PRESENTATION OF A SYSTEM FOR THE IMPROVEMENT OF THE ONLINE THERMAL MONITORING ON 900 MW TURBOGENERATORS FOR PREDICTIVE MAINTENANCE PURPOSES

This paper firstly presents an acquisition system which has been devel oped in order to rapidly detect the overheating in the stator bars of large turbogenerators in nuclear power plants; it has been installed i n 1994 on the stator of Tricastin1. Secondly, associated to it, the de velopment of a new monitoring system is presented, based on the determ ination of the fingerprints of the stator from the thermal modelling o f the water cooling circuit. Validated by the measurements made on sit e, it also enables the following in real time of the status of the mac hine, provides a detailed diagnosis of it, and optimises the load in a bnormal conditions. It may turn the actual maintenance made on those m achines towards the concept of predictive maintenance.