15.03.2015 Views

Structural Design of Pavements PART VI Structural ... - TU Delft

Structural Design of Pavements PART VI Structural ... - TU Delft

Structural Design of Pavements PART VI Structural ... - TU Delft

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CT 4860<br />

<strong>Structural</strong> <strong>Design</strong> <strong>of</strong> <strong>Pavements</strong><br />

January 2006<br />

Pr<strong>of</strong>.dr.ir. A.A.A. Molenaar<br />

<strong>PART</strong> <strong>VI</strong><br />

<strong>Structural</strong> Evaluation and<br />

Strengthening <strong>of</strong> Flexible <strong>Pavements</strong><br />

Using Deflection Measurements and<br />

Visual Condition Surveys


<strong>Structural</strong> <strong>Design</strong> <strong>of</strong> <strong>Pavements</strong><br />

<strong>PART</strong> <strong>VI</strong><br />

<strong>Structural</strong> Evaluation and Strengthening<br />

<strong>of</strong> Flexible <strong>Pavements</strong><br />

Using Deflection Measurements<br />

and Visual Condition Surveys<br />

January 2006<br />

Pr<strong>of</strong>.dr.ir. A.A.A. Molenaar


2<br />

Table <strong>of</strong> contents:<br />

Preface 3<br />

1. Introduction 4<br />

2. Usage and condition dependent maintenance 6<br />

3. Deflection measurement tools 7<br />

3.1 Falling weight deflectometer 7<br />

3.2 Benkelman beam 8<br />

3.3 Lacroix deflectograph 9<br />

3.4 Factors influencing the magnitude <strong>of</strong> the measured FWD deflections 10<br />

4. Measurement plan 13<br />

4.1 Estimation <strong>of</strong> the number <strong>of</strong> test points per section 13<br />

4.2 Development <strong>of</strong> a measurement plan 14<br />

5. Statistical treatment <strong>of</strong> raw deflection data and selection <strong>of</strong> a location representative<br />

for the (sub)section 18<br />

6. Back calculation <strong>of</strong> layer moduli 26<br />

6.1 Surface modulus 26<br />

6.2 Back calculation <strong>of</strong> layer moduli 28<br />

6.3 Example 29<br />

7. Analysis <strong>of</strong> Benkelman beam and Lacroix deflectograph deflection bowls 33<br />

8. Estimation <strong>of</strong> the remaining life using an empirical based method 38<br />

9. Mechanistic procedures for remaining life estimations and overlay design 43<br />

9.1 Basic principles 43<br />

9.2 Extension <strong>of</strong> the basic principles 45<br />

10. Extension <strong>of</strong> the simplified procedure to estimate critical stresses and strains 51<br />

10.1 Relations between deflection bowl parameters and stresses and strains<br />

at various locations in the pavement 51<br />

10.2 Temperature correction procedure 54<br />

10.3 Relationships with other strength indicators such as SNC 54<br />

10.4 Relationships between falling weight deflections and deflections<br />

measured with the Benkelman beam 55<br />

11. Remaining life estimation from visual condition surveys 56<br />

12. Procedures to estimate material characteristics 58<br />

12.1 Fatigue characteristics <strong>of</strong> asphalt mixtures 58<br />

12.2 Deformation resistance <strong>of</strong> unbound base materials 59<br />

12.3 Subgrade strain criterion 59<br />

12.4 Maximum tensile strain at bottom <strong>of</strong> the bound base 59<br />

13. Overlay design in relation to reflective cracking 61<br />

13.1 Overlay design method based on effective modulus concept 61<br />

13.2 Method based on stress intensity factors 63<br />

13.3 Ovelay design method based on beam theory 64<br />

13.4 Effects <strong>of</strong> reinforcements, geotextiles, SAMI’s and other interlayer systems 69<br />

13.5 Load transfer across cracks 70<br />

14. Effect <strong>of</strong> pavement roughness on the rate <strong>of</strong> deterioration 72<br />

References 73


3<br />

Preface:<br />

<strong>Pavements</strong> deteriorate due to damaging effects <strong>of</strong> traffic and environmental loads and at a<br />

given moment in time maintenance is needed. Maintenance activities can grossly be divided<br />

into two categories.<br />

The first category is the so called routine maintenance which is mainly applied to keep the<br />

pavement surface in such a condition that it provides good service to the public but also to<br />

limit the effects <strong>of</strong> ageing. Routine maintenance consists e.g. <strong>of</strong> crack filling, local repairs and<br />

the application <strong>of</strong> surface dressings. Normally this type <strong>of</strong> maintenance is not too expensive.<br />

The costs <strong>of</strong> a surface dressing are approximately fl 6/m 2 while filling <strong>of</strong> cracks costs<br />

approximately fl 2.5/m ’ . Routine maintenance is done on a regular basis; the time period<br />

between two successive applications depends <strong>of</strong> course on the rate <strong>of</strong> deterioration which in<br />

turn is affected by the damaging power <strong>of</strong> traffic and climate and by the workmanship <strong>of</strong> the<br />

maintenance crews.<br />

The second category is much more capital intensive. Now we are dealing with strengthening<br />

<strong>of</strong> the pavement for which overlays are needed or partial or complete reconstruction. This<br />

type <strong>of</strong> maintenance is less <strong>of</strong>ten required than routine maintenance.<br />

Because pavement strengthening is such a costly affair, investigations to determine precisely<br />

the extent and severity <strong>of</strong> the damage and the rate <strong>of</strong> progression are strongly recommended.<br />

If a pavement surface e.g. shows severe cracking, removing this layer and replacing it by a<br />

new one seems to be a sensible solution. If however the cracking is due to the very low<br />

stiffness <strong>of</strong> the base and no measure are taken to improve the bending stiffness <strong>of</strong> the base<br />

layer, then the cracking will soon reappear.<br />

This simple example already illustrates that, in order to be able to make a proper selection <strong>of</strong><br />

the maintenance treatments available, one not only should know where something is going<br />

wrong but also why.<br />

Understanding why the pavement fails means that one needs knowledge on the stresses and<br />

strains in the pavement as well as the strength <strong>of</strong> materials. The process <strong>of</strong> gaining this<br />

knowledge is called “evaluation <strong>of</strong> the structural condition <strong>of</strong> pavements”.<br />

As it will be shown in these lecture notes, deflection measurements are an extremely useful<br />

tool in the assessment <strong>of</strong> the structural condition <strong>of</strong> the pavement. During a deflection<br />

measurement, the bending <strong>of</strong> the pavement surface due to a well-defined test load is measured.<br />

This is called the measurement <strong>of</strong> surface deflections. It is clear that the magnitude <strong>of</strong><br />

the deflections and especially the curvature <strong>of</strong> the deflection bowl reveal important information<br />

on the bending stiffness <strong>of</strong> the pavement.<br />

In the notes ample attention is paid to the techniques for measuring deflections, the way how<br />

the measurement results can be processed to obtain information on the stiffness <strong>of</strong> the<br />

individual pavements layers and how they can be used to determine the required thickness <strong>of</strong><br />

the overlays to be applied.<br />

Although all possible care has been given during the preparation <strong>of</strong> these notes to avoid<br />

typing errors etc., it is always possible that some “bugs” are still present. Furthermore the<br />

reader can have suggestions about certain parts <strong>of</strong> the material presented. It would be highly<br />

appreciated if you could send your comments to the author using the following email address.<br />

a.a.a.molenaar@citg.tudelft.nl


4<br />

1. Introduction:<br />

These lecture notes are dealing with deflection measurements, how they should be performed<br />

and how the results can be used to determine the remaining life <strong>of</strong> the pavement and the<br />

maintenance that has to be performed.<br />

The importance <strong>of</strong> deflection measurements can be described by means <strong>of</strong> the following<br />

example. When children have to build a bridge across a creek using a wide variety <strong>of</strong> wooden<br />

beams, their instinct will tell them that they better select those planks that show the lowest deflection<br />

under load. They also know that it is wiser to place the beams like shown in figure 1a<br />

than in figure 1b.<br />

A<br />

B<br />

Figure 1: Children know by instinct that placing a beam according to A is more effective than<br />

placing it according to B.<br />

As civil engineers we know that the selection by the children is a correct one because beam A<br />

has lower stresses and strains at the outer fibres than beam B when both beams are<br />

subjected to the same load. However as civil engineers we also know that the question “is it<br />

safe or not to cross a beam which shows a maximum deflection <strong>of</strong> 2 mm” cannot be<br />

answered without knowledge <strong>of</strong> the span <strong>of</strong> the beam, the load applied and the strength <strong>of</strong> the<br />

material from which it is made. This clearly indicates that measurement <strong>of</strong> only the maximum<br />

deflection gives some information about the strength <strong>of</strong> the beam but that more information is<br />

needed. We would already be in a much better shape if the curvature <strong>of</strong> the deflection bowl<br />

due to the load was known.<br />

The same is true for pavements. In order to get useful information about the flexural stiffness<br />

<strong>of</strong> the pavement one should measure the deflection due to a test load at various distances<br />

from the load centre.<br />

We know that the flexural stiffness is determined by the stiffness <strong>of</strong> the subgrade and the<br />

stiffness modulus and thickness <strong>of</strong> the layers placed on top <strong>of</strong> the subgrade. It will then be<br />

obvious that it must be possible to back calculate the stiffness modulus <strong>of</strong> each <strong>of</strong> the individual<br />

layers if the deflection bowl due to a defined test load is known as well as the thickness<br />

<strong>of</strong> each pavement layer.<br />

If the stiffness modulus <strong>of</strong> each layer is known together with its thickness, then the stresses<br />

and strains in any location in the pavement can be calculated.<br />

Knowledge on the strength <strong>of</strong> materials however is absolutely needed for the determination <strong>of</strong><br />

whether or not the pavement is capable <strong>of</strong> carrying the traffic loads expected in the future and<br />

whether or not it should be strengthened.<br />

All this means that the usefulness <strong>of</strong> a deflection measurement program without paying<br />

proper attention to the strength <strong>of</strong> materials can be doubted.<br />

In order to determine to what extent traffic loads have resulted in a deterioration <strong>of</strong> the pavement<br />

strength, deflections should be measured regularly during the pavement life. Since<br />

deflection measurements are fairly costly, one should make a realistic estimate <strong>of</strong> the number<br />

<strong>of</strong> measurements to obtain a picture <strong>of</strong> the deterioration trend line that develops in time. One<br />

should however be aware <strong>of</strong> the fact that the trend lines one wants to establish are influenced<br />

by variations in temperature (effect on stiffness modulus <strong>of</strong> the asphalt layers) and moisture<br />

(effect on stiffness modulus <strong>of</strong> the subgrade) and that the deflections measured over a certain<br />

stretch <strong>of</strong> road might show a considerable variation because <strong>of</strong> variations in layer thickness<br />

and stiffness modulus. Another question, which then arises, is how many measurement loca-


5<br />

tions should be tested in a certain section in order to obtain a realistic picture <strong>of</strong> the flexural<br />

stiffness <strong>of</strong> the pavement.<br />

In these lecture notes we will deal with all these aspects. The structure <strong>of</strong> the notes is as follows.<br />

Attention will be paid to the development <strong>of</strong> a measurement program. This will be<br />

followed with a discussion on the determination <strong>of</strong> the number <strong>of</strong> measurements required per<br />

section and the statistical treatment <strong>of</strong> the deflection data.<br />

Although the Benkelman beam was developed some 40 years ago, it is still in use in many<br />

countries. This is also the case with the automated version <strong>of</strong> the Benkelman beam called the<br />

Lacroix deflectograph. A chapter has been devoted to these devices and especially<br />

procedures to correct the measured deflections to true deflections are discussed.<br />

After that attention will be paid to some simple techniques allowing the overall stiffness <strong>of</strong> the<br />

pavement structure to be assessed and potential problem layers to be identified. Then the<br />

back calculation <strong>of</strong> stiffness moduli will be treated.<br />

This will be followed by a discussion on the design <strong>of</strong> overlays in which probabilistic principles<br />

are introduced.<br />

After that ample attention will be paid to an analysis method which allows critical strains to be<br />

evaluated without the need to back calculate layer moduli. This method is <strong>of</strong> special interest in<br />

case accurate information on the layer thickness is not available.<br />

Then attention is paid on the importance <strong>of</strong> visual condition surveys. A method will be<br />

presented that allows the remaining life to be estimated from such surveys.<br />

This chapter is followed by a chapter on the estimation <strong>of</strong> material strength characteristics like<br />

the fatigue resistance <strong>of</strong> asphalt mixtures and the resistance to permanent deformation <strong>of</strong><br />

unbound granular materials.<br />

Reflective cracking is an important issue and the commonly used overlay design methods<br />

don’t take into account this important phenomenon. Therefore a chapter dealing with the<br />

design <strong>of</strong> overlays controlling reflective cracking is presented.<br />

Finally the effect <strong>of</strong> pavement roughness on pavement deterioration will be discussed and<br />

simple procedures to estimate pavement roughness will be given.<br />

First <strong>of</strong> all however attention will be paid to the question why pavement maintenance has to<br />

rely on regular monitoring <strong>of</strong> the pavement condition and why the decision on applying<br />

maintenance cannot be taken simply on the basis <strong>of</strong> the number <strong>of</strong> years the pavement is in<br />

service or the number <strong>of</strong> loads that have been applied to the pavement.


6<br />

2. Usage and condition dependent maintenance:<br />

<strong>Pavements</strong> deteriorate due to the combined influences <strong>of</strong> traffic and environmental loads.<br />

This means that at a given moment maintenance activities should be scheduled in order to<br />

restore the level <strong>of</strong> service the pavement should give to the road user. It will be obvious that<br />

careful consideration should be given to the planning and the selection <strong>of</strong> the maintenance<br />

activity. The right strategy should be applied on the right spot at the right time.<br />

Planning <strong>of</strong> maintenance can be sometimes a rather simple task to perform. If we consider<br />

e.g. the maintenance <strong>of</strong> our illumination systems, we observe that in a number <strong>of</strong> cases (e.g.<br />

hospitals) the bulbs are not replaced after failure, but after a certain number <strong>of</strong> burning hours.<br />

This way <strong>of</strong> maintenance is called “usage dependent maintenance”, because the replacement<br />

is done after a certain time period the object to be maintained is used.<br />

There are three important reasons why such a type <strong>of</strong> maintenance is possible and accepted<br />

for illumination.<br />

a. For some reasons we don’t accept to be in the dark (safety, interruption <strong>of</strong> work).<br />

b. We know quite precisely what the mean lifetime is <strong>of</strong> the light bulbs.<br />

c. We know quite precisely what the variation is <strong>of</strong> the lifetime <strong>of</strong> the light bulbs and we<br />

know that this variation is small.<br />

This way <strong>of</strong> performing maintenance is not very suited to be applied on pavements for the<br />

following reasons.<br />

a. In most cases some degree <strong>of</strong> failure is acceptable on pavements. Traffic can e.g. drive<br />

at a fairly high speed level although there is a substantial amount <strong>of</strong> cracking. This implies<br />

that some damage types can be allowed to occur over a significant area and with a<br />

significant severity before an unacceptable level <strong>of</strong> service is reached.<br />

b. Although pavements have been subjected to extensive research, the predictive capability<br />

<strong>of</strong> our performance models is still limited. Even the accuracy <strong>of</strong> our models to predict the<br />

mean pavement life is quite <strong>of</strong>ten disappointing.<br />

c. <strong>Pavements</strong> exhibit a substantial amount <strong>of</strong> variation in performance mainly due to the<br />

variation in layer thickness, material characteristics etc.. This means that two pavements<br />

which are nominally the same and which are loaded under nominally the same conditions<br />

can show a significant difference in initiation and progression <strong>of</strong> damage.<br />

All in all a strategy which implies maintenance to be performed after the pavement has been<br />

in service for a certain number <strong>of</strong> years is not applicable for road networks. A certain amount<br />

<strong>of</strong> damage can mostly be allowed because pavement failure seldom results in catastrophic<br />

events. Furthermore the variation in pavement life is such that usage dependent maintenance<br />

cannot be made cost effective.<br />

This implies that the planning and selection <strong>of</strong> maintenance strategies for pavements heavily<br />

relies on input coming from condition observations and predictions based there on. Such an<br />

approach to maintenance is called “condition dependent maintenance”.<br />

This immediately means that tools should be available to monitor the condition <strong>of</strong> the<br />

pavement. An overview <strong>of</strong> such tools is already given in [1]. The lecture notes we have in<br />

front <strong>of</strong> us are dealing with one <strong>of</strong> the most important evaluation tools being the deflection<br />

measurement device.


7<br />

3. Deflection measurement tools:<br />

The deflection device that currently receives the highest popularity is the falling weight<br />

deflectometer (FWD). Nevertheless other deflection measuring devices like the Benkelman<br />

Beam (BB) and the Lacroix Deflectograph (LD) are still used at different places at the world.<br />

Especially the Benkelman Beam deserves attention since this low cost device (the price is<br />

approximately 1/30 th <strong>of</strong> the price <strong>of</strong> a falling weight deflectometer) is used in many<br />

developing countries. The principles <strong>of</strong> these three devices are given elsewhere [1], here only<br />

the main features will be described.<br />

3.1 Falling weight deflectometer:<br />

The principle <strong>of</strong> the FWD is schematically shown in figure 2.<br />

Figure 2: Principle <strong>of</strong> the falling weight deflectometer.<br />

A weight with a certain mass drops from a certain height on a set <strong>of</strong> springs (normally rubber<br />

buffers) which are connected to a circular loading plate which transmits the load pulse to the<br />

pavement. Load cells are used to monitor the magnitude and duration <strong>of</strong> the load pulse. The<br />

magnitude <strong>of</strong> the load pulse can vary between the 30 and 250 kN depending on the mass <strong>of</strong><br />

the falling weight and the falling height. The duration <strong>of</strong> the load pulse is mainly dependent on<br />

the stiffness <strong>of</strong> the rubber buffers. Usually pulse duration between 0.02 and 0.035 s are<br />

measured.<br />

The surface deflections are measured with so called geophones. These are velocity transducers<br />

which measure the vertical displacement speed <strong>of</strong> the surface. By integration the displacements<br />

are obtained.<br />

Since the electronic circuits are only opened a very short moment before the weight hits the<br />

buffers, the influence <strong>of</strong> passing traffic on the magnitude <strong>of</strong> the deflections is eliminated; only<br />

the displacements due to the impact load are measured.<br />

The advantage <strong>of</strong> the FWD is the short duration <strong>of</strong> the load pulse comparable to the duration<br />

<strong>of</strong> the load pulse caused by a truck driving at approximately 50 km/h. Because <strong>of</strong> the short<br />

pulse duration, the influence <strong>of</strong> viscous effects can be neglected.<br />

One should however be cautious when the modulus <strong>of</strong> a saturated subgrade with a high<br />

ground water level is determined from the deflection measurement results. In that case one<br />

might measure the bulk modulus K <strong>of</strong> the subgrade which, in case <strong>of</strong> a fully saturated<br />

subgrade, can be high. Because road materials are very much sensitive for shear, this high<br />

bulk modulus value gives a wrong idea about the real stiffness <strong>of</strong> the material. This can be<br />

illustrated with the following simple example.


