Prognostic analysis of defects in manufacturing

Abstract

Purpose

Since works-in-process (WIPs) are highly vulnerable to defects because of the variety and complexity of manufacturing processes, the purpose of this paper is to describe how to utilize existing analytics techniques to reduce defects, improve production processes, and reduce the cost of operations.

Design/methodology/approach

Three alternatives for diagnosing causes of defects and variations in the production process are presented in order to answer the following research question: “What are the most important factors to be included in prognostic analysis to prevent defects?”

Findings

The key findings for the proposed alternatives help explain the characteristics of defects that have a great impact on manufacturing yield and the quality of products. Consequently, any corrective action and preventive maintenance addressing the common causes of defects and variations in the process can be regularly evaluated and monitored.

Research limitations/implications

Although the focus of this study is on improving shop-floor operations by reducing defects, further experimentation with business analytics in other areas such as machine utilization and maintenance, process control, and safety evaluation remains to be done.

Practical implications

This study has been validated with several scenarios in a manufacturing company, and the results demonstrate the practical validity of the approach, which is equally applicable to other manufacturing sub-sectors.

Originality/value

This study is different from the others by providing alternatives for diagnosing the root causes of defects. Control charts, costs of defects, and clustering-based defect prediction scores are utilized to reduce defects. Additionally, the key contribution of this study is to demonstrate different methods for understanding WIP behaviors and identifying any irregularities in the production process.

Citation

Chongwatpol, J. (2015), "Prognostic analysis of defects in manufacturing", Industrial Management & Data Systems, Vol. 115 No. 1, pp. 64-87. https://doi.org/10.1108/IMDS-05-2014-0158