TABLE OF CONTENTS This case involves manufacture of a polymer product to be used for the food industry. Consequently, stringent regulations had to be met, requiring the removal of residual monomers from the liquid polymer product. In this case, the use of a novel catalyst required designing a new back-end separation process for removing the undesired monomers.
Designing a system for removing a particular component from the liquid phase requires an understanding of the liquid-vapor equilibrium behavior of each component in the mixture. However, no fundamental thermodynamic models were available for determining the liquid-vapor properties (hereafter referred to as partition coefficients) of the three monomers to be removed from the liquid phase of the non-ideal mixture in this particular case. The partition coefficient for a component in a mixture is instrumental in determining its liquid-vapor phase equilibrium properties that govern the amount of that component present in the liquid phase. Therefore, determining the most suitable operating conditions for removing a component from the liquid mixture requires knowledge of the partition coefficients under a wide range of mixture conditions.
Partition coefficients of components in non-ideal mixtures are complex functions dependent on the temperature, pressure, and composition of the mixture. As no fundamental thermodynamic models were available for the system in this case, pattern recognition technology was harnessed to develop data-driven models from experimental laboratory results, thereby enabling the process designers to develop the commercial process. The models had to be robust as they would be required to extrapolate beyond laboratory conditions.
