A review of fatigue damage in bituminous mixtures: Understanding the phenomenon from a new perspective

Construction and Building Materials 113 (2016) 927–938

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Review

A review of fatigue damage in bituminous mixtures: Understanding the phenomenon from a new perspective F. Moreno-Navarro ⇑, M.C. Rubio-Gámez Construction Engineering Laboratory of the University of Granada (LabIC.UGR), Granada, Spain

h i g h l i g h t s
Innovative fatigue phenomenon approach to study bituminous materials.
Molecular mobility could play a significant role in the appearance of fatigue damage in bituminous materials.
Various materials under different test conditions were performed through UGR-FACT.

a r t i c l e

i n f o

Article history: Received 19 June 2015 Received in revised form 23 March 2016 Accepted 23 March 2016

Keywords: Fatigue Bituminous mixtures Thixotropy Asphalt Permanent deformations Review

a b s t r a c t Fatigue cracking constitutes one of the main distresses responsible for the decline in the service life of asphalt pavements. The study of fatigue phenomena is therefore a field of research that has become crucially important for enhancing the durability of these structures. In spite of the advances achieved in the understanding of the fatigue phenomenon in bituminous materials, there remain some questions that are in need of further research. Firstly, the majority of studies do not consider the influence that permanent deformations can exert on the mechanical response of materials. Secondly, reversible phenomena that co-exist with damage during the development of fatigue processes make it difficult to accurately measure the latter. Further, given that the fatigue phenomenon has both global and local effects that cannot be dissociated, the analysis and failure criteria used could lead to non-homogenous results and incorrect fatigue life predictions. This research therefore constitutes a deeper examination of these issues and proposes a new approach that allows for a global analysis of the fatigue phenomenon. This approach has been tested through the study of various types of materials under different test conditions using the UGR-FACT device. Results have shown that using this approach it is possible to distinguish between the different phenomena that appear during cyclic loading and to establish a homogenous failure criterion. In addition, it has been demonstrated that molecular mobility could play a significant role in the appearance of fatigue damage in bituminous materials. Ó 2016 Elsevier Ltd. All rights reserved.

Contents 1. 2. 3. 4.

5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Previous considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Description of the new approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Validation of the new approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Materials and testing plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Analysis of the results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. Influence of molecular mobility on fatigue damage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. Comparison of the new approach with a traditional test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3. Sensitivity of the new approach to different types of materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

⇑ Corresponding author. E-mail addresses: [email protected] (F. Moreno-Navarro), [email protected] (M.C. Rubio-Gámez). http://dx.doi.org/10.1016/j.conbuildmat.2016.03.126 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.

928 929 931 933 933 934 934 935 935 936 937

928

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

1. Introduction Roads and highway pavements are designed to support the traffic loads and climatological events (rain, thermal changes, solar radiation, etc.) that they will be expected to endure during their service life. Traffic loads are cyclical and their magnitudes are considerably inferior to those that cause the breakage of the asphalt mixtures used to build them. Nevertheless, the repeated passage of these loads (combined with the effects caused by environmental agents) induce a fatigue process that leads to the appearance of cracks, which, in the long-term, is one of the main causes that can bring an end to the service life of a road. It is therefore important to develop materials that can offer greater resistance to this distress, and recent years have seen more studies conducted with the aim of offering a better understanding of the fatigue phenomenon that occurs in bituminous materials. Based on these studies, the fatigue that occurs in bituminous mixtures due to cyclic efforts can be considered as a global process (Fig. 1) which involves three main phenomena [1]: (i) accumulation of permanent deformations; (ii) reversible degradation (thixotropy) and initiation of irreversible damage (micro-cracks); (iii) crack propagation (the coalescence of micro-cracks produces the localization and propagation of macro-cracks). During the study of the fatigue behavior of asphalt mixtures, the occurrence of these phenomena can be identified by the changes produced in their mechanical properties (traditionally expressed through the changes produced in the phase angle and modulus [2]). Thus, the results obtained in a typical cyclic loading test can be divided into three stages (Fig. 2a, [3]): (1) a rapid decrease of the modulus and increase of the phase angle (which is related to the occurrence of plastic deformations, along with other viscoelastic reversible phenomena such as heating or thixotropy [1]); (2) a quasi-stationary stage where the changes produced in these parameters are small (due to the effect of the reversible phenomena and the initiation of the fatigue damage in the form of micro-cracks); (3) a rapid decrease of modulus and phase angle (due to the occurrence and propagation of the macro-crack). The study of the fatigue behavior of asphalt mixtures should therefore be approached as a global study that takes into account the developments and changes that asphalt materials suffer during the entire process. Nonetheless, as several authors have pointed out [4–8], this type of analysis is not easy to accomplish, and more research is needed in order to offer a better understanding of this phenomenon. One of these aspects is the effect caused by the permanent deformations that appear during the cyclic loading process, which in turn leads to fatigue. Whilst these deformations cannot be considered as fatigue damage, their appearance changes the viscoelastic properties of the material (making it more elastic and rigid, due

to the strain hardening phenomenon [9,10]), and therefore they can exert a significant influence on its mechanical response. Thus, when a controlled stress fatigue test is used, the initial decrease of the modulus can be largely due to the effect of permanent deformations, which can conceal the real damage produced by the fatigue process (Fig. 2b) [11]. Given this possibility, the majority of fatigue tests are conducted under controlled strain conditions (with the aim of avoiding the effects caused by permanent deformations when studying fatigue damage) [12–15]. However, this type of test does not reproduce the same load conditions that affect bituminous materials during their service life (a stress relaxation and fatigue process under constant strain occurs in the laboratory tests, whilst creep, strain hardening, and fatigue process occurs in the roads due to the presence of constant loads). Therefore, significant differences can be obtained between laboratory fatigue-life predictions and real fatigue lives [16,17], and between laboratory tests carried out at controlled stress or strain conditions [18]. Thus, despite the fact that permanent deformations do not cause damage, their effect on the mechanical properties of the bituminous mixture cannot be neglected when analysing fatigue processes. For this reason, it is necessary to use tests that take into account the influence of this phenomenon and its relationship with fatigue behavior. Another aspect that limits the analysis of fatigue in bituminous mixtures is the presence of other phenomena that co-exist with damage during cyclic loading (heating, thixotropy, etc.) [1,19,20]. Indeed, many studies have demonstrated that in the stages where these phenomena co-exist (stages 1 and 2, Fig. 2a), it is very difficult to distinguish which of them causes the changes in the mechanical properties of the material, and similarly, it is difficult to quantify the changes that are due only to real damage [21]. Other authors have stated that, due to the large and relatively sudden recoveries produced in the modulus during a short rest period [22–24], the observed recoveries cannot only be related to the healing of the damage (as it is not possible to produce such an amount of healing in such a short time period) [25,26]. These studies have shown that a considerable part of the loss in modulus that occurs during the first stages of a cyclic loading test is due to these reversible phenomena. During the loading process, thixotropy causes the bitumen to change progressively from a gel to a sol structure (ascribed to the dissociation and deformation of interand intra-molecular bonds), which reduces the viscosity and the modulus of the material (on the cessation of the loads, viscosity and modulus increase again) [27,28]. Heating is caused by molecular friction during the loading process; it produces chain separation due to thermal expansion and a consequent reduction of the secondary intermolecular forces (this reduces the modulus of the material, which is recovered when the loads disappear and the molecular temperature is restored) [29–31]. Therefore, the fatigue life calculated in traditional tests (where damage is measured

Fig. 1. Sketch of the global process due to the action of cyclic loading.

