Pavement texture characterisation using wavelets analysis in relation to pendulum skid tester

Abstract

ABSTRACT Pavement texture is an important characteristic affecting user safety and satisfaction. Despite the large number of studies on pavement texture, there is still a need to further explore the models relating texture to tyre-pavement interaction. In this study, 29 cores were extracted from 15 different pavement surfaces and tested in the laboratory by Ohio department of transportation. The data set includes high density texture scans, modified mean texture depth (MMTD) values estimated using a modified sand patch test, and friction measurements acquired using dry and wet British pendulum (BP) testes. The texture scans were characterised using the mean profile depth (MPD) and wavelets energy. The correlation between MPD values and the coefficients of friction (COF) estimated using BP tests were not as strong as the correlation between the wavelet energies and the wet and dry COFs. Statistically significant models were derived using the wavelet energies to predict the MMTD (R 2 = 0.94) as well as the dry and wet COF (R 2 Dry = 0.49/R 2 Wet = 0.60). The principle component analysis was used to decorrelate the wavelet energies and overcome the multicollinearity, while the least absolute shrinkage and selection operator was used to reduce the number of variables in the model.