(550d) Methodology for Rapid Development of a Continuous Loss-in-Weight Feeding Process

The pharmaceutical industry has been increasingly shifting towards continuous manufacturing, specifically for the production of tablets. When compared to batch processing, continuous tableting exhibits higher efficiency, flexibility and product quality. The first unit operation in a continuous tableting process is the feeding step, where each raw material is fed through a loss in weight (LIW) feeder at a predefined rate. In order to ensure the quality of the final product, it is crucial to maintain an accurate and stable mass flow, making it essential to understand the behavior of the individual raw materials during the feeding process [1, 2]. Despite the available research correlating material properties with feeding performance, the implementation of this knowledge is not straightforward due to the diversity and inherent variability of raw materials, equipment, external factors, and modeling strategies [3]. Additionally, there is still a lack of understanding of how different setup options and process parameters affect the feeding performance, thus making it challenging to use the models already available.

The present work focuses on creating a methodology for rapid development of a continuous feeding process by creating predictive models that correlate raw material properties with their feeding performance. Additionally, the impact of various setup options and process parameters was assessed, enabling the creation of a workflow to predict the operating space and ideal setup for a new material based solely on material properties.

To obtain a broad range of material properties, 18 raw materials were selected: nine commercially available excipients, two model APIs, and seven processed excipients. The materials were then submitted to various characterization techniques, such as bulk and tapped density, true density, shear cell (yield locus and wall yield locus test), FT4 (stability and variable flow rate, permeability, aeration, and compressibility tests), angle of repose, particle size distribution, dynamic vapor sorption, and Faraday Cup charge measurement. In total, 63 material property variables were obtained, describing density, flowability, cohesion, particle size, hygroscopicity, and electrostatic charge. Each material was tested in a GEA compact feeder using a predefined setup and feeding parameters, to obtain the feeding performance responses (i.e., mass flow RSD, maximum feeding capacity, feed factor decay constant, and dead mass). Multivariate data analysis (MVDA) was used to create predictive models correlating raw materials properties with feeding performance parameters. Due to the considerable number of characterization variables, careful attention was taken during the analysis to avoid model overfitting. To evaluate the impact of altering the setup and feeding parameters on the feeding responses, a design of experiments (DoE) was performed by changing the screw pitch, mesh, agitator, baffle and screw speed.

Results from the MVDA show that feeding performance is mainly affected by variables related to density, cohesion, and flowability. For instance, materials with higher density and lower cohesion have lower mass flow RSD, as well as a higher maximum feeding capacity (FFmax), due to a more uniform screw filling. On the contrary, more cohesive materials led to an earlier decay of the feed factor (FFdecay constant) since they will gradually stick to the screws, reducing the free screw volume over time. Additionally, these materials also tend to stick to the hopper walls, leading to a higher dead mass.

Results from the DoE within the tested ranges show that the mass flow RSD is higher when running at higher screw speeds, as well as when using a mesh or the asymmetric agitator. For more cohesive materials, the asymmetric agitator also leads to an increase in dead mass due to the accumulation of the material on the hopper walls caused by the two vertical shafts. This accumulation can lead to powder falling from the walls over time, causing spikes in the feeder weighing scale and thus affecting the mass flow variability. The FFmax is only affected by the screw pitch, and no impact is observed when using the vertical baffle.

The predictive models created and the conclusions from the DoE were incorporated into a workflow to aid in process development. Since the models have an associated error, this workflow was validated with five external materials by predicting their feeding performance and ideal setup (including refill strategy) using only raw material characterization data. Despite these errors, the selected setup was the same for the experimental and predicted feeding performance. Overall, in order to predict the feeding apparatus and performance, only four characterization techniques are deemed relevant, reducing the analysis time from two days to one hour. As a result, the quantity of material needed to make predictions was also significantly reduced.

Although the implementation of this workflow does not eliminate the need to perform laboratory feeding trials before transitioning to a continuous tableting line, it serves as a fit-for-purpose tool that enables the reduction of raw material and time spent in development. Additionally, it provides deeper process knowledge, enabling the transition from a trial-and-error approach to a data-driven approach.

References:

[1] Wang, Y., Li, T., Muzzio, F. J., & Glasser, B. J. (2017). Predicting feeder performance based on material flow properties. Powder Technology, 308, 135–148. https://doi.org/10.1016/j.powtec.2016.12.010

[2] Van Snick, B., Holman, J., Cunningham, C., Kumar, A., Vercruysse, J., De Beer, T., Remon, J. P., & Vervaet, C. (2017). Continuous direct compression as manufacturing platform for sustained release tablets. International Journal of Pharmaceutics, 519(1–2), 390–407. https://doi.org/10.1016/j.ijpharm.2017.01.010

[3] Hörmann-Kincses, T., Beretta, M., Kruisz, J., Stauffer, F. J., Birk, G., Piccione, P. M., Holman, J., & Khinast, J. G. (2022). Predicting powder feedability: A workflow for assessing the risk of flow stagnation and defining the operating space for different powder-feeder combinations. International Journal of Pharmaceutics, 629, 122364. https://doi.org/10.1016/j.ijpharm.2022.122364