CAN WE BUILD PERPETUAL PAVEMENTS?

or

can we build pavements that fulfill our expectations

André A.A. Molenaar*

*Emeritus professor Delft University of Technology, the Netherlands; a.a.a.molenaar@tudelft.nl

(corresponding author).

Abstract

In this paper problems related to building perpetual pavements are discussed. It will be shown

that not only the average quality of the pavement layers and materials is important but that

variability in material quality and layer thicknesses are key parameters ensuring a pavement to be

perpetual. Ample attention will be paid to how variability builds up when constructing asphalt

concrete layers and what the consequences are of these variations. Then the paper will discuss

how these variations can be kept under control.

Attention is not only paid to variations in the quality of asphalt concrete layers. Attention will

also be paid to the effect polymer modifications can have on extending the fatigue life of asphalt

layers and it will be shown that significant savings in asphalt layer thickness can be made when

using the right polymer modification.

Making unbound granular base courses perpetual is also a topic that is discussed in the paper. It

will be shown that compaction is key factor, next to moisture content and gradation, in realizing

a perpetual base course.

Finally it will be shown that contractual aspects can greatly affect the realization of a perpetual

pavement.

1. INTRODUCTION

Perpetual pavements, meaning pavements with a long lifetime, are especially needed for very

busy major roads in urban areas simply because traffic intensities are so high during the entire

day that the time slots available for doing maintenance are very short. Actually road authorities

are looking for solutions such that the number and duration of the maintenance moments are

reduced to a minimum in order to limit the hinder to traffic as much as possible. The question

however is whether we are able not only to design but also to build such pavements. Furthermore

we should know about the factors that are making building such pavements difficult or even

impossible because this knowledge can help us in finding solutions to overcome these problems.

The prospects of being able to build perpetual pavements are a bit gloomy because quite often

the pavements we build are not really giving the service we were expecting. Too often, the

pavement life is less than originally anticipated and too often major maintenance has to be

applied at a too early stage. Therefore this author believes that we first have to find out why

pavements are failing so often too early before we can start thinking about building perpetual

pavements. If we have found the reasons for this premature failure then we have to determine

whether we can come up with solutions for this problem.

To the opinion of this author, there are some reasons why pavements do not give the service they

are expected to give and why they need too early maintenance. The first reason is that the

warrantee period that is agreed in contracts is too short. In our private life we will seldom buy a

product on which no warrantee is given. We consider such a product as not trustful. Car

manufacturers have fully understand the necessity of giving a good warrantee. Some producers

even give a warrantee of 7 years while all producers will call their cars to the garage if some

unexpected defect appears. Such a client friendly attitude we don’t see in the pavement field.

Unfortunately, warrantee periods for pavements are much shorter. Quite often a warrantee period

of only one year is given and in some cases a 3 years warrantee period applies. There are

however also cases where no warrantee period applies at all. Although this author has

participated in the pavement industry for 40 years he never managed to understand why clients

seem not to be able to enforce sufficiently long warrantee periods and why contractors seem not

to be willing to accept extended warrantee periods.

The second reason why pavements are too often not performing to expectation is, according to

this author, the fact that the contractor is taking far too less risk when building a pavement. In

many contracts, the contractor has to build what is agreed upon in the contract and he has to

comply with specifications set to the required quality of the materials laid, layer thicknesses and

so on. Complying to the specifications however does not necessarily imply that the pavement

will perform as expected. When it doesn’t cases the contractor can easily defend himself by

saying “I did what you ask me to do so don’t blame me for disappointing performance of the

pavement I built”. If contracts would have been made such that the contractor would be much

more responsible for what he was building, implying that he would take a much greater risk, he

would act pro-active rather then re-active. Such a pro-active attitude implies that he would do

everything to solve problems before they even could occur rather then re-act on them when they

do occur.

The third reason is that contactors are not really given an incentive to produce a better product.

Contracts are awarded on the lowest price and not on the highest lifetime/cost ratio. Furthermore

a bonus to the contractor when the pavement performs better than anticipated is rarely awarded.

All this implies that there is not really an incentive for contractors to produce a better product

than the minimum required which immediately implies that some parts of the constructed road

will most likely be substandard (one cannot control each and every mm of a pavement).

This author is of the opinion that the chances that the client really gets what he paid for will not

be high when the conditions mentioned above are not improved.

