(145a) Optimization-Based Azeotropic Distillation System Synthesis Using Geometric Insights

Distillation is the most prevalent technology to separate mixtures in chemical industries. In multicomponent distillation, encountering highly nonideal mixtures which form azeotropes is common. The presence of azeotropes increases the difficulty of separation, leading to complex separation networks with several columns. Due to the energy- and capital-intensive nature of a distillation system, early decisions such as the number and design of columns and their connections become important and greatly affect the final economics of the process.

There are two major approaches in distillation system synthesis: sequential and simultaneous. Evolutionary algorithms can also be incorporated to assist the configuration search. In these approaches, most studies apply rigorous column design algorithms for pre-generated system configurations [1], [2]. In simultaneous approaches, decisions regarding the network configuration and column design are obtained simultaneously by solving an optimization problem. In these approaches, efficient representation of the system (columns and their connections) is essential for computational tractability. Available studies on simultaneous approaches assume linear distillation boundaries, require fully specified system inlet stream information, and employ local optimization solvers [3].

In this work, we present a simultaneous approach to synthesizing distillation systems for multicomponent azeotropic mixture separation. The approach employs a generalized problem statement [4] enabling the integration with reactor network synthesis. We approximate the residue curve map as ideal compartments [5] and apply the SA matrix method [6] for each compartment to represent promising configurations. In these compartments, we treat both the pure components and the azeotropes as pseudo-components whose relative volatilities are constant. We employ the modified Underwood equations [7] to describe the columns in the pseudo-component basis. The synthesis problem is formulated as a mixed-integer nonlinear programming (MINLP) model where binary variables are used to represent discrete decisions (e.g., connections between units and activation of units).

In the ideal compartments approximation, distillation boundaries are treated as linear. However, when the boundaries are highly nonlinear, the resulting design may become infeasible. To account for the curvature of the distillation boundaries, we apply composition corrections using piece-wise linear approximating functions and the multicollinearity property. The piece-wise linear functions are employed to represent the boundary accurately while maintaining the tractability of the model, whereas the multicollinearity property is used to enforce the column material balance after the composition correction. These corrections enable the distillation boundary crossing in our approach allowing an azeotropic mixture to be separated into pure components. Furthermore, we incorporate split fractions for recycling streams in the system which allows significantly better solutions to be obtained. We showcase the applicability of the approach via two examples.

References

[1] K. Kraemer, S. Kossack, and W. Marquardt, “Efficient optimization-based design of distillation processes for homogeneous azeotropic mixtures,” Ind Eng Chem Res, vol. 48, no. 14, pp. 6749–6764, 2009.

[2] S. Brüggermann and W. Marquardt, “Conceptual design of distillation processes for mixtures with distillation boundaries. II. Optimization of recycle policies,” AIChE journal, vol. 57, no. 6, pp. 1540–1556, 2011.

[3] X. Yang, H.-G. Dong, and I. E. Grossmann, “A framework for synthesizing the optimal separation process of azeotropic mixtures,” AIChE journal, vol. 58, no. 5, pp. 1487–1502, 2012.

[4] J. Ryu, L. Kong, A. E. Pastore de Lima, and C. T. Maravelias, “A generalized superstructure-based framework for process synthesis,” Comput Chem Eng, 2020, doi: 10.1016/j.compchemeng.2019.106653.

[5] A. Vogelpohl, “On the relation between ideal and real mixtures in multicomponent distillation,” Chemical Engineering & Technology: Industrial Chemistry–Plant Equipment–Process Engineering–Biotechnology, vol. 25, no. 9, pp. 869–872, 2002.

[6] V. H. Shah and R. Agrawal, “A matrix method for multicomponent distillation sequences,” AIChE journal, vol. 56, no. 7, pp. 1759–1775, 2010.

[7] G. Liu, M. Jobson, R. Smith, and O. M. Wahnschafft, “Shortcut design method for columns separating azeotropic mixtures,” Ind Eng Chem Res, vol. 43, no. 14, pp. 3908–3923, 2004.