Numerical and Experimental Investigation into the Effect of Retention Properties of Partially Saturated Soil for the Stability Analysis of River Embankments

Stability assessment for river embankment represents a fundamental issue in safety assessment towards geo-hazard; several topical issues are involved with this study. Primary, unsaturated soil conditions are experienced during the entire lifetime of these structure; transient analysis are often required in regards to time-variability of external conditions; uncertainties related to soil hydraulic, mechanical and retention parameters are intrinsic for this problem, as well as those related to the knowledge of initial conditions. To overcome these problems, several simplified hypothesis are usually adopted both for the design and assessment of embankments. With the purpose to comprehensively deal with main part of the involved geotechnical issues, numerical and experimental investigations has been performed for the stability analysis of river embankments. The effect of retention properties of partially saturated soils on the hydraulic and stability conditions have been investigated, accounting for hydrometric water-level time-variability; numerical studies have been performed with increasing level of sophistication, using different mathematical methodologies for probabilistic numerical analysis. Various experimental studies and site monitoring data are presented, sharing the purpose of investigating retention properties for partially saturated soils. In its entirety, the final purpose of this study is to analyse the assessment process of riverbanks stability towards global instability mechanism triggered by variable hydraulic boundary conditions. Indications and suggestions, which can eventually be referenced for similar problems or implemented in the standard engineering practice, are provided; the use of site monitoring is justified to improve present flood early-warning system able to define in continuous progressive alert level combining measured and forecasted data with numerical probabilistic analysis results.