FATIGUE BEHAVIOUR OF FIBRE REINFORCED BITUMINOUS MIXTURES FROM INDIRECT TENSILE FATIGUE TEST

FATIGUE BEHAVIOUR OF FIBRE REINFORCED BITUMINOUS MIXTURES FROM INDIRECT TENSILE FATIGUE TEST

FATIGUE BEHAVIOUR OF FIBRE REINFORCED BITUMINOUS MIXTURES FROM INDIRECT TENSILE FATIGUE TEST

Ibrahim Kamaruddin Universiti Teknologi Petronas, Malaysia

ABSTRACT

Fatigue behaviour of bituminous mixtures is characterised from the relationship between stress or strain level and the number of load repetitions to failure This paper presents the results of a laboratory investigation to assess the influence of polymer fibres on the fatigue characteristics of Hot-Rolled Asphalt (HRA) mixtures Polyester and polypropylene are the two types of fibres added to the bituminous mixtures The fibrous HRA are subjected to two types of laboratory fatigue tests; indirect tensile and the beam flexural test A constant stress test regime was adopted with all the mixtures tested under identical conditions of applied stress, frequency of loading, temperature and support conditions Fatigue equations obtained from both tests are presented It appears that the incorporation of synthetic fibres in bituminous mixtures improves the fatigue performance of the mix despite the higher void content of the specimens incorporating the fibres This takes place despite the fact that fatigue performance generally deteriorates when the void content is increased The higher strain capacity

of the fibre-modified mixtures is attributed to the higher bitumen content and the thicker bitumen film coating the aggregates Comparison of the fatigue relationship obtained from the Indirect Tensile Fatigue test and that of the Beam Flexural test revealed that the fatigue lines for both tests do result in approximately a single line defining the fatigue behaviour of the mixtures tested

Keywords: Bitumen, Bituminous Mixtures, Fatigue, Hot-Rolled Asphalt, Polypropylene and Polyester Fibres

1.0 INTRODUCTION

Excessive permanent deformation and cracking are generally accepted as the main forms of distress in bituminous road pavements While permanent deformation occurs predominantly at elevated temperatures, thermal cracking is normally a low-temperature phenomenon

In addition to temperature, cracking can also be brought about by traffic loading Load associated fatigue cracking is the phenomenon of fracture as a result of repeated or fluctuating stresses brought about by the traffic loads Traffic loads can cause a pavement structure to flex and the maximum tensile strain will occur at the base of the bituminous layer

Cracking occurs when the thermally induced tensile stresses, together with stresses caused by traffic, exceeds the tensile strength of the material If the structure is inadequate for the imposed loading conditions, the tensile strength of the material will

be exceeded and cracks are likely to initiate, which will be manifested as cracks on the surface of the pavement As a result, it is generally assumed that there is a significant reduction in the load distribution capacity within the pavement

This paper describes and presents the results of a laboratory investigation to assess the influence of polymer fibres on the fatigue characteristics of Hot-Rolled Asphalt (HRA) mixtures

2.0 MATERIALS USED IN THE INVESTIGATION

2.0.1 Mineral Aggregates, Filler and Bitumen

Limestone aggregates and Ordinary Portland Cement (OPC) filler and a binder of nominal penetration 50 were used Some relevant properties of these materials are shown in Table 1

Table 1: Properties of the Mineral Aggregates, Filler and Bitumen Used in the Study

BY WEIGHT (%)

RELATIVE DENSITY

ABSORPTION (%)

BS SPECIFICATION

Coarse

BS 594: Part 1:1992 Table 3, Type F Wearing Coarse Designation 30/14

Fine Aggregate

Bitumen Penetration (dmm)

Softening Point (oC)

