Correlating pavement texture, noise and friction properties

Access full-text files




Li, Ruohan (M.S. in civil engineering)

Journal Title

Journal ISSN

Volume Title



This thesis uses texture, friction, and noise data collected along eight asphalt pavements with different surface types across Texas to explore the intercorrelation between the three properties, both within each pavement surface type and across different types. It was found that across all surface types, the entire frequency band of noise from 400 to 5000 Hz correlates the strongest with texture of wavelengths from 31.5 mm to 2.5 mm positively. This means that regardless of the surface type, pavements with a higher texture level in the wavelength spectrum of 31.5 mm to 2.5 mm tend to generate a higher level of noise in the frequency band of 400 to 5000 Hz. When noise is broken down into 1/3 octave bands in frequency, the strongest positive correlation is found between noise of 630 Hz and texture of 50 mm wavelength. A negative correlation, however, is found between higher frequency (f > 1000 Hz) noise and shorter wavelength ([lambda] < 10 mm) texture. The slope of noise vs. texture is similar across different pavements, but the intercept can be different, indicating that with a unit increase in texture level, the additional noise generated by different pavement types is of similar magnitude, but they might be at different levels of loudness given the same texture level. Across all pavement types, when texture level is the same, pavements surfaced with thin overlay mixtures (TOM) tend to generate a consistently lower level of noise at both high and low frequencies. While no strong correlation was found between noise and friction, this finding is consistent with the conclusions from studies by previous researchers. The correlation between friction and texture using the original data has not been found to be strong, which can be partially due to the inconsistency in location of the corresponding measurements. With the capability of measuring texture and friction simultaneously to ensure that the data are collected under the same condition and location using the equipment developed at UT Austin, a much stronger correlation between friction in terms of Grip Number (GN) and texture in terms of root mean square (RMS) was found. Speed, meanwhile, also plays an important role in predicting friction, with a significantly negative coefficient in the model. Statistically different friction levels are also observed among different mix types of pavement surface when other variables are held constant, indicating that different surface types can provide different levels of friction given the same texture at the same speed.


LCSH Subject Headings