How do you run thousands of virtual experiments a week?
Many process development challenges require extensive lab work to solve. But lab space is tight, and labs often do not run as productively as they could because trial and error experiments are taking up time and space.
Virtual experiments, especially those powered by computational fluid dynamics (CFD) are zeroing in on the solutions to process development challenges, meaning lab-based experimental programs are getting shorter and faster.
Small and mid-sized companies, that lack large sale process development infrastructure, can now run experiments that make actual lab work more productive and deliver processes to CMOs that run right first time.
Process development lab space and expertise is at a premium for small-to-medium biotech and pharmaceutical companies. This means that they must get smarter and more productive about the lab work they are doing in internally or at their external partners. Many process development approaches such as Design of Experiments (DoE) is inherently based on trial and error, a key component of the scientific method, but an approach that leads to inefficiencies when poorly designed. These challenges negatively impact speed to the clinic either through delays bringing processes into production or through issues arriving after introduction that slow things down.
With this challenge in mind, APC scientists developed a CFD platform that can run virtual experiments on real or imagined equipment sets allowing small-to-medium size companies to zero in on real solutions to process development challenges before going into the lab. This CFD platform comes together across 4 key steps which build vritual equipment sets and run simulations that describe how heat and mass transport is occurring in the system.
Step 1: Create a 3D computational domain of the unit operation. By modeling the exact unit operation and configuration there is no reliance on correlations for a similar process or best-guess.
Step 2: Divide geometry into minute volumes or mesh. These volumes must be refined to capture the flow features in areas of rapid change (i.e. high velocities and shear around an impeller).
Step 3: Solve transport equations for each of the minute volumes such as the Navier-Stokes equations. Can model how the process evolves over time.
Step 4: Analyze the results to inform further simulations or recommended at-scale operating conditions
This CFD approach is especially critical for newer modalities where the manufacturing equipment is often bespoke (a wave bag for example) and not subject to the standard engineering calculations which govern scale up.
This breakthrough enabled one small biotech company to design a custom wave bag for a cell therapy in just 5 weeks. Before any lab-based experiments were conducted various virtual experiments were conducted that provided information about mixing times of media and shear effect which could negatively impact product yield. Using this information rocking wave bag dimensions, rocking angle and fill level were all selected before a narrow set of experiments were run in the lab to validate the results. This process resulted in choosing a suitable wave bag in half the time it normally took, while ensuring valuable laboratory resources were focused on other high priority objectives.