Model-based fault diagnosis of a two-pump system

This paper outlines the design and initial testing of a quantitative model-based fault detection and isolation (EDI) scheme for a water pumping system. The work has been motivated by the desire to develop a fault diagnosis and condition monitoring scheme for an industrial cooling water pumping system in operation at the 'Heysham 2' power station (British Energy). Methods of residual (fault indicator) generation using observers and parameter estimation are appraised as to their suitability for industrial systems, where sensor availability and limited sampling rates must be taken into consideration. A detailed model of the industrial pumping system has been developed and the performance of the methods considered is illustrated using this model. Due to the industrial motivation the task of analysing the information provided by complementary methods is described and a practical scheme for the cooling system is proposed which considers the currently available methods and the particular challenges presented by the actual system.