Falling Weight Deflectometer Backcalculation Research Articles

Study on Dynamic Modulus Prediction Model of In-Service Asphalt Pavement

The dynamic modulus of in-service asphalt pavements serves as a critical parameter for the computation of residual life and the design of overlays. However, its acquisition is currently limited to laboratory dynamic modulus testing using a limited number of core samples, necessitating a reassessment of its representativeness. To facilitate the prediction of dynamic modulus design parameters through Falling Weight Deflectometer (FWD) back-calculated modulus data, an integrated approach encompassing FWD testing, modulus back-calculation, core sample dynamic modulus testing, and asphalt DSR testing was employed to concurrently acquire dynamic modulus at identical locations under varying temperatures and frequencies. Dynamic modulus prediction models for in-service asphalt pavements were developed utilizing fundamental model deduction and gene expression programming (GEP) techniques. The findings indicate that GEP exhibits superior efficacy in the development of dynamic modulus prediction models. The dynamic modulus prediction model developed can enhance both the precision and representativeness of asphalt pavement’s dynamic modulus design parameters, as well as refine the accuracy of residual life estimations for in-service asphalt pavements. Concurrently, the modulus derived from FWD back-calculation can be transmuted into the dynamic modulus adhering to a uniform standard criterion, facilitating the identification of problematic segments within the asphalt structural layer. This is of paramount importance for the maintenance or reconstruction of in-service asphalt pavements.

Adaptation of an alternative neural network instead of ANN for backcalculating pavement layer moduli

The structural condition of the pavement is often evaluated using deflection data from the non-destructive test such as the Falling Weight Deflectometer (FWD). Young's moduli calculated from surface deflections are used to characterize pavement layers. One of the most popular techniques for analyzing FWD data to determine pavement layer moduli is backcalculation. Artificial Neural Network (ANN) has been explored or used in studies to backcalculate layer moduli from FWD data, although its applications in this area are relatively recent. However, ANN suffers from limitations in terms of convergence accuracy and generalization capability. This research aims to develop an alternative neural network instead of ANN for backcalculating pavement layer moduli. Based on FWD data, Recurrent Neural Network (RNN) and Long-Short Term Memory (LSTM) networks were used to overcome the shortcomings of the traditional ANN. The LSTM, RNN, and ANN networks were developed and trained with the same training properties. The results revealed that LSTM attained high convergence accuracy and fastest convergence speed than ANN and RNN. LSTM network produced reasonable and accurate layer moduli values with determination coefficient (R2) of 0.9298 when compared to the measured values, while R2 values of the RNN and ANN networks were 0.906 and 0.8326, respectively. There was evidence that the LSTM network can learn the continuity pattern between the deflection basin points and enhance the FWD backcalculation accuracy.

RNN-based pavement moduli prediction for flexible pavement design enhancement

In order to facilitate the effective implementation of the MEPDG, researchers concentrate on quantifying local material properties, with a particular emphasis on pavement layer moduli. The layer modulus is a critical parameter necessary for calculating pavement responses (stress, strain, and deflections) resulting from traffic loading. Accurately determining the layer modulus is crucial for enhancing pavement design as it directly impacts the required pavement layer thicknesses and associated costs. Backcalculation is a commonly used method for analyzing Falling Weight Deflectometer (FWD) data to determine pavement layer moduli, with Artificial Neural Networks (ANNs) being the traditional choice. However, ANNs have limitations in terms of convergence accuracy and generalization capability. The aim of this study is to improve the backcalculation of layer moduli to enhance pavement design. By utilizing FWD data, Recurrent Neural Network (RNN) was employed to address the limitations of conventional ANN. Both ANN and RNN networks were developed and trained using identical properties. The findings demonstrate that RNN achieved faster convergence and higher convergence accuracy compared to ANN. The RNN network generated reasonable and precise layer moduli values, exhibiting a determination coefficient (R) of 0.95 in comparison to the measured values, while the ANN network had an R-value of 0.79. The results indicate that the RNN network can learn the continuity pattern between deflection basin points, thereby enhancing the accuracy of FWD backcalculation.

