This paper deals with an extensive research program into the upgrading of the fatigue transfer functions of asphalt mixes in use in South Africa. Parts of the program have recently been completed and the aim of the paper is to summarise and discuss the results obtained to date. Two stages of the program are discussed in this paper: a comparison of stiffnesses of asphalt mixes obtained from different sources, and the evaluation of the fatigue life of laboratory asphalt mixes and development of laboratory fatigue regression models.

In the first of the above-mentioned stages, stiffnesses of asphalt mixes were determined in the laboratory (using the Indirect Tensile Test) and from two theoretical stiffness prediction models (Shell method and Asphalt Institute method). Attempts were also made to backcalculate field stiffness using the Multi-Depth Deflectometer (MDD) results of four Heavy Vehicle Simulator (HVS) field tests. In the second stage fatigue tests were undertaken in the third- point beam fatigue test apparatus in the constant strain mode at various strain levels. Regression models were developed for the following independent variables: strain, temperature, initial flexural stiffness, load frequency, aggregate grading and binder type.

The findings indicate that the Shell BANDS- program and the Asphalt Institute's equation (AI) can be used for an approximate determination of asphalt stiffness, provided that the methods are used correctly. There is also some indication that there is a correlation between the resilient modulus obtained in the Indirect Tensile Test and the actual effective field stiffness, but this has to be verified by further studies. In the second part of the investigation it was found that the fatigue regression models obtained from this study gave a higher life than other models such as those of the Asphalt Institute, TRRL and NCHRP. The coefficients of determination obtained were, however, very good.