Determination of uncertainty of automated emission measuring systems under field conditions using a second method as a reference

This report presents a procedure to determine the uncertainty of an automated emission measuring system (AMS) by comparing the results with a second method (REF). The procedure determines the uncertainty of AMS by comparing the final concentration and emission results of AMS and REF. In this way, the data processing of the plant is included in the result evaluation. This procedure assumes that the uncertainty of REF is known and determined in due form. The uncertainty determination has been divided into two cases; varying and nearly constant concentration. The suggested procedure calculates the uncertainty of AMS at the 95 % confidence level by a tabulated t-value. A minimum of three data pairs is required. However, a higher amount of data pairs is desirable, since a low amount of data pairs results in a higher uncertainty of AMS. The uncertainty of AMS is valid only within the range of concentrations at which the tests were carried out. Statistical data processing shows that the uncertainty of the reference method has a significant effect on the uncertainty of AMS, which always becomes larger than the uncertainty of REF. This should be taken into account when testing whether AMS fulfils the given uncertainty limits. Practical details, concerning parallel measurements at the plant, and the costs of the measurement campaign, have been taken into account when suggesting alternative ways for implementing the comparative measurements.