COMPARISON OF DOUBLE LOG-LOG, MIXTURE RESPONSE-SURFACE AND COMBINED NIBS REDLICH-KISTER SOLUBILITY MODELS/

The accuracy and prediction capability of the linear double log-log (L DL-L), mixture response-surface (MR-S) and the combined nearly ideal b inary solvent/Redlich-Kister (CNIBS/R-K) solubility equations have bee n compared using the model parameters calculated from either the whole data or a minimum number of data in an experimental set. The CNIBS/R- K model produced better prediction for some experimental sets than the other two models when the parameters obtained from the whole data in a set were employed, whereas the LDL-L model was superior to the other models when the parameters calculated from a minimum number of data w ere used, indicating its greatest prediction capability.