Análises comparativas de erro de medição através de estudos GR&R cruzado e expandido

Abstract

In the quest for the quality of a manufacturing process, it is necessary to control and evaluate a unavoidable source of variation, the measurement error. The Gage Repeatability and Reproducibility (GR&R) studies is most commonly used for sources of variation into the Measurement System Analysis (MSA). The aim of this research is to compare the GR&R crossed and the GR&R expanded from their estimation of the indexes of classification of the measurement system, the first is limited to estimate the variation of the manufacturing process only by the difference between the pieces, the second is more efficiently using more factors in a variance analysis model. This comparison was performed in a coating welding experiment by analyzing five cases, each with one critical-to-quality (CTQ), three one-dimensional, the high reinforcement, the penetration and Width of weld bead and two two-dimensional, the Penetration Area and the Total Area. The contributions are to point out in the context of this experiment what is the best experimental arrangement for the classification of the measurement system and which is the most efficient to diagnose reproducibility and repeatability errors. According to the statistical and graphical analyzes, it is evident that the GR&R expanded with interactions is more efficient and complete than the GR&R expanded without interactions and the GR&R crossed.