials [Khosla and Ali, 1989]

Fall 4000 0.55 Crushed rock [Pandey and Naidu, 1994] in kPa, MR =MPa 3.47 0.7375

MR, - are in psi

In addition to the above mentioned studies, May and Witczak [1981] and Zaman et al

[1994] observed that the resilient modulus values for granular materials varied from 51 to

159 MPa for the corresponding variation of the sum of the principal stresses from 100 to

690 kPa. Smith and Nair [1973] observed an increase of 50% in the MR value when value increased from 70 to 140 kPa. Hicks [1970] suggested that the MR value is unaffected by

22

the magnitude of deviator stress applied, provided the specimen is not subjected to

excessive deformation. Hicks and Monismith [1971] reported a slight softening of the

material at low deviator stress levels and slight stiffening at higher stress levels.

Though the K- model is extremely useful, it has some deficiencies. The effect of shear stresses induced due to the shear resistance provided by strong confinement is not

considered. As the deviator stress increases, the MR value decreases initially before showing

an increasing trend. This phenomenon is also not explained by the K- model. Studies conducted by May and Witczak [1985], and Uzan [1985] resulted in the following model

which takes into account the effect of shear stress on MR value.

32 k

octk

1R kM = (2.22) where oct= octahedral shear stress; k1, k2 and k3 = material constants

Studies by Trollope et al (1962), Hicks (1970), Robinson (1974), Rada and Witczak (1981)

and Kolisoja (1997) suggested that MR value generally increases with increasing density.

Kolisoja [1997] included the effect of material density in the K- model, which is represented by Equation 2.23.

MR = A (nmax- n) po (/po) 0.5 (2.23a) MR = B (nmax- n) po (/po) 0.7(q/po) 0.2 (2.23b)

where A, B = constants; nmax, n = maximum porosity and material

porosity respectively

The above equations are based on laboratory triaxial tests conducted with constant

confining pressure.

Nataatmadja and Parkin (1989) and Nataatmadja (1992) proposed the following equations

for MR values for constant confining pressure (CCP) and variable confining pressure (VCP).

MR = / q (A+Bq) for CCP (2.24a)

MR = / 1 (C+Dq) for VCP (2.24b)

where A, B, C, D = constants

Itani [1982] developed a multiple regression equation to arrive at a model that included

bulk stress, shear stress and confining stress for estimation of MR.

MR = k9 (/3) k10 d k11 3 k12 (2.25)

where k9, k10, k11, k12 = regression constants

Feliberti [1991] developed another model where axial strain (d) is used rather than deviator

stress in evaluating MR value.

23

MR = k13 () k14 dk15 (2.26)

where k13, k14, k15 = regression constants

Estimation

A widely used expression for estimating the modulus of granular layer adopted in the Shell

design procedure [1978] is given as

45.0RR hx2.0)Subgrade(M/)Granular(M = (2.27)

where MR (granular), MR (subgrade) are in MPa; h = thickness of granular layer (mm)

Extensive studies were carried out by various researchers for establishing the ratio of the

granular layer modulus to subgrade modulus. The modular ratio values (ratio of elastic

modulus of granular layer to elastic modulus of subgrade) suggested by various researchers

are given in Table 2.5.

Table 2.5 Modular Ratio Values Suggested by Various Researchers

Investigator Modular

Ratio Remarks

Heukelom and Klomp [1962] 2 to 4 Shell Criteria 5 On strong base (Vibratory test) Smith and Witczak [1981] 3 to 4 On normal base (Vibratory test)

Brown et al [1982] 1.5 to 7.5 Finite Element Analysis Deen et al [1971]

Dependent on moduli of asphalt layer and subbase independent of base course thickness

Smith and Witczak [1981]

Modular ratio increases as h1, E1, h2 decrease, E3 increases

Shook et al [1982] 1.9 to 6.7. Varies with Traffic, subgrade modulus and asphalt concrete thickness

Bose [1993]

Modular ratio increases as h1, h2 decreases and E3 increases for granular layers with asphalt concrete surfacing

Kumar [2001] 3.47 to 4.0 Modular ratio more in monsoon season compared to other seasons

E1= Surface Modulus; E2= Base Modulus; E3= Subgrade Modulus; h1= Surface thickness; h2= Base thickness Smith and Witczak [1981] proposed the following equations for the estimation of subbase

and base moduli from thicknesses and other layer moduli.