Based on his 40 years’ experience, this author is of the opinion that the main reasons for

unsatisfactory pavement performance are the following.

Too low average pavement lifetime.

Too high variability in layer quality and layer thickness.

Not paying attention to important details.

Not collecting important information.

Ignoring important information.

In the remainder of this paper an example will be given first of all to illustrate the effects on

pavement performance of the last three mentioned items. The main body of the pavement

however will discuss extensively the effects of the first two mentioned items. It will be shown

how a too low average quality and especially a too high variability negatively affect pavement

life and how they reduce the chance on being able to build a perpetual pavement.

2. EFFECTS OF NOT COLLECTING CQ IGNORING IMPORTANT

INFORMATION ON PAVEMENT LIFE AND PERFORMANCE

The author has been involved in a road project in Africa where a pavement had to be built in an

area with an annual rainfall of 800 – 1000 mm per year which precipitated in a wet season

running from November to March. The specific problems that occurred during and immediately

after construction was the appearance of longitudinal cracking in hard shoulder. Figure 1 shows

the type of cracking that was observed. Extensive investigations showed that the longitudinal

cracks most probably were due to shrinkage – swell of the material under the shoulder. This was

initially believed to be a bit remarkable because the borrow pit material which was used to

construct the unbound base was initially classified as an A-2-4 or A-2-6 soil and in the materials

report prepared for this project there was no mentioning of swelling and or shrinking material.

However the report did quote the geological report in which it was mentioned that “the schist

derived from sedimentary material is largely composed of flaky mica and quartz”. However the

consequences of the presence of Mica in the soil to be retrieved from the borrow pits was not

mentioned and discussed in the materials report .

It is remarkable that this presence of Mica did not ring alarm bells because presence of Mica

implies that the material is highly resilient and difficult to compact.

The next thing that was alarming was that although Proctor compaction tests were done, no effort

was made to compare the gradation of the material after Proctor compaction with the gradation

of the material before Proctor compaction. Making such a comparison would immediately have

indicated whether or not the material is prone to crushing during compaction.

That crushing was a problem appeared when material taken from test pits was analyzed. It

appeared that most of the material was classified as A-4 or even A-6 clearly indicating that it had

degraded from a coarse to a fine graded materials. The analyses also showed a significant

increase in the amount of fines making the material sensitive for shrinkage and swell.

The reason for crushing of the material was over-compaction. Because the material contained

Mica, it was very difficult to achieve to prescribed density levels. They could only be achieved

by “hammering” the soil with heavy vibratory rollers which caused the material to crush.

In conclusion, three things went wrong. The indication that Mica was present was not taken into

account. Furthermore the crushing resistance of the material was not investigated and the swell

and shrinkage properties of the fines were not investigated. Ignoring and not collecting important

information was the main reason for the unexpected premature damage that had occurred and

which made the pavement not perform as was expected.

Figure 1. Longitudinal cracking in the hard shoulder

3. STRENGTH, COSTS, AND EFFECTS OF VARIABILITY

It is a well-known fact that the stronger we build our pavements, the less maintenance they

require. However building a stronger pavements also implies larger investmenst. If pavement

strength is related to construction and maintenance cost a trend line like the one shown in figure

2 is obtained. From this trend line one can conclude that there is some optimum strength value

from a cost point of view.

Pavements however never have a constant strength. The strength over a certain length of

pavement will vary because of a variation in layer thickness and material characteristics. So the

strength of a certain pavement section is not a constant but will show some kind of distribution.

This distribution however will affect the total costs as is schematically indicated in figure 3.

From this figure one can easily conclude that pavements which show a large amount of variation

in their strength are more costly than pavements which only show a limited amount of variation.

Figure 2. Construction, maintenance and total costs as a function of pavement strength.

Figure 3. Effect of variability in strength on total costs.

However, the decision on the pavement structure to be built does not only depend on the

construction and maintenance costs. Especially for heavily trafficked roads, perpetual pavements

are needed simply because of the fact that maintenance is hardly possible without creating a huge

amount of hinder to the road users. Experience in the Netherlands has shown that public fully

understands that roads need to be maintained but they do not understand why maintenance

activities have to be done over and over again. Public does accept that a certain road is only

limited available for a short period of time because of maintenance, but then they also expect that

the road is fully operational again for a long period of time after maintenance works have been

completed. Figure 4 shows the pavement performance people is ready to accept and which

performance is not acceptable to them.

