Application of Artificial Neural Networks for Estimating Dynamic Modulus of Asphalt Concrete

Abstract

Researchers developing the MEPDG implemented a hierarchical input structure in recognition of the fact that ΗE*Η values for materials used in a particular design might not be available when the analysis is performed. In the lowest level of this structure, users may choose to employ a predictive equation that is based on mixture volumetric and binder properties to predict the mixture modulus (1). In addition to the original predictive model, the Witczak model, others have developed predictive models. Those that have gained some national interest are the Hirsch model (3), a modified form of Witczak's original model (4) referred to as the modified Witczak model, and a modified form of the Hirsch model referred to as the Al-Khateeb model (5). Various researchers have reviewed these models by using independent data sets and concluded that, while both have advantages and disadvantages in relation to necessary inputs and ease of use, the predictive capabilities of each could be improved (6-9).The North Carolina State University (NCSU) research team is currently developing a more robust Η E*Η prediction algorithm with the goal of using the model to predict the moduli of certain layers in the Long-Term Pavement Performance (LTPP) database. The team's current research focuses on improving the prediction accuracy by using the artificial neural network (ANN) technique. Such an approach has been used successfully in the pavement field for such diverse problems as moduli backcalculation (10-12) and predictions of roughness deterioration (13). The primary advantage of this approach over statistical regression techniques is that the functional form of the relationship is not needed a priori. When one considers that many individual and interacting variables affect the ΗE*Η values, such an advantage may be significant and may offer the potential for capturing complicated nonlinear relationships between the ΗE*Η and mixture variables better than regression analysis. The drawback most often cited to these approaches is their inability to extrapolate when a situation arises that is beyond those used to train the ANN. This shortcoming is overcome by using an extensive training data set that covers the entire range of expected conditions. The use of ANN techniques for predicting the mixture Η E This paper presents outcomes from a research effort to develop models for estimating the dynamic modulus (ͦ E*ͦ ) of hot-mix asphalt (HMA) layers on long-term pavement performance test sections. The goal of the work is the development of a new, rational, and effective set of dynamic modulus ͦ E*ͦ predictive models for HMA mixtures. These predictive models use artificial neural networks (ANNs) trained with the same set of parameters used in other popular predictive equations: the modified Witczak and Hirsch models. The main advantage of using ANNs for predicting ͦE*ͦ is that an ANN can be created for different sets of variables without knowing the form of the predictive relationship a priori. The primary disadvantage of ANNs is the difficulty in ...

Sensitivity Evaluation of MEPDG Performance Prediction

“…The Witczak E* empirical model incorporated into the MEPDG is just one of many similar models that have been proposed in the literature (e.g., Al-Khateeb et al, 2006;Christensen et al 2003, Ceylan et al, 2007Ceylan et al, 2009;Far et al, 2009). The prediction errors and lack of sensitivity to mix parameters of the Witczak dynamic modulus model have been amply documented in the literature (e.g., Schwartz, 2005;Ceylan et al, 2009).…”

“…It is important to keep in mind that the NSI z values in Table 8.30 and related discussion are all predicated on the validity of the Witczak dynamic modulus model. Although this is the model currently implemented in the MEPDG, there are several other alternatives that have been proposed in the literature (e.g., Al-Khateeb et al, 2006;Christensen et al, 2003, Ceylan et al, 2007Ceylan et al, 2009;Far et al, 2009) …”

Utilising neural networks and closed form solutions to determine static creep behaviour and optimal polypropylene amount in bituminous mixtures

“…Xiao and Amirkhanian 51 in another look explored the utilization of the artificial neural networks in predicting the stiffness behaviour of rubberized asphalt concrete mixtures with reclaimed asphalt pavement. A paper by Far et al 52 presented outcomes from a research effort to develop models for estimating the dynamic modulus (|E*|) of hotmix asphalt layers on long-term pavement performance test sections. Tapkın et al 20 presented an application of neural networks for the prediction of repeated creep test results for polypropylene modified asphalt mixtures.…”

Optimum Design of High-Strength Concrete Mix Proportion for Crack Resistance Using Artificial Neural Networks and Genetic Algorithm

“…Artificial neural networks (ANN) have already been applied in the field of concrete research, mainly to predict various properties of concrete (Yeh, 2007;Far et al, 2009;Mangalathu et al, 2018;Abellan Garcia et al, 2020). For example, Chithra applied regression analysis and ANN, respectively, to predict the compressive strength of concrete mixed with copper slag.…”