UC Davis

Fatigue cracking is the most common distress in asphalt pavements. Currently, no performance-related laboratory tests exist for fatigue cracking to use in routine asphalt mix design to approve job mix formula (JMF) or quality control and quality assurance (QC/QA) in California. The existing four-point bending (4PB) test was developed to evaluate the fatigue performance of asphalt materials, but it is not necessarily appropriate for use in routine JMF, and it takes too long to complete for QC/QA. The first objective of this doctoral dissertation is to evaluate potential surrogate fatigue performance-related testing methods and identify a test that is simple and easy to perform and also provides a guidance for asphalt mix design on routine projects and for QC/QA on all projects.Potential performance-related tests evaluated in this study included monotonic loading fracture tests: semicircular bend (SCB) test, indirect tensile asphalt cracking test (IDEAL-CT), and repeated loading fatigue testing on fine aggregate matrix (FAM) mixes with linear amplitude sweep (LAS) testing configuration. These tests were conducted on a variety of asphalt materials, and they were assessed based on simplicity, repeatability (variability), and their relationship to flexural stiffness and fatigue life from 4PB tests. Fracture parameters obtained from SCB tests and IDEAL-CT and fatigue parameters from FAM mix fatigue tests were assessed as potential fatigue cracking indicators. Linear regression analysis was performed to correlate these indicators with the initial flexural stiffness and fatigue life from 4PB tests. The regression analysis results demonstrated that the SCB tests and IDEAL-CT provided the similar fracture information. In addition, fracture parameters from SCB tests and IDEAL-CT were found to be well correlated with the initial flexural stiffness from 4PB tests. Meanwhile, the initial flexural stiffness from 4PB tests showed a moderate nonlinear correlation with the fatigue life from 4PB tests. Among all fracture parameters, material strength obtained from IDEAL-CT was found to have low variability, strong correlation with flexural stiffness and a moderate correlation with fatigue life from 4PB tests, therefore, strength was proposed as a surrogate indicator for flexural stiffness and an indication of fatigue performance. The relationship identified in this study between flexural stiffness and flexural fatigue life, and the one between flexural stiffness and material strength from IDEAL-CT were used to develop a preliminary specification for fatigue performance. The strength from IDEAL-CT should meet both upper and lower specification limits to ensure required fatigue performance met for asphalt mixtures. However, since there was no strong relationship found directly between strength from IDEAL-CT and fatigue life from 4PB tests, fracture tests did not provide sufficient information to predict fatigue life performance. The repeated loading FAM mix fatigue test showed promising comparison results with both initial flexural stiffness and fatigue life from 4PB tests. The comparison between master curves of FAM shear stiffness and the ones of full graded hot mix asphalt (HMA) flexural stiffness indicated that FAM mixes were more sensitive to temperature and loading frequency than HMA as expected because of higher binder contents in FAM mixes. Linear correlations with R2 values of 0.63 and 0.59 were found between FAM shear stiffness and HMA flexural stiffness at intermediate frequencies (100 Hz and 10 Hz) at a reference temperature of 20 °C. In addition to the comparison between flexural stiffness of HMA and shear stiffness of FAM mixes, the dynamic compressive stiffness of HMA obtained from the asphalt mixture performance tester (AMPT) was also included to explore the effect of different loading configurations on the relationship between HMA stiffness and FAM mix stiffness. The shear stiffness of FAM and dynamic compressive stiffness of HMA were found to be moderately correlated at frequencies of 1 Hz, 10 Hz, 100 Hz and 1000 Hz. Furthermore, these three different types of stiffness: flexural stiffness of HMA, dynamic compressive modulus of HMA and shear stiffness of FAM mixes also indicated that the addition of rejuvenator to asphalt materials containing up to 50% RAP effectively reduced the stiffnesses almost to the same level of the virgin control mix. Given these findings, an attempt was made to upscale the shear stiffness of FAM mixes to the flexural stiffness and dynamic moduli of HMA with two methods. The comparison between predicted and measured moduli showed that the shear stiffness of FAM mixes provided reasonable estimates of both flexural stiffness and dynamic modulus of HMA at intermediate frequencies (1 to 10 Hz) with the error percentage less than 10%. On the other hand, overprediction was noted from both methods at higher frequencies. The comparison of fatigue performance between HMA and FAM mix was further investigated based on damage curves. The viscoelastic continuum damage (VECD) model, which depicts the reduction of material integrity under repeated loading as a function of damage accumulated in asphalt materials, was used to formulate damage curves based on the FAM LAS testing results and HMA 4PB fatigue testing results. Comparison results demonstrated that similar damage characteristics were observed between HMA and FAM mixes. The FAM mixes also showed lower material integrity at failure compared to the