Abstract: A chemometric method for predicting unknown properties in polymers using an on-line NMR system comprising the steps of producing a predictive data set and using the predictive data set to obtain unknown amounts of properties in a polymer. In a preferred embodiment, the process begins by obtaining free induction decays for samples of polypropylene with measured concentrations of xylene soluble polypropylene from a xylene soluble polypropylene data set to produce a free induction decay data set. The free induction decay data set is analyzed using PCA to produce a principle component data set. The principle component data set, the xylene soluble polypropylene data set, and the free induction decay data set, are analyzed using partial-least squares analysis to produce a training data set and the training data set is subsequently validated to produce a predictive data set.

Abstract: A chemometric technique for predicting styrene content in Butadiene-Styrene resin (K-Resin) using an on-line NMR system comprising the steps of producing a predictive data set and using the predictive data set to obtain unknown concentrations of styrene in samples of K-Resin. The predictive data set is generated by obtaining free induction decays for samples of K-Resin with measured concentrations of styrene to produce a free induction decay data set; analyzing the free induction decay data set using PCA to produce a principle component data set; analyzing the styrene concentrations, the free induction decay data set and the principle component data set using PLS to produce a training data set; and validating the training data set to produce a predictive data set. Using the predictive data set involves the steps of obtaining free induction decays of samples of K-Resin with unknown concentrations of styrene and applying the free induction decays to the predictive data set to predict the unknown concentrations.