Modo de falha crível em barragens de rejeito de mineração: uma análise das variáveis condicionantes e proposição de um modelo preditor
保存先:
| 第一著者: | |
|---|---|
| 出版日付: | 2023 |
| フォーマット: | Bachelor thesis |
| 言語: | por |
| ソース: | Repositório Comum do Brasil - Deposita |
| Download full: | https://deposita.ibict.br/handle/deposita/497 |
要約: | The significance of understanding credible failure modes in tailings dams stems from the significant environmental and economic devastation such failures can cause. Moreover, these failures represent critical threats to human life. Consequently, grasping the credible failure mode of a particular structure holds importance in its risk management. Depending on the specific failure mode, the geotechnical stability analysis varies. Several failure modes exist, encompassing slope instability, internal erosion, liquefaction, overtopping, structural or foundation inadequacies, earthquakes, and external erosion. In this study, an examination of variables related to this phenomenon was conducted, and predictive models for failure modes were formulated. To achieve this, the K Nearest Neighbors machine learning technique was employed using the Python language. The study's foundation relied on a database featuring 66 instances of failed tailings dams across the globe. Notable variables encompassed the ore type generating the stored tailings, dam construction materials, construction methods, height, volume, seismic risk, climate, and failure mode. The findings demonstrated the significance of all variables in predicting failure modes, with the exception of the "volume" variable, which exhibited minimal influence on failure patterns in the analyzed dams. A total of 35 predictive models for dam failure modes were developed, designated as M1 through M35. The most accurate model, M11, achieved a 71% accuracy rate. This model excluded the volume variable, configured with three neighbors (k=3), and a training/test sample size ratio of 90% to 10%. This model was subsequently applied to a database containing information on 10 Brazilian dams classified with emergency levels II and III by the National Mining Agency. Within this set, eight were identified as susceptible to credible failure due to slope instability, and two due to internal erosion. |
このレコードへの初めてのコメントを付けませんか!