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

929

Fig. 2. (a) Typical results obtained in a strain-controlled fatigue four-point bending test [3]. (b) Main differences between strain and stress controlled tests.

Fig. 3. Sketch of the random propagation of fatigue damage that appeared in the bituminous mixtures.

through the loss in modulus without considering these non-fatigue related phenomena) could be over-predicted. Based on these considerations, it is clear that the influence of these phenomena should not be ignored when studying the fatigue behavior of bituminous mixtures and that it is very difficult to isolate them from real damage. It is therefore of interest to develop new analyses or test conditions where the impact of these phenomena can be reduced. Finally, another of the factors that hinders analysis of the fatigue response of bituminous mixtures is the selection of an accurate failure criterion. Fatigue damage begins with a micro-crack network (global damage), which develops randomly in the 3 dimensions of a certain volume of the material where the fatigue process takes place (Fig. 3). Following this, at a certain level of damage, the coalescence of these micro-cracks creates a macrocrack that propagates inside this volume of material until causing its total failure (localized damage). During the cyclic loading tests these damage phenomena co-exist with permanent deformations and visco-elastic phenomena (thixotropy, heating, etc.), and thus the amount of micro-cracks that generates the macro-crack or the propagation of global damage will directly depend on many variables [11,18] such as the type of bituminous mixture evaluated, the test conditions (amplitude, frequency, and temperature), the type of test used, or the geometry of the specimen. Due to this, it is very difficult to clearly separate and identify a damage limit that defines a homogenous failure criterion (global or local) that could be generalized to any fatigue test [12,18]. Consequently, in recent years many studies have focused on the definition of a criterion to minimize the influence of all these variables during the assessment of the fatigue behavior of bituminous materials [32,33]. Given these concerns, this research aims to analyze the phenomenon of fatigue in bituminous mixtures, with particular

emphasis on the major constraints that limit their study. For this purpose, a new approach (which combines the analysis of the geometrical changes and the energy dissipated by the material in each load cycle, from the appearance of permanent deformations to the macro-crack propagation) has been developed using the UGR-FACT (University of Granada-Fatigue Asphalt Cracking Test) method [34,35]. In this respect, different materials and tests conditions have been used in order to validate this new tool.

2. Previous considerations Bituminous mixtures are viscoelastic materials whose mechanical properties are highly dependent on the temperature of service and the applied loading rate [36–38]. In this respect, at high temperatures and under low loading rates (i.e. low frequencies), bituminous materials behave in a more viscous way (ductile fracture, offering high values of phase angle d) and they are susceptible to flow, which causes the appearance of plastic deformations (Fig. 4). In contrast, when the temperature of service is low or the applied load has a high frequency, these materials behave in a more elastic way (brittle fracture, with low values of phase angle d) and thus they have a greater capacity to support the stresses without flow. Based on the assumption of good adherence at the bitumen/ aggregate interface, it can be said that these responses are mainly governed by the molecular mobility of the bituminous binder [39–40], which is highly influenced by the temperature and the asphaltenes/maltenes relationship [41]. At low temperatures, the non-polar fractions of the bitumen (maltenes) cannot move and create a rigid network, whose mechanical response is controlled by the stretching and bending of intermolecular bonds [31], and thus by chain scission (brittle fracture). Therefore, as plastic

930

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

Fig. 4. Sketch of the viscoelastic response of bituminous mixtures.

Fig. 5. Schema of the effects caused by the stresses transmitted by the traffic over bituminous mixtures.

deformations are limited under these circumstances, the energy introduced in each load cycle is dissipated through fatigue damage (stress excites a main-chain bond to the state of separation), which propagates by means of molecular rupture [42]. Conversely, at higher temperatures, the separation of chains due to thermal expansion allows for the sliding of the molecular network [43,44], which causes the appearance of plastic deformations in the material. In this case, the energy of each load cycle is also dissipated by the maltenes through friction (thermal energy) and dissociation/deformation of molecular bonds (thixotropy) during plastic flow. After that (strain hardening phenomenon), the energy is stored in the bonds until the dislocation of micro-structural segments is avoided and molecular scission is produced (ductile fracture). Accordingly, for a given repetitive loading (such as produced due to vehicle circulation), if the molecules can move easily (as occurs at high temperatures), the mechanical energy introduced into the material due to the presence of such loading is consumed by molecular movements (which mainly produces non-recoverable deformations, and other secondary phenomena such as thixotropy). However, when the molecules have less mobility, as occurs at low temperatures, the same amount of mechanical energy cannot induce such phenomena and it is absorbed at molecular bonding level, producing stretching and bending of intermolecular bonds (‘‘small” recoverable deformations). Thus, the material absorbs this energy without producing deformations and showing

a more elastic response. In this respect, it is interesting to highlight that molecular mobility in bituminous binders not only depends on temperature, but is also influenced by other variables such as the load frequency or amplitude (thixotropy), as well as the asphaltenes and maltenes content [45]. Some authors state that under controlled-strain cyclic tests, viscoelastic materials with higher molecular mobility (those which are tested at higher temperatures, lower frequencies, or manufactured with bitumens with a weaker asphaltenes/maltenes relationship) are less susceptible to cracking and are therefore resistant to more load cycles than stiffer materials [18]. For the same reason, stiffer materials with reduced molecular mobility allow for a small decrease in modulus before fatigue failure, whilst ductile materials allow for a considerable decrease in modulus [11,16]. However, when controlled stress conditions are used, bituminous materials that are tested at high temperatures and low frequencies (with a higher molecular mobility) have a shorter fatigue life than those which are more rigid [46,47]. When a load is applied over a material (as occurs when the wheel of a vehicle passes over the pavement), it generates a stress that is initially homogenously distributed (Fig. 5). This stress induces a mechanical response in the material in the form of strain, which could be low in the case of elastic materials with low molecular mobility (these strains are considered to be recoverable as they are produced at molecular bonding level), or high in the case of viscous materials that are more susceptible to molecular