The Application of Frequency-Temperature Superposition Principle for Back-Calculation of Falling Weight Deflectometer

The falling weight deflectometer (FWD) is a widely used nondestructive test (NDT) device in pavement infrastructure. A FWD test measures the surface deflections subjected to an applied impact loading and the modulus of pavement layers can be determined by back-calculating the measured deflections. However, the modulus of asphalt layers is significantly influenced by temperature; hence, the temperature correction must be considered in back-calculation to evaluate the moduli of asphalt layers at a reference temperature. In addition, the in situ temperature at various pavement depths is difficult to measure. A model for evaluating the temperature at various depths must be established to estimate the in situ temperature of asphalt layers. This study collected the temperature data from a FWD test site to establish a temperature-evaluation model for various depths. The cored specimens from the test site were obtained to conduct dynamic modulus tests for asphalt layers. The FWD tests were applied at the FWD test site and the back-calculation was performed with temperature correction using the frequency-temperature superposition principle. The back-calculated moduli of asphalt layers were compared with the master curve of dynamic modulus to verify the application of the frequency-temperature superposition principle for FWD back-calculation. The results show that the proposed temperature-evaluation model can effectively evaluate the temperature at various depths of pavement. Moreover, the frequency-temperature superposition principle can be effectively employed to conduct temperature correction for FWD back-calculation.

Visco-Elastic Back-Calculation of Traffic Speed Deflectometer Measurements

The traffic speed deflectometer (TSD) has proven a valuable tool for network level structural evaluation. At the project level, however, the use of TSD data is still quite limited. An obstacle to the use of TSD at the project level is that the standard approaches to back-calculation of pavement properties are based on the falling weight deflectometer (FWD). The FWD experiment is similar, but not equivalent, to the TSD experiment, and therefore it is not straightforward to apply the traditional FWD back-calculation procedures to TSD data. In this paper, a TSD-specific back-calculation procedure is presented. The procedure is based on a layered linear visco-elastic pavement model and takes the driving speed of the vehicle into account. This is in contrast to most existing back-calculation procedures, which treat the problem as static and the pavement as purely elastic. The developed back-calculation procedure is tested on both simulated and real TSD data. The real TSD measurements exhibit significant effects of damping and visco-elasticity. The back-calculation algorithm is able to capture these effects and yields model fits in excellent agreement with the measured values.

Damage Assessment for ME Rehabilitation Design of Modified Asphalt Pavements: Challenges and Findings

The overall objective of this study was to assess the use of Level 1 analysis for mechanistic-empirical (ME) rehabilitation designs of deteriorated polymer-modified asphalt concrete (AC) pavements in Nevada using the AASHTOWare® Pavement ME software. This research also explored the possible implementation of a hybrid approach for AC damage characterization to overcome the challenges associated with the use of the Witczak model for estimating the undamaged dynamic modulus master curve of the existing AC layer. Two rehabilitation field projects were used as part of this study. The experimental plan involved falling weight deflectometer (FWD) testing in the right wheelpath before rehabilitation, analysis of core samples, estimation of an equivalent undamaged dynamic modulus, and estimation of equivalent damaged dynamic modulus from FWD backcalculation. The proposed hybrid approach consisted of conducting laboratory dynamic modulus testing on the collected core samples and estimating an equivalent undamaged dynamic modulus at the same FWD testing temperature and loading frequency. The pre-overlay damage, characterized based on the approach in Pavement ME Design software (i.e., using a Witczak prediction model and backcalculated modulus), showed overly high values that did not match with the collected pre-overlay distress data on either of the rehabilitation projects. Based on the findings from this study, the hybrid approach was recommended for implementation by Nevada Department of Transportation (NDOT) when designing AC overlay over polymer-modified asphalt pavements in Nevada. Recommendations for user inputs were also provided for future consideration in Pavement ME Design software.

A Comparison of Dynamic Soil Models for Backcalculation of Engineering Soil Properties from Falling Weight Deflectometer Data

The falling weight deflectometer (FWD) test is one of the most widely used nondestructive tests for backcalculation evaluation of pavement and soil properties. Despite the advantages in accounting for the time-dependent load and responses, the accuracy of evaluating elastic moduli based on dynamic backcalculation analyses is, however, significantly depended on the forward dynamic soil models employed in the backcalculation process. To serve this challenge, a comparison for advantages, limitations and the reliability of different dynamic soil models employed in the backcalculation of engineering soil properties is presented in this paper. Two types of dynamic soil forward models in the FWD backcalculation, namely, dynamic single-degree-offreedom (SDOF) and dynamic half-space models are considered. The study in this paper provides a better understanding of dynamic backcalculation processes which is essential for the development of dynamic backcalculation program and its applications.

