Accurate prediction of the critical submergence for intakes in open channels has been mainly based on experimental studies and theoretical equations developed are empirical in nature. In the present study estimation of critical submergence for a 90° horizontal intake in open channels based on experimental results have been attempted by using empirical equations, the multi linear regression, and SVM (Polynomial and Rbf) based modelling. Two different locations of intake from the channel bed-one with clearance from the bottom equal to zero and the other having half the intake diameter have been taken in the present work. The data set has intake pipes of diameter equal to 4.25, 6.25 and 10.16 mm for the critical submergence under a wide range of flow conditions in a flume for 90° horizontal intakes. The study shows that the soft computing technique Support vector machine (Polynomial & Rbf) have emerged as an alternative to the analytical equations as suggested by previous investigators on present data.
Critical submergence, intake, modeling, prediction, Support Vector Machines
