Empirical Assessment of Backward Erosion Piping via Blanket Thickness, Lower Mississippi Valley

Abstract

Backward erosion piping (BEP) is a common hazard along artificial levees in the Lower Mississippi Valley (LMV). This study applies an empirical assessment of overbank deposit thickness, also referred to as blanket thickness, and the probability of BEP initiation within the LMV. A dataset composed of 800 boreholes, 500 piezometers, and 2,691 sand boils is compiled from US Army Corps of Engineers (USACE). The data are assessed using a staged evaluation methodology. Results indicate blanket composition has no clear relationship with BEP initiation. Blanket thickness correlates with BEP initiation, even when variations for vertical hydraulic gradient are accounted. BEP also appears to initiate more readily than expected through thin blankets. Finally, using a Naïve Bayes algorithm, a model is presented indicating probabilities of BEP initiation as a function of blanket thickness and factor of safety (FS). The model indicates blanket thickness influences probability of BEP initiation independently from FS. The model could prove useful for BEP hazard assessment along levees as both an evaluation and design tool.