Critical gradient for internal erosion in earthen d ams : a comparative analysis of two predictive methodologies

Abstract

Minimizing the uncertainty in predicting the critical gradient of a dam (i.e. the critical reservoir pool level) is important during the risk analysis of dams. Uncertainty leads to inexact relative risk in portfolio management; therefore it is essential to get as accurate a risk estimation as possible for each project in a portfolio. To understand the uncertainty inherent in the predictive methodologies, this thesis sets out to compare the two most commonly used predictive methodologies, Sellmeijer and Schmertmann, in the USACE portfolio in order to make a suggestion of when to use which. Both methodologies have been calibrated for a small range of ideal soil characteristics that may not reflect of existing conditions of the portfolio. This thesis concludes with the recommendation to broaden the range of applicability through additional experiments that include anisotropic conditions.