Backcalculation of Flexible Pavement Layer Moduli from Traffic Speed Deflectometer Data

In the past few years, state transportation departments in the United States have been investigating the potential use of the traffic speed deflectometer (TSD) in routine network-level pavement structural evaluation, including efforts to compute parameters typically obtained with the falling weight deflectometer (FWD). However, differences in loading configuration (single plate versus dual tires) and mechanism (impact versus moving load) between the FWD and the TSD necessitate modification of conventional analysis methodologies for use with TSD data. The TSD is a trailer truck that uses Doppler lasers to measure the pavement deflection velocities at normal traffic speed. Deflection velocities are then converted to surface deflections with a computational algorithm. This study proposes two approaches to backcalculating layer moduli from the TSD data. The first method uses 3D-Move software and deflection velocities measured directly by the TSD. The 3D-Move software can closely simulate TSD loading by considering moving load, nonuniform contact pressure of the dual tires, and viscoelastic material properties of asphalt layers. The second method uses TSD-computed surface deflections in a linear elastic analysis backcalculation program modified to incorporate a dual-tire configuration. The two methods were evaluated with field data from the MnROAD test facility and two in-service pavement sections in Pennsylvania and Idaho. The results show that the second method can be reliably used in network-level applications until the first method becomes computationally practical. The limitations of use of the TSD deflection measurement in the conventional FWD backcalculation program was also demonstrated.

Influence of Cracking Damage on Deflection Basin Test Data of FWD

With the popularization of falling weight deflectometer (FWD) to calculate the stiffness related parameters of the pavement structures, non-destructive evaluation of physical properties and performance of pavements has taken a new direction. FWD backcalculation is mathematically an inverse problem that could be solved either by deterministic or by probabilistic approach. A review of the currently used backcalculation procedures indicates that the calculation is generally based on a homogeneous, continuous, and linear elastic multi-layer system. Identifying effective data of dynamic deflection basins seems to be an important task for performing modulus backcalculation. Therefore, the main objective of this paper was to discuss the distribution features of dynamic deflection basins of asphalt pavements with crack distresses, and present the reasonable criteria to filter the testing data of FWD deflection basins. Finally, the study aims to validate the established criteria by conducting in-situ case study.

Effect of Cross-Anisotropy of Hot-Mix Asphalt Modulus on Falling Weight Deflections and Embedded Sensor Stress–Strain

In this study, the effects of cross-anisotropy on asphalt pavement responses are examined. A dynamic finite element model (FEM) was developed in ABAQUS to simulate pavement responses under falling weight deflectometer (FWD) and truck loads on a pavement section on I-40 at Mile Post 141 in New Mexico. This section was recently instrumented with strain gauges, moisture probes, and pressure cells. Pavement response (i.e., stress, strain, deflection) from the instrumented section is compared with the FEM-predicted values. Two combinations of cross-anisotropy are considered. The first combination considers cross-anisotropy of modulus in every layer of the pavement, and the second combination considers it only in the hot-mix asphalt layer. Time–deflection histories, stress, and strain are predicted from the FEM under FWD and truck loads. Results show that predicted deflections, stress, and strain are highly sensitive to cross-anisotropy. Predicted deflections, stress, and strain increase with a decrease in n-value, defined by the ratio of horizontal to vertical modulus of elasticity. Analyses are performed for two shapes of loading area: semicircle–rectangle and rectangle. Predicted stress and strain are shown to be larger for the rectangular shape of the loading area than for the semicircle–rectangle area. This study recommends including modulus anisotropy in FWD backcalculation and in pavement design.

