A series of dynamic centrifuge model tests was conducted in order to evaluate some factors affecting the seismic performance of hillside
embankments consisting of sandy or silty soils and resting on stiff base slope. The effects of seepage water elevation in embankments, toe drain,
embankment height, base slope inclination, soil compaction, and fill materials on the seismic behaviour of embankments were investigated. The
test results showed that: the seepage water was one of the most important factors for earthquake-induced embankment failure; the seismic
performance of both the smaller and higher embankments was remarkably improved by installing the toe drain; larger base slope inclination
produced larger earthquake-induced deformation of embankments; well-compacted embankments were not vulnerable to earthquake-induced
damage; and the seismic performance of well-compacted embankments consisting of well-graded silty soils with large fines content was
higher than that of poorly graded sands under otherwise the same condition. In some tests, as observed during past strong earthquakes, delayed
flow failure occurred due possibly to the multiple effects of upward seepage associated with the redistribution of excess pore water pressures
generated during main shaking, continued small vibration after main shaking, and driving static shear stresses caused by the embankment weight.
A series of triaxial compression and cyclic triaxial liquefaction tests was also conducted to evaluate undrained behaviour of the fill materials. The
correlation between the model and laboratory element test results was presented.