Pavements and Materials: Modeling, Testing, and Performance

The main objective of this paper is to present the state-of-the-art in the anisotropic characterization of unbound aggregate layers to support the development of the next generation of highway as well as airport pavement infrastructure by employing advanced technology and detailed knowledge gathered within the last one and a half decades in characterization of the actual stress- and direction- (anisotropic) dependent aggregate behavior. From advanced triaxial tests conducted in the laboratory, unbound aggregate materials are clearly shown to have greater modulus in the vertical direction, which can be adequately modeled using cross-anisotropic, stress-dependent modulus characterization models. Field validations of the anisotropic aggregate layer behavior have involved comparing measured full-scale pavement test section responses with the predicted ones using the anisotropic aggregate modulus characterization. A simplified approach of obtaining level of anisotropy as an aggregate quality indicator is summarized to indicate use of high quality aggregates with superior properties, including size, gradation and shape, critical to successful pavement designs. Recent efforts as well as the need to develop simplified design and analysis methodologies that will enable optimized use of engineered aggregate structural layers for improved performance and increased use in flexible pavements are also discussed in detail.

This paper presents a model for predicting damage behavior of asphalt mixtures in flexible pavement systems. Experimental protocols to measure material properties and fracture characteristics of mixture components and their incorporation into constitutive models based on the micromechanical finite element analyses are developed and presented herein. The model employs only directly-measured material and fracture properties of small scale mixture components as model inputs necessary to predict damage-dependent behaviour and life of global scale asphalt concrete mixtures that typically show significant heterogeneity and inelasticity. The presented methodology is therefore expected to provide a significant savings in experimental costs and time over traditional approaches.

This paper proposes a method for determining Burger's model parameters using creep-recovery data. With the determined Burger's model parameters, the viscoelastic behaviors of an asphalt binder (PG64-28) and its two mastics (a binder PG64-28 blended with mineral fines passing the ASTM 100# sieve and 200# sieve, respectively) are investigated. The total deformation of the mastics was separated into three parts: the instantaneous elasticity, delayed elasticity, and viscous flow. The ratios of the viscous flow to the elasticity were calculated and analyzed. It was found the proposed method is promising where the average errors of the creep section are 3.53%, 5.28% and 8.02% for the three materials, and the errors in the recovery section are 0.69%, 1.286% and 1.80%, respectively.

This study evaluates cracking in asphalt concrete under wet and dry conditions. Three Superpave mixes are considered. Semi-circlular notched asphalt concrete specimens are fractured in the laboratory using a strain-based load in three point bending. Crack opening displacement (COD) data is captured in real-time along the length of the crack using linear variable displacement transducers (LVDTs). Three parameters derived from the laboratory data are used to characterize the mix cracking in terms of resistance to crack initiation and crack propagation in dry and wet conditioned samples. These parameters are namely ultimate load, slope of the crack propagation curve, and crack velocity. It is shown that wet samples can carry a higher ultimate load than the dry samples. Wet samples tend to show higher load decrease (higher slope) during crack propagation than the dry samples. Mix type has very little effects on the crack velocity, although wet samples have slower cracks than dry samples.

This paper demonstrates a simplified method to capture the effect of temperature non-linearity on rigid pavement stress responses utilizing a single parameter (NOLA) and a graphical understanding of temperature non-linearity. This method allows users to easily post-process rigid pavement solutions with linear temperature assumptions to account for these self-equilibrating stresses at any depth in the slab. In addition, this paper also reveals the impact of accounting for self-equilibrating stresses in terms of projected structural fatigue damage at locations along both the top and bottom of the transverse joint and longitudinal edge of a concrete pavement section using a rigid pavement analysis program named RadiCAL.

This paper discusses the results obtained from small scale accelerated pavement testing in the laboratory. The small scale loading equipment has been used in the laboratory to perform fatigue tests on HMA test slabs compacted using vibratory compactor in the laboratory. The real time analysis of data obtained from strain gauges and thermocouples indicated the effect of loading and temperature on the strain response. The fatigue characteristic equation has been developed and compared with the standard fatigue equation.

