Process Capability Analysis With GD&T Specifications

Abstract

Geometric dimensioning & tolerancing (GD&T) and process capability indices (PCIs) play critical roles in quality assurance. Conventional PCIs, when used together with GD&T, strongly rely on certain assumptions (e.g. normality and regularity of specification region). GD&T requirements often involve interrelated tolerances, creating irregular tolerance regions. Violation of these assumptions misleads the results (18) and interpretation in applications. A non-conformity (NC) index is developed based on nonparametric distribution model and numerical assessment techniques. Kernel is used for probability density (pdf) estimation and Monte Carlo integration algorithm is adopted for NC analysis, i.e. integration of a pdf over a specification region. The method is validated by case study.