931

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

re-orientation (where the strains are considered to be permanent). Given that bituminous mixtures are visco-elastic materials (which behave more elastically or more viscously in relation to the frequency of the load or the temperature of service), the stress transmitted by the traffic loading will induce strains that could be produced by stretching and bending of intermolecular bonds (which are smaller and ‘‘recoverable”) or by molecular movements (which mainly produce permanent deformations and other secondary phenomena such as thixotropy). When the molecular mobility in the material is reduced (for example at low temperatures or when low penetration grade binders are used), the strains produced by traffic-induced stresses occur primarily at bonding level, until they cause molecular scission (appearance of damage in the material). As the broken bonds cannot support stresses, the neighboring bonds suffer higher strains thereby creating a zone with higher stresses (stress concentration). The bonds from this zone are more prone to breaking, creating an area of weakness in the material that will grow as the number of load cycles increases (propagation of damage in the material). In contrast, when conditions are such that molecular mobility in the material is favored, the strains induced by traffic stresses produce molecular re-orientation that is mainly manifest in the form of plastic deformations and other phenomena such as thixotropy (that do not cause damage in the material). In this case, as these deformations grow, they will create a phenomenon known as strain hardening [9,10], which changes the mechanical properties of the material throughout successive cycles, rendering it more rigid and elastic and reducing its molecular mobility. As the molecular mobility is reduced, the strains produced in each load cycle begin to appear at molecular bonding level (the molecules have already been re-oriented and its mobility is limited), and they cease to induce the appearance and propagation of damage in the material (molecular rupture). As seen in other materials such as metals [48,49], the strain hardening phenomenon prompts the appearance of zones where the stresses are concentrated and the energy introduced due to loading is dissipated on molecular scission, which will propagate until forming a crack that induces the failure of the material. Thus, the development of plastic deformations creates zones in the material that are susceptible to cracking under a given load (molecular dislocation is limited by the stretching or bending of intermolecular and main-chain bonds, and thus chain failure occurs). The addition of modifiers such as polymers (SBS, crumb rubber, etc.) to the asphalt binders could help to recover part of those plastic deformations [50], and therefore retard strain hardening, which in turn would retard the appearance of fatigue damage. It is therefore clear that plastic deformations and other molecular mobility phenomena (such as thixotropy) will have a considerable influence on the appearance of fatigue damage in bituminous mixtures. Some authors have already shown that creep strain energy should be considered during fatigue analysis, but it should be separated from the damage energy [51]. At high temperatures of service or low frequency loads (where molecular mobility is favored) the stresses transmitted by the traffic will produce more plastic deformations in the bituminous mixture. Thus, many concentrated zones of stress (due to strain hardening) and weak points susceptible to cracking would appear in the material (presenting a ductile fracture failure, which would be wide and ramified, Fig. 5). In contrast, at low temperatures or high frequencies, molecular mobility is very limited and the stresses induced by the traffic are firstly absorbed in stretching and bending of intermolecular bonds and are thereafter dissipated in the creation of damage due to molecular rupture (presenting a brittle fracture, which would be thin and un-ramified, Fig. 5). It is therefore critical to consider fatigue as a global process, the study of which should

take into account the influence of all the various phenomena that occur during cyclic loading. 3. Description of the new approach Based on the previous discussion, a new approach has been developed using the UGR-FACT method [34,52]. This methodological approach aims to evaluate the movements produced at molecular and bonding level, as well as the amount of damage produced due to breakage of molecular bonds. This test procedure reproduces the conditions that lead to the appearance of fatigue cracking in pavements (traffic loads and thermal gradients), by using a simple device composed of a sliding support (with a recovery spring), and two elastic elements under these support plates (rubber pads). The latter are reminiscent of both the bending and shear stresses commonly caused by traffic loading, and the tensile strains that are a consequence of thermal gradients (Fig. 6). Four LVDTs (one vertical and one horizontal in each side of the specimen) are used in order to control the vertical and horizontal displacements produced in the material in each load cycle (Fig. 6). Based on the measures taken, two different types of displacements can be observed in each direction (horizontal and vertical) and load cycle: a ‘‘permanent” displacement (hi, vi) that remains after the load cycle and is related to the non-recoverable deformations or the damage produced in the material; and a ‘‘relative” displacement (Hi, Vi) that is related to the consistency (stiffness) or damage state of the material in the given cycle (Fig. 7). Thus, if both types of displacements produced in each load cycle are used for a combined analysis, it is possible to conduct a precise evaluation of the evolution of the different phases appearing during the fatigue damage process. On the one hand, the hysteresis loop described for the ‘‘relative” displacements (Hi, Vi) produced in the material is used to define the dissipated energy in each load cycle, which is obtained as the addition of the dissipated energies calculated in the vertical and horizontal directions (Eq. (1)). These energies are obtained from the values of the areas inside the hysteresis loops. In this respect, the use of the areas are more accurate than the use of the absolute value of the ‘‘relative” displacements (which are commonly used to define parameters such as modulus), as the areas take into account the viscous and elastic nature of the material.

xi ¼ xhi þ xv i

ð1Þ 3

where xi is the dissipated energy in cycle i (in J/m ); xhi is the horizontally-dissipated energy in cycle i (in J/m3); and xvi is the vertically-dissipated energy in cycle i (in J/m3). On the other hand, the ‘‘permanent” displacements (hi, vi) can be used to define the variation of the geometry (Dei) of the material in the zone where the fatigue phenomenon takes place. This variation in the volume (measured in percentage) is calculated from the variations of the dimensions produced in the vertical and horizontal directions of the material (Eq. (2)).

Dei ¼

jð1 þ dhi Þ  ð1  dv i Þj  1  100 1

ð2Þ

where Dei is the variation of the geometry of the specimen in the cycle i for an initial unitary volume; dhi and dvi are the horizontal and vertical dimension changes measured in the specimen in the cycle i. Fig. 8 shows a typical graph obtained from the representation of these two parameters (xi in the x axis, and Dei in the y axis). As can be observed, the different stages of the fatigue process (plastic deformations, thixotropy, etc.; micro-damage; and macrodamage) can be clearly identified in the material. During the first part of the test, high variations are produced in the geometry of the material due to their molecular mobility, and these are reduced

932

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

Fig. 6. UGR-FACT test device.

Fig. 7. Outline of the efforts and displacements produced in the bituminous material during the UGR-FACT test.

