TABLE OF CONTENTS As discussed in Secs. 8-5 and 8-6, activity coefficients in binary liquid mixtures can often be estimated from a few experimental vapor-liquid equilibrium data for the mixtures by using some empirical (or semiempirical) excess function, as shown in Table 8-3. The excess functions provide a thermodynamically consistent method for interpolating and extrapolating limited binary experimental mixture data and for extending binary data to multicomponent mixtures. Frequently, however, few or no mixture data are at hand, and it is necessary to estimate activity coefficients from some suitable prediction method. Unfortunately, few truly reliable prediction methods have been established. Theoretical understanding of liquid mixtures is limited.
Therefore, the few available prediction methods are essentially empirical. This means that estimates of activity coefficients can be made only for systems similar to those used to establish the empirical prediction method. Even with this restriction, with few exceptions, the accuracy of prediction is not likely to be high whenever predictions for a binary system do not utilize at least some reliable binary data for that system or for another that is closely related. In the following sections we summarize a few of the activity-coefficient prediction methods useful for chemical engineering applications.
Following ideas first introduced by van der Waals and van Laar, Hildebrand and Scatchard working independently (Hildebrand and Scott, 1962), showed that for binary mixtures of nonpolar molecules, activity coefficients ? 1 and ? 2 can be expressed by
where V
