Temporal and spatial temperature predictions for flexible pavement layers using numerical thermal analysis and verified with large datasets

Abstract

The pavement temperature patterns play a significant role in the performance of flexible pavements containing asphalt concrete surface layers for several reasons. Since asphalt concrete behaves as a viscoelastic material at medium to high service temperatures, the structural response of the pavement is affected by the temperature history and the magnitude and duration of the wheel loads applied on the pavement. Therefore, predicting the pavement temperature patterns in a scalable manner (i.e., daily, seasonal, annual) is vital for a more accurate simulation of pavement response and performance, particularly the permanent deformation (rutting). This can result in more effective pavement designs. A numerical model was developed in this paper to predict the temperatures in asphalt layers by using the heat transfer theory and the Finite Element method with the help of proprietary software. A model that can predict the pavement temperatures continuously for one year covering all four seasons at time intervals that allow pavement distress prediction for each axle load cycle and calibrated with high-quality measured data is the main contribution of this study. Among the parameters that significantly affect pavement temperature prediction, the convective heat transfer at the air-pavement interface is of utmost importance. Therefore, the Convection Heat Transfer Coefficient (CHTC) for the air-pavement interface was also selected as an exploratory focus area for this paper. Two approaches were used to estimate the CHTC needed for the heat-transfer model. An error analysis was conducted to assess the accuracy of the pavement temperature prediction by using a large, measured dataset from a field test. The error of temperature predictions for the whole year was found to be minimum when the value of CHTC was equal to 21 W/(m2K). The error observed in winter predictions with the wind speed model of CHTC was minimized when a constant value of 60 W/(m2K) was used for CHTC. The significance of the convective heat transfer component in the FE model was identified through the improvements obtained by adapting the value of CHTC. Correlation analysis also identified a connection between the prediction error and relative humidity. Further research is underway to improve the temperature prediction model using additional parameters (i.e., relative humidity, cloud cover) and more refined data analytics approaches.