Fig. 8. Example of the values obtained in the representation of De and load cycles as a function of the dissipated energy.

as the number of applied load cycles increases due to strain hardening. In spite of this considerable variation in the geometry of the material, the dissipated energy measured in each load cycle during this first part does not correspondingly change (the first part of the curve almost represents a vertical descent). This implies that the variations produced in the material due to the cyclic loads do not produce damage, as the levels of dissipated energy do not change considerably from the first cycle (which represents the viscoelastic response of the undamaged material). Based on these considerations, in the example shown in Fig. 8, the first 15,000 load cycles

do not cause fatigue damage in the material, rather they induce the appearance of plastic deformations and other phenomena such as thixotropy. The value of initial dissipated energy represents the molecular mobility capacity of the material evaluated under the test conditions used (frequency and temperature). As this initial dissipated energy decreases, the material shows a lower molecular mobility and therefore the stresses applied in each load cycle will be absorbed mainly in the form of strains at molecular bonding level (reducing the variations produced in the geometry of the specimen due to plastic deformations), until they generate

933

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938 Table 1 Properties of the mixtures studied. Properties

Bituminous mixtures

Type of Mineral Skeleton Coarse Aggregates Fine Aggregates Type of Bitumen Bitumen Content (% over the total weight of the mixture) Bulk Density (g/cm3, EN 12697-6 [55]) Air Voids (%, EN 12697-8 [56])) Marshall Stability (kN, EN 12697-34 [57]) Marshall Flow (mm, EN 12697-34 [57]) Stiffness at 20 °C, rise time value 124 ± 4 ms (MPa, EN 12697-26, C [58])

AC

BBTM-SBS

BBTM-CR

BBTM-B

AC Limestone Limestone B1 5.1 2.534 3.3 14.671 2.5 8937

BBTM Ophite Limestone PMB-SBS 4.8 2.464 5.2 9.301 3.9 3168

BBTM Ophite Limestone PMB-CR 4.8 2.469 5.1 8.473 3.1 4310

BBTM Ophite Limestone B2 4.8 2.496 4.8 8.662 3.3 4784

Table 2 Properties of the binders used in the manufacture of the mixtures. Property

Bituminous binders

Type of modifier Softening Point (°C, EN 1427 [59]) Penetration at 25 °C (mm/10, EN 1426 [60]) Fraass breaking point (°C, EN 12593 [61])

B1

PMB-SBS

PMB-CR

B2

– 66.4 22 12

SBS 68.4 62 17

Crumb Rubber 66 55.8 14

– 53.2 44 8

100

material passing (%)

90 80 70 60 50 40 30 20 10 0 0.01

0.1

1

10

100

Sieve (mm) AC

BBTM

Fig. 9. Grain size curves of the mineral skeletons used in the manufacture of the mixtures.

molecular breakage. In contrast, a high initial level of dissipated energy means that there is a high molecular mobility in the material. In this case, the stresses transmitted in each load cycle produce a rapid development of plastic deformations and thixotropy, which cause the appearance of strain hardening that will induce the ductile failure of the material. In the second part of the test (and due to the strain hardening phenomenon), the variations of the geometry of the material become very small (less than 0.1% in the example shown), and the energy introduced in each load cycle is absorbed mainly at bonding level (producing stretching and bending of intermolecular bonds). Thus, when the intermolecular bonds begin to fracture, the dissipated energy measured in each additional cycle begins to increase (Nmd, showing that the internal properties of the material are being altered). Strain hardening causes a reduction in the molecular mobility capacity and each additional load cycle does not produce plastic deformations or thixotropy – rather, they begin to produce micro damage due to molecular rupture. Because of this fact, the values measured of Dei are small (between 0.1 and 0.01% in the example shown), and the dissipated energy gradually increases. Finally, after a certain number of cycles, the dissipated energy begins to increase considerably by maintaining the changes produced in the

geometry (NMD, which means the initiation of the macro-crack due to the coalescence of micro-cracks), until a point where the values of Dei begin to increase again (due to the propagation of the macro crack), until total failure of the specimen is reached (Nf). This approach therefore permits accurate measurement of the propagation of the damage appearing in the material at the two levels: damage in the volume (micro-cracks) and localized damage (macro-crack). This information could be very useful for considering macro-damage in the estimation of fatigue life (which traditionally cannot be included because of its randomness and dispersion), or for the definition of a more accurate failure criterion (global or local) that could offer an analysis under the same level of damage (irrespective of the type of materials tested or the testing conditions used). In addition, this approach also allows for a distinction to be made between the various phenomena that occur during cyclic loading, and to identify which of these is responsible for real damage in the material. The real fatigue life (Nfl) of the material can thus be defined as Nfl = Nf  Nmd, as it marks the number of cycles from the initiation of damage until its total propagation throughout the specimen. Hence, the calculus of the mean damage parameter (which determines the susceptibility of the material to damage by cyclic loads [35]) must be obtained from the values measured during those cycles. 4. Validation of the new approach 4.1. Materials and testing plan In order to offer a representative analysis of the fatigue phenomenon in bituminous mixtures, and to assess the potential of the approach presented, different types of materials have been tested during this study. Four types of bituminous mixture have been evaluated: one AC mixture EN 13108-1 [53], and three BBTM mixtures EN 13108-2 [54]) with different types of aggregates (Limestone and Ophite), binders (neat binders of several penetration grade and modified binders with various modifiers) and mineral skeletons (continuous and gap-graded). The main characteristics of these materials are shown in Tables 1 and 2, and Fig. 9.

934

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

Table 3 Test procedure carried out in the study. Material tested

Test method

Test conditions

Number of specimens tested

AC

UGR-FACT

800 kPa; 5 Hz; 5 °C 800 kPa; 5 Hz; 15 °C 800 kPa; 5 Hz; 30 °C 200 le; 5 Hz; 5 °C 200 le; 5 Hz; 15 °C 200 le; 5 Hz; 30 °C

3 3 3 3 3 3

Fatigue Four-Point Bending Test BBTM-SBS

UGR-FACT

800 kPa; 5 Hz; 5 °C 800 kPa; 5 Hz; 15 °C

3 3

BBTM-CR

UGR-FACT

800 kPa; 5 Hz; 5 °C 800 kPa; 5 Hz; 15 °C

3 3

BBTM-B

UGR-FACT

800 kPa; 5 Hz; 5 °C 800 kPa; 5 Hz; 15 °C

3 3

mixture AC at different temperatures (which shows the varying levels of molecular mobility in the material). As can be observed, for the given test conditions, an increase in the test temperature produced an increase in variations of the geometry of the material during the early stage of the loading process. This occurs because the molecular mobility of bituminous materials increases as the temperature increases (offering a more viscous response, as indicated by the observed increase in the initial dissipated energy at the different temperatures). In this case, most of the energy introduced into the material in each cycle (which is provided by the constant stress loading, and is the same at all test temperatures) is consumed by the re-orientation of the molecules (mainly producing plastic deformations and other secondary phenomena such as thixotropy). In contrast, as the temperature decrease, the molecular mobility of the AC mixture was reduced, and therefore the stresses produced in each load cycle were mainly absorbed in small movements at molecular bonding level (the material is stiffer and behaves more elastically). If these stresses are not sufficiently high to produce movements that cause the rupture of molecular bonds, the material could support a higher amount of load cycles, as no energy is consumed in the creation of damage (the dissipated energy in each additional cycle does not increase, Fig. 10). The molecular re-orientation was in this case rather slow and limited, and the initial dissipated energy measured was therefore lower.

