Fault Diagnosis and Prognosis in Rotating Machines Carried Out by Means of Model-Based Methods: A Case Study
Diagnostic methods, based on mathematical models, can be used to identify the most common faults and malfunctions of rotating machines by minimizing the error between experimental vibration data and the corresponding theoretical response of the rotor system caused by a specific set of excitations. These techniques allow the severity and location of the fault to be estimated. Moreover, depending on the fault characteristics, model-based prognostic techniques can be used to study appropriate corrective actions that can eliminate the cause of the malfunctions or reduce the machine vibration levels. This paper shows the results of a diagnostic analysis carried out to investigate the cause of the high vibration of the HP-IP steam turbine of a large power unit that occurred, during the runups, when approaching the first balance resonance. The numerical results confirmed the suspect that this high vibration was caused by a shaft bow. Then, the machine model was used also to study and optimize a corrective action that allowed the operating speed to be reached and the shaft bow to be eliminated by means of the turbine heating caused by a load rise. The successful results obtained with the machine maintenance carried out considering the indications provided by the model-based diagnostic and prognostic analyses are shown and discussed.