Considering the developments of new accurate constitutive models and numerical methods, a comparison between the classical and modern numerical methods of dynamic analysis of earthfill dams seems necessary. A three steps procedure was followed in this study. First, the accelerations and displacements of a typical homogenous dam were calculated using Sarma, Makdisi-Seed, and Sarma-Ambraseys methods, the results were compared with the results of Quake/W and Slope/W software using both linear elastic and equivalent linear models. Then, a special code was developed to calculate the deformations of a given wedge of the dam body. The studies showed that using linear elastic behavior model, the critical acceleration estimated in classic methods is high and leads to oversized design. However, the more realistic constitutive models give better results. Also, in classic methods the shape of the wedges was unrealistic as it was assumed to pass through the core of the dam, while in numerical analysis it was shown that the critical wedge passes through downstream shell. Therefore, the critical acceleration of wedges was different from the predictions of the classic models, for example it was estimated to be 0.41g in the Ambraseys-Sarma method compared to 0.31g in the linear method for a wedge depth of 0.6h. It was concluded that numerical methods with more realistic constitutive models must be used. Finally, the Kheyrabad embankment dam was analysed using the developed method. It was concluded that the dam would show a permanent deformation of 4.0 cm under scaled Naghan earthquake.
Oulapour et al. (Sun,) studied this question.