Fig. 10. Mean results of the AC mixture tested at different temperatures.

In order to analyze the influence of molecular mobility on the fatigue behavior of bituminous mixtures, the AC mixture was tested at different temperatures (5, 15 and 30 °C), under stresscontrolled conditions (stress amplitude of 800 kPa, and a frequency of 5 Hz) using the UGR-FACT method. These test conditions were selected in order to simulate the real stress conditions usually endured by the pavement [62,63], and high-speed traffic (around 100 km/h), by assuming a ‘‘vehicle type” with a mean distance between axes of 6.5 m [64]. The AC bituminous mixture was also used to compare the fatigue life obtained using the new approach of UGR-FACT method, and that obtained in a traditional fatigue test performed under strain-controlled conditions. For this purpose, the AC mixture was also tested using the four-point bending fatigue test (EN 12697-24, part D [65]) at the same temperatures and under a strain amplitude of 200 le (which is a common deflection registered in asphalt pavements [66]), using a failure criterion of a reduction of 30% of the initial modulus. Finally, the sensitivity of the new method to the type of materials tested and their molecular mobility was also analyzed using the BBTM mixtures manufactured with different types of binders (neat, modified with SBS polymers, and modified with crumb rubber). The test method and conditions used were the same as those described in the previous study (UGR-FACT with a load amplitude of 800 kPa and a frequency of 5 Hz), and the temperatures evaluated were 5 and 15 °C. As a summary, Table 3 show the testing plan carried out in this study. 4.2. Analysis of the results 4.2.1. Influence of molecular mobility on fatigue damage Fig. 10 displays the mean values obtained in the parameters xi and Dei from the various UGR-FACT tests conducted with the

Fig. 11. Definition of the characteristic cycles for the AC mixture at different temperatures: (a) 30 °C; (b) 15 °C; (c) 5 °C.

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

935

Fig. 12. Fatigue life observed for the AC mixture (a) at different temperatures under stress and strain controlled conditions; (b) for different initial dissipated energies under stress and strain controlled conditions.

Accordingly, permanent deformations that precede the appearance of damage in the material (molecular scission) were smaller, and following the fatigue process the fracture produced was brittle. This aspect can be observed in the tests conducted at a temperature of 5 °C, which were stopped at 2,000,000 cycles, and no signs of fatigue damage were found in the AC material (due to this, the dissipated energy remained constant). Fig. 11 shows the mean results obtained in the three AC mixture specimens tested at the different temperatures, as well as their characteristic cycles (Nmd, NMD, Nf). Based on them, it can be said that the specimens evaluated under temperature conditions where molecular mobility is favored, reached its fatigue failure (Nf) more rapidly (in accord with previous suggestions). In this respect, it is clear that once the molecules of the material have been reoriented (and plastic deformations and/or thixotropy are produced), the energy introduced in each additional load cycle is consumed by the production of bonding movements (stretching and bending of intermolecular bonds), that in the end will induce the creation of micro-damage (molecular rupture), and ultimately, the macro-crack propagation. Given this fact, it is worth pointing out that molecular mobility could play a significant role in the development of fatigue phenomena in bituminous mixtures, as no damage would appear whilst the molecules can be re-oriented. 4.2.2. Comparison of the new approach with a traditional test Traditional tests that study fatigue damage in bituminous mixtures are typically conducted under strain-controlled mode [1,2]. As these tests are based on the measurement of the modulus of the material, conducting them under controlled-stress mode could mean that the real damage produced by the fatigue process could be masked due to the undetected effect of permanent deformations and thixotropy. Thus, in order to limit such an effect, the tests are conducted by controlling the strain produced in each additional load cycle. Nonetheless, this load mode also limits the stresses that cause molecular mobility and bonding failure in the material. Thus, given that materials with greater molecular mobility (or which are tested under conditions that favor this molecular re-organization) can be easily deformed, the stresses produced in each load cycle to induce the controlled strain imposed on them are lower. Because of this fact, these types of materials can support a higher number of load cycles (as each load cycle applies a limited level of stress) and hence they would offer a longer fatigue life. Fig. 12a shows the fatigue life of the AC mixture as a function of temperature, when the material is tested under stress controlled conditions (using the UGR-FACT method and the Nfl values for each specimen) and under strain controlled conditions (using the fourpoint bending test and the values of the cycle when the initial modulus is reduced to 30%). As the material tested is the same (AC mixture), the resistance to fatigue damage afforded as a function of the temperature should be the same in both tests, irrespective of the control mode used. Nonetheless, it is observed that when tested under stress-control mode, the material exhibits a

completely opposite response to that observed when tested under strain controlled mode. This then raises the question of what exactly is the real mechanical response. It is clear that, if we take a prismatic specimen of a bituminous mixture and we try to break it into two pieces, it will be easier to do so as the temperature of the specimen increases. Therefore, it might be assumed that as the temperature increases, fewer load cycles of the same stress intensity are needed to cause the breakage of the specimen. Under strain-controlled conditions, the stress applied in each additional load cycle to produce the imposed strain is not constant, becoming smaller as the temperature increases, the stiffness of the material decreases, or the degradation produced in it increases. Thus, the energy introduced in each additional load cycle to produce the failure of the material also becomes smaller under these circumstances and the fatigue life is increased. In contrast, under stresscontrolled conditions, the stress applied remains constant in each additional load cycle, independently of the stiffness of the material tested or the test temperature (as occurs in a real pavements, where the stress applied by the trucks does not depend on the type of material or environmental conditions). Fig. 12b demonstrates that if the results of the tests are shown as a function of the initial dissipated energy [67], both tests (UGRFACT and Four-Point Bending) and control modes (stress and strain) display the same trend (as would be expected given that the material tested is the same). This fact confirms that an approach based on dissipated energy is compatible with the real mechanical performance of the bituminous mixture, regardless of the control mode used. Thus the new procedural approach presented in this paper, which is based on the changes produced in the geometry of the material and the dissipated energy in each load cycle (distinguishing these changes from the damage produced by other phenomena), can offer an accurate fatigue life prediction under more realistic test conditions (stress-controlled mode). 4.2.3. Sensitivity of the new approach to different types of materials Figs. 13 and 14 show the mean results obtained in the three specimens of the BBTM mixtures manufactured with different bituminous binders (polymer-modified, crumb rubber, and conventional), at 5 °C and 15 °C respectively. As observed in the previous tests conducted with the AC mixture, as the temperature decreases, the molecular mobility of the three bitumens also decreases (the initial dissipated energies offered by the mixtures at 5 °C are lower than at 15 °C), and therefore they can support a higher number of load cycles (all three offer a longer fatigue life at 5 °C than at 15 °C). These results demonstrate that the findings obtained for continuous mineral skeletons (AC) and conventional binders (B1) are also consistent for gap-graded mineral skeletons (BBTM), other types of neat binders (B2), and modified binders (PMB-SBS and PMB-CR). However, it has been observed that under the same test conditions (stress amplitude, frequency, and temperature), the lowest

