(637c) Membrane-Assisted Processes for Energy-Efficient Separations: An Optimisation Based Design Approach Using Evolutionary Algorithms

The separation of azeotropic multicomponent mixtures is still a challenging issue in chemical process design. The replacement of conventional, energy-intensive separation methods such as extractive distillation with energy-efficient, membrane-assisted processes can lead to significant energy savings and cleaner processes and thus constitutes an example of how synergies can be used for process intensification. Nevertheless, this signifi­cant economic and ecological potential is hardly exploited in industry due to the lack of a general process design methodology and missing detailed process know-how.

To promote their industrial application, an optimisation based design approach for hybrid processes comprising distillation, pervaporation and vapour permeation is developed and verified for the separation of a ternary azeotropic mixture of acetone, isopropyl alcohol (IPA) and water.

The optimisation approach is based on a generic process model including rate-based models for distillation and membrane separation, which are essential to describe the increased complexity of hybrid processes and strong interactions between the unit operations involved. The parameters for prediction of trans-membrane flux for both membrane separations are determined in lab-scale experiments using different hydrophilic polymeric membranes. The results using different empirical and semi-empirical models for calculation of membrane permeances shows a very good agreement between calculated and measured permeate fluxes.

The optimal design of a membrane-assisted hybrid separation process involves the determination of several structural parameters, e.g. the number of apparatuses and their connection, and corresponding operational parameters, e.g. the flow rates of connecting streams. This complex mixed-integer non-linear design problem is optimised by incorporating the detailed models into a process superstructure, which represents several possible process configurations appropriate for the given separation task. The optimisation problem is solved by applying an evolutionary algorithm to minimize the annualized total production costs as objective function. This approach facilitates the simultaneous determination of the process configuration, dimension of apparatuses and operating conditions required for the optimal hybrid process.

This approach is demonstrated for the determination of the optimal process configuration for superstructures of membrane-assisted processes comprising up to four apparatuses. The results show that the developed methodology is applicable without detailed simulation studies at the early design phase. Finally, the results will be compared to a conventional distillation process. Major achievements concerning the energy efficiency can be reached by integration of membranes into the process.

The support of the “Deutsche Forschungsgesellschaft” for the joint project „Optimisation based framework for mem­brane-assisted hybrid processes” (cooperation with Prof. Marquardt, RWTH Aachen) is gratefully acknowledged.