Loading...

Estimating the Scour Depth Downstream of Ski-Jump Bucket Spillways Using Soft Computing Models

Shafagh Loron, Reza | 2023

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57549 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Ghaemiyan, Mohsen; Shamsai, Abolfazl
  7. Abstract:
  8. Spillways are among the most critical hydraulic structures constructed to discharge excess water beyond the dam reservoir's capacity. Ski-jump bucket spillways are among the most frequently used types of spillways . In such spillways, water flows rapidly from the crest's edge to a point on the downstream riverbed of the dam and impacts the ground at high speed. The resulting high-velocity water jet can create a scour hole downstream of the spillway. Therefore, an important and fundamental issue in engineering hydraulic structures is the accurate estimation of the scour depth downstream of ski jump bucket spillways and ensuring the stability of the dam and its surrounding structures. In recent years, the use of soft computing models in hydraulic parameter modeling has seen significant growth. In this study, soft computing models, including Artificial Neural Networks (ANN), Extreme Learning Machines (ELM), M5 Model Trees (M5MT), and Multivariate Adaptive Regression Splines (MARS), have been employed to estimate the scour depth. For this purpose, models for estimating the scour depth downstream of ski jump bucket spillways have been developed using available field data (82 data points) and laboratory data (95 data points). The ELM model, with CC=0.969, RMSE=0.206, and MAE=0.157, had the highest accuracy among soft computing models for scour depth estimation using field data. The M5 model, with CC=0.946, RMSE=0.273, and MAE=0.188, had the lowest accuracy among soft computing models for field data scour depth estimation. In addition, the ELM model had the highest accuracy with CC = 0.989, RMSE = 0.040, and MAE = 0.033 among soft computing models and empirical formulas for estimating scour depth using laboratory data and data derived from empirical formulas. The M5 model showed the least accuracy among the soft computing models and empirical formulas in scour depth estimation, with CC = 0.877, RMSE = 0.127, and MAE = 0.084. The results from these models indicate their successful application in estimating the scour depth downstream of ski jump bucket spillways. The CWPRS formula (1986) has the highest accuracy among the empirical formulas studied in this research
  9. Keywords:
  10. Scour Depth ; Artificial Neural Network ; Extreme Learning Machine ; Multivariate Adaptive Regression Analysis Spline ; Soft Computing Models ; Ski Jump Bucket Spillway

 Digital Object List

 Bookmark

No TOC