Numerical analysis of rapid drawdown: Applications in real cases

In this study, rapid drawdown scenarios were analyzed by means of numerical examples as well as modeling of real cases with in situ measurements. The aim of the study was to evaluate different approaches available for calculating pore water pressure distributions during and after a drawdown. To do that, a single slope subjected to a drawdown was first analyzed under different calculation alternatives, and numerical results were discussed. Simple methods, such as undrained analysis and pure flow analysis, implicitly assuming a rigid soil skeleton, lead to significant errors in pore water pressure distributions when compared with coupled flow-deformation analysis. A similar analysis was performed for the upstream slope of the Glen Shira Dam, Scotland, and numerical results were compared with field measurements during a controlled drawdown. Field records indicate that classical undrained calculations are conservative but unrealistic. Then, a recent case of a major landslide triggered by a rapid drawdown in a reservoir was interpreted. A key aspect of the case was the correct characterization of permeability of a representative soil profile. This was achieved by combining laboratory test results and a back analysis of pore water pressure time records during a period of reservoir water level fluctuations. The results highlight the difficulty of predicting whether the pore water pressure is overestimated or underestimated when using simplified approaches, and it is concluded that predicting the pore water pressure distribution in a slope after a rapid drawdown requires a coupled flow-deformation analysis in saturated and unsaturated porous media.