BREACH HYDRO - Water Balance Suite for Stochastic Dam Break

For more detail around this study, refer to the Breach Hydro Manual.

Four (4) dams of real world sites, with varying storage, embankment heights and embankment types (i.e. earthen or rockfill) were selected. For each of these dams, six (6) scenarios were developed with varying failure types, conditions, and breach parameters. This is documented in Table 2.

Breacher and HEC-RAS models were developed for each of these scenarios, with breach hydrographs and changes to dam storage levels compared to confirm Breacher replicates the breach process of HEC-RAS. The results from this are shown in Figure 20, Figure 21, Figure 22 and Figure 23.

Breach progression in HEC-RAS, especially for a piping failure, was very difficult to replicate in Breacher and required substantial testing and verification, an example from this is shown in Figure 19. Once the team could replicate the breach progression, and on review of publications (relating to the HEC-RAS breach progress and general literature on dam breach progression), it was understood why this approach was adopted in HEC-RAS.  

Our take away from this exercise was breach progression and the influence it has on breach width, elevation, hydrograph shape and peak flow is highly influenced by assumptions made by the programmer of the software tool chosen by the user. We strongly recommend other software developers of similar dam breach tools provide similar validations to provide confidence in programmed assumptions.  

Some interesting observations were found by the Forward Hydro team during the validation process: 

- Outflow velocity during piping in HEC-RAS seemed to have an upper limit. During testing this was around 11 m/s for several of the scenarios. This capped velocity may be due to one of the following:

    - In HEC-RAS, like other hydrodynamic solvers, velocity and timestep impact stability (relating to Courant number). An upper limit to velocity would reduce the frequency of hydrodynamic models going unstable, and ~ 11 m/s might be reasonable given HEC-RAS’ minimum timestep (0.1 seconds).

    - An established relationship in the solver between piping erosion and velocity that hasn’t been documented.

    - Influence of “lateral structure flow stability factor” in smoothing (or dampening) estimated computed flows

- Instabilities during a HEC-RAS model run can compound causing substantial changes to hydrograph shape and peak flow, Site 3 in this validation had minor instabilities that while time consuming to resolve, were left in the model result to showcase this effect. See Figure 22. 

- In hydrodynamic modelling, adaptive timestep influenced hydrograph shape and peak flow, largely due to small mass balance errors for larger timesteps compounding prior to breach peak flow. These mass balance errors influenced volume within the dam as it approached peak flow, thus influencing peak flow. When using adaptive timestep with hydrodynamic models, it’s strongly recommended to sensitivity test the lowest allowable timestep. Similarly with Breacher, it’s highly recommended to sensitivity test timestep, which can rapidly be done by randomising the command input.

Results overall demonstrate, for the scenarios tested, Breacher was able to replicate HEC-RAS.