A Case Study in Mixture Design: Multi Response Optimization of Glaze Formulation

ABSTRACT

The main purpose of this study was to optimize the glaze formulation of a company that produces vitrified products. In this type of process, the chemical composition of the glaze has a critical effect on quality indicators such as coefficient of expansion, R value, and creeping value. The procedure used in this study included design, modeling, optimization, and confirmation steps. By the selection of desired ranges of multiple responses, the optimum glaze formula that minimizes the cost was determined by Derringer-Suich optimization. The verification results for the optimum glaze formula exhibited good agreement between the regression models and experimental measurements.

KEYWORDS:

• D-optimal design
• glaze formulation
• mixture
• optimization

ACKNOWLEDGMENTS

The experimental procedure was performed with contributions of ESVIT Inc. and the Ceramic Research Center (CRC). The authors would like to thank the employees of CRC and Professor Alpagut Kara, the Head of CRC. The participation of Professor A. Sermet Anagun from the Industrial Engineering Department at Eskisehir Osmangazi University is specially acknowledged.

ABOUT THE AUTHORS

Ezgi Aktar Demirtas has been working at the Department of Industrial Engineering of Eskisehir Osmangazi University. She received her MS degree and Ph.D. in Industrial Engineering from the same university. She has several national and international publications and research projects in her main research interests. Her main research interests are QFD, Response Surface Methodology, multi criteria decision making and Service Systems.

Onur Ayva has been working as industrial engineer in Esvit Ceramic Company. He holds BS degree from Eskisehir Osmangazi University. His main research interests include quality control and production planning.

Yeliz Buruk is research assistant in Industrial Engineering Department. She holds BS and MS from Eskisehir Osmangazi University. Her research interests include design of experiments and response surface methodology.