DON'T CALIBRATE! DON'T EVEN VALIDATE! COULD NEW TECHNOLOGIES ELIMINATE ANALYSER MAINTENANCE?

摘要

Initial calibration followed by a structured programme of regular validation tests is the well-established norm for ensuring accuracy in the measurements delivered by online analysers. Statistical Quality Control methodology ensures that random effects are ignored and maintenance intervention, including recalibration, only takes place when there is evidence of assignable errors. This approach meshes well with modern strategies such as Risk Based Maintenance, which seek to minimise maintenance activities and respond only when it is clear that there is an economic benefit to carrying out a maintenance task. This paper raises the question of whether some new analyser technologies have developed to the point where their calibration can be said to be fundamental, and hence inevitably accurate. Furthermore, it considers whether manipulation of the data which are generated in real-time from process streams can provide continuous validation of analysers, thus eliminating the need for SQC checks. An analyser which is always accurate, always validated, seems like an impossible goal... but are there ways in which we could approach that dream? This paper is intended as thought provoking discussion starter.