Modern measurement techniques in combination with advanced computerized data processing and acquisition show new ways in the field of machine surveillance by use of spectral and correlation analysis of accelerations, displacements and vibrations. Time domain analysis using characteristic values to determine changes by trend setting, spectrum analysis to determine trends of frequencies, amplitudes and phase relations, correlation analysis to evaluate common sources and excitations of different vibrations and cepstrum analysis to detect periodical components of spectra are used in diagnostic systems.

Case studies of fault detection at turbines, fresh water and feed water pumps, blowers and safety valves using a portable multichannel vibration measuring system are presented. The influence of different faulty operation conditions to the characteristic parameters are described as well as trend settings and comparative measurements between similar machine arrangements.

Operating experiences with an on-line vibration control systems for PWR main coolant pumps is presented in the paper. The vibration control system is an effective tool for early failure detection and predictive maintenance of PWR main circulation pumps. Diagnostical results of different vibration behaviour of dissimilar pump shaft constructions were compared to particular anomalies caused by incorrect mounting of some components after a repair of one main coolant pump.

For implementation of vibration supervision and condition monitoring systems as tools for predictive maintenance, the vibrational behaviour of the machinery has to be investigated in order to get references for further trend settings. Another possibility is obtained by comparing several machine arrangements of similar design. The performance of reference measurements are of importance for fault detection during operation by trend settings. The comparison with thresholds given by norms and standards is only a small section of available possibilities. Therefore, for each machinery own thresholds should be determined using statistical time values, spectra comparison, cepstrum analysis and correlation analysis for source localization corresponding to the machine operation conditions.