A Once and Future CFD

Above: Streamlines off a Harrier jet in flight shows air turbulence while surface stress is analyzed with pro-STAR, CD adapco’s STAR-CD pre- and post-processing package.

A roundtable discussion about the state of CFD foresees an exciting future where it may be used in everything from aerospace to sneakers.

By Louise Elliot

2002, Daratech predicted a 14 percent annual growth rate for computational fluid dynamics (CFD) through 2007. That prediction might have been conservative. What’s really behind the high rate of interest and adoption of technology that was once regarded as highly specialized and more the province of scientists than engineers? We asked representatives of the three Tier One CFD development companies to address the issue in a roundtable discussion.

Desktop Engineering hosted the discussion and invited the following participants:

Paul Bemis, vice president Product Planning and eBusiness for Fluent, has more than 20 years’ experience in high technology, including assignments with HP, Dell, and Apollo. He has a BSME from the University of New Hampshire and an MSEE and MBA from Northeastern University.

Chris Reid, vice president and GM of the Fluids Business Unit for ANSYS (CFX), spent 10 years with Honeywell International Industrial Automation and Control and held several senior management positions. He has a BASc. in civil engineering from the University of Waterloo.

Bill Clark, director of Advanced Methods for the CD adapco Group, leads a team that applies CFD to “real world” uses and facilitates product design and innovation. As a Research Assistant Professor at Duke University he studied numerical simulation of the unsteady aerodynamic response of turbomachinery blade rows for GE Aircraft.

Didier Halbronn, vice president of sales for CD adapco Group, earned a civil engineering degree in France and has a M.Sc. degree from Virginia Polytechnic. With a 20-year track record in the industry, Halbronn worked for ANSYS before joining CD adapco in 2001.

Clark was in transit to a meeting in Europe at the time of the conference call and Paris-based Halbronn took over for him for part of the discussion. The group’s comments are enlightening and resulted in the following QandA:

DE: Let’s start with the current rapid growth in use of CFD, the areas where you find the most growth, and the reasons for it.

Bill Clark: We’re finding a lot more acceptance at the design level. Design engineers now know that the information they get from CFD is viable and can impact design. They have been able to access CFD only for the last five years, because of increased computer power, easier-to-use software tools, and more robust modeling. Many different steps go into doing a CFD analysis, including bringing in the geometry, the computer, the physics of the problem, finding all the pieces of the puzzle that will create a successful model. Each of these elements is challenging, but nowadays all the tools have become tightly integrated within a CAD environment, making it much easier for a design engineer to be successful.

Chris Reid: I agree. Certainly the constantly improving price-performance ratio of computing provides the horsepower needed to run these large-scale problems. Ease-of-use is also improving, particularly as it applies to the less sophisticated user. And another key factor has been the recognition by leading-edge companies that sophisticated tools such as CFD enable them to have more innovative, breakthrough product designs. For our customers, many of the products they create could not have been done pragmatically without the core usage of CFD.

Paul Bemis: All those things are true, and there are also new application areas are also opening up. The three factors you’ve mentioned, greater computer power, improved and easier-to-use code, and a need for tools to assist innovation, have all increased demand and use for CFD. With labor moving around the world in pursuit of the lowest-cost provider, the only way companies can differentiate themselves from competitors is innovation, a better product, and better quality at a lower overall cost.

DE: Where do you find new interest coming from?

Clark of CD adapco: We’re seeing a lot of interest from the medical industry. It’s an interesting market because users there aren’t classically trained CFD specialists, so they need the new ease of use in the products. Offshore oil exploration is another new area. The search for alternative sources for petroleum involves many CFD applications in harvesting, manufacturing, and transport.

Reid of ANSYS: I agree that biomedical applications are growing, and it’s also an area where multiphysics applications also come into play, to simulate fluid-structure interaction. I think growth will be higher than 14 percent in those industries now starting to embrace CFD as a mainstream application.

Bemis of Fluent: New growth centers, such as pulp and paper, production of glass, medical simulation of blood flow through stents, and so on, are asking more of CAE. We recently had a sports user studying swimsuit designs. Even the traditional markets, such as the internal combustion engine, are growing. Nowadays people use CFD to predict engine emissions, and for assistance in designing cylinders, heads, valves, and acoustics.

DE: Can new users run such software?

