Complex Variable Filtering: The Next Step in Rotor Diagnostics

Abstract

Currently most rotor diagnostic algorithms focus on extracting as much information as possible from the signal of a single transducer. This limits these algorithms to only processing information about the projection of the rotor motion onto the probe axis. Utilizing single probes is adequate for isotropic systems where the projections are the same for any probe orientation. However, for anisotropic systems the results of the algorithm can be different depending on the orientation of the measurement transducer. This leads to the blind men and elephant conflict, multiple descriptions of the same animal, or in our case rotor motion. For example, if the transducer is located on the minor axis a great running machine, on the major axis a machine that needs balancing. The phase is also inconsistent, rotate the transducer a few degrees and the indicated phase jumps drastically. If the transducer is on the ellipse axis the high spot and heavy spot are together below the resonance, move the probe off the axis and the high spot and heavy spot are no longer related. More sophisticated diagnostic methodologies which correlate data from multiple axial locations, such as relative phase and modeshape, are affected even more than balancing by the phase inconsistency. What is really desired, is information about the actual motion of the shaft in the two dimensional plane. Up until recently the only data presentation format that addressed the planer motion was orbit. Orbits are powerful diagnostic tools, but have the limitation of only showing motion at one speed. What is needed are similar processing techniques that present complete motion over the entire speed range and report the same information for all probe orientations, such as complex variable filtering. This paper presents the concept of complex variable filtering and explores some situations in which it provides superior results to those obtained utilizing conventional single probe techniques.