This paper presents the results of large scale physical model embankment failure
experiments. The model simulates the failure by overtopping of three embankments
that are varying in dimensions and soil type. The soil is either homogeneous pure
sand or mixture of sand, silt and clay. A new measuring technique consists of the
labeled rod bridge; grids lined and captured instantaneous pictures are utilized to
monitor the failure. It was used to analyze the morphological evolution during
embankment failure. The flow pattern and progressive failure of embankment
overtopping were investigated. The results showed four types of phenomenon
during embankment failure: erosion; erosion with sliding; erosion with mass failure
and head cut migration. The results revealed that the recorded time to reach the full
breach failure in case of mixture materials (sand, silt and clay) is seven times that
with pure sand. Also, reducing the embankment height to 50% (0.9m), led to
increase the time to peak discharge 8.4 times that of large one (1.8m). Adding 15%
clay to pure sand changes soil characteristics to (c and o) soil which makes soil to
have a critical vertical cut height. Set of dimensionless regression equations are
developed by using the experimental data. In conclusion, types of embankment
failure are defined. Adding a percentage of clay (15%) to the pure sand led to the
maximum remain vertical cut height of about 50% of the calculated critical vertical
cut height (Z). Bearing in mind the limited number of large scale testes for
embankment breach in literatures, the set of data could be used for calibration of
mathematical breach models and provides a reference for the flood risk management
of embankments.