The impact of data noise on permanent deformation behaviour of asphalt concrete mixtures

Abstract

Flow number (FN) test, which evaluates permanent deformation (PD) of asphalt concrete (AC) mixtures, contains data noise that leads to a false representation of the tertiary stage of flow. Most studies, however, ignore this aspect and have not quantified the amount of noise in the data. This, in turn, leads to unrealistic characterisation, ranking of AC mixtures and correlation of FN with other factors such as mix volumetric, aggregate parameters and dynamic modulus. As such, this study quantifies the extent of noise in the FN data and its consequent impact on characterisation of PD behaviour of AC mixtures. Twelve AC mixtures were prepared with different aggregate gradations and binder grades, and FN test was performed at a single temperature. Statistical analysis confirms that the corrected FNs and the original FNs are statistically different from each other. Moreover, all of the AC mixtures are characterised with and without data noise, and it is concluded that noise in the data has a significant influence on the ranking of AC mixtures. Furthermore, the results indicate data noise either reverses the expected sign or reduces coefficient of determination of correlation. The results of this study would have insightful implications on findings of earlier studies.