Continuous manufacturing of pharmaceutical products has the potential to enhance product quality, reduce environmental footprint and improve overall production efficiency. Apart from broadly established quality-by-design principles (QbD) (ICH, 2009, ICH, 2005, ICH, 2008, ICH, 2012), the potential of continuous manufacturing can be profoundly exploited by employing novel quality-by-control (QbC) methodologies (Su et al., 2019). However, most pharmaceutical facilities still rely on traditional approaches, i.e., the production plants usually operate using constant process parameters and only limited process monitoring and control. This operating approach reduces the efficiency of production plants and increases their sensitivity to disturbances, such as equipment failures or raw material variability. Hence, the full potential of continuous manufacturing is not yet exploited. Such drawbacks could be significantly reduced or completely mitigated by adopting novel QbC approaches, e.g., active process- and quality control supported by soft sensors and operator assistant concepts. This research demonstrates the application of QbC concepts and explores their potential on the twin-screw wet granulation unit in the industrial ConsiGmaTM-25 manufacturing line (GEA, 2023).

Twin-screw wet granulation is a size enlargement process critical for suppressing segregation and improving flowability, compressibility, and homogeneity of raw materials (Shirazian et al., 2018). The raw materials (fine particles) interact with the granulation liquid inside the twin-screw granulator and wet granules are formed due to compression and shear forces. The resulting active pharmaceutical ingredient (API) content in wet granules is identified as an intermediate critical quality attribute (CQA). Long-term offsets or extensive variations in the API content, caused by material variability or equipment failures, could negatively affect the final product characteristics, such as tablet content uniformity and assay. Furthermore, variations in liquid content could also negatively affect the performance of subsequent unit operations and final product quality. For instance, wet granules with excessive liquid content could be insufficiently dried, thus negatively impacting tablet properties such as hardness (Zupančič et al., 2023). Implementing real-time process monitoring, along with appropriate control strategies, is essential for preventing undesired occurrences and ensuring production efficiency.

Development of such process monitoring and control strategies has gained significant interest over the last decade (Su et al., 2017, Bhaskar and Singh, 2018, Lakerveld, 2018, Singh et al., 2014, Robinson, 2016, Kruisz et al., 2017, Rehrl et al., 2017, Rehrl et al., 2018, Kirchengast et al., 2019). A multitude of these studies focuses on residence time distribution (RTD) understanding and RTD-based monitoring of the continuous plants. The captured RTD information is employed by quality control concepts to perform corrective actions to remove out-of-specification material from the production stream. However, certain failures unreported in equipment data or the impact of material disturbances, such as segregation at intermediate process points, cannot be captured in this manner. Hence, the RTD-based approaches are not a suitable alternative for inline PAT equipment in terms of advanced process control strategies. The results presented in Kirchengast et al., 2019, Singh et al., 2013, Simonaho et al., 2016 and Rehrl et al. (2016) are promising regarding control-based disturbance mitigation and enhanced process efficiency for a continuous direct compaction (CDC) line. However, prior to designing such control concepts for alternative manufacturing lines, such as the investigated ConsiGmaTM-25, reliable process monitoring solutions must be available. Traditionally, widely-established methods such as near-infrared (NIR) spectroscopy (suitable for CDC lines) exhibit high liquid sensitivity and would not be suitable for monitoring the API content in wet granules. Nicolaï et al. (2018) even exploited this limitation and employed the NIR method for real-time monitoring and control of granule moisture. Hence, the limited studies on control concepts in this field can be attributed to the insufficient availability of reliable process monitoring equipment.

The paper at hand continues a recent study (Celikovic et al., 2023), which introduced a novel process monitoring approach based on Raman spectroscopy. This approach provides a real-time measurement of the granule composition after the twin screw wet granulation unit in the ConsiGmaTM-25 manufacturing line. The present research builds upon this process monitoring solution and proposes a control platform to ensure the production of wet granules with target composition, i.e., with target API- and liquid content. The proposed platform integrates model-based process control for real-time adjustment of process settings and fault-detection-based digital operator assistant offering real-time support and instructions for plant operators. Both algorithms are initially designed in a simulation environment and then transferred, fine-tuned, and validated on the ConsiGmaTM-25 manufacturing line. The present manuscript encompasses comprehensive development steps and can serve as a guide for the implementation of the proposed control platform on alternative production sites.