8<br />

When a swimmer makes a nice dive from the diving tower he will hit the water in a gentle<br />

way, without too much <strong>of</strong> splash and without hurting himself. We can say that with such a nice<br />

dive he experiences the shear modulus G <strong>of</strong> water which, as we all know, is very low.<br />

However when he falls flat on his stomach, his dive is causing him much pain and probably a<br />

blue stomach. In this case he experiences the bulk modulus K <strong>of</strong> water which, as we know, is<br />

very high. A fluid with no air bubbles is in fact incompressible.<br />

3.2 Benkelman beam:<br />

The principle <strong>of</strong> the Benkelman beam, invented by A.C. Benkelman is schematically shown in<br />

figure 3.<br />

Figure 3: Principle <strong>of</strong> the Benkelman beam.<br />

The measuring system consists <strong>of</strong> a beam that can rotate around a pivot attached to a<br />

reference frame. The load is supplied by a truck that slowly moves to or from the tip <strong>of</strong> the<br />

beam.<br />

The advantage <strong>of</strong> the BB is the fact that the device is simple and cheap. The disadvantage is<br />

the slow speed <strong>of</strong> the truck that can cause all kinds <strong>of</strong> viscous effects making the<br />

measurements difficult to interpret. Furthermore the effects <strong>of</strong> passing vehicles on the<br />

magnitude <strong>of</strong> the deflection cannot be neglected. Finally it should be mentioned that the<br />

supports <strong>of</strong> the reference frame could stand in the deflection bowl. This means that the frame<br />

is not a true reference and corrections for movement <strong>of</strong> the support system have to be made<br />

in order to obtain the true deflections.<br />

Quite <strong>of</strong>ten only the magnitude <strong>of</strong> the rear axle load <strong>of</strong> the truck used as loading vehicle for<br />

the BB measurements is reported. This is absolutely insufficient; precise knowledge <strong>of</strong> the<br />

tyre pressure, tyre spacing and area <strong>of</strong> the tyre print is necessary in order to allow proper<br />

analyses to be made.<br />

Different measurement procedures exist and one should strictly adhere to the guidelines for<br />

doing the measurements when one <strong>of</strong> such procedures is used.<br />

Furthermore one should realise that the dimensions <strong>of</strong> the BB can differ. There are devices<br />

with shorter and longer measuring beams. One should take good notice <strong>of</strong> this in order to<br />

overcome that a beam is used that doesn’t comply with the requirements set in the procedure<br />

to be used.


9<br />

3.3 Lacroix deflectograph:<br />

Figure 4 shows the Lacroix deflectograph (LD). The principle <strong>of</strong> the measurement is the same<br />

as that <strong>of</strong> the Benkelman beam. The major difference however is that the measuring system<br />

is attached to the loading vehicle and that it is moved automatically to the next measuring<br />

position. This procedure is schematically shown in figure 5.<br />

It is obvious that the LD has large advantages over the BB. First <strong>of</strong> all the measurements are<br />

continuously taken and are far less affected by the varying speed <strong>of</strong> the loading vehicle. With<br />

the BB measurements the speed <strong>of</strong> the truck varies between 0 (at the beginning <strong>of</strong> the<br />

measurements) and approximately 5 km/h when the truck drives at constant speed. The<br />

speed <strong>of</strong> the LD vehicle is more or less constant at 5 km/h.<br />

The LD however suffers from the same disadvantages as the BB. The low speed can cause<br />

that the viscous behaviour <strong>of</strong> the asphalt surfacing cannot be neglected and corrections for<br />

movement <strong>of</strong> the reference frame need to be applied.<br />

Because the entire measurement procedure is automated, much more measurements can be<br />

taken with the LD as with the BB in the same time period. This however has its price; the LD<br />

has about the same price level as the FWD.<br />

Figure 4: Principle <strong>of</strong> the Lacroix deflectograph.


10<br />

Figure 5: Principle <strong>of</strong> the automatic positioning <strong>of</strong> the measuring system <strong>of</strong> the LD.<br />

3.4 Factors influencing the magnitude <strong>of</strong> the measured FWD deflections:<br />

When civil engineers are dealing with measurements they quite <strong>of</strong>ten show a bad habit which<br />

is that they accept the measurement result as “the truth”. They seldom realise that the measurement<br />

result is affected by a large number <strong>of</strong> factors and that the magnitude <strong>of</strong> the<br />

influence <strong>of</strong> these factors should be known in order to avoid misinterpretations. A number <strong>of</strong><br />

such influence factors on deflections measured with a FWD will be discussed here. The<br />

material presented is based on the excellent work done on this topic by van Gurp which is<br />

reported in [10].<br />

When a number <strong>of</strong> FWD devices are used on the same pavement to measure the deflections,<br />

one will notice that all these devices will not measure the same value. This is even true when<br />

the deflections are corrected to a particular load level. Some reasons for that are described<br />

hereafter.<br />

It is a well-known fact that the stiffness <strong>of</strong> rubber is temperature dependent. At higher<br />

temperatures the stiffness will be lower than at lower temperatures. This is nicely shown in<br />

figure 6 where the stiffness <strong>of</strong> a particular rubber buffer used in a particular FWD is given in<br />

relation to the load level and the temperature.<br />

It will be obvious that the temperature in the rubber buffers will vary when a FWD survey is<br />

done starting early morning and ending late afternoon. This is not only because <strong>of</strong> the<br />

variation in air temperature but also because <strong>of</strong> the cumulative energy that is collected in the<br />

buffer, and that is transformed in heat, because <strong>of</strong> the large number <strong>of</strong> measurements that<br />

are taken during the day. This means that the stiffness <strong>of</strong> the rubber buffer will vary during the


11<br />

day. The effect is <strong>of</strong> course more pronounced if measurements done in the winter have to be<br />

compared with those done in the summer.<br />

Figure 6: Static spring constant <strong>of</strong> a particular rubber buffer used in a particular FWD.<br />

If for some reason the spring stiffness decreases, the shape <strong>of</strong> the load pulse changes. Its<br />

peak value will decrease while the duration <strong>of</strong> the pulse will increase. The longer duration <strong>of</strong><br />

the pulse might cause a somewhat s<strong>of</strong>ter response (lower stiffness) <strong>of</strong> the pavement. More<br />

important <strong>of</strong> course is the fact that differences between the devices occur if they have<br />

different buffers and if the deflections have to be corrected to a predefined load level.<br />

Furthermore one has to be careful when using the FWD for studies on the non linearity <strong>of</strong><br />

pavements. Especially pavements where the main body is formed by unbound materials, will<br />

show non linear behaviour. One might try to analyse this by doing deflection measurements at<br />

different load levels but from the text given above it will be clear that at least some <strong>of</strong> the non<br />

linearity that is measured is caused by the device itself!!<br />

Research in [10] has shown that it is wiser to correct the deflections based on the area<br />

enclosed by the load vs time plot rather than based on the peak load.<br />

Other effects, which are unfortunately more <strong>of</strong> the “black box” nature, are the following. As<br />

mentioned, geophones are used to measure the deflections. The nature <strong>of</strong> the geophones<br />

however is that their sensitivity reduce with decreasing frequency. Especially below 10 Hz,<br />

the sensitivity decreases rapidly. However these low frequencies contribute significantly to the<br />

frequency spectrum <strong>of</strong> a single deflection pulse. Especially the frequency spectrum <strong>of</strong><br />

deflection pulses measured on thin pavements laid on s<strong>of</strong>t subgrades will show the great<br />

contribution <strong>of</strong> the low frequencies. If the geophones don’t pick up these low frequencies, a<br />

too low deflection will be recorded and one would expect the pavement to have a higher<br />

flexural stiffness than it really has.<br />

This effect can be compensated by using high gain factors for the low frequencies. The way in<br />

which this is done depends however on the manufacturer and information on this is usually<br />

confidential information.<br />

It has also been shown in [10] that the system processor can deform the deflection readings.<br />

For one FWD system, the influence <strong>of</strong> the system processor appeared to be so large that it<br />

did not pass the calibration procedure and could therefore not be used in FWD surveys.<br />

Another influence factor is the smoothing <strong>of</strong> signals that is applied on the FWD deflections.<br />

This smoothing is done in order to get rid <strong>of</strong> high frequency disturbances. The question then<br />

always is what the cut-<strong>of</strong>f frequency should be. Studies reported in [10] have shown that if f =<br />

60 Hz is chosen as cut-<strong>of</strong>f frequency, the effect <strong>of</strong> the smoothing is minimal. Again it is noted<br />

that one should ask the FWD supplier to give details on this important aspect.


12<br />

From the text given above it is clear that there are several influence factors which cause that<br />

the deflections measured with one device are different from those measured with an other<br />

device. It is clear that calibration is vital in order to avoid unexpected and unacceptable<br />

differences between devices to occur.


13<br />

4. Measurement plan:<br />

The question always is how many measurements should be taken and where should the<br />

measurements be taken on a specific stretch <strong>of</strong> pavement in order to get a reliable picture <strong>of</strong><br />

the flexural stiffness <strong>of</strong> the pavement. Some guidelines for this will be given in this chapter.<br />

4.1 Estimation <strong>of</strong> the number to test points per section:<br />

In this section the method presented in [2] is described which allow the number <strong>of</strong> tests to be<br />

determined that are needed on a particular road section to obtain a proper insight in the<br />

bearing capacity <strong>of</strong> the pavement.<br />

One can calculate a statistical quantity R, called the limit <strong>of</strong> accuracy, which represents the<br />

probable range the true mean differs from the average obtained by “n” tests at a given degree<br />

<strong>of</strong> confidence. The larger n is, the smaller value will be obtained for R which means that the<br />

mean value calculated from the data obtained from the tests will differ less from the true mean<br />

value. The mathematical expression is:<br />

R = K α . ( σ / √ n )<br />

Where: K α<br />

σ<br />

= standardised normal deviate which is a function <strong>of</strong> the desired confidence<br />

level 100 . (1 - α),<br />

= true standard deviation <strong>of</strong> the random variable (parameter) considered.<br />

If the confidence level is chosen and if a proper estimate for σ is obtained, R is inversely<br />

proportional to the square root <strong>of</strong> the number <strong>of</strong> tests.<br />

Figure 7shows the basic shape <strong>of</strong> the relation between n and R.<br />

Figure 7: Typical limit <strong>of</strong> accuracy curve for all pavement variables showing general zones.


14<br />

As shown in figure 7, 3 zones can be discriminated. In zone I a small increase in the number<br />

<strong>of</strong> tests reduces the value <strong>of</strong> R tremendously and the accuracy <strong>of</strong> the predictions will increase<br />

drastically. In other words a small increase in budget to increase the number <strong>of</strong> data points is<br />

really value for money.<br />

In zone III, R hardly reduces with an increasing number <strong>of</strong> tests. This means that in this case<br />

very little extra value is obtained from an increased measurement budget.<br />

The optimal number <strong>of</strong> tests can be found in zone II.<br />

The main problem in calculating R is the assessment <strong>of</strong> the standard deviation σ. Since the<br />

magnitude <strong>of</strong> the deflections can vary quite considerably within one pavement section and<br />

between pavement sections (thick pavements compared with thin pavements), it is not<br />

possible to give a single value for σ. Nevertheless it is possible to give values for the<br />

coefficient <strong>of</strong> variation CV for the measured deflections which are observed in practice.<br />

Typical values are:<br />

CV = standard deviation / mean =<br />

0.15 low variation, typical for pavements which are in<br />

good condition,<br />

0.30 medium variation, typical for pavements which<br />

show a fair amount <strong>of</strong> damage,<br />

0.45 high variation, typical for pavements which<br />

show a large amount <strong>of</strong> damage.<br />

By using these CV values and adopting confidence levels <strong>of</strong> 95% (α = 0.05) and 85% (α =<br />

0.15), figure 8 has been constructed.<br />

The use <strong>of</strong> the procedure is illustrated by means <strong>of</strong> the following example. A deflection survey<br />

has to be performed on a road that is in reasonable condition and the question is how many<br />

measurements need to be taken to obtain a reliable picture <strong>of</strong> the flexural stiffness <strong>of</strong> the<br />

pavement. Because a reliable picture is desired the average deflection is allowed to differ 8%<br />

from the true mean. The required confidence level is 95%. Since the pavement shows some<br />

damage a CV is estimated <strong>of</strong> 20%. By interpolation, the position <strong>of</strong> the line for CV = 20% is<br />

estimated in figure 8a. Using this line and the R-value <strong>of</strong> 8%, the number <strong>of</strong> observations to<br />

be taken is equal to 7.<br />

4.2 Development <strong>of</strong> a measurement program:<br />

Before one decides on where and how many deflection measurements should be taken, a<br />

visual condition survey should preferably be performed. It is e.g. important to know which<br />

types <strong>of</strong> defects are present on the pavement and how the various defect types are<br />

distributed over the pavement surface. Is the damage evenly distributed or is the damage<br />

concentrated in a limited number <strong>of</strong> locations.<br />

A visual condition survey is not only needed to develop an effective measurement plan, but<br />

the condition data are also needed in the evaluation phase when decisions on the<br />

maintenance strategy to be applied need to be taken.<br />

The most important damage types to consider in the structural evaluation <strong>of</strong> pavements are <strong>of</strong><br />

course cracking and deformations because they are related to lack <strong>of</strong> flexural stiffness.<br />

If cracking and deformations occur rather locally it is not recommended to use an equal<br />

spacing between the measurement points but to locate them in such a way that an as good<br />

as possible sample <strong>of</strong> both sound and cracked cq deformed areas is obtained.<br />

For reasons that will be discussed later on, it is recommended to measure both the outer<br />

wheel track as well as the area between the wheel tracks, the latter being representative for<br />

the flexural stiffness <strong>of</strong> the undamaged pavement. These measurements are <strong>of</strong> course only<br />

useful if the area between the wheel tracks is not damaged.<br />

In case <strong>of</strong> severe longitudinal or transverse cracking, it is recommended to perform some<br />

measurements across the crack. This can be done very easily with the FWD using the<br />

geophone positions schematically shown in figure 9.


15<br />

Figure 8a: Graph to estimate the number <strong>of</strong> observations required at<br />

a confidence level <strong>of</strong> 95%.<br />

Figure 8b: Graph to estimate the number <strong>of</strong> observations required at<br />

a confidence level <strong>of</strong> 85%.


16<br />

Figure 9: Placement <strong>of</strong> loading plate <strong>of</strong> FWD and geophones for load transfer measurements.<br />

Deflection measurements across the crack are important in order to be able to determine the<br />

amount <strong>of</strong> load transfer. This parameter has a significant influence on the thickness <strong>of</strong> the<br />

overlay; if there is e.g. no load transfer at all, additional maintenance work like milling and<br />

filling <strong>of</strong> the cracked area might be necessary.<br />

The magnitude <strong>of</strong> the measured deflections is dependent on the temperature, which affects<br />

the stiffness <strong>of</strong> the asphalt layers, and the moisture content, which can have a significant<br />

effect on the stiffness <strong>of</strong> the subgrade and other unbound layers. This means that if measurements<br />

are taken at various periods <strong>of</strong> the year, corrections are needed in order to be able<br />

to compare them. In order to avoid the rather complex corrections due to moisture variation, it<br />

is recommended to take the measurements in the so-called “neutral” period. During such<br />

periods the moisture content in the unbound materials is approximately at its mean level. In<br />

the Netherlands that is the late April – early May period and the October month.<br />

Because BB and LD measurements are taken at relatively low speeds, one should not<br />

perform these measurements at too high temperature levels because otherwise viscous<br />

effects will have a significant influence on the measurements which makes interpretation<br />

there-<strong>of</strong> complicated. Also the temperatures should not be too low because then the<br />

deflections might be so small that accuracy problems occur in the measurement and<br />

monitoring <strong>of</strong> the deflections. For that reason the Transport and Road Research Laboratory<br />

(TRRL) in the UK has suggested the temperature ranges shown in table 1 at which the BB<br />

and LD measurements should preferably be taken.<br />

Maximum temperature 30 o C if bitumen has a penetration lower or equal than 50<br />

25 o C if bitumen has a penetration higher than 50<br />

Minimum temperature 5 – 10 o C depending on the structure<br />

Table 1: Maximum and minimum temperature for deflection measurements<br />

as specified by TRRL.


17<br />

One should realise that the influence <strong>of</strong> temperature always has to be taken into account and<br />

that the deflections measured always should be corrected to a reference temperature. The<br />

temperature correction procedure will be presented in an other chapter.


18<br />

5. Statistical treatment <strong>of</strong> raw deflection data and selection <strong>of</strong><br />

a location representative for the (sub)section:<br />

Statistical treatment <strong>of</strong> the data as measured is always needed in order to be able to recognise<br />

trends and in order to limit the amount <strong>of</strong> work that should be done in the evaluation<br />

process. It is e.g. not necessary and even not useful to back calculate the layer moduli for<br />

each measurement location simply because <strong>of</strong> the fact that it is impossible to obtain accurate<br />

layer thickness information for each and every location. It is therefore much more effective to<br />

concentrate the analysis on locations which can be taken as representative for a particular<br />

section or sub-section.<br />

Simple statistical procedures have shown to be very effective to discriminate homogeneous<br />

sub-sections within a larger section. A homogeneous sub-section is defined as a section<br />

where the deflections and so the flexural stiffness are more or less constant. When such<br />

homogeneous sub-sections have been determined, one has to take a point which can be<br />

taken as being representative for that sub-section. That point can be the location where the<br />

measured deflection bowl comes closest to e.g. the average deflection pr<strong>of</strong>ile or the 85%<br />

deflection pr<strong>of</strong>ile. The 85% pr<strong>of</strong>ile is the pr<strong>of</strong>ile that is exceeded by 15% <strong>of</strong> all the measured<br />

pr<strong>of</strong>iles.<br />

The so-called homogeneous sub-sections can be determined by means <strong>of</strong> the method <strong>of</strong> the<br />

cumulative sums. The cumulative sums are calculated in the following way.<br />

First <strong>of</strong> all the mean <strong>of</strong> a variable over the entire section is calculated (e.g. the mean <strong>of</strong> the<br />

maximum deflection). Then the difference between the actual value <strong>of</strong> the variable and the<br />

mean is calculated. Next these differences are summed. In formula the cumulative sums are<br />

calculated using:<br />

S 1 = x 1 - µ<br />

S 2 = x 2 - µ + S 1<br />

S n = x n - µ + S n-1<br />

Where: S n = cumulative sum at location n,<br />

x n = value <strong>of</strong> the variable considered at location n,<br />

µ = mean <strong>of</strong> variable x over entire section.<br />

The method is illustrated by means <strong>of</strong> an example. Table 2 shows the deflections that were<br />

measured by means <strong>of</strong> a FWD on a particular road in the Netherlands. The load applied was<br />

50 kN, the diameter <strong>of</strong> the loading plate was 300 mm. The table gives values for d 0 , d 300 , etc.;<br />

these are the deflections measured at a distance <strong>of</strong> 0 and 300 mm etc.. An important value is<br />

the surface curvature index SCI, which is the difference between the maximum deflection d0<br />

and the deflection, measured at 600 mm from the loading centre (d 600 ). Also the logarithm <strong>of</strong><br />

the SCI values is reported. Also this is an important characteristic as will be shown later on.<br />

As one will observe from the table, high deflections are measured and the amount <strong>of</strong> variation<br />

in the measured deflections is very high. It should be noted that the pavement considered<br />

was a polder road on a very weak subgrade and showed a significant amount <strong>of</strong> damage.<br />

It should be noted that the example presented is a rather extreme one; normally such large<br />

variations in deflections are not observed.<br />

Figure 10 is a graphical representation <strong>of</strong> the measured deflections, while figure 11 shows the<br />

variation <strong>of</strong> the SCI over the section. Figure 12 shows in a graphical form the variation <strong>of</strong> the<br />

cumsum (cumulative sum) as determined for the SCI. The SCI is selected as parameter<br />

decisive in the determination <strong>of</strong> the homogeneous subsections since the SCI can be<br />

considered to be the most important deflection parameter.<br />

Homogeneous sub-sections can easily be recognised from figure 12 since by definition an<br />

area over which the slope <strong>of</strong> the cumulative sums plot is more or less constant indicates an<br />

area where the differences between the actual measured values and the overall mean value<br />

are approximately the same.


