Start Date

4-29-2025 1:30 PM

Description

The EDF funded Performance Assessment of Industry Applicable IE Initiated Breach Prediction Models project was initiated with a workshop at HERU in Stillwater in October 2019. In the following 4 years a team of modelers comprising both model developers and industry model users analyzed the performance of 6 different breach modelling codes against a range of different data sets, including hypothetical, field and real dam failure case studies. Each of the data sets reflected a dam or levee failure initiated by internal erosion, and which subsequently developed into an open breach. Modelers were required to simulate breach formation from the initial ‘pipe’ flow through to open breach and the associated dam or levee failure. Four phases of modelling work were undertaken, with the initial Phase 0 focusing on hypothetical data to test the team approach. Phase 1 focused on assessing performance against sets of field data, whilst Phase 2 focused on assessing performance against observed Lawn Lake and Big Bay dam failure data. The final Phase 4 work took a different approach by focusing on data uncertainty (both case study and modelling data uncertainty) to see how the observed and predicted conditions potentially overlapped. This final stage of work allowed us to determine whether a model might recreate the observed conditions given the correct combination of modelling parameters.

Share

COinS
 
Apr 29th, 1:30 PM

Conclusions From the Performance Assessment of Industry Applicable Internal Erosion Initiated Breach Prediction Models

The EDF funded Performance Assessment of Industry Applicable IE Initiated Breach Prediction Models project was initiated with a workshop at HERU in Stillwater in October 2019. In the following 4 years a team of modelers comprising both model developers and industry model users analyzed the performance of 6 different breach modelling codes against a range of different data sets, including hypothetical, field and real dam failure case studies. Each of the data sets reflected a dam or levee failure initiated by internal erosion, and which subsequently developed into an open breach. Modelers were required to simulate breach formation from the initial ‘pipe’ flow through to open breach and the associated dam or levee failure. Four phases of modelling work were undertaken, with the initial Phase 0 focusing on hypothetical data to test the team approach. Phase 1 focused on assessing performance against sets of field data, whilst Phase 2 focused on assessing performance against observed Lawn Lake and Big Bay dam failure data. The final Phase 4 work took a different approach by focusing on data uncertainty (both case study and modelling data uncertainty) to see how the observed and predicted conditions potentially overlapped. This final stage of work allowed us to determine whether a model might recreate the observed conditions given the correct combination of modelling parameters.