ESTIMATION OF ASH SOFTENING TEMPERATURES USING CROSS TERMS AND PARTIAL FACTOR-ANALYSIS

The relationship between elemental composition of coal ash and ash softening temperature has been studied, using a set of 70 ashes prepared from blends of seven source coals. Simple multiple linear regression analysis is of limited value. The inclusion of second-order cross terms, representing acid-base and other metathetic reactions, markedly improves the precision of predictive regressions. Excessive collinearity, notably among five major components of these coal ashes, limits the combinations of regressor terms usable in a predictive equation. A strategy for avoiding collinearity problems by means of partial factor analysis is illustrated. Best estimates of ash softening temperature show root mean square errors of 45 °F (25 K) or less. © 1990, American Chemical Society. All rights reserved.