(368b) Validation of the Coalescence-Dispersion Model for Complex Chemical Reactions

"Validation of the Coalescence-Dispersion Model for Complex Chemical Reactions"

Gary K. Patterson, Professor Emeritus

Missouri University of Science and Technology

Chemical and Biological Engineering

Rolla, Missouri 65401

The integration of the coalescence-dispersion (c-d) model of mixing into a CFD simulation code has been extensively tested. The c-d model simulates mixing by assuming that the fluids may be represented by mobile fluid points which carry the mass, concentrations and other properties of the fluids. Upon coalescence of two randomly chosen fluid points, their properties are averaged (concentrations, enthalpies, etc.). Upon dispersion (separation) the two fluid points have identical properties. In a c-d simulation of a chemical reaction, the reactions occur for the length of the time increment between flow events as if the fluid points are batch reactors. Use of the c-d model avoids the need for a closure model for the mixing in the presence of chemical reactions. Such closures have been successful only for single reactions and for two competing reactions.

Integration of the c-d model into a CFD simulation requires that the mobile fluid points move, mix and disperse in a way dictated by the CFD simulation. Each volume segment of the CFD simulation is assigned a number of fluid points which move from segment to segment as dictated by the CFD simulation. This is accomplished by a subroutine linked to the CFD code that accesses the velocity and mixing rate predictions of the CFD. A critical aspect of the c-d model is in the most effective choices of the time increment and number of fluid points in each CFD volume segment. This is dependent on the number of c-d events during each time increment and flow volumes from segment to segment. Of course, the accuracy of the CFD simulation is critical for the accuracy of the c-d simulation.

Validations of c-d simulations have been done by comparing results with well-established experimental data for single and multiple reactions in various types and sizes of chemical reactors. Examples of validation data used are experiments by Paul and Treybal, Penney and coworkers, Middleton and coworkers, Bourne and coworkers and by the author and students. The validations show that the c-d simulation method can be a valuable addition to experimental research on multiple competitive reactions where there is interest in yields of chosen products.