A class of inhomogeneous shear models for seismic
response of dams and embankments
Panos Dakoulas and George Gazetas
Department of Civil Enffineering, Rensselaer Polytechnic Institute, Troy, N. Y 12180, USA
Results of linear and nonlinear static analyses of gravity-induced stresses in several typical dam cross-sections,
in conjunction with published experimental correlations of shear modulus versus confining pressure for a
large variety of soils, reveal that the average shear modulus across the width of earth/rockfill dams may be
expressed as a power m of depth, with m ranging from 0.35 to 0.90 and depending on material and geometric
parameters. A general inhomogeneous shear beam model is developed to account for any possible such
variation of modulus with depth. Perhaps somewhat surprisingly, closed-form analytical expressions are
derived for natural frequencies, modal displacements, participation factors, and steady-state response
functions for all values of the inhomogeneity factor m. Parametric results are presented in tabular and
graphical form and conclusions are drawn of practical significance. Finally, a comprehensive comparative
study is undertaken to investigate the validity of the inhomogeneous shear beam (SB) models. For five
different dam cross-sections, each excited by four recorded accelerograms, it is shown that plane-strain finite-
element analyses yield fundamental periods and peak displacements within the dam which are in very good
accord with the predictions of a 'consistent' inhomogeneous SB model. A companion paper 1 extends the
present work and focuses on seismic shear strains and seismic coefficients within dams and embankments.
INTRODUCTION
The shear beam (SB) model has been used extensively over
the years to estimate the lateral seismic response of earth/
rockfill dams and embankments, 2-7 its popularity
stemming mainly from its simplicity. A number of
rigorous studies a ~o have largely corroborated one of the
crucial assumptions of the SB model, namely that
horizontal shear strains, shear stresses and displacements
are essentially uniformly distributed across the width of
the dam, and have moreover shown that the SB model
yields natural periods and modal shapes which are quite
realistic. These conclusions are also confirmed by the
results of this and the companion paper. 1 To account for
the dependence of soil stiffness on confining pressure, an
improved version of the SB model has been developed
which considers the (average across the width) shear
modulus as increasing with the 2/3 power of the
depth. ~°'~1 Direct and indirect evidence, including in-situ
measurements of S-wave velocities and observations of
the actual response of several dams 2° 12 have shown the
general validity of such a modulus variation with depth.
Other researchers 13 15 have also suggested a shear
modulus increasing with depth, although not quite as
steeply as with the 2/3 power.
This paper first outlines the results of a comprehensive
investigation on the factors influencing the variation of
stiffness in earth and rockfill dams. It is concluded that the
(average across the dam) shear modulus may be
realistically considered as increasing with the m power of
the depth z, where m may usually take values in the range
0.40<~m~<0.75. Then an exact analytical solution is
derived for free and forced vibrations of a truncated shear
wedge (as sketched in Fig. 1) characterized by an arbitrary
value of the power m. The third objective of the paper is to
Accepted May 1985. Discussion closes December 1985.
0267-7261/85/0401661752.00
© 1985 CML Publications
166
present the results of a comprehensive comparative study
in which the seismic response of several dam sections are
computed by plane finite-element (FE) and by inhomo-
geneous SB analysis. For each dam section the l~ower m of
the SB model is selected such that its modulus variation is
consistent with the spatial variation of stiffness in the
specific plane-strain model. It is concluded that such a
'compatible' inhomogeneous SB predicts fundamental
periods and peak seismic displacements which are in a
very close agreement with those computed with the
corresponding FE models. T~he present study is further
extended in a companion paper 1 which focuses on
distribution of peak seismic shear strains and seismic
coefficients in dams and embankments.
MODULUS VARIATION WITH DEPTH
While several (material, geometry and excitation related)
factors influence the spatial distribution of effective shear
moduli within a dam, it has been found in the course of
our investigation that the average modulus, G(z), over a
horizontal plane at a depth z from the origin (Fig. 1),
depends mainly on three factors:
(a) the dependence on confining pressure of the shear
modulus Ge of each constituent material
(b) the size and the relative overall stiffness of the
impermeable cohesive core
(c) the inclination of the slopes and the truncation ratio.
The effect of these factors on G(z) is quantitatively
illustrated in this section.
The shear modulus of a particular soil element at small
levels of shear strain can be expressed as:
Ge = F(e,OCR). a~'~ (1)
where: a~t =the effective normal octahedral stress; and e
Soil Dynamics and Earthquake Enoineerino, 1985, Vol. 4, No. 4