ABSTRACT
This paper contains the results of the reanalysis of the previous work by Patron (1989) as published in Bergado et al (1992). In this study, empirical autocorrelations rather than idealized functions were used. In the case study, six slope failures were observed at random locations along a 10 km embankment adjacent to an irrigation canal. The irrigation canal, 20 m wide and about 2.5 m below mean sea level (msl) at its deepest runs adjacent and parallel to the embankment. Slope failures occurred when the embankment was raised to 2.05 m above msl from an average elevation of 1.7 m above msl coupled with the lowering of the canal water level at the end of the dry season. The spatial variability of undrained strength, the actual variation in embankment geometry, and the varying water level in the canal were considered in the analyses. An analysis using the plane strain factor of safety based on deterministic soil profile defined by the mean undrained strength resulted in the prediction favoring a reverse failure pattern along the embankment. Using probabilistic analysis, the results conform to the actual failure pattern along the embankment. The use of empirical autocorrelation function (ACF) seems to confirm and explain better the failure zones than utilizing idealized theoretical ACF.