For subbase course material

Esb = Esg(1+0.003 * hsb) (2.28a)

For base course material

Eb = Esg(1+0.067 * hb) (2.28b)

where Eb , Esb, , Esg = moduli of base, subbase and subgrade (MPa) respectively;

24

hb , hsb, are base and subbase thicknesses in mm.

Generalized equations were developed for the estimation of base and subbase moduli by

USACE based on the investigations of Barker et al, [1977]. The equations are given as:

For base course material

))tlog(x)Mlog(x10.2)tlog(x52.101(MM 1Rn1RnRn ++ += (2.29a) For sub-base course

))tlog(x)Mlog(x56.1)tlog(x18.71(MM 1Rn1RnRn ++ += (2.29b) where MRn= elastic modulus of the nth layer; MRn+1= elastic modulus of the

(n+1)th layer; t= thickness of the nth layer(inches)

Smith and Witczak [1981] carried out extensive analytical investigations on the elastic

moduli of granular layers used in flexible pavements. The equation developed for the

estimation of base modulus as a function of different layer thicknesses and moduli is given

as

)klog(x888.0)Elog(x279.0

)Elog(x155.0)hlog(x008.0)hlog(x511.0079.1)Elog(

13

121Granular

++=

(2.30)

where h1 = thickness of the asphalt concrete layer (inch); h2 = thickness of the

granular layer (inch); E1 = modulus of the asphalt concrete layer (psi); E2 =

modulus of the granular layer (psi); E3 = modulus of the subgrade (psi); k1 =

material constant for granular layer obtained from repeated triaxial test

The equation used for the estimation of granular layer moduli (psi) in the DAMA computer

program of Asphalt Institute [DAMA, 1983] is of the form given by Equation 2.31.

65432 kkR

kR

k2

k11R )F(x))Subgrade(M(x))ousminBitu(M(x)h(x)h(xk)Granular(M

= (2.31) where h1, h2 = thicknesses of bituminous and granular layers (inches);

k1 to k6= regression constants with the following values k1 = 10.447,

k2= 0.471, k3 = 0.041, k4 = 0.139, k5 = 0.287, k6 = 0.868, F = a factor representing

the type of unbound aggregate layer

AASHTO [1993] recommends the following relationships for the estimation of MR value of

unbound granular materials from CBR values.

psi10forCBRx250;psi340forCBRx340;psi30forCBRx440;psi100forCBRx740)psi,Granular(MR

=====

(2.32)

where is the sum of principal stresses.

25

Austroads [1992] recommends the following options for estimation of granular layer

modulus (i) laboratory triaxial testing (ii) backcalculation from deflection bowls (NDT) and

(iii) presumptive values in the absence of any data. In the case of cemented materials, the

following equations were developed relating elastic modulus (MPa) with Unconfined

Compressive Strength (UCS).

E (MPa) = 1814 UCS 0.88 +3500 for cemented crushed rock (2.33a)

E (MPa) = 2240 UCS 0.88 +1100 for cemented natural gravel (2.33b)

where UCS is in MPa

Structural Evaluation of Pavements

Badu et al, [1989] suggested the following equation from multiple regression analysis of

FWD data for the estimation of base course layer moduli.

