Statistical mixture experiment techniques were applied to a waste glass data set to investigate the effects of the glass components on Product Consistency Test (PCT) sodium release (NR) and to develop a model for PCT NR as a function of the component proportions. The mixture experiment techniques indicate that the waste glass system can be reduced from nine to four components for purposes of modeling PCT NR. Empirical mixture models containing four first-order terms and one or two second-order terms fit the data quite well, and can be used to predict the NR or any glass composition in the model domain. The mixture experiment techniques produce a better model in less time than required by another approach.
Origin
Pacific Northwest Labs, Usa
Journal Title
J Am Ceram Soc 80 12 1997 3038-3044
Sector
Special Glass
Class
S 1312