EARLY CRACKED SHAFT DETECTION IN PUMPS USING ROTOR LATERAL VIBRATION ANALYSIS

摘要

This paper presents a chronological account of a successful crack detection on a large vertical pump using rotor lateral vibration data analysis. It then looks at one of the primary forces occurring in a pump, the steady state radial load force generated by moving the fluid, and explores the interaction of this force with the changes in the dynamic stiffness of the system caused by a shaft crack. The vibration behavior of any system depends on the complex interaction between the forces applied to the system and the dynamic stiffness of the system. These effects depend heavily upon the original dynamic stiffness parameters, the operating speed of the pump, and the location of the crack. The primary effects can be categorized into creating additional vibration components at harmonics of operating speed, and shifting the values of selected parameters in the stiffness matrix. These interactions can produce profound changes in the rotor lateral vibration response which can be used to diagnose crack propagation in the rotor. Unfortunately, other mechanisms that can produce similar results exist, preventing the construction of simple vibration analysis rules for shaft crack detection, such as, "when the twice rotative frequency (2x) lateral vibration increases, the shaft is cracking." To solve the above dilemma, methods are presented for evaluating the possibility that a particular vibration component change is produced by a cracked shaft, or is the result of other system parameter changes. To accomplish this, traditional single axis signal processing is expanded to include complex variable processing of the data from two orthogonal transducers at one measurement plane, and modeshape analysis which concurrently uses the vibration information from several axial locations. The use of these advanced rotordynamic signal processing techniques provides much more insight into the primary cause of changes in the vibration response of any machine, including pumps. As shown by Muszynska et al. (1997), complex variable processing decomposes the response at any location into a forward and backward circular rotating response. The relative magnitudes and phases of these components can give much more insight into the system dynamic stiffness than using measurements from a single axis alone. The effectiveness of these signal processing techniques is demonstrated using data obtained from numerical simulations, rotor rigs capable of modifying individual parameter values within the stiffness matrix while being subjected to the common forces produced in a pump, and vibration data from pumps with cracked rotors in the field. By using good rotordynamic principles and some advanced signal processing techniques, analysis of the lateral vibration of pump rotors can be a powerful tool in the early detection of shaft cracks.