)34.2()DDlog(8423.4)DDlog(0167.9

)DDlog(3562.3)Dlog(1179.0)t(03326.0280.3Elog

5141

2171Base

+=

where EBase = base course modulus (ksi); D1, D2, D4, D5, D7 = measured deflections

at 0, 200, 500, 800, 1600 mm from center of the loading plate; t1= thickness of

surface course (inch)

Roque et al, [1998] presented the following equation for the prediction of subbase moduli

from deflections measured with a FWD having dual load configuration.

)60/xD/09.321260/x

60/x

t686.0t0498.06706.360/x

t/302.5D/6202.11609.236/x

12/x0/yD/4888.202523.6

60/x36/x0785.1

2R

)D(x]D15.1

)DD[(x)DD()t(81136.105)ksi,subbase(M

+

+=

(2.35)

where all the thicknesses (t) are in inches and deflections (D) are in mils (0.001 inch)

2.8.3 Bituminous Layer Modulus

Determination of bituminous layer modulus is quite complex, as its value is affected by a

large number of factors including temperature and loading time. Modulus of bituminous mix

can be determined either by laboratory tests on cores obtained from the field or on samples

prepared under representative conditions. The moduli can also be estimated from

nondestructive evaluation of in-service pavements. The laboratory tests are usually

conducted under repeated load conditions in constant load or constant strain mode.

Rada et al [1991] developed a relationship for estimating the asphalt concrete layer

modulus using SHRP (Strategic Highway Research Program) data, which is given as

Equation 2.36.

26

Eac= 0.553833 +0.28829 * P200 * f -0.17033 -0.03476*Va + 0.070377

*70.10 6 +0.000005 * [tp 0.3+ 0.49825 log (f) * Pv 0.5] 0.00189

[tp 0.3+ 0.49825 log (f) * Pv0.5 *f 4.4] +0.931757 *f0.02774 (2.36)

where Eac = asphalt concrete modulus (105 psi); Va = percentage of air voids in mix; f =

test frequency; tp = mid-depth AC layer temperature (oF); P200 = Percentage of aggregate

weight passing #200 sieve; = Asphalt viscosity at 700F; Pv = percentage of asphalt

content by volume of mix.

Badu et al [1989] suggested the following equation from multiple regression analysis of

FWD data for the estimation of asphalt layer moduli.

)37.2()37.2()DDlog(871.5

)DDlog(205.17)DDlog(445.12)t(2481.0215.2Elog

41

31211AC

+=

where EAC = asphalt layer Modulus (ksi); D1, D2, D3, D4 =measured deflections at 0,

200, 300 and 500 mm radial distances respectively from center of the loading plate;

t1= thickness of surface course layer (inch)

Roque [1998] developed a regression equation for estimating the bituminous layer modulus

from FWD test conducted with 300 mm diameter loading plates (dual load configuration).

Es (ksi) = 78.2254 (t1) 0.5554 (Dy/0 - Dy/305) (-0.7966-19.1332/t1) * (Dy/0

- Dx/200) 17.4791/t1 (2.38)

where t1 =thickness of surface layer (inches); Dy/0, Dy/305 = deflections (mils) at 0,

305 mm radial distances respectively from centre of the loading plate in longitudinal

direction; Dx/200 = Deflections (mils) at a distance of 200 mm from centre of the

loading plate in the lateral direction

2.9 EFFECT OF TEMPERATURE ON BITUMINOUS LAYER MODULUS

Properties of bituminous mixes vary with temperature. Modulus values determined at

different temperatures are normally adjusted to correspond to a standard temperature for

design of pavements and overlays. Different temperature adjustment factors and equations

were given by various researchers for adjusting the modulus and or deflections for

temperature.

Ullidtz and Peattie [1982] utilized the deflection data from AASHO road test and the SHELL

procedure for estimation of mix stiffness and developed the following equation for

comparing the moduli obtained at two different temperatures.

27

2Tlog38.16277.21Tlog38.16277.2

EE

10

10

2T

1T

=

(2.39)

where ET1, ET2 = moduli of bituminous mix at T1 and T2 temperatures (0C)

respectively

Rada et al [1988] gave the following expression for modeling the variation of stiffness with

temperature.

