Modeling of Flow Friction Factor in Irrigation Pipes using Machine Learning Methods and Comparing with Empirical Equations

The implicit Colebrook–White equation has been extensively used to estimate the friction factor of turbulent flow in irrigation pipes. In the meantime, a practical and accurate solution for Colebrook–White equation is, in particular, necessary for hydraulic computations of pressurized irrigation systems. In this paper, the performance of some machine learning methods such as support vector regression (SVR), genetic programming (GP) and M5 model trees have been evaluated and compared to the empirical equations in friction factor estimation. The obtained results from statistical analysis of studied methods showed that Buzzelli and Haaland empirical equations with root mean squared error (RMSE) of 0.00002 and 0.00015, respectively and also genetic programming with RMSE of 0.00031, had better performances among the others. Also, it was concluded that the M5 model trees and SVR with RMSE of 0.00204 and 0.00417, respectively, had lower accuracy in comparison with the empirical equations and genetic programming methods in estimating friction factor of irrigation pipes.