Figure 4. Pavements with a long life showing less variability are favored by public.

If pavements perform like indicated by the white curve, then maintenance needs to be performed

regularly over relative short sections. However, when pavements perform according to the

orange line then the number of maintenance moments is far less and when maintenance needs to

be performed it can be done over a large section making planning and execution of maintenance

activities much more effective.

An example which proves this to be the case is given in figure 5. This figure shows that 10% of

the porous asphalt wearing courses as applied in the Netherlands are failing within the first 7

years after placement and that 90% of all sections fail after 16 years. One will agree that this is a

wide time window and that it indicates quite some variability in performance. A study was made

to investigate what the effect would be of limiting the amount of variation. The effects of having

10% failed sections after 9 instead of 7 years was investigated and the results showed that this

minor change in variability resulted in a reduction of maintenance costs of 20% and a reduction

in delay hours due to maintenance with 10%.

Figure 5. Effect of reduced variability on maintenance costs and delay hours due to maintenance.

A reduction with 10% in delay hours also brings a major environmental benefit since because of

that cars will use less petrol and will produce less fumes and air pollution. Especially reduction

of the truck delay hours is beneficial because trucks do have a far bigger environmental impact

then construction and maintaining a road does. This is shown in figure 6.

In conclusion one can state that building pavements with a longer life and less variability will

reduce maintenance costs, delay hours and pollution caused by traffic.

Figure 6. Contribution of truck and car traffic as well as construction and maintenance to the

environmental loading caused by a road traffic system.

4. HOW DOES VARIABILITY BUILD UP

When producing and laying asphalt mixtures, variation in the composition and mixture

properties can build up at various moment during the production and laying process. It starts

already when weighing the different ingredients and heating them to the required temperature.

This becomes more critical when large amounts of moist aggregates have to be dried and heated

or when large amounts of RAP at ambient temperature and with a certain moisture content are

added to the mixture. In a study performed at the Delft University of Technology [1], the

temperature of the mixture prepared with a double barrel drum mixer was measured. In this

mixture 50% RAP at ambient temperature and around 4% moisture was added to the virgin

materials. The concept of the double barrel drum is that virgin aggregates are heated in the inner

drum and then discharged in the outer drum where also the RAP, virgin filler and virgin bitumen

are added. Because of the fact that a large amount of moist RAP is added, the virgin aggregates

have to be preheated to very high temperatures, temperatures of 400 – 500 oC are not uncommon,

in order to arrive to the desired mixing temperature of 180 oC. The question of course is whether

a homogeneous temperature distribution is obtained given the short mixing times of 30 – 40

seconds. If such a homogeneous temperature distribution is not achieved then the chances that

the RAP has fully blended with the virgin materials are small.

Figure 7 shows the double drum and the location at which the infrared temperature

measurements were made.

Figure 7. Double barrel drum and location where the temperature measurements were made [1].

Figure 8 shows a thermo-graphic image of the asphalt mixture when it was falling on the

conveyor belt on its way to the silo. Superheated particles can be seen in the picture as well as

rather cool material which most probably is the RAP that was added to the mixture.

Figure 9 shows how the temperature was changing during a 30 minute period (1800 seconds) in

which approximately 250 tons of mixture was produced. The figure nicely shows that the

production temperature was initially too low but then became too high. This is because the plant

operator increased the temperature based on the information he was getting from the person who

was checking the temperature of the mixture on the conveyor belt by means of a thermometer.

These measurements are not continuous measurements but are done every now and then. The

figure shows that the mixing temperature more or less stabilized later on but that the temperature

of the mixture was not becoming homogeneous. The bottom part of the figure shows that the

temperature was not only varying in time (time is the horizontal axis) but also to some extent

over the width of the conveyor belt.

The conclusions of this investigation were as follows:

- superheated aggregates are still visible indicating a non-perfect temperature blending;

- a fair amount of variation in mixture temperature continues to occur during the

production process;

- continuous temperature measurements using infrared equipment will give a better control

on the mixture temperature and should therefore replace spot measurements using

thermometers.

Figure 8. Thermo graphic image of the material falling on the conveyor belt [1].

Figure 9. Temperature variations during a 30 minutes production period [1].