Penetration Index (PI) BS 4699:1985

2.0.2 Synthetic Fibres

Two types of synthetic polypropylene and polyester fibres were used in this study The fibres were used as a partial replacement of the filler, on an equal volume basis,

at two different concentrations of 0.5% and 1 % filler/bitumen ratio by weight of mix The fibres; in chopped form; were the by-products of the textile industry and thus their potential use was desirable on environmental grounds Some characteristics of the fibres used are shown in Table 2 In order to maintain thermal stability when using the polypropylene fibres, it was decided that the mixing temperature during the preparation of the HRA mixtures should not exceed 140°C and compaction be done at

130°C

Table 2: Characteristics of Fibres Used

FIBRE

TYPE

SPECIFIC

LENGTH (mm)

AVERAGE DIAMETER (µµµµm)

DEGRADATION TEMPERATURE ( o C)

Polypropylene

Polyester

*Values obtained from 20 readings using a light microscope at 400x magnification

3.0 FATIGUE RELATIONSHIP

Fatigue tests can be carried out in two principal methods, namely the constant stress tests where the stress level is kept constant throughout the test and the constant strain tests where the magnitude of the peak cyclic strain is kept constant throughout the

test Before the fatigue performance of a bituminous material can be assessed, the failure of the specimen tested must be consistently defined

Defining the failure criterion in the constant stress mode is relatively easy as the specimens undergo a relatively short crack propagation period Hence, the failure point is taken as when the specimen has completely failed However, in the constant strain mode of loading, the failure point is not very well defined, due to the large amount of crack propagation included in the test An arbitrary point of failure must thus be assumed which is normally defined as the point when the specimen has reached a reduction in its initial stiffness of 50% or in practical terms is the point when the stress applied has been halved to achieve a constant strain

A linear relationship exists between the log of stress σ, or strain ε, and the log of the

number of load repetitions, N f to failure The failure criteria can therefore be expressed as:

Log stress against log load applications

Log strain against log load applications

This can be written in the form:

Log (σ or ε) = a + b log N f

For the strain controlled tests, the results are normally presented in the form

N f =A

b













ε 1

while in the stress controlled tests, the results are presented in the form

N f =A

d

σ









 1

where N f = Number of load applications to failure

σ, ε = Tensile strain or stress repeatedly applied load

A,b,C,d = Material coefficients

4.0 INDIRECT TENSILE FATIGUE TEST (ITFT)

The indirect tensile fatigue test was employed in order to characterize the fatigue behaviour of the HRA specimens in the laboratory The test was conducted in the Universal Material Testing Apparatus (UMATTA) with the appropriate force, rise time and pulse repetition period being selected For each loading pulse, the accumulated displacements were continuously being calculated and displayed Read and Brown (1994) were of the opinion that the Indirect Tensile Fatigue Test is able to characterize the fatigue life of a bituminous mixture by testing a small number of specimens (less than 10) at high temperatures (in excess of 25oC) and at high stress levels (greater than 450 kPa) This means that the fatigue testing time needed to produce an adequate fatigue relationship for a bituminous material is significantly shorter than other traditional laboratory fatigue testing methods

In the indirect tensile fatigue test, a repeated line of loading (generally controlled stress) is applied along the vertical diameter of a cylindrical specimen The vertical

load produces both a vertical compressive stress and a horizontal tensile stress on the diameters of the tested specimen The magnitude of the stresses changes along the diameter of the specimen with a maximum occurring at the center Assuming that the specimen is homogeneous, isotropic, behaves in a linear elastic manner and subjected

to a plane stress condition; the stress conditions in the specimen can be calculated for

a known Poisson's ratio ν when the force, P applied is a line load

σxmax =

dt

P

π

2

(Equation 1.0)

σymax =

dt

P

π

6

(Equation 1.1)

where: σxmax = Maximum horizontal tensile stress at the center of the specimen

σymax = Maximum vertical compressive stress at the center of the specimen

d = Specimen diameter

t = Specimen thickness

By simple linear elastic stress analysis (Hooke's Law):

εxmax =

m

x S

max σ

=

m

y S

where: εxmax = Maximum initial horizontal tensile strain at the center of the specimen

v = Poisson's ratio

S m = Stiffness modulus of the specimen

By substitution;

