by Jill R. Aitoro

In building a flying replica of the Wright Brothers' first plane, the Los Angeles Chapter of the American Institute of Aeronautics and Astronautics (AIAA) had a luxury that Wilbur and Orville couldn't have imagined: Computer simulation for analyzing the airflow around the aircraft's propellers.

Combining old with new

The first flight will be recreated by AIAA in partnership with the NASA Dryden Flight Research Center. It will take place next spring in the Rogers Dry Lake Bed at Edwards Air Force Base in California.

The pilot's safety will be more ensured than Orville Wright's was a hundred years ago, thanks in part to AIAA's collaboration with Fluent Inc. (www.fluent.com) to simulate propeller aerodynamics for a range of flight conditions.

"One of the Wright Brothers' great achievements was to join the momentum theory with existing blade element theories to arrive at what we still routinely use today in the design of airplane propellers and helicopter rotors," says Dr. Christoph Hiemcke, an aerospace CFD engineer at Fluent's headquarters in Lebanon, N.H.

Orville and Wilbur performed tests with a 28-inch diameter fan in their wind tunnel. They likely spun the fan in quiescent air, Hiemcke says, which would be the equivalent of a static test of a propeller.

The brothers later performed stationary tests of an 8.5-foot propeller, and eventually tests of the full-scale propellers by pivoting one wingtip of the assembled airplane, running the engine at the desired throttle setting, and measuring the resulting thrust at the other wingtip.

Findings from these experiments might have provided some degree of confidence in their blade element theory, but the brothers recognized that a propeller performs differently during forward flight than it does while static. Unfortunately, the technology wasn't available for them to test under in-flight conditions.

"The Wrights simply tested the fan and miniature as prototypes - they were static tests, with no forward velocity," Hiemcke says. "Before the actual flight, they never tested any propellers at non-zero forward speed."

The AIAA wanted to perform its own analyses to ensure that necessary changes made to the original components wouldn't affect the performance of the replica.

"The replica being readied for flight incorporates several deviations from the historic design that were introduced to improve the safety of the airplane," Hiemcke says. "The results of the study were used to decide on whether to use authentic replicas of the propeller, and whether to spin them faster than the historic 350 RPM."

Recreating the 1903 Flyer

Referring to original blueprints, Fluent engineers created a solid computer model of the propeller of the 1903 Flyer, using Fluent's Gambit pre-processing software to draw the shape of the blade's profile at five spanwise sections.

The problem posed with modeling a spinning propeller is that when an observer is up close the passage of each blade causes flow to be variable.

"But, if we create a relative reference frame in such a way that the observer is located at the hub and spinning with the blade, then the flow is steady," Hiemcke says. This adjustment allowed Fluent engineers to solve 3D Navier-Stokes equations - the principal equations used in computational fluid dynamics (CFD) - in their steady form.

Engineers modeled one blade of the propeller in a 180-degree rotational domain, which was embedded in a hemisphere with a radius of 20 times the blade span. The hemisphere was split into two halves to provide inflow and outflow surfaces.

A triangular surface mesh was generated on the single modeled blade. Fluent's Tgrid software was used to generate five layers of wedges in the boundary layer, and tetrahedral cells filled the remaining volume. The final volume mesh consisted of 1.79 million cells.

Once the mesh was generated, engineers read the data into the Fluent solver, which performed the calculations for generating simulations of the propeller under variable flight conditions.

To encourage deep convergence - required in aerodynamics to get accurate predictions of coefficients such as lift and drag - simulations were performed with angular speeds that increased from 60 to 120, 240 and 350 RPM about the negative x-axis, reflecting the propeller's clockwise spin.

"Convergence is much easier if one ramps up the propeller's rotational speed in a few steps," Hiemcke says. In addition, increasing the number of iterations - or times the algorithm is repeated - encourages convergence. Each iteration brings with it an approximate solution that results in a small imbalance. That imbalance typically decreases as the solution progresses with each iteration.

Read More

as seen in...

Desktop Engineering, December 2004