Having discussed the temperature inhomogeneity that can occur at the mixing plant, we will now

discuss the inhomogeneity that occurs when laying the mixture. Figure 10 shows the

measurements that were taken at a specific site [1]. In order to increase homogeneity, the client

demanded that a feeder between the truck and the paver was going to be used. Furthermore

using a feeder would avoid contact between the truck and the paver and would guarantee a

constant flow of material from the feeder to the hopper of the paver in this way avoiding the need

of the paver to stop because of not enough material being available. It was believed that all this

would help in getting a more homogeneous quality of the mixture.

Figure 10. Temperature measurements during a paving operation [1].

Figure 11 and figure 12 show the results obtained during paving an 8 m wide lane with 45 mm of

a binder course material. The speed of the paver was 3.5 m/min. Figure 10 shows that the

temperature varied between 118 and 158 oC. It should be mentioned that these temperatures are

surface temperatures and that the temperature in the middle of the asphalt layer will probably be

higher. Figure 11 shows that the temperatures varied between 125 and 163 oC. Both figures

further more show that the temperature varied quite a bit not only in the longitudinal direction of

the section (the horizontal axis of the bottom figure) but also in the transversal direction (the

vertical axis of the bottom figure). Figure 12 furthermore shows some kind of sinusoidal

variation in the temperature which, as is shown in the figure, could be related to the various truck

loads entering the hopper of the paver.

All in all the conclusion is that the temperature distribution of the paved mat was far from

homogeneous which would have consequences for compaction because a hot mixture is far

better to compact than a cool mixture.

Figure 11. Surface temperatures of a paved mat.

Figure 12. Surface temperatures of a paved mat.

Figure 13 shows the results of temperature measurements AND roller pass measurements on

another project.

Figure 13. Surface temperature and number of roller passes of a paved mat.

Figure 13 shows some worrying results. The upper part of the figure shows that temperatures

went down to 100 oC at the location where the paver had to stop because of the fact that no

trucks were arriving at the site. It will be obvious that this area is prone to premature damage

since it will not be possible to compact the material properly at such low temperatures. The

problem however significantly increases when we also consider the plot which shows the

number of roller passes. Although rolling was done by experienced operators, the picture shows

that a very uneven roller pass distribution was obtained. There is an area which received as many

as 20 passes (the area with the pink color in the upper left hand part of the picture) and there are

areas were only a few roller passes were applied (the blue colored areas at the bottom of the

picture). Figure 13 clearly shows that there are locations where the temperature was low AND

which only received a few roller passes. The question now is how all this affects the quality of

the material as laid.

5. EFFECTS OF VARIABILITY IN TEMPERATURE AND COMPACTION

EFFORT ON THE QUALITY OF THE LAID MIXTURE

At the Delft University of Technology a study [2] was made on the factors influencing the

compaction cq void content of porous asphalt wearing courses as a function of the mixture

temperature, mixture composition and compaction effort. The model that resulted from that study

is shown in figure 14.

Figure 14. Compaction model for porous asphalt concrete.

For the sake of completeness it should be mentioned that porous asphalt concrete is currently

used on all major highways in the Netherlands for noise reducing purposes. Figure 14 shows the

factors that play a role in the compaction of this type of wearing course but it should be

mentioned that the temperature of the mixture as well as the compaction effort are the most

important parameters. This model has been applied on the temperature and roller passes data

shown in figure 13 and the results are shown in figure 15. Figure 15 shows that the occurring

combinations of temperature and number of roller passes will result in an area with a void

content of 18% as well as in an area where the void content is 28%. The later area is the area

where the mixture temperature was low and where only a few roller passes were applied.

One of course might argue that this is just an example of an application of the model and one

might wonder whether such a wide range in void contents really occur in practice. Because of

this, figure 16 is added to this paper which shows the void contents as determined on an actually

constructed porous asphalt layer. The figure shows that the predicted void contents also occur in

practice and furthermore that the variation in void content over the width of the paved lane as

well as over the longitudinal direction of the paved lane is quite high. The figure also shows that

the variation in void content over the width of the paved lane is as high as the variation over the

length of the paved lane!

Figure 15. Effect of temperature and number of roller passes on the void content.

Figure 16. Variation in void content over the width and length of a porous asphalt concrete layer

[2].

Also the influence of the void content on the stiffness and tensile strength of the mixture that was

produced in the project from which figure 16 originated, was investigated. The influence of the

void content on the mixture stiffness, as determined with repeated load indirect tension tests, is

shown in figure 17.