19<br />

Table 2: Deflection testing results obtained on a particular section and summary statistics.


20<br />

Figure 10: Results <strong>of</strong> a deflection survey.


21<br />

Figure 11: Surface curvature index.


22<br />

Figure 12: Cumulative sum <strong>of</strong> the surface curvature index.


23<br />

The following sections are discriminated.<br />

Section<br />

Locations<br />

1 0.05-0.1-0.15<br />

2 0.2-0.25-0.3<br />

3 0.35-0.4-0.45<br />

4 0.5-0.55-0.6<br />

5 0.65 this is a single point clearly visible in the SCI plot<br />

6 0.7-0.75-0.8-0.85-0.9-0.95<br />

7 1 this is a single point clearly visible in the SCI plot<br />

8 1.05-1.1-1.05<br />

9 1.1-1.15-1.2-1.25-1.3-1.35-1.4-1.45<br />

10 1.5 this is a single point clearly visible in the SCI plot<br />

11 1.55-1.6<br />

12 1.65-1.7-1.75-1.8-1.85-1.9-1.95-2<br />

By means <strong>of</strong> the cumsum method we have arrived to a set <strong>of</strong> successive sub-sections, each<br />

<strong>of</strong> them having more or less a certain flexural stiffness. Now it is interesting to determine if we<br />

can combine a few sections. If this is possible we would reduce the work load. The question<br />

now is how to achieve that.<br />

If we compare the slopes <strong>of</strong> the different sections we notice that the slopes <strong>of</strong> sections 2, 4<br />

and 12 are about the same. This means that they can be taken as one section in the further<br />

analysis. This also holds for sections 1 and 6, so also these can be treated as one section.<br />

The same is true for sections 3, 8 and 11.<br />

Then we have a look to the single points that are discriminated and we try to assign them to a<br />

particular subsection. We observe that location 0.65 is clearly an isolated peak value and<br />

should therefore be treated as such. Location 1 however could very well be combined with<br />

section 2. Also location 1.5 is better treated as a single point.<br />

All in all we arrive to the subsections given below.<br />

Section<br />

Locations<br />

1 0.05-0.1-0.15 and 0.7-0.75-0.8-0.85-0.9-0.95<br />

2 0.2-0.25-0.3 and 0.5-0.55-0.6 and 1.65-1.7-1.75-1.8-1.85<br />

-1.9-1.95-2 and 1<br />

3 0.35-0.4-0.45 and 1.05-1.1-1.15 and 1.55-1.6<br />

4 0.65<br />

5 1.1-1.15-1.2-1.25-1.3-1.35-1.4-1.45<br />

6 1.5<br />

The statistics <strong>of</strong> the sub-sections mentioned above are tabulated below.<br />

Section Mean Value SCI Standard Deviation SCI Var. Coeff.<br />

1 420 111 26%<br />

2 175 65 37%<br />

3 494 62 13%<br />

4 962<br />

5 423 77 18%<br />

6 96<br />

As one will notice, rather high values for the coefficient <strong>of</strong> variation are still obtained for<br />

sections 1 and 2. We have to look then in table 2, in order to find out what the possible<br />

reasons for this could be. By doing so we observe that location 0.8 doesn’t really fit in section<br />

1 and should better be moved to section 2. The high variation in section 2 is probably caused<br />

by the inclusion <strong>of</strong> locations 1.7 and 1.75; also location 1.9 could contribute to the high<br />

variation. Therefore it is suggested to move location 1.9 to section 1 and to combine locations<br />

1.7 and 1.75 with location 1.5. We then obtain the sections and summary statistics as shown<br />

in table 3.<br />

As one can observe a better result in terms <strong>of</strong> lower coefficients <strong>of</strong> variation are obtained. The<br />

division in subsections as shown in table 3 will be used for further treatment.


24<br />

Section Locations Mean SCI SD SCI Var. Coeff.<br />

1 0.05-0.1-0.15-0.7-0.75-0.85-0.9<br />

-0.95-1.9 434 87 20%<br />

2 0.2-0.25-0.3-0.5-0.55-0.6-0.8-1<br />

-1.65-1.8-1.85-1.95-2 181 40 22%<br />

3 0.35-0.4-0.45-1.05-1.1-1.15-1.55-1.6 494 62 13%<br />

4 0.65 962<br />

5 1.1-1.15-1.2-1.25-1.3-1.35-1.4-1.45 423 77 18%<br />

6 1.5-1.7-1.75 87 11 13%<br />

Table 3: Homogeneous sub-sections based on SCI<br />

An other approach to the reduction <strong>of</strong> the data is to make a frequency plot <strong>of</strong> the deflections<br />

measured. Figure 13 is an example <strong>of</strong> such a plot based on the measured SCI’s. In making a<br />

frequency plot one has to decide about the number <strong>of</strong> classes to be used. A practical<br />

guideline for this is to take the number <strong>of</strong> classes equal to the square root <strong>of</strong> the number <strong>of</strong><br />

observations.<br />

From figure 13 it is clear that we have 1 observation in the range SCI = 0 – 72 µm, 13<br />

observations in the range SCI = 73 – 220 µm, 6 observations in the range SCI = 221 – 369<br />

µm, 13 observations in the range SCI = 370 – 517 µm, 6 observations in the range SCI = 518<br />

– 665 µm and one extreme value which is the SCI = 962 µm measured at location 0.65. The<br />

locations which belong to the frequency classes and the summary statistics are given in table<br />

4.<br />

Frequency Locations SCI.<br />

Class Mean St. Dev. Var. Coef.<br />

0 – 72 1.75 72<br />

73 – 220 0.2-0.25-0.3-0.5-0.8-1-1.5-1.65-1.7-1.8-1.85<br />

-1.95-2 157 37 24%<br />

221 – 369 0.05-0.55-0.6-1.2-1.3-1.9 306 42 14%<br />

370 – 517 0.1-0.15-0.35-0.45-0.7-0.85-0.95-1.15-1.25<br />

-1.35-1.4-1.45-1.55 430 40 9%<br />

518 – 665 0.4-0.75-0.9-1.05-1.1-1.6 550 33 6%<br />

higher 0.65 962<br />

Table 4: Frequency classes for the SCI, locations and summary statistics.<br />

As one can observe from table 4, this approach results in a grouping <strong>of</strong> the deflection data in<br />

such a way that the coefficient <strong>of</strong> variation in one group is limited to very small.<br />

From the description given above it will be clear that several techniques are available for<br />

reduction <strong>of</strong> the raw deflection data. In principle the cumulative sum technique is a very<br />

powerful tool to discriminate homogeneous sections. However situations might occur that<br />

even the cumsum technique results in sections which exhibit a rather high degree <strong>of</strong> variation.<br />

In that case reduction <strong>of</strong> data through an analysis <strong>of</strong> the frequency distribution can result in<br />

data sets which are rather homogeneous in nature.<br />

The big advantage <strong>of</strong> the cumsum technique is that it results in physical section units ready to<br />

receive maintenance whereas the other approach doesn’t result in such units.<br />

All in all this means that the data reduction process and the statistical analysis <strong>of</strong> the raw data<br />

is not a straightforward process. Each time the data set should be treated carefully in order to<br />

select the most appropriate way to reduce the data.<br />

The selection <strong>of</strong> the location which can be considered to be representative for the entire<br />

(sub)section is done in the following way. First <strong>of</strong> all one has to decide whether one wants to<br />

base the analysis on the mean conditions or whether one wants to do the analysis using a<br />

deflection pr<strong>of</strong>ile that is exceeded by only 15% <strong>of</strong> the measured pr<strong>of</strong>iles. In the first case one<br />

selects a measured pr<strong>of</strong>ile that comes closest to the mean pr<strong>of</strong>ile while in the second case<br />

one selects a measured pr<strong>of</strong>ile that comes closest to the 85% pr<strong>of</strong>ile.


25<br />

In section 1 <strong>of</strong> table 3, location 0.85 has the SCI value (453) that comes closest to the mean<br />

SCI value <strong>of</strong> that section being 434, while location 0.75 has the SCI value (525) that comes<br />

closest to the 85% pr<strong>of</strong>ile <strong>of</strong> that section being 521 (mean plus one standard deviation).<br />

These locations are then selected as being the representative locations for this section. Cores<br />

are taken at those locations to obtain accurate information on the thickness <strong>of</strong> the layers. This<br />

information is needed to allow accurate back calculations <strong>of</strong> the layer moduli to be made.<br />

Figure 13: Frequency distribution <strong>of</strong> the measured SCI values.


26<br />

6. Back calculation <strong>of</strong> layer moduli:<br />

Back calculation <strong>of</strong> layer moduli is quite <strong>of</strong>ten considered as an important step in pavement<br />

evaluation. The reason for this is quite simple; the magnitude <strong>of</strong> the back calculated stiffness<br />

modulus quite <strong>of</strong>ten reveals whether or not the pavement layer is damaged or not. If e.g. a<br />

stiffness modulus <strong>of</strong> 600 MPa is back calculated for a cement treated layer, this layer should<br />

be in a rather deteriorated state because the modulus <strong>of</strong> a sound cement treated layer is<br />

substantially higher.<br />

One <strong>of</strong> the drawbacks <strong>of</strong> back calculating layer moduli is the fact that accurate information on<br />

the thickness <strong>of</strong> the various layers should be available. We know that the deflections are<br />

heavily influenced by the product E.h 3 , which means that a small error in the layer thickness<br />

can have a large effect on the magnitude <strong>of</strong> the back calculated modulus.<br />

Although computer programs are available that back calculate the layer moduli automatically<br />

when the deflections, the load configuration and the thickness <strong>of</strong> the different layers is known,<br />

back calculation <strong>of</strong> layer moduli is certainly not as straightforward as it may look like because<br />

in many cases the solution is not unique. This implies that some pre-treatment <strong>of</strong> the data is<br />

necessary before the actual back calculation process is started.<br />

In the sections hereafter the surface modulus diagram will be discussed first <strong>of</strong> all. This<br />

diagram provides insight in how the overall stiffness <strong>of</strong> the pavement develops from bottom to<br />

top and whether or not weak interlayers are present. After that the actual back calculation<br />

process will be discussed.<br />

It should be noticed that the procedures described are especially valid for the analysis <strong>of</strong><br />

FWD measurements. They can however also be used for the analysis <strong>of</strong> BB and LD<br />

measurements provided that the appropriate corrections are applied. These correction<br />

procedures will be described in a later section.<br />

6.1 Surface modulus:<br />

According to Boussinesq’s theory, the elastic modulus <strong>of</strong> a homogeneous half space can be<br />

calculated from the deflection measured at a given distance following:<br />

E = σ . a 2 . (1 - µ 2 ) / d r . r<br />

E = 2 . σ . a . (1 - µ 2 ) / d 0<br />

Where: E = elastic modulus,<br />

a = radius <strong>of</strong> loading plate,<br />

µ = Poisson’s ratio,<br />

σ = contact pressure under loading plate.<br />

The question now is whether this formula can be <strong>of</strong> use in analysing the stiffness development<br />

in a pavement. Let us consider therefore figure 14.<br />

geophones<br />

a<br />

b<br />

Figure 14: Distribution <strong>of</strong> the vertical stress in a pavement.


27<br />

The way in which the load is distributed depends on the thickness and the stiffness <strong>of</strong> the<br />

layer. In figure 14, the top layer is the stiffest followed by the base and the subgrade. It is<br />

obvious that only that part <strong>of</strong> the pavement that is subjected to stresses, will deform; that is<br />

the area enclosed by the cone. This means that the geophone that is farthest away from the<br />

load centre (geophone a) only measures deformations in the subgrade while the geophone in<br />

the load centre (geophone b) measures the deformations in the subgrade, base and top layer.<br />

This implies that if the Boussinesq formula is applied using the deflection value measured by<br />

geophone a as input, the modulus <strong>of</strong> the subgrade is calculated. In case Boussinesq’s equation<br />

is used using the reading <strong>of</strong> geophone b as input, an overall effective stiffness <strong>of</strong> the<br />

pavement is calculated.<br />

So the stiffness calculated from the geophone readings going from a to b give information<br />

about: the subgrade, the subgrade plus some effect <strong>of</strong> the base, the subgrade plus the base<br />

plus some effect <strong>of</strong> the top layer, the subgrade plus the base plus the top layer; in short:<br />

increasing moduli value will be calculated.<br />

All this means that the deflection readings taken at a certain distance from the load centre<br />

give in fact information on the stiffness <strong>of</strong> the pavement at a certain depth.<br />

Using this information a so-called surface modulus plot is constructed. On the vertical axis<br />

one plots the surface modulus calculated using the Boussinesq formulas and on the<br />

horizontal axis one plots the equivalent depth which is equal to the distance <strong>of</strong> the geophone<br />

considered to the load centre. The principle <strong>of</strong> the plot is schematically shown below.<br />

Surface<br />

Modulus<br />

Equivalent Depth<br />

Figure 15 shows the surface modulus plots as calculated using the deflections measured at<br />

locations 0.65 and 1 (see table 2). The figure indicates that we are dealing with a weak<br />

pavement because the surface modulus values are very low and because the stiffness hardly<br />

increases from bottom to top. Only in location 1 some stiffening due to the base and top layer<br />

is visible.<br />

As shown below, different shapes <strong>of</strong> the surface modulus plot can be obtained.<br />

Surface Modulus<br />

Equivalent Depth


28<br />

The drawn line indicates a pavement where the stiffness gradually increases from bottom to<br />

top while the dashed line indicates a pavement which has layers with a low stiffness on top <strong>of</strong><br />

the subgrade. The reason for this might be stress dependent behaviour, lack <strong>of</strong> compaction,<br />

moisture effects etc.. It might very well be that the material with the lower stiffness is in fact<br />

the same material as the subgrade material. This is e.g. the case with fill material that cannot<br />

be compacted to the density <strong>of</strong> the existing subgrade.<br />

Figure 15: Surface modulus plots for locations 0.65 and 1.<br />

The surface modulus plot assists in deciding how many layers should be taken into account in<br />

the back calculation analysis. As indicated, the number <strong>of</strong> layers to be considered is not only<br />

the number <strong>of</strong> physical layers, top, base, sub-base and subgrade; one also has to take into<br />

account the fact that within one layer, sublayers may occur with a different stiffness.<br />

6.2 Back calculation <strong>of</strong> layer moduli:<br />

Back calculation <strong>of</strong> layer moduli from measured deflection bowls is done in an iterative way.<br />

The input for the calculations consists <strong>of</strong> the measured deflection pr<strong>of</strong>ile, the load geometry<br />

used to generate the deflections and the thickness <strong>of</strong> the layers. Furthermore the cores that<br />

are taken from the pavement to determine the thickness <strong>of</strong> the layers give information on the<br />

materials used and the quality <strong>of</strong> the materials.<br />

From the surface modulus plot an estimate is obtained for the modulus <strong>of</strong> the subgrade and<br />

furthermore the surface modulus plot provides information that helps to decide whether or not<br />

low stiffness sublayers should be introduced in the analysis.<br />

Then moduli values are assigned to the various layers and the deflections are calculated.<br />

Next the calculated deflections are compared with the measured ones. If the differences are<br />

too large, a new set <strong>of</strong> moduli is assumed and the deflections are calculated again. This<br />

process is repeated until there is a good match between the calculated and measured


29<br />

deflections. Normally the analysis is stopped when the difference between the measured and<br />

calculated deflections is 2%.<br />

As has been mentioned before, the iterative back calculation procedure can either be an<br />

“automatic” or a “hand operated” one. In the “automatic” procedures the computer program<br />

automatically performs the iterations while in the “hand operated procedure” it is the<br />

experienced engineer who controls the iteration process. Both procedures have their<br />

advantages. The automatic procedure is fast but might sometimes result in “funny” results. By<br />

“funny” it is meant that the set <strong>of</strong> moduli that is back calculated results in a good fit between<br />

the measured and calculated deflections but the moduli value themselves cannot be true<br />

given the type and condition <strong>of</strong> the materials in the pavement, given the temperature<br />

conditions etc.. Such results can occur because the back calculation procedure doesn’t<br />

necessarily result in a unique answer. In such cases the hand operated procedure is more<br />

powerful because the experienced engineer can adjust the moduli values to such levels which<br />

are reasonable for the type and condition <strong>of</strong> the pavement materials present and still result in<br />

a good fit between measured and calculated deflections.<br />

Problems with back calculating layer moduli can occur when the top layer is thin (< 60 mm) or<br />

when the base layer has a higher stiffness than the top layer.<br />

A golden rule in the back calculation analyses is that one never must vary the moduli values<br />

<strong>of</strong> all layers in the same time. This should be done in a step by step procedure. First <strong>of</strong> all one<br />

should try to find a modulus value for the subgrade by finding a good fit between the<br />

deflections measured and calculated at the largest distance to the load centre (see also figure<br />

14). Then one tries to fit the deflections at intermediate distance from the load centre; this will<br />

result in the moduli for the sub-base and base. Finally one should fit the deflections at the<br />

shortest distance to the load centre and the maximum deflection; this results in the modulus<br />

for the top layer.<br />

Furthermore one should realise that a good fit <strong>of</strong> the measured SCI is <strong>of</strong> great importance<br />

since this value gives a lot <strong>of</strong> information on the strain levels in the pavement. Sometimes the<br />

measured deflection pr<strong>of</strong>iles are not easy to match. In such cases one should notice that a<br />

good match <strong>of</strong> the SCI is to be preferred over a good match <strong>of</strong> the deflections measured at a<br />

greater distance from the load centre.<br />

6.3 Example:<br />

The example that will be given here is taken from the OECD FORCE test pavements that<br />

were built at the LCPC test facilities in Nantes, France. These pavements were tested by<br />

means <strong>of</strong> the accelerated load testing device <strong>of</strong> the LCPC. The aim <strong>of</strong> the project was to<br />

study pavement response and performance <strong>of</strong> three types <strong>of</strong> pavements under accelerated<br />

loading. The results <strong>of</strong> the FWD data evaluation <strong>of</strong> two test pavements are discussed here [3,<br />

4].<br />

Figure 16 shows the two pavement sections analysed.<br />

I<br />

II<br />

60 mm asphalt<br />

120 mm asphalt<br />

300 mm base<br />

300 mm base<br />

subgrade<br />

Figure 16: Structures I and II <strong>of</strong> OECD’s FORCE project.