)TT(10x245.3

2T

1T798.1

2798.1

14

10EE = (2.40)

Antunes [1993], based on the analysis of backcalculated moduli obtained from the FWD

data collected at different temperatures, proposed the following Equations.

For Asphalt Concrete 2

1

2T

1T

T0317.0635.1T0317.0635.1

EE

= (2.41a)

For Bituminous Macadam 21

2T

1T

T0398.0795.1T0398.0795.1

EE

=

(2.41b)

Kim et al [2000] gave the following equations for adjusting deflections and moduli for

temperatures.

For Deflection

(2.42)

where D68 = deflection (inches) corresponding to a temperature of 680F

DT = deflection (inches) corresponding to a temperature of T 0F

=3.67 x 10-4 x t 1.4635 for wheel paths and 3.65 x 10-4 x t 1.4241 for lane centers; t= thickness of the Asphalt Concrete (AC) layer (inch) and T = AC

layer mid-depth temperature (0F) at the time of FWD testing.

For Modulus

(2.43)

where E68 = AC Modulus at temperature of 680F (psi);

ET = backcalculated AC Modulus (psi) from FWD testing at temperature of T 0F

Chen et al [2001] suggested the following equation for adjusting the layer modulus for a

given temperature.

ETw = ETC/ [(1.8Tw +32)2.4462. (1.8Tc +32)-2.4462] (2.44)

]10[xDD )T68(T68=

]10[xEE )T68(0153.0T68=

28

where ETw = modulus adjusted for a temperature of Tw (oC); ETc = modulus at a

temperature of Tc (0C)

Johnson and Baus [1992] recommended the following equation for adjusting the bituminous

layer modulus for a standard temperature of 700 F.

)T70(0002175.0E

856.1886.1

10 = (2.45) where E =adjustment factor and T= temperature at which the modulus has been obtained

Ullidtz [1987] developed a model for temperature correction based on backcalculated moduli

values obtained from AASHO Road Test deflection data. The model is given as

ETo= (1/3.177-1.673 log10 T) ET (2.46)

where ETo= Asphalt Concrete (AC) modulus value at temperature To;

ET = backcalculated AC modulus at temperature T (0 F)

Baltzer and Jansen [1994] developed the following temperature correction model based on

statistical analysis of backcalculated moduli and measured AC temperatures.

ETo= 100.018(T-20) x ET (2.47)

where ETo and ET have the same meaning given for Equation 2.46

Ali and Slezneva [2000] developed a relationship for estimating AC layer modulus as a

function of average AC layer temperature (oC) and temperature gradient in the AC layer

(oC/m).

)T(*0018.0)T(*033.053.9(AC

Gpe934E ++= (2.48) where EAC= AC Layer Modulus (MPa); TP = average AC layer temperature (oC); TG =

temperature gradient in the AC layer ( oC/m).

2.10 POISSON'S RATIO

For most of the pavement materials, the influence of Poissons ratio () on various critical parameters is normally small. This allows the use of typical values in the analysis rather

than resorting to complex laboratory testing [Mitchell and Monismith, 1977]. For clayey

subgrades, Poisson's ratio varies from 0.4 to 0.5. A value of 0.5 is normally adopted for wet

condition [Brown and Pell, 1972]. Mitchell and Monismith [1977] recommended a value of

0.5 for saturated clays and 0.35 for sands. A value of 0.4 was selected for subgrade soils.

Typical values of for unbound granular material vary from 0.2 to 0.5. Allen and Thompson [1974] recommended a range of 0.35 to 0.4 for Poisson's ratio of granular material. Uzan et

al,[1972] discussed about the non-linearity of value of granular material and reported that

29

it may even assume a value of 0.6 to 0.7. Poissons ratio () of bituminous mixtures ranges from 0.35 to 0.50 and a value of 0.50 is relevant for higher temperatures [Pell, 1987]. The

value of Poissons ratio considered for analysis of flexible pavements in different pavement

design procedures are given in Table 2.6.

