Determination of thermal conductivity of asphalt paving mixtures using finite element method

Abstract

Thermal conductivity k is an important input parameter to the study of pavement thermal behaviors in a number of practical engineering problems. Laboratory determination of k requires special-purpose devices which are commonly not available in a standard pavement engineering laboratory. Another complication is that the k of an asphalt mixture changes as its air voids content reduces with time under traffic loading. This makes laboratory determination of k impractical for analyzing problems related to pavement thermal behavior. To overcome the mentioned difficulties, this study develops a computer-aided procedure to determine by means of a numerical model the k of asphalt mixtures with known volume proportions of mix constituents. The asphalt mixture analyzed is first represented by a two-dimensional row-column structure with a unit thickness, with each constituent randomly assigned to occupy the number of cells based on its volume proportion in the mix. Next, based on the row-column structure, a computer simulation model of heat conduction in an asphalt mixture is developed to derive the k of the mixture studied. In cases where the k values of asphalt binder and aggregate are unknown, the model can be calibrated using genetic-algorithm to obtain all the needed k values. Once calibrated, the numerical model can be repeatedly applied to determine the k of the asphalt mixture with different air voids contents, with no further requirements for laboratory testing. Examples are presented to demonstrate that the proposed model is able to estimate k of asphalt mixtures with accuracy within practically acceptable limits.