(61b) Process Synthesis Using Grid Superstructure

Process synthesis recognizes the optimal process flowshseet among many feasible alternatives to minimize/maximize economic, environmental, social or other objectives [1-2]. A process superstructure is often used to represent all plausible process configurations within a large flowsheet. Superstructure optimization, a systematic method to address process synthesis problems, is then employed to find the optimal processing routes [3]. While superstructure-based process synthesis has been very successful in systematic design of chemical processes, a key remaining challenge is that whenever a new problem is addressed, a different superstructure needs to be postulated.

To this end, we have recently proposed a building-block based representation of chemical processes that allows us to dissect classical unit operations into different phenomena, represent these phenomena using fundamental building blocks, and then recombine the building blocks to generate new and intensified unit operations, equipment and process flowsheets [4]. In this work, we generalize the idea towards a comprehensive superstructure for process synthesis, in which all plausible unit operations are represented using finite blocks that are arranged in a two dimensional grid. This grid-based representation of a superstructure facilitates the automated selection of different unit operations as required. We have also formulated a mixed-integer nonlinear optimization (MINLP) problem for process synthesis model. Furthermore, the connectivity among different units is identified via the connectivity between different blocks, and the interaction of blocks is achieved through intra-block material and energy transfer. While reaction, mixing, heating and cooling operations are modeled within the block, separation operations are achieved via the block boundary between adjacent blocks. The systematic arrangement of building blocks results in as many process flowsheets as possible without a priori postulation of their connectivity.

In this presentation, we will also address the model complexity for large grid superstructure which may make it difficult for a global solver to find the optimal solution in reasonable time. To tackle this challenge, we have developed an iterative solution strategy based on the expansion of the grid superstructure to facilitate the solution of the MINLP model. During the initialization step, the initially-set grid superstructure is solved for a feasible solution. This feasible solution is further utilized as an initial guess for an expanded superstructure. Through a range of applications, we will demonstrate that the proposed method is capable of simplifying the procedure for superstructure-based process synthesis, and unifying the mathematical formulation for various synthesis problems.

References:

[1] Chen, Q. and Grossman, I.E., 2017. Recent Developments and Challenges in Optimization-Based Process Synthesis. Annual Review of Chemical and Biomolecular Engineering, 8(1).

[2] Babi, D.K., Holtbruegge, J., Lutze, P., Gorak, A., Woodley, J.M. and Gani, R., 2015. Sustainable process synthesis–intensification. Computers & Chemical Engineering, 81, pp.218-244.

[3] Wu, W., Henao C.A, Maravelias C.T. A superstructure representation, generation, and modeling framework for chemical process synthesis. AIChE Jounal, 2016, 62(9): 3199-3214.

[4] Demirel, S.E., Li, J., Hasan, M.M.F. Systematic Process Intensification using Building Blocks. Computers & Chemical Engineering, http://dx.doi.org/10.1016/j.compchemeng.2017.01.044.