εxmax =

m

x S

max

Equation (1.3) is used for the calculation of the maximum tensile strain (εxmax) at the centre of the specimen that is dependent on the stiffness modulus of the material This parameter was obtained from the Indirect Tensile Stiffness Modulus test at the same stress level and test temperature as the ITFT

The specimen geometry selected for all the ITFT was 100 mm in diameter by 40 mm thick The specimen was first conditioned at the test temperature of 20+0.5oC for between 2-4 hours before testing commenced A cyclic load pulse was applied to the specimen with a time to the peak of the load pulse being 120 ms A loading rate of 40+1 pulse/minute at the test stress level were employed and the permanent vertical deformation measured by a linear variable differential transducer (LVDT) Figure 1 relates the strain-number of cycles to failure that was typical of the output obtained on the UMATTA during the test which shows an initial period of comparatively large displacement followed by a part that represents a constant rate of strain amplitude Finally the curve started to concave indicating the point of failure

Figure 1 Typical Indirect Tensile Fatigue Test Output from MATIA Testing Machine Nikolaides (1997) defined the point of failure in the ITFT as the point where the straight line obtained between the number of loading cycles to transient deformation started to concave In this test however, the point of failure was taken as the point on the curve that indicated that the test specimen had ruptured

4.1 Discussion of Results

The fatigue characteristics of the control and the fibre incorporated mixes were tested

in the Indirect Tensile Fatigue Test at their optimum bitumen content The number of specimens tested for each mix type in order to determine its fatigue characteristics ranges from 12-15 specimens (BS DD ABF/95; British Standard Institution, 1993)

Table 3 presents a summary of the fatigue characteristics of the mixes that were tested

at the optimum bitumen content For the control mix, the life at 100 microstrains was 172,232 load applications and the strain at 106 cycles was 52 microstrains This is considerable poorer than the fibre reinforced mixes where the life at 100 microstrains varied from 700,026 load applications (0.5PP) to 2,589,372 load applications (1POL) The results from the ITFT also showed that the 1% fibre mixes have a better fatigue behaviour than those of the 0.5% fibre mixes The higher bitumen content of the 1% fibre mixes could possibly have been responsible for the superior fatigue characteristics of these mixes

The polyester (POL) mixes also exhibited superior fatigue properties as compared to that of the polypropylene (PP) mixes This may be due to the higher bitumen content

in the polyester mixes as compared to the polypropylene mixes In addition, the higher viscosity of the polyester incorporated binder resulted in harder bitumen thus responsible for greater stiffness of the polyester fibre mixes The Indirect Tensile Fatigue test (ITFT) ranked the fatigue behaviour of the mixes as follows:

1) 1%POL 2) 1%PP 3) 0.5%POL 4) 0.5%PP 5) Control

Table 3: Summary of Fatigue in the ITFT at Optimum Bitumen Content

Mix Type Equation for

Strain

Strain @

106 cycles

Equation for Cycles

to Failure

Cycles @

100 microstrains Control ε = 8751.3Nf-0.3709 52 Nf = 4.251x1010(1/ ε )2.696 172,382 0.5PP ε = 12596N f-0.3593 88 Nf = 2.577x1011(1/ ε )2.783 700,026 0.5POL ε = 8912Nf-0.3303 93 Nf = 9.095x1011(1/ ε )3.028 799,471 1PP ε = 6238.7N f-0.2922 110 Nf = 9.728x1012(1/ ε )3.422 1,393,232 1POL ε = 7158.9Nf-0.2892 132 Nf = 2.134x1013(1/ ε )3.458 2,589,372

5.0 FLEXURAL BEAM FATIGUE TEST

The beam test that was used in this study is essentially a simple arrangement and does not simulate the more complicated stress regime at the crack tip that occurs during a load pulse caused by traffic loading Despite this limitation, the test produced useful data in showing the benefit, or otherwise, of reinforcing bituminous mixtures with the synthetic fibres

