456350 Optimal Reactor-Separator-Network Synthesis

Monday, November 14, 2016
Grand Ballroom B (Hilton San Francisco Union Square)
Nicolas Maximilian Kaiser, Otto-von-Guericke-University, Magdeburg, Germany and Kai Sundmacher, Max Planck Institute for Dynamics of Complex Technical Systems, Magdeburg, Germany; Process Systems Engineering, Otto-von-Guericke University Magdeburg, Magdeburg, Germany

Main challenges in process design is finding the optimal pathway to form a specified product in the most economical way and building a process which is able to approximate this theoretical pathway best. For latter one can basically choose between two main operating modes - discontinuous and continuous processes. First is often associated with smaller scale productions like for pharmaceuticals and fine chemicals and latter for larger scale productions of e.g. bulk chemicals. Due to that the design approaches for both kinds of processes are often distinguished as well. However, it would be more promising to figure out the potential of different process pathways or realizations without restricting it in advance to either discontinuous (batch) or continuous process types. The concept of elementary process functions (EPF) allows in particular for that, the analysis of the maximum potential of a process without preliminary restrictions to the optimal process [1]. Core of the EPF methodology is the optimal manipulation of mass and energy fluxes of matter elements during their travel through the process. These fluxes are determined by means of dynamic optimization calculations in an analogous batch process.

Within this scope it is intended to develop an approach to find an optimal design for any arbitrary process by means of the EPF methodology, which corresponds to optimizing it within a batch process optimization with optimal dosing of all attending components. Thereby any characteristics and beneficial control of the reaction progress can be identified, e.g. advantages of mixing, back-mixing, extraction, heating, cooling, etc. The resulting time-dependent set of fluxes is directly related to aforementioned functions and/or the corresponding unit operations, or at least allow for concluding realization options. Analyzing these fluxes provides basic information for optimal discontinuous and continuous realization of the process: A discontinuous process can be realized by optimal time control of a single batch reactor or as optimal control or scheduling of several interconnected batch reactors and separator steps; a continuous process can be developed by translation of the fluxes into an optimal network of continuous reactor and separator steps, whereby also the back-mixing characteristics of the reactor steps and the benefits of recycles can be revealed. For the realization of those optimization results in a batch process several interesting work has already been published, see e.g. [2]. Within this work the focus lies on using those optimization results for developing an optimal continuous reactor-separator-network.

The translation of the fluxes into an optimal reactor-separator-network results in a superstructure of aforementioned continuous reactor types, dosing/mixing spots and extraction spots together with the corresponding cooling/heating strategies for all units. The dosing/mixing spots can thereby be also realized by closed recycle loops. Since this translation only provides a qualitative structure of the process, this superstructure has to be optimized then on its own to obtain a quantitative design of the process. This can be done for different target compositions resulting in an area in the concentration space in which for each point an optimal reactor-separator-network is associated. This is closely connected to the idea of determining the Attainable Region (AR) of a process and then to find the optimal operation point in this region depending on the overall process, the specified process requirements and the objective of the design. But the AR approach is limited to the combination of only reaction and mixing in continuous reactors, i.e. PFR, CSTR and DSR, and mixing units, respectively [3].

Within this work the proposed reactor-separator-network synthesis method is exemplified on two processes. First on an extended van-de-Vusse reaction, which is often taken as example process in literature, as well for the Attainable Region approach. Therefore it allows for a direct comparison of our method to the state-of-the-art Attainable Region concept and other network synthesis approaches. The second example is the hydroformylation of long chain olefins in thermomorphic solvent systems [4]. The hydroformylation is a very important homogeneously catalyzed process in chemical industry. For a profitable, highly selective hydroformylation of long chain olefins a Rhodium/Biphephos catalyst complex is used which is supposed to be recovered by an innovative solvent system, whose separation characteristics can be controlled by temperature.


[1] Freund, Hannsjörg; Sundmacher, Kai (2008): Towards a methodology for the systematic analysis and design of efficient chemical processes. In Chemical Engineering and Processing: Process Intensification 47 (12).

[2] Peschel, Andreas; Hentschel, Benjamin; Freund, Hannsjörg; Sundmacher, Kai (2012): Design of optimal multiphase reactors exemplified on the hydroformylation of long chain alkenes. In Chemical Engineering Journal188, pp. 126–141.

[3] Glasser, David; Crowe, Cameron; Hildebrandt, Diane (1987): A geometric approach to steady flow reactors. The attainable region and optimization in concentration space. In Ind. Eng. Chem. Res. 26 (9).

[4] Schäfer, Elisabeth; Brunsch, Yvonne; Sadowski, Gabriele; Behr, Arno (2012): Hydroformylation of 1-Dodecene in the Thermomorphic Solvent System Dimethylformamide/Decane. Phase Behavior–Reaction Performance–Catalyst Recycling. In Ind. Eng. Chem. Res. 51 (31), pp. 10296–10306.

Extended Abstract: File Not Uploaded
See more of this Session: Interactive Session: Systems and Process Design
See more of this Group/Topical: Computing and Systems Technology Division