Options
Projekt Titel
Building Block Based Automatic Process Synthesis for Intensified Separation Processes
Förderkennzeichen
SK 291/4-1
Funding code
945.03-1027
Startdatum
September 1, 2023
Enddatum
August 31, 2026
Gepris ID
Loading...
Current energy shortages and the objective to drastically reduce greenhouse gas (GHG) emissions, require innovative solutions in the chemical industry, exploiting the full potential of process intensification (PI). Considering that separation processes are responsible for the majority of the energy requirement of the chemical industry, which requires one third of the energy demand of the manufacturing industry in Germany and even one sixth of the overall energy consumption in the U.S., it is especially important to identify intensified separation processes for individual applications, exploiting the various means for PI developed in recent years and innovative options beyond that level. Yet, existing methods for process synthesis are unable to develop truly novel processes and exclude many PI configurations, either relying on simplified models or restricted search space. The aim of this project is to develop an advanced optimization-based synthesis method for the automatic synthesis of intensified separation processes with rigorous thermodynamic and kinetic models, on the basis of abstract phenomena building blocks (PBB). This is accomplished by a combination of automatic code generation, successive model refinement, and superstructure optimization in order to overcome the common trade-offs between generality, fidelity, and tractability. Synthesis problems are posed as generic state-space superstructures that connect several PBB in the platform-independent meta-language MathML/XML as logic-algebraic equation systems, enabling automatic generation and export of model code for different versions of the sub-layers (generality). Different implementations of the individual PBB allow for a polylithic modeling approach that enables the design of multistage processes based on rigorous thermodynamic and kinetic models (fidelity). Individual optimization problems are solved in a reduced space, making use of external/implicit functions, while further reducing problem size through the exploitation of paradigms from generalized disjunctive programming (tractability), following either a direct logic-based deterministic or hybrid optimization approach. This innovative approach will first be developed and demonstrated for the separation of non-ideal and azeotropic mixtures, by means of an energy-efficient separation process. Thereby, the overall approach will for the first time provide a transferable and automated solution to the general process synthesis problem for relevant separation tasks, which synthesizes highly integrated process designs on the basis of rigorous thermodynamic models matching the accuracy of industrially applied process simulators. This enables the direct application of optimization-based process synthesis to practical problems without requiring in-depth knowledge and experience in formulating and solving optimization problems and extensive model simplification.