The bituminous mixtures for the fabrication of the beams were prepared in the laboratory The amount of materials required to produce a 500x100x100 mm beam that gave similar density to that of the mix compacted with the Gyratory Testing Machine (GTM) were calculated and prepared The aggregates were heated to 140oC

in the oven, the desired amount of bitumen, also preheated to 140oC, was added to the aggregate in the mixer Each tray of mix was returned to the oven after mixing before they were placed into the moulds and compacted

The moulds were themselves kept in the oven at the appropriate compaction temperature Compaction of the bituminous material was done using a Kango 638 vibrating compactor with a 98x98x20 mm tamping foot moving sequentially along the length of the mould The material was compacted in three layers, the number of passes and tamping duration would determine the resulting mix density and porosity The kneading type action produced by the tamping foot is felt to give the specimen an aggregate orientation similar to that developed by the GTM in the laboratory and in the field under the action of a roller compactor Prior to testing, the beams were cut to size using a 'Kipper' concrete cutter

5.1 Arrangements and Condition for Testing

Figure 2 shows the general experimental arrangement used in this study Once the beams were cast, they were placed onto two plywood sheets 18 mm thick, with a 10

mm gap in between to act as a crack initiator The gap simulated an existing crack or joint in a pavement that needed overlaying A 10 mm metal spacer was used to set the gap in the experimental set-up and was removed just before testing begun This gap would induce any cracks that were envisaged to occur as the beam was being subjected to the cyclic loads

Figure 2 Details of Crack Initiator and how Horizontal Movement across Crack

Recorded Figure 3 shows how the horizontal movement across the crack was recorded A linear variable differential transducer (LVDT) was placed 20 mm from the bottom edge of the beam to be tested The LVDT was held in position by a brass holder with the tip

of the LVDT resting against an L-shaped bracket Both the brass holder and the bracket were glued to the surface of the HRA beam using araldite glue that positions the LVDT during the test

Figure 3 General Arrangement for Reflection Cracking Test

Each of the beams tested rested on a rubber pad 25 mm thick with a resilient modulus

of 43 MPa The rubber provided a resilient support, simulating the response encountered in real road pavements, allowing the bituminous beam to flex under the action of the applied cyclic loading Cracks on both sides of the beam were monitored visually as the vertical distance from the base of the bituminous beam layer The sides of the beams were painted with white emulsion to facilitate this observation In order to account for any displacements at the end of the beam during the test, a strapping tape was used on both ends of the beam to ensure that this displacement was minimized

5.2 Test Equipment

The testing machine used in this study was a Servo-hydraulic Universal testing machine with a waveform generator and a data logging facility that can be controlled

on load, position or strain It has a 200 kN minimum load capacity and is capable of producing a repeated half sine shaped loads on the beam specimens The test rig comprised of a 50 ton spring jack and a 50 ton load cell

A haversine cyclic load was applied to the specimen through a rubber loading platen measuring 40 mm wide and 1 10 mm in length attached to the load cell at a frequency

of 4 Hz The load was cycled between a maximum and a minimum peak load and monitored using an internal load cell of the testing machine with the results displayed

on a digital panel meter Testing of the beams was done at ambient temperature ranging from 17-220C

6.0 INTERPRETATION OF BEAM FATIGUE TEST RESULTS

In the laboratory fatigue tests, the prime factor controlling the onset of cracking in the bituminous material is the maximum dynamic tensile strain in the specimens The tensile strain therefore seems to be the most logical parameter to be used for the purpose of analysing the fatigue behaviour of the specimens However, difficulties arise in defining a satisfactory mean value of tensile strain that would reflect the mean level during all or the major part of the life of the specimens Because the best correlation with fatigue life is obtained with the tensile strain at the beginning of the test (Goddard et al, 1978), this factor is similarly adopted in describing the fatigue behaviour of the specimen tested