30<br />

The clayey subgrade was covered with a 300 mm thick base on which 60 mm resp. 120 mm<br />

asphalt was placed.<br />

Figure 17 shows the maximum deflection level as measured on the top <strong>of</strong> the base as well as<br />

the maximum deflections that were measured after placing the asphalt layers. Figure 18<br />

shows the thickness <strong>of</strong> the top and base layer as determined by means <strong>of</strong> the Penetradar.<br />

Figure 17: Deflections measured on top <strong>of</strong> the base and top <strong>of</strong> the asphalt layer.<br />

Figure 18: Thickness <strong>of</strong> the layers <strong>of</strong> sections I and II.<br />

Figure 19 shows the surface modulus plots representative for both sections determined from<br />

the deflections measured on top <strong>of</strong> the completed sections.


31<br />

Figure 19: Surface modulus plots representative for the OECD FORCE sections.<br />

Three things appear from this figure. First <strong>of</strong> all the additional 60 mm asphalt which is present<br />

on section II contributes significantly to the stiffness <strong>of</strong> the pavement. Secondly, the modulus<br />

<strong>of</strong> the base and subgrade seems to be highly sensitive to the stress level. In both sections<br />

materials were used which are nominally the same. In section II however, the stresses in the<br />

base and subgrade are much smaller because <strong>of</strong> the thicker asphalt layer on top. The effect<br />

<strong>of</strong> the lower stress level in base and subgrade results in higher values for the surface<br />

modulus.<br />

Furthermore one should realise that the plot was made based on measurements which were<br />

taken at a temperature <strong>of</strong> approximately 6 0 C which means that the stiffness <strong>of</strong> the asphalt<br />

layer was fairly high and the stress levels in the base and subgrade are rather low .<br />

Thirdly the figure shows that on top <strong>of</strong> the subgrade, layers are present with a much lower<br />

stiffness. It appeared that a fill had to be placed in order to have the pavement surface at the<br />

right level. The fill was made with the subgrade material but problems during compaction had<br />

occurred. This lack <strong>of</strong> density <strong>of</strong> the fill has <strong>of</strong> course a direct effect on the density and so the<br />

stiffness <strong>of</strong> the base layer placed on top. The low surface modulus values could, in this case,<br />

easily be explained from the construction history.<br />

Based on this knowledge it was decided to divide the base layer in two sublayers, each being<br />

50% <strong>of</strong> the total base thickness, and to divide the subgrade in two sublayers. This was done<br />

by assuming a thickness <strong>of</strong> 500 mm <strong>of</strong> low stiffness subgrade material on top <strong>of</strong> the stiff deep<br />

subgrade. The selection <strong>of</strong> this thickness is based on experience, sometimes a thickness <strong>of</strong><br />

1000 mm is chosen.<br />

All in all it means that for the back calculation analysis, the pavement was divided in 5 layers<br />

(top layer, two base layers, two subgrade layers).<br />

The results <strong>of</strong> the analysis are shown in table 5.


32<br />

Section I<br />

Temp Force Layer E-mod Position Meas. Calc. Diff.<br />

Thickn. Defl. Defl.<br />

[ 0 C] [kN] [mm] [Mpa] [mm] [µm] [µm] [%]<br />

6.4 57.0 56 15980 0 1049 1050 0.1<br />

146 106 300 655 655 0<br />

146 150 600 318 318 0<br />

500 37 900 158 163 3.2<br />

171 1200 92 92 0<br />

1500 63 60 -4.8<br />

1800 46 46 0<br />

Section II<br />

Temp Force Layer E-mod Position Meas. Calc. Diff.<br />

Thickn. Defl. Defl.<br />

[ 0 C] [kN] [mm] [Mpa] [mm] [µm] [µm] [%]<br />

6.8 58.0 145 10514 0 415 417 0.5<br />

130 117 300 329 326 -0.9<br />

130 239 600 217 216 -0.5<br />

500 48 900 133 135 1.5<br />

276 1200 83 83 0<br />

1500 50 51 2.0<br />

1800 34 33 -2.9<br />

Table 5: Results <strong>of</strong> the back calculation analysis for the OECD FORCE sections.<br />

It should be noted that the FORCE examples are complicated ones; normally one has to deal<br />

with less complicated deflection pr<strong>of</strong>iles.<br />

6.4 Computer program:<br />

As had been mentioned before, several computer programs are available that allow the<br />

values for the layer moduli to be backcalculated in an automatic way. One <strong>of</strong> those programs<br />

is the program MODCOMP 5 developed by pr<strong>of</strong>. Irwin <strong>of</strong> the Cornell university in the USA.<br />

The program can be found on the cd which is part <strong>of</strong> these lecture notes. At the end <strong>of</strong> these<br />

lecture notes an appendix, appendix I, is given which contains a description <strong>of</strong> how the<br />

program has to be used.<br />

7. Analysis <strong>of</strong> Benkelman beam and Lacroix deflectograph<br />

deflection bowls:


33<br />

BB and LD measurements are usually related to empirical evaluation and overlay design<br />

methods. However an elegant procedure has been developed [5] which allows these<br />

deflection readings also to be used for back calculation purposes. The procedure is correcting<br />

the measured deflections that might be influenced by the movement <strong>of</strong> the support system to<br />

true deflections. One drawback <strong>of</strong> the method is that it doesn’t take into account viscous<br />

effects that might occur due to the slow speed <strong>of</strong> the loading vehicle.<br />

The basis <strong>of</strong> the method is the Hogg model which consists <strong>of</strong> a plate (E 1 , h, µ 1 ) resting on an<br />

elastic foundation (E 2 , µ 2 ). The assumption that the top layer behaves like a plate implies that<br />

no vertical displacements are developed in this layer. The characteristics <strong>of</strong> the pavement<br />

structure are characterised by:<br />

D = E 1 . h 3 1 / {12 . (1 - µ 2 1 )} stiffness <strong>of</strong> the top layer(s)<br />

R = 2 . E 2 . (1 - µ 2 ) / {(1 + µ 2 ) . (3 - 4µ 2 )} reaction <strong>of</strong> the subgrade<br />

L 0 = ( D / R ) 0.33 critical length<br />

The shape <strong>of</strong> the deflection pr<strong>of</strong>iles is described following<br />

d 0 / d r – 1 = γ + α . (r / L 0 ) β<br />

Where: d 0<br />

d r<br />

= maximum deflection,<br />

= deflection at distance r from the load centre.<br />

This equation is graphically represented in figure 20. Using the specific dimensions <strong>of</strong> both<br />

the deflectograph and the BB (figure 21) as well as the above mentioned pavement<br />

characteristics, true deflection pr<strong>of</strong>iles as well deflection pr<strong>of</strong>iles that would be measured were<br />

calculated; typical results are shown in figure 22.<br />

Figure 20: Graphical representation <strong>of</strong> an equation used to describe<br />

the shape <strong>of</strong> deflection pr<strong>of</strong>iles.<br />

In the development <strong>of</strong> the model the following values were assumed for the wheel and axle<br />

loads as well as contact pressures.<br />

Axle P fa /P ra P 1 I 1 W 1 σ 1 P 2 I 2 W 2 σ 2 -σ 1 P axle


34<br />

[N] [mm] [mm] [MPa] [N] [mm] [mm] [MPa] [N]<br />

rear 0.6 1750 250 177 0.05 21750 192 136 1.062 94000<br />

front 0.6 336o 320 220 0.061 25040 270 186 0.635 56800<br />

Figure 21: Dimensions <strong>of</strong> the LD and BB as well as <strong>of</strong> the loading vehicle.


35<br />

Figure 22: Recorded and true LD (lac) and BB (ben) deflections.


36<br />

Based on these calculations, evaluation diagrams were developed which allow true<br />

deflections to be calculated from the measured LD and BB deflections. These diagrams are<br />

shown in figure 23. In this figure some abbreviations are used which are not explained in the<br />

figure; the meaning there<strong>of</strong> is described hereafter.<br />

D CGRA = maximum deflection according to the Canadian Good Roads Association method,<br />

D AI = maximum deflection according to the Asphalt Institute method,<br />

D TRRL = maximum deflection according to the Transport and Road Research Laboratory<br />

method.<br />

The method will be illustrated with some examples. Let us assume that a maximum deflection<br />

was measured with the LD <strong>of</strong> 393 µm. From the measured deflection pr<strong>of</strong>ile it was determined<br />

that the distance at which the deflection was 50% <strong>of</strong> the maximum deflection (L x , x = 50%)<br />

was 368 mm. From the evaluation charts one can derive that L 0 = 178 mm and the ratio<br />

D 00 /D lac = 1.226. This means that the true maximum deflection is 482 µm.<br />

The ratio D 00 . R / P ra equals 0.47 and with a rear axle load P ra = 91.6 kN this results in an R<br />

value <strong>of</strong> 89.5 Mpa and a subgrade modulus <strong>of</strong> 149 Mpa (assuming µ 2 = 0.35). Since L 0 and R<br />

are known, D can be calculated.<br />

Furthermore we can determine the maximum BB deflection that would be obtained following<br />

the TRRL procedure. One observes that D TRRL /D lac = 0.98 which means that the value that<br />

has to be used in the TRRL evaluation procedure equals 385 µm.<br />

It is stressed that figure 23 is only applicable for the load and LD and BB geometries shown in<br />

figure 21.<br />

One should keep in mind that the moduli obtained in this way are quasi-static moduli. It is a<br />

well-known fact however that for most materials there is a difference between the static and<br />

the dynamic modulus. From an extensive correlation study it was observed that the subgrade<br />

modulus as determined by means <strong>of</strong> the BB or LD and the FWD relate to each other<br />

following:<br />

1.4576 (t – 0.255)<br />

E FWD / E lac = 10<br />

Where: t<br />

= loading time <strong>of</strong> the LD or BB [s].


37<br />

Figure 23: Evaluation chart to determine true LD and BB deflections from measured<br />

deflections.


38<br />

8. Estimation <strong>of</strong> the remaining pavement life using an empirical<br />

based approach:<br />

A number <strong>of</strong> empirical pavement evaluation and overlay design methods have been<br />

developed in time. Well known are the methods developed by the Asphalt Institute and the<br />

Transport and Road Research Laboratory. Although extensively used all over the world, this<br />

author believes strongly that one has to be very cautious in using these methods for situations<br />

they have not been developed for. The hart <strong>of</strong> the TRRL method e.g. are the performance<br />

charts developed for several pavement types. An example <strong>of</strong> such a chart is given in figure 24<br />

[6]. For the sake <strong>of</strong> completeness the load and load configuration used for the BB<br />

measurements according to the TRRL procedure are shown in figure 25.<br />

The point is that pavement performance is dependent on the traffic, the materials and<br />

structures used, and the climate, all <strong>of</strong> them are typical British in case <strong>of</strong> the TRRL method.<br />

This means that the chances are very small that the method can be used without modifications<br />

in countries like e.g. Pakistan or Yemen where traffic, climate, and materials are<br />

significantly different from UK conditions.<br />

Another severe problem with the TRRL method is that an important input parameter, being<br />

the number <strong>of</strong> equivalent 80 kN single axles that have passed the pavement, is not known in<br />

many cases.<br />

Nevertheless the author also believes that the TRRL method can be used in other conditions<br />

as well provided this is done by making the evaluation charts dimensionless. The procedure<br />

to do so is outlined hereafter.<br />

Let us define the following variables:<br />

DeltaDef n<br />

DeltaDef c<br />

n<br />

N c<br />

= increase in deflection since time <strong>of</strong> construction,<br />

= difference between the initial deflection and the critical deflection, this latter<br />

value depends on the probability <strong>of</strong> achieving life that is used to define pavement<br />

failure,<br />

= applied number <strong>of</strong> load repetitions,<br />

= number <strong>of</strong> load repetitions at which critical deflection level is reached.<br />

Work presented in [7] has shown that performance curves like the one presented in figure 24<br />

can be written in a dimensionless shape following:<br />

DeltaDef n / DeltaDef c<br />

= (n / N c ) b<br />

The shape parameter b seemed to be dependent on the initial deflection level following:<br />

for granular bases: b<br />

0.4639<br />

= 0.06 Def 0<br />

for bituminous bases: b<br />

0.7186<br />

= 0.0185 Def 0<br />

An important question in all this is how DeltaDef c and the initial deflection Def 0 are related.<br />

From the analysis in [7] it appeared that for pavements with granular bases and accepting<br />

50% <strong>of</strong> achieving life as the failure condition, the ratio DeltaDef 0 / Def 0 can be expressed as<br />

follows:<br />

DeltaDef c / Def 0 = 0.4767 – 0.000299 Def 0<br />

Where: Def 0<br />

= maximum deflection measured with the BB according to the TRRL<br />

procedure [µm] <strong>of</strong> the pavement when not subjected tot traffic loads.<br />

For bituminous bases this relation can be written as:<br />

DeltaDef c / Def 0 = 0.34833 – 0.000198 Def 0<br />

If we don’t know the number <strong>of</strong> load repetitions applied to the pavement, how do we derive<br />

DeltaDefn? It will be shown hereafter that we can obtain that value in a relatively simple way.


39<br />

Figure 24: Example <strong>of</strong> a TRRL performance chart.


40<br />

Figure 25: Load configuration used for the BB measurements according to TRRL.


41<br />

Normally BB measurements are only taken in the wheel tracks. These values are in fact the<br />

Def n values since that pavement area has been subjected to n load repetitions. If we also take<br />

deflection measurements between the wheel tracks, then we get a good estimate <strong>of</strong> the<br />

flexural stiffness <strong>of</strong> that part <strong>of</strong> the pavement that is not subjected to traffic loads. These<br />

deflections can be taken as representative for Def 0 .<br />

Assume that the deflection measured between the wheel tracks is 350 µm and that the<br />

deflection in the wheel tracks is 390 µm. The pavement has an unbound base. Then we arrive<br />

to:<br />

DeltaDef n = 390 – 350 = 40<br />

and<br />

DeltaDef c / Def 0 = 0.4767 – 0.000299 x 350 = 0.372<br />

so<br />

DeltaDef c = 0.372 x 350 = 130<br />

We also calculate:<br />

b = 0.91<br />

so<br />

DeltaDef n / DeltaDef c = (n / N) b<br />

40 / 130 = (n / N) 0.91<br />

n / N = 0.27<br />

Normally road authorities are not interested in a damage ratio or a remaining pavement life<br />

expressed in a number <strong>of</strong> allowable load repetitions but much more in a remaining life in<br />

years. This can be estimated in the following way.<br />

Assume the traffic composition has not changed in time and for reasons <strong>of</strong> simplicity we also<br />

assume that no growth in the number. <strong>of</strong> vehicles per day has taken place. This means that<br />

the area indicated in figure 26 is representative for the cumulative amount <strong>of</strong> traffic n that has<br />

passed the road during time period t.<br />

Traffic intensity<br />

n<br />

N<br />

t<br />

T<br />

Time<br />

Figure 26: Procedure to estimate the remaining life in years from the n/N ratio.<br />

In the same way the allowable number <strong>of</strong> load repetitions N is arrived after T years. From this<br />

simple example it is clear that in this case:<br />

t / T = n / N


42<br />

If we assume e.g. that the deflection survey <strong>of</strong> the above mentioned example was taken 5<br />

years after the pavement has been put in service, we calculate that:<br />

t / T = n / N = 0.27, t = 5 so T = 18 years and the remaining life is 13 years.<br />

The procedure described above cannot be used if variations occur in the cross section <strong>of</strong> the<br />

pavement due to variations in the thickness <strong>of</strong> the layers and because different types <strong>of</strong><br />

material are used over the width <strong>of</strong> the pavement. Those conditions can occur if e.g. ruts are<br />

filled, the pavement is widened or <strong>of</strong> mill and fill operations have been carried out.<br />

Of course unknown changes in the traffic growth, composition <strong>of</strong> the traffic and the axle loads<br />

have also a negative effect on the results obtained by the procedure described above.


43<br />

9. Mechanistic procedures for remaining life estimations and<br />

overlay design:<br />

Mechanistic overlay design methods are based on the analysis <strong>of</strong> stresses and strains in the<br />

existing pavement. The calculated values are then compared with the allowable values and<br />

based on this comparison, conclusions are drawn with respect to the most appropriate<br />

maintenance strategy.<br />

One <strong>of</strong> the most important differences between a mechanistic and an empirical approach is<br />

the fact that in the latter, the interactions between stresses, strains, strength, fatigue,<br />

resistance to deformation etc are not visible; they are hidden in the procedure. This makes<br />

the empirical methods unreliable as soon as different materials and structures are used than<br />

those for which the procedure was developed. On the other hand empirical methods are<br />

based on observed performance which is an advantage over mechanistic models especially if<br />

these models are used in a too simplistic way.<br />

The big advantage <strong>of</strong> the mechanistic models <strong>of</strong> course is that they are based on sound<br />

analyses <strong>of</strong> stresses, and strength <strong>of</strong> the materials used.<br />

9.1 Basic principles:<br />

In classical mechanistic overlay design methods, only the strain levels in the existing<br />

pavement are considered as well as the required reduction in those strain levels in order to<br />

obtain the required extension <strong>of</strong> the pavement life. The overlay is designed in such a way that<br />

the necessary reduction <strong>of</strong> the strain level in the existing pavement is realised. The effect <strong>of</strong><br />

damage in the existing pavement on the performance <strong>of</strong> the overlay is normally not<br />

considered. This makes the classical mechanistic methods rather straightforward.<br />

The following steps can be recognised. First <strong>of</strong> all the moduli <strong>of</strong> the various layers are<br />

calculated in the way described earlier. Secondly the asphalt layer modulus is corrected to a<br />

reference temperature; for Dutch conditions this is 18 0 C. This correction can be applied using<br />

the asphalt mix stiffness vs temperature chart as developed by Shell [8]; this chart is given in<br />

figure 27. Then the stresses and strains due to an equivalent single axle load are calculated.<br />

The tensile strain calculated at the bottom <strong>of</strong> the asphalt layer is introduced in a fatigue<br />

relation and the allowable number <strong>of</strong> load repetitions is calculated. The same is done for the<br />

subgrade strain. The amount <strong>of</strong> damage, being the ratio n/N, is then calculated where n is the<br />

applied number <strong>of</strong> load repetitions and N is the allowable number. The remaining life ratio is<br />

calculated as 1 – n/N.<br />

If the pavement life should be extended, the number <strong>of</strong> load repetitions that are expected<br />

needs to be calculated. This results in a figure n + ∆n. Then the pavement thickness should<br />

be increased in order to decrease the tensile strain at the bottom <strong>of</strong> the asphalt layer and to<br />

increase the allowable number <strong>of</strong> load repetitions from N to N + ∆N. The appropriate overlay<br />

thickness is obtained if:<br />

1 – n/N = ∆n / (N + ∆N)<br />

The procedure is illustrated with an example.<br />

Assume that the tensile strain that is calculated at the bottom <strong>of</strong> the asphalt layer due to a<br />

standard axle load equals:<br />

ε = 2 . 10 -4 [m / m]<br />

Fatigue tests carried out on the material resulted in the following fatigue relation.<br />

Log N = -13 – 5 . log ε<br />

The allowable number <strong>of</strong> load repetitions is then N = 312500.<br />

If we assume that the pavement has already carried 200000 standard loads, then the damage<br />

ratio equals n / N = 0.64.


