(285a) Fast Fault Detection and Identification Using Grey Box Models

Fault detection and identification applications are mostly based on empirical models that quickly degenerate in time due to slow changes in what operators consider normal process conditions. The extensive use of such algorithms is always rare due to the variations that a traditional industrial facility undergoes in time. This work illustrates the use of a monitoring system that takes advantages of grey box models to distinguish the parameters that are expected to shift in time versus those that are expected to remain constant or invariant. A change in the invariant parameters indicates the existence of a potential fault. The identification of the fault is closed related to a parameter that has shifted in time. This detection and identification methodology is considerably faster than the traditional process in the sense that the direction in which a fault could bring the process operation is well-defined. A well-defined fault space allows the identification of such a fault in a more precise and decisive way than in cases where several samples are needed to corroborate the presence of abnormal operating conditions.

This technique is applied to an Air Heater Lab Experiment (AHLE) in which obstruction of the air flow, changes in the heat throughput and air leaks are considered faults that should be detected and identified, while variations in the ambient temperature or adjustments in the forced flow are expected to be part of normal operating changes.