Figure 17. Influence of the void content on the stiffness of a porous asphalt concrete mixture [2].

If the information from figure 17 is applied on the data shown in figure 15, then figure 18 is

obtained. Figure 18 clearly shows that the area with the high void contents has a significantly

lower stiffness than the well compacted area. Furthermore the figure shows that the area with a

void content of 18% has a tensile strength ft which is 1.4 times higher than that of the area where

the void content is 28%. Information on how this difference in tensile strength was determined

will not be given here; it is sufficient to say that there is a strong relation between the stiffness of

the material and its tensile strength. For further information about this topic the reader is referred

to [3].

It needs no explanation to understand that areas with high void contents, low stiffness and a low

tensile strength are prone for showing premature damage.

The question now is what kind of damage will result from too low compaction levels. It is a

well-known fact that raveling is the dominant defect type of this very open wearing course

mixture. Figure 19 shows what is meant by raveling of the pavement surface.

By means of an extensive study on the performance of porous asphalt concrete wearing courses

in the Netherland, researchers of the Delft University of Technology [4] were able to develop a

model that predicts the amount of raveling 8 years after construction. The model was developed

by using field data and by applying Artificial Neural Networks. This research showed that there

are two major factors that influence the resistance to raveling of porous asphalt concrete being

the void content and the bitumen content. These effects are shown in figures 20 and 21.

Figure 18. Stiffness and strength in relation to void content.

Figure 19. Raveling of porous asphalt concrete.

Figure 20. Influence of the void content on the amount of raveling of porous asphalt concrete 8

years after construction [4].

Figure 21. Influence of the bitumen content on the amount of raveling of porous asphalt concrete

8 years after construction [4].

Figures 20 and 21 clearly show that porous asphalt wearing courses having void contents of

higher than 19% and bitumen contents lower than 4% are prone to significant raveling. It should

be mentioned that the bitumen content in a porous asphalt layer is quite often not equally

distributed over the entire height of the layer. Very often the bitumen content in the upper part of

the layer is lower than that of the lower part and the bitumen content in the upper part appeared

to be quite often less than 3.8% by weight. This is because of drainage of the binder due to

gravity. No surprise that quite a number of porous asphalt wearing courses suffer from early

damage.

The lesson we have learned so far is that building a perpetual pavement is not as simple as it may

look. Because of a large number of reasons, the construction process might not proceed as

desired resulting in locally weak areas which are prone to premature maintenance. These areas

might occur even when the quality control performed on such pavements did not give rise to

doubt the quality of the layer.

One might say at this moment “this is all interesting but we don’t build porous asphalt concrete

layers so our problems will be less severe”. In order to be able to answer such a statement, this

author made an analysis of the effects of a too high void content and a too hard bitumen on the

quality of a dense mixture. The results of this analysis are shown in figure 22.

Figure 22. Effects of variations in void content and binder properties on the mechanical

characteristics of a dense mixture.

Figure 22 clearly shows that void content as well as bitumen properties have a large effect on the

mechanical properties of dense mixtures. The data in the figure also imply that keeping the

variability under control is required to obtain a mixture with homogeneous characteristics.

6. METHODS TO REDUCE THE PROBABILITY OF FAILURE AND

VARIABILITY IN CHARACTERISTICS

Figure 23 shows the concept of designing structures with a certain probability of failure (Pfailure)

cq a certain probability of survival (P=surv). The probability of survival can be written as:

Psurv = 1 - Pfailure

Figure 23. Methods to affect the probability of survival.

The figure shows that the probability of survival, so getting a pavement which has a higher

chance that it will perform to expectations, can be achieved by increasing the strength of the

materials involved without changing the variability or by decreasing the variability without

changing the average strength of the materials. Although not shown in this figure, similar effects

are obtained if the stresses in the pavement are decreased by increasing e.g. the layer thicknesses

or by reducing the variation in stress levels by limiting the layer thickness variation and the

variation in stiffness characteristics. Building a thicker pavement or using a stronger material

however is a costly option; reducing the variability is a much more cost effective option because

it means doing the same things in a more accurate way and giving attention to details.

Reducing the variability in asphalt mixture properties can be achieved by using a shuttle buggy

(figure 24) between the truck and the paver. The purpose of this device is to remix the mixture

such that both the composition and temperature gets much more homogenized resulting in a

lower variability of the mixture properties.