936

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

Fig. 14. Mean results obtained in the three specimens tested of the BBTM mixtures at 15 °C: (a) BBTM-SBS; (b) BBTM-CR; (c) BBTM-B. Fig. 13. Mean results obtained in the three specimens tested of the BBTM mixtures at 5 °C: (a) BBTM-SBS; (b) BBTM-CR; (c) BBTM-B.

molecular mobility (measured from the initial dissipated energy) does not guarantee the longest fatigue life when comparing different materials. Fig. 13 shows that in spite of BBTM-SBS having the higher initial dissipated energy, it is the mixture that resists the highest number of load cycles. Similarly, the mixture manufactured with the crumb rubber modified bitumen has a higher initial dissipated energy than that manufactured with the neat binder, whilst also exhibiting a longer fatigue life. This effect could be due to the reversibility of the molecular movements and the absorption of part of the energy produced by the presence of the modifiers, which retards the appearance of plastic deformations (strain hardening phenomenon) and the propagation of the damage. Therefore, on the basis of these results, the novel approach presented here appears to be sensitive to the evaluation of different materials. This procedure could therefore provide an interesting tool for resolving some of the current problems associated with the analysis of fatigue damage in bituminous mixtures. It should be noted that as the response of the material becomes more viscous (for example, if it is tested at higher temperatures), it is more difficult to distinguish between the initial phase (where plastic deformations and thixotropy occur) and the microdamage phase. This is due to the co-existence of molecular mobility in one part of the specimen, and chemical bonding fracture in other parts (as the process of fatigue is not homogenous in all

sections of the specimen when a ductile fracture is produced). Therefore, it is suggested that real fatigue damage can be clearly identified at lower temperatures, where the co-existence of the various phenomena is less plausible due to the presence of a brittle fracture. Further, it has been demonstrated that regardless of the test temperature, this method is able to identify the proliferation of different types of damage as a function of the type of material tested. Materials of a higher stiffness (BBTM-CR or BBTM-B) offer a longer initial stage (due to them having lower molecular mobility). Materials modified with polymers offer a longer phase of micro-damage, as the elastic polymers absorb part of the energy introduced in each cycle without increasing the damage caused at chemical bonding level. Therefore, it is clear that fatigue damage analysis must identify not only the entire number of cycles, but also the number of cycles consumed by the various phenomena appearing in the materials. In this way, the analysis proposed by this approach – combining the variations in the geometry of the material and the energy dissipated in each load cycle – allows for a homogenous failure criterion that can consider separately any state of damage (micro or macro) produced in the specimen. 5. Conclusions This paper reviews the main factors that affect the study of fatigue in bituminous materials, and proposes a new methodological approach that could offer a valuable tool for resolving some of

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

the current difficulties associated with its analysis. Based on the results obtained in the present study (which involved the evaluation of different types of bituminous materials and fatigue test conditions), the following conclusions can be drawn: – Molecular mobility, which induces phenomena such as plastic deformations or thixotropy, is the initial response of bituminous materials during cyclic loading. Once the molecules have been re-oriented and its mobility is reduced, the energy introduced in each load cycle is consumed by molecular bonding level displacements that ultimately lead to their breakage. Therefore, if molecular mobility is favored, the appearance of fatigue damage is accelerated in the material under constant stress. Similarly, if modifiers such as elastomer are used to recover part of the molecular mobility, the fatigue damage can be delayed. – Phenomena related to molecular mobility such as plastic deformations or thixotropy that co-exist with damage, cannot be neglected during fatigue analysis. Fatigue tests conducted under strain-controlled conditions can limit the development of these phenomena (as they induce a stress relaxation process in the material, thereby limiting molecular mobility in any additional load cycle), which could in turn modify the real response of the bituminous materials. Due to this fact, materials tested under strain-controlled mode and under conditions that favor molecular mobility could considerably retard the appearance of damage and therefore offer a longer fatigue life. Nonetheless, these conditions are different to those that prevail during real traffic conditions, where the stresses transmitted to bituminous mixtures are not relaxed and each passing axel induces the reorientation of the molecules until they reach the appearance of damage at molecular bonding level. – It has been demonstrated that the development of the different stages of the fatigue process (different damage states reached by the specimens; i.e. phenomena based on molecular mobility, micro and macro-damage) depends on the characteristics of the material tested and the test conditions used. Thus, it is very difficult to define a homogenous failure criterion (either global or local) that could clearly separate and identify a damage limit to be used to compare the fatigue resistance of different materials. In this respect, it can be argued that the study of the variation produced in the geometry of the material seems to be an interesting tool to define each stage of the fatigue process, and therefore to establish certain limits that can be used to determine the fatigue life of different bituminous mixtures under similar circumstances. – The new approach presented in this paper, which combines the study of the changes produced in the geometry and the energy dissipated by the material in each load cycle, allows for the identification of the different phases appeared during fatigue loading conditions. Thus, it provides a more refined analysis of fatigue damage in bituminous materials, without the confounding effects of other phenomena that are not related to damage. Similarly, it is possible to establish a homogenous failure criterion that considers a similar level of damage in each material studied (even when the level of macro-damage level is taken into account). Finally, the new approach presented here appears to be sensitive to the variation of the test parameters and the type of materials evaluated.

References [1] H. Di Benedetto, Q.T. Nguyen, C. Sauzéat, Non-linearity, heating, fatigue and thixotropy during cyclic loading of asphalt mixtures, Road Mater. Pavement Des. 12 (1) (2011) 129–158.