Table 2.6 Poissons Ratio Values adopted in Different Design Procedures

Design Method

Layer Range Typical Value

Remarks

Asphalt

0.15-0.45

0.35

Dependent upon temperature; low value for cold temp. (

30

and overlays. The following sections briefly present the salient features of a few design

methods.

2.11.1 Asphalt Institute Method [1981]

In this method, the pavement is considered as a multi-layer elastic system and each layer is

characterized by its modulus of elasticity and Poisson ratio. Traffic is expressed in terms of

repetitions of 80 kN axle load having dual tyres with radius of contact of 115 mm spaced at

345 mm apart with a tyre pressure of 0.483 MPa. The failure modes considered are fatigue

cracking in bituminous bound layer and rutting along wheel paths.

The performance criteria used in the Asphalt Institute design method are: -

Fatigue:

]69.0)VV/V[(84.4M

)E(x)(x004325.0x)10(x4.18N

nbb

854.0

ac291.3

tM

f

+==

(2.49)

where Nf = number of load repetitions to failure; t= magnitude of tensile strain at

the bottom of asphalt concrete layer; Eac = dynamic modulus of elasticity of asphalt

concrete (psi); Vv= percentage volume of air voids; Vb= percentage volume of

asphalt

Rutting: 477.4

c9

p x10x365.1N = (2.50)

where Np = number of load applications; c = vertical compressive strain

Selection of layer moduli for analysis is done in the following manner.

Subgrade resilient modulus: - (a) laboratory tests (AASHTO Method T 193 or ASTM D 1883)

(b) Estimation of MR from either of the following relationships MR = 10.342 x CBR or

Resistance (R) value (ASTM D 2834 or AASHTO method T 190) - MR = 7.963 + 3.826 R.

Unbound granular layer modulus is estimated from the predictive equation [Shook et al,

1982]. Asphalt concrete modulus is estimated from the regression equation incorporated in

the DAMA program.

2.11.2 Shell Pavement Design Method [1978]

In this procedure also, the pavement structure is considered as a linear elastic multi-layer

system. 80 kN standard axle load (circular contact area radius of 105 mm and a centre to

centre spacing between tyres of 310 mm) is considered for analysis. The performance

criteria adopted are: -

31

Fatigue: 363.2

ac671.5

tf )E(x)(0685.0N= (2.51)

where Nf = number of load repetitions to failure; Eac = modulus of asphalt concrete

in psi; t= magnitude of tensile strain at the bottom of the asphalt concrete layer

Rutting: 4

c7

r )(x10x15.6N = (2.52)

where Nr = number of load repetitions to rutting failure ; c = vertical compressive strain at the top of the subgrade.

Subgrade modulus is estimated using the Equation MR = 10 x CBR. Granular layer modulus

is obtained using Shell relationship given by Equation No. 2.27. Modulus of bituminous layer

is determined from binder properties, mix proportions, temperature and loading time.

2.11.3 Australian Road Design Method (AUSTROADS) [1992]

This is another procedure similar in framework to the Shell method. The performance

criteria are given by Equations 2.53 and 2.54.

N= [6918(0.856 Vb +1.08/s 0.36 t)]5 (2.53)

where N = allowable number of the load applications before an unacceptable level

of cracking develops; Vb = percentage by volume of bitumen in the asphalt; s =

stiffness of mix (MPa) ; t = tensile strain (in units of micro strain) at the bottom of

the asphalt layer.

N = [8511/ s] 7.14 (2.54)

where s = vertical compressive strain (in units of micro strain) at the top of the

subgrade; N = allowable number of the load applications before an unacceptable

level of rutting develops.

