Table of Contents Author Guidelines Submit a Manuscript
The Scientific World Journal
Volume 2014, Article ID 515467, 16 pages
http://dx.doi.org/10.1155/2014/515467
Research Article

Prediction of Frequency for Simulation of Asphalt Mix Fatigue Tests Using MARS and ANN

Department of Civil Engineering, K.N.Toosi University of Technology, No. 1346, Vali Asr Street, Mirdamad Intersection, Tehran 19967-15433, Iran

Received 25 August 2013; Accepted 24 October 2013; Published 4 February 2014

Academic Editors: N. Avdelidis and J. Zheng

Copyright © 2014 Ali Reza Ghanizadeh and Mansour Fakhri. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Fatigue life of asphalt mixes in laboratory tests is commonly determined by applying a sinusoidal or haversine waveform with specific frequency. The pavement structure and loading conditions affect the shape and the frequency of tensile response pulses at the bottom of asphalt layer. This paper introduces two methods for predicting the loading frequency in laboratory asphalt fatigue tests for better simulation of field conditions. Five thousand (5000) four-layered pavement sections were analyzed and stress and strain response pulses in both longitudinal and transverse directions was determined. After fitting the haversine function to the response pulses by the concept of equal-energy pulse, the effective length of the response pulses were determined. Two methods including Multivariate Adaptive Regression Splines (MARS) and Artificial Neural Network (ANN) methods were then employed to predict the effective length (i.e., frequency) of tensile stress and strain pulses in longitudinal and transverse directions based on haversine waveform. It is indicated that, under controlled stress and strain modes, both methods (MARS and ANN) are capable of predicting the frequency of loading in HMA fatigue tests with very good accuracy. The accuracy of ANN method is, however, more than MARS method. It is furthermore shown that the results of the present study can be generalized to sinusoidal waveform by a simple equation.