A second tool which is important is the intelligent compactor (figure 24). Such a device monitors

the homogeneity in compaction effort and it monitors the areas which are covered by the roller.

The device shown in figure 24 is based on measuring accelerations and it in fact monitors the

response of the entire pavement. This most not the most effective way of measuring the

compaction homogeneity of the asphalt wearing and base course. Most probably a method which

is based on measuring density using nuclear principles is to be preferred. Such a method is under

development at Troxler.

Figure 24. Reducing variability can be achieved by using shuttle buggies and intelligent

compaction.

Reducing variability however starts already when storing the raw materials (figure 25). Using

sufficient bins to store the various aggregate fractions is one thing. Covering the bins by a shed

to prevent the aggregates from getting wet and preventing overfilling of the bins resulting

aggregate fractions to get mixed up is another thing that can help significantly in keeping the

variation in mixture composition and so in mixture quality under control. These things are just a

matter of paying good attention to details and will hardly add to the costs.

7. BUILDING LONG LIFE PAVEMENTS

Reducing variability is not enough to build long life pavements, in order to achieve this also the

average pavement life needs to be increased. It is a well-know fact that modifying asphalt

mixtures by means of modifiers does increase the resistance to permanent deformation and

fatigue cracking. In many cases only the wearing course is modified to give it a better resistance

to permanent deformation but it is very advantageous also to modify the base course with

polymers. A very nice example of successfully modifying the entire asphalt thickness with

polymers is the pavement structure that is used for runways and taxiways at Amsterdam’s

Schiphol airport. Originally these pavements were made of a 60 cm thick lean concrete base on

top of which 27 cm of traditional base and wearing course mixtures were applied. Because these

Figure 25. Covering the bins with a shed would help in reducing the variability in mixture

properties.

pavements suffered from transverse shrinkage cracks having crack spacing’s between 15 and 45

m. Because of the resulting wide crack openings, particles broke loose from the cracked edges

which in turn increased the chance on foreign object damage to aircraft. Therefore the pavements

were redesigned starting in the early 1990’s resulting in using 20 cm of polymer modified

asphalt on top of the lean concrete base together with pre-cracking by means of cutting joints in

the lean concrete base. Since then the transverse cracking problem has been minimized. This

successful application of polymer modifications was then also adopted at other large

international airports where the specifications for the modified mixtures were copied from the

Schiphol specifications.

In 2008 a large research project was initiated by Kraton in which the Delft University

participated by means of testing asphalt mixtures modified by different types of polymers

supplied by Kraton [5]. Figure 26 gives an overview of the mixtures tested. On these mixtures

stiffness testing was performed as well as fatigue testing and the fatigue tests resulted in a so

called endurance limit which is the strain level below which no fatigue damage will develop. The

most important results of these tests are shown in figure 27.

The data shown in figure 27 were then used in the analysis of a pavement structure, the

characteristics of which are shown in figure 28. The objective of these analyses was to determine

which minimum asphalt thickness would be needed to arrive at the endurance limit. The results

of these analyses are shown in figure 29.

Figure 26. Mixtures tested at the Delft University of Technology as part of the research program

initiated by Kraton.

Figure 27. Endurance limit for two polymer modified and one reference mixture.

Figure 28. Analyzed pavement structures.

Figure 29. Pavement design analyses results.

Figure 29 shows that significant savings can be made in asphalt thickness when using certain

types of polymer modifications. The results shown in the figure also imply that almost

everlasting pavements can be built if the asphalt thickness is not reduced. Not reducing the

asphalt thickness might be needed to provide sufficient cover for the base and subbase to

avoid permanent deformation to occur in those layers.

The conclusion of this research therefore was that long lasting pavements can indeed be built

by modifying asphalt mixtures with the right type of modifier.

8. BUILDING PERPETUAL GRANULAR BASE COURSES

Until now the attention in this paper is focused on asphalt concrete layers and asphalt

concrete mixtures. Many pavements however have base and subbase layers which are made

of unbound granular materials. The question now is whether we also can build perpetual

unbound layers knowing that such layers show a much higher degree of variability than

asphalt concrete layers do.

An example of the variability that might occur in unbound layers is shown in figure 30 which

pictures the variation in degree of compaction of a particular unbound base.

Figure 30. Variation in degree of compaction of a particular base course layer.