Consistency and accuracy of selected FWD backcalculation software for computing layer modulus of airport pavements

Falling weight deflectometer (FWD) data are generally processed through backcalculation software for computing layer modulus of airport pavements. Though a number of backcalculation software are currently available, it is not known that software produces the most consistent and accurate results. Inconsistent and inaccurate backcalculation software can take a significant toll on the overlay and rehabilitation design of airport pavements as modulus being the inputs for such design calculations. Most backcalculation software requires an initial value of modulus, known as seed modulus, to initiate simulations. Unfortunately, in many of the software, the outputs vary with the initial choice of seed modulus (consistency). In addition, depending on the software algorithm, the final output differs from the laboratory measured value of modulus (accuracy). In this study, three backcalculation software programs are evaluated for consistency and accuracy. They are BAKFAA, MODULUS and EVERCALC. The major goal is to determine the most consistent and accurate software for determining layer modulus of airport pavements in New Mexico. Consistency of software is determined by comparing outputs from three sets of input data from FWD tests at identical locations with three magnitudes of loads: 40·03, 53·38 and 71·17 kN. Laboratory tests are conducted for modulus and tensile strength of asphalt and soil samples collected from the same locations of FWD tests. Software accuracy is determined based on the distribution and magnitude of the variance of laboratory and FWD modulus. Laboratory tensile stress is also compared with the backcalculated tensile stress using KENLAYER, a multilayer elastic analysis software. Study results show that EVERCALC software produces more consistent and accurate modulus values than the BAKFAA and MODULUS software. Therefore, EVERCALC is recommended for analyzing layer modulus of airport pavements.

Evolution of Asphalt Mixture Stiffness under Combined Effects of Damage, Aging, and Densification under Traffic

The long-term evolution of asphalt mixture stiffness in the field represents a complex process in which three factors play an essential role: damage, aging, and densification under traffic. To take the factors into account adequately, not only must the level of each factor be known, but also how each level modifies the modulus of the asphalt mixture for the temperature and frequency range corresponding to field conditions. The research presented is based on experimental data from four flexible sections tested at the test track of the Centro de Estudios y Experimentación de Obras Públicas (CEDEX), in Spain. The stiffness of the asphalt layer was periodically evaluated with falling weight deflectometer back-calculation and with dynamic modulus testing in the laboratory. During a 28-month period, 1,323,600 loads were applied on the pavements. This loading caused a high level of deterioration in all sections and significant asphalt densification. The loading also provided time for some aging effects to develop. From the results of this full-scale test, different conclusions could be deduced concerning the evolution of damage, aging, and densification and how this evolution modifies the stiffness of the asphalt mixture. The effects of changes in the parameters of the original master curve have also been quantified. The model and methodology incorporated in the CalME design procedure were successfully used to reproduce the evolution of the asphalt layer modulus during the test. In particular, the assumptions of this model regarding how the dynamic modulus master curve was modified by each of the three factors were found to be valid for this experiment.

Shift Factors for Asphalt Fatigue from Full-Scale Testing

Asphalt fatigue lives determined in the laboratory must be shifted to predict the performance of in-service pavements. Determination of the shift factor to be applied constitutes one of the main challenges of flexible pavements design. Experimental data from four sections tested for 28 months at the CEDEX test track provided detailed information regarding accumulation of asphalt fatigue damage as a function of loads and temperature. The data were used to draw conclusions about the damage process and to determine shift factors for two cracking levels in the field. An effort was made to quantify the beneficial effect of rest periods in the field through use of a variable shift factor that incorporates the methodology for studying this effect and the concept of reduced rest period that was developed in NCHRP 9-38 and 9-44. Results showed that rest periods were not only the main factor responsible for the need for shift factors but also the primary explanation for why most damage took place at medium to low temperatures and not during the warmest season. The asphalt fatigue model in the CalME design procedure has been used successfully to reproduce actual performance observed in the tested sections in terms of modulus reduction, which was previously determined from falling weight deflectometer backcalculation.

