Statistical analysis of viscosity, electrical resistivity, and further glass melt properties


Alexander Fluegel, David A. Earl, Arun K. Varshneya, and Dolun Oksoy

New York State College of Ceramics at Alfred University

Alfred, New York 14802


Chapter 9 in: "High temperature glass melt property database for process modeling"; Eds.: Thomas P. Seward III and Terese Vascott; The American Ceramic Society, Westerville, Ohio, 2005, ISBN 1-57498-225-7 (announcement flyer, PDF, 180 kB)


Abstract:

The application of multiple regression techniques for glass property modeling is described. Multiple regression was applied for the development of models for the viscosity at log(viscosity/Pa*s) = 1.5 to 12.0 and for the high temperature electrical resistivity at T = 1000°C to 1400°C. The models are based on 150 composition-property data of commercial glasses, including low-expansion borosilicate, textile fiber type E, fiber wool, TV panel, container, and float glasses. The model errors were found to depend on the glass type and on the property measured. The best predictions were possible for soda lime container and float glass viscosities (error 1-8°C), and less accurate results were obtained for borosilicate and TV panel glass viscosities (error 2-30°C). High temperature electrical resistivities could be modeled with errors of log(resistivity/Ohm*cm) = 0.005-0.015 for container and float glasses, and log(resistivity/Ohm*cm) = 0.01-0.13 for borosilicate glasses. The models were evaluated in comparison to experimental data from the scientific literature. The influences of all glass components are discussed, and relations between the high temperature viscosity and electrical resistivity are examined. In Appendix (C) of this chapter initial modeling results for the water solubility and surface tension of glass melts are presented.


Database available for purchase from Wiley Publishers and Amazon.com.


Models* for the viscosity, electrical conductivity/resistivity, and water solubility can be downloaded here. Selected experimental data are online in a preliminary report (2003) to the US Department of Energy.


* The models derived from this database are only recommended for the viscosity of soda-lime glasses and the water solubility in glass melts. Otherwise, the global viscosity and electrical resistivity models are superior.