Evaluation of Pavement Layer Modulus:

Subgrade support from: CBR (field or laboratory) or elastic parameters. Modulus of unbound granular materials is to be determined from repeated triaxial

tests.

Asphalt concrete modulus: Using Shell nomographs or from laboratory tests. 2.11.4 Indian Roads Congress Method [IRC: 37, 2001]

Linear elastic layer theory is used in this method to analyse the pavements. Two major

structural distresses, namely rutting and fatigue cracking of bituminous layer, are

considered.

32

The fatigue criterion adopted is

Nf = 2.21X10-4 (t) 3.89 (1/ Eac) 0.854 (2.55)

where Nf = number of cumulative standard axles to produce 20 % cracked

surface area; t = Tensile strain at the bottom of the bituminous concrete layer

(micro strain); Eac = Elastic Modulus of bituminous surfacing (MPa)

Rutting criterion is given as

Nr = 4.1656X10-8 (1/s) 4.4337 (2.56)

where Nr = number of cumulative standard axles to produce rutting of 20 mm

s = vertical subgrade strain (micro strain)

Subgrade modulus is estimated using the Equation Es = 10 x CBR for CBR 5 % and 17.6

x CBR0.64 for CBR>5 %. Equation 2.27 is used for estimating granular layer modulus.

Bituminous modulus values suggested for different temperatures are given in Table 2.7. The

values suggested by IRC are presented in the table after rounding off the same to the

nearest 10.

Table 2.7 Suggested Elastic Modulus Values for Bituminous Materials

[IRC: 37, 2001] Elastic Modulus (MPa) values at Temperature oC

Mix Type 20 25 30 35 40 BC and DBM for 80/100 Bitumen 2300 1970 1450 980 800 BC and DBM for 60/70 Bitumen 3600 3130 2580 1700 1270 BC and DBM for 30/40 Bitumen (75 blows compaction and 4 % air void)

6000 4930 3810 2950 2280

BM for 80/100 Bitumen 500 BM for 80/100 Bitumen 700

BC: Bituminous Concrete; DBM: Dense Bituminous Macadam; BM: Bituminous Macadam

2.12 FLEXIBLE OVERLAY DESIGN PROCEDURES

The purpose of overlay design is to determine the requirement of the thickness of either the

bituminous layer or the granular layer which, when placed on the existing pavement, will

overcome the strength deficiencies of the pavement and ensure that the pavement retains

its structural integrity throughout the design period. An important aspect of any overlay

design procedure is to assess the structural soundness or strength of the existing pavement.

Various equipment are available to evaluate the structural condition of the pavements.

Some popularly used overlay design procedures are discussed briefly in the following

sections.

33

2.12.1 Overlay Design Procedures based on Structural Evaluation using FWD AASHTO pavement design guide [AASHTO, 1993] recommends the use of FWD for the

evaluation of roadbed soil resilient modulus (MR) and the effective Structural Number (SNeff)

of in-service pavements. MR of the subgrade is estimated using the Equation 2.9.

The overlay thickness required to increase the structural capacity to carry the future traffic

is determined by the following equation.

SNol = aol * Dol = SNf - SNeff (2.57a)

where, SNol= required overlay structural number; aol= structural coefficient for the

AC overlay; Dol= required overlay thickness, inches; SNf = structural number

required to carry future traffic; SNeff = effective structural number of the existing

pavement

3peff ED0045.0SN = (2.57b)

where, D = total thickness of all pavement layers above the subgrade

(inches); Ep= effective modulus (psi) of all pavement layers above the

subgrade computed from Equation 2.57c.

+

+

+

=P

2

3

R

pR

o E

aD

1

11

M

E

aD

1M

1ap5.1d (2.57c)

where p = load contact pressure (psi); a = FWD load plate radius (inches); do= deflection

measured at the centre of the load plate (inches) corrected for a temperature of 68oF.

Sidess et al [1992] developed an overlay design procedure based on FWD measurements.

