The thesis discusses the tool qualification procedure on the automated Uhlmann 1660
packaging line, which is used in pharmaceutical production for packaging tablets into blisters
and subsequently packing the blisters into cartons. Tool qualification is extremely important,
as it ensures high-quality and uninterrupted production in accordance with Good Manufacturing Practice standards.
The thesis first presents the packaging line, which is divided into a primary and a secondary
section. The primary section of the line is intended for forming and filling blisters, while the
secondary section is used for packing blisters into cartons together with patient information
leaflets. The thesis then describes the tool qualification procedure itself, which consists of
several consecutive tests that are also described in more detail.
The central part of the thesis is dedicated to improving the determination of MSR station values. The MSR parameter defines the distance between the location of a defective blister and the rejection station, and is crucial for the correct rejection of nonconforming blisters. The aim of the thesis was to develop a model that calculates MSR values for each station for new products with previously unknown dimensions, based on known data from previously performed qualifications and the dimensions of the new blister.
Two models were developed to predict MSR values. The first is based on the rule-of-three
calculation model, where one reference product is used. The second is based on modelling with several reference products and selecting the appropriate polynomial model using cross
validation. The results showed that both approaches follow the actual measurements well, with each model having certain advantages. On average, the rule-of-three model achieved a lower absolute error, while the cross-validation model was less sensitive to larger individual
deviations.
The developed models will help engineers qualify tools for new product dimensions more
quickly, as they will reduce the time required to find suitable settings and decrease the number of trial runs. However, they cannot completely replace practical verification on the line, since the final MSR station values must still be confirmed through the actual operation of the machine and the execution of the prescribed tests.
|