Comparing Performance of a New Statistical Model with Contemporary Models of Unbound Basecourse Permanent Deformation

Abstract

The use of unbound granular materials (UGM) in road pavements is very common in New Zealand (NZ) and sparsely populated countries as they are economical and provide a good load bearing foundation for the wearing course.  The load bearing capacity of UGM materials defines the performance of these types of pavements, which are normally used on the lower traffic volumed roads spectrum. The most common performance test for unbound granular materials is the repeated load triaxial (RLT) test that simulates the dynamic loading of traffic on UGM in actual pavements.  Given that this test can be a good indicator of the actual performance of these materials under simulated cyclic loading, forecasting models have been developed using RLT data. In this paper, various statistical models based on single input criteria (number of loading cycles, or stress state) as well as combined effects of loading and stress criteria are assessed for the RLT test data obtained in this research. A comparison of how existing models fit the empirical data is undertaken. Furthermore, an alternative model is proposed that takes account of both the number of loading cycles and stress conditions in the sample. This new model shows promising results, especially since it is able to predict aggregate performance for different moisture and drainage conditions.