Navigation
 
CFD

CFD Goes Mainstream

Leslie Gordon, Senior Editor, Machine Design

8/4/2008

Ten years ago, technologies such as computation- fluid dynamics (CFD) were just a gleam in the eye of everyday engineers.

No longer. The advanced analysis is moving mainstream, said participants at CDadapco’s recent Star American Conference in Dearborn, Mich. The company develops CFD software for analyzing fluid flows, temperatures, and thermal effects in everything from large parking garages to Formula 1 race cars, and even nuclear reactors.

 
courtesy of Giorgio Pagliara at Pro-S3, Italy
Results of aerodynamic analysis of a motorcycle, rider in Star-CCM+ predicted straight-line speed of 120 kph with the motorcycle well balanced without excessive lift or downforce, but sharp turns generated large amounts of lift. Modifications to bike shape produced a negative lift/drag ratio to hold the bike to the road, allowing faster cornering.

“Engineers still do the same kind of calculations and use the same equations as in the past, but the how is changing,” says Tony De Vuono, vice president and chief technical officer of Modine Manufacturing Co., “For example, our company is a Tier-One supplier of heat-transfer devices that has shifted to computational engineering. We now only hire what we call ‘virtual engineers,’ individuals well versed in simulation and digital design. We have implemented Star-CMM+ V 3.04 CFD software that even engineers straight out of school with just a Bachelor of Science can use. Experienced personnel oversee them, as well as handle more complex flow analyses.”

The increasingly widespread use of CFD is even saving dying parts of the nuclear industry, says senior development engineer of CD-adapco Emilio Baglietto. “There are a large number of older, operating reactors with efficiencies around 50 to 55%,” he says. “New reactors must be more efficient, and at higher temperatures. Here, CFD software is critical. It creates a user-independent high-quality grid that has improved fuel-bundles, an important part of reactor design. Engineers can also model random pebble distributions using the discrete element method, a technique that employs points instead of a mesh. The software also analyzes buoyancy-driven two-phase flows and boiling heat transfers.”

“CFD lets you model an entire nuclear reactor,” says W. David Pointer, a developer in the Reactor Analysis and Engineering Division of Argonne National Laboratory. “The software provides a good turbulence model and fully conformal meshing across several physical domains. It ensures constraints such as fuel burn-up, vibration, and cavitation are met based on a given set of design parameters such as duct dilation, inlet and outlet temperatures, and cooling circulation. Recently, as part of a multi-scale thermal-simulation concept, we compared the use of the Reynolds-Averaged Navier- Stokes technique with that of large eddy simulation to predict the effects of steady and unsteady fuel assemblies,” he says.

A good mesh is obviously critical to support such complex designs, according to Kurt Hamman, engineer with the Idaho National Laboratory. “So we benchmarked different meshes using a complex geometry, in this case a wirewrapped fast reactor,” he says. “In designing next-generation reactors, there are really no hard and fast rules, only general guidelines. One of our goals was to convince governmental regulators that simulation closely represents reality. We tested polyhedral and trimmed-surface meshes with 65 to 100 million elements, using a serial mesher with 65 Gbytes of memory. Results were similar, but the 65 million element polyhedral mesh took 36 hours to solve, while the 100 million trimmed mesh took 24 hours.”

Article reprinted by permission of Penton Media, publisher of Machine Design