Flowsheet modelling of a powder continuous feeder-mixer system

Abstract

In this study, an integrated flowsheet model of the continuous feeder-mixer system was calibrated, simulated and compared against experimental data. The feeding process was first investigated using two major components (ibuprofen and microcrystalline cellulose (MCC)), in a formulation comprised of: 30 wt% of ibuprofen, 67.5 wt% MCC, 2 wt% of sodium starch glycolate and 0.5 wt% of magnesium stearate. The impact of a refill on feeder performance was experimentally evaluated for different operating conditions. Results showed that it had no influence on feeder performance. While simulations with the feeder model fairly reproduced the material behaviour observed in the feeder, unintended disturbances were underpredicted due to the model’s low complexity. Experimentally, mixer’s efficiency was assessed based on ibuprofen residence time distribution.

Mean residence time pointed to a higher mixer’s efficiency at lower flow rates. Blend homogeneity results showed that for the entire set of experiments, ibuprofen RSD < 5%, irrespective of process variables. A feeder-mixer flowsheet model was calibrated, after regressing the axial model coefficients. The regression curves exhibited a R2 above 0.96, whereas the RMSE varied from 1.58×10-4 to 1.06×10-3 s−1 across all fitted curves. Simulations confirmed that flowsheet model captured the powder dynamics inside the mixer and qualitatively predicted the mixer’s filtering ability against feeding composition fluctuations, as well as ibuprofen RSD in blend, in line with real experiments.

Introduction

The pharmaceutical industry is on the verge of a technologic revolution. An increasing demand to bring in innovative medicines in a fierce competitive market while coping with emergencies (e.g. pandemics) (Gernaey et al., 2012), has spurred it up to adopt advanced manufacturing technologies, such as continuous manufacturing (CM). In brief, CM pertains to a mode of process manufacturing, whereby raw materials and intermediate/final products are continuously charged into and discharged from each unit operation in the manufacturing process train, during a specified amount of time (Lee et al., 2015). Recently, the ICH guideline Q13 on CM has brought companies’ uncertainties to an end, while providing in-depth regulatory considerations over CM implementation (ICH, 2021). The advantages of CM have long since been acknowledged by the pharmaceutical industry: manufacturing flexibility, shorten and secure supply chain, reduction in production costs and footprint, process streamlining and robustness (Boukouvala et al., 2012, Byrn et al., 2015, Lee et al., 2015, Moghtadernejad et al., 2018, Vanhoorne and Vervaet, 2020).

CM hinges upon an enhanced understanding of how material’s attributes, process parameters and operating conditions influence the critical quality attributes of the drug product (Boukouvala et al., 2012). Such process knowledge can be gained in process modelling, by means of mathematical equations, able to accumulate vast amounts of information collected over time and experimentation (Djuris and Djuric, 2017).

In this way, flowsheet modelling comes up as a modelling framework to expedite process understanding. Indeed, flowsheet modelling is a major breakthrough in computer aided simulation tools permitting the design, simulation, analysis and optimization of an integrated manufacturing process (Boukouvala et al., 2012, Dosta et al., 2020). That is, in a real powder-based CM process, each unit operation is connected to the next by suitable interfaces to accomplish an uninterrupted powder-to-tablet process. Likewise, in mathematical terms, the process integration resides in linking one unit’s operation model to the next, whereby the output of a preceding model becomes the input of a subsequent one. Thus, the “flow of variables” carried over by unit models imitates the flow of materials across unit operations (Wang et al., 2017).

A developed flowsheet model is capable of simulating specific instances of the process (e.g. recycle streams, process start-up shut-down), as well as different process scenarios leading to the selection of the best design configuration and operating conditions that comply with product’s quality criteria (Boukouvala et al., 2012). Moreover, as the flowsheet model is able to evaluate integrated processes in silico, less experiments are required to validate the models (Kleinebudde et al., 2017, Rogers et al., 2013).

In the last decade, flowsheet modelling has been applied to perform dynamic simulations of a direct compression process, to design and evaluate control strategies (Tian et al., 2021, Tian et al., 2019) and to establish a quality risk assessment strategy (Boukouvala et al., 2012, Rogers et al., 2014, Wang et al., 2017). Design and simulation of a flowsheet model has also been conducted in dry–granulation (Park et al., 2018, Singh et al., 2012) and wet-granulation manufacturing routes (Metta et al., 2019). The potential to combine a DEM–PBM models in an integrated flowsheet model for downstream processing has been successfully deployed by Sen et al., 2013. However, flowsheet modelling application in powder-based CM processes is still limited (Ierapetritou and Muzzio, 2016) and validation of a flowsheet model against experimental data is solely covered in a handful of articles (Galbraith et al., 2020, Galbraith et al., 2019, Park et al., 2018, Rogers and Ierapetritou, 2015, Tian et al., 2019).

This work focus on solid oral dosage forms (e.g. tablets), since they are by far the most common drug products in the market (∼80%) (Rogers et al., 2014a). Within a continuous tablet manufacturing line, the feeder-mixer system is crucial to deliver a uniform composition blend down to the processing line (Engisch and Muzzio, 2012, Karttunen et al., 2020). Thus, the integrated feeding-mixing process must be examined as whole in terms of its ability to successfully reduce inflow composition variation (Gao et al., 2011) in order to ensure high-quality tablets are consistently delivered under the Quality-by-Design concept (Djuris and Djuric, 2017).

Hence, the present study looks at the integrated feeder-mixer system in a continuous manufacturing environment and sets out to compare an integrated flowsheet model thereof against experimental data. To this end, firstly loss-in-weight feeders are experimentally characterised and individually simulated; secondly the feeder-mixer system is experimentally studied towards mixing efficiency and blend homogeneity; thirdly, feeder-mixer system model parameters are regressed to fit experimental data and lastly a calibrated flowsheet model is employed with a view to simulating the feeder-mixer system and predicting blend homogeneity.