Dam breach parameters: from data-driven-based estimates to 2-dimensional modeling

This research focuses on improving the estimations of key dam breach parameters (peak discharge, breach formation time, and average breach width). These parameters are crucial for understanding and mitigating the impact of dam failures. The study acknowledges the difficulty in estimating these parameters due to limited historical data and existing empirical models’ uncertainties. To overcome these challenges, the research follows a systematic approach: (1) Data Compilation and Cleaning: multiple datasets of historical dam break events were compiled and cleaned. Missing data were filled in, and anomalous outliers were removed to ascertain the data quality. (2) Data Fusion approach was then considered for (2-a) Identifying the most suitable empirical equations for each breach parameter. (2-b) Developing regression-based models based on the selected empirical equations. (2-c) Conducting uncertainty analyses to assess the reliability and confidence of the models’ results. The newly developed data fusion-based models were claimed to outperform existing empirical equations, reducing uncertainty across all breach parameters. (3) The research finally applied the proposed models to four real-world dam cases via a two-dimensional hydraulic model. This application demonstrated how the models can be used in practical engineering scenarios, aiming to enhance our ability to predict and consequently mitigate the impact of dam failures, ultimately contributing to improved dam safety and risk management.