ASTM D6621 Performance Testing of Process Analyzers for Aromatic Hydrocarbon Materials
1. Scope
1.1 This practice serves as a practical guide for the performance testing of process stream analyzers specifically for measuring chemical or physical characteristics of liquid aromatic hydrocarbon materials for production or certification of these materials. The practice may be applicable to other hydrocarbon stream analyzers as well.

1.2 Only external methods (complete substitution of the process stream with a standard) of control sample introduction are included. Internal methods are beyond the scope of this practice.

1.3 Methods for resetting key operational parameters of analyzers to match predefined limits are provided by vendors and are not included in this practice.

1.4 Analyzer validation procedures are covered in Practices D3764 and D6122, not in this practice.

1.5 Procedures for statistically interpreting data from automatic sampling process stream analyzers are outlined.

1.6 The implementation of this practice requires that the analyzer be installed according to APIRP-550 (1), and be in agreement with the analyzer supplier's recommendations. Also, it assumes that the analyzer is designed to monitor the specific material parameter of interest, and that at the time of initial or periodic validation, the analyzer was operating at the conditions specified by the manufacturer and consistently with the primary test method.

1.7 The units of measure used in this practice shall be the same as those applicable to the test primary method used for analyzer validation.

1.8 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this standard to establish appropriate safety and health practices, and determine the applicability of regulatory limitations prior to use.

2. Referenced Documents
2.1 ASTM Standards:
D3764 Practice for Validation of the Performance of Process Stream Analyzer Systems
D4177 Practice for Automatic Sampling of Petroleum and Petroleum Products
D6122 Practice for Validation of the Performance of Multi-variate Online, At-Line, and Laboratory Infrared Spectrophotometer Based Analyzer Systems
E456 Terminology Relating to Quality and Statistics
E1655 Practices for Infrared Multivariate Quantitative Analysis

3. Terminology
3.1 Definitions:
3.1.1 accuracy, n - closeness of agreement between a test result and an accepted reference value.

3.1.2 analyzer output, n - signal that is proportional to the quality parameter being measured and suitable for input to readout instrumentation.
3.1.2.1 Discussion - It may be pneumatic, electrical, digital, etc., and expressed as psi, mv, sec., etc.

3.1.3 analyzer result, n - numerical estimate of a physical, chemical, or quality parameter produced by applying the calibration model to the analyzer output signal.

3.1.4 bias, n - the difference between the expectation of the results and an accepted reference value.

3.1.5 control sample, n - material similar to the process stream that is stable over long periods of time so that its parameters may be measured reproducibly in performance tests to characterize analyzer precision and accuracy.
3.1.5.1 Discussion - May be a pure compound, standard mixture, or a sample from the process stream. Its parameters are used to plot statistical process control charts to define analyzer precision in normal operation.

3.1.6 external performance testing, n - procedure involving complete substitution of the process/product stream measured by the analyzer with a control sample stream to measure the analyzer's precision and possibly accuracy (if the control sample's true value is known).

3.1.7 internal performance testing, n - procedure involving the addition of a known quantity of a standard material homogeneously into the process/product stream measured by the analyzer to measure the analyzer's precision and possibly accuracy (if the sample material's true value is known).

3.1.8 linearity, n - parameter ranges where the analyzer's results do and do not approximate a straight line.

3.1.9 performance testing of an analyzer, n - mechanical and statistical procedure for routinely checking the accuracy and precision of an analyzer's results against historical accuracy and precision for a control sample.

3.1.10 precision, n - closeness of agreement of independent test results of the same chemical or physical property of a given material obtained under stipulated conditions.
3.1.10.1 Discussion - Expressed in terms of dispersion of test results around the arithmetic mean, usually as variance, standard deviation, repeatability or reproducibility, or both.

3.1.11 repeatability of an analyzer, n - difference between two successive analyzer results measured in a short time interval that would be exceeded in the long run in only 1 case in 20 (5 % of the time) when the analyzer is operated on a flowing sample of uniform quality.

3.1.12 reproducibility of an analyzer, n - difference between a single result from each of two identical analyzer systems that would be exceeded in the long run in only 1 case in 20 (5 % of the time) when the two systems are operated at different sites by different operators, but on identical samples.

3.1.13 rule violation, n - condition when a point value or pattern of points in a statistical process control chart statistically exceeds the defined probability of its occurrence, as defined by the Western Electric rules (2) being used.

3.1.14 spot sample, n - representative material resembling the stream being monitored, an identical portion of which is analyzed both in a process analyzer and by a laboratory test on a non-scheduled basis for periodic validation testing.
3.1.14.1 Discussion - May be the same material as the control sample.

3.1.15 validation of an analyzer, n - process to identify how comparable an analyzer's results are statistically to results from the primary method, or to define how the analyzer's results compare to the primary method's results in precision and accuracy.
3.1.15.1 Discussion - Must be done when the analyzer is first configured or reconfigured (initial validation), and then on a periodic basis (periodic validation), as described in Practice D3764.

3.2 For additional definitions, see Appendix X1.