Hybrid Semi-Quantitative Monitoring and Diagnostics of Energy Conversion Processes

Abstract

This paper presents a novel development in the field of automatic, “intelligent” process monitoring. It is possible to show that some of the limitations of a totally qualitative approach may be overcome by judicious use of reconciled experimental data and/or of the results of numerical simulations. The major problem in basing the response of a monitoring & diagnostic system on a process simulator is that, to run efficiently in real time, the simulator must introduce some simplification in the process model, and therefore its reliability as a source of “process data” is negatively affected. The approach proposed in this paper consists of adopting a mix (thence the attribute “hybrid” in the title) between reconciled data and physical modeling, to extract a limited number of numerical coefficients that introduce a sufficient degree of “quantification” in a qualitative monitoring system. The result is a fast and reliable intelligent procedure that assists human operators by presenting them a preliminary fault analysis based on a limited set of relevant reconciled process variables. An application to a regenerated gas turbine expansion process is discussed in detail.

Keywords

• diagnostics, prognostics, process health monitoring