Francis Melvin Rogallo is now known for the eponymous double-conical hang-glider wings that he developed for NASA in the 1960s. Long before that, and before NASA even came into being, he worked for the sainted National Advisory Committee for Aeronautics-NACA-and in the late 1930s conducted a wind-tunnel study of inlet and outlet designs. From this work, the results of which appeared in 1941 as Technical Report No. 713, Internal-Flow Systems for Aircraft, emerged a few maxims that seem obvious today-though not so obvious that they are not still frequently violated. Intakes, Rogallo said, are best placed at stagnation points, that is, points where the air encountering the moving aircraft momentarily comes to a stop against it. One big inlet is better than several little ones. Outlet velocity should match free-stream velocity as closely as possible. Outlets with projecting lips produce a lot of drag. And so on.
Along with a good many protruding inlets and scoops, Rogallo included in his study a couple of flush or semi-flush inlets, which violated the “stagnation” principle. Their performance was unremarkable. But a few years later, as World War II was nearing its end and the age of the jet airplane was dawning, NACA went back for another look at flush inlets.
The reason for the reawakening of interest in them was that the performance of jet engines is quite sensitive to ram recovery. You get the best results from a jet engine if air arrives at the engine face with the full energy of the free stream. With few exceptions the engines of early jet airplanes were enclosed inside their fuselages. The exceptions were German. Although both their inventor of the jet engine, Hans Pabst von Ohain, and ours, Frank Whittle, had employed radial-flow centrifugal compressors, a Junkers designer, Anselm Franz, correctly reasoning that a small frontal area would be a virtue, developed a slimmer axial-flow design, the Jumo 004, which could be placed in external nacelles like those of the Me262. Nevertheless, the barrel-shaped centrifugal style, once it had secured a foothold in Britain and America, persisted for years after the fall of the Axis, and obliged airframe designers to house their bulky engines within the fuselage. (Franz, by the way, like Whittle and Ohain, emigrated to the United States after the war; he worked for Lycoming.)
Buried engines presented a problem. How was air, of which jets inhaled a great deal, to get to the engine without losing a lot of its energy? The earliest American designs used a bifurcated duct, called a Y duct, with inlets ahead of the wing roots; Lockheed’s P-80 is an example of this approach. North American’s Sabre and Republic’s Thunderjet, along with a series of Russian MiG models, placed the intake at the nose of the fuselage and led air back to the engine, which was situated slightly behind the wing, through a long tunnel that detoured, in one way or another, around the tub in which the pilot sat.
Now, you could not send a lot of air at high speed through a long tunnel without robbing it of some of its energy, and Y ducts, though shorter, required sharp bends. Neither approach was clearly superior to the other. But it did occur to designers that a flush inlet, in addition to reducing drag by cutting down the cross-sectional area of the fuselage, might also allow for a gentler Y duct. The question was, how do you get air to flow energetically into a duct that does not present itself head-on to the free stream?
The original submerged scoops of Rogallo’s study consisted of a sloping ramp between parallel sidewalls. The Ames Aeronautical Laboratory team investigating submerged ducts began with that configuration. The first results were disappointing. The duct captured only a little more than half the free stream pressure, as measured by a rake of small pitot tubes located a short way down the duct. The inlet performed best at a velocity ratio of .5-that is, when the rate of flow through the duct was half the rate of flow through a similar-sized section of the free stream. Some of the air flowing down the duct spilled out over the sides as it approached the duct entrance, forming a characteristic S-shaped wave. The investigators decided to try matching the ramp walls to the local streamlines, and thus the familiar-in fact, iconic-shape we now call the “NACA scoop” was born.
Reshaping the inlet had accidentally yielded a marked improvement in pressure recovery. They went on to test inlets with and without raised ridges along the edges of the ramp, and with variously shaped lips on the “roof” of the duct where it disappears into the surface. They tried different ramp angles and entry shapes; a seven-degree slope and a rectangular inlet with an aspect ratio of three to five seemed best. The end result was pressure recoveries of around 80 percent at velocity ratios from .6 to .8. The ducts seemed to be relatively insensitive to misalignment with the local flow. Considering that conventional nose and Y ducts yielded pressure recoveries at the jet engine face on the order of 65 percent, submerged ducts looked promising.
