Asphalt Pavement Skid Resistance Prediction Based On In-situ High-resolution Three-dimensional Surface Texture Data

Traffic accidents are closely related to pavement surface skid resistance,which gives rise to a high degree of adhesion coefficient for tires,contributing to stability,reliability,and controllability for traveling vehicles,and lowering down the traffic accident rate.Pavement texture highly correlates to the skid resistance.A high-quality texture morphology can help to drain standing water and break through the water film,thus to improve the tire pavement contact condition and reduce the vehicle hydroplaning risk.Equipped with the high-resolution3D(three-dimension)laser scanning technology,this study obtains the high-resolution pavements’ in-situ 3D texture point-cloud data,and investigates the inner relationship between texture and skid resistance,thus to predict the pavement skid resistance based on the areal 3D areal texture parameters.Based on the LTPP SPS10(Long-term pavement performance;Specific pavement study)sites in Oklahoma state,USA,the LS-40 portable 3D surface analyzer and DFT(Dynamic friction tester)is adopted to acquire pavement surface data from 2015 to 2017,which contains108 data sets of pavement surface texture and pavement skid resistance data.This research is conducted as follows:1.the conventional pavement surface texture and skid resistance testing method are introduced,the limitations of the conventional pavement surface texture characterizing system is pointed out,thus the novel 3D areal texture characterizing frame is proposed,the pavement skid resistance multi-regression prediction model is established,and the pavement surface textures’ influence principle on skid resistance is investigated based on 3D areal texture indicators.2.the pavement skid resistance prediction model is established via support vector machine(SVM)and artificial neural network,which are both machine learning methods.It is revealed that both models’ prediction ability is stronger than the multi-regression prediction models and the models in the precious studies.It is also found that the artificial neural network reveals better performance than the support vector machine in this case.Furthermore,it is shown that non-linear relationship exists between the pavement surface texture and skid resistance,from which confirming the effectiveness of predicting the pavement skid resistance based on the 3D areal texture indicators.3.a pavement 3D texture FE model is re-constructed based on high-resolution 3D texture point-cloud data.A coupled rubber-pavement interaction FE model is established and validated according to the DFT testing mechanism.The rubber-pavement interface friction mechanism is simulated by adopting exponential decay friction model,of which the kinetic friction parameter,static friction parameter and decay coefficient are back-calculated using binary search approach.The high-resolution 3D texture data is separated into macro-and micro-scales using butterworth filtering,and various areal texture indicators are calculated at macro and micro levels.4.the correlation among the 3D areal texture indicators is evaluated at macro and micro levels respectively.The principal component analysis regression is conducted to quantify the relationship between texture characteristics and the interface friction parameters.It is found that the texture indicators have superposition and joint effect on the skid resistance.The texture’s influence on pavement skid resistance varies with the texture characterizing manner changing from macro level to micro level.The influence of the interface friction and the rubber material model on the pavement skid resistance is discussed,and the significance analysis of each influencing factor is carried out.It is found out that influence sequence is as follows:kinetic interface friction coefficient,rubber material model,static interface friction coefficient,interface friction exponential decay coefficient.5.the two-dimension tire model of the locked-wheel friction tester is established,of which the tire geometric parameters are adopted from the ASTM E524 standard.The tire’s threedimension FE(finite element)model is established based on the ABAQUS software platform.The tire’s linear elastic property parameters are acquired from the tire tread imprint analysis,which is conducted under static loading tests after the tire’s inflation simulation.The coupled tire-water film-pavement FE model is validated through the energy conservation,water flowing tracing,and NASA hydroplaning equation;6.the field friction data acquired by DFT is considered as the inputting parameters of the coupled tire-water film-pavement FE model and thus the tire’s skid resistance in different sliding speeds is computed by inputting corresponding friction data.The tire’s braking distances are calculated based on the newton’s second law of motion,and validated by comparing the previous research and AASHTO(American Association of State Highway and Transportation Officials)standards.Considering the 3D areal texture indicators as explanatory variables,the tire braking distance prediction model is established via the artificial neural network method,and it is found out that the pavement’s 3D areal texture indicators can highly explain the tire braking distance.This section provides a reference for tire braking distance prediction in terms of a contactless manner;This research provides a theoretical basis for the intelligent evaluation of the contactless pavement skid resistance,significantly contributing to informationizing,automizing and intelligentizing the pavement safety management.It also proposes a guide for pavement design,material selection,and contribute new ideas for enhancing road traffic safety management and improving pavement maintenance level.