Online FEA and CFD Classes Becoming the Norm

In 2020, as the spread of COVID-19 accelerated and businesses began shutting down, Dr. Anthony Petrella, director of the Computational Biomechanics Group, launched a graduate-level certificate program at the Colorado School of Mines in Golden, CO. The program, called FEA Professional (FEA Pro), is entirely online and asynchronous. The certificate that is granted upon completion is the equivalent of completing four graduate-level engineering classes.

“At the Colorado School of Mines, we’ve been developing more online degree options here since 2017,” recalls Petrella. “COVID-19 simply accelerated our activities.” The move at Colorado School of Mines reflects a growing trend—online education and training that is removing the time and space barriers for learners.

Instructor-Led, 100% Asynchronous

Before his current academic life, Petrella worked for Johnson & Johnson, managing the computational biomechanics research group for DePuy Orthopaedics, Inc., which designs and manufactures joint-replacement implants.

“I designed FEA Pro based on my experience and industry perspectives,” says Petrella, “so the classes heavily leverage Abaqus FEA software [from Dassault Systèmes].” For all things analytical, Petrella says classes use MATLAB [from MathWorks], Wolfram Mathematica and Excel spreadsheets. Additionally, classes use Python “because it is the main scripting language for Abaqus.”

This year, Petrella began adding more software titles to the curriculum, expanding to cover Nastran, MSC Apex [from Hexagon], HyperX [from Collier Aerospace], and Ansys Fluent. “We’re software-agnostic, but we also want to give students a pathway that emphasizes certain software used in their industry of interest, like automotive or aerospace,” says Petrella.

To complete assignments, students can access relevant software by logging onto a virtual desktop preloaded with the programs. “Going forward, we are gravitating toward using more cloud-based software,” says Petrella.

Petrella prerecorded his lectures in 8- to 10-minute instruction clips and uploaded them to Canvas, the university’s course-hosting platform. Since the course is 8 weeks long, students are expected to complete the eight modules assigned to them in Canvas. The discussion board allows students to pose questions, start discussion threads and ask for help.

“Most of the answers are coming from instructors, but the students’ peers also tend to chime in,” observes Petrella.

The classes are 100% asynchronous, Petrella says, but they’re also instructor-led, in the sense that students and instructors can negotiate and select a time of mutual convenience to meet online. This allows learners from different time zones to join the class.

“Instructors like me are there to meet with students one on one, address their questions and have dialogues about specific applications of the technology,” says Petrella.

Unlike years ago, many learners and even hobbyists can now pick up basic and advanced computational fluid dynamics (CFD) software skills from YouTube. Software vendors like Dassault Systèmes, Autodesk, Ansys, COMSOL and Siemens also post tutorial videos to their YouTube channels.

“I believe students should leverage all available resources, including those vendor tutorials,” says Petrella.

AI Poses Questions

Many schools and universities are now grappling with the emergence of artificial intelligence (AI). Its potential to automate some class functions (such as grading quizzes and tests) encroaches on what has always been the instructors’ territory. Its ubiquitous access also challenges the reliability of conventional testing methods to assess a student’s output. For example, can Petrella and his colleagues detect the work of a student who used ChatGPT to complete his or her assignment?

“[Forbidding students to use AI] is a losing battle,” says Petrella. “What we need to do is to invent new ways to assess the students’ knowledge that are not susceptible to AI. We have to set up the problem in such a way so that the students need to use what they have learned to solve it, with all the resources available.”

“Teaching students about the software UI [user interface], the mechanics of the software … these are areas that can be automated with AI,” adds Petrella. “AI might be able to support students through making engineering decisions based on the CFD-generated data and making difficult choices, but the human element remains essential in that process.”

Petrella believes industry will continue to seek university graduates with traditional 4-year and graduate degrees for many years to come. But he also thinks there’s growing demand for college graduates with job-ready engineering skills who can solve practical design and analysis problems more quickly with less investment in post-hire training. This is the motivation behind his course, FEA Pro.

Tailormade, Synchronous, Remote

When Krystian Link was a student learning CFD, he went to his class in person like most people. But by the time he began teaching CFD at RandSim as an application engineer, things changed.

“There might have been a few isolated cases where a customer wanted us to teach onsite, so we did, but ever since we launched our training and enablement services, they have been predominantly online,” recalls Link.

RandSim, an authorized Ansys reseller, provides Ansys software-based simulation training ranging from 101-style courses to custom training-engagement tailored to the customer’s needs. Classes are remote but synchronous, where instructors like Link and the learners meet on Microsoft Teams.

“As part of our CFD 101 course, we have beginners’ classes for those who need to learn basic CFD, from CAD import to meshing to simulation and post-processing. We also have our Fast Start Program to address the customer’s discrete problems as well as mentoring opportunities. They are more like a college or university professor’s office-hour visits, where the engineers come to us with a set of questions, or a problem that has stumped them,” explains Link.

According to Adam Weaver, director of simulation at RandSim, class sizes are usually between six and 10 people, but some popular classes have seen their sizes grow to 15 to 20 students. The company typically runs about six classes per year, but Weaver hopes to double that number this year.

In the virtual format, when a learner raises an issue, Link uses the screen-share mechanism to allow the entire class to benefit from his instructions. “It facilitates learning across the board,” remarks Link. “For in-person training classes, the struggle used to be stopping the class to walk over to the student’s computer, solve the problem there, hope the rest of the class heard you and hope they were able to execute the steps you verbalized,” he says.

