THE POSSIBLE CAUSE AND EFFECT GRAPHS (PCEG) MODEL FOR FAULT-DIAGNOSIS.2. APPLICATIONS

Probable cause and effect graphs (PCEG) have been proposed by Wilcox a nd Himmelblau (Computers chem. Engng 18, 103-116, 1993) as an extensio n of and improvement over the use of signed digraphs for fault diagnos is. Here we provide two detailed examples of the application of PCEG t o processes of interest to chemical engineers, a heat exchanger and a chemical plant. The examples indicate that the PCEG size, and hence th e computational effort, depends primarily on the degree of detail soug ht (and available) for the fault model of the process. Increased resol ution of diagnosis via additional sensor measurements is feasible, but relating explanations to small sets of measurements would be preferre d for efficiency in diagnosis.