Diagnostics of Anomaly Steam Turbine Behavior in Terms of Remote SHM and Cloud Computing

Abstract

Monitoring of structural health is the important task in terms of ensuring the reliability of the operated technology. It is important especially in case of the steam turbine operation where the unplanned outage is associated with the high financial losses. The conventional approach for the evaluation of the normal turbine operation is based on precise specification of the trigger limits for each of the measured signals made by the technical expert. The abnormal operation is then determined when one of the signals is outside of the limits. However, it is a non-trivial task to specify trigger limits for all measured signals and sometimes it is even not practicable. The method described in this paper is based on the automatic detection of anomalies in the turbine behavior without the need to precisely specify the trigger limits manually. The approach is based on two steps. The first is to find the behavior of the turbine that is related to the normal turbine operation. The signal trigger limits are evaluated automatically with probabilistic assessment. The second step is to investigate the actual operation state using the signal measurements. Then if the actual behavior of the turbine is out of the boundary of the normal turbine operation, the anomaly is detected. The described method was validated using the measurement data acquired in operation of a steam turbine with a nominal power of hundreds MW. As an example of the method application, the paper shows the detection of an anomaly, which was subsequently identified as a contact between the stator and the rotor turbine part. This contact is potentially dangerous because it can change the structure of the machine. The described method is being integrated into the remote monitoring system that is based on cloud computing and being developed by the authors of this paper. The remote sensing of the turbine operation is nowadays the key to reduce costs in terms of maintaining the installed system and frequent visits of technical personal support to collect the data. The architecture of the monitoring system itself is described in the paper. The system is important in terms of providing an early warning in case of an unexpected behavior of the turbine. This provides the maintainability and reliability of the operated technology. Nowadays more than 30 operated turbines are part of this monitoring system worldwide.