“Conversely, what the virtual format lacks is that interpersonal dialogue that tends to happen in in-person classes, like the conversation over the coffee break. However, this is largely tied to the engagement of the students,” Link notes. “If you have a group of students solely focused on consuming the material, others may miss out on the opportunity to learn best practices and skills from other industries they can apply. If you get a batch of really talkative and engaged students, the entire class (instructor included) can learn a lot of new things. Training is the necessary first step where students can be empowered to ask questions and try new approaches when they return to their day-to-day work. That’s why it’s really fitting we call our training offerings Enablement Services. We enable engineers to be better through our training.”

The virtual format also allows the attendees to remain at work, with the flexibility to address an urgent email if needed. “My students don’t need to incur hotel costs or catch a red-eye flight to come to class,” he quips.

On YouTube, LinkedIn and other social media, CFD tutorial documents and clips are growing. For many beginners, such training may very well be the starting point. But Link cautions, “With independently created simulation tutorial videos not created by an accredited source (university, simulation provider, etc.) you have to be careful. You can’t always gauge the expertise of the individual making the video. You hope they have a good grasp of the physics being portrayed and are using best practices for the given industry, but there’s no guarantee.

“The viewer may not know offhand why a given assumption works in one case and not the other, or why it can’t be applied to a certain situation they are seeing in their workday,” Link explains. “Our enablement services offer that peace of mind where you know the content is robust, trustworthy, and relevant. Our engineers have expertise in the tools and a multitude of industries. With our synchronous classes, it’s a two-way engagement. The learners get the chance to ask us questions where we can not only give them an answer but qualify it for their given industry, situation, or application.”

Whereas many university classes on simulation focus on theories, Link says, “Here, we do teach fundamentals, but heavily focus on direct application. You’re learning to deploy simulation in a product-development environment.”

Access to Mentors

Kade Beck was trained as a civil engineer. His master’s thesis was “Assessing the Effects of Local Turbulence and Velocity Profiles on Electromagnetic Flow Meter Accuracy” (2018). And his Ph.D. dissertation was “An Analysis of Electromagnetic Flowmeters: A Numerical Study” (2021). Both are published at Utah State University’s (USU) page for All Graduate Theses and Dissertations (digitalcommons.usu.edu).

“Magnetic flow meters are widely used in water distribution systems. Many flowmeters that measure your residential water use are magnetic flowmeters,” he explains. Since his professor was well-known for his research in this field, Beck, as an undergraduate, decided to apply CFD to better understand the behaviors of the same measuring devices.

As an undergraduate, he began learning CFD using the STAR-CCM+ user manual. He also borrowed a professor’s copy of “An Introduction to Computational Fluid Dynamics: The Finite Volume Method” (H. Versteeg, Pearson, 2007)—and has since bought his own copy.

Beck took his first formal CFD course at USU in spring 2020, during the COVID shutdown. The first half of the semester was an in-person lecture, but the remaining half he had to contend with virtual classes and prerecorded lectures.

“I personally prefer in-person classes. I’m the kind of student who will stay behind to talk to the professor after the class,” Beck says. “That’s when I’ve had some of the most meaningful interactions.”

He also recognizes the advantages of remote and virtual classes. “I’ve interviewed experts from all six continents for my podcast. It would not have been possible to do in person,” he says.

Publicly available resources and vendor-produced videos make learning CFD much easier today, he points out, but something crucial is missing. “The wisdom to know what matters most and how to tackle industrial problems is a skill in high demand and not easily taught,” Beck says. “I wanted to talk to the professionals who have been doing CFD for a long time, to people who could be mentors.”

This desire was the catalyst for his own podcast, “CFD 4 Industry,” available on YouTube. Launched in August 2023, the podcast features video interviews with veteran and aspiring CFD practitioners. His guests include John Chawner, a meshing guru who has held executive posts at Pointwise and Cadence Design Systems; Akshai Runchal, former president and founding partner of CFD Virtual Reality Institute; and newcomers to CFD, sharing their thoughts and journey. The podcast has gained enough tracking to allow Beck to get sponsorship—something he didn’t anticipate when he launched it.

“Every month, I get about 300 to 500 new subscribers. It’s steadily growing,” says Beck. The episode with the 60-year CFD veteran Runchal is his most popular episode. “It outperforms others both in view count and in view time,” says Beck.

Beck had conducted CFD coaching sessions online. He realizes a nominal fee of $60 is nothing more than the cost of a night out for a U.S.- or Europe-based attendee, but for many in the developing world, this amount still represents a barrier. “I’ve given my session away for free, because I know the attendees just can’t afford it,” he reveals.

More Ansys Coverage

More Dassault Systemes Coverage

Unified Modeling And Simulation (MODSIM) For Sustainable Product Development

Companies in all sectors must address the need for greater sustainability to meet customer demands. The development of new products necessitates rigorous testing and evaluation, while efforts must be made to decrease emissions and operate in a more economical fashion.

More Hexagon Coverage

Accelerating Your Journey to Smarter Industrial Facilities: Design

Transform your design processes for this new digital world by identifying efficiencies that can be gained that can help improve project delivery and overall quality in your projects.

More MathWorks Coverage