Qualitätsorientierte Automatisierung für den kontinuierlichen Betrieb pharmazeutischer Tablettenproduktionsanlagen

Aachen (2018, 2019) [Dissertation / PhD Thesis]

Page(s): 1 Online-Ressource (xiii, 136 Seiten) : Illustrationen, Diagramme


The pharmaceutical industry is currently facing profound changes in the production concerning its technology as well as the processes of quality assurance. The production conditions of the status quo are characterized by high regulatory efforts and low flexibility of the process control. The production is performed in the conventional batchmode, which is predominantly operated experience-based and thus with a low degree of automation, requiring a high level of manual adjustment and testing. In contrast to this, today’s efforts focus on the continuous production. Additionally, manufacturers are encouraged to systematically gather and make use of process understanding as early as during the development of novel production technologies, but also for the realization of new processes of quality assurance. In contrast to the conventional production with fixed operating points and manual testing, critical quality characteristics are now to be guaranteed with the proper design of processes and operating strategies, which are to be applied knowledge-based with the help of suitable automation solutions. For the first time, manufacturers of continuous production technology are facing the task to consider effects of the coupling of machines in the process automation, as from now on the sub-processes operate simultaneously in cooperation and not as individual units any more. This thesis is conceptualized as a pioneering contribution for the establishment of such automation solutions for the pharmaceutical industry. Therefore, a major aim is the exemplary application and practical demonstration of tools and solution strategies for the development of automation algorithms. For this purpose, the thesis focuses firstly on the development of an automation solution for basic tasks concerning the process control of a demonstration plant. The object of study features a modular concept that can connect machines of different manufacturers with individual process automations in order to realize different manufacturing processes. Using the example of the process of direct compaction of pharmaceutical tablets, the whole toolchain consisting of modeling, identification and validation with measurement data and simulation-based development and testing of observer and controller algorithms is applied. For this purpose, the description of material streams is focused first, as this is the basis for the synchronisation of the machines during operation which is realized by means of a model predictive controller. Furthermore it is shown which tasks can be performed by the automation system additionally, if product quality is to be taken into account as a target variable. A major precondition for this is process understanding, which is required in the form of mathematical models that describe the correlation of material properties, process control and the quality attributes of the product. A procedure is presented for the analysis of such models in order to identify operation regions that guarantee compliance with quality specifications. Subsequently, the presented methodology for the model-based synchronisation of sub-processes of a continuous production plant is adjusted such that modelled correlations with quality attributes can be taken into account to enable optimal control of the plant concerning quality-related objectives. The thesis concludes with a demonstration of the practical application of the presented methodologies for the production process of direct compaction. The main contribution of this thesis is the practically tested optimal control strategy for the synchronisation of the machines of a continuous production plant in combination with the methodology for the sensitivity analysis of the required models, which can be used to optimize safe operation regions of critical input variables (concerning quality attributes) and for the optimal choice of nominal operation points of the machines.



Groß-Weege, Christopher Andreas


Abel, Dirk
Horn, Martin


  • REPORT NUMBER: RWTH-2019-00308