Comparative Study of Static and Dynamic Falling Weight Deflectometer Back-Calculations Using Probabilistic Approach

The falling weight deflectometer (FWD) test is the most widely accepted and used technique for nondestructive evaluation of pavements. In this test, the pavement response to the impulse load of a falling weight is analyzed or back-calculated to evaluate the in situ pavement layer moduli. FWD back-calculation is mathematically an inverse problem that can be approached deterministically or probabilistically. In this paper, the probabilistic approach to FWD back-calculation is used as a tool to investigate the reliability of different dynamic and static back-calculation procedures. The probabilistic approach to FWD back-calculation, a relatively new approach, is initially introduced and a brief overview of its numerical implementation is presented. The approach is then applied to back-calculate moduli from a set of synthetic FWD data using different back-calculation procedures, such as static and dynamic deflection bowl or dynamic time history procedures. Based on the probabilistic solution for back-calculated moduli from each procedure, the reliability of each procedure is then compared to the others. It is shown that, although widely used, the static FWD back-calculation procedures fail to capture the essential dynamic nature of the test and consequently cannot be relied upon to correctly back-calculate the layer moduli. It is also demonstrated that there is little redundancy in the dynamic deflection bowl back-calculation procedures, potentially producing relatively large uncertainties in the obtained back-calculation results. Finally, it is shown that the dynamic deflection time history back-calculation is the most reliable approach for FWD back-calculation. Using the back-calculated moduli from different procedures, effect of different back-calculation approaches on the pavement life is illustrated.

Field and Lab Investigations of Prematurely Cracking Pavements

A forensic study was conducted to identify the cause of the premature cracking on three recently completed projects that were built with the same design. Nondestructive [ground penetration radar, falling weight deflectometer (FWD), GeoGauge, and Portable FWD], nuclear density gauge, dynamic cone penetration, and extensive laboratory tests were performed. It was found that the initial stiffness of the treated base was found to be excessively high by FWD backcalculation. Some sections of the backcalculated base moduli were over 20.7 GPa. This indicates that the layer is excessively brittle for a base material, similar to lean concrete. Six specimens (that were made without a mellowing period) exhibited cracks. There was no cracking for six specimens that had two days of mellowing. It was concluded that the culprit of the transverse cracking in the main lanes was the shrinkage of lime treated base layers. The longitudinal cracks are related to the edge drying and the transverse cracks are related to the insufficient mellowing period. Based on the findings of this study, the District implemented a 2-day mellowing period for Quicklime treated caliche base. Three newly constructed pavements (age 8, 5, and 2 months) were surveyed. No cracking can be observed so far, and the District thinks the cracking problem has been mitigated by the 2-day mellowing period. Without the mellowing period, cracking had normally occurred 1 to 2 months after construction.

Practical Methodology of Determining the In Situ Dynamic (Complex) Moduli for Engineering Analysis

The viscoelastic behavior of asphalt mixtures can be conveniently presented in the form of a stiffness master curve in various engineering applications. However, the methods to obtain the in-service mixture stiffness are limited to the falling weight deflectometer (FWD) testing and back-calculation procedure because the pavement layer thicknesses in many cases are too thin to allow laboratory testing. The test protocol to measure the axial dynamic modulus requires cylindrical specimens with a height-to-diameter ratio higher than 2, while the cores need to be at least 38 mm thick for measuring shear stiffness. This technical note describes and compares the methods for constructing asphalt mixture master curves utilizing stiffness predictive equations for asphalt binders and mixtures, in addition to using laboratory measurements, as an alternative to FWD backcalculation analysis. This technical note also recommends constraining the low and high mixture stiffness values measured from simple shear testing to obtain stiffness comparable to the uniaxial dynamic modulus testing.

Use of FWD Data for Pavement Material Characterization and Performance

Conventionally used FWD backcalculation and pavement stress-strain computation procedures are based on linear static layered elastic theory. This paper presents a simple hypothetical case study in which the applicability of the assumptions made in the foregoing procedures is investigated using the finite-element method. The pavement displacements computed using static/dynamic falling-weight deflectometer (FWD) loading and constant and stress-dependent material properties were used as input for the backcalculation analysis. The assumed and backcalculated layer moduli were compared. The backcalculated asphalt concrete (AC) moduli values were within acceptable limits for all pavement sections studied except for the case of the thin AC pavement with a shallow rigid base. The subgrade moduli were consistently underpredicted; the base course moduli were both underpredicted and overpredicted. Widely used traffic-induced pavement strains that are related to pavement performance that were obtained under different loading and material property characterizations were also compared. Results indicate that the conventional procedures are not conservative, especially for thick AC pavement sections.