Application of machine learning methods on investigation the effect of interrupted deflector’s height and downstream depth changes on energy dissipation at flip bucket spillways

Flip bucket is a type of energy dissipator structure. Flip buckets can sometimes be improved by adding wedge-shaped deflectors. This research introduced the best height proportion of used deflector on the flip buckets to increase energy dissipation. It used 4 types of deflector series including 32, 45, 47 and 55 degrees toward horizon with two different regular and irregular layouts and 19 different heights from 2.31 to 5.6 cm and in different hydraulic condition and the results were compared with a flip bucket without a deflector. The characteristics of laboratory flume were: length= 7.5 m, width= 0.58 m and height= 1.6 m. The results illustrates that the energy dissipation in the model with deflectors increased from 11.83 to 19.38 percent as compared with model without a deflector. The greatest percentage of energy dissipation was 80.74% which observed at a deflector angle of 55° and a discharge of 10 L/s, at deflector’s ratio of n=0.8 and in non-uniform layout and in free hydraulic jump. Larger deflector angles and side lengths initially boosted energy dissipation, but this effect plateaued or even reversed at very large angles. For calculating energy dissipation and hydraulic jump length parameters, the regression relations were extracted in this research and results of this relations were compared with results of the gene expression programming (GEP), random forest (RF) and multivariate adaptive regression splines (MARS) methods. The results showed that the RF method is the most accurate method for calculating energy dissipation and hydraulic jump length parameters.