Global optimization for integrated solvent and process design

The project is part of the SFB/TR 63 joint research center with partners from TU Dortmund and TU Berlin. From the application side, focus is on the development of novel chemical processes from renewable resources in integrated liquid multiphase systems. The focus of our project has mainly been on hydroformylation of long chain olefines using thermomorphic solvents. Over time, the focus of our project has shifted from optimal process design to integrated process and solvent design, which is investigated in cooperation with the PSE group.

Challenges posed by optimal process design include the global solution of the resulting mixed-integer nonlinear programs (MINLP) in reasonable time using deterministic global optimization (e.g. BARON [R1], SCIP [R2], or ANTIGONE [R3]). For this purpose, model reformulations, bound tightening and surrogate models with iterative refinement have been developed resulting in reductions in the computational effort by orders of magnitude. A good example of the latter is our work on the global optimization of distillation columns [1,2], which have a strong influence on the overall costs of many chemical processes including the present class of novel process concepts.

Fig. 1: Screening approach for integrated solvent and process design.

Fig. 1: Screening approach for integrated solvent and process design.

The motivation for an extended integrated solvent and process design stems from the fact that DMF, the standard solvent for hydroformylation, is toxic and needs to be replaced by ’greener’ solvents. To identify green solvents which meet economic and ecologic requirements in a systematic way, two different approaches have been considered as illustrated in Fig. 1. In the screening approach, suitable solvent candidates are selected from a database using characteristic physical properties as well as environmental, health and safety (EHS) criteria. The essential phase behavior is predicted using molecular simulation and is validated experimentally in a second step. Afterwards, a rigorous economic process optimization is performed to identify the best solvent candidate. In [3], the approach was applied to the hydroformylation process, and a new green solvent was found, which performs in a very similar manner to DMF without being toxic.

Fig. 2: CAMD approach for integrated solvent and process design.

Fig. 2: CAMD approach for integrated solvent and process design.

The second approach in Fig. 2 uses computer-aided molecular design (CAMD). In principal, this leads to an extended MINLP optimization problem, where solvent molecule structure, process structure, and optimal operating conditions are determined simultaneously. The advantage, compared to the screening approach, is that new solvents can also be found which were not already in the screening database. To predict the physical and EHS properties depending on the molecule structure group contribution methods are used. Since a reliable solution of such an extended optimization problem is very challenging, we proposed a hierarchical approach. First, suitable solvent molecules are generated using the CAMD approach [4,5]. To narrow down the search space, the moments of the σ-profiles determined by molecular simulation are used. We found that solvent molecules with σ-profiles similar to some reference components show similar physical and EHS behavior, although the molecule structure can be quite different. Using this approach, four further solvent candidates for the replacement of DMF were identified.

Cooperation partners

Prof. Sundmacher, PSE group
Jun. Prof. Michaels, TU Dortmund
Prof. Engell, TU Dortmund
Prof. Górak, TU Dortmund

Publications

[1] T. Keßler, C. Kunde, Kevin McBride, K. Sundmacher, N. Mertens, D. Michaels, and A. Kienle. Global optimization of distillation columns using explicit and implicit surrogate models. Chem. Eng. Sci., 197:235–245, 2019.
[2] T. Keßler, C. Kunde, N. Mertens, D. Michaels, and A. Kienle. Global optimization of distillation columns using surrogate models. SN Applied Sciences, 1:11, 2019.
[3] T. Keßler, C. Kunde, S. Linke, K. McBride, K. Sundmacher, and A. Kienle. Systematic selection of green solvents and process optimization for the hydroformylation of long-chain olefines. Processes, 7(12):882, 2019.
[4] T. Keßler, C. Kunde, S. Linke, K. McBride, K. Sundmacher, and A. Kienle. Computer aided molecular design of green solvents for the hydroformylation of long-chain olefines. In Computer Aided Chem. Eng., 48:745-750, 2020.
[5] T. Keßler, C. Kunde, S. Linke, K. Sundmacher, and A. Kienle. Integrated molecular and process design of green solvents for the hydroformylation of long-chain olefines, 2021. Preprint. DOI: 10.5281/zenodo.5137004.

[R1] M. Tawarmalani and N. V. Sahinides. A polyhedral branch-and-cut approach to global optimization. Math. Prog., 103:225–249, 2005.
[R2] T. Achterberg. Constraint integer programming, 2007. Ph.D. Thesis Technische Universität München.
[R3] R. Misener and C. A. Floudas. Antigone: Algorithms for coNTinuous / Integer Global Optimization of Nonlinear Equations. J. Glob. Opt., 559:503–526, 2014.