values of HMA mixtures, which indicated that FAM mixes were more damage tolerant than HMA. In addition to the VECD model, the FAM mix fatigue testing results also showed a good fitting result on the damage model implemented in the California Mechanistic-Empirical pavement design software (CalME). Similar ranking result among the CalME damage curves of different material types was found between FAM mix and HMA. Based on this study, it seems promising that FAM mix fatigue testing can be developed to supplement/replace 4PB fatigue testing on HMA due to its relatively more economical, faster and simpler procedure than conventional 4PB tests. More importantly, linear regression analyses on the selected fatigue parameters from FAM mix LAS fatigue test results and HMA 4PB fatigue results indicated that there was a strong correlation between the shear strain value at failure of FAM mixes and the strain value corresponding to one million cycles of fatigue life of HMA. The shear strain value at the failure of FAM mixes also showed a low variability with a coefficient of variation (COV) of 11.2%, therefore, the FAM mix LAS fatigue testing with the fatigue parameter of shear strain value at failure was recommended as a promising surrogate test for 4PB tests on HMA. Fatigue performance was then studied in the context of pavement structure, which is the second objective of this dissertation: develop numerical models using finite element method (FEM) with the software ABAQUSTM to estimate the pavement responses under traffic loading and daily thermal variation. Specifically, composite pavements containing of an asphalt concrete (AC) overlay on top of portland cement concrete (PCC) slabs was taken into consideration to investigate both traffic loading-induced and thermal loading-induced fatigue cracking or reflective cracking performance in this study. As this study only focused on the damage and crack initiation stage of reflective cracking, terms of fatigue cracking and reflective cracking were used in an interchangeable manner. FEM was firstly applied to investigate the impacts from the pavement bonding condition between AC overlays and PCC slabs, tire loading location, pavement material properties and joint properties between PCC slabs on the pavement response under traffic loading. The tensile strain value at the bottom of the AC overlay was considered as the primary fatigue damage parameter. A preliminary simulation study showed that the critical strain type that causes damage in the AC layer was dependent on the bonding condition between the AC overlay and the PCC layer. When the AC overlay is fully bonded with the PCC slabs, debonding between the AC and PCC layers will firstly take place due to the separating tension, and the damage is expected to initiate at the bottom of the AC layer above the joint corner between two PCC slabs. When the debonding area forms and starts to expand between the AC and PCC layers, damage in the AC overlay will then be primarily caused by the bending tensile strain at the bottom of the AC overlay. A full factorial with 2,700 simulation cases was then carried out with varying AC thickness, AC stiffness, bonding condition, stiffness of base layers (k-value), load transfer efficiency (LTE) between PCC slabs, and traffic loading value. Due to the different damage mechanisms of fully-bonded pavement and partially-bonded pavement, two separate regression models were established based on the simulation results to predict the maximum principal tensile strain. The comparison between the predicted strain value from these two models and the value obtained from FEM simulations demonstrated the accuracy of the regression models. In addition to traffic induced reflective cracking, the daily temperature variation induced reflective cracking was also investigated. In contrast to extreme cold temperatures which cause one time fracture cracking, moderate temperatures can induce repeated tensile strain and stress in the AC overlay all year around due to daily temperature variation, which is a more common situation in California. To address potential thermal reflective cracking under moderate temperatures, composite pavement structures under only thermal loading were simulated with FEM, and the critical thermal stress and strain values were calculated. Among the selected six climate regions, the yearly temperature parameters (average yearly maximum, average yearly minimum and average seasonal change) and daily extreme temperature difference indicated that composite pavement structures in the representative climate cities (Reno (NV), Daggett (CA) and Sacramento (CA)) were more prone to thermal reflective cracking at moderate temperatures. Two composite pavement structures with different AC overlay materials were modeled based on a HVS test track. The viscoelastic properties of the AC overlay material were obtained from 4PB frequency sweep tests. The movements in the PCC slabs and the AC overlay showed a decent agreement between simulation results and actual measurements with a relative error of 15%. For the purpose of simulation efficiency, a data clustering method was implemented to strike a balance between obtaining sufficient information with representative temperature profiles and minimizing the computation efforts. As a result, the temperature profiles in the year of 2011 in Davis, CA were divided into five groups based on the K-means clustering algorithm. Then, a single day was selected from each group as a