Projects per year
Abstract
Process Systems Engineering (PSE) is a discipline which connects a wide range of chemical engineering topics in a systems view approach. The reason for this systematic view of this scientific field is the need of computational concepts, numerical methods and computer-aided tools which can be applied to different use cases in industry. The multi-scale framework developed in this work encompasses four levels of application: (I) A server based property prediction software prototype which quantifies the uncertainty of group contribution methods and quantitative structure property relationship models and provides the confidence bounds of the estimates. (II) A modelling development level which allows the user to develop models
in a flexible way by using common programming languages for fast prototyping (Python) and high performance computing (Fortran). (III) An interface to process simulators to analyse and optimise entire flowsheets with advanced routines. (IV) A superstructure optimisation layer where surrogate models generated from unit operations or process models can be embedded in a superstructure formulation and solved for the optimal process structure and operating point. The contributions presented in this work show how the developed framework allows to tackle research in machine learning, optimisation and Monte Carlo driven methods such as sensitivity analysis. The developed tools were applied to the oleochemical domain with selected processes. In conclusion, this work demonstrates that a modular approach to process systems engineering, combined with tools integration from various vendors, allows to gain new knowledge in a time-efficient and augmentable manner.
in a flexible way by using common programming languages for fast prototyping (Python) and high performance computing (Fortran). (III) An interface to process simulators to analyse and optimise entire flowsheets with advanced routines. (IV) A superstructure optimisation layer where surrogate models generated from unit operations or process models can be embedded in a superstructure formulation and solved for the optimal process structure and operating point. The contributions presented in this work show how the developed framework allows to tackle research in machine learning, optimisation and Monte Carlo driven methods such as sensitivity analysis. The developed tools were applied to the oleochemical domain with selected processes. In conclusion, this work demonstrates that a modular approach to process systems engineering, combined with tools integration from various vendors, allows to gain new knowledge in a time-efficient and augmentable manner.
Original language | English |
---|
Place of Publication | Kgs. Lyngby |
---|---|
Publisher | Technical University of Denmark |
Number of pages | 150 |
Publication status | Published - 2019 |
Fingerprint
Dive into the research topics of 'Design and Optimisation of Oleochemical Processes'. Together they form a unique fingerprint.Projects
- 1 Finished
-
Design and optimization of selected oleochemical processes
Jones, M. N. (PhD Student), Sin, G. (Main Supervisor), Gernaey, K. V. (Supervisor), Sarup, B. (Supervisor), von Solms, N. (Examiner), Pantelides, C. C. (Examiner) & Camarasa, A. E. (Examiner)
15/04/2016 → 19/08/2019
Project: PhD