Due to difficulty and radiation exposure associated with examination of the internals of large commercial nuclear reactor coolant pumps, it is necessary to be able to diagnose the cause of an excessive vibration problem quickly without resorting to extensive trial and error efforts. Consequently, it is necessary to make maximum use of all available data to develop a consistent theory which locates the problem area in the machine. This type of approach was taken at Three Mile Island, Unit #1, in February 1984 to identify and locate the cause of a continuously climbing vibration level of the pump shaft. The data gathered necessitated some in-depth knowledge of the pump internals to provide proper interpretation and avoid misleading conclusions. Therefore, the raw data included more than just the vibration characteristics. Pertinent details of the data gathered is shown and is necessary and sufficient to show that the cause of the observed vibration problem could logically only be a cracked pump shaft in the shaft overhang below the pump bearing.
crack catia v5r20 32 bit
The Approximate Entropy is a statistical calculus used primarily in the fields of Medicine, Biology, and Telecommunication for classifying and identifying complex signal data. In this work, an Approximate Entropy algorithm is used to detect cracks in a rotating shaft. The signals of the cracked shaft are obtained from numerical simulations of a de Laval rotor with breathing cracks modelled by the Fracture Mechanics. In this case, one analysed the vertical displacements of the rotor during run-up transients. The results show the feasibility of detecting cracks from 5% depth, irrespective of the unbalance of the rotating system and crack orientation in the shaft. The results also show that the algorithm can differentiate the occurrence of crack only, misalignment only, and crack + misalignment in the system. However, the algorithm is sensitive to intrinsic parameters p (number of data points in a sample vector) and f (fraction of the standard deviation that defines the minimum distance between two sample vectors), and good results are only obtained by appropriately choosing their values according to the sampling rate of the signal. 2ff7e9595c
Comments