DEFINITION of EFFECTIVE ALARM THRESHOLDS on VIBRATION SIGNALS for PREDICTIVE MAINTENANCE in PRODUCTION PROCESSES

The thesis aims to develop an effective strategy for condition monitoring of an existing production plant and thus set a framework for predictive maintenance. The condition monitoring will be based on vibration measurement, which uses a set of accelerometers. The goal is to detect significant changes that may lead to developing fault using a limited amount of data, i.e., vibration measurement. One of the main challenges of this thesis work is the definition of an effective methodology to set warning thresholds for the accelerometers combined with different equipment, e.g., compressors, bearings, and speed reducers. Some of the cases mentioned above are covered by ISO standards, while others lack specific guidance. The first part of the thesis will be devoted to analyzing ISO standards related to industrial maintenance and predictive maintenance. In the second part of the thesis, the case study from Stellantis CRF (Italy) will be analyzed by establishing the general alarms and warning levels for the roller bearing of the gantry system, considering the data coming from acceleration sensors and maintenance activity list. Here a standard-inspired methodology will be applied. In the third part of the thesis, the case study from Rompetrol (Romania) will be analyzed to show the predictive maintenance approach in the industrial sector and develop an effective strategy for condition monitoring for the reciprocating compressor, considering the data from accelerometers by using different software.