A guide to breach prediction

A guide to breach prediction.
West, M.and Morris, M. and Hassan, M.
Discussion Paper. HR Wallingford Ltd, Wallingford. (2018)

document citation PDF (3931Kb)
Abstract:Earthen embankments that are categorised as large (15m or greater in height) number in the tens of thousands globally. Embankment dam risk assessment is a vital measure that has been adopted throughout the industry to assess the potential impact that catastrophic dam failures can carry. A critical part of this assessment is the prediction of the breach process, which will determine the reservoir outflow hydrograph. This is crucial for the following stage of flood routing, which aids in flood risk assessment, evacuation planning and land-use planning. This report provides details of the breach prediction methods available to users, ranging from simple parametric equations to complex multi-dimensional erosion models. These are commonly divided into three categories; parametric models, semi-physically based models and physically based models. Parametric models, such as Froehlich (2016a, b), Xu & Zhang (2009) and Von Thun & Gillette (1990), allow breach geometry, formation time and peak outflow to be estimated through the regression analysis of historical dam failure data. These have advantages in their ease and speed of use, but were found to have great uncertainty in their application and are therefore not typically suitable for high risk applications, where uncertainties will have a large impact. Appropriate applications include initial appraisal-level breach modelling and the study of low-risk scenarios where uncertainties will have a minimal impact. Semi-physically based models, such as HEC-RAS, take breach geometry and formation time, or soil erosion rates, as input values to produce a breach hydrograph. No physical processes are modelled; rather the flow of water through the use-defined breach is calculated using simple fluid dynamic equations, such as weir and orifice flow. These provide no improvement in the accuracy of predicting a breach over parametric models, but improve on the process of converting these results into outflow hydrographs, which may be required for further use. Physically based models, such as EMBREA, DL Breach and WinDAM consider the complex geotechnical, structural and hydraulic behaviour of an embankment dam and its impounded reservoir. While generally more time-consuming than parametric and semi-physical models, physical models tend to provide results with a greater certainty and accuracy. These, as a whole, are suitable for high risk breach scenarios, where accuracy and reliability are critical in providing results within the acceptable bounds of uncertainty. The descriptions and recommendations of breach prediction methods in this report are intended to guide users towards the most appropriate breach model for a given scenario. A recommended approach to choosing a model type (parametric, physical etc.) is given, taking into consideration the type of analysis and associated risks, amongst other factors. It was concluded that, while parametric models have tended to be used by industry in the past, technological advancements and practical field testing have allowed rigorous physically based models methods to become more feasible. Further developments are likely to reduce the reliance of physical models on simplifications and improve their accuracy and usability.
Item Type:Monograph (Discussion Paper)
Subjects:Floods > Dams and reservoirs
ID Code:1469
Deposited On:12 Mar 2018 11:06
Last Modified:12 Mar 2018 11:06

Repository Staff Only: item control page