44<br />

Figure 27: Relationship between the stiffness <strong>of</strong> asphalt mixtures and temperature for a<br />

loading time <strong>of</strong> 0.02 s.


45<br />

The remaining life ratio equals:<br />

1 – n/N = 0.36<br />

Assume that another 500000 standard axles should be carried by the pavement. This means<br />

that:<br />

∆n = 500000<br />

The tensile strain at the bottom <strong>of</strong> the asphalt layer should be decreased to a level where N +<br />

∆N load repetitions can be taken. This value is calculated from:<br />

N + ∆N = ∆n / (1 – n/N) = 500000 / 0.36 = 1.39 . 10 6<br />

By using the fatigue relation we calculate that this new number <strong>of</strong> allowable load repetitions<br />

can be obtained if the strain is reduced to ε = 1.48 . 10 -4 [m / m]. This means that the overlay<br />

needs such a thickness that the strain at the bottom <strong>of</strong> the existing asphalt layer is reduced to<br />

this value.<br />

The approach described here gives rise to some comments. It is quite clear that a very large<br />

overlay thickness is needed when the ratio n/N approaches 1. The reason is that the fatigue<br />

relation is based on beam fatigue tests. This implies that failure means that the specimen is in<br />

two parts if the allowable number <strong>of</strong> load repetitions is reached (at least in load controlled<br />

fatigue tests) which implies that the beam lost its functionality. In reality however the cracked<br />

asphalt slab is still supported by the base and other layers; the cracked slab is still functional.<br />

All this indicates that the procedure results in unrealistic designs in case <strong>of</strong> high values <strong>of</strong> the<br />

damage ratio.<br />

Furthermore the example indicates that in general fairly small strain reductions are needed<br />

which results in rather thin overlays.<br />

Because the overlay design is only based on the reduction <strong>of</strong> the strain level in the existing<br />

pavement, only the thickness and the stiffness <strong>of</strong> the overlay are <strong>of</strong> importance. From practice<br />

one knows that this cannot be true. The existing pavement normally exhibits a certain amount<br />

<strong>of</strong> cracking when an overlay is applied and these cracks tend to propagate through the<br />

overlay. This means that reduction <strong>of</strong> the strain level in the existing pavement cannot be the<br />

only design criterion for overlays; also the resistance to crack reflection <strong>of</strong> the overlay should<br />

be considered. This aspect will be discussed later in these lecture notes.<br />

Finally the procedure described above doesn’t take into account the large amount <strong>of</strong> variation<br />

in deflections and material characteristics that can occur in pavements.<br />

9.2 Extension <strong>of</strong> the basic principles:<br />

In this section an extension <strong>of</strong> the basic principles presented in the previous section will be<br />

given. The extension is dealing with the fact that in case the n/N ratio reaches 1, realistic<br />

values for the overlay thickness should still be obtained. Furthermore the extension takes into<br />

account the variation in deflection level and material characteristics that occur in practice.<br />

If there was no variation in deflection level along the section under consideration, and if there<br />

was no variation in the thickness <strong>of</strong> the pavement layers, then there would be no variation in<br />

the elastic modulus <strong>of</strong> the layers and there would be no variation in strain level. If there also<br />

would be no variation in the fatigue characteristics, then the pavement would fail precisely at<br />

the number <strong>of</strong> load repetitions predicted and the pavement would fail from one moment to the<br />

other. This particular behaviour is illustrated in figure 28a. Such a performance however is<br />

never observed, pavements don’t collapse in the way indicated by this figure. In reality a more<br />

gradual deterioration is observed as is indicated by figure 28b.<br />

If we use the mean strain level <strong>of</strong> figure 28b as design criterion and we use this strain value<br />

together with the mean fatigue characteristic (the solid fatigue line in figure 28b) then we<br />

determine the mean number <strong>of</strong> load repetitions. At that number <strong>of</strong> load repetitions there is a<br />

50% chance that the pavement is failed. It can easily be shown that this means that 50% <strong>of</strong><br />

the trafficked pavement surface shows cracking. Because <strong>of</strong> the variation in the fatigue


46<br />

resistance, some parts <strong>of</strong> the pavement will live longer and some shorter. Furthermore the<br />

strain level in some parts <strong>of</strong> the pavement are lower than at other parts because <strong>of</strong> e.g. the<br />

variation in thickness. The variation in strain level combined with the variation in fatigue<br />

resistance results in a variation <strong>of</strong> pavement life over the section considered. This is shown in<br />

figure 28b. Figure 28b also clearly shows that pavements don’t fail in a catastrophic way but<br />

show a gradual deterioration. The overlay design procedure should take this into account.<br />

Log n<br />

Thickness <strong>of</strong> the pavement layers<br />

and the layer moduli are constant,<br />

so strain is constant.<br />

N<br />

Fatigue<br />

characteristics<br />

show no<br />

variation<br />

Condition<br />

Logε<br />

N<br />

Log n<br />

Figure 28a: Condition deterioration when there is no variation in pavement properties.<br />

Log n<br />

Fatigue characteristics<br />

show variation<br />

Thickness and modulus <strong>of</strong><br />

the layers show variation so<br />

strain is variable.<br />

N<br />

Condition<br />

Log ε<br />

50% failed and<br />

50% sound Mean strain<br />

level<br />

N<br />

Log n<br />

Figure 28b: Condition deterioration when there is variation in pavement properties.


47<br />

In order to take the variation <strong>of</strong> input parameters into account, probabilistic analyses should<br />

be made. Several procedures are available to determine which combinations <strong>of</strong> layer<br />

thickness, layer modulus and fatigue relation should be used in the calculations in order to enable<br />

to estimate the variation in strain level and pavement life. A far more effective approach<br />

is to make use <strong>of</strong> simple relations that exist between e.g. the surface curvature <strong>of</strong> the deflection<br />

pr<strong>of</strong>ile on one hand and the tensile strain at the bottom <strong>of</strong> the asphalt layer, the tensile<br />

strain at the bottom <strong>of</strong> the bound base or the vertical compressive strain at the top <strong>of</strong> the<br />

subgrade, on the other hand. This will be shown in the following part.<br />

Let us consider the bending <strong>of</strong> a slab as shown in figure 29.<br />

Figure 29: Bending moments in a slab.<br />

The magnitude <strong>of</strong> the bending moments can be calculated a follows:<br />

M 1x = E h 3 ( 1/R x + µ 1/R y ) / 12 ( 1 - µ 2 ) and M 1y = E h 3 ( 1/R y + µ 1/R x ) / 12 ( 1 - µ 2 )<br />

Where: M 1x = bending moment in the x direction,<br />

M 1y = bending moment in the y direction,<br />

R x = radius <strong>of</strong> curvature in the x direction,<br />

R y = radius <strong>of</strong> curvature in the y direction,<br />

E = elastic modulus <strong>of</strong> the slab,<br />

h = thickness <strong>of</strong> the slab,<br />

µ = Poisson’s ratio.<br />

The stresses can be calculated as σ x = 6 M 1x / h 2 and σ y = 6 M 1y / h 2 . If we are dealing with a<br />

circular load in the centre <strong>of</strong> a large slab, R x = R y and σ x = σ y .<br />

Because: ε x = ( σ x - µ σ y ) / E = ( 1 - µ ) σ x / E we can now develop a relation between the<br />

curvature and the tensile strain by substitution <strong>of</strong> σ x by M 1x and by substitution <strong>of</strong> M 1x by the<br />

equation that relates the bending moment to the radius <strong>of</strong> curvature. We obtain then:<br />

ε x = 6 ( 1 - µ ) M / E h 2 = h / 2 R x ≅ 1 / R x<br />

This indicates that the strain at the bottom <strong>of</strong> the asphalt layer is related to the radius <strong>of</strong><br />

curvature <strong>of</strong> the deflection bowl due to the applied load.<br />

Extensive research [9,10], has shown that there exists a direct relation between the tensile<br />

strain at the bottom <strong>of</strong> the asphalt layer and the surface curvature index following:<br />

log ε<br />

= C 0 + C 1 log SCI<br />

For pavements with an asphalt thickness ≥ 150 mm the relation becomes:


48<br />

Log ε = 0.481 + 0.881 log SCI 300<br />

Where: SCI 300 = difference in maximum deflection and the deflection measured at a distance<br />

<strong>of</strong> 300 mm,<br />

ε = tensile strain at the bottom <strong>of</strong> the asphalt layer [µm / m].<br />

This relation is shown in figure 30.<br />

Figure 30: Relation between SCI 300 and the tensile strain at the bottom <strong>of</strong> the asphalt layer.<br />

Since log N = A 0 + A 1 log ε, we can write:<br />

log N = A 0 + A 1 C 0 + A 1 C 1 log SCI<br />

It can be shown that the variance <strong>of</strong> log N (the squared standard deviation <strong>of</strong> log N) can be<br />

calculated from:<br />

S 2 logN<br />

= A 1 2 . C 1 2 . S 2 logSCI + S 2 l<strong>of</strong><br />

Where: S logSCI = standard deviation <strong>of</strong> the logarithm <strong>of</strong> the measured SCI’s (see also table 2)<br />

S l<strong>of</strong> = standard deviation <strong>of</strong> log N at a given log ε; it describes the variation in<br />

fatigue life.<br />

We can now write:<br />

log N P = log N – u . S logN<br />

Where: log N = logarithm <strong>of</strong> the mean number <strong>of</strong> load repetitions to failure,<br />

log N P = logarithm <strong>of</strong> the number <strong>of</strong> load repetitions to failure at level <strong>of</strong> confidence P<br />

u = factor from the tables for the normal distribution related to confidence<br />

level P


49<br />

From the equations given above it becomes clear that the quality <strong>of</strong> the predictions increases<br />

when S logN decreases. This means that S logSCI and S l<strong>of</strong> should be as low as possible. A low<br />

S logSCI stresses the need to pay ample attention to the discrimination <strong>of</strong> homogeneous subsections.<br />

The only factor that cannot be easily assessed is the variation in fatigue characteristics.<br />

Although this value can be estimated (see e.g. lecture notes CT4850 part III<br />

Asphaltic Materials) if mixture composition data are available, extensive fatigue testing has<br />

shown that S l<strong>of</strong> = 0.25 is a reasonable first estimate.<br />

Overlay calculations based on the confidence level or probability <strong>of</strong> survival level P are made<br />

in the following way. As is shown above, the number <strong>of</strong> load repetitions until a certain<br />

probability <strong>of</strong> survival level P 1 is reached can be calculated using:<br />

log N P1 = A 0 + A 1 C 0 + A 1 C 1 log SCI 1 – u 1 S logN<br />

If the pavement life has to be extended to N + ∆N load repetitions and after that number <strong>of</strong><br />

load repetitions, the probability <strong>of</strong> survival should be P 2 , the needed SCI level to achieve this<br />

can be calculated using:<br />

Log (N + ∆N) P2 = A 0 + A 1 C 0 + A 1 C 1 log SCI 2 – u 2 S log(N+∆N)<br />

After subtracting <strong>of</strong> both equations one obtains:<br />

Log {N P1 / (N + ∆N) P2 } = A 1 C 1 log {SCI 1 / SCI 2 } – u 1 S logN + u 2 S log(N+∆N)<br />

By writing<br />

N P1 / (N + ∆N) P2 = 1 / X<br />

I 1 = 10**(u 1 S logN )<br />

I 2<br />

= 10**(u 2 S log(N+∆N)<br />

We arrive to<br />

Log {1 / X} = A 1 C 1 log {SCI 1 / SCI 2 } – log I 1 + log I 2<br />

This can be written as:<br />

SCI 2<br />

= SCI 1 (X I 2 / I 1 ) 1/A1C1<br />

In these equations SCI 1 can be considered as the SCI before the overlay is placed and SCI 2<br />

as the SCI after overlaying. In the same way S logN is valid before overlaying and S log(N+∆N) is<br />

valid after the overlay is placed.<br />

We still need equations to predict the SCI 2 in relation to the overlay thickness and stiffness as<br />

well as the SCI 1 . Furthermore an equation is needed to predict S logSCI2 because from this<br />

value S log(N+∆N) can be calculated. These equations are given below:<br />

Log SCI 2 = b 0 + b 1 E o + b 2 h o + b 3 log SCI 1 + b 4 E o log SCI 1 + b 5 h o log SCI 1<br />

+ b 6 h o log E o log SCI 1<br />

S 2 logSCI2<br />

= {b 1 + b 4 log SCI 1 + b 6 h o log SCI 1 / E o } 2 S 2 Eo<br />

+ {b 2 + b 5 log SCI 1 + b 6 log E o log SCI 1 } 2 S 2 ho<br />

+ {b 3 + b 4 E o + b 5 h o + b 6 h o log E o } 2 S 2 logSCI1<br />

Where: SCI 1<br />

SCI 2<br />

h o<br />

E o<br />

= surface curvature index (d 0 – d 300 ) before overlaying [µm]<br />

= surface curvature index (d 0 – d 300 ) after overlaying [µm]<br />

= overlay thickness [mm]<br />

= elastic modulus <strong>of</strong> the overlay [Mpa]<br />

b o = -0.0506<br />

b 1 = 1.178 10 -5<br />

b 2 = 0.0094


50<br />

b 3 = 1.0153<br />

b 4 = -7.73 x 10 -6<br />

b 5 = -3.778 x 10 -4<br />

b 6 = -1.4971 x 10 -3<br />

With respect to the procedures discussed above, it is once again stressed that they are based<br />

on limiting the strains in the existing pavement. Also it should be noted that it is assumed that<br />

the overlay is fully bonded to the existing pavement. This however is not always the case<br />

especially in cases where, because <strong>of</strong> reasons to be discussed later, an interface layer is<br />

placed between the overlay and the existing pavement allowing the overlay to behave more<br />

or less independently from the existing pavement. Furthermore the effect <strong>of</strong> cracks in the<br />

existing pavement on the performance <strong>of</strong> the overlay is not taken into account. This effect<br />

however cannot be ignored in cases where the existing pavement shows moderate to severe<br />

cracking. Also this will be discussed in a later chapter.<br />

One important point remains to be discussed which is the estimation <strong>of</strong> the probability <strong>of</strong><br />

survival <strong>of</strong> the existing pavement P.<br />

Without going into all the details (for these the reader is referred to [9]), it can be shown that P<br />

can be estimated from the ratio <strong>of</strong> the surface curvature index measured in and between the<br />

wheel tracks following:<br />

P<br />

= (SCI b / SCI in ) q<br />

Where: SCI b = SCI measured between the wheel tracks (d 0 – d 500 )<br />

SCI in = SCI measured in the wheel tracks (d o – d 500 )<br />

q = dependent on the type <strong>of</strong> structure taking a value between 0.6 and 0.4 for<br />

pavements with an unbound base and between 0.7 and 0.5 for pavements<br />

with a bound base; the higher values are for a 150 mm thick base, the<br />

lower values are for a 300 mm thick base.<br />

If for reasons mentioned earlier, the SCI ratio cannot be used, P can also be estimated from<br />

the percentage <strong>of</strong> the wheel track area that shows cracking following:<br />

P = 1 – percentage cracked area / 100<br />

It should be noted that a substantial part <strong>of</strong> the cracking that is visible at the pavement is<br />

surface cracking. This type <strong>of</strong> cracking is initiated at the pavement surface and normally<br />

progresses downwards to approximately 40 mm. It is clear that this type <strong>of</strong> cracking cannot be<br />

associated to the fatigue type cracking for which the above mentioned procedures are developed.<br />

All in all this means that P values estimated in this way might be too high, the real<br />

structural condition might be better than it appears from the P value estimated in this way.<br />

If P is known as well as S logN , the damage ratio n / N can easily be determined using the<br />

equations given above or by means <strong>of</strong> figure 31.


51<br />

Figure 31: Relation between P, S logN and n / N.


52<br />

10. Extension <strong>of</strong> the simplified procedure to estimate critical<br />

stresses and strains:<br />

In many cases the thickness <strong>of</strong> the pavement layers is unknown or highly variable. In that<br />

case a pavement evaluation that relies on the back calculation <strong>of</strong> layer moduli is less effective<br />

and estimation <strong>of</strong> critical stresses and strains using simple methods as described in the<br />

previous chapter are extremely useful. In a joint research effort by the Government Service<br />

for Land and Water Use (LWU) <strong>of</strong> the Dutch Ministry <strong>of</strong> Agriculture, Nature Management and<br />

Fisheries, KOAC consultants and the <strong>Delft</strong> University <strong>of</strong> Technology, a pavement evaluation<br />

and overlay design method was developed which completely relies on such simple relations<br />

[11]. The hart <strong>of</strong> the method being the relations to estimate the stresses and strains will be<br />

reproduced here.<br />

The basis <strong>of</strong> the method is the large number <strong>of</strong> calculations on stresses and strains in on four<br />

layer pavement systems due to a FWD load. The calculated values are schematically shown<br />

in figure 32.<br />

FWD load 50<br />

kN, φ = 300 mm<br />

Asphalt<br />

Unbound or<br />

Bound<br />

Base<br />

Subbase<br />

Subgrade<br />

1. Tensile strain at<br />

pavement surface.<br />

2. Tensile strain at<br />

bottom asphalt.<br />

3. Compressive<br />

stresses in top<br />

unbound base.<br />

4.Tensile strain at<br />

bottom bound base<br />

5. Vertical compressive<br />

strain at<br />

top subbase.<br />

6. Vertical compressive<br />

strain at<br />

top subgrade.<br />

Figure 32: Analysed structures and locations where stresses and strains were calculated.<br />

The analyses have been made for pavements with E asphalt > E base > E subbase > E subgrade and for<br />

pavements where E subbase < E subgrade .<br />

One will notice that the equations are much more complex than the ones described until now.<br />

The reason for this is that thin asphalt surfacings had to be considered and for those<br />

pavements the simple relations between e.g. the SCI and the tensile strain at the bottom <strong>of</strong><br />

the asphalt layer are not valid anymore.<br />

Also one will notice that in a number <strong>of</strong> cases information on the thickness <strong>of</strong> some layers is<br />

required. From the type <strong>of</strong> equation one will notice however that the influence <strong>of</strong> the thickness<br />

information on the magnitude <strong>of</strong> the estimated strains and stresses is limited.<br />

10.1 Relations between deflection bowl parameters and stresses and strains at various<br />

locations in the pavement:<br />

From the extensive analyses, the following results were obtained:<br />

Tensile strain at the bottom <strong>of</strong> the asphalt layer:<br />

log ε r1,0 = -1.06755 + 0.56178 log h 1 + 0.03233 log d 1800 + 0.47462 log SCI 300<br />

+ 1.15612 log BDI – 0.68266 log BCI


53<br />

Where: ε r1,0 = maximum horizontal strain at the bottom <strong>of</strong> the asphalt layer [µm/m]<br />

h 1 = thickness <strong>of</strong> the asphalt layer [mm]<br />

d r = deflection at distance r <strong>of</strong> the load centre [µm]<br />

SCI 300 = d 0 – d 300 [µm]<br />

BDI = base damage index = d 300 – d 600 [µm]<br />

BCI = base curvature index = d 600 – d 900 [µm]<br />

Tensile strain at pavement surface:<br />

Many cracks that are visible at the pavement surface are initiated at the top <strong>of</strong> the pavement.<br />

