CFD Improves Scale-Up of Biological Mixing Tanks

ANSYS Fluent can reduce expensive experimentation scaling-up pharmaceutical mixing tanks for manufacturing.

The Value of CFD to Scale-Up Production

CFD is vital for the design of mixing tanks. Image courtesy of ANSYS.

CFD is vital for the design of mixing tanks. Image courtesy of ANSYS.

In the biotech manufacturing industry, computational fluid dynamics (CFD) plays an important role linking bench-scale processes to manufacturing.

“The manufacturing scale doesn’t have the luxury of experimentation and trial and error,” said Yogesh Waghmare, staff scientist at Genzyme, a biotechnology company.


Phenomena

  • Ensuring microbe survival in mixing tanks

 

Applications

  • Pharmaceuticals
  • Food processing
  • Green energy
  • Wastewater treatment

Main Software

  • ANSYS Fluent

Analysis Type

  • CFD

Computing Power 

  • Typically eight cores

Mesh

  • Two-three cells across impeller blade thickness
  • Typically automated tetrahedral-mesh

Models

  • Single phase
  • Multiphase Eulerian
  • Discrete phase model

Findings

  • Mixers must reply on efficiency to ensure materials are mixed without risking cells
  • CFD can allow engineers to better scale-up production
  • CFD also useful for quality-by-design and troubleshooting production

“Engineers will want to limit the amount of experimentation while scaling up a process, and CFD plays a large role in that,” he said.

CFD, like ANSYS Fluent, isn’t exclusively used just for scaling up processes. It can just as easily be used to better understand your bench-scale process or trouble shoot an error in your production system.

However, Waghmare states that the biggest value of CFD in the pharmaceutical industry is establishing a relationship between that bench scale and the production scale.

Essentially, CFD allows engineers to minimize the risk associated with scaling up their system and build it right the first time.

Once your system is scaled up, engineers can then validate the simulation to better link the bench-scale model to the production scale equivalent. Waghmare explained that, once you have this connection between the bench and production scale, you can continue to iterate the bench scale to predict outcomes in the manufacturing scale.

This quality-by-design mentality will expand the operational conditions of your production scale and assist your ability to investigate an error in production. After all, it is cheaper to troubleshoot the root cause of an error in the bench scale model.

Simulation Challenges in the Biotech/Pharmaceutical Industry

Challenges designing bioreactors: efficient mixing and ensuring preservation of organisms. Image courtesy of ANSYS.

Challenges designing bioreactors: efficient mixing and ensuring preservation of organisms. Image courtesy of ANSYS.

Designing a mixing tank in any industry typically has the same goal: to ensure components are well mixed at limited energy input. Whether mixtures involve liquid-liquid combinations, suspended solids, or dissolving gasses, this efficiency goal holds true.

“Having a well-mixed material is easy,” said Waghmare. “Just mix it very intensely. However, each industry has added constraints. For biological mixers you can’t mix too rigorously. In fact, our mixing is so slow that you might not mix well enough before your cells start to suffer. So you have to be smart and efficient about how you design your equipment to minimize turbulence, shear and velocities. You want things to be barely sufficient enough to get right level of mixing.”

This is why CFD plays such a big role in designing bioreactors and mixers. You can’t rely on classical industry wisdom to ensure your components are well mixed. CFD allows you to be smarter with how you mix by assessing design parameters and geometry before you build your optimal mixing tank.

How to Set Up Your Bioreactor Model Geometry

ANSYS’ mixing template can be used to set up the bioreactor’s geometry. Image courtesy of ANSYS.

ANSYS’ mixing template can be used to set up the bioreactor’s geometry. Image courtesy of ANSYS.

“There is a synergy for how to design a bioreactor,” explained Waghmare. “The first step is to know your design criteria and key parameters. This takes process experience and understanding of your bench scale system.”

In other words, the engineer needs to know which factors have the strongest impact on the survival of microbes in a bioreactor system: the even distribution of nutrients and pH buffers, microcarrier distribution, or shear and turbulence of the fluid flow. For each process, the answer might differ.

The engineer then needs to model an initial reactor geometry in the ANSYS Fluent software. This can be based off laboratory-scale equipment or existing production-scale equipment. Alternatively, the geometry can be produced by ANSYS’ mixing template if you are new to the software. Once the model is created, the engineer can change parameters in the model to see how it will affect the larger scale production.

“Engineers can play with the rpm and whatever else is easy to change first,” suggested Waghmare. “If those don’t work, you can make more complicated changes like adding baffles, changing number of impellers and geometry.”

Setting Up and Validating the Physics in the Bioreactor Simulation

There are two ways that an engineer can model their mixers: as a single or multiphase flow. During the initial stages of design, an engineer will typically want to start with a single phase model and learn as much as they can. They will still need to run the more complicated and computationally expensive multiphase simulation, but it is best to start off your design cycle with a simplified model.

Setting up the physics for the bioreactor. Image courtesy of ANSYS.
Setting up the physics for the bioreactor. Image courtesy of ANSYS.

