| Abstract: | An assessment of the accuracy of emission measurements is fundamental to an understanding of the ability of an organization to comply with emission regulations. Without a knowledge of the measurement accuracy, the significance of observed differences cannot be assessed. In addition, the average emission level that a group of engines must achieve for compliance is a function of the accuracy of the measurement of the exhaust constituents; i.e., the less accurate the measurement, the lower the average level must be to assure that essentially all engines pass the compliance tests. The accuracy assessment specifically included: 1) test instrument precision, 2) calibration gas accuracy and 3) sample error due to the distribution of data resulting from gas concentrations that are spatially non-uniform. In addition, the accuracy assessment required the consideration of the following general topics: the propagation of errors in a calibration hierarchy, the proper choice of error units, surveys of instrument precision, the evaluation of the significance of various instrument errors, the assessment of the significance of sampling error, and the consideration of methods for demonstrating representative samples and improving instrument precision. An accuracy assessment considering those factors resulted in the following conclusions: 1) Emission instrument precision is best expressed as percent of full scale. 2) The proper units for bias errors are percent of point. 3) Calibration-to-calibration repeatability may be utilized to estimate instrument precision. 4) Instrument precision shows wide variations from day to day. 5) Correcting past data by post-test calibration adjustments will not, in general, improve emission data altered by instrument drift. 6) Sampling error due to the distribution of emission concentrations in space is the single largest source of uncertainty in gas turbine emission measurements. |
