Statistical modelling of pavement friction: analyzing the effects of texture characteristics and measurement resolution

This study aims to develop robust predictive models for the friction coefficient (μFAP) by examining the influence of pavement texture characteristics, measurement resolution, and sample treatment histories. Road pavement samples were subjected to varying treatments, including natural weathering and controlled polishing using a Friction After Polishing (FAP) device, to simulate traffic effects.The analysis demonstrates that higher texture measurement resolutions significantly enhance the predictive accuracy of μFAP, with models using parameters from 200 µm, 500 µm, and 1000 µm resolutions explaining 80-90% of the variability in friction coefficients. In contrast, models based on lower resolutions of 10 µm and 100 µm performed less effectively. Additionally, consistent treatment of samples led to more reliable predictions, highlighting the importance of uniform sample conditions in friction coefficient estimation. These findings underscore the potential of non-contact measurement techniques for real-world applications in assessing road safety.