Solving biopharma design challenges with computational fluid dynamics part one: Designing a mixing tank with high-performance computing in the cloud

Solving biopharma design challenges with computational fluid dynamics

Computational fluid dynamics and cloud computing can make complex models more cost-effective and less time-consuming.

With the biopharmaceutical industry facing patent expirations and increasing competition, now more than ever, processing equipment designs, such as mixing tanks, need to be developed quickly and efficiently and work correctly the first time.

Having the right software and hardware approach helps engineers complete designs faster in a cost-efficient manner. In this three-part series, we will identify three specific design challenges and the benefits of using computational fluid dynamics (CFD).

Design challenge

When designing a mixing tank, it is important to know how its components interact to generate the desired hydrodynamic properties for adequate mixing. Sometimes, the tank mixers have tight physical tolerances and rotate at high speeds. To model such mixers properly, a very fine mesh is required, resulting in models with greater than 14 million cells with sizes of 8 gigabytes plus. With multiple similar jobs, it can be challenging to process all of them on time during peak seasons. Having an external partner to complement internal resources allows us to deliver on time, while keeping costs down.

How computational fluid dynamics solved it

These are the actions we took to achieve the desired results:

  • We built the model geometry on in-house workstations.
  • We completed model meshing for large models using Penguin Computing® On Demand™ (POD™) high-performance computing (HPC) cloud clusters.
  • We solved the model using Fluent on POD clusters for faster completion time.
  • We collaborated using the Penguin Computing Scyld Cloud Workstation™ to visualize and analyze results, communicate findings and solicit input.
  • We updated and reran the model directly on the cloud if required after having shared results.
  • We performed post-processing and completed engineering design on in-house workstations.

Results

Having an external HPC cloud partner has provided the following benefits:

  • 100 percent on-time project completion by solving models during peak seasons
  • 100 percent client satisfaction on delivery time and results provided
  • 10 percent reduction in project execution time, allowing more design time for engineers
  • Elimination of lost revenue due to computational resource limitations
  • Complete flexibility and customization: allocated storage capacity, processor type and speed, number of cores
  • 20 percent cost avoidance and cost savings compared to procuring or upgrading in-house equipment
  • Ease of use: quick adoption, minimal training needs
  • Good collaboration and communication tools

At CRB, we use ANSYS Mechanical and ANSYS CFD to help us develop innovative designs for processes and equipment. As consultants, our workload varies throughout the year in ways that can be difficult to predict. During peak seasons, we can turn to our partner, Penguin Computing, to process simulation jobs faster or to handle additional jobs in conjunction by expanding our in-house capacity. We have found that using ANSYS Fluent in POD HPC cloud clusters is a winning combination to quickly solve complex models in a cost-effective manner. Having a cloud partner enables us to size our internal resources for our usual workload instead of peak seasons, reducing yearly costs by over 20 percent.

 Return to top