| The objective of this work is, on the one hand, to develop numerical techniques to overcome the difficulties encountered in the steady-state and dynamic simulation of azeotropic distillation towers, and on the other, to provide some valuable insight on azeotropic distillation.; Steep fronts are encountered in the temperature and mole fraction profiles of an azeotropic distillation tower. Due to these fronts the algorithms for the simultaneous solution of the MESH (Material balance, Equilibrium, Summation of the mole fractions, and Heat balance) equations either fail to converge, or require many iterations to give undesirable solutions. In the first Chapter of this work modifications to such an algorithm are presented and they lead to reliable convergence. Results of simulation studies show the existence of three distinct regimes of operation in azeotropic distillation. However, these results were obtained without considering the material balances and equilibrium of the decanter, which provides the entrainer reflux.; There is a narrow range of design variables, or specifications, that give high recovery of alcohol in the azeotropic tower and an overhead vapor that can be condensed into two liquid phases with positive flow rates in the decanter and decanter bypass stream. There is a small window of overhead vapor composition within which this objective is achieved. In this region, the overhead vapor condenses into two liquid phases, but is in equilibrium with a single liquid phase on the top tray. This region can easily be located with bubble point calculations.; A non-linear programming problem has been formulated, in which the design variables are adjusted to position the overhead vapor composition within a feasible window, while satisfying the decanter equilibrium and material balances. The optimization problem is solved efficiently and reliably using Powell's algorithm; given its solution, the flow rate of the aqueous phase from the decanter is specified to permit routine cleanup by a stripping tower.; A case of steady-state multiplicity has been reported in the literature for the system ethanol-benzene-water, assuming constant molal overflow and using activity coefficient parameters obtained from vapor-liquid equilibrium data. These parameters predict an erroneous binodal curve and consequently, the non-linear programming formulation could not be used to confirm the steady-state multiplicity. . . . (Author's abstract exceeds stipulated maximum length. Discontinued here with permission of school.) UMI... |
