Physical modelling of flow behavior in a stirred glass system

Abstract

As stirring efficiency made the use of isothermal models plausible, dimensional analysis was used to extend model laws to include stirrers, gobbing cams and to indicate the conditions required to get data representative of actual or proposed units. Based on fullsize modelling, the production department installed stirred glass set-ups and made excellent products. Correlation between modelling and production was excellent and included mechanical variables and temperatures. Furthermore, a new delivery system, including a new gobbing cam designed for a very fluid photosensitive glass, was based on modelling and behaved exactly as predicted. Upsets in the stirred glass, delivery Systems resulted in slow cyclic weight variations of about 1 %. As the weights of the items being made were specified to within ±1.0%, it was clear that control had to be improved. The weight changes were not caused by compositional changes in the glass but to some Instrumentation problems and to some unique, unknown problem(s) associated with stirring. Thus, a better understanding of how temperature and the stirrers controlled flow was needed. Assuming that flow is inversely proportional to viscosity showed that the change in weight should be about 0.9%/°C. This number is about 1 0% too low compared with the empirical value of 1 %/°C quoted by Operators. Next data from a full-size, stirred model were summarized in an equation which relates flow to the viscosity of the fluid and to the rotational speed and vertical position of the stirrer. Partial differentiations and evaluations of this equation at a nominal operating condition showed that the weight should first, vary by 1.02%o/°C, second, decrease by 0.90% per rpm increase and third, increase by 0.80% when the stirrer was raised (0.006") or 1 % when it was at the 5/8" position.