A novel noncontact method for the pavement skid resistance evaluation based on surface texture. (January 2022)

• Record Type: Journal Article
• Authors: Lu, Jiale; Pan, Baofeng; Liu, Quan; Sun, Minghao; Liu, Pengfei; Oeser, Markus
• Is Part Of: Tribology international. Volume 165(2022)
• Publication Date: 2022-01
• Subjects: Pavement skid resistance -- Pavement surface texture -- British Pendulum Test -- Convolutional neural network; Tribology -- Periodicals; 621.89
• DOI: 10.1016/j.triboint.2021.107311
• Languages: English
• ISSNs: 0301-679X

Abstract

Abstract: A convolutional neural network is proposed to feature the primary relationship between pavement surface texture and in-situ skid resistance measurement namely British Pendulum Test. The influence of texture collection interval was firstly analyzed. Then, effective contact textures were extracted. Finally, the sample size required for model training, which could serve as a reference for further texture-related research was addressed. Results indicated that textures with wavelengths above 2.40 mm is key for the wet friction. Textures below the cross section whose area is 0.6 times the nominal area do not contact the rubber. The sample size should be more than 100. The newly developed non-contact method shows high feasibility in predicting the skid resistance and effectively control the relative error within 14%. Highlights: A novel non-contact method for skid resistance evaluation is proposed in the paper. Textures with wavelengths above 2.40 mm are critical for the wet friction. Textures below the cross section of 0.6 A do not have contact with the rubber. The sample size of 100 is required for model training. The newly developed method shows high feasibility and precision.