Laboratory Evaluation of Free Overfall Equations for Predicting Water Profile Over Weir

Abstract

Laboratory evaluation of three free overfall equations for predicting water profile over hydraulic structure – weir was conducted in Irrigation laboratory of Higher National College of Agriculture of Montpellier (ENSAM) , France. A broad-crested weir was constructed as a reduced physical model to simulate the gated-structure (Begemann gate) for upstream water control in the Main Canal of Hadejia Valley Irrigation Project (HVIP), Northern Nigeria. Instrumentation in the ENSAM laboratory allows precise variation of channel flows with incremental changes of 5 l/s. Three recently developed models in a progressive succession were used for comparison between measured and predicted data to establish degree of accuracy in the prediction of water profile passing over weir in open channel.

Results of the evaluation show that the three equations performed differently in predicting water profile over weir at the selected channel flows of 60 l/s, 30 l/s and 10 l/s respectively. These stream sizes were chosen because they were the maximum, medium and minimum for the canal capacity.  Although the stream size directly influenced the water profile as a result of internal pressure due to convergence of stream filaments between upper and lower laminar flow. Longer velocity jumps were recorded at the higher stream size (60 l/s) of 35 cm from weir crest end while less than 25 cm distance was covered by the least stream size of 10 l/s. Generally, based on the R² values and percentage errors for the three models at all the three selected stream sizes, Cavailhé model predicted the water profile with high degree of accuracy followed by that of Davis and lastly the Hager model. The same order was observed on ease of manipulation and user friendliness. The choice of water profile prediction model with high degree of accuracy is important in the design of hydraulic structures for water conveyance, diversion and measurements in irrigation canal network.