Figure 30 shows that base course layers suffer from the same problem as asphalt concrete

layers do being the large variation in degree of compaction. Although not shown in the

picture, granular layers also suffer from the fact that the gradation might vary significantly

from location to location. All this of course has an effect on the performance of the base

course.

Triaxial testing (figure 31) has proven to be an excellent way in determining the effects of

moisture, degree of compaction and gradation on the stiffness and resistance to shear failure

and permanent deformation of granular materials.

Figure 31. Triaxial testing, an excellent way to characterize unbound granular materials.

Figure 32 shows how the cohesion and angle of internal friction of an unbound material is

affected by gradation and degree of compaction. The figure shows that a higher compaction

level results in improved cohesion, like a higher fines content does, but that the angle of

internal friction is hardly affected by gradation and compaction level.

The amount of permanent deformation that occurs in a granular base depends on how the

stresses are compared to the stresses at failure. It appeared that if the ratio applied vertical

stress at a particular confinement level : vertical stress at failure at the same confinement

level is less than 0.35, only a very limited amount of permanent deformation will occur [6].

It has already been mentioned that gradation is another important factor controlling the

resistance to permanent deformation. This is clearly shown in figure 33 which shows the

relationship between the vertical compressive strain at the top of the subgrade and the

number of load repetitions to 4% permanent deformation of the base layer. The figure clearly

indicates that the gradation plays an important role and that it is important to limit the amount

of variation in the gradation of the material. This asks for proper workmanship during

placement and compaction of granular base layers.

Figure 32. Effect of gradation and compaction level on the shear resistance of a granular

material [6].

Figure 33. Relation between the vertical compressive strain at the top of a base course and

the number of load repetitions to 4% permanent deformation and its dependency on

gradation.

Proper compaction also has a positive effect on the resilient modulus of unbound base layers;

this is perfectly shown in figure 34 which indicates that increasing the degree of compaction

from 97% to 105% results in an increase of resilient modulus with a factor of 1.5.

Figure 34. Effect of compaction on the resilient modulus of a granular material [6].

All in all the conclusion with respect to building perpetual unbound base courses is that, next

to gradation and moisture content, compaction is a key issue and every effort should be made

to get a high as possible and a as homogeneous as possible compaction level of the unbound

base. Intelligent compaction (figure 35 and figure 36) also plays in this case an important role

especially in getting an as homogeneously compacted base course as possible. But we should

keep in mind that using such compactors should always be combined by regular

measurements on density and moisture content.

Figure 35. Ways to get base courses with a high and homogeneous quality [7].

Figure 36. Example of the output generated by an intelligent compactor [7].

9. CONTRACTUAL ASPECTS

Contracts can be written in such a way that one will only get what is required as a minimum

or the contract is written in such a way that the producer will do everything to produce the

best. The traditional recipe based contracts are examples of contracts that will result in the

minimum needed while design-build-finance-maintain (DBFM) contracts are stimulants to

provide the best possible.

The only way, at least to the opinion of this author, to get the best out of a recipe based

contract is to extend the warrantee period. This however implies that the contract should be

performance based although recipe aspects can still be included. As mentioned before, the

warrantee period in the Netherlands has been extended from 3 to 7 years and it is carefully

described in those contracts with an extended warrantee what type of defects as well as their

severity and extent are allowed after 7 years. This implies that contract is in fact a

performance based contract. However, in order to ensure that the contractor is not proposing

“wild” solutions, recipe aspects still are part of the contract.

The best way, at least to the opinion of this author, to get “the best possible” is awarding

DBFM contracts which cover a period of e.g. 30 years. In the Netherlands such contracts are

given for large projects. The idea is that the contractor takes full responsibility for the design

and maintenance and has full freedom of choice in the materials and types of structures to be

used. In order to ensure that he “gets what he wants” the client evaluate the proposals on the

following aspects:

- hinder to traffic during the construction period;

- hinder to traffic during maintenance periods;

- environmental aspects.

The client has defined his “wishes” for each of the three aspects and when the proposal is not

fully complying with the wishes, an artificial amount of money is added to the bid. This

implies e.g. that proposals that include the use of environmentally hazardous products that

give a long life to the pavement will not have a chance to be accepted because they will score

very bad on the “environmental wish” although they might score good on the “traffic hinder

wishes”. The total bid is therefore the sum of the actual bid and the sums added to the bid

because of not completely fulfilling the “wishes”. The project will be awarded to the one

with the lowest total bid.

