Sediment transport modeling in circular smooth and rough rainwater transport pipes using factorial analysis, intelligence and empirical methods

One of the common phenomena in water and urban wastewater transport pipes is sedimentation. Due to complexity of sedimentation phenomenon and influence of various parameters, determining the governing equations are difficult and classical mathematical models are not sufficiently accurate. In this study, capability of Gaussian Process Regression (GPR) in predicting sediment discharge in circular rainwater transport pipes with smooth and rough beds was investigated. In this regard, hydraulic and sediment parameters which had most correlation with sediment discharge were determined using factorial analysis. Then, different models were developed and using three experimental data series were investigated. Then, the accuracy of the results was compared with the classical sediment transport models. The results showed the high efficiency of the intelligent GPR model in prediction of sediment discharge in rainwater transport pipes compared to the classical methods. In both pipes the model with input parameters λs, Fm, Dgr, d50/y which are relative sediment size, non-dimensional sediment size, fraud number of sediments, and total roughness coefficient, respectively, was obtained as superior model. It was also observed that pipe diameter affected the models efficiency and with increasing pipe diameter, model accuracy increased. The pipe with a diameter of 305 mm led to more accurate results. Factorial and omitted sensitivity analysis showed that d50/y was the most effective parameter in estimation of sediment discharge in both smooth and rough pipes.