The probability of structural failure of flood defences is an important component in the modelling of flood risk. Flood embankments are a very common type of defence whose reliability assessment shows some problematic aspects. In particular it is difficulty to model the process of piping through the earthfill and, when assessing the safety of an extended flood defence network, a diffuse lack of knowledge (epistemic uncertainty) about some of the characteristics of the structures has to be addressed. This thesis presents a new methodology, named the Reliability Rating System, which makes possible the rapid quantification of the expected performance of flood embankments in extreme hydraulic conditions. The proposed methodology ranks and compare flood embankments according to their proneness to breaching using a performance indicator, called the Probability of Breaching in Extreme hydraulic conditions PBE. This performance indicator is related to probability of structural failure for a limited range of water levels above and immediately below the crest. The Reliability Rating System takes into account three failure modes: breaching initiated by grass cover failure in overtopping conditions, breaching induced by piping through the earthfill (through-piping) and breaching due to piping in the founding soil (under-piping). Credible mathematical models are available for the grass cover failure and under-piping and the probability of breaching induced by these failure modes can be calculated with the methods of reliability analysis. By contrast no credible processbased mathematical model is currently available for through-piping. In this work the probabilities of breaching by through-piping have been estimated with a rigorous process of subjective judgement elicitation. In the case of uncertainty on some of the relevant characteristics the final users can rapidly study different possible scenarios thanks to the tabulated solutions presented here. Upper and lower bounds on the performance indicator PBE can be readily determined in order to handle the epistemic uncertainty typical of flood defences safety assessments.

It has been known since studies carried out by Cooling and Marsland for the 1953 North Sea Floods that desiccation fissuring of clay fill can play a major role in the failure of flood embankments under overflow conditions. However, very little scientific research has been carried out into the onset of desiccation fissuring in flood embankments and the potential failure mechanisms. To overcome this shortage of information, field and laboratory studies have been carried out into the desiccation fissuring of clay fill for flood embankments. The preliminary results confirmed that desiccation fissuring occurs to a depth of typically 60 cm within the outer surface of a flood embankment constructed from clay fill and can occur within 2 years of construction. The critical condition occurs when desiccation creates an interconnected network of sub-vertical and sub-horizontal fissures that increases the mass permeability of the fill material similar to that of coarse sand or gravel and hence allows rapid seepage of flood water through the surface layer of the embankment (crest and sides slopes). The aim of the thesis is to investigate the influence of desiccation fissuring on the stability of flood embankments. In particular, the main objectives of this thesis are as follows: 1. Survey of flood embankments in UK constructed from clay fill to determine the extent and nature of desiccation fissuring. 2. Construction and investigation of the behaviour of macro-scale embankmenmt odel subjected to desiccation processes. 3. Investigation of the innovative, non-invasive geophysical technique for desiccation cracking detection. 4. Investigation of the embankment model behaviour under different flooding conditions.

Wave overtopping is one of the most important processes for the design of seawalls. During the past 50 years methods for predicting wave overtopping of coastal structures have continuously been developed. However, it is evident from the existing literature that additional investigations into overtopping of small positive, zero and negative freeboard are needed. The present thesis describes numerical investigations based in this background. Wave overtopping is dependent on the processes associated with wave breaking. Therefore, a two dimensional breaking wave numerical model has been developed and used to study the phenomena of wave overtopping. The model is based on the Reynolds averaged Navier-Stokes equations for the mean flow and k-epsilon equations for turbulent kinetic energy, k, and the turbulence dissipation rate, epsilon. The model accuracy in simulating propagation of linear and irregular waves has been evaluated. The overall performance of the model is considered satisfactory. The development of guidelines for calculating overtopping discharge for different seawall slopes is presented. All slopes have been subjected to a wide range of irregular waves. The influence of how the slopes of seawalls, wave type (breaking and non-breaking) and crest freeboard affect the overtopping discharges has been investigated. Based on the numerical data, a new expression for breaking and non-breaking waves on smooth impermeable slopes is proposed. With the new expression it is possible to predict overtopping discharges of smooth seawalls in small positive, zero and negative freeboard.