In order to ensure that the project will not become too costly, the client has set a maximum to

the amount of money he is willing to pay.

Furthermore the client has carefully specified the extent and severity of damage allowed.

In the contract the number of times the contractor is allowed to do maintenance and the

duration of the maintenance periods is specified. In cases maintenance has to be done at

moments which are not agreed upon in the contract (e.g. urgent repairs) the contractor has to

pay a “lane rental fee” which is in fact a penalty because the pavement is not performing as

agreed upon. These fees/penalties are very high as is shown in figure 35.

If however less maintenance is needed during the contract period than was anticipated, the

contractor receives a bonus because by realizing less maintenance needs he is reducing

hinder that traffic would experience because of maintenance works.

Finally the pavement should be at a certain condition at the end of the contractual period. If

these conditions levels are not met, then the contractor should pay himself for the works

needed to get the pavement to the desired condition.

Figure 37. Lane rental fees/penalties for performing not scheduled maintenance.

The effect these types of contracts have on the attitude of the contractor is enormous. His

attitude changed from re-active (I do what I am told to do) to pro-active (how can I fulfill in

the best way the wishes of my client). Furthermore contractors became immediately more

quality conscious and they became much more aware of the risks involved which they now

have to carry themselves instead of leaving them on the clients side. As a result of this,

contractors are spending much more money on research on arriving to products which have a

long life, are fail safe and are easy to apply.

This author believes that this type of contracts and this type of approach is a boast to

realizing perpetual pavements since this is in the own interest of the contractor.

10. CAN WE BUILD PERPETUAL PAVEMENTS

The title of this paper is “can we build perpetual pavements”. Now, at the end of the paper

we have come to the point that this question should be answered. From the material presented

in this paper it has become clear that a perpetual pavement can only be obtained if a lot of

attention is given to pavement construction. It has become clear that we need to achieve a

high and homogeneous quality of the pavement layers which implies a.o. the following.

Proper storage of the aggregates for asphalt mixtures including using sheds on the storage

bins to protect the aggregates from getting wet and using sufficient amounts of bins

allowing proper fractionizing of the aggregates.

Strict temperature control at the mixing plant avoiding too high and too low mixing

temperatures to occur.

Using shuttle buggies and intelligent compactors on site to ensure the asphalt mixture is

as homogeneous in temperature and composition when being laid and to ensure that an as

homogeneous compaction is achieved.

Selection of the right polymer modification for asphalt mixtures will certainly help in

getting a much longer life out of the asphalt layers.

Strict control of gradation, moisture content and using intelligent compactors to ensure a

high and constant quality of the granular base course.

Enforcing longer warrantee periods and setting up contracts in such a way that the

contractor is made much more responsible for the pavement as built.

Paying a bonus when the pavement is performing better than expected.

REFERENCES

1. Mohajeri, M.; Hot mix asphalt recycling. PhD thesis Delft University of Technology;

Delft, 2014 (to be published in December 2014).

2. Molenaar, A.A.A.; A.J.J. Meerkerk; M.Miradi and T. van der Steen. Performance of

porous asphalt concrete. Paper presented at the Annual Meeting of the Association of

Asphalt Technologists; Savannah, 2006.

3. Li, N.; Asphalt mixture fatigue testing; influence of test type and specimen size. PhD

thesis Delft University of Technology; Delft, 2013.

4. Miradi, M.; Knowledge discovery and pavement performance. PhD thesis Delft

University of Technology; Delft, 2009.

5. Molenaar, A.A.A.; E.J. Scholten; M.F.C. van de Ven; M. Poot and N. Li. SBS polymer

modified base course mixtures for heavy duty pavements. Paper presented at the 14th

AAPA Conference; Sydney, 2011.

6. Van Niekerk, A.A.; Mechanical behavior of granular bases and sub-bases. PhD thesis

Delft University of Technology; Delft, 2002.

7. Åkesson, F.; Dynapac compaction analyzer and optimizer. Power point presentation at

www.intelligentcompaction.com/downloads/Presentation/Akesson_Dynapac%20IC.pdf

The PhD theses of the Delft University can be downloaded via

www.citg.tudelft.nl/en/about-faculty/departments/structural-engineering/sections/pavement-engineering/publications/dissertations/