Dynamic response of rigid pavement subjected to moving vehicle load with variable velocity (accelerating or decelerating) is investigated in the paper. Rigid pavement is modeled by an infinite beam on viscoelastic foundation. Double Fourier Transform and inverse Fourier Transform technique and Convolution Theorem are employed to obtain the dual integral solution for the dynamic deflections of beam. A numerical calculation has been performed by Fast Fourier Transform (FFT) to illustrate the dynamic responses of the pavement under different moving conditions of vehicle. The effect of acceleration and deceleration of a moving wheel on the dynamic deflections of rigid pavement is discussed in detail. Numerical example results show that the velocity of the moving vehicle has greater effect on the maximum dynamic deflection of pavement when compared with the acceleration and deceleration, which has less effect on the maximum dynamic deflection. The latter mainly has less influence on the position and the timing of the maximum.

The objective of this paper is to present a new approach to determine the flow number of asphalt mixtures during dynamic creep test. A number of research studies have focused on the relationship between material properties and pavement performance. The Superpave Simple Performance Tests (SPT) is part of the efforts in past few years. One of the Superpave SPT was the repeated loading or dynamic creep test. The output of this test is the flow number, which is the initiation of tertiary flow. A common method in examining the flow number is to locate the lowest point in the strain rate versus cycle number curve. However this method may provide confusion due to the variation of the testing data. A new method was developed and evaluated in this paper. In this process, it was assumed that the strain would only increase along with time. Then the flow number at the minimum point of new strain rate versus cycle number was determined. It was found that the flow number result from this new method is more consistent and practical compared to other methods for determining the flow number.

Asphalt matrix (AM) properties have been used for many years to relate the behavior and performance of Hot Mix Asphalt (HMA) mixture. The AM is commonly defined as the combination of asphalt binder and fine aggregates. Understanding the mechanics and rheological characteristics of AM can facilitate better mixture design, and provide an insight into the visco-elastic behavior of HMA mixtures. In this study, the effects of aggregate gradation and asphalt binders on the visco-elastic response of AM were studied. Two fine aggregate gradations and two asphalt types were used to make AM samples. Dynamic mechanical analyses were conducted to investigate the creep response and effect of rate of loading at various temperatures. The dynamic shear modulus (G*) and shear creep compliance (J[t]) were obtained over a range of temperatures to construct master curves at 25°C. The master curves for G* and J[t] were modeled using Sigmoidal function, and generalized visco-elastic model, respectively. The analysis of master curves and models showed that the finer gradation exhibited a lower G*, and higher J[t], for the neat binders. In general, the modified binders and coarser gradation produced a stiffer material for a range of frequencies, but the J[t] showed some variations in the AM creep response. In addition, rutting susceptibility of AM was evaluated using the Superpave® rutting parameter.

The characteristics of concrete joint roughness and crack width are of paramount importance when studying the shear behavior of joints and cracks. These characteristics play a fundamental role in determining the contribution of each of the three shear mechanism; namely friction, shearing through and the dilation components in the overall shear strength at the crack or joint interface. However, in almost all prior studies of shear transfer across joints and cracks, no attempt was made to accurately quantify the roughness of concrete joint or crack surface. In the past, substantial levels of empiricism were used to model and predict shear behavior along discontinuities. This empiricism can be attributed mainly to the lack of an accurate measure for quantifying roughness. A new method that captures all critical characteristics of joint roughness is proposed. The new method is based on the logical assumption that an accurate description of roughness should reflect the following three important aspects of roughness: the irregularity of the surface, the amplitude of different roughness features, and the frequency of occurrence of roughness features with different amplitudes. Based on laboratory testing, the proposed method was found to capture all three aspects of a rough surface successfully. It is proposed that the surface roughness should be described using the form RI(D, k_l, h_l), where RI will be called the Roughness Index, D is the fractal dimension, k_l is the Korcak Number, and h_l is the Hurst Number with the subscript l indicating the stride length. These parameters are described in detail in this paper.

