OCR Text |
Show form required by the viscosity prediction formula. The Watt-Fereday formula requires the ash composition to be in terms of the weight fractions of Si02 , A£203, Fe203, FeO, and CaO. (The formula assumes that MgO makes up the balance.) The calculated composition, on the other hand, consists of the above species but also multiple oxides and silicates such as 3AZ203«2S102, CaO«Si02, CaO-A£203-2Si02, etc. These multiple compounds are divided into the six Watt-Fereday species for the purpose of obtaining the appropriate formula inputs. Following this, the viscosity of the liquid phase of the slag mixture is calculated using the Watt-Fereday formula, and the mixture viscosity is then obtained by use of Equation (1) to account for the presence of solids. Results of the Viscosity Predictions The methodology of extending viscosity prediction formulas to sub-liquidus temperatures has been applied to two slags for which experimentally measured viscosities were available (7). One slag, designated slag A, has a viscosity behavior which is predicted very well by the original Watt-Fereday formula, even at sub-liquidus temperatures. The second slag, designated slag B, has a viscosity behavior which deviates markedly from the Watt-Fereday prediction for temperatures below the liquidus temperature. The ASTM analysis of the slags are given in Table 1. Note that slag B has a much higher iron content and a substantially lower silica content than slag A. The equilibrium code compositions for the liquid phase of the two slags as a function of temperature are presented in Table 2. These are the compositions of the slags after being converted to the Watt-Fereday input format by the procedure described in the previous section. The viscosities predicted by the extended viscosity method, as well as the experimentally measured viscosities and the Watt-Fereday formula predictions are shown in Figures 1 and 2. Figure 1 for slag A shows that the Watt-Fereday formula prediction matches the experimental measurements very well, while the extended viscosity method predicts higher viscosities than the experimental measurements. The extended viscosity method however predicts the same qualitative behavior of the viscosity in that no sudden change in the slope of the curve is present, such as would occur for a slag which is characterized by a definite 3-4 |