Method validation and MU evaluation of the air pollution source by online-analyzer

What is it about?

The air pollution (PM or SO2) is usually monitored by the online-analyzer, however, this analyzer, belonging to the non-standard test method, need to be validated in comparision with the well-recognized method. Whereupon a large number of real-time data, in multi-sites at different levels, were accumulated under site precision (sR') in-statistical-control condition. In view of the heterogeneour variance, the top-down approach, using closeness sum of squares (CSS), can be applied to qualify for a linear relationship. The data-transformed-residuals-system must submit to the hypothesis testing, the Anderson Darling (AD) statistics, for the statistical-control condition. As the system under investigation can be accepted as none of AD were failed in acceptance at the 95% confidence level for the hypothesis of normality and independence. Therefore, it is appropriate to use the routine data under site precision condition for estimating the top-down uncertainty, which is superior, with the easily simple and directly way, to the bottom-up. In addition, the analyzer system, comprised of dynamic-contributed-uncertainties, can be poolde the combination of the effects on various variances, refrained from the complicated relativity, to an ultimate extent, without complicated computation in the bottom-up approach. Finally, if the online-analyzer system cannot achieve the statistical control, all measurements are to be in vain. This paper survey has paid a very important practical significance at home and abroad, and the practice appears to be convenient and preferable in applications, especially in testing community.

Featured Image

Why is it important?

Our assessors only perform oversights of management system, strict adherence to documented procedures, but the technical mentioned lightly. IUPAC said that “QA is the essential organizational infrastructure that underlies all reliable analytical measurements. In recent years, the management practices have been recognized as essential, however, the prevalence of these favorable circumstances by no means ensures the attainment of appropriate data quality unless IQC is conducted”. Our assessors should have paid close attention to the Technical (QA and QC), in comparison with the Management (QM). 7.7 of ISO 17025-2017 ( Ensuring the validity of results) is a heart in the Technical, I think, just as IUPAC mentioned, “IQC comprises the process of checking that the required uncertainty is achieved in a run”. The relevant requirements on measurement performance should be the result of a dialog between the laboratory and the customer, this objective will be achieved by QA or QC (both is analogous). In the audit, we often come across the intermediate checks on equipment or CRM, but how to tackle this check still puzzled us. We know there isn’t a strict or specified rule. IUPAC’s word remind me that IQC is a final check of the correct execution of all of the procedures (including calibration). IQC is therefore necessarily retrospective. IQC is recognized as such by accreditation agencies. Now that IQC is an essential aspect of ensuring that data released from a laboratory are fit for purpose, if properly executed, it can monitor the various aspects of data quality on a run-by-run basis for AAS, ICP-AES, GC, XRF, etc. By this way for the intermediate check, I insist that ASTM doing can meet and achieve this purpose. Strictly speaking, the intermediate check, should refer to an “analytical system” rather than “method” or “equipment”. At present, the international popular way for QA or QC is the Sheward control ( ISO 8258 ), but with some shortcomings I think. How to guarantee the system under real control, how to operate step by step for IQC, I extremely recommend to adopt ASTM practices. Referring to 7.6 of ISO 17025-2017 ( Evaluation of measurement uncertainty ), GUM approach not for testing laboratories as it is too complicated and impossible to use when essential sources of error are unknown or impossible to quantify and that is usually the case in testing laboratory. Actually, QA/QC approach is, with Type A (data driven) more favored and perfect for estimating uncertainty in the industrial world. I felt curious about why touch on ISO 21748 (top-down) without deeply.

Perspectives