In this procedure, the measured deflections are used to calculate Surface Curvature Index

(SCI). Equivalent modulus of the pavement section is determined using DEFMOD

backcalculation program and is correlated to SCI. Pavement is classified as weak, medium

and strong based on the deflection measured at 1800 mm (D6) and the SCI value. Using

these values, pavements are again classified in a quantitative manner such as the Structural

Index (SI). Overlay thickness charts were developed for different SI values and for weak,

34

medium and strong subgrades for different traffic levels. Figure 2.3 shows a typical overlay

design thickness chart for medium subgrade.

Figure 2.3 Overlay Thickness Versus Structural Index (SI) for Medium Subgrade

Mamlouk [1990] proposed a rational overlay design method for flexible pavements in

Arizona State based on roughness, fatigue and plastic deformation models. FWD tests were

conducted and the backcalculated moduli values were used to develop fatigue and plastic

deformation models. These models were incorporated in the microcomputer program CODA

which calculates overlay design thickness and the remaining life of the pavement.

PAVMAN computer program was developed by Richer and Irwin [1988] to calculate required

overlay thickness and remaining life of the existing pavement. MODCOMP 2 was used to

determine the pavement layer moduli from FWD data, which is also a part of the overlay

design program.

Abdallah et al, [2000] developed Artificial Neural Network (ANN) models for the

determination of critical strains at layer interfaces for three and four layer flexible pavement

systems from which the remaining life and overlay thickness required for an existing

pavement can be estimated using available fatigue and rutting criteria. The deflections

measured using FWD are main among the inputs to artificial neural network models.

35

Brown et al, [1987] proposed an approach for overlay design based on FWD deflection

measurements. In this method, FWD deflection studies are conducted at two places i.e, at

the center of the lane and along the wheel track. These deflections are used to

backcalculate the layer stiffnesses which are then adjusted for temperature and speed. The

life of the pavement against fatigue cracking and permanent deformation modes of failure is

calculated using Nottingham performance criteria [Brown and Brunton, 1986]. The

backcalculated layer moduli obtained using center lane deflections are used to determine

the original fatigue life of the pavement. Fatigue damage is computed as

Fatigue damage = Np / Nt (2.58a)

where Np = traffic carried by the existing pavement in msa; Nt = estimated fatigue

life of the existing pavement.

Overlay thickness requirement is obtained using Miners principle in which the total damage

must be less than or equal to 1. If the design traffic is N then

Np / Nt + N / Nn =1 (2.58b)

where Nn = new total fatigue life of the pavement at the reduced level of tensile

strain due to overlay thickness.

Thickness requirement from permanent deformation consideration is also determined

similarly. Using the backcalculated moduli values adjusted to the design conditions,

maximum vertical compressive strain at the top of the subgrade is calculated. The desired

overlay thickness is obtained from a plot of traffic-induced strain versus thickness. The

larger of the two thicknesses obtained from fatigue and rutting criteria is chosen as the

overlay thickness.

Arnold [1999] proposed the following limiting subgrade strain criterion for the design of

granular overlays for thin surfaced pavements in New Zealand.

23.0subcvs Rx

= (2.59) where cvs = limiting design vertical compressive strain at the top of the subgrade

R = ratio of future to past traffic (Nf/NP); sub = backcalculated (before overlay or

in-situ stabilization) vertical compressive strain at the top of the subgrade.

The backcalculated strains are computed using the backcalculated moduli obtained from

FWD measured deflections.

Idaho Flexible Overlay Design Method

This overlay design method proposed by Bayomy et al [1996] adopts the Asphalt Institute

fatigue and rutting failure criteria. In-situ moduli are evaluated either by backcalculation

36

using FWD deflection basins or by suitable laboratory tests. Seasonal adjustment factors for

subgrade, base, subbase and asphalt materials are considered in the design approach.

