Learn how large-scale scans are transformed into a CFD model for NASA evaluation of new analytical tools.

Variable vector nozzles on experimental F-15 are shown here. Source: NASA

Using computer-aided design (CAD) models, designers can evaluate any flow process-fluid, mass or thermal-via computational fluid dynamics (CFD). But the CAD model must resemble the object closely enough so that the simulated flow produces realistic data.

Designers have two sources for CAD models. For recent products, original design files can serve as the model. However, when the CAD file differs from the actual hardware, unrealistic data results.

The other route to a CAD model is to reverse engineer it from the physical object.

Softening a shock wave

Supersonic flight over the continental United States is prohibited because of noise and damage caused by the impact of pressure waves at ground level. At Edwards Air Force Base in California, NASA engineers are exploring noise- reduction techniques using a much-modified F-15 fighter. Known as LANCETS (lift-and-nozzle-change-effect-on-tail-shock), the program measures the benefits of redistributing lift on the aircraft to reduce shock wave pressure.

However, taking pressure measurements emitted by a supersonic plane from a chase plane that is less than 200 feet away is risky and expensive. NASA wanted to see if its CFD tool would simulate in-flight conditions closely enough to predict how adjustments to the lift could lower noise levels.

To do this, they needed a digital stand-in for the F-15 in the simulated wind tunnel. CAD drawings of the plane were available-in sections-but assembling them into a representation of the real airframe would not result in a model that was accurate enough to produce creditable data. Also, the test airplane had been hard landed, hard enough to make dimensions differ fractionally from original drawings. What NASA needed was something like a 3-D photo of the F-15 that could be produced quickly and would be dimensionally accurate.

Enter the laser scanner. Recently, lasers for large-scale imaging have become available that offer the ability to capture the shape of something as big as an aircraft, in extreme detail. To see if this could produce a CFD-quality model, NASA turned to Direct Dimensions Inc. (Owings Mills, MD), a metrology consulting engineer firm and a pioneer in 3-D digital reverse engineering. Accuracy of the computer model needed to be ±0.25 inch, with a resolution of 0.125 inch. And the final CAD model could be a “dumb solid”-a rapidly produced solid body in CAD that can only be modified in limited ways.

At the core of the project was a 3-D image-capturing system-Laser Scanner LS 880-from Faro Technologies (Lake Mary, FL). Operating on a tripod, the spherical laser scanner can digitize surrounding space to a distance of 250 feet. The 64-foot-long F-15 presented no problem.

The forte of the laser is data saturation of a big space; it captures so many points in a given scene (120,000 points per second) that the resulting files are like 3-D gray-scale photographs. Then, in less than 1 minute, the scanner takes an 8-megapixel image of its surrounding area, with a resolution of 0.009 degree vertical and 0.00076 degree horizontal. Packaged with its administrative software, Faro Scene, the system generates point-cloud images that can be viewed from any angle, or zoomed in to observe detail as small as data on the head of a bolt. Its uses are varied, from digitizing 3-D details of a nuclear containment room to capturing crime and accident scenes for forensic investigation.

History tells us that on December 17, 1903, Orville and Wilbur Wright were the first men to achieve powered flight. What history fails to add is that the enterprising brothers also were the first pilots to fail to properly tie down their airplane. After its fourth flight of the day, a gust caught the 40-foot Wright Flyer and flipped it, smashing the airplane and propellers. What was an instant national treasure would never fly again.

A century later, when a new generation of aeronautical enthusiasts set out to recreate the original 1903 Flyer, they wanted to use replicas of the original 8.5-foot propellers. However, all that remained of them was a fragment-barely more than half-of one that was stashed at the National Archives.

New propellers were created by reverse engineering them from the original fragment. Direct Dimensions technicians captured the shape of the fragment with a FaroArm (contact digitizer), then recreated the missing section in software. From the new CAD file, a machining program was drawn to cut duplicate propellers from a blank of laminated wood.

Five steps to "image engineering"

To generate a workable model of the F-15, NASA and the scan team took a five-step approach:

• Eliminate the surprises. Decisions had to be made as to how sections of the image would be presented. For instance, the engine intakes and exhausts represent large black holes to a laser. Other unknowns were the landing gear and the panels that cover the gear compartments when the gear is retracted. Because the analyses would be for the plane in supersonic flight, a way had to be found to image the parked plane and then digitally retract the gear.

• Get a clean target to scan. The laser scanner sees everything within its field of vision, from a wrench left on a workbench to power conduits on the wall. To the extent possible, the F-15 team cleaned up the background, eliminating surfaces that would have to be scrubbed out later in software. Another consideration was the canopy. To keep the laser beam from passing right through it, technicians treated the canopy with an opaque material to reflect the light. Curvature of the plane’s surface presented another problem; there is a natural fall-off of reflected data over curved surfaces so technicians made additional scans over these areas to be more nearly perpendicular to points around bends. Finally, most of the scans were shot in the evening to minimize “washout” of the laser beam by intense sunlight.

• Clean up the data. After two days, Direct Dimensions had more than 50 scans of the airplane-taken over and under, fore and aft, and port and starboard-for a total point cloud harvest of approximately 50 million points. These scans were initially cleaned up in Faro Scene to remove overlapping data, then stitched together to form a 3-D point cloud of the complete aircraft. Cleanup reduced the final file size dramatically, to about 1 million points.

• Extract the geometry. The cleaned point cloud image was then transferred into PolyWorks (Innovmetric, Incl, Quebec City, Canada) where the geometric shapes of the airframe were extracted, essentially reverse engineering the airframe from the point cloud. The centerline was established, cross-sections put in, axes of rotation located and contours established-all to ultimately build a wire-frame suitable for CAD. Then, as a wireframe geometry, smooth surfaces were added to form the final 3-D model. The images of the engines were eliminated altogether; NASA would add them later.

• Generate an appropriate format. Some file formats offer features unavailable in others, and for this reason, Direct Dimensions supplied six versions of two models to NASA, including formats in IGES, STEP and one containing NURBS surfaces.

X-38…Why design files can’t always cut it as CFD models

When NASA set out to use the original design files of the lifting body X-38 escape vehicle as a CFD model, the analytical data produced did not match in-flight results. At first puzzled, engineers finally measured the X-38 with a CMM and found that dimensions of the manufactured airframe differed from the vehicle’s CAD files. “The difference was significant enough to skew the CFD results,” says Ed Haering of NASA. Source: NASA

Realism: The secret to flying in the CFD environment

The F-15-837 at the center of the LANCETS program is not your grandfather’s Cessna. It is equipped with a small canard wing below the cockpit and engine nozzles that can alter thrust vectors. By changing the vectors in flight, a pilot can increase the lift on the canard, and at the same time reduce the lift from the primary wing. The net effect is a softening of the shock boom.

Up the coast from Edwards at the Ames Research Center, the CFD team assembled the pieces of the virtual wind tunnel. According to John Melton, the learning curve to adapt the F-15 models to NASA’s CFD program-CART 3-D-was short, in part because of the realism of the model. “The model gave us the asymmetries that are on the real airplane,” explains Melton.

The question of how well the laser-generated model would work was soon answered. After adjustments to make it suitable for a surface grid, the team laid a trial grid on the model and began making calculations.

“We will never be able to soften the F-15’s sonic boom to the point of making it legal to over fly the continental United States, but these CFD tools may enable us to unearth techniques that can be applied to future aircraft,” says Melton.

FARO Technologies Inc.
(800) 736-0234
Reply 15