2019 – Modelling Reservoirs with a 2D Hydraulic Model

Andrew Northfield, David Stephens, Tim Craig, Mitchell Smith

In recent times two dimensional (2D) hydraulic modelling has become the most common type of modelling for undertaking dambreak assessments. Direct map outputs such as depth and depth-velocity product are very useful in assessing risk across a floodplain. The temporal output from 2D models also enables the tracking of flow across a floodplain, helping practitioners and dam owners alike make informed decisions on warning time and evacuation routes. These outputs form essential input to packages such as HEC-LifeSim an agent-based simulation model for estimating life loss by simulating population redistribution during an evacuation.

A number of investigations have shown the hydraulic model, TUFLOW, is able to simulate the hydraulic conditions expected in a dambreak flood wave, giving confidence in the model’s ability to correctly capture the flood wave propagation. Notwithstanding this ability, there remains uncertainty over the best methodology to adopt when assigning a breach hydrograph to the model and in turn the impact this choice has on assessing downstream populations at risk.

A commonplace method of assigning dam breach hydrographs is to model the reservoir and dam structure with a 1D model or spreadsheet, where the storage is represented with a stage storage relationship and outflow through a time-varying breach is calculated using level-pool routing. The resulting hydrograph is then applied directly to a 2D model immediately downstream of the dam to model the propagation of flow downstream.

An alternative approach consists of representing the entire reservoir, dam and downstream floodplain in the 2D model. This allows for the dynamic effects of bathymetric constrictions in the reservoir to be accounted for which could greatly impact on the timing and shape of the dam breach hydrograph. However, this comes at a cost, as representing the reservoir in 2D requires bathymetry data which can be expensive to capture and also may require a major extension of the model domain.

In this paper the ‘Fully 2D’ and ‘Stage storage relationship 1D/Spreadsheet’ approaches are compared for a number of case studies.

Buy this resource