Over the past decades, the pharmaceutical industry has witnessed a transformative shift from traditional batch processes to adopting continuous drug manufacturing methods. Twin-screw wet granulation (TSWG) was proposed as a continuous alternative for conventional fluidized bed and high shear granulation processes (Keleb et al., 2004). The design of the first continuous twin-screw wet granulator is derived from the design of a twin-screw extruder. An α-lactose monohydrate formulation was used as the first case for optimizing the modifications. Based on this successful application, TSWG was found to be a reproducible and stable production process. Nevertheless, the number of regulatory filings that include this production pathway is still limited, illustrating the complexity of this innovative manufacturing process (Stauffer et al., 2022, Lee et al., 2015).

Due to the promising nature of TSWG, various vendors have developed and commercialized a continuous TSWG system. One of those commercialized production devices is the ConsiGmaTM-25 (GEA Pharma Systems, Belgium). Typically, four-unit operations can be distinguished in an industrial TSWG line: granulation, drying, milling, and tableting. Although granulation is just one of the sub-processes, most of the research effort has been spent on the characterization of this core unit operation (Fonteyne et al., 2013, Kumar et al., 2014, Kumar et al., 2016, Verstraeten et al., 2017, Nicolaï et al., 2018, Peeters, 2023). Furthermore, researchers have also focused on optimizing the granulation settings (Fonteyne et al., 2015, Meier et al., 2017, Vercruysse et al., 2012) or screw configurations (Li et al., 2014, Vercruysse et al., 2015, Djuric and Kleinebudde, 2008). This comprehensive process knowledge was subsequently used to develop mathematical prediction models that can describe the granulation process in detail (Barrasso et al., 2015, Barrasso and Ramachandran, 2016, Barrera Jiménez et al., 2021, Barrera Jiménez et al., 2023, Van Hauwermeiren et al., 2019). This model type provides the ability to perform model-based process optimizations and control actions, increasing the efficiency of drug product development challenges (Wang et al., 2021).

In contrast to granulation, the second unit operation, the drying step, was investigated much less intensively. Specifically for the ConsiGmaTM-25 system, the drying of the granules is performed using a (semi-)continuous fluidized bed dryer. This dryer uses six sequentially operating cells or compartments. Efficient operation necessitates precise control of the drying cycles, with a critical need to complete the drying and discharging phases in the first cell before subsequent cells are filled, ensuring the uninterrupted flow of wet granules.

The granular drying process ensures that the residual moisture content (RMC) of the granules decreases after they are constituted through the wet granulation process. The physical underlying drying process is called evaporation and induces structural transformations in semi-permanent aggregates. This physical process converts weak liquid bindings into enduring particle interactions through binder solidification (Cheong et al., 2007, Iveson et al., 2001). The latter particle property has already been associated with the granular flowability of the resulting product. The drying process step is essential as the granular RMC plays a crucial role in adjusting the granular flowability (Emery et al., 2009, Crouter and Briens, 2014). For the ConsiGmaTM-25 system, De Leersnyder et al. (2018) gives a comprehensive overview of the risks associated with excessively brief drying times, including potential blockages in the dry transfer line and the formation of smears of broken granules. Furthermore, stated (Ryckaert et al., 2021) that the extent of breakage and attrition caused by the dry transfer line is positively correlated to the RMC. Consequently, to produce high-quality drug products and avoid production interruptions, it is necessary to achieve a high level of process understanding for the drying operation as well (Lee et al., 2015).

Analogous to the approach followed to improve the process understanding of the first unit operation, several experimental investigations on the ConsiGmaTM-25 fluidized bed dryer have been conducted over the last decades (Fonteyne et al., 2014, De Leersnyder et al., 2018, Ryckaert et al., 2021, Ghijs et al., 2021, Ghijs, 2020, Vandeputte et al., 2022). These studies show that the granular RMC decreases disproportionately with the size of the granules to be dried. A positive correlation between granule size and the remaining moisture content is mentioned by Fonteyne et al. (2014). Further elaborating on this last observation, Ghijs et al. (2021) also reported significant variations in moisture content within the same granule size classes after short drying times. Furthermore, this study also concluded that as the total drying time increased, the spread of the moisture content within one size class decreased, or the spread of the moisture distribution became narrower. After long drying, an equilibrium with the inlet drying air conditions and all granules (independent of the size) are formed.

In summary, many variables describing the heterogeneity in the granule bed, such as particle size distribution and the RMC distribution, must be considered during production process optimization assignments. To address this challenge, Vandeputte et al. (2023) proposed a mathematical model to estimate the RMC variation based on process parameters and material properties. The proposed mathematical framework combines several previously published drying model structures of Mezhericher et al. (2007), Peglow et al. (2007), Mortier et al. (2012) and Ghijs et al. (2019). Combining the different model structures allows the model to simultaneously account for physical drying phenomena occurring at a different process level, such as the system or particle levels. In other words, the proposed model framework is a multi-compartmental model. However, despite the high model complexity, some model deficiencies were reported, including under- and overpredictions of evaporation rates.

These typical prediction errors were subsequently linked to the incomplete implementation of typical fluidization and granule segregation phenomena. Vandeputte et al. (2024) investigated both phenomena experimentally to increase the process understanding further. This study then seeks to leverage the insights from Vandeputte et al. (2024) to develop a phenomenological submodel addressing the observed impact of granule-size-based segregation phenomena on the general drying behavior of pharmaceutical granules. Furthermore, this submodel is integrated into the existing structure proposed by Vandeputte et al. (2023), enhancing the fluidized bed drying model’s overall robustness and predictive power.