Applications of Fault Diagnosis techniques for a multishaft centrifugal compressor

In the present paper the design and implementation of a Fault Diagnosis system for a compression´s process integrated in an IGCC (Integrated Gasification & Combined Cycle) section of a refinement plant is described. Both single and multiple faults have been considered which may cause errors in the sensor readings and/or in the actuators used in the process. A multivariable data-driven approach, that is a principal component analysis (PCA) technique has been adopted for monitoring the chemical process performances. A new procedure for the determination of number of principal components based on the statistical test ANOVA is introduced which constitutes the original contribution of the paper. The proposed approach on detection and isolation of faults have been tested and validated on the plant and its goodness and effectiveness could be proven.