imPACts (K-Project)

The primary objective of this K-Project coordinated by RECENDT GmbH
(Coordinator: Robert Holzer) is to investigate and develop new process analytical measurement techniques as well as measurement-based modeling methods for improving the optimization and control of on-line production processes, especially in the chemical and biochemical industry.

As a successor of the K-project “Process Analytical Chemistry – Data-acquisition and Data-processing” (PAC), imPACts is based on the scientific experience and industrial requirements of PAC. It is intended to significantly expand the metholdogies developed in PAC to new forms of more generic applicability and installability at industrial partners. Its main missions can be summarized as follows:

• Knowledge generation about chemical production processes through novel analytics, multivariate data analysis, fluid dynamic simulations and mechanistic modelling.
• Development of advanced optical and MEMS based sensor systems for ultrasensitive, rapid and spatially resolved measurement of chemical and physical sample properties.
• Transfer of methodological approaches to industrial partners enabling for application on different but similar production processes.
   

The whole project is divided into 3 multi-firm and 2 strategic research projects:

• Multifirm-Project 1: Development of novel measurement technologies for process analytics
• Multifirm-Project 2: Characterization and in-depth understanding of multi-phase processes
• Multifirm-Project 3: Monitoring & control of target parameters in industrial processes
• Strategic Project 1: Advanced strategies for information mining and knowledge generation
• Strategic Project 2: Advancing spectroscopy for the analysis and characterization of materials
   

As setting the key researcher for MP3, the role of our department is to provide valuable scientifc inputs for and to actively contribute in the projects MP3 and SP1 in terms of own developed methodologies in the fields of chemometric modeling using soft computing and machine learning methods, dimensionality reduction, auto-diagnostics and re-calibration, fault detection and diagnosis, batch classification and others.
The ultimate goal is to establish a Computational Model Life-Cycle Management (CMLCM) system which can be used on-line by the operators working at the production sites, supporting sufficient flexbility for designing model work-flows, performing re-calibration phases and making time-intensive off-line modeling phases obsolete.