Simulation and optimization of three-phase distillation processes

The relationship between two-phase and three-phase solutions to heterogeneous distillation simulation and optimization problems is studied. Simulation results illustrate that bifurcation points can occur in the two-phase solutions and that, in regions of steady-state two-phase solution multiplicity, not all two-phase solutions can be used to compute a three-phase solution. Numerical difficulties are attributed to marked differences in the separation performed by the two- and three-phase solutions and to ill-conditioning when a second liquid phase is present in small amount. The optimization study shows that multiple local optima with different numbers of three-phase trays can occur. These local optima need not be spurious and indicate that some global optimization method is required to guarantee the correct solution is located. Accordingly, the tunneling algorithm for global optimization is used to solve a three-phase distillation optimization problem. Results show that tunneling can be used to successfully find the global solution.