Gross Error Detection Based on Serial Elimination: Applications to an Industrial Gas Turbine

Abstract

Gross Error Detection (GED) is a technique used to identify possible systematic errors in measurements and validate data for a further diagnostic phase. It is always applied along with Data Reconciliation (DR), a technique to improve the accuracy of process data by adjusting the measured values to fit the process equations describing the physical phenomena. They have been applied for a long time to chemical plants with balance equations (mass and composition) and recently extended to industrial power plants [1,2,3]. In this paper a well-known GED technique based on serial elimination has been applied in a gas turbine plant operating in a combined cycle power plant. In a first analysis errors have been imposed manually in the field data to understand the minimum error amplitude avoiding the smearing effect: at the beginning a single gross error has been imposed on the fuel flow rate and on the compressor discharge temperature respectively, then multiple gross errors have been imposed simultaneously on the same measurements. The single gross error tests showed a high capacity of detection and localization, while the multiple gross error analysis highlighted the problems due to the smearing effect (the minimum error intensity to detect and locate errors increased with respect to the single error case). In a second analysis the GED technique has been used to detect and locate a gross error among the three sensors measuring the compressor discharge temperature. The main objective was to analyze the ineffectiveness in error detection and localization of using the mean for redundant measurements.