Clark of CD adapco: It’s not easy to find very qualified users, so we all share the goal of designing the technology to incorporate industry best practices, but still allow for reproducible results. Our strategy has been to develop vertical applications for new market niches. We also want to guide new users through the process and restrict the functionality enough to allow less-skilled people to use it.

DE: Does that mean making two levels of tools: one for qualified and another for less-qualified users?

Clark: No. Even our vertical application tools have the full functionality of our standard package at the core. Any company using the software has a full spectrum of users, with research scientists at the high end and design engineers who want to turn over as many permutations as possible at the other.

Reid of ANSYS: Historically, the ratio of CFD users to the overall engineering community has been small. So, one of the challenges for growth has been to enable the technology to be used by a much broader audience and that does require simplifying and packaging the technology to make it applicable for the design community and other engineers who aren’t masters of CFD. I think the notion of vertical applications tailored to a specific industry requirement is valid, along with offerings that integrate CFD into other tools so that communities will collaborate on solving multiphysics problems with a common model.

Bemis: In addition to different tools for different people, we need torecognize that there’s a continuum of CAE users that ranges proportionate to each person’s understanding of physics, or his training. We need to create interfaces that allow people of any educational background to do some CFD. After all, our society has delivered very complex machines and systems to people with high school educations without problems. Just think of automobiles. We can give beginners a beginning level tool, so that they can do fundamental first-pass studies, and have someone with experience watch their results and make sure they’re using the tools properly.

Above: This ANSYS image was created with CFX-5 to show surface pressure mean contours on a model of a 152-meter-tall Buddha (one of the largest in the world) to be built in northern India with a design life of 1,000 years.

This ANSYS simulation (left) shows concentration levels of fragrance clouds of perfume delivered by the ventilation system of a Swiss theater for an “interactive” performance of Rimsky-Korsakov’s Scheherezade.

Didier Halbronn of CD adapco: I think that CAD integration is a key component, because if users can change parameters easily, they can replay the whole scenario quickly in a familiar environment.

DE: What does this technology most need to maintain its momentum?

Bemis of Fluent: I think it needs greater affordability. We’ve talked about computers being fast enough, code being good enough to handle the problems, and tight coupling with CAD, making it possible to interface with other geometry and applications. Now we need to get the price to a point where CFD can be used broadly, and not just the price of the software, but the price of the whole solution, including the computer and parallel processing. I think we need a pay-as-you-go model, usage pricing, where buyers pay for what they need and no more. And for that we need to rethink licensing methods.

Reid of ANSYS: Paul has hit upon a key aspect of the pricing and delivery model for software such as CFD. Customers may tell you they want it to be less expensive. They will certainly say they want far more flexibility to deal with peaks and valleys in usage, or that they want to move the software around the company depending on the needs of the day. But this isn’t like Microsoft tools, intended to be used easily by everyone. Most importantly, companies need to look at their return on investment in terms of improved product design.

This image (above) shows the wall pressure stresses in an abdominal aortic aneurysm. The radiologist’s image, using Fluent, indicates weak sections of the vessel. After deploying a stent-graft (right) in a simulation, pathlines through the artery show a much-improved bloodflow and tells surgeons what size stent is needed and where it should be placed.

Bemis: I think you’re right and that people get a tremendous value out of our software. But we need to come up with a model that scales up on the high end, and down on the low. Up to now we’ve sold it to people who have no choice. They’re building planes, trains, and automobiles, and have no choice but to use CFD. But for those who don’t have to, such as swimsuit manufacturers and tennis ball manufacturers, we have to find a way to allow them to experiment and prove that it works, while retaining the high value returns from the high end.

Halbronn of CD adapco: Another way to make it happen is to deliver limited versions of the high-end code by defeaturing some of the high-end features that most engineers might not need. That would be in addition to the approaches suggested by Paul and Chris.

Bemis: Of course, we all have a vested interest, but many of us believe that analysis should lead design, and not the other way around. We’ve been saying for 10 years that people really need to simulate first, before they go off and draw detailed drawings. We are finally, finally beginning to see a shift, where engineers say that if a design doesn’t provide lower noise, higher quality, cleaner air, and better comfort, then we can’t win in the marketplace. That’s where an analysis tool such as CFD starts to shine.

Contributing editor Louise Elliott is a freelance writer based in California. Offer Louise your feedback on this article through Desktop Engineering Feedback.