Bjorklund's (1985) definition of fatigue failure was adopted in all the tests carried out Figure 4 shows a plot relating the tensile strain and the number of cycles to failure A linear part of the curve representing an initial period of large strain build-up or accumulation and the part that represents a constant rate of increase in strain amplitude was extrapolated The strain value corresponding to the intersection of these two extrapolations is defined as the initial tensile strain The same principle was applied in determining the number of cycles to failure as shown in the figure The laboratory fatigue performance of the mixes may then be characterized by the relationship between the initial tensile strain, c and the number of load cycles to

failure N f When plotted on a log-log basis, this takes the form:

logε = a - blogN f (Equation 1.5)

which can be expressed as:

Nf = K

n











 ε

1

(Equation 1.6)

where: N f = Fatigue life of the beam given in number of cycles to failure

ε = Initial strain (in microstrains)

K = Material constant which determines the position of the line (= 10a/b)

n = Material constant and is the slope factor of the fatigue line (=l/b)

Regression analysis of the test data will yield the values of K and n For a given

beam, the stress amplitude at the location where the strains amplitude are measured are calculated from the following:

I

My

=

where: M = Bending moment

y = Location of the LVDT from the centroid of the beam

I = moment of inertia of the beam 













=

12

3

bd I

Figure 4 Initial Strain and Number of Cycles to Failure Determination Following

Bjorklund’s Definition of Fatigue Failure (After Napiah, 1994)

Equation (1.7) developed from beam theory is based on the assumption that the material is homogeneous, sections that are plane before loading remain plane after loading and the stress-strain relationship is linear for the loading considered For a

beam on an elastic rubber foundation acted upon by a concentrated load P, the bending moment M was calculated from Hetenyi's (1946) equation:

25 0

4











=

EI k

P

where: k = Modulus of rubber foundation (=4.3MPa)

E = Beam stiffness modulus = Smix (MPa)

I = Moment of inertia of rectangular beam

Using the BANDS computer program (Shell Bitumen, 1990), Smix, was calculated by keying in values of Sbit volume of binder and volume of aggregate Sbit was also determined from the first part of the BANDS program based on the penetration index and softening point These in turn depend on the bitumen type, temperature (20oC) and time of loading (0.25 sec, based on a 4 Hz frequency of loading)

7.0 CRACK PROPAGATION IN FLEXURAL FATIGUE BEAMS

In order to inhibit or reduce the propagation of cracks, the bituminous beam must be capable of absorbing the high level of strains that are imposed on it through the fatigue test Earlier work (Kamaruddin I., 1998) on the indirect tensile tests have shown that the addition of fibres in the HRA mixtures caused a reduction in tensile strength of the mix and an increase in tensile strain (elongation) at failure as a result

of the added bitumen in the mix This resulted in a more flexible bituminous mix as shown by the higher toughness and energy that were characteristic of the fibre mixes This combination of properties may mean that more energy is required to produce a crack in the fibre-modified mixes as compared to that of the control It is also

significant to note the reduction in severity of the crack in the fibre-reinforced mixes when compared with that of the control A tendency for some hairline cracks to heal after testing was also noticed in the fibre modified beams that were not apparent in the control hewn Figure 5 shows the typical crack patterns that were observed from the beam fatigue tests for beams without the fibres (control mix) and those reinforced with fibres

Figure 5 Restriction of crack propagation by fibers (after Tons and Krokosky, 1963) Specimens containing the fibres cracked over a wider area than those without fibres, demonstrating the load spreading ability of the fibres in the mix It is conceivable that the fibres can act as crack arresters in the bituminous mixtures As the crack propagates, the tip of the crack is deflected by the presence of the fibres as shown in Figure 6 (Tons and Krokosky, 1963) The synthetic fibres because of their shape tend

to be more effective because they present a larger surface area to deflect the cracks The main conclusion drawn from the beam reflection cracking test was the ability of the fibre mixes to reduce the propagation and severity of the crack through the bituminous beam

Figure 6 Typical Cracking Patterns of Overlay Test Specimen With and Without

Fibers