direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

There is no English translation for this web page.

Methods for Process Design and Operations


Objective of this research area is the development and application of methods for the operation and design of processes. This entails the incorporation of model uncertainty in optimization tasks, the design of frameworks for parameter estimation, data reconciliation, dynamic real-time optimization as well as the decomposition and reformulation of models to improve their convergence behavior.

Contact Persons: Dr.-Ing. Erik Esche, Christian Hoffmann

Process Design


Optimal Design of Process Plants: The synthesis of chemical processes and plants remains a challenge. At d|b|t|a, we are looking at different sequential and simultaneous optimization techniques for solving superstructure optimization problems (MINLP) either based on detailed process simulation in Aspen Plus and CHEMCAD or on reduced order models in software such as AMPL or GAMS.


Optimization under Uncertainty: There are various approaches to include uncertainty information into optimisation problems. At d|b|t|a we focus on chance constraints, which exploit the probabilistic information to compute probabilities of holding certain constraints under uncertainty. We have our own python framework to quickly and reliably compute chance constraints for both dynamic NLP and MINLP problems. The efficient solution of the chance constraints is obtained via an integration over a sparse grid, a high degree of parallelisation, and the efficient reuse of prior solutions.

Process Operation


Optimal Experimental Design and Parameter Estimation: Given our department's extensive experimental expertise, d|b|t|a is also researching advanced techniques for optimal experimental design and parameter estimation. This includes online applications on our own mini-plants and smaller experimental set-ups. State-of-the-art techniques for globalization and regularization (e.g. subset selection) have been implemented and are frequently used in all our experimental projects.


Robust Estimation: To support the analysis of experimental data obtained in our test rigs and mini-plants and to continuously improve their operation, d|b|t|a works on a number of different techniques for data reconciliation and state estimation. This includes the filtering and elimination of gross error and a multi-rate state estimation framework. The latter is able to handle flow, temperature and pressure data in the millisecond range while also processing quality data at an hourly frequency.


(Dynamic) Real-time Optimization: Some of d|b|t|a's numerous mini-plants are continuously operated for up to 300 hours. To improve the performance of these plants, methods are being developed to (re-)optimize online. These methods feature an online connection to each mini-plant, while retaining the basic control structure of the distributed control system, the application of fast surrogate or substitute models for dynamic or steady-state optimisation in real-time, as well as model-adaption based on plant data.

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe