Detecting, Identifying and Managing Sources of Variation in Production and Product Development

Abstract: It is well known that customers have varying needs and wants and thus expect to have a broad spectrum of products to choose from. However, they hardly tolerate variation in products produced to the same specifications. This variation, which originates from multiple sources and adversely affects product quality, should be detected, identified and managed throughout the Product Realization Process (PRP). While the PRP is a widely defined concept, this thesis focuses on its two distinct parts, namely production and product development. There are essentially two strategies for dealing with variation. One is to detect and eliminate sources of variation. This is usually done using Statistical Process Control (SPC) procedures. The second is to minimize the impact that the sources of variation have on important product characteristics. This is usually done using Robust Design Methodology (RDM) tools. The main objective of this thesis has been to develop statistical methods that support the execution of both strategies. More specifically, in order to identify and assess the risks related to variation from early phases of product development, a new statistically based engineering method, Variation Mode and Effect Analysis (VMEA), has been developed. Furthermore, to effectively address variation-related problems in production, especially when traditional control charts cannot detect assignable causes of variation, a number of modifications to Shewhart?s criteria and Exponentially Weighted Moving Average (EWMA) control charts have been proposed and applied in practice. Finally, some contributions have been made to the understanding and definition of predictability as an important notion in statistical quality control.