Gage R2&E2: an effective tool to improve the visual control of products

Abstract

Purpose The variability of the results of a visual control is often high. This paper aims to propose a new tool to give information about what improvement actions can be carried out to reduce this variability. Design/methodology/approach The variability of a visual control can be measured by Kappa's Fleiss which measures the level of agreement between appraisers and experts. The R&R Gage is then classically used to give information about corrective actions which can be carried out in order to improve this level of agreement. The paper demonstrated that this information is not always sufficient. Findings By considering the two essential steps of a visual control (exploration and evaluation), the R2&E2 Gage proposed gives more precise information about the improvement actions to carry out to reduce the variability of a visual control. Repeatability and reproducibility, for detection and evaluation purposes, are considered separately. Research limitations/implications This R2&E2 gage is one result of a European research program called INTERREG. The aim of this program, which brings together two laboratories from the University of Savoy and EPFL, two institutional partners (CTDEC and CETEHOR) and some Swiss and French industrial companies, is to create methodological support and the tools needed to improve the visual control of high added‐value products. Practical implications This R2&E2 gage has been used in six industrial companies involved in the European program INTERREG. Significant improvement of the visual control has been observed over a short time. Originality/value The paper fulfils an identified need of industrial firms to have efficient tools improving the visual control of their products.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.