As a result of economic growth, the value which is protected by flood defence systems increased significantly. The population at risk significantly increased too. Numerous cases (e.g. North Sea flood in the Netherlands 1953; New Orleans, USA 2005) led to devastating losses. Risk management of flood zones is hence on the top priority globally. A few regional or flood-event related databases of dike breaching cases were published; however, the availability of dike breaching data is still limited. Most commonly failure cases can only be found on a one-by-one basis. As a result of the lack of a comprehensive database, dike specific physically-based models are sometimes validated with the aid of dam breaching cases, and only a few dike-specific empirical equations are available for est...[ Read more ]

As a result of economic growth, the value which is protected by flood defence systems increased significantly. The population at risk significantly increased too. Numerous cases (e.g. North Sea flood in the Netherlands 1953; New Orleans, USA 2005) led to devastating losses. Risk management of flood zones is hence on the top priority globally. A few regional or flood-event related databases of dike breaching cases were published; however, the availability of dike breaching data is still limited. Most commonly failure cases can only be found on a one-by-one basis. As a result of the lack of a comprehensive database, dike specific physically-based models are sometimes validated with the aid of dam breaching cases, and only a few dike-specific empirical equations are available for estimating dike breaching parameters. Sometimes dam specific models are used without validation. Available physically-based models cannot be applied to analyse historical dike failures due to their complex cross-section layouts. The main objectives of this thesis are to compile an international database of dike breaching cases, study failure mechanisms, propose dike specific empirical equations for estimating dike breaching parameters, and develop a generalised physically-based framework suitable for simulating overtopping erosion of historical river dikes.

An international database has been compiled which consists of over 1000 records including a number very detailed cases. Failure mechanisms of dikes are evaluated based on this database. Most important failure mechanisms of dikes are found to be external erosion (68.6%), internal erosion (14.3%), and slope failure (7.5%). It can be concluded that different failure mechanisms may lead to different breaching lengths. Statistics also show that more dikes failed in the 19^th century than thereafter likely as a result of the improvement in compaction methodologies and quality control.

New empirical equation sets are proposed for estimating dike breaching parameters based on the compiled database. The breaching length, breaching depth and peak discharge can be approximated with multiplicative multivariable models. Two sets of equations are proposed based on different datasets: a global-scale equation set is delivered based on the global database, and a Hungary specific equation set is derived based on the Hungarian cases in the database. Extensive comparison of specific models for dikes, man-made dams, and landslide dams shows that the dam models are not suitable for estimating dike breaching parameters.

A physically-based model has been proposed for simulating overtopping erosion of historical earthen dikes, and/or dikes with berms. The generalised framework is time-step based. The dike breaching stages of historical dikes are identified. The new model consists of three modules: a hydraulic module, an erosion module and a geometric module. A new feature of the hydraulic module is that it is capable of describing hydraulic conditions related to multiple headcuts. The erosion module is shear-stress based and considers both slope and headcut erosion. Two applications of the model are presented for validation purposes. The Lizhou Polder Dike breach is used to evaluate whether the methodology can estimate properly the breach growth process or not. Both the predicted breaching length and total discharge agree well with the measured values. The second application studies the Tivadar Dike breach in Hungary. Great details of this breach case are available, so both early and later stages of the breaching process can be evaluated and compared. The new framework shows promising results. The breaching process in the early stages is well simulated: the predicted breaching lengths agree well with the observations at two different times; the total discharge is also well estimated.