representative, resulting in a total of five simulation cases in comparison to 365 cases. In the composite pavement structure, the maximum principal tensile stress was found to be located at the surface of the AC overlay right above the joint between PCC slabs. The largest tensile stress was calculated to be 10 kPa which occurred on the coldest day while the lowest tensile stress of 0.6 kPa took place on the day with the highest temperature. On the other hand, the critical tensile strain was always located above the joint with a negligible difference between the surface and bottom of the AC overlay. The highest tensile strain value of 100,000 με happened on the hottest day and the lowest tensile strain was approximately 10,000 με which occurred on the coldest day. To develop a laboratory test for moderate temperature induced fatigue cracking, modified 4PB fatigue tests were performed at high strain values and low frequencies. Two high strain levels (4,000 με and 6,000 με) were determined for 4PB testing based on the thermal strains obtained from FEM simulations and the testing machine constraint. The test frequency was set to 0.05 Hz to simulate the low frequency of daily temperature variation. After calculating the respective fatigue life at each strain level for the five temperature clusters, Miner’s law was used to obtain a quick estimation of the fatigue life under thermal loading. It was shown that when the composite pavement consisting of an AC overlay (64 mm thickness) on top of PCC slabs (178 mm) was only exposed to daily temperature variations in Davis, CA, the predicted fatigue life for the pavement was approximately 1.3 years, which agreed with the observation from the HVS section. In order to incorporate the moderate temperature effect on reflective cracking to pavement design, the damage model in CalME was utilized to fit the thermal fatigue 4PB testing results. The root mean square (RMS) value from the fitting analysis demonstrated that thermal fatigue had a high goodness of fit with the CalME damage model. In addition, the damage curve revealed that within the same loading cycles, thermal strain induced damage was considerably greater than the one caused by traffic loading. According to the findings from this study, the daily temperature variation at climate regions with moderate temperatures contributed to much larger values of thermal strain relative to those caused by traffic loading. However, it is important to point out that such high thermal strain values were obtained from the simulation condition that the AC overlay and PCC slabs were fully bonded. As the bonding starts to deteriorate, the strain and stress values caused by temperature changes are reduced substantially. Since the pavement is subjected to separating damage at the early stage of cracking initiation under traffic loading, the high thermal strain values will only exist before the separation/debonding between the AC overlay and PCC slabs. Reflective cracking performance will be a combined result from the moderate temperature induced damage and the traffic loading induced damage. If the debonding takes place faster than the damage from thermal strain, the impact from temperature will quickly reduce and the fatigue performance will mainly be controlled by traffic loading. On the other hand, if the damage from thermal strain accumulates faster than the debonding, the pavement will develop thermal reflective cracks quickly. This study raises a number of concerns that reflective cracking relies heavily on the interaction between thermal loading and traffic loading especially at the early stage after construction, and that the initial bonding condition as well as the deterioration of the bonding between the AC overlay and the existing bottom layer play an important role in determining the rate of reflective cracking. In summary, the following conclusions were obtained from this study: (1) IDEAL-CT with the parameter of material strength was recommended to be a surrogate fatigue performance-related test due to its simplicity, low variability and strong correlation with the initial flexural stiffness from 4PB tests, however, none of the monotonic loading fracture tests showed a strong correlation with fatigue life information from 4PB tests; (2) Repeated loading FAM mix fatigue testing is not as simple as IDEAL-CT, but it showed strong correlation with fatigue life information from 4PB tests, therefore further exploration of developing FAM mix fatigue testing as a replacement for 4PB tests for JMF and QC/QA in routine projects is worth investigation; (3) According to FEM simulation of traffic-induced reflective cracking on composite pavements, tensile strain values at a fully bonded composite structure were much larger than the ones from a debonded composite structure. Therefore, it is recommended that for reflective cracking modeling and prediction, separate regression models should be implemented for the different bonding conditions; (4) The FEM simulations on moderate temperature induced reflective cracking demonstrated that the moderate daily temperature variations led to relatively high strain values in the AC overlay in composite pavements, which makes composite pavements susceptible to premature reflective cracking; (5) The current CalME damage model was found to be suitable to describe the moderate temperature induced reflective cracking, therefore, it is recommended to incorporate the moderate temperature effect into the future ME pavement design.