(173d) Pyosyn Graph: New Representation and Systematic Generation

Recent developments have brought new opportunities as well as new challenges for the chemical process industry, including availability of inexpensive feedstocks from the shale gas revolution in the U.S., rising awareness of environmental impacts, evolving regulatory landscapes, and renewed volatility in market conditions (1). The industry must decide how to best adapt its business practices to most effectively provide society with the fuels, polymers, pharmaceuticals, and other chemical goods necessary to sustain increasing standards of life across the world. Integral to these analyses are the process flowsheet designs themselves.

Superstructure optimization (2) aims to provide a systematic search over the entire relevant design space, by postulating all relevant permutations of the process alternatives and their interconnections (captured as a superstructure), then formulating a mathematical programming model and solving it to identify the optimal design. The use of mathematical programming confers one of the main advantages of superstructure-based synthesis—a mathematical optimality guarantee of the maximum gap with respect to the user-specified objective function, between a feasible flowsheet and the best possible design embedded in the superstructure. This approach relies on the success of three main steps:

1. Generation of a superstructure embedding the relevant flowsheet alternatives
2. Formulation of a tractable mathematical programming model
3. Solution with a mathematical programming code to obtain optimal flowsheet design

The first step is generation of a superstructure representation that embeds the optimal flowsheet. The optimal design must be a subgraph of the superstructure to be successfully identified (3, 4). The choice of representation can also have an impact on the tractability of the later solution step (5).

This work introduces a new representation and superstructure generation approaches for Pyosyn, an open-source framework for systematic superstructure-based process synthesis. The new Pyosyn Graph (PSG) representation consists of units, ports, and streams, and includes support for nested units, including new “single-choice” units and modular superstructure construction. We introduce superstructure generation strategies based on both library-assisted and direct-hierarchical means-ends analysis. For the library-assisted approach, we describe generalized port annotations that describe conditions for compatibility between connected unit ports. We extend literature methods (6) to present seven screening rules based on new material port annotations that categorize process chemical species as primary, secondary, or residual. In doing so, we improve inherent process safety by distinguishing species that are residual from those that are infeasible due to material incompatibility or safety concerns. We demonstrate the ability of the new screening rules to reduce superstructure complexity without loss of generality, while requiring fewer assumptions.

We also discuss implications of the PSG representation on modeling and solution strategies in Pyosyn. Using a bipartite graph view of the PSG, we illustrate common modeling simplifications that are possible for process synthesis, and demonstrate their implementation in Pyomo.Network.

We finally demonstrate the use of Pyosyn through a set of diverse case studies, including methane to syngas, syngas to methanol, acid gas scrubbing, and Kaibel column design.

References

1. Grossmann IE, Harjunkoski I. 2019. Process systems Engineering: Academic and industrial perspectives. Comput. Chem. Eng. 126:474–84
2. Mencarelli L, Chen Q, Pagot A, Grossmann IE. 2020. A review on superstructure optimization approaches in process system engineering. Comput. Chem. Eng., p. 106808
3. Westerberg AW. 2004. A retrospective on design and process synthesis. Comput. Chem. Eng. 28(4):447–58
4. Agrawal R. 1996. Synthesis of Distillation Column Configurations for a Multicomponent Separation. Ind. Eng. Chem. Res. 35(4):1059–71
5. Yeomans H, Grossmann IE. 1999. A systematic modeling framework of superstructure optimization in process synthesis. Comput. Chem. Eng. 23(6):709–31
6. Wu W, Henao CA, Maravelias CT. 2016. A superstructure representation, generation, and modeling framework for chemical process synthesis. AIChE J. 62(9):3199–3214