Rutting in flexible pavements is often associated with permanent deformation of the unbound granular layer. The current permanent deformation models are applicable only to a single stress path in repeated-load tri-axial (RLT) tests, in which the load pulses are of constant amplitude. In this paper, a general approach using the time-hardening concept was introduced to model the permanent deformation of unbound granular materials (UGMs) continuously in multistage (MS) RLT tests, in which load pulses of a range of different amplitudes are applied, to represent field conditions realistically. With this formulation, three existing permanent deformation models were reconstructed, and one of the models was slightly modified, to suit MS loading conditions better. The material parameters of these models were then optimized for three UGMs used in pavement construction with data from MS RLT tests and application of a least squares curve-fitting method to the test data. The goodness-of-fit statistics were computed to evaluate and compare the quality of fit achieved with these models. The shakedown ranges were also calculated for each stress path of the MS RLT tests to compare these models in simulating these ranges. Generally, the time-hardening approach, for which the quality of fit is dependent on the chosen model, was found to work successfully. In this study, the selected models performed quite well, with the modified model showing the closest agreement to the test data. Thus, this approach has the potential to be applied for better modeling and prediction of pavement performance.

To reliably predict the permanent deformation behaviour of unbound granular materials (UGM) in a pavement structure, the material parameters of the constitutive models used in design should be evaluated using a multi-stage (MS) loading approach. This paper investigated the prediction of the accumulation of permanent strain in UGM using some current models, extended applying the time-hardening approach, based on MS repeated load triaxial tests (RLTTs). The material parameters of these models were optimised for five different UGM used in pavement construction using the MS RLTT data with a specific set of stress levels. With these models, the accumulations of permanent strain in the same materials for MS RLTTs with a different set of stress levels were predicted. Using this approach, three out of the four models performed very well, which may be further developed for field conditions for better prediction of rutting.

One of the key components in analysing pavement structural behaviour is the response model which is used to estimate the stresses, strains and displacements of the pavement structure subjected to the existing traffic, taking into account the material properties and prevailing environmental conditions. Multilayer elastic theory (MLET) is often preferred over other methods such as the finite element method, due to its computational performance for repeated applications. A new elastic response analysis program has been developed based on the Burmister MLET theory to calculate the response of flexible pavement structures. In the development of the program, the time-consuming part of MLET processes was optimised. To improve the convergence and accuracy of responses in the vicinity of the surface of the top layer, an approach based on Richardson's extrapolation was employed. Moreover, an iterative approach to model stress dependency of unbound granular materials was incorporated. A comprehensive comparison of the program with two frequently used programs demonstrated an excellent agreement and improved performance.

A simple model has been proposed to characterise the accumulation of permanent deformation (PD) in unbound granular materials (UGMs) under cyclic loading of variable magnitudes. The model was developed based on multi-stage (MS) repeated-load triaxial (RLT) tests. The material parameters of this model can be evaluated using an MS RLT test. The model was validated by calibrating it for a few UGMs with a range of grain size distributions, moisture contents and degrees of compaction. The calibrated model was further validated by predicting the PD behaviour of some of these UGMs for different stress conditions. Generally, quite satisfactory predictions were obtained with this model with the advantage of reduced effort required for its calibration compared to some existing models. Additionally, the sensitivity of the parameters of this model to moisture, degree of compaction and grain size distribution was investigated with the aim of incorporating them into the model in future.