The conclusions of the Ames group’s 1945 report were upbeat, but they included a caution that designers have been ignoring ever since. “The submerged inlet is essentially a high inlet-velocity-ratio type in contrast to wing-leading-edge and fuselage-nose inlets,” they wrote. This characteristic limited its most efficient use to systems “that require only a small amount of diffusion, such as the internal ducting for jet motors of the axial-flow type.” Submerged inlets were unsuitable for “oil coolers, radiators, or carburetors of … reciprocating engines,” the report continued, because “the required diffusion of the air and the range of inlet-velocity ratios is too great to give desirable characteristics at all flight conditions.”
Diffusion, in this context, means slowing down a jet of high-velocity air. It requires a long, gently expanding duct to prevent excessive turbulence; the greatest angle allowable between the sides of a diffuser is usually said to be seven degrees. As the duct expands the flow slows down, since the same amount of air must pass by each station of the duct in the same time. The lost velocity becomes pressure; if an ideal diffuser brought flow to a more or less complete halt in a large plenum chamber, the static pressure there would be equal to the dynamic pressure of the free stream. The trouble is that an ideal diffuser turns out to be extremely long-much too long to enclose in an airplane. But radiators and engine air filters don’t like a small, high-velocity jet; they want their air moving relatively slowly but over a broad front. They are therefore inherently mismatched to NACA scoops.
The essential difference between the mechanism of a submerged inlet and a pitot-type inlet-that is, one that protrudes from the surface and faces directly into the wind-is that flow deceleration and pressure recovery take place in the internal ducting of the flush inlet, whereas they occur in the free air ahead of the protruding inlet. A protruding inlet with a well-formed lip can have very high pressure recovery over a wide range of flow rates, from wide open to completely throttled. The NACA scoop works well only when air flows through it fairly rapidly, because its divergent sides create twin vortices that draw high-energy air down from above the boundary layer and inject it into the center of the duct. Without the proper throughput there are no vortices, and the duct loses its magic.
After the end of World War II, Ames conducted further tests on a full-scale jet fighter model in its 40-by-80-foot wind tunnel. Very high pressure recoveries were obtained-up to 94 percent. Various inlet locations were tried, from well ahead of to above the wing. A more aft location was theoretically desirable because it reduced the length of internal ducts; but it turned out that the accelerated flow over the wing’s upper surface triggered transonic problems. The best location was ahead of the wing.
In 1950, North American, builder of the highly successful Sabre series, flew a prototype, the YF-93, that mated Sabre wings and empennage to a slightly larger fuselage and engine. It was the first airplane to use NACA scoops to supply air to a jet engine. The Air Force thought well enough of the design to order 118 while the airplane was still on paper, but they canceled the order in 1949. I believe the money eventually went to buy B-47s from Boeing instead.
North American completed and flew two XF-93 prototypes in 1950. The novel air intakes didn’t work. Flow instability at high angles of attack led to compressor stalls and one dead-stick landing. North American tried modifying the ducts in various ways, including bulging the entrance somewhat, so that it was only semi-submerged. Eventually, they replaced the submerged inlet with a conventional protruding one, but it was too late for the XF-93; North American turned to the supersonic F-100 Super Sabre instead.
North American’s bad experience-and those of others, such as Long-EZ builders who tried to use them for engine cooling air-did not put paid to the NACA scoop. Today we see them everywhere, and they are widely used in precisely the applications-engine induction and cooling air inlets, oil radiator inlets, fuel tank vents, cabin ventilators, all sorts of mysterious apertures on Firebirds and Lamborghinis-for which the original researchers deemed them unsuitable.
Why? Different designers probably have different motives. Many applications require peak performance at only a narrow range of flow rates anyway, and so efficiency losses at off-design points can be tolerated. Even though they are not really drag-free, submerged inlets certainly produce less drag than protruding ones do. They have practical advantages; one is that they’re easy to fabricate-much easier than a pitot-style scoop with a boundary-layer channel. NACA scoops are popular with amateur builders, some of whom are probably in the dark about the mechanism of their operation, but assume that if they’re used on some airplanes, they must be good for all.
Other reasons are more frivolous. NACA scoops look nice. They don’t appear to clutter up the surface of the airplane the way protruding inlets would. They at least seem as if they produce no drag. And that double-ogive with its gently sinking floor is an elegant shape that just feels aerodynamic in an undefined sort of way. Looks matter. When reason fails, follow your heart.