These cracks are the result <strong>of</strong> the complex stress distribution under tyres; especially the<br />

horizontal shear stresses are <strong>of</strong> importance. These are not caused by braking but by the fact<br />

that free horizontal expansion <strong>of</strong> the tyre when loaded can not occur due to friction forces. In<br />

order to take these stresses into account the stress conditions under a tyre were modelled in<br />

the way shown in figure 33.<br />

Position [mm] Radius Stress [kPa]<br />

Load X y [mm] X y Z<br />

1 +60 +90 52.57 -200 0 +400<br />

2 +70 0 42.57 -200 0 0<br />

3 +60 -90 52.57 -200 0 +400<br />

4 -60 +90 52.57 -200 0 +400<br />

5 -70 0 42.57 -200 0 0<br />

6 -60 -90 52.57 -200 0 +400<br />

7 +90 0 22.57 -180 0 0<br />

8 -90 0 22.57 -180 0 0<br />

9 0 0 112.57 0 +150 +750<br />

10 0 0 50.00 0 -60 +750<br />

Figure 33: Schematisation <strong>of</strong> the contact stresses under a tyre.<br />

The following relation was found:<br />

ε r1,b = 194.895 – 20.7769 SCI 300<br />

0.5<br />

Where: ε rt,b<br />

= tensile strain at pavement surface [µm/m]


54<br />

Compressive vertical strain at the top <strong>of</strong> the unbound base:<br />

The vertical compressive strain at the top <strong>of</strong> the subgrade is a well known design criterion.<br />

Such a criterion doesn’t exist for e.g. unbound base materials. Nevertheless it can be<br />

expected that if the compressive strains at the top <strong>of</strong> the unbound base become too large,<br />

excessive deformations might develop there as well.<br />

In order to develop an estimation procedure for the compressive strain at the top <strong>of</strong> the<br />

unbound base, Alemgena [25] analysed the same structures as were analysed by van Gurp.<br />

It appeared that the development <strong>of</strong> such a relation was rather complicated and was only<br />

possible for particular types <strong>of</strong> pavement.<br />

Alemgena found the following predictive equation:<br />

Log ε vb = 1.5615 + 0.3743 log SCI 300 + 1.0067 log BDI + 0.8378 log d 0<br />

- 1.9949 log d 1800 + 0.6288 log d 300<br />

This equation is only valid for the following conditions:<br />

a. the pavement shouldn’t be an inverted pavement so E 1 > E 2 > E 3 > E 4 ,<br />

b. the stiffness <strong>of</strong> the upper layer shouldn’t exceed four times the underlying<br />

layer (e.g. E 2 ≤ 4 E 3 ),<br />

c. applicable only for weak bases (i.e. E 2 < 1000 Mpa).<br />

Tensile strain at the bottom <strong>of</strong> the bound base:<br />

The following relation was developed:<br />

log ε r2,o = 0.0931 + 0.4011 log d 0 + 0.3243 log d 1800 + 0.4504 log d 300 – 0.9958 log d 900<br />

+ 0.8367 log BDI<br />

Where: ε r2,o<br />

= tensile strain at the bottom <strong>of</strong> the bound base [µm/m]<br />

Compressive vertical strain at the top <strong>of</strong> the subbase and subgrade:<br />

Two cases have to be considered which are the case where the stiffness <strong>of</strong> the subbase is<br />

higher than that <strong>of</strong> the subgrade and the case where the stiffness <strong>of</strong> the subbase is smaller<br />

than that <strong>of</strong> the subgrade. In the first case the surface modulus plot will shown an increase in<br />

stiffness going from bottom to top while in the second case the surface modulus plot indicates<br />

the presence <strong>of</strong> low stiffness layers on top <strong>of</strong> the subgrade.<br />

The following results were obtained:<br />

a. Subbase is stiffer than the subgrade:<br />

log ε v3 = 2.48589 + 0.34582 log SCI 300 + 0.16638 log d 1800 – 0.68746 log (h 1 + h 2 )<br />

+ 0.47432 log BDI<br />

b. Subbase is less stiff than subgrade:<br />

log ε v3,s = 1.52887 + 0.39502 log SCI 300 – 0.84168 log d 1800 – 0.60888 log (h 1 + h 2 )<br />

+ 0.43195 log BDI – 0.78407 log BCI + 1.73707 log d 600<br />

c. Subgrade:<br />

log ε v4 = 2.48589 + 0.34582 log SCI 300 + 0.16638 log d 1800 – 0.68746 log (h 1 + h 2 + h 3 )<br />

+ 0.47432 log BDI<br />

Where: ε v3<br />

ε v4<br />

ε v3,s<br />

= vertical compressive strain at the top <strong>of</strong> the subbase [µm/m]<br />

= vertical compressive strain at the top <strong>of</strong> the subgrade [µm/m]<br />

= vertical compressive strain at the top <strong>of</strong> the subbase when this layer<br />

has a lower stiffness than the subgrade [µm/m]


55<br />

10.2 Temperature correction method:<br />

As mentioned before, temperature has a large influence on the magnitude <strong>of</strong> the measured<br />

deflections. In order to be able to use the simplified relations between SCI and strain in the<br />

asphalt layer which were discussed in the previous paragraph, a temperature correction<br />

procedure adaptable to these relations should be available. Furthermore the correction<br />

procedure should take into account the effect <strong>of</strong> cracks present in the pavement. A fully<br />

cracked pavement e.g. acts like a block pavement and in such conditions a temperature<br />

correction is not needed on the measured deflections. On the other hand it is obvious that the<br />

effect <strong>of</strong> temperature is the largest on a sound asphalt layer.<br />

A procedure taking into account both effects is described in [10] and is discussed hereafter.<br />

The surface curvature index measured at a specific temperature can be corrected to a<br />

reference temperature using:<br />

TNF = 1 + {(a 1 + a 2 / h 1 ) . (T A – 20) + (a 3 + a 4 / h 1 ) . (T A – 20) 2 } . (1 – SR t )<br />

Where: TNF = temperature normalisation factor,<br />

T A = asphalt temperature [ 0 C],<br />

h 1 = thickness <strong>of</strong> the asphalt layer [mm],<br />

SR t = percentage area cracked / 100.<br />

TNF takes values smaller than 1 if the measurements are taken below the reference<br />

temperature <strong>of</strong> 20 0 C (which is the reference temperature in the Netherlands). Consequently<br />

TNF is larger than 1 if the measurements were taken above 20 0 C.<br />

The constants a 1 to a 4 take the following values:<br />

Variable a 1 [ 0 C -1 ] a 2 [mm / 0 C] a 3 [0.001 0 C -2 ] a 4 [mm / 0 C 2 ]<br />

D 0 0.01661 -0.67095 0.28612 -0.01408<br />

SCI 225 0.05955 -2.73223 1.48011 -0.08171<br />

SCI 300 0.05398 -2.61130 1.28439 -0.07493<br />

SCI 450 0.04720 -2.39175 1.05022 -0.06371<br />

SCI 600 0.04190 -2.15168 0.87228 -0.05301<br />

The correction is applied in the following way. The SCI 300, T measured at temperature T is<br />

corrected to a SCI 300, 20C at 20 0 C following:<br />

SCI 300, 20C<br />

= SCT 300, T / TNF<br />

A simple but highly effective technique to estimate the temperature in the asphalt layer is<br />

given below. The procedure has been developed in [12] and is slightly modified in [10].<br />

T 3 = 8.77 + 0.649 T 0 + (2.20 + 0.044 T 0 ) . sin {2 π (h r – 14) / 24}<br />

+ log (h 1 / 100) . [-0.503 T 0 + 0.786 T 5 + 4.79 sin {2 π (h r – 18) / 24}]<br />

Where: T 3 = temperature at third point in the asphalt layer [ 0 C]<br />

T 0 = pavement surface temperature [ 0 C]<br />

T 5 = prior mean five days air temperature [ 0 C]<br />

h 1 = asphalt thickness [mm]<br />

= time <strong>of</strong> the day in 24 hour system [hr]<br />

h r<br />

10.3 Relationships with other pavement strength indicators such as SNC:<br />

Also in [11], valuable relationships are presented which relate the deflection bowl to the<br />

modified structural number SNC as used in the Highway <strong>Design</strong> Model. The relationship that<br />

was developed is shown here-after.<br />

log SNC<br />

= 1.82472 + 0.03344 log h 1 + 0.11832 log BCI – 0.16207 log BDI<br />

+ 0.12659 log d 0 – 0.57878 log d 900 + 0.19996 log d 1800 - 0.19829 log SCI 300


56<br />

This relationship opens possibilities for characterising pavement strength by means <strong>of</strong> a well<br />

known physical quantity.<br />

10.4 Relationships between the falling weight deflections and deflections measured<br />

with the Benkelman beam:<br />

Furthermore an extensive study was made in [11] <strong>of</strong> the relationships that could exist<br />

between the deflections as measured by means <strong>of</strong> a BB and those by means <strong>of</strong> a FWD. The<br />

relations that were developed are reported hereafter.<br />

It should be noted that the BB measurements were done with a rear axle load <strong>of</strong> the loading<br />

vehicle <strong>of</strong> 63.5 kN (this is the same axle load as used in the TRRL procedure). As mentioned<br />

before the FWD measurements were taken at a load level <strong>of</strong> 50 kN.<br />

It should also be noted that the relations shown are those between the BB values which are<br />

not corrected for the movement <strong>of</strong> the support system and the FWD values. Table 6 gives the<br />

results.<br />

FWD deflection [µm]<br />

Variable Unit log d 0 log d 300 log d 600 log d 900 log d 1200 log d 1500 log d 1800<br />

Constant µm + 1.61 + 1.44 +1.40 + 1.31 + 1.23 + 1.19 + 1.14<br />

log BB 0 µm + 0.49 + 0.29 0 0 0 0 0<br />

log BB 500 µm + 1.23 + 1.11 + 0.83 + 0.38 + 0.23 + 0.13 + 0.33<br />

log BB 1000 µm - 1.53 - 1.06 - 0.43 0 0 0 - 0.48<br />

log BB 2000 µm + 0.47 + 0.32 + 0.31 + 0.34 + 0.55 + 0.67 + 0.96<br />

log BB 3500 µm 0 0 - 0.08 - 0.10 - 0.14 - 0.15 - 0.16<br />

log h 1 mm - 0.33 - 0.27 - 0.25 - 0.25 - 0. 26 - 0.27 - 0.27<br />

Table 6: Regression coefficients <strong>of</strong> the conversion formulas BB values to FWD values.<br />

The variables BB x are related to the deflections which are measured when the rear axle <strong>of</strong> the<br />

loading truck is at a distance <strong>of</strong> x mm from the tip <strong>of</strong> the beam. An example how the equations<br />

should read is given below.<br />

log d 900 = + 1.31 + 0.38 log BB 500 + 0.34 log BB 2000 - 0.1 log BB 3500 – 0.26 log h 1<br />

It should be mentioned that these relations have been developed using a BB with the<br />

following dimensions.<br />

pivot<br />

610 mm<br />

2695 mm<br />

915


57<br />

11. Remaining life estimation from visual condition surveys:<br />

As has been indicated in the previous chapters, visual condition surveys give important information<br />

on the condition <strong>of</strong> the pavement. With respect to the structural condition <strong>of</strong> the<br />

pavement, two damage types are <strong>of</strong> importance which are cracking and permanent<br />

deformation especially when the deformation is due to deformation <strong>of</strong> the base subbase or<br />

subgrade.<br />

In the past, several condition prediction models using visual condition surveys as input have<br />

been developed (e.g. [9]). Mostly these models suffer from accuracy because in practice the<br />

damage is seldom allowed to grow to such an extent and severity that models describing the<br />

progression <strong>of</strong> the damage completely could not be developed. Fortunately such information<br />

can be obtained from sections tested by accelerated loading facilities. In this chapter the<br />

models will be discussed which have been developed from observations made on test<br />

sections at the outside facilities <strong>of</strong> the Road and Railways Research Laboratory <strong>of</strong> the <strong>Delft</strong><br />

University, which were tested by means <strong>of</strong> the <strong>Delft</strong> University accelerated pavement testing<br />

facility called LINTRACK [13, 14].<br />

Before going into the discussion <strong>of</strong> the models developed, attention is called for the fact that<br />

in the analysis <strong>of</strong> visual condition survey data one always has to consider the way in which<br />

the information is obtained.<br />

The models for the prediction <strong>of</strong> the development <strong>of</strong> the amount <strong>of</strong> cracking that are going to<br />

be presented are based on the visual condition survey system used by the Road and<br />

Hydraulics Engineering Division <strong>of</strong> the Dutch Ministry <strong>of</strong> Transport. The unit section length is<br />

100m. The length over which longitudinal cracking is visible in the left and right hand wheel<br />

track is determined and divided by 200; the ratio obtained is called LC. In the same way the<br />

amount <strong>of</strong> alligator cracking is determined and again this number is divided by 200 in order to<br />

obtain the ratio AC. The amount <strong>of</strong> cracking is then calculated from the sum LC + AC.<br />

It has been shown that the progression <strong>of</strong> cracking can very well be described by means <strong>of</strong> a<br />

Weibull function following [15, 16]:<br />

F w (n) = 1 – exp [-( n/µ) β ]<br />

Where: F w (t) = probability that failure has occurred before n load repetitions,<br />

n = number <strong>of</strong> load repetitions,<br />

µ = number <strong>of</strong> load repetitions at which 63% <strong>of</strong> the area considered is cracked,<br />

β = curvature parameter.<br />

Analysis <strong>of</strong> the World Bank cracking models incorporated in the HDM III design system [7]<br />

showed that β was dependent on the stiffness <strong>of</strong> the pavement. The LINTRACK experiments<br />

indicated that β was dependent on the asphalt thickness following:<br />

log β<br />

= -0.08 + log h<br />

Where: h<br />

= asphalt thickness [mm].<br />

In the LINTRACK test sections also permanent deformation was observed. It was shown that<br />

this deformation was due to deformation <strong>of</strong> the subgrade. The permanent deformation was<br />

measured at several locations under a 1.2 m long straight edge and the mean value was<br />

determined. The maximum allowable rut depth was set at 18 mm and the number <strong>of</strong> load<br />

repetitions needed to arrive to this depth was determined. The rut formation could then be<br />

described using the following non dimensional model:<br />

S n / S N = ( n / N ) 0.41<br />

Where: S n<br />

S N<br />

n<br />

N<br />

= rut depth after n load repetitions [mm],<br />

= rut depth at which pavement is considered to be failed = 18 mm,<br />

= number <strong>of</strong> load repetitions applied,<br />

= number <strong>of</strong> load repetitions needed for a rut depth <strong>of</strong> 18 mm.


58<br />

The remaining pavement life can easily be predicted by means <strong>of</strong> these normalised<br />

equations. One measures the amount <strong>of</strong> damage that is present and one sets the maximum<br />

amount <strong>of</strong> damage which is just acceptable before maintenance is needed. From the ratio<br />

present amount <strong>of</strong> damage over allowable amount <strong>of</strong> damage the pavement life ratio can be<br />

determined. By using the procedure outlined in chapter 8, the damage ratio can be translated<br />

in a number <strong>of</strong> years before maintenance is required.


59<br />

12. Procedures to estimate material characteristics:<br />

In the previous chapters ample attention has been paid to the assessment <strong>of</strong> the stresses and<br />

strains at critical locations in the pavement. It has also been stressed that a proper evaluation<br />

<strong>of</strong> the remaining life and determination <strong>of</strong> the required overlay thickness cannot be made<br />

without knowledge on the strength <strong>of</strong> materials. Especially knowledge on the fatigue<br />

characteristics <strong>of</strong> the asphalt and the resistance to permanent deformation <strong>of</strong> the unbound<br />

base, subbase and subgrade is <strong>of</strong> importance.<br />

In this chapter transfer functions that allow the pavement life to be assessed will be<br />

presented.<br />

12.1 Fatigue characteristics <strong>of</strong> asphalt mixtures:<br />

The fatigue resistance <strong>of</strong> asphalt mixtures is usually described following:<br />

Log N = log k 1 – n log ε<br />

Where: N<br />

k 1 , n<br />

ε<br />

= number <strong>of</strong> load repetitions to failure,<br />

= material parameters,<br />

= applied strain level.<br />

It has been shown that the exponent n strongly depends on the slope <strong>of</strong> the master curve <strong>of</strong><br />

the stiffness modulus. Figure 34 is an example <strong>of</strong> such an relationship.<br />

Figure 34: Example <strong>of</strong> the relationship between the loading time and the stiffness <strong>of</strong> an<br />

asphalt mixture.<br />

Relationships like those shown in figure 34 can be determined experimentally by means <strong>of</strong><br />

e.g. repeated load indirect tensile tests. If such tests cannot be performed, the stiffness<br />

modulus <strong>of</strong> the asphalt mixture can also be estimated using the Shell nomographs for the


60<br />

estimation <strong>of</strong> the bitumen and mixture stiffness. Input that is needed to feed those<br />

nomographs is the T R&B and PI <strong>of</strong> the bitumen as well as the volumetric composition.<br />

If we call the slope <strong>of</strong> the relationship between log t and log S mix , m, then this value can be<br />

calculated using the following relationship.<br />

m = d (log S mix ) / d (log t)<br />

The exponent <strong>of</strong> the fatigue relationship, n, can then be calculated using [17]:<br />

n = 2 / {m . (0.541 + 0.346 / m – 0.0325 V a )<br />

Where: V a = void content <strong>of</strong> the asphalt mixture [%]<br />

The intercept value log k 1 is calculated in the following way [17].<br />

log k 1<br />

= 6.589 – 3.762 n + 3209 / S mix + 2.332 log V b + 0.149 V b / V a + 0.928 PI<br />

-0 .0721 T R&B<br />

Where: n = slope <strong>of</strong> the fatigue relation,<br />

S mix = stiffness <strong>of</strong> the asphalt mixture [MPa],<br />

V b = volume percentage <strong>of</strong> bitumen [%],<br />

V a = void percentage [%],<br />

PI = penetration index <strong>of</strong> the bitumen,<br />

= s<strong>of</strong>tening point <strong>of</strong> the bitumen [ 0 C].<br />

T R&B<br />

The relationship for log k 1 was established using the fatigue test results performed on over<br />

100 mixtures. Results involved were those reported by the SHRP A-003 team, by Shell<br />

researchers, researchers <strong>of</strong> the <strong>Delft</strong> University and by researchers <strong>of</strong> the Road and<br />

Hydraulics Engineering Division <strong>of</strong> the Dutch Ministry <strong>of</strong> Transport.<br />

The relationship for n was established using the results obtained for over 30 mixtures. All<br />

tests considered were displacement controlled tests.<br />

12.2 Deformation resistance <strong>of</strong> unbound base materials:<br />

The vertical permanent deformation in unbound base materials is usually described using:<br />

