Synthesis and analysis of flexible and complex heat-integrated distillation sequences

This dissertation considers the synthesis problem of flexible integrated separation systems and complex distillation systems.; First, the synthesis of flexible heat integrated and non-heat integrated distillation sequences for multiperiod operation is addressed. This problem considers the case of operation subject to a specified set of discrete changes in the feed flowrate and/or composition to the separation system. A superstructure representation of the problem is constructed and mixed-integer linear programming MILP techniques are used to model and solve for the optimal design.; A methodology is then presented for algorithmic development and implementation of discrete continuous optimization problems through a high-level modeling language. Emphasis is given to mixed-integer nonlinear programming MINLP problems and algorithms requiring decomposition. This methodology serves as an important tool for the remainder of the research.; Next, a systematic approach for the synthesis of single period heat integrated distillation sequences is presented based on a superstructure and mixed-integer nonlinear programming formulation. Discrete variables represent the selection of individual columns and utility matches. Nonlinearities are introduced by treatment of the column pressures explicitly as continuous decision variables. Theoretical aspects and an implementation based on the proposed algorithmic development methodology are discussed.; For the synthesis of multiperiod heat integrated distillation sequences, a theoretical framework and solution procedure based on a combined stochastic and mixed-integer nonlinear programming approach is also proposed.; Optimal synthesis of conventional and complex distillation columns is then addressed. Design strategies based on a superstructure approach and a corresponding mathematical formulation that combines rigorous model representations of the distillation column(s) with discrete-continuous nonlinear optimization. Theoretically based treatment of the discrete nature of the equilibrium stage process is provided through introduction of binary decision variables to represent each tray.; Finally, a set of novel optimization approaches are described for the solution of phase and chemical equilibria problems through minimization of the Gibbs free energy. Discrete variable optimization is employed in which the appearance and disappearance of phases in the system are posed as binary decision variables within the nonlinear programming formulation of the free energy minimization.