Analyzing Factors Influencing the Discharge Coefficient of Morning Glory Spillways Using Artificial Neural Networks: A Case Study of San Luis Dam, USA

Morning glory spillways are essentially closed conduits typically used to pass floods from higher elevations to lower elevations. These types of spillways are utilized in reservoir dams located in narrow valleys and areas with steep reservoir wall slopes. The advantage of these spillways lies in their relatively high capacity to convey flow with relatively low energy losses. This characteristic makes them highly efficient for flow rates below the design discharge, making them an ideal structure for flood conveyance in scenarios where sufficient time is available for temporary storage in the reservoir to attenuate subsequent flood intensities. In this study, nearly ۳۰۰ experiments conducted on the physical model of the spillway of San Luis Dam in the USA were reviewed to derive the hydraulic flow parameters, including Froude number and submergence limits. Using dimensional analysis, a more precise formulation for the discharge coefficient was developed, encompassing the effects of the number of steps and vortex breaker types. Subsequently, with the help of an artificial neural network (based on the Levenberg-Marquardt algorithm), the influence and importance of each parameter were evaluated using the Root Mean Square Error (RMSE) as the benchmark. According to the obtained results, the parameter associated with the number of vortex breakers had the most significant impact, with an RMSE value of ۲۱۵۴.

کلیدواژه ها:

Levenberg-Marquardt Algorithm ، Sensitivity analysis ، Artificial Neural Network ، Morning Glory Spillway ، Discharge coefficient