June 28, 2021
Thanks to modern computers and the exponential boom in computational power, CFD software brings fluid flow testing to an engineer’s computer screen. Computational fluid dynamics software can be used in a variety of industries and for many applications, allowing engineers to test, simulate, and solve physics problems and equations that follow physical laws.
Here’s why computational fluid dynamics software has become so popular.
Engineers, scientists, and researchers use computational fluid dynamics software to predict and simulate fluid flow, heat transfer, species transport, chemical reactions, particle transport and rigid-body dynamics, and other physical processes that would otherwise be too time consuming, complex, and expensive to investigate using experimental approaches.
For example, in the pharmaceutical industry, biomanufacturing processes which produce biologic drugs are an important area of study. These biologic drugs are typically produced by living organisms within stirred tank bioreactors, requiring a continuous supply of sparged oxygen.
However, oxygen transfer in bioreactors is governed by complex fluid mechanics, which causes challenges when it comes to scaling up from lab-scale, tabletop bioreactors with small operating volumes to production-sized bioreactors that can hold up to tens of thousands of liters.
CFD software helps solves problems like this by providing a physics-based approach to modeling oxygen transfer, predicting fluid flow behavior—like oxygen transfer at multiple vessel scales, in this example—so manufacturers can more effectively and accurately scale up production.
Interested in learning more about this example?
Check out our academic paper with Bristol-Myers Squibb on the topic.
The benefits can be summed up in three simple ways:
To further understand why CFD software works, let’s look at how it has evolved—and how it should work today.
Scientists and researchers have always been fascinated with fluid dynamics, and through the years many have tried to mathematically describe the motion of fluids. One could argue that the roots of computational fluid dynamics software stretch all the way back to the 17th century when Sir Isaac Newton tried to quantify and predict fluid flow phenomena through his Newtonian physical equations.
Fast forward to the early 20th century—when many agree the modern definition of CFD begins. This was when mathematical and numerical methods were being improved upon and refined. By the mid-1900s, these models and methods could be integrated to generate numerical solutions based on hand calculations. These calculations transitioned to computer-based computations with early computers in the 1950s.
Ultimately, the development of computational fluid dynamics software for commercial use began in the 1980s after Boeing, NASA, and other organizations released codes to the public. These codes have been continuously improved upon—and transformed completely—ever since. CFD software has truly come a long way.
Today, it’s available for laptops, desktops, GPU clusters, or even directly on the cloud on various platforms including Windows and Linux. Here’s how modern CFD works.
Not all computational fluid dynamics software are created equal. To be able to solve complex problems faster, three things need to be true about the solution you deploy:
Modern computational fluid dynamics software is based on Lattice Boltzmann algorithms, which solves the time-dependent Navier-Stokes equations.
Unlike legacy CFD tools that describe time-average/steady-state flow fields using low-fidelity turbulence models, Lattice Boltzmann is a mesoscopic modeling tool that describes the space-time dynamics of a probability distribution function across phase-space.
This approach, which is inherently time-dynamic, enables superior turbulence modeling capabilities, faster run times, and scales more favorably with increasing system complexity.
For modern algorithms to work, they need to be paired with modern graphical processing card architectures (GPUs).
GPUs provide significant computational power, allowing users to maximize the amount of science they can do per unit man-hour.
Grid fineness informs physics complexity—which is what modern algorithms provide—and speed informs practicality. That’s where GPUs come in. GPU-based algorithms help users model transient, three-dimensional physics in real-time.
The problem? Many “modern” CFD solutions are coded for CPUs (central processing units). So if you want more accurate results faster, you need CFD software that has been coded specifically with GPU-based algorithms.
Finally, CFD software is only as useful as it is usable. It’s no good if it takes too long to learn how to add geometries, setup models, and generate results. You need something that can help you start visualizing your simulation after just minutes of processing time. Here are capabilities that indicate ease of setup:
When it comes to setting up the simulation, your tool should allow you to focus on what matters without getting bogged down by CAD shape healing or meshing complex geometries.
By itself, the combination of a Lattice Boltzmann CFD solver and high-resolution meshes should provide the basis fluid model for your simulations that you can then layer and enrich with additional, extraordinary physics. This helps you solve more problems faster, because you already have a starting place.
Imagine how much faster you could get results if you could analyze data as it’s generated in real time by the solver. With modern CFD software that offers an integrated post-processing suite, you can.
Computational fluid dynamics software has come a long way. Modern CFD software that solves Lattice Boltzmann algorithms on GPUs has transformed the industry, helping engineers, scientists, and researchers in the pharmaceutical space and beyond solve more complex physics problems—faster and at an accuracy that rivals experimental data.
Build advanced fluid models in minutes, predict real-time dynamics with precision, and solve more complex fluid flow problems faster with M-Star CFD—CFD software for the real world.
This week, guest Aaron Sarafinas, Principal of Sarafinas Process & Mixing Consulting LLC, joins John to discuss one loaded question: “Will CFD or some CFD parallels ever replace experiment, and in an increasingly digitized world, what’s the role of experiment?”
This week, guest Johannes Wutz from M-Star Center Europe GmbH joins John to discuss turbulence modeling.
This week, guest Brian DeVincentis, M-Star’s Lead Algorithm Developer, joins John to discuss the question, “What’s the big deal about GPU computing?”
