Performance Evaluation of Discharge Coefficient in Physical Models of Labyrinth Fusegate Spillways with Intellectual and Statistical Models

Fusegate spillways dispart into two models of straight and labyrinth crest according to their plan view. Labyrinth Fusegates are consists of three types: Narrow Low Head (NLH), Wide Low Head (WLH), and Wide High Head (WHH). In this study, the effect of well with different heights and floor slope in WLH model of Fusegate spillways investigated on discharge coefficient over the spillway, and the amount of discharge coefficient calculated by artificial neural network and statistical method of multivariate regression by SPSS software, and compared with the result of physical modeling tests. Also, the sensitivity analysis of effective factors on the flow discharge coefficient has been done. To predict the flow discharge coefficient in artificial neural network, the best fit was obtained from Levenberg-Marquardt algorithm application as training function, TANSIG as transfer function for hidden layer and linear function as output layer. The results of sensitivity analysis showed that among the dimensionless parameters, the ratio of upstream water head to height of bucket is more effective than the other input variables. To estimate the discharge coefficients, the amount of relative error in statistical model is approximately 30% and in neural network is less than 5%. Therefore, the neural network model is considered as a suitable tool for estimating the discharge coefficient in Fusegate Spillways.