Accelerating drug development through in-silico mass transfer modeling

Meeting the moment in Pfizer’s Digital Transformation

Pfizer, an American multinational pharmaceutical and biotechnology corporation headquartered in New York City, United States (U.S.), innovates across diverse therapeutic areas, including internal medicine, oncology, immunology and vaccines.

In 2020, the rise of emerging viruses and variants placed global pharmaceutical leaders under unprecedented pressure. This race against time created an urgent demand for the rapid scale-up of biologic antibody production, requiring high fidelity tools to guide drug development and manufacturing at speed.

Although these scaling challenges are a perennial bottleneck even under normal circumstances, the global call for rapid vaccine production transformed them into an urgent priority.

The following case study explores research conducted jointly with Pfizer, published in Chemical Engineering Science by authors Hooman Farsani, Johannes Wutz, Brian Devincentis, John A. Thomas and Seyed Pouria Motevalian.

Their challenge

Microbioreactors are essential to the production of biologic medicines. Bioreactor processes are responsible for 10 of the top 15 best-selling drugs worldwide. Their value lies in their efficiency: conducting development and quality characterization at a small scale is far more cost-effective than pilot-scale testing.

However, the vast range of operating parameters still necessitates experimental sets. The challenge intensifies during scale-up; reproducing a specific fluid mechanical environment at a larger scale is a complex feat, as physical parameters rarely harmonize perfectly across different scales.

Consequently, Pfizer researchers sought a method to accurately predict microbioreactor behavior, aiming to minimize physical experiments and streamline the path to large-scale manufacturing.

Our solution

To meet this challenge, Pfizer catalyzed its process with Simcenter™ M-Star™ software – a next-gen, graphics processing unit (GPU)-native computational fluid dynamics (CFD) software. Simcenter, which is part of the Siemens Xcelerator business platform of software, hardware and services, provides the robust framework necessary for such complex simulations.

Simcenter M-Star enables the precise prediction of fluid flow and transport inside two-phase bioreactors, significantly accelerating the path to market for critical compounds. By leveraging the massive power of GPU resources to solve lattice-Boltzmann-based transport algorithms, this technology provides the high-fidelity insights necessary to streamline process scale-up.

For decades, most two-phase bioreactor simulations have relied on the time-averaged representations of both the flow field and bubble size distribution. While these simplifications made simulations possible on standard central processing unit (CPU) hardware, they introduced significant challenges regarding parameter tuning, correlation specification and overall model trustworthiness. The maturation of GPU hardware and massively parallelizable algorithms has triggered a step-change in simulation capabilities.

While hardware costs are comparable, a massively parallel algorithm on a GPU workstation performs orders-of-magnitude faster than a similarly priced CPU environment. This performance leap allows researchers to roll back the simplifying physics assumptions once required to make CPU-based models practical.

By harnessing advanced tools like Simcenter M-Star, Pfizer developed a Digital Twin to conduct laboratory experiments. This digital-first approach accelerated manufacturing innovation while significantly reducing the scope of physical experimental studies.

With Simcenter M-Star, the team was able to mechanistically decompose gas transfer into distinct contributions from the free surface and individually sparged bubbles. Building on this, they identified the precise sensitivity of the mass transfer coefficient around each bubble relative to local fluid mechanics. By understanding these micro-level interactions, the team established a robust, physics-based foundation for predicting how these dynamics would shift at production scales, thereby providing a definitive road map for process scale-up.

Results

Simcenter M-Star provided a step-change in computational speed and algorithmic precision, enabling a level of simulation fidelity that was previously out of reach for traditional CPU-based tools. The resulting modeling approach was validated by studying an Ambr 15 microbioreactor across a diverse range of operating conditions. By leveraging this high-fidelity Digital Twin, Pfizer researchers achieved a level of granular insight previously unattainable through physical testing alone.

The insights garnered from this approach provided a definitive road map for process scale-up, significantly reducing the number of physical experiments required to characterize fluid flow. By replacing exhaustive trial-and-error with targeted in-silico engineering, Pfizer was able to streamline the transition from lab-scale discovery to global manufacturing.

Ultimately, Simcenter M-Star provided advanced digital tools, which served as a primary driver behind Pfizer’s ability to navigate complex engineering hurdles and bring life-saving drug products to market with unprecedented speed.