Discharge Performance of Plan View of Multi-cycle Triangular and Circular arc Labyrinth Weir

Weirs have been widely used for the purpose of flow measurement and flow control in open channels. For the purpose of diversion of flow side weirs or skew weirs are used. In the past various weirs of modified plan form have been used to enhance the discharging capacity with minimum head over the weirs. Labyrinth weirs are also a modified plan view of weirs, used to increase the discharge compared to linear weir. The current work focusses on experimental investigation of discharge capacity of labyrinth weir and predicting the results using Artificial Neural networks. The structures used in this study, are designed to carry large flows at lower head by increasing its effective length. Two different geometric construction namely triangular and circular arc labyrinth is used. The weir height ranges from 5 cm to 30 cm with gap of 5 cm and cycles changes from one to five for triangular labyrinth weir. For circular arc crested weir arc lengths is changed by changing the central angle to 30˚, 60˚, 90˚, 105˚ and 120˚ and the crest height from 5cm to 30 cm in steps of 5cm. The equation for the discharge capacity of a labyrinth weir as proposed by Tullis et al. (1995) has been used. The performance of the feed forward back propagation neural networks or multi-layer perceptron of linear regression model and nonlinear regression equation was compared with our physical model. To this purpose, 212 experimental data set were used for multi-cycle triangular labyrinth weir and 210 data set for circular arc weir. Correlation Coefficient (R), Root Mean Square Error (RMSE), and Co-efficient of Efficiency (CE) were used for the evaluation of the model’s performance. The results of ANN model shows that R was 0.93, RMSE was 0.08 and CE was 0.89 for multi-cycle triangular labyrinth weir. In the case of circular arc labyrinth weir these values were 0.98, 0.07 and 0.97, respectively, indicating satisfactory performance of the ANN model.