New Approach to Estimate Velocity at Limit of Deposition in Storm Sewers Using Vector Machine Coupled with Firefly Algorithm

One of the crucial issues regarding a storm sewer system is the ability to avoid sediment depositions on the pipe invert. In this study, the mean flow velocity under the limit of sediment deposition conditions in partially filled circular storm sewers is evaluated through the use of a support vector machine (SVM) model coupled with the firefly algorithm (FFA). The aforemetioned velocity, defined as the velocity at the limit of deposition, and the parameters upon which it depends have been nondimensionalized using the Buckingham Π theorem. Therefore, once the dimensionless parameters are identified, six different functional relationships in terms of dimensionless groups can be obtained. The effects of each of these functional relationships on the dimensionless velocity at limit of deposition, defined as the densimetric particle Froude number at the limit of deposition, have been analyzed by using, respectively, the SVM-FFA model, SVM model, genetic programming (GP) model, and artificial neural network (ANN) model. Five statistical indices have been used for evaluating the performance of each model (both in training and test phases) and, later, for comparing the performance of the different models between them. Finally, the predicted densimetric particle Froude number values obtained through the proposed SVM-FFA model have been compared with those obtained by three different dimensionless equations for velocity at the limit of deposition. The results indicate that SVM-FFA predicts the densimetric particle Froude number at limit of deposition fairly accurately.