The development and commercialization of new biologics for therapeutic applications is accompanied by the need to scale up the bioprocess to production capacity. This implies adapting the parameters and procedures for a process that is developed at bench-scale (typically 15 mL to 2 L) to industrial-scale volumes (of the order of a few thousand liters) at required product quality and quantity to fulfill patient needs. Bioprocess scale-up and technology transfer can be particularly challenging for bioreactors due to the inherent complexity of bioprocesses combined with the heterogeneity of the environment in larger vessels with multiple hydrodynamic zones, which can cause significant cell-to-cell variability (Alavijeh et al., 2022). As the scale of a bioprocess changes, the geometry and configuration of bioreactor vessels may also change. Even in situations where geometric similarity is maintained (e.g. when using single-use vessels from the same manufacturer), the difference in volume and surface area can increase the influence of fluid turbulence on cells, impacting their growth and productivity. This makes it difficult for scale-down models to mimic the environment that cells will face in large-scale bioreactors. Additionally, there is a large cost incentive to utilize scale-down models for process development, as the cost per batch increases greatly as developmental runs are executed at scale (Farid et al., 2020). Thus, it is imperative that the lab-scale systems that these bioprocesses are developed within are characterized completely and accurately for efficient comparison with large-scale systems.

One method that is widely applied for technology transfer during bioprocess scale-up is the calculation of mathematical parameters that represent one or more aspects of the hydrodynamics of the process being studied. Traditionally, this includes scale-up criteria such as the volumetric mass transfer coefficient (kla), specific power consumption/dissipation, and mixing time for vessels; these parameters are based on vessel similarity and dimensional analysis. The kla describes how quickly oxygen in a system is transferred from the gas phase (typically in the form of bubbles sparged into a bioreactor) to the bulk liquid. This is an important characteristic that describes the ability of the bioreactor vessel to transfer gaseous oxygen to the cell culture and can significantly affect the growth and metabolism of the cells in the culture (Klöckner, et al., 2013). The energy dissipation rate (EDR) and its associated shear rate have the potential to lyse cells in the culture at high values (Bujalski et al., 1987). Mixing time or blend time describes the length of time to homogenize a fluid inside of a vessel. This influences how quickly all cells in the culture will receive adequate nutrients as feed boluses are added to a bioreactor (Anane et al., 2021). These process parameters have important impacts on culture performance and can be very useful for designing bioprocess scale-up, but are conversely challenging and often cumbersome to measure experimentally. This can be due to the absence of standardized procedural guidelines for such characterizations, and the opportunity cost that is associated with the characterization of large-scale vessels. Furthermore, these experimental calculations provide only a global summary of the parameter, and often miss the variabilities of the microenvironments that are present in large-scale vessels (Bylund et al., 1998).

Cell culture performance can be significantly impacted by the scaling method chosen, so understanding typical characteristics of mammalian cell culture is necessary. For mammalian cultures, the cell-specific oxygen uptake rate is usually 3 pmol/cell-day to 10 pmol/cell-day, depending on the biological status of the culture, bioreactor operating conditions, and cultivation mode (Ducommun et al., 2022, Goudar et al., 2011, Wagner et al., 2011). The most widely used scale-up criterion in mammalian cell cultures is constant specific power input to maintain a consistent oxygen transfer rate across scales (Paul and Herwig, 2020, Xing et al., 2009). The agitation rate (RPM) and sparger gas flow rates (vvm, or volume of gas per volume of liquid per minute) may vary with a variety of aeration strategies and vessel geometries explored for mammalian cultures (Xing et al., 2009). In process development of scale-down models, specifically reactors of 200 mL working volume, ranges of air sparging rate (0 vvm to 0.013 vvm), agitation speed (400–1000 RPM), and the corresponding power per unit volume (37 W/m3 to 590 W/m3) have been explored to understand the impact on cell culture performance (Zhang et al., 2019). In most large-scale mammalian cell cultures, the power input range of 20 W/m3 to 80 W/m3 is typically used (Zhang et al., 2019). As cell density increases (and for high density cultures), oxygen demand can increase from approximately 0.001 vvm to 0.1 vvm, though this highly depends on the design of spargers, as well as impeller and vessel geometry, used for cell culture (Xu, et al., 2017).

To mitigate the challenges associated with scale-up and process transfer, researchers have demonstrated that computational fluid dynamics (CFD) is effective for characterizing and optimizing bioprocess scale-up across a variety of process types (Alavijeh et al., 2022, Villiger et al., 2018, March 15, Kuschel et al., 2023, January 18, Scully et al., 2020). CFD entails the mathematical analysis and simulation of bioprocess conditions by solving for the conservation of momentum and mass, typically via the Navier-Stokes equations (Chen, 1992). Particle motion, such as that of bubbles, can be tracked using the Euler-Lagrange approach (Krýsa and Šoóš, 2022). This approach can help users account for the local environment that cells experience both within manufacturing vessels and laboratory-scale vessels (Delvigne et al., 2017). For example, gradients in dissolved oxygen or nutrients across a vessel can be demonstrated with CFD, while these microenvironments are poorly understood through most empirical analyses. Thus, CFD can allow for a more granular and localized comparison between different conditions and bioprocess scales, as well as aid in the optimization of process parameters during scale-up (Amer et al., 2019). Furthermore, in-silico analyses can be performed to simulate multiple conditions, reducing the usage of raw materials and downtime from equipment characterization while providing more in-depth insights into the hydrodynamics of the bioprocess.

In this study, a time-accurate computational fluid dynamics simulation to characterize the Ambr® 250 mammalian vessel was developed. Using a computational tool that employs lattice-Boltzmann-based solving techniques allows for better turbulence modeling, which is valuable for the characterization of the Ambr® 250, as it has fast blend times and other hydrodynamic phenomenon due to its small scale. The Ambr® 250 is a single use 250 mL miniature stirred-tank bioreactor created by Sartorius and is widely used across the biopharmaceutical industry due to its high throughput capabilities and compact design. The Ambr® 250 is also routinely used as a scale-down model during biologics process development. Due to the advantages, standardization, and widespread usage of the Ambr® 250, there is also more publicly available information on mixing conditions that can be used to validate in-silico models, such as the characterization work by Xu et al. (Xu, et al., 2017). The Ambr® 250 exists in multiple bioreactor configurations for mammalian and microbial process types. This study will focus on mammalian cell bioreactors, which contain two pitched-blade (i.e. “elephant ear”) impellers. While other studies have explored the hydrodynamic profile of the Ambr® 250 microbial vessel, which contains two Rushton impellers, a similar hydrodynamic characterization of the Ambr® 250 mammalian vessel is novel (Li et al., 2018). In this work, the use of CFD for the comprehensive characterization of scale-down models within bioprocess development is validated by performing single-phase, species-advection, two-phase, and gas-liquid characterizations of the Ambr® 250 mammalian vessel and by demonstrating comparability with existing experimental studies. Additionally, the analyses developed also assess the effects of operating conditions on cell health. By tracking individual particles through the fluid domain and analyzing their exposure to fluid strain, the accumulated cell damage is quantified, thus providing an approach to inform bioprocess design with optimized parameters that are suitable for minimum cell lysis and maximum cell growth. Finally, the differences between large-scale and small-scale vessels in terms of bulk and surface mass transfer are explored, focusing on the differences in overall oxygen mass transfer that play a key role in accurate process scale-up. The combination of experimental measurements and modeling approaches provides valuable insights into the complex interactions and phenomena occurring in the system, contributing to the understanding and optimization of industrial processes.