• August 24th, 2023

Strength loss in high plasticity clay soils is a common occurrence along roadways in western and central Alabama and has considerable impact on pavement distress and slope failures. Approximately $16 million was spent to repair slope failures in western and central Alabama between 2005 and 2015 with many of these failures occurring in areas with high plasticity Prairie clays. Selecting strengths for these soils to use in slope stability analyses is often a key source of uncertainty and the selected strengths must account for effects of loading conditions and any potential changes in strength over time. For high plasticity clay, such as the Prairie clays in Alabama, repeated cycles of wetting and drying can reduce the available drained shear strength to the fully softened condition before large movements occur. This reduction in strength can lead to failure without any change in loading and accounting for this potential strength loss when analyzing the stability of these slopes is critical to obtain accurate results. The torsional ring shear test can measure both the fully softened strength (used for first time failures) and the residual strength (used for ongoing or reactivated failures) of high plasticity clays. For this study, clay samples were collected at six landslide sites around Alabama. Samples were tested to determine the index properties, electrical resistivity, mineralogy, and fully softened and residual strengths. The results from these tests were compared with existing correlations to determine which correlations were the most appropriate for use in Alabama. For three of the sites, slope stability analyses were performed to compare with observed performance at the site. The results show that the strength envelopes from ring shear testing are consistent with the observations at all three sites, although nonlinear envelopes are needed for some sites. Recommendations are provided for correlations that can be used to estimate strengths when ring shear testing is not available and for cases where existing correlations do not provide good estimates.

Read the full study here.