ε p<br />

= 10 a . N b<br />

Where: ε p = permanent strain [µm/m],<br />

a, b = material constants,<br />

N =number <strong>of</strong> load repetitions.<br />

The parameters a and b are dependent on the type <strong>of</strong> material, the gradation, the degree <strong>of</strong><br />

compaction and the moisture content. It is common practice to determine these parameters<br />

by means <strong>of</strong> repeated load triaxial tests.<br />

At the <strong>Delft</strong> University, an extensive testing program has been performed by van Niekerk [26]<br />

on base materials composed <strong>of</strong> crushed concrete and crushed masonry. Recycling <strong>of</strong> old<br />

concrete and masonry is a very important issue in the Dutch road industry. Van Niekerk’s<br />

results were used by Alemgena to develop base compressive strain criteria. This was<br />

possible since both permanent deformation as well as resilient modulus tests were performed.<br />

From the permanent deformation tests it was determined at which number <strong>of</strong> load repetitions<br />

a permanent deformation <strong>of</strong> 4% occurred. This number <strong>of</strong> load repetitions is <strong>of</strong> course<br />

dependent on the material type and the stress conditions. For each stress condition applied<br />

also the resilient modulus could be determined and so the elastic strain. Using all this<br />

information relations between the elastic strain and the number <strong>of</strong> load repetitions at which a<br />

permanent deformation <strong>of</strong> 4% occurred was determined. Typical examples <strong>of</strong> such criteria are<br />

shown in figure 35 while figure 36 gives the gradations. The code UL-65-100 e.g. means that<br />

the gradation is the UL gradation, that the mixture is composed with 65% crushed concrete


61<br />

and 35% crushed masonry (mass percentages) and that the samples were compacted to<br />

100% <strong>of</strong> standard proctor.<br />

Allowable vertical compressive strain at top <strong>of</strong> base<br />

log eps [mum/m]<br />

3,4<br />

3,3<br />

3,2<br />

3,1<br />

3<br />

2,9<br />

2,8<br />

2,7<br />

2,6<br />

2,5<br />

2 3 4 5 6 7 8<br />

log N<br />

CO-65-100<br />

AL-65-100<br />

UL-65-100<br />

LL-65-100<br />

Figure 35: Base strain criteria.<br />

100<br />

cummulative percentage passing<br />

[%]<br />

80<br />

60<br />

40<br />

20<br />

0<br />

UL<br />

UN<br />

LL<br />

CO<br />

AL<br />

FL<br />

0.063 0.125 0.25 0.5 1 2 4 8 16 31.5<br />

sieve diameter [mm]<br />

Figure 36: Gradations tested by van Niekerk.<br />

Figure 35 clearly shows that base strain criteria can be developed but that there doesn’t exist<br />

a single base strain criterion. The criterion is clearly dependent on the gradation but also on


62<br />

the composition and the degree <strong>of</strong> compaction (these later two influence factors are not<br />

shown here).<br />

12.3 Subgrade strain criterion:<br />

Research with the LINTRACK [13] has shown the following criterion to be applicable for a<br />

typical Dutch fore shore sand. The relationship is based on allowing a maximum rut depth <strong>of</strong><br />

18 mm.<br />

log N<br />

= - 7.461 – 4.33 log ε v<br />

Where: ε v<br />

= subgrade strain [µm/m]<br />

12.4 Maximum tensile strain at bottom <strong>of</strong> the bound base:<br />

The terminology “bound base” is used for any base materials to which some kind <strong>of</strong> binding<br />

agent is added or for base materials which shown some kind <strong>of</strong> self cementing action. Such<br />

materials will always show cracks due to shrinkage. With appropriate measures, the influence<br />

<strong>of</strong> that type <strong>of</strong> cracking can be kept under control. Nevertheless also deterioration due to<br />

traffic loads will occur. Given the shrinkage cracks that are already present in the base, it is<br />

not realistic to assume that the base is a homogeneous material and it is not realistic to<br />

analyse the possibility <strong>of</strong> fatigue cracking as is usually done for asphalt layers. It is however<br />

wise to limit the tensile strains due to traffic in order to avoid extensive traffic related damage.<br />

For that reason it is proposed to keep the tensile strain due to traffic loads below 50% <strong>of</strong> the<br />

tensile strain at failure, so below a level <strong>of</strong> approximately 60 µm/m.


63<br />

13. Overlay design in relation to reflective cracking:<br />

The main purpose <strong>of</strong> the overlay design procedure that was presented in one <strong>of</strong> the previous<br />

chapters was to limit the tensile strain at the bottom <strong>of</strong> the existing asphalt layer and the<br />

compressive vertical strain at the top <strong>of</strong> the subgrade. As was indicated such a method<br />

doesn’t take into account the effect <strong>of</strong> cracks in the existing pavement on the performance <strong>of</strong><br />

the overlay. This is a serious issue since these cracks tend to propagate through the overlay<br />

and can reduce the effective life <strong>of</strong> the overlay significantly. The conclusion therefore must be<br />

that crack reflection must be considered when designing an overlay.<br />

Linear elastic theory applied on homogeneous, isotropic layers can be used in overlay design<br />

procedures which are based on limitation <strong>of</strong> the stresses and strains in the existing pavement.<br />

Cracked pavement however cannot be analysed in this way. In fact principles <strong>of</strong> fracture<br />

mechanics have to be applied to analyse the effects <strong>of</strong> cracks. This immediately implies that<br />

finite element programs need to be used for the analysis <strong>of</strong> crack propagation. Although such<br />

programs can easily be used on today’s personal computers, they are still considered to be<br />

not practical for every day’s use. Therefore there has always been a strong need for so called<br />

“engineering tools” which allow the complex phenomenon <strong>of</strong> reflective cracking to be analysed<br />

with rather simple tools.<br />

Although the author fully understands this need <strong>of</strong> practice, he also likes to stress that each<br />

model is a schematisation <strong>of</strong> reality and that too simple models will be a too simple schematisation<br />

<strong>of</strong> reality which can result in less optimal or even wrong results.<br />

In spite <strong>of</strong> these drawbacks, some simplistic models are presented here-after because they<br />

are based on sound analyses <strong>of</strong> pavement structures using fracture mechanics principles.<br />

13.1 Overlay design method based on effective modulus concept:<br />

The first method to be presented is based on the effective modulus concept. This concept is<br />

schematically shown in figure 37.<br />

A B layer has reduced, effective modulus<br />

Figure 37: Concept <strong>of</strong> effective modulus method.<br />

Figure 37a shows the condition one is dealing with in reality when designing an overlay. The<br />

overlay is placed on the cracked pavement and this crack wants to propagate through the<br />

overlay because <strong>of</strong> stress concentrations at the tip <strong>of</strong> the crack due to the bending and<br />

shearing action <strong>of</strong> the load. The stress concentrations due to the bending action are indicated<br />

by K 1 (the horizontal arrow), those due to the shearing action are indicated by K 2 (the vertical<br />

arrows). It should be noted that in fracture mechanics K is called the “stress intensity factor”.<br />

The growth <strong>of</strong> the crack due to K is described using:<br />

dc / dN = A . K eff<br />

n<br />

Where: dc / dN = increase in crack length c per load repetiton,<br />

K eff = effective stress intensity factor combining the bending and shearing<br />

effects and taking into account the fact that the K 1 and K 2 are not constant<br />

when the crack progresses through the overlay,<br />

A, n = material parameters.


64<br />

The life <strong>of</strong> the overlay N can simply be calculated using:<br />

N = h o / { dc / dN }<br />

Where: h o = overlay thickness [mm].<br />

In the effective modulus method (figure 37b), the tensile strain at the bottom <strong>of</strong> the asphalt<br />

layer is calculated, indicated by the horizontal arrow, and the fatigue life <strong>of</strong> the overlay is<br />

calculated using the appropriate fatigue relation. The magnitude <strong>of</strong> the tensile strain at the<br />

bottom <strong>of</strong> the overlay, and so the life <strong>of</strong> the overlay, is <strong>of</strong> course dependent on the modulus <strong>of</strong><br />

the existing asphalt layer. This modulus value should be reduced to such a level that the<br />

fatigue life <strong>of</strong> the overlay, calculated according to the principle <strong>of</strong> figure 37b, equals the life <strong>of</strong><br />

the overlay based on the crack propagation principles shown in figure 37a. The reduced<br />

modulus so obtained is called the effective modulus <strong>of</strong> the existing asphalt layer.<br />

Using these principles, figure 38 was developed [19]. In principle this figure is only valid for<br />

the following conditions.<br />

E 1 = modulus existing asphalt layer = 3000 MPa,<br />

h 1 = thickness <strong>of</strong> existing asphalt layer = 100 and 300 mm,<br />

h 2 = thickness <strong>of</strong> the base = 300 mm,<br />

E 2 = modulus <strong>of</strong> the base = 200 MPa,<br />

E 3 = subgrade modulus = 100 MPa,<br />

= modulus <strong>of</strong> the overlay = 5000 MPa.<br />

E o<br />

The graph shows that if the effect <strong>of</strong> a 60 mm overlay is to be analysed when placed on a<br />

severely cracked pavement where load transfer takes place across the crack, that an effective<br />

modulus for the existing asphalt has to be used <strong>of</strong> approximately 900 MPa.<br />

Figure 38: Effective modulus <strong>of</strong> the existing asphalt layer in relation to the thickness <strong>of</strong> the<br />

existing asphalt layer and overlay, and the amount <strong>of</strong> load transfer across a crack.


65<br />

13.2 Method based on stress intensity factors:<br />

This method is in fact an extension <strong>of</strong> the crack growth calculations that were made to<br />

develop the effective modulus method presented in the previous section. For this method [20]<br />

a number <strong>of</strong> pavement structures was considered and the propagation through the overlay <strong>of</strong><br />

cracks which were fully developed through the existing asphalt layer as well as 50 mm deep<br />

surface cracks was analysed. Figure 39 shows the analysed pavements as well as the three<br />

load conditions considered.<br />

Figure 39: Analysed pavement structures.<br />

Table 7 gives the regression equations and values for the regression parameters for the<br />

calculation <strong>of</strong> K 1eq and K eff for all three loading conditions. First <strong>of</strong> all the K 1eq should be<br />

estimated. This value represents the combined effect <strong>of</strong> K 1 and K 2 . Than K eff is determined;<br />

this value takes into account the variation <strong>of</strong> the K 1eq over the height <strong>of</strong> the overlay.<br />

Table 7a: Relationship between K 1eq and several pavement parameters.


66<br />

Table 7b: Relationship between K eff / K 1eq and several pavement parameters.<br />

When using these equations for the determination <strong>of</strong> the thickness <strong>of</strong> the overlay, values for A<br />

and n should be available.<br />

It has extensively been shown [ e.g. 9] that the value <strong>of</strong> the exponent n <strong>of</strong> the crack growth<br />

relation is equal to the value <strong>of</strong> the exponent n <strong>of</strong> the fatigue relation. For the estimation <strong>of</strong> n,<br />

the reader is therefore referred to section 12.1.<br />

Furthermore A and n appear to be strongly correlated following [17]:<br />

log A<br />

0.273 log Smix<br />

= - 2.890 – 0.308 n – 0.739 n<br />

Where: n<br />

S mix<br />

= slope <strong>of</strong> crack growth relation which is equal to the slope <strong>of</strong> the fatigue<br />

relationship, see section 12.1,<br />

= stiffness <strong>of</strong> the asphalt mixture [MPa].<br />

13.3 Overlay design method based on beam theory:<br />

The disadvantage <strong>of</strong> the method presented in section 13.2 is that it is only applicable for the<br />

conditions for which it has been developed. This means that there is a big chance that the<br />

real conditions are different from the conditions for which the method is developed which<br />

implies that the method only has a limited field <strong>of</strong> application.<br />

A more general applicable simple design system has therefore been developed in [21]. This<br />

method is based on the propagation <strong>of</strong> cracks in fully supported beams as described in [22].<br />

In the text hereafter the equations given in [22] will be given first <strong>of</strong> all. This is followed by an<br />

explanation how this method can be generalised to pavement systems.<br />

Let us consider the two loading conditions as shown in figure 40.<br />

The stress intensity factors at the tip <strong>of</strong> the crack due to bending and shearing can be<br />

calculated in the following way.<br />

K bending = k b . q . e -β/2 . sin (β . l / 2) / β 2 d 1.5<br />

K shearing = k s . q [1 + e -βl . [sin (β . l) – cos (β . l) / 4 β √ d<br />

β<br />

Where: k b<br />

k s<br />

q<br />

l<br />

= (E s / E) 0.33 / 0.55 d<br />

= dimensionless stress intensity factor due to bending,<br />

= dimensionless stress intensity factor due to shearing,<br />

= contact pressure [MPa],<br />

= width <strong>of</strong> loading strip [mm],


67<br />

c<br />

d<br />

E<br />

E s<br />

= length <strong>of</strong> the crack [mm],<br />

= thickness <strong>of</strong> the beam [mm],<br />

= modulus <strong>of</strong> the beam [MPa],<br />

= modulus <strong>of</strong> the supporting layer [MPa].<br />

l<br />

q<br />

E<br />

c<br />

d<br />

E s<br />

Bending<br />

Shearing<br />

Figure 40: Crack propagation in a fully supported beam as a result <strong>of</strong> bending and shearing.<br />

Figure 41 shows how the dimensionless stress intensity factors change in relation to the ratio<br />

c / d. As one observes the stress intensity factor due to shearing increases with increasing<br />

crack length. This is logical because with increasing crack length, the area that has to transfer<br />

the load decreases so the stresses in that area increase.<br />

Figure 41 however also shows that the stress intensity factor due to bending increases first<br />

with increasing crack length but then decreases to a value <strong>of</strong> zero. This is because <strong>of</strong> the fact<br />

that the crack reaches the neutral axis <strong>of</strong> the pavement at a given moment and penetrates the<br />

zone where horizontal compressive stresses are acting. Then the cracks stops to grow since<br />

the driving force has disappeared.


68<br />

Figure 41: Relationship between c / d and the dimensionless stress intensity factors.


69<br />

The question now <strong>of</strong> course is how this beam approach can be used for the design <strong>of</strong><br />

overlays on cracked pavements. The first step how to schematise a cracked pavement with<br />

overlay is shown in figure 42.<br />

overlay<br />

existing<br />

asphalt<br />

bound<br />

b<br />

d<br />

base<br />

subgrade<br />

Figure 42a: Pavement structures to be schematised.<br />

c = h existing asphalt + h bound base c = h existing asphalt<br />

d = h overlay + h existing asphalt + h bound base d = h overlay + h existing asphalt<br />

E s = E subgrade E s = combined modulus <strong>of</strong><br />

E = combined modulus <strong>of</strong> overlay, existing subgrade and base<br />

asphalt and bound base E = combined modulus <strong>of</strong><br />

overlay and existing<br />

asphalt<br />

Figure 42b: Schematised structures.<br />

The question now is how to arrive to the combined modulus values. This is done in the<br />

following way. First <strong>of</strong> all the layer moduli <strong>of</strong> the existing pavement are back calculated. In<br />

case the modulus <strong>of</strong> the subgrade and the unbound base have to be combined, the following<br />

equation has been suggested by Odemark.<br />

1 / E s = (1 / E 2 ) . {1 - √ [(a 2 + h e1 2 ) / (a 2 + (h e1 + g e2 ) 2 )]}<br />

+ (1 / E 3 ) . √ [(a 2 + h e1 2 ) / (a 2 + (h e1 + h e2 ) 2 )]<br />

Where: E s = combined modulus <strong>of</strong> subgrade and base,<br />

E 2 = modulus <strong>of</strong> the unbound base,<br />

E 3 = modulus <strong>of</strong> the subgrade,<br />

a = radius <strong>of</strong> loading area,<br />

h e1 = 0.9 h 1 (E 1 / E s ) 0.33 ,<br />

h 1 = thickness <strong>of</strong> the existing asphalt layer,<br />

E 1 = modulus <strong>of</strong> the existing asphalt layer,<br />

g e2 = 0.9 h 2 ,<br />

h 2 = thickness <strong>of</strong> the unbound base,<br />

h e2 = 0.9 h 2 (E 2 / E 3 ) 0.33 .


70<br />

From the nature <strong>of</strong> the equation it is clear that it has to be solved by iteration since E s can<br />

only be calculated if an initial value for E s is assumed.<br />

The combined modulus <strong>of</strong> the existing asphalt layer and the overlay can be calculated using<br />

Nijboer’s equation.<br />

E = E a . {[b 4 + 4 b 3 n + 6 b 2 n + 4 b n + n 2 ] / [n . (b + n) . (b + 1) 3 ]}<br />

Where: E<br />

E a<br />

b<br />

n<br />

= combined modulus <strong>of</strong> existing asphalt layer and overlay,<br />

= modulus <strong>of</strong> the existing asphalt layer,<br />

= thichkness <strong>of</strong> existing asphalt layer / thickness overlay,<br />

= modulus <strong>of</strong> overlay / modulus <strong>of</strong> existing asphalt layer<br />

In case one has to determine the combined modulus <strong>of</strong> the base, existing asphalt and<br />

overlay, then the combined modulus <strong>of</strong> the base and existing asphalt layer has to be<br />

determined first <strong>of</strong> all. Then the combined modulus <strong>of</strong> this value and the overlay has to be<br />

determined using the same equations. This means that in that case E a = combined modulus<br />

<strong>of</strong> the existing asphalt layer and the base, n = modulus <strong>of</strong> overlay / combined modulus <strong>of</strong><br />

existing asphalt and base, b = total thickness <strong>of</strong> existing asphalt layer and base / thickness <strong>of</strong><br />

the overlay.<br />

The procedure is illustrated by means <strong>of</strong> an example.<br />

Example:<br />

Assume a given pavement consisting <strong>of</strong> a 100 mm thick asphalt layer on a 300 mm thick base<br />

which in turn is placed on a subgrade. From the back calculation analysis it appeared that the<br />

modulus <strong>of</strong> the existing asphalt layer was 9000 MPa. The base had a modulus <strong>of</strong> 130 MPa<br />

and the subgrade a modulus <strong>of</strong> 50 MPa.<br />

First <strong>of</strong> all the E s value had to be calculated using the above mentioned equation. As a<br />

starting value for E s a value <strong>of</strong> 130 MPa was assumed. This resulted in a calculated E s value<br />

<strong>of</strong> 74 MPa. This value was used as starter for a second iteration, then a value for E s <strong>of</strong> 71<br />

MPa was obtained. A third iteration resulted in the same E s value so E s = 71 MPa.<br />

Then the stiffness <strong>of</strong> the overlay was determined from the mixture composition, the bitumen<br />

characteristics and the temperature and loading conditions. This procedure will not be<br />

illustrated here. The interested reader is referred to the lecture notes on Asphalt Materials<br />

CT4850. The modulus <strong>of</strong> the overlay was determined to be 8000 MPa.<br />

Using Nijboer’s equation a combined modulus for the existing asphalt layer and the overlay<br />

was calculated using:<br />

n = E overlay / E existing asphalt = 8000 / 9000 = 0.89<br />

b = h existing asphalt / h overlay = 100 / 50 = 2<br />

The combined modulus <strong>of</strong> existing asphalt and the overlay was calculated to be E = 8496<br />

MPa.<br />

The question now is what the stress intensity factors are at the tip <strong>of</strong> the crack that wants to<br />

penetrate the overlay. The pavement is severely cracked so only a limited amount <strong>of</strong> load<br />

transfer through aggregate interlock will occur.<br />

From the pavement geometry we know:<br />

c =length <strong>of</strong> the crack = thickness <strong>of</strong> the existing asphalt layer = 100 mm,<br />

d = thickness <strong>of</strong> existing asphalt layer + thickness <strong>of</strong> the overlay = 150 mm,<br />

so<br />

c / d = 0.66.<br />

From figure 41 it appears that one only has to take into account the shearing action.


