(460g) Generalized Superstructure-Based Distillation Network Synthesis

Due to a vital role of distillation in chemical facilities and its significant energy use, multiple methods have been proposed to synthesize an optimal distillation network. Superstructure-based optimization methods, where a network that embeds all potentially useful elements of the system and their interconnections is optimized, have received a lot of attention¹. Most existing superstructure-based methods for distillation network synthesis are developed to be used within sequential approaches², where the distillation network synthesis problem is solved separately, after the reactor network is synthesized. Henceforth, it is assumed that information from the reactor network, such as the set of components to be separated and their flow rates, is given. Also, only a single stream to be separated is typically considered, and target outlet streams of the distillation network (e.g., final products or recycle streams into the reactors) are typically given as almost pure components.

However, when we consider the synthesis of an integrated reactor-distillation network, the feed information can vary depending on decisions related to the reaction network; for instance, catalyst/reaction selection can change the components in the effluent of a reactor, potentially leading to different components to be separated in the distillation network. Furthermore, the number of streams to be separated and the number/compositions of outlet streams of the distillation network can also vary depending on decisions in the reactor system^3,4. Considering these additional degrees of freedom and the interactions between different subsystems can lead to superior solutions for the synthesis of the integrated system. Unfortunately, these solutions cannot be found using existing methods.

To address this challenge, we propose a generalized superstructure-based distillation network synthesis model (non-convex Mixed-Integer-Non-Linear-Programming model). The key features of the proposed model are the following: (1) it allows multiple streams to be separated, and components present in the streams can vary; (2) thermal coupling between columns^5,6 and stream bypasses are simultaneously considered, resulting in a significant reduction in total cost; and (3) it allows compositions of outlet streams to vary. The first feature is enabled by adopting flexible distillation column models⁷ which can detect components in the stream and calculate the corresponding minimum vapor flow rates in the column. The second feature extends the search space, so superior solutions, which cannot be found with the existing methods, can be found. The third feature enables the optimization of recycle streams to the reactor subsystem, facilitating a better integration with reactor network synthesis. Due to the unique features, the proposed model can not only be seamlessly integrated with reactor network synthesis models but also incorporate additional configurations in the superstructure, leading to superior solutions.

We present a number of examples to illustrate the applicability of the proposed model. First, we show a simple reactor-distillation network synthesis without recycle streams to highlight the first feature of the model. Then, we consider systems with recycle streams to show how reactor and distillation networks can benefit from the optimization of the compositions of the recycle streams (and the corresponding distillation network configuration). Notably, interactions between stream bypasses and thermal couplings result in interesting solutions. Finally, we show an example with multiple streams to be separated, showing the advantages of considering multiple feed streams.

References

1. Chen Q, Grossmann IE. Recent developments and challenges in optimization-based process synthesis. Annu Rev Chem Biomol Eng. 2017;8:249-283.
2. Douglas JM. A hierarchical decision procedure for process synthesis. AIChE J. 1985;31(3):353-362.
3. Kong L, Maravelias CT. Expanding the scope of distillation network synthesis using superstructure-based methods. Comput Chem Eng. 2020;133:106650.
4. Ryu J, Kong L, de Lima AEP, Maravelias CT. A generalized superstructure-based framework for process synthesis. Comput Chem Eng. 2020;133:106653.
5. Halvorsen IJ, Skogestad S. Minimum energy consumption in multicomponent distillation. 2. Three-product Petlyuk arrangements. Ind Eng Chem Res. 2003;42(3):605-615.
6. Fidkowski ZT, Agrawal R. Multicomponent thermally coupled systems of distillation columns at minimum reflux. AIChE J. 2001;47(12):2713-2724.
7. Ryu J, Maravelias CT. Efficient generalized shortcut distillation model with improved accuracy for superstructure‐based process synthesis. AIChE J. 2020;66(11), e16994.