A Mechanistic–Empirical Model for Top–Down Cracking of Asphalt Pavements Layers (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 CHAPTER 2. SYNTHESIS OF CURRENT KNOWLEDGE Top-down cracking of asphalt pavements initiates at pavement surface and propagates downward through the asphalt layer. It has been identified worldwide, such as the United States (1-4), Canada (5), United Kingdom (6), South Africa (7), France (8), the Netherlands (9), and Japan (10). Top-down cracking usually develops in the longitudinal direction both within the wheel path and outside of the wheel path. Top-down cracks that develop at different locations on the pavement surface may have different mechanisms. According to its mechanism, top-down cracking can be grouped into two categories: 1) construction-related top-down cracking; and 2) load-related top-down cracking. The construction-related top-down cracking results from the segregation of aggregates in the asphalt mixture during construction. Three locations with construction-related top-down cracking have been identified: on the outside edges of the two slat conveyors in the laydown machine, and under the gearbox between the two slat conveyors (2-3). The load-related top-down cracking usually develops in the wheel paths, produced by the non- uniform three-dimensional tire-pavement contact stresses and accelerated by thermal stresses and mixture aging effects. This project has focused on the load-related top-down cracking only. Load-related top-down cracking is significantly affected by numerous factors induced by the material itself, the pavement structure, the traffic, and the climate. A comprehensive literature review is conducted to identify these factors as well as existing models to simulate each factor. These factors must be considered when developing top-down cracking models in order to predict accurately the growth of top-down cracks. Figure 2.1 lists the identified major items responsible for the development of top-down cracking in four categories: material, structure, traffic, and climate. Figure 2.1. Major Factors Affecting Top-Down Cracking in Asphalt Pavements Top‐Down Cracking Material •Mixture composition •Modulus variation •Fracture properties •Thermal properties Structure •Thin/thick asphalt layer •Stabilized/unstablized base course Traffic •Load mechanism •Load magnitude & distribution •Load spectrum Climate •Pavement temperature •Aging •Thermal stress

5 Based on the factors identified from the literature search, a complete top-down cracking model should include the following submodels to accurately predict the growth of top-down cracks:  Mixture material property models  Mixture aging models  Traffic traction stress and strain models  Traffic load spectrum models  Thermal stress models  Pavement temperature models  Crack initiation models  Crack propagation models  Finite element crack growth models  Artificial Neural Networks (ANN) models  Cumulative damage models  Observed geometry of a crack as it propagates downward They are discussed in more detail as follows. Mixture Material Property Models The mixture material property models documented in 22 relevant journal articles and reports are reviewed. In general, the mixture material property models can be categorized into: 1) Effects of mixture composition; 2) Effects of fracture properties; and 3) Effects of environmental conditions. The mixture composition refers to the asphalt binder, asphalt mastic, aggregates, and air voids. It is found that air voids play an important role in top-down cracking since high air void content induces age hardening and thus the likelihood of cracking (11-14). In addition, the binder content, aggregate gradation, and binder-aggregate adhesion also affect the initiation of top- down cracking (15-16). The fracture properties measured to study top-down cracking refers to the properties necessary to evaluate the susceptibility of asphalt mixtures to top-down cracking. The examples include the crack growth rate (17), tensile strength (18-19), and low temperature cracking characteristics (20-22). In all of these models, the viscoelastic properties of asphalt mixtures, e.g. dynamic modulus or creep compliance, are required to study cracking behaviors of the material. The effects of environmental conditions refer to the influence of temperature and moisture, which are considered in studying top-down cracking by adding mixture temperatures (10, 15, 23-24), thermal properties (25-28), or moisture damage properties (29). Mixture Aging Models The mixture aging models documented in 13 relevant journal articles and reports are reviewed. Aging causes the asphalt mixtures near the pavement surface become brittle and consequently they are more prone to cracking. A common way to consider aging is to measure the effect of aging on the material properties of asphalt mixtures (e.g. modulus, tensile strength)