The hollow cylindrical asphalt mixture specimens were modeled with discrete element method. The dynamic moduli of the samples were predicted. The microstructure of asphalt concrete specimen was captured from an X-ray tomography images. The asphalt concrete hollow cylindrical images were divided into aggregate, mastic and air voids domain. The particle size of the mastic was less than 2.36mm in this paper. Aggregate, mastic and air voids domain were identified from the density index of the X-ray computed tomography images. The coordinates of aggregate, mastic and air voids domain was defined by density index to develop two dimensional (2D) and three dimensional (3D) discrete element models (DEM). The dynamic modulus of the asphalt mixture was calculated from stress divided strain in each temperature and frequencies. The comparison of predicted and laboratory measurement dynamic modulus has been investigated in this paper. When compared with the dynamic modulus of 2D DEM and 3D DEM against the laboratory measurement, the average difference was 6.25% for 3D DEM; the average difference was 17.72% for 2D DEM; and the average difference of dynamic modulus was 22.19% when compared with 3D DEM and 2D DEM. It was found that 3D DEM had better predicted dynamic modulus comparing 2D DEM.

The paper presents the finite incremental constitutive equations for the Viscoplasticity Constitutive Model of Asphalt Concrete developed under the Strategic Highway Research Program (SHRP). The viscoplasticity model is highly non-linear and numerically stiff. The implicit method is not well suited for the implementation of this type of constitutive models; the explicit method requires very small time step and therefore is very expensive. A semi-implicit method using Taylor series expansion and first-order-derivative approximation was used to derive the finite incremental equations. The equations were in vector format and convenient for implementation.

Subgrade soil compaction can generate a volumetric locking phenomena if standard displacement-based finite element formulation is used to model the their nearly incompressible limit. This is a potentially problematic issue for pavement compaction for either asphalt or subgrade soil. This paper presents a stabilized large deformation finite element formulation for compaction simulation. The stabilization technique employed here for the mixed displacement-pressure formulation is based on the variational multiscale idea. It is written in the spatial description and is integrated with a smooth surface Cap model for the analysis of materials that can take large compressive stresses but very low tensile stresses. Numerical experiments using hexahedral elements with equal low-order interpolations for both displacement and pressure fields show that the new stabilized formulation can successfully remove volumetric locking. This formulation effectively predicts the compaction density induced by the external loads, which is an attractive feature of the formulation for practical applications in geotechnical/pavement engineering.

The composite pavement of CRCP+AC has often been given credit for its combination of high bearing capacity and desired surface feature like typical flexible pavement has. It has full potential to make the most out of the desired features of both rigid and flexible pavements to make a sound pavement system. In this paper, 3D FEM models of CRCP and CRCP+AC were developed respectively by using FEM software named ABAQUS. The vehicle load is simplified to be a moving load and has been realized through a user subroutine VDLOAD. The differences of the dynamic response between CRCP and CRCP+AC under traffic loading were compared. Influencing factors, including AC thickness, base course type, longitudinal reinforcement content and slab thickness, were evaluated to analyze the dynamic stress and displacement of pavements. The results of analysis can provide theoretic basis and reference for the structural design and construction technique of the CRCP and CRCP+AC.

Rheological properties of asphalt binders are important parameters for prediction of major distresses in asphalt pavements. An experimental study of three Performance Grade (PG) asphalt binders (PG 64-22, PG 70-28, and PG 76-28) commonly used in Oklahoma was undertaken to examine their visco-elastic properties at high grade temperatures by using a Dynamic Mechanical Analyzer (DMA). The test results obtained from the DMA were compared with those from a Dynamic Shear Rheometer (DSR). The DSR tests were conducted in accordance with the AASHTO T-315 test method. Nine PG binder samples (three for each binder type) were examined by using the DMA, and an equal number of samples were tested by using the DSR. Additional tests were conducted to compare DMA and DSR results at different temperatures ranging from 58°C to 73°C at 3°C intervals. The DMA-based test results were found to be in agreement with the DSR-based results. This study reveals that the DMA can be used efficiently to obtain reliable rheological properties of PG binders.

This paper comprehensively analyzed the several existing typical IRI models, and evaluated three existing models that describe the relationship between the pavement age and IRI values for selection of a best one as initial model. For those highly correct variables to the establishment of IRI model, this paper explored the data combination and interpolation. Finally, this paper tested the established IRI prediction model using LTPP data. In order to evaluate the effect of site factors and climate conditions on the established IRI model, this paper used LTPP data from three states: New Jersey, Washington DC/Virginia, and Maryland to check their effects. This paper has compared our IRI model with other existing models, and analyzed the relationship between the IRI model and distresses. The results have demonstrated that the distress characteristics identified in this paper are directly related to incremental changes in the IRI and have a significant effect on incremental changes in IRI with time.