71<br />

The pavement is loaded by truck wheels having a contact pressure q = 0.7 MPa. The radius<br />

<strong>of</strong> the loaded area = 150 mm, this means that l = 300 mm.<br />

We calculate:<br />

β = (E s / E) 0.33 / 0.55 d = (71 / 8496) 0.33 / 0.55 . 150 = 0.0025<br />

K shearing = k s . q . [1 + e -βl (sin βl – cos βl)] / 4 β √d<br />

= k s 0.7 [1 + e -0.0025 x 300 (sin 0.0025 x 400 – cos 0.0025 x 400)] / 4 x 0.0025 x √150<br />

= k s 0.7 [ 1 + 0.472 (0.841 – 0.540)] / 0.122 = k s 6.553<br />

Please note that in the calculation <strong>of</strong> the sin and cos, βl is in radians.<br />

If the K values are known, the number <strong>of</strong> load repetitions that is needed to allow the crack to<br />

reflect through the overlay can be calculated using the procedures given earlier.<br />

The question now <strong>of</strong> course is to what extent beam theory is representative for real pavement<br />

problems. This is <strong>of</strong> course not the case and some shift factors resulting in similar stress<br />

conditions in the beam as in the real pavement are therefore necessary. The easiest way is to<br />

do is to compare the stresses at the bottom <strong>of</strong> the beam with the stresses that would occur at<br />

the bottom <strong>of</strong> the top layer in the two layer system. Most probably the stresses at the bottom<br />

<strong>of</strong> the beam are higher than the stresses at the bottom <strong>of</strong> the layer. The correction factor that<br />

is needed to fit the stresses at the bottom <strong>of</strong> the beam to the stresses at the bottom <strong>of</strong> the<br />

layer can also be used as correction factor for the stress intensity factors.<br />

13.4 Effects <strong>of</strong> reinforcements, geotextiles, SAMI’s and other interlayer systems:<br />

In order to retard reflective cracking, various systems have been developed in time which can<br />

be used to do so. Examples <strong>of</strong> such systems are:<br />

1. Application <strong>of</strong> polymer modifications in the overlay mixture to enhance the crack<br />

resistance <strong>of</strong> the overlay material.<br />

2. Reinforcement <strong>of</strong> the overlay material in order to improve the crack growth resistance <strong>of</strong><br />

the material.<br />

3. Application <strong>of</strong> a low stiffness material between the existing pavement and the overlay in<br />

order to let the overlay behave independently from the existing pavement.<br />

Re 1: Polymer modifications have shown to be very effective in improving the crack resistance<br />

<strong>of</strong> asphalt mixtures. Especially SBS modifications have proven to be very useful. It is<br />

beyond the scope <strong>of</strong> these lecture notes to discuss in detail the selection <strong>of</strong> the most<br />

appropriate polymer modification. Nevertheless some practical guidelines will be given.<br />

It had been shown (e.g. in the lecture notes on asphalt materials) that a material has a high<br />

crack resistance when its tensile strength is high and when its fracture energy is high.<br />

Materials with such characteristics can easily be discriminated by tests like the indirect tensile<br />

test. This is schematically shown in figure 43.<br />

By measuring the load and the displacements, one can derive a plot showing the growth <strong>of</strong><br />

the tensile stress in relation to the growth <strong>of</strong> the tensile strain. A picture like figure 40 is then<br />

obtained. The peak value represents the tensile strength σ t , while the area enclosed by the<br />

plot represents the energy that is needed to fracture the specimen. This parameter is<br />

indicated by Γ.<br />

A modification should preferably have a positive effect on both the tensile strength and the<br />

fracture energy. In practice however it has been observed that modifiers that increase the<br />

tensile strength, decrease the fracture energy and vice versa. Only a limited number <strong>of</strong><br />

modifiers produce an improvement <strong>of</strong> both. By comparing plots like figure 43, the most<br />

effective modifier can easily be determined.


72<br />

σ<br />

σ t<br />

Γ<br />

ε<br />

Figure 43: Strength and fracture energy obtained in a (indirect) tensile test.<br />

Re 2: Asphalt mixtures can be reinforced in the same way as cement concrete can be<br />

reinforced. Vital aspects with respect to reinforcement are the modulus <strong>of</strong> the reinforcing<br />

material, its total cross sectional area, and the bond between the reinforcement and the<br />

surrounding asphalt.<br />

Materials like meshes made <strong>of</strong> polypropylene, glass fibres and steel are <strong>of</strong>ten propagated as<br />

reinforcing materials. The question however is whether they really can act as a reinforcing<br />

material. There are two reasons to doubt this. First <strong>of</strong> all the mesh might be a woven material<br />

which means that not the stiffness <strong>of</strong> the material from which the mesh is made is <strong>of</strong><br />

importance, but the stiffness <strong>of</strong> the mesh which might be fairly low. Secondly many meshes<br />

have a low physical thickness and are glued to the pavement by means <strong>of</strong> a tack coat. The<br />

question in such cases is whether the tack coat is stiff enough to provide a good bond<br />

between the reinforcing material and the surrounding asphalt. All this doesn’t necessarily say<br />

that such products are useless; what it really says that these products certainly can have an<br />

effect but that the effect is not likely to be a reinforcing effect. In some cases the effect <strong>of</strong> such<br />

materials is somewhere between reinforcing and separating.<br />

Although the effect <strong>of</strong> reinforcements including the effect <strong>of</strong> the bond stiffness should be<br />

analysed by means <strong>of</strong> finite element programs, the procedures presented above can be used<br />

as well. In such cases it is common practice to describe the effect <strong>of</strong> the reinforcement by<br />

using a lower value for the crack growth parameter A for the reinforced overlay than for the<br />

unreinforced overlay. No general applicable values for the way in which reinforcing materials<br />

reduce the A values when compared with reference unreinforced mixtures. These values<br />

should be derived by means <strong>of</strong> properly designed experiments. Excellent guidelines for such<br />

tests can be found in [23] and [24].<br />

Re 3: Cracks will not propagate into the overlay if the overlay behaves independently from the<br />

existing pavement. This can be accomplished by placing a chewing gum type layer between<br />

the existing pavement and the overlay. Such a chewing gum layer might be a 1.2 mm layer <strong>of</strong><br />

polymer modified bitumen sprayed on the existing pavement, but it might also be a non<br />

woven geotextile soaked with bitumen. In this case the geotextile acts as a container for the<br />

bitumen.<br />

The effect <strong>of</strong> such interlayers can easily be assessed by assuming that such layers have a<br />

thickness <strong>of</strong> 1 mm and having a stiffness <strong>of</strong> about 50 MPa.<br />

In general one will observe that the overlay thickness that should be used on top <strong>of</strong> such a<br />

chewing gum interlayer system is limited in thickness. This is because thick overlays attract<br />

tensile strains and will therefore not perform as good as one would expect.<br />

13.5 Load transfer across cracks:<br />

Especially in cases where the pavement has a cement treated base, ample attention should<br />

be given to the load transfer that takes place across a crack. This is because at low


73<br />

temperatures, the cement treated base will shrink. This shrinkage not only introduces extra<br />

stresses in the overlay but also has a significant effect on the load transfer that takes place<br />

across the crack. This load transfer will be zero when the crack is so wide that the crack faces<br />

don’t touch each other.<br />

A typical example <strong>of</strong> how the load transfer can vary during the year is given in figure 44. This<br />

figure shows the deflection bowls measured with a FWD (load was 85 kN) around a specific<br />

crack in the winter and in the summer. The pavement consisted <strong>of</strong> 200 mm asphalt on top <strong>of</strong><br />

a 300 mm thick sand cement base. The pavement showed transverse cracking due to<br />

shrinkage.<br />

It should be noted that in position a, six <strong>of</strong> the seven geophones are on one side <strong>of</strong> the crack<br />

where the loading plate is placed. In position f, only one geophone is on the side <strong>of</strong> the crack<br />

where the loading plate is.<br />

From the figure it is clear that in summer the load transfer is rather good. In general the<br />

deflection bowl is a fluid line. However in the winter the deflection bowls show that almost no<br />

load is transferred across the crack indicating that in that period <strong>of</strong> the year the shearing<br />

conditions <strong>of</strong> an overlay placed on top <strong>of</strong> such a crack will be severe.<br />

In such conditions only thick overlays or overlays with a heavy reinforcement have a chance<br />

to survive.<br />

Figure 44a: Surface deflections at a transverse crack in the summer.<br />

Figure 44b: Surface deflections at a transverse crack in the winter.


74<br />

14. Effects <strong>of</strong> pavement roughness on the rate <strong>of</strong> deterioration:<br />

It is a well known fact that driving over a rough pavement results in dynamic axle loads which<br />

can be fairly high. This is <strong>of</strong> special importance for rather thin pavements because repeated<br />

high dynamic wheel loads on one particular spot can result in premature failure at that<br />

location. Some knowledge on the effect <strong>of</strong> pavement roughness on pavement deterioration is<br />

therefore needed.<br />

The magnitude <strong>of</strong> the dynamic axle load depends on the pavement roughness, the speed <strong>of</strong><br />

the vehicle and characteristics <strong>of</strong> the vehicle like size, weight and properties <strong>of</strong> the spring<br />

suspension system. All this means that no unique relationship can be given between the<br />

pavement roughness and the dynamic axle loads.<br />

This means that the relationships given hereafter must be taken as indicative and not as hard<br />

predictions.<br />

It is not the intention to give here all the backgrounds <strong>of</strong> vehicle pavement interactions. Only<br />

some useful formula’s will be given. For further details the reader is e.g. referred to [15].<br />

In [21] two relationships are derived for the standard deviation <strong>of</strong> the dynamic axle loads <strong>of</strong> a<br />

particular truck with particular characteristics, which had a static axle load <strong>of</strong> 10 tons. The<br />

relationships are as follows:<br />

log σ<br />

log σ<br />

= - 0.5184 + 0.4075 log SV<br />

= 0.892 – 2.151 log PSI<br />

Where: σ = standard deviation <strong>of</strong> the dynamic loads due to a static axle load <strong>of</strong> 10 tons<br />

<strong>of</strong> a truck driving at 63 km / h [tons],<br />

SV = slope variance <strong>of</strong> the road pr<strong>of</strong>ile multiplied by 10 6 [rad 2 ],<br />

PSI = present service ability index = 3.27 – 1.37 (log SV – 0.78).<br />

In many countries <strong>of</strong> the world however the international roughness index IRI is used to<br />

characterise the pavement roughness.<br />

According to [15] the relation between PSI and IRI is as follows.<br />

PSI<br />

IRI<br />

-0.18 IRI<br />

= 5.0 e<br />

= 5.5 ln (5.0 / PSI)<br />

Where: IRI<br />

= in [m/km].<br />

The problem in these analysis is how to obtain the PSI or IRI; normally quite sophisticated<br />

equipment is used to measure pavement roughness and to derive PSI or IRI value from these<br />

measurements. Fortunately it is shown in [15] that the IRI can be obtained using straightedge<br />

measurements. The relationships given in [15] are:<br />

2m straightedge: PD mean = 1.23 IRI<br />

3m straightedge: PD mean = 1.58 IRI<br />

Where: PD mean = mean deviation <strong>of</strong> the pr<strong>of</strong>ile from the straightedge [mm].<br />

It is believed that these relationships help in identifying locations where high dynamic axle<br />

loads occur so where rapid deterioration might occur as well.<br />

In order to allow a more precise analysis <strong>of</strong> the effects <strong>of</strong> a rough road on the dynamic axle<br />

loads, the computer program ROUGHNESS, has been developed by Huurman <strong>of</strong> the <strong>Delft</strong><br />

University. The program can be found on the cd which is part <strong>of</strong> these lecture notes. The<br />

user’s manual for this program is given in appendix II.


75<br />

References:<br />

1. Molenaar, A.A.A.<br />

Pavement management systems, Part I, II and III.<br />

Lecture notes e54; Faculty <strong>of</strong> Civil Engineering; <strong>Delft</strong> University <strong>of</strong> Technology;<br />

<strong>Delft</strong> – 1993.<br />

2. AASHTO<br />

AASHTO guide <strong>of</strong> design <strong>of</strong> pavement structures 1986.<br />

Washington D.C. - 1986<br />

3. Holster, A.M.; Molenaar, A.A.A.; Van den Bosch, H.G.; Van Gurp, C.A.P.M.<br />

Comparison between observed and predicted pavement response.<br />

Report 7-91-209-14; Road and Railway Research Laboratory; <strong>Delft</strong> University <strong>of</strong><br />

Technology; <strong>Delft</strong> – 1991<br />

4. Stas, W.F.; Molenaar, A.A.A.; Van Gurp, C.A.P.M.<br />

Evaluation <strong>of</strong> the structural condition <strong>of</strong> test pavements FORCE project.<br />

Report 7-91-209-16; Road and Railway Research Laboratory; <strong>Delft</strong>University <strong>of</strong><br />

Technology; <strong>Delft</strong> - 1991<br />

5. Hoyinck, W.; Van den Ban, R.; Gerritsen, W.<br />

Lacroix overlay design by three layer analyses.<br />

Proc. 5 th Int. Conf. <strong>Structural</strong> <strong>Design</strong> <strong>of</strong> Asphalt <strong>Pavements</strong>.<br />

Vol 1, pp 410 – 420; <strong>Delft</strong> – 1982<br />

6. Kennedy, C.K.; Lister, N.W.<br />

Prediction <strong>of</strong> pavement performance and the design <strong>of</strong> overlays.<br />

TRRL Laboratory Report 833.<br />

Crowthorne – 1978<br />

7. Groenendijk, J.; Molenaar, A.A.A.<br />

Pavement design methods, a literature survey into linear elastic theory and condition<br />

prediction models.<br />

Report 7-93-209-31; Road and Railway Research Laboratory; <strong>Delft</strong> University <strong>of</strong><br />

Technology; <strong>Delft</strong> – 1993<br />

8. Shell International Petroleum Company Ltd.<br />

Shell pavement design manual; asphalt pavements and overlays for road traffic.<br />

London – 1978.<br />

9. Molenaar, A.A.A.<br />

<strong>Structural</strong> performance and design <strong>of</strong> flexible pavements and asphalt concrete overlays.<br />

PhD dissertation; <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> – 1983.<br />

10. Van Gurp, C.A.P.M.<br />

Characterization <strong>of</strong> seasonal influences on asphalt pavements with the use <strong>of</strong> falling<br />

weight deflectometers.<br />

PhD dissertation; <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> – 1995.<br />

11. Van Gurp, C.A.P.M.; Wennink, P.M.<br />

<strong>Design</strong>, structural evaluation and overlay design <strong>of</strong> rural roads (in Dutch).<br />

KOAC-WMD consultants; Apeldoorn - 1997.<br />

12. Stubstad, R.N.; Lukanene, E.O.; Baltzer, S.; Ertman-Larsen, H.J.<br />

Prediction <strong>of</strong> AC mat temperatures for routine load/deflection measurements.<br />

Proc. 4 th Int. Conf. On Bearing Capacity <strong>of</strong> Roads and Airfields.<br />

Vol. 1, pp 661 – 682; Minneapolis – 1994.<br />

13. Groenendijk, J.<br />

Accelerated testing and surface cracking <strong>of</strong> asphaltic concrete pavements.<br />

PhD Dissertation; <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> – 1998.<br />

14. Molenaar, A.A.A.; Groenendijk, J.; Van Dommelen, A.<br />

Development <strong>of</strong> performance models from APT.<br />

Proc. 1 st Int. Conference on Accelerated Pavement Testing; Reno – 1999.<br />

15. Paterson, W.D.O.<br />

Road deterioration and maintenance effects.<br />

John Hopkins University press; Baltimore – 1987.


76<br />

16. Bekker, P.C.F.<br />

Pavement performance modelling.<br />

Proc. Closing conf. TEMPUS JEP-1180 “Educational developments in pavement<br />

management systems”, pp 263 – 288: <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> – 1993.<br />

17. Medani, T.O.<br />

Characterisation <strong>of</strong> crack growth and fatigue behaviour <strong>of</strong> asphalt mixtures using simple<br />

tests.<br />

MSc Thesis; International Institute for Infrastructural, Hydraulic and Environmental<br />

Engineering; <strong>Delft</strong> – 1999.<br />

18. Kloosterman, H.J.; Molenaar, A.A.A.<br />

Model for the prediction <strong>of</strong> permanent deformation <strong>of</strong> unbound granular materials.<br />

Report WB-13 (7-79-115-5); Road and Railroad Research Laboratory; <strong>Delft</strong> University <strong>of</strong><br />

Technology; <strong>Delft</strong> – 1979.<br />

19. Van Gurp, C.A.P.M.; Molenaar, A.A.A.<br />

Simplified method to predict reflective cracking in asphalt overlays.<br />

Proc. 1 st RILEM Conf. On Reflective Cracking in <strong>Pavements</strong>, pp 190 – 198; Liege – 1989.<br />

20. Jacobs, M.M.J.; De Bondt, A.H.; Molenaar, A.A.A.; Hopman, P.C.<br />

Cracking in asphalt concrete pavements.<br />

Proc. 7 th Int. Conf. on Asphalt <strong>Pavements</strong>; Vol 1, pp 89 – 105; Nottingham – 1992.<br />

21. Molenaar, A.A.A.; Nods, M.<br />

<strong>Design</strong> method for plain and geogrid reinforced overlays on cracked pavements.<br />

Proc. 3 rd Int. RILEM Conference on Reflective Cracking in <strong>Pavements</strong>, pp 311 - 320.<br />

Maastricht – 1996.<br />

22. Lytton, R.L.<br />

Use <strong>of</strong> geotextiles for reinforcement and strain relief in asphalt concrete.<br />

Geotextiles and Geomembranes, Vol. 8, No. 3, 1989.<br />

23. De Bondt, A.H.<br />

Anti-reflective cracking design <strong>of</strong> (reinforced) asphaltic overlays.<br />

PhD Dissertation; <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> – 1999.<br />

24. Francken, L.; Vanelstraete, A.<br />

Prevention <strong>of</strong> reflective cracking in pavements.<br />

RILEM Report 18; E & F.N. Spon; London – 1997.<br />

25. Alemgena Alene Araya<br />

Estimation <strong>of</strong> maximum strains in road bases and pavement performance prediction.<br />

MSc thesis TRE 127.<br />

International Institute for Infrastructural Hydraulic and Environmental Engineering.<br />

<strong>Delft</strong> – 2002.<br />

26. Van Niekerk, A.A.<br />

Mechanical behavior and performance <strong>of</strong> granular bases and sub-bases in pavements.<br />

PhD Dissertation; <strong>Delft</strong> University <strong>of</strong> Technology; <strong>Delft</strong> - 2002

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!