Options
A novel approach to hybrid evolutionary-deterministic optimization in process design
Publikationstyp
Conference Paper
Date Issued
2013-06
Sprache
English
TORE-URI
First published in
Number in series
32
Start Page
961
End Page
966
Citation
Computer Aided Chemical Engineering 32: 961-966 (2013)
Contribution to Conference
Publisher DOI
Scopus ID
Publisher
Elsevier
ISBN
978-0-444-63234-0
Optimization-based process design can be accomplished by the formulation of superstructures and the use of metaheuristics as well as deterministic optimization. This paper proposes a novel hybrid optimization approach, which combines an evolutionary algorithm (EA) and a sophisticated deterministic optimization strategy. In contrast to related approaches, the EA provides an initial superstructure, which results in a MINLP and is solved by a local deterministic algorithm. This combination facilitates an extensive inspection of the search space, while the sophisticated deterministic optimization leads to a reduction in the number of individuals that need to be evaluated within the evolutionary approach. The application of the novel hybrid optimization approach is illustrated by a case study, i.e., the separation of an ethanol/water mixture by means of an entrainer-enhanced pressure swing distillation process.
Subjects
Evolutionary algorithm
Hybrid optimization
MINLP
Process design
DDC Class
600: Technology