Date of Award

Summer 2000

Document Type


Degree Name

Doctor of Philosophy in Chemical Engineering - (Ph.D.)


Chemical Engineering, Chemistry and Environmental Science

First Advisor

Piero M. Armenante

Second Advisor

Gordon Lewandowski

Third Advisor

Robert Benedict Barat

Fourth Advisor

Norman W. Loney

Fifth Advisor

Pushpendra Singh


The optimization of reaction processes to maximize the yield of a desired product while minimizing the formation of undesired by-products is one of the most important steps in process development for drug manufacturing or fine chemical production. In many situations the kinetics of product formation can be quite complex and involve a number of intermediate steps as well as parallel and serial reactions. This renders the systems sensitive to the operating conditions. Often, upon scale-up, a decrease in the yield of the desired product is experienced, while more undesired by-products are produced. A large number of interrelated variables influence the outcome of a process, and furthermore, most systems of process interest take place under turbulent conditions. Present methods for process design are inadequate because they involve the use of lumped parameters which fail to capture essential flow details and the rapid changes in the local concentration of reactants.

In recent years Computational Fluid Dynamics (CFD) has been successfully used to model the fluid dynamics of complex vessels (such as agitated reactors) and to predict the velocity distribution in turbulent systems such as mixers and reactors. In this work a novel approach based on the use of CFD coupled with micromixing models was used to predict the behavior of a multiple, competitive reaction system in cylindrical stirred tank reactors fitted with a variety of agitators. In particular, the following fast parallel competing reactions scheme (Bourne and Yu 1994) was modeled, and the results compared with original experimental data:

NaOH (A) + HCl (B)--k1--> NaCl(P) + H2O

NaOH(A) + CH2ClCO2C2H5(C)-- k2--> CH2ClCO2Na(Q) + C2H50H(S)

The reactor was operated in semi-batch mode, with the limiting reagent (A) being slowly added to the contents of the reactor in which the other reagents (B and Q were already dissolved. The final yield of the undesired product (S) was experimentally measured. The flow field in the reactor was simulated using the Reynolds Stress (RSM) turbulence model. The full impeller geometry was incorporated in the CFD simulation using the Multiple Reference Frames (MRF) model. The reaction zone was modeled in a Lagrangian way using a multi-phase Volume of Fluid (VOF) model (Hirt and Nichols 198 1). The interaction of turbulence and reaction was accounted for by means of the engulfment-based models for micro-mixing (Baldyga and Bourne 1989a; Baldyga and Bourne 1989b; Baldyga et aL 1997). The agreement between experimental velocity distribution data and the results of the simulations was generally good. The micro-mixing models, in conjunction with CFD, predicted a final yield in close agreement with the experimental data, demonstrating that the proposed approach can be successfully used to model turbulent reactive systems without the need for experimental input.