The airline industry can save millions of dollars and dramatically reduce carbon emissions with just a one percent improvement in the fuel efficiency of commercial airliners. Small, incremental improvements, such as winglets that reduce wingtip vortices, are made to aircraft for this reason. To take a giant leap forward, NASA and MIT researchers, in collaboration with industry partners, are using HPC and wind tunnel experiments to revolutionize the design of aircraft as we know it with the goal of reducing fuel burn to 60 percent compared to a Boeing 737 made in 2005, along with substantial reductions in emissions and noise.
To achieve this ambitious goal, NASA’s Advanced Air Transport Technology project is exploring various technology improvements. One exciting example is a set of improvements to the Boeing 737 aircraft that result in the MIT D8 aircraft design. A group of researchers led by MIT and including NASA, Pratt & Whitney, and Aurora Flight Sciences is challenging what most people think of as commercial aircraft. The Boeing 737, for example, has a tube-shaped body with wings mounted mid-fuselage, and engines mounted under the wings with horizontal stabilizers, and a large vertical stabilizer at the rear.
The D8 design uses a revolutionary fuselage shape that combines two tubes to make what has been dubbed a “double-bubble” fuselage. This new body shape makes it possible to generate additional lift to reduce the burden on the wings. And the two-aisle design means that passengers can board the aircraft more quickly. The D8 also has two vertical stabilizers (instead of one on current civil aircraft) at the rear of the fuselage, with a horizontal stabilizer mounted on top of the vertical tails, producing a shape like the Greek letter Pi (π), as shown in the figures. The most interesting aspect of the design is that the engines are partially embedded into the rear-fuselage between the vertical tails.
Among the many advantages to this aircraft architecture are: reduced structural weight, smaller wings, and smaller vertical stabilizers. But what sets the D8 apart most is a technology called boundary layer ingestion (BLI). Due to the fact that air is viscous, kinetic energy is lost when the air slows down as it brushes against the surface of a traditional aircraft. A large amount of this lost energy is in a thin layer of air, called the boundary layer, which hugs the surface of the aircraft.
Traditionally, engines are mounted on a pod under the wing or on a pod on the side of the rear fuselage to avoid ingesting the boundary layer in favor of allowing only clean flow entering the engine. Employing BLI by closely integrating the engines into the fuselage, as in the D8 design, has the potential to provide huge benefits—if aircraft and engine designers can figure out some of the challenges associated with such an integration. Long used for marine propulsion, BLI technology assures that the excess kinetic energy in the boundary layer and jet is not wasted. Instead, when the slower-moving boundary layer air is ingested by the engine, less energy needs to be added to the airflow to achieve the thrust required for cruise. The engine is re-energizing the slower flow, resulting in a substantial reduction of wasted kinetic energy.
To put the conventional wisdom to the test, two open questions must be answered: Will BLI technology require less fuel? And, can a distortion-tolerant fan (necessary to handle the variation in the speed of the airflow entering the engine due to ingestion of the boundary layer) be built without a large increase in weight or loss of performance while maintaining its life cycle? While the second effort is being addressed by a separate team, the MIT-led team is now working to answer the first question, focusing on quantifying the aerodynamic benefit of BLI. The answer will tell us how much fuel would be saved if the D8 engine integration is used and the fan ingests the boundary layer. We attack the problem on two tracks in parallel:
- On one track, we conducted an experiment in the 14- by 22-foot wind tunnel at NASA’s Langley Research Center (LaRC) using a 1:11 scale model of the D8. The electrical power required to achieve cruise conditions was measured in the wind tunnel. The same experiment was then conducted on a modified D8 where the engines were removed from their BLI position on top of the fuselage and moved to the side of the fuselage, as on a conventional business jet. We refer to this configuration as the non-BLI or podded configuration. The electrical power was then converted to mechanical power to determine if the D8 requires less power than the podded version. A lower power requirement directly translates to less fuel burned.
- On the second track, the D8 and the podded configuration are modeled on a supercomputer to answer the same question. The surface of the aircraft is represented with grids that break up the surface into small quadrilateral cells. These surface grids are then “grown” into the surrounding space to model the airflow with some off-body grids made of cube-shaped cells covering the air in the rest of the wind tunnel. To obtain an accurate grid, we reproduce computationally the conditions in the wind tunnel as closely as possible.
Simulations like this require many millions of grid cells to obtain an accurate answer. Modeling the D8 configuration and the surrounding air requires 166 million cells in 69 separate grids — putting it well beyond the ability of a smart phone, tablet or even a small cluster of very good desktop computers. While the mesh generation and setup are done on a desktop system, all the flow simulations are carried out on the Pleiades supercomputer at the NASA Advanced Supercomputing (NAS) facility at NASA’s Ames Research Center in Silicon Valley. A typical solution requires 540 cores of Pleiades for just over 24 hours of wall-clock time for each case.
Because the engine fan setting that will balance the forces on the aircraft and attain conditions suitable for cruise speed are not known beforehand, we must run several cases with the modeling equivalent of various fan speed settings to find the cruise condition. The flow simulation solutions are then post-processed to compute the mechanical power required to attain cruise. A comparison similar to the experimental effort allows us to determine the benefit of going to this new configuration.
In comparing the podded configuration to the D8, we found that the D8 burns five to nine percent less fuel than the podded version. So, along with the other improvements in performance, the D8 is an attractive possibility for saving fuel. In the coming year, the research team plans to ramp up the simulations to real-world airliner speeds to demonstrate the benefits of using HPC to shape the future of passenger aircraft.
Shishir Pandya is a member of the Applied Modeling and Simulation Branch at NASA Ames Research Center.