Jens Folke's presentation on measurement system analysis provides a detailed exploration of how organizations can validate and optimize their measurement capabilities to support quality assurance and Six Sigma initiatives.
With a background in environmental chemistry and 25 years of consulting experience, Folke presents a systematic approach to understanding measurement systems before they are deployed in production environments.
The core message emphasizes that measurement system variation must be small relative to product specifications, ideally consuming no more than one-tenth of the tolerance band.
The presentation distinguishes between continuous data, which contains rich information for statistical analysis, and discrete or attribute data, which provides limited information but is often unavoidable in practice.
Folke highlights the importance of questioning data sources, including whether measurements are passive (collected automatically) or active (influenced by human awareness), and whether data sources are trustworthy and free from bias.
The speaker outlines multiple measurement system analysis methodologies, from simple Type 1 studies using repeated measurements of calibrated standards to complex Gauge R&R studies involving multiple operators and parts.
Key technical concepts include accuracy versus precision, bias detection through external standards, linearity assessment across measurement ranges, and the impact of hysteresis and dead band effects on system performance.
Practical considerations such as calibration procedures, environmental conditions, equipment maintenance, and warm-up time are presented as essential elements of measurement system design.
The presentation addresses both destructive and non-destructive testing scenarios, explaining how nested study designs accommodate situations where measurements cannot be repeated.
Throughout the discussion, Folke emphasizes that investing time in proper measurement system analysis prevents costly errors in quality decisions and ensures that process improvements are based on reliable data rather than measurement system artifacts.
From the meeting "Certain When Uncertain".
