Multivariate Process Capability Assessment and Improvement: A Case Study

Abstract

In today's competitive manufacturing environment, the challenge is to responsively produce products with minimum cost and high quality. Achieving and controlling the targeted quality level in manufacturing processes does not only increase customer satisfaction, but it can also result in significant cost and time savings. Further, measuring the process performance is a critical issue in process improvement initiatives. The common practice in several industries is using the Univariate Process Capability Indices (UPCIs) to measure the process performance, which are based on only a single quality characteristic. In most of the applications, it is not acceptable to judge the performance based on a single quality characteristic as it actually relies on more than one characteristic. In this paper, univariate and multivariate PCIs are used to measure the performance of the flare making process. This process is a critical step in the straight fluorescent light bulb production line. In addition, multivariate control charts such as the Hotelling ܶ ଶ as well as the Multivariate Exponentially Weighted Moving Average (MEWMA) are constructed for the collected data to verify that the process is in control before assessing its capability. Besides, Principal Component Analysis (PCA) and Joint Normal Distribution (JND) techniques are applied in the multivariate process capability assessment. In this paper, Multivariate Process Capability Indices (MPCIs) have been evaluated to compare the process performance before and after improvement efforts. In the considered case study, MPCIs provide the user with an overall assessment of process capability regardless of the fluctuations in the individual variables capabilities.