Paul Kubera

Scale-up (and down) modeling of bioreactors is a valuable tool to support process characterization and improvement in a predictable, timely, and cost-effective manner. The emergence of a variety of geometrically dissimilar single-use bioreactor platforms provides users with both challenges and opportunities as they look to incorporate this technology across the range of scale from bench-top through production. In the ideal case, geometrically-scaled, single-use models of existing production equipment would be available to provide accelerated process development over a range of organisms, media formulations, and operating conditions. More often, one is faced with aligning the performance of what may be different hardware designs at the bench, pilot, and production scales. 

	A combination of approaches can be used to reduce risk and ensure predictable performance as processes are transferred across hardware platforms and scale. Simple calculations, model experiments, and simulations complement each other to characterize equipment capability and define operating conditions that will deliver desired results independent of platform and scale.

	A primary objective for cell culture and bacterial processes is to provide a consistent and uniform environment across scales. Calculation of energy dissipation, shear rate, blend-time, and mass transfer (e.g., of oxygen and carbon dioxide) at small and large scale can be used to assess how critical parameters change with scale (because it is impossible to hold all parameters constant) and how these critical parameters can be managed by changing other design variables (e.g., hardware configuration). Spreadsheet-based analysis readily extends the range of options that can be examined to achieve an optimal solution.

	Performance aspects such as oxygen transfer can be highly dependent upon the configuration of what is often a proprietary hardware design. Small-scale model tests can be used to generate baseline data when at-scale performance correlations are not available. These test results can be used to generate hardware-specific correlations and to evaluate the relative performance of different approaches. Different single-use platforms will have different strengths-having relative information available allows one to best match performance to process requirements.

	While a mixing-dependent parameter (e.g., oxygen transfer) may be relatively independent of scale, blend-time (i.e., the time to achieve batch uniformity) is sensitive to both hardware selection and scale. Most bioreactor scale-up is based on hardware geometric similarity and equal energy dissipation to drive equal volumetric mass transfer; this approach is straightforward but it guarantees that blend-time will increase as the batch volume increases. For processes where uniformity is a critical parameter, model-scale tests can be used together with computational simulation to quantify and manage blend-time. 

	Simple small-scale blend-time tests can be conducted using either visual or instrumentation-based techniques to establish performance correlations; multiple hardware configurations can be assessed in a short period of time to evaluate both configuration and operating condition options. Computational fluid dynamics (CFD) can be employed to model the flow pattern and simulate blend-time for preferred configurations. Presuming that there is good agreement between the small-scale CFD simulation and experimental results, simulation can then be used with confidence to model expected performance at full-scale.

	Depending upon need, one or more of the above tools can be employed to address scale-up and hardware platform transfer concerns. As an example, in a case where three different single-use platforms, all at different scales, were being evaluated for large-scale production, at-scale characterization experiments were run to establish performance correlations, and CFD simulation (validated at small-scale) and desktop calculations were used to project expected performance at large scale.  

	Case study: comparing a small-scale single-use bioreactor with a large-scale stainless-steel bioreactor

	As another example, a multi-pronged approach can also be used to (i) demonstrate equivalent performance between traditional alloy and proposed single-use equipment, (ii) validate CFD simulation, and (iii) extend performance projections to intermediate and full-scale equipment. In this case, a single-use bioreactor with hydrofoil impellers (ABEC Custom Single Run [CSR]) was proposed as a development platform for an existing stainless-steel production bioreactor that was more than 10 times the size of the CSR and was equipped with a proprietary agitator design. 

	Laboratory tests at 50% of the development-scale were conducted using two different impeller geometries to compare blend-time for the single-use configuration with the stainless-steel alloy configuration. The laboratory data were also used to validate the results of CFD blend-time simulation of the lab-scale CSR single-use geometry (see 

), allowing use of CFD to project performance of the proposed development-scale (

) platforms. The velocity contours in the three plots shown in 

 illustrate the near identical flow structures across the range of scale for the alloy impeller and its single-use counterpart. 

	Key bioreactor performance indicators (e.g., blend-time, oxygen mass transfer coefficient, and oxygen mass transfer rate [OTR]) can be measured in the laboratory and the resulting correlations used to predict performance at development and full-scale. A spreadsheet application (see 

) allows for ease of comparison across scale and platform type for these parameters and others that may impact cell growth and product yield, such as energy dissipation, shear rate, and others. 

	Equal power per volume (P/V) is a commonly applied approach for bioreactor scale-up because it provides equal volumetric oxygen transfer and thus equivalent cell growth/mass. When equal P/V is used together with geometric similarity, 

 illustrates that most other parameters will change; the key is to establish which ones may have a positive or negative impact on product yield and then manage them accordingly. For example, if experience indicates an improved product yield at lower shear rate or shorter blend time, one can change impeller selection and operating speed with scale to achieve the desired result while maintaining P/V and OTR. Because the mass transfer coefficient is a function of both agitator energy input and gas velocity, this side-by-side comparison shows that air sparge rate must be increased as scale decreases to achieve the equal OTR.

	Oxygen transfer was identified as the controlling element for the production bioreactor in this example; other parameters (e.g., shear rate, blend-time) were reviewed to ensure that they did not exhibit order-of-magnitude change with scale-up. The volumetric OTR for the alloy production unit was calculated using correlations for similar-style equipment; operating set-points for agitator speed and air sparge rate to achieve the same OTR were proposed for the development-scale single-use bioreactor based on measured performance. Plots of cell density and product titer versus time for a typical single-use development-scale run are presented in 

, along with similar information for runs in the production-scale bioreactor.

	Cell mass and product titer for the development-scale single-use bioreactor track closely with historic values for the alloy production-scale unit. The ability to effectively characterize and compare the performance of the different platforms and scales before initiating a development or production campaign serves to reduce at-scale experimentation, saving both cost and time. 

	When referring to this article, please cite it as P. Kubera, " Testing and Simulation Approaches for Single-Use Bioreactor Scale-up," 

	Paul Kubera is vice-president, Process Technology, at ABEC.

Articles

Tools aid scale-up and comparison of single-use and stainless-steel bioreactors.