6 or material responses (e.g. tensile stress and strain) (7, 11-12, 14). Moreover, the effect of aging can be formulated in the constitutive models for asphalt mixtures (30), or is quantified by the change of the viscosity of the asphalt binder (7, 9, 13, 31-32). The viscosity of the binders from the wearing course is also used to investigate the variation in aging with depth (33). The aging models currently used in the Pavement ME Design are originally developed based on the viscosity data from different sources (31). Traffic Traction Stress and Strain Models The traffic traction stress and strain models documented in 27 relevant journal articles and reports are reviewed. In general, these models can be categorized into: 1) Experimental models; 2) Numerical models; and 3) Analytical models. It is believed that non-uniform stress distribution due to the interaction between the tire and pavement (34-36) is one of the primary causes. The contact stresses between the tire and pavement are measured for different tire types (21, 37-38). Significant tangential forces can be imparted to the pavement surface, which result in large tensile and shear stresses responsible for surface crack initiation (21-22, 39-44). In addition, the tire-pavement interaction is numerically simulated and a variety of computational analyses are performed using CIRCLY, BISAR, 2D plane strain elastic model, 3D multi-layer elastic finite element model, etc. (9-10, 45-51). The distributions of the tire-pavement contact stresses are well visualized by the outputs of the computational analyses. Based on the measured and numerically simulated results, analytical models are also developed to study tire-pavement contact stresses based on continuum damage mechanics or micro-mechanics (52-53). The produced results demonstrated the importance of shear stresses in generating top-down cracking in addition to the exclusively focused tensile stresses (53). Traffic Load Spectrum Models The load spectrum models documented in seven relevant journal articles and reports are reviewed. These models can be categorized into: 1) Effects of load spectra; 2) Mathematical load spectrum distribution model; and 3) Computational load spectrum model. It is reported that the load spectra have one of the most instrumental effects on the development of surface cracks. The critical load position could induced the greatest tensile stresses in the pavements (4). Several mathematical load spectrum models are formatted to determine the load spectra caused by traffic loading, such as the lognormal multiple-mode model along with the probability density function (54-55) and the composition method with composite random variables (56). The axle load data collected at Weigh-in-Motion (WIM) stations are also used to develop a truck load spectra model (57-58). In the Pavement ME Design, a computer program named WEIGHT.xls is incorporated to generate axle load spectra (59).

7 Thermal Stress Models The thermal stress models documented in 18 relevant journal articles and reports are reviewed. Thermal cooling in the pavement intensifies the tensile stresses for surface cracking (4, 23). The thermal stress models either adopt thermal properties, such as thermal coefficient of contraction (8, 30, 61-62), or utilize the fundamental properties like the creep compliance (63- 64), or reflect the effects of the thermal stress on fracture resistance of materials (7-8, 12, 21). The model types include the mechanical formulations based on the theory of viscoelastoplastic continuum damage (65), computational simulations (7, 66), and experimental evaluations of thermal fatigue cracking (67-69). Pavement Temperature Models The pavement temperature models documented in 30 relevant journal articles and reports are reviewed. The pavement temperature models consider various weather conditions, such as daily and annual temperature variation, wind speed, and solar radiation (70-74). The prediction of pavement temperatures changes from yearly maximum and minimum to a smaller time scale (e.g. daily, hour) (11, 74-80). Besides the steady-state temperature distribution, the transient temperature modeling is becoming important because the environmental and weather exposure affecting pavements is often random and of short duration (26-28, 81-82). The existing pavement temperature models generally can be classified as numerical models (85-86) and mathematical models. The mathematical models are formulated based on the theory of heat transfer and energy balance are regarded to be fundamental and efficient (70-71, 77-78, 82-84). The temperature variations with time and with pavement depth have been combined with age hardening and rainfall to predict the stiffness and crack growth within asphalt layers (4, 21, 30, 87-88). Crack Initiation Models The crack initiation models documented in 13 relevant journal articles and reports are reviewed. In general, the methods to predict crack initiation can be categorized into: 1) Experiment and mechanics-based models; and 2) Numerical models. The experiment and mechanics-based models rely on laboratory cracking tests and the theory of fracture mechanics (89-91) or viscoelastic continuum damage (30, 93). They are able to account for different mechanisms for top-down cracking, for instance, the crack initiation in asphalt layers of thin to medium thickness due to bending-induced surface tension and that in thicker layers due to shear-induced near-surface tension (30, 93-95). To account for the visco- elasto-plastic nature of asphalt mixtures, an energy-based crack initiation criterion is developed on the basis of Griffith’s crack initiation criterion for elastic materials (96). The numerical models make use of the techniques of the displacement discontinuity method (DDM) (97-98) or the embedded process zone (EPZ) (99) to simulate crack initiation and subsequent growth. The CRACK computer program (7, 66) can predict the initiation potential of surface cracks due to both thermal and load induced stresses. It is concluded that the combination of thermal and traffic effects is the cause of top-down cracking.

