Evaluation of an advanced fault detection system using Koeberg nuclear power plant data

Abstract

The control and protection system of early nuclear power plants (Generation II) have been designed and built on the then reliable analog system. Technology has evolved in recent times and digital system has replaced most analog technology in most industries. Due to safety precautions and robust licensing requirements in the nuclear industry, the analog and digital system works concurrent to each other in most control and protection systems of nuclear power plants. Due to the ageing, regular maintenance and intermittent operation, the analog plant system often gives faulty signals. The objective of this thesis is to simulate a transient using a simulator to reduce the effects of system faults on the nuclear plant control and protection system, by detecting the faults early. The following steps will be performed: • validating the simulator measurements by simulating a normal operation, • detecting faults early on in the system. These can be performed by resorting to a model that generates estimates of the correct sensors signal values based on actual readings and correlations among them. The next step can be performed by a fault detection module which determines early whether or not the plant systems are behaving normally and detects the fault. (Baraldi P. et al, 2010.)