Subgrade soil characterization in flexible pavement design is
necessary for estimating the structural capacity of flexible pavements
and then determining tbe pavement layer thicknesses. Several non-linear
and linear models have been used to characterize the subgrade
soils. The current bulk and deviatoric stress models have limited
applications since they are essentially developed for a particular
soil type. Few other models are restricted by the multiple collinearity
problems in the regression analysis. A regression model with confining
and deviatoric stresses as attributes is adopted in this paper
to analyze the repeated load triaxial test results. This model
is simple, easy to analyze the repeated load triaixial test results
and does not have any structural type collinearity problems. This
paper presents the regression model analysis results on different
subgrade soils including mixed soil types. The repeated load triaxial
tests conducted on granular, cohesive and mixed soil types at
various moisture content levels are first analyzed to determine
the model constants. The model constants are then correlated with
the basic soil properties. The correlations are evaluated in their
reliability in predicting the resilient properties of the same
subgrade soils. This evaluation, which is fairly successful, showed
that this approach can be used to predict the resilient properties
of similar type soils. This approach will save significant project
costs since core sampling and resilient modulus testing tasks
can be averted.