A prediction model of the friction coefficient of asphalt pavement considering traffic volume and road surface characteristics

ABSTRACT

In order to build an appropriate prediction model of pavement friction coefficient attenuation, the effects of aggregate texture, on a basis of texture scanning and pavement testing of the dynamic friction coefficient, traffic volume, and rock characteristics on pavement slide decay resistance were studied, and the accuracy of the developed prediction model of friction coefficient was thereafter compared and analysed. First, based on the tire-pavement dynamic friction analyzer independently developed by the research group, accelerated loading tests under different traffic conditions were conducted by the orthogonal test method, and the dynamic friction coefficient of pavement during the abrasion process was recorded. At the same time, the non-contact laser profile measurement system was used to collect three-dimensional micro-texture data of coarse aggregate surface at different abrasion stages, and characteristic parameters such as height, wavelength, and shape were calculated. Second, the factors such as traffic volume and rock characteristics were incorporated, and the algorithms such as multiple linear stepwise regression, feedforward neural network, and random forest regression were adopted to develop estimation models of pavement friction. The results showed that the model constructed based on only texture feature parameters poorly represented pavement skid resistance. After introducing the two indicators of total traffic volume and rock type, the correlation coefficients of the model significantly improved, reaching 0.63, 0.58, and 0.82 respectively, indicating that the latter two factors have significant effects on the anti-skid attenuation of pavement. Moreover, the above three models can achieve satisfactory prediction results. In addition, when a large sample size can be obtained, a random forest algorithm is recommended to acquire higher prediction accuracy of pavement friction coefficient attenuation.

KEYWORDS:

• Friction coefficient
• traffic volume
• pavement texture
• accelerated loading test
• machine learning

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by China Postdoctoral Science Foundation [grant number 2018M633444]; National Natural Science Foundation of China [grant number 51608085]; Open Fund of Key Laboratory of Road Structure and Material of Transport Ministry, Chang'an University [grant number 300102210506]; Natural Science Foundation of Chongqing Municipality [grant number cstc2021jcyj-msxmX0554]; Key R & D Program of Guangxi [grant number GUI KE 22 AB20159036]; Shaanxi Province Postdoctoral Science Foundation [grant number 2018BSHEDZZ123].