8 Crack Propagation Models The crack propagation models documented in 24 relevant journal articles and reports are reviewed. In general, the methods to predict crack growth can be categorized into: 1) Statistical regression models; 2) Experiment and mechanics-based models; and 3) Numerical models. The statistical regression models are constructed based on the regression analysis on historical distress data of top-down cracking (79, 100). Such models are only valid within the limit of the analyzed data and conditions. A more reliable model is the one developed based on experimental and analytical analyses, namely experiment and mechanics-based models. Some models are developed based on the fracture mechanics (30, 89-91; 94-95). To overcome the limitation that fracture mechanics focuses on a single crack, an energy-based mechanistic (EBM) approach is developed with the concept of distributed continuum fracture (DCF) for numerous cracks in asphalt mixtures (101-103). The common parameters or variables used to evaluate the fracture resistance include the stress intensity factor (8, 104), energy release rate or fracture energy (105-107), J-integral (108); and pseudo J-integral (109). The crack propagation models for top-down cracking are also developed using different numerical approaches, such as cohesive zone model (110), Coupled element free Galerkin (EFG) and finite element modeling (111), and Displacement discontinuity method (DDM) (97-98). The computer programs ABAQUS and CRACK are used perform the simulation of top-down cracking in asphalt pavements (39, 112). Finite Element Crack Growth Models The finite element crack growth models documented in 16 relevant journal articles and reports are reviewed. These models are also reviewed above, including the finite element (FE) analysis (4, 7-8, 22, 66, 93, 112-113), displacement discontinuity (DD) method (97-98, 114), and coupled element free galerkin (EFG) and finite element modeling (111). Artificial Neural Networks (ANN) Models The ANN models documented in nine relevant journal articles and reports are reviewed. The ANN technique has gained wide applications in pavement analysis and modeling. For example, it is used for surface cracking detection, classification, and quantification (115-119). The approach has the potential to classify pavement cracking images by type, severity, and extent of cracking present in the images. Based on this information, the network is able to determine the pavement condition and identify the condition ratings (120-121). Another example is the application of the ANN models to determine stress intensity factors that are used to predict crack growth through the Pairs’ Law (70). Cumulative Damage Models The cumulative damage models documented in four relevant journal articles and reports are reviewed. The purpose of the cumulative damage models is to combine different mechanisms of top-down cracking to produce a net prediction of crack initiation and growth. For example,

9 top-down cracking caused by thermal loading and traffic loading are combined to calculate the cumulative damage in asphalt layers (7, 66). Another example contains more factors such as aging, healing, failure criteria, viscoplasticity, and thermal stress, which are incorporated into the existing viscoelastic continuum damage (VECD) model as submodels to predict top-down cracking (30). Furthermore, the modified Miner’s law (122), as the sum of damage ratios, is useful to determine the total damage due to the combined effects of a number of factors. Observed Geometry of a Crack as It Propagates Downward Top-down crack severity level is recorded in the LTPP database as the width of the crack at the surface. The following table gives the crack width and the corresponding severity of the distress. Table 2.1. Crack Width and the Corresponding Severity of the Distress (154) Severity of Distress Range of Surface Crack Width (mm) Low 3 - 6 Medium 6 - 19 High > 19 The computer program developed in this project predicts the depth of penetration of the top-down crack. The geometry of top-down cracks as revealed in cores was used as a basis for relating the computed crack depth to the observed surface crack width and crack distress severity. The details of this relationship are provided in Chapter 3. The top-down crack severity that is recorded in the LTPP database has very few crack severity ratings greater than “low”. This is because as these cracks become more severe, they are recorded as “medium severity” alligator cracking.