WSDOT Overlay Design Method

EVERPAVE, a computer program developed by Washington State Department of

Transportation (WSDOT) for mechanistic - empirical overlay design [Mahoney and Pierce,

1996; Pierce and Mahoney, 1996] makes use of the effective layer moduli backcalculated

from FWD deflections using EVERCALC backcalculation program. The failure modes

considered are (i) rutting and (ii) fatigue cracking.

Nf = 1.05 X 10-2 /[ v] 4.4841 (2.60)

where Nf = Allowable number of 80 kN single axles so that rutting does not exceed

12.7 mm; v = Vertical compressive strain at the top of subgrade layer

The fatigue cracking failure criterion is based on a laboratory-based model. The laboratory

fatigue life is calibrated (shifted) using a shift factor (SF) to correspond to field

performance.

Nfield = (Nlab) (SF) (2.61)

where Nfield = number of load applications to cause 10 % or less fatigue cracking in

wheel path; Nlab = laboratory fatigue life = 10(14.82-3.291 log t-0.854 Eac; t = horizontal

tensile strain at the bottom of AC layer (X 10-6); Eac = modulus of AC layer (ksi) that

changes with seasonal temperature; SF = shift factor (3 to 10) depending on the

asphalt concrete thickness, equivalent standard axle loads, climate and construction

quality.

WSDOT overlay design flow chart is shown in Figure 2.4.

North Carolina DOT Method

The North Carolina DOT method [Corley, 1996] is based on FWD measured deflection bowls

and performance data obtained in North Carolina. Asphalt layer strain is calculated using the

curvature of the deflection bowl. The expressions for the calculation of radius of curvature

and asphalt strain are given below.

R = -a2/[2*(Do-Dedge)] (2.62a)

AC strain = ac thick/(2*R) (2.62b)

where Do = deflection under the load plate, Dedge = deflection at edge of load plate

calculated from curve fit to individual deflection bowl; a= radius of load plate; R =

reciprocal of radius of deflection bowl.

37

Figure 2.4 WSDOT Overlay Design Flow Chart

The strain values computed using Equation 2.62a are used in Equation 2.63 to

estimate the life of the pavement.

N = (5 x 10-6)(1/ac strain) 3 (2.63)

where N = number of equivalent single axle loads (ESAL) to failure.

2.12.2 Overlay Design Using Other Methods

Austroads [1992] overlay design procedure uses the curvature of the deflected pavement

surface in addition to single maximum deflection as acceptable criteria since deflection alone

does not give reliable indication of the likelihood of fatigue cracking. For deflection

measurement, Benkelman beam and Lacroix Deflectograph are used. Charts are available

for determining the overlay thickness required from characteristic deflection and curvature

only adjusted for temperature. The design thickness is checked against fatigue cracking

and permanent deformation also.

In the Shell method [1978] for flexible overlay design, fatigue cracking of bound layers

and rutting are the two modes of failure considered in design. BISAR computer program,

FWD Deflection Data Pavement Layer Data

Backcalculation Programme (EVERCALC)

Pavement Layer Moduli

Traffic Data- ESAL for Design Period or ESAL per

year or Average Daily Traffic

General Data: - Load, Tyre Pressure & Spacing,

Shift factors, Seasonal Temperature & Adjustments

Pavement Data:- Poissons Ratio &Overlay

Moduli, Initial Overlay thickness & increment, Existing layer moduli&

Poisson ratio,

Overlay Design (EVERPAVE)

Overlay Thickness

38

which is based on linear elastic layer theory, is used for the analysis of pavements. The

residual life of the existing pavements is determined as the difference between the original

design life of the pavement and the life used prior to testing [Shell, 1978]. The main steps

followed in the design method are:

Select an appropriate mix code for the bituminous mix of the existing structure or

determine from measurement.

Determine the subgrade modulus from FWD measurements on the existing pavement.

Calculate the original life of the pavement (ND1) from the chart using other data such as effective thickness, weighted mean average air