“A typical single phase analysis can be carried on in few hours on HPC clusters,” said Shitalkumar Joshi, senior manager at ANSYS. “This provides information about impeller performance, blend-time [liquid-liquid mixing] and shear rate profiles inside the bioreactor.” 

Joshi explains that, unfortunately, single-phase simulation cannot predict the mass-transfer rate, which is vital to bioreactor designs. To determine this value, a multiphase simulation is required.

“Multiphase CFD models here are Eulerian multiphase models with population balance,” said Joshi. “This simulation option will enable users to model gas sparging, gas-liquid interaction, bubble-size distribution and the mass-transfer co-efficient. This simulation is CPU intensive and will require about a day’s turnaround time with about 48 CPUs. [Fortunately,] today’s exponentially increasing computing power have made such simulation routine.”

Simulating the cell biology itself is rather complicated, and there are not many options available to perform such a procedure. Therefore, Joshi suggests an alternative. He said, “You don’t need to involve the biology to determine the mass transfer coefficient. Experimentally, you measure the mass transfer coefficient measuring the air and water. You can do that with CFD just as easily.”

By adding more multiphysics into the simulation, oxygen dissolution, biological reaction and carbon-dioxide stripping can all be assessed. However, Joshi notes that these more complicated models will be extremely expensive computationally and much more difficult to set up — typically limiting their use to academic research. In industry, engineers rarely go beyond single phase and multiphase simulations.

Validating the Simulation Reduces the Risks of Scale-Up

Tracer blending simulation can help to validate your design with experimentation. Image courtesy of ANSYS.

Tracer blending simulation can help to validate your design with experimentation. Image courtesy of ANSYS.

Validating your designs is important to limit the risk of scaling up of your system. If you rely solely on the model and simulation, your risk tolerances will be too high. However, Waghmare explains that validation is expensive and therefore there is a cost versus risk assessment to be made.

Additionally, many of the important physical phenomena are impracticably expensive to assess in real life, such as fluid turbulence and shear.

At some point, validation requires either using a variable, which is easily measured, or relying on empirical calculations. Once you have reasonable accuracy, Waghmare admitted that at some point you have to have a leap of faith.

However, Waghmare noted that validating your model has become more important as software allows for more complex Multiphysics simulations. One could suggest that the Multiphysics trend is a double-edged sword. Though the technology allows for more accurate simulation results, finding the right parameters to get those accurate results is becoming ever more challenging.

Waghmare said, “One challenge with CFD is that you have a series of model options and set up parameters. These model parameters can also have three to four selections to choose from. If you use an Eulerian model, you can have multiple force, turbulence and dispersion settings. Each setting will describe a different model, so which ones should you choose? You can rely on literature, but often, with complicated models, you won’t get your results to match your validation testing. That is why validation is so important.”

CFD Simulations Allow Engineers to Incorporate Quality-by-Design

In the past, biotech/pharmaceutical companies registered a specific process with the FDA. These companies typically had to stick to tight production specifications to ensure their product met with the FDA’s guidelines.

However, an added benefit of simulation is that you are able to assess more operational conditions that your products may face. This is called quality-by-design. As a result, the biotech/pharmaceutical industry can reduce their waste batch runs by widening the production margins the FDA will allow.

“Companies undertake quality-by-design to understand a wider range of operation conditions that still lead to valid products. You can have multiple parameters when you run your process. To assess which are valid, you need to explore the design space of your process,” said Marc Horner, technical lead at ANSYS for the healthcare sector.

“When we have quality-by-design we are talking about a range rather than a single point of operation,” said Waghmare. “The concept is that certain parameters scale independently and some [are] dependent on size. For example, temperature would remain the same but stir speed is typically higher at small scale than in a big scale. You need to run experiments in the small scale, figure out that design space and then compare it to the scaled-up version to ensure similar conditions are met.”

This is another reason why it is important to correlate your bench scale to your production scale after the design is complete. Just as CFD can help to assess the design space of the equipment in the concept stage, it can also help to assess the design space to operate said equipment. It is still expensive to perform experiments at the production scale. Transferring that experimentation to CFD and the bench-scale is a great way to reduce that cost.

To learn more about how CFD can improve your biotech/pharmaceutical designs, follow this link.

ANSYS has sponsored this post. They have no editorial input. All opinions are mine. —Shawn Wasserman

Written by

Shawn Wasserman

For over 10 years, Shawn Wasserman has informed, inspired and engaged the engineering community through online content. As a senior writer at WTWH media, he produces branded content to help engineers streamline their operations via new tools, technologies and software. While a senior editor at Engineering.com, Shawn wrote stories about CAE, simulation, PLM, CAD, IoT, AI and more. During his time as the blog manager at Ansys, Shawn produced content featuring stories, tips, tricks and interesting use cases for CAE technologies. Shawn holds a master’s degree in Bioengineering from the University of Guelph and an undergraduate degree in Chemical Engineering from the University of Waterloo.