Authors:

Adam Georgopoulos

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Continuous manufacturing of pharmaceutical tablets offers more economic and scalable production with improved quality control over traditional batch processes, but this requires adequate process monitoring techniques. Principal component analysis (PCA), a multivariate statistical method, is more commonly used in process monitoring than the simpler univariate statistical process monitoring approach for its robustness and ability to reduce data on a large amount of process variables to only 2 or 3, allowing easier visualization of relationships between them. However, it has not been widely implemented in a pharmaceutical manufacturing setting. This project investigates the use of principal component analysis (PCA) as a means of process monitoring in a dry granulation process, a key step in tablet manufacturing. Dry granulation process data from 15 experiments investigating ribbon splitting in a roller compactor were used to construct plots of all process variables over time. PCA plots for each experiment were also made, accompanied by diagnostics plots showing details of the underlying variable relationships and the PCA model's goodness of fit. Each experiment required >10 univariate plots to display the data, showing relationships between at most 2 variables at a time. Preliminary PCA plots of each experiment were concise yet captured the variable relationships; a PCA plot of all 15 experiments highlighted discrepancies in process behavior between different experiments, while the univariate method did not. More investigation is being done into how PCA captures differences between startup, steady-state, and shutdown conditions, and the causes behind the distinct regions on the plots of all 15 datasets.

Source:

Purdue University / 2023

Topics:

No topics listed

Co-authors:

Adam Georgopoulos