937

[2] H. Di Benedetto, C. De La Roche, H. Baaj, A. Pronk, R. Lundstrom, Fatigue of bituminous mixtures, Mater. Struct. 37 (2004) 202–216. [3] F. Moreno-Navarro, Aplicación de ensayos de laboratorio en el estudio del comportamiento a fatiga de mezclas bituminosas, 2013 (Copicentro, ISBN 97884-15814-62-7). [4] S. Shen, S.H. Carpenter, Application of the dissipated energy concept in fatigue endurance limit testing, Transp. Res. Rec. 1929 (2005) 165–173. [5] S.H. Carpenter, S. Shen, Dissipated energy approach to study hot-mix asphalt healing in fatigue, Transp. Res. Rec. 1970 (2006) 178–185. [6] X. Shu, B. Huang, D. Vukosavljevic, Laboratory evaluation of fatigue characteristics of recycled asphalt mixture, Constr. Build. Mater. 22 (2008) 1323–1330. [7] C. Maggiore, J. Grenfell, G. Airey, A.C. Collop, Evaluation of fatigue life using dissipated energy methods, in: Proceedings of the 7th RILEM International Conference on Cracking in Pavements, Delft, 2012, pp. 643–652. [8] M. Pasetto, N. Baldo, Fatigue performance of asphalt concretes made with steel slags and modified bituminous binders, Int. J. Pavement Res. Technol. 6 (4) (2013) 294–303. [9] M.K. Darabi, R.K. Abu Al-Rub, E.A. Masad, D.N. Little, Cyclic hardeningrelaxation viscoplasticity model for asphalt concrete materials, J. Eng. Mech. 139 (7) (2013) 832–847. [10] T. Chang, W. Guo, Effects of strain hardening and stress state on fatigue crack closure, Int. J. Fatigue 21 (1999) 881–888. [11] R. Lundstrom, H. Di Benedetto, U. Iacsson, Influence of asphalt mixture stiffness on fatigue failure, J. Mater. Civ. Eng. 16 (2004) (516–515). [12] M. Pasetto, N. Baldo, Fatigue performance of asphalt concretes with RAP aggregates and steel slags, in: A. Scarpas et al. (Eds.), 7th RILEM International Conference on Cracking in Pavements, RILEM Bookseries, vol. 4, Springer, Netherlands, 2012, pp. 719–727. [13] M. Pasetto, N. Baldo, Fatigue characterization of asphalt rubber mixtures with steel slags, in: A. Scarpas et al. (Eds.), 7th RILEM International Conference on Cracking in Pavements, RILEM Bookseries, vol. 4, Springer, Netherlands, 2012, pp. 729–737. [14] H. Di Benedetto, C. de La Roche, H. Baaj, A. Pronk, R. Lundstron, Fatigue of bituminous mixtures: different approaches and RILEM group contribution, in: Proceedings of the 6th International RILEM Symposium, Zurich, Swiss, 2003. [15] I. Artamendi, H. Khalid, Characterization of fatigue damage for paving asphaltic materials, Fatigue Fract. Eng. Mater. Struct. 28 (2005) 1113–1118. [16] F. Pérez-Jimenez, R. Miró, A. Martínez, R. Botella, O. Reyes, G. Valdés, False failure in flexural fatigue tests, in: Pais (Ed.), Proceedings of the 2nd Workshop on Four Point Bending, University of Minho, Portugal, 2009. [17] R. Guo, Predicting in-service life of flexible pavements based on accelerated pavement testing (Doctoral dissertation), The University of Texas at Austin, USA, 2007. [18] K. Mollenhauer, M.P. Wistuba, Comparison of fatigue test procedures for asphalt mixes, in: Proceedings of the 5th Int. Conference of the European Asphalt Technology Association (EATA), Braunschweig, Germany, 2013. [19] F. Pérez-Jimenez, R. Botella, R. Miró, Differentiating between damage and thixotropy in asphalt binder’s fatigue tests, Constr. Build. Mater. 31 (2012) 212–219. [20] F. Pérez-Jimenez, R. Botella, R. Miró, Damage and thixotropy in asphalt mixture and binder fatigue tests, Transp. Res. Rec. 1 (2012) 8–17. [21] L. Shan, Y. Tan, B. Underwood, Y.R. Kim, Separation of thixotropy from fatigue process of asphalt binder, Transp. Res. Rec. 2207 (2011) 89–98. [22] D. Breysse, C. De la Roche, V. Domee, JJ Chauvin, Influence of rest time on recovery and damage during fatigue tests on bituminous composites, in: Proceedings of 6th RILEM Symposium PTEBM, Zurich, 2003. [23] A. Soltani, D.A. Anderson, New test protocol to measure fatigue damage in asphalt mixtures, Road Mater. Pavement Des. 6 (4) (2005) 485–514. [24] E. Santagata, O. Baglieri, L. Tsantilis, D. Dalmazzo, Evaluation of self healing properties of bituminous binders taking into account steric hardening effects, Constr. Build. Mater. 41 (2013) 60–67. [25] J.S. Daniel, Y.R. Kim, Laboratory evaluation of fatigue damage and healing of asphalt mixtures, J. Mater. Civ. Eng. 13 (6) (2001) 434–440. [26] D.N. Little, A. Bhasin, Exploring mechanisms of healing in asphalt mixtures and quantifying its impact. Self healing materials, an alternative approach to 20 centuries of materials science, Springer Ser. Mater. Sci. 100 (2007) 205–218. [27] D. Lesueur, J.F. Gérard, P. Claudy, J.M. Létoffé, D. Martin, J.P. Planche, Relationships between the structure and the properties of paving grade asphalt cements, J. Assoc. Asphalt Paving Technol. 66 (1997) 486–505. [28] L. Shan, Y. Tan, S. Underwood, Y.R. Kim, Application of thixotropy to analyse fatigue and healing characteristics of asphalt binder, J. Transp. Res. Board 2179 (2010) 85–92. [29] R.N. Traxler, C.E. Coombs, The colloidal nature of asphalt as shown by its flow properties, J. Phys. Chem. 44 (1935) 149–165. [30] A.B. Bazyleva, M.D.A. Hasan, M. Fulem, M. Becerra, J.M. Shaw, Bitumen and heavy oil rheological properties: reconciliation with viscosity measurements, J. Chem. Eng. Data 55 (3) (2010) 1389–1397. [31] L.H. Sperling, Introduction to Physical Polymer Science, John Wiley & Sons, New Jersey, 2006. [32] M. Abojaradeh, Development of fatigue failure criterion for hot-mix asphalt based on dissipated energy and stiffness ratio, Jordan J. Civ. Eng. 7 (1) (2013) 54–69. [33] N. Tapsoba, C. Sauzéat, H. Di Benedetto, Analysis of fatigue test for bituminous mixtures, J. Mater. Civ. Eng. 25 (6) (2013) 701–710.

