Experimental Characterization of Silt Erosion of 16Cr-5Ni Steels and Prediction Using Artificial Neural Network

摘要

Hydropower generation from the Himalayan rivers in India face challenge in the form of sand-laden water. These sediments contain abrasive particles which can erode the turbine blades and reduce turbine life. This calls for the development of newer materials for turbine blade. To address this issue in the present investigation, 16Cr-5Ni martensitic stainless steel has been selected. Silt erosive wear tests were done at various test conditions determined by Taguchi design of experiments of impact velocity, impingement angle, erodent size and silt concentration. Analysis of variance studies of erosion rate and roughness indicated that impact velocity is the single most important parameter and interaction of impact velocity and impingement angle are proved to be significant. The optimized artificial neural networks are finally used to estimate the erosion rate for different combinations of the test conditions in conjunction with optimization techniques like Genetic algorithm were employed to arrive at the worst possible scenario (impact velocity 20 m/s, impingement angle 30A degrees, erodent size 245 A mu m and silt concentration 60 kg/m(3)).