938

F. Moreno-Navarro, M.C. Rubio-Gámez / Construction and Building Materials 113 (2016) 927–938

[34] F. Moreno-Navarro, M.C. Rubio-Gámez, UGR-FACT test for the study of fatigue cracking in bituminous mixes, Constr. Build. Mater. 43 (2013) 184–190. [35] F. Moreno-Navarro, M.C. Rubio-Gámez, Mean damage parameter for the characterization of fatigue cracking behavior in bituminous mixes, Mater. Des. 54 (2014) 748–754. [36] H.J. Lee, Y.R. Kim, Viscoelastic constitutive model for asphalt concrete under cyclic loading, J. Eng. Mech. 124 (1998) 32–40. [37] Bahia HU, H. Zhai, K. Bonnetti, S. Kose, Non-linear viscoelastic and fatigue properties of asphalt binders, in: Proceedings of the Annual Meeting of the Association of Asphalt Paving Technologists, USA, 1999. [38] Q.T. Nguyen, H. Di Benedetto, C. Sauzéat, Linear and nonlinear viscoelastic behaviour of bituminous mixtures, Mater. Struct. (2014), http://dx.doi.org/ 10.1617/s11527-014-0316-5. [39] D.A. Netzel, F.P. Miknis, K.P. Thomas, J.C. Wallace, C.H. Butcher, Molecular Motions and Rheological Properties of Asphalts: An NMR Study, Asphalt Science and Technology, Marcel Dekker, New York, USA, 1997, pp. 11–58. [40] D.A. Netzel, T.F. Turner, F.P. Miknis, J. Soule, A.E. Taylor, Molecular dynamics and the structure of asphalts and modified asphalts at low temperatures, Prepr. Pap. Am. Chem. Soc., Div. Fuel Chem. 41 (4) (1996) 1260–1266. [41] D.A. Netzel, Apparent activation energies for molecular motions in solid asphalt, Energy Fuels 20 (2006) 2181–2188. [42] B. Rosen, Fracture Processes in Polymeric Solids: Phenomena and Theory, Wiley, New York, 1964. [43] C.Y. Cheung, D. Cebon, Deformation mechanisms of pure bitumen, J. Mater. Civ. Eng. 9 (1997) 117–129. [44] J. Gaskin, On bitumen microstructure and the effects of crack healing, University of Nottingham, 2013 (Doctoral Thesis). [45] L. Zhang, M.L. Zhang, Relaxation time, diffusion and viscosity analysis of model asphalt system using molecular simulation, J. Chem. Phys. 127 (9) (2007), http://dx.doi.org/10.1063/1.2799189. [46] H. Baaj, H. Di Benedetto, P. Chaverot, Effect of binder characteristics on fatigue of asphalt pavements using an intrinsic damage approach, Road Mater. Pavement Des. 6 (2) (2005) 147–174. [47] H. Wu, B. Huang, X. Shu, Characterizing fatigue behavior of asphalt mixtures utilizing loaded wheel testers, J. Mater. Civ. Eng. 26 (1) (2014) 152–159. [48] M.V. Borodii, S.M. Shukaev, Additional cyclic strain hardening and its relation to material structure, mechanical characteristics, and lifetime, Int. J. Fatigue 29 (2007) 1184–1191. [49] M. Liakat, M.M. Khonsari, Entropic characterization of metal fatigue with stress concentration, Int. J. Fatigue 70 (2015) 223–234. [50] F. Moreno, M. Sol, J. Martín, M. Pérez, M.C. Rubio, The effect of crumb rubber modifier on resistance of asphalt mixes to plastic deformations, Mater. Des. 47 (2013) 274–280. [51] J. Kim, R. Roque, B. Birgisson, Interpreting dissipated energy from complex modulus data, Road Mater. Pavement Des. 7 (2) (2006) 223–245. [52] F. Moreno-Navarro, M. Sol-Sánchez, M.C. Rubio-Gámez, Exploring the recovery of fatigue damage in bituminous mixtures: the role of healing, Road Mater. Pavement Des. (2015), http://dx.doi.org/10.1080/14680629.2015.1029706.

[53] EN 13108-1: Bituminous mixtures, Material Specifications – Part 1: Asphalt Concrete, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2008. [54] EN 13108-2: Bituminous mixtures, Material Specifications – Part 2: Asphalt Concrete for Very Thin Layers, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2007. [55] EN 12697-6: Bituminous mixtures, Test Methods for Hot Mix Asphalt – Part 6: Determination of Bulk Density of Bituminous Specimens, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2012. [56] EN 12697-8: Bituminous mixtures, Test Methods for Hot Mix Asphalt – Part 8: Determination of Void Characteristics of Bituminous Specimens, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2003. [57] EN 12697-34: Bituminous mixtures, Test Methods for Hot Mix Asphalt – Part 34: Marshall Test, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2013. [58] EN 12697-26, annex C: Bituminous mixtures, Test Methods for Hot Mix Asphalt – Part 26: Stiffness, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2012. [59] EN 1427: Bitumen and bituminous binders, Determination of the Softening Point – Ring and Ball Method, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2007. [60] EN 1426: Bitumen and bituminous binders, Determination of Needle Penetration, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2007. [61] EN 12593: Bitumen and bituminous binders, Determination of the Fraass Breaking Point, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2007. [62] M. De Beer, C. Fisher, L. Kannemeyer, Tyre pavement interface contact stresses on flexible pavements – quo vadis?, in: Roads the Arteries of Africa: 8th Conference on Asphalt Pavements for Southern Africa, Sun City, North West Province, South Africa, 2004 [63] L.A. Myers, R. Roque, B.E. Ruth, C. Drakos, Measurement of contact stresses of different truck tire types to evaluate their influence on near-surface cracking and rutting, Transp. Res. Rec. 1388 (1991) 108–119. [64] F. Moreno-Navarro, Design of a laboratory test method for the analysis of asphalt mix response to fatigue cracking (Ph.D. thesis), University of Granada, Spain, 2013. [65] EN 12697-24, annex D: Bituminous mixtures, Test Methods for Hot Mix Asphalt – Part 24: Resistance to Fatigue, AENOR, Asociación Española de Normalización y Certificación, Madrid, 2012. [66] Consejería de Obras Públicas y Transportes de la Junta de Andalucía. Instrucción para el diseño de firmes de la red de carreteras de Andalucía. N° de Registro: JAOP/GIASA-47-2007, Sevilla, 2007, p. 158. [67] G.M. Rowe, Application of the dissipated energy concept to fatigue cracking in asphalt pavements (Ph.D. thesis), Department of Civil Engineering, University of Nottingham, UK, 1996.