ch5-3

THE HIGH SPEED FRONTIER


Chapter 5: High-speed Cowlings, Air Inlets and Outlets, and Internal-Flow Systems

HIGH-SPEED INLETS AND OUTLETS



[146] While the 8-foot high-speed tunnel was shut down from October 1937 to March 1938 after the drive-fan accident described in Chapter II, page 26, I was assigned temporarily to the Atmospheric Wind Tunnel (AWT). At first I feared this would be time lost but it turned out to be very fortunate.

The AWT was noted for its vast but rather uninspiring production of low-speed test data and it had accumulated a staff which included several older individuals who seemed content with this type of work. However, off in a corner of the building a bright-eyed young engineer was engaged in a special investigation of what were then called "scoops and vents" (later known by the more dignified name "auxiliary air intakes and exits"). F. M. Rogallo had acquired a better-than-average understanding of propulsion theory from the extensive propeller research carried on by Durand and Lesley, his teachers at Stanford. He was now making good use of this background in setting up a meaningful theoretical framework for the scoop-and-vent investigation (ref. 176). I was assigned to work with Rogallo on this interesting project. Of special interest to [147] me was a part of the analysis which could be applied to the internal flow system of the NACA cowl. It showed that the drag power expended by the airplane to propel an internal flow system was always significantly greater than the "pumping power" required to force the internal flow through the cowling as considered in Theodorsen's cowling analysis (ref. 174), the difference being the power represented by the velocity of the wake. It was obvious to me almost at once that direct calculation of the internal drag of the cowling system from the pressure loss data would have been much more useful than the pumping power of the PRT cowling reports, in which the drag due to the coolant air flow could be found only indirectly by analysis of the total drag measurements. Rogallo had shown that a previous outside study of scoops and vents (ref. 177) also suffered from analytical flaws. Nearly all of these small openings were found to have high drag coefficients, especially if uncontrolled intake or exhaust of air was involved, and it was clear that aircraft of that day, many of which carried a multiplicity of these small openings, were paying a severe penalty.

In mid-December, I was asked to work with Abe Silverstein in the Full-Scale Tunnel on boundary-layer measurements for a family of full-scale wings. I learned a great deal about turbulence, transition, and hot-wire techniques which was used repeatedly on my return to 8-foot (ref. 85). Furthermore, collaboration with Silverstein was an interesting education in itself (ref. 178).

Eastman N. Jacobs paid an unexpected visit to our office in the 8-foot tunnel early in 1939, shortly after Stack had become Section Head. He was in the early stages of the Campini system investigation (Chapter III, p. 68ff.) and was concerned about how best to design an inlet at the fuselage nose to handle the large propulsive airflow. None of the performance numbers had been firmly fixed, but a flight speed on the order of Mach 0.8 or higher was contemplated. I described our cowling work leading to Cowl "C'' with its critical speed of Mach 0.64. Industry engineers I had talked to previously were delighted with Mach 0.64, which in all cases had been well beyond their level-flight speeds. But here was a man who wanted Mach 0.8 plus! Jacobs was also hoping for substantial runs of low-drag laminar flow over the fuselage forebody, another requirement which seemed to me then to be impossible. We had [148] never been able to avoid a suction pressure peak at the nose of the cowls and this, it seemed to me, would trigger transition. I mentioned the turbulent pulsations found in the cowlings in the PRT; these also would tend to prevent laminar flow. Jacobs fidgeted with characteristic impatience at these objections. He noted that the Campini system had no propeller and thus a basic requirement of the cowled engine-that the plane of the propeller be close to the face of the engine-did not apply. There was no limitation either on the size of the inlet opening as there had been in the cowl work. In principle, it should be possible to improve on Cowl "C." The same approach which improved the critical speed should also favor longer runs of laminar flow. As Jacobs departed, he said that his colleague Ira H. Abbott had developed a family of streamlined body shapes with falling pressures back to their maximum thickness stations. He would send Abbott to talk to me about using one of them in high-speed tests to develop an inlet for the Campini application.

Discussing the problem with Abbott I learned that he was dubious about the Campini system, but he argued that critical speeds beyond that of Cowl "C" would eventually be needed for more orthodox radial-engine installations. We chose a basic body shape from Abbott's family having a fineness ratio of 5-more representative of a radial-engine nacelle than a Campini fuselage. It was obvious that inlet velocity would have a large effect on inlet performance and critical speed and this suggested an inlet size substantially smaller than the cowling inlets; a diameter half that of the "C" cowl was selected.

By now it was quite clear to me that the dimensional restrictions of the current radial engines which we had arbitrarily imposed in developing Cowl "C" were artificial and undesirable. As this mental roadblock was dispelled, my imagination expanded and I began to think in larger terms. Suppose all restrictions were lifted and the question was phrased in the broadest possible terms, "What is the most drag-effective way to ingest or expel air into or out of a streamlined body at very high speeds?" To answer this question, the investigation would have to be greatly broadened. I felt a mounting enthusiasm at the prospect of contributing fundamental new knowledge. Inlet size was made a primary variable. Two types of outlet opening in various sizes were also selected (fig. 38). Both the cowling work and Rogallo's tests had indicated that interference effects [149] existed when inlets and outlets operated in combination, which were usually indeterminable; to avoid this problem, I would test all the inlets and exits separately. This meant that the ingested air would have to be removed by a large blower, or, in the case of the outlets, supplied by a blower. The blower offered an additional feature of great importance. It would ensure that the very high velocity ratios we desired would actually be attained. The blower system was a large complication requiring a flexible seal for drag measurements and careful evaluation of airflow momentum changes in analysis of the results. My experience with Rogallo's setup was valuable in designing this equipment. Jacobs' inlet test, which was now only a detail of the investigation, would be delayed a couple of months to allow for design and procurement of the more elaborate equipment.

After a week of "shakedown" and learning how best to conduct the testing with the rather complicated blower system, we were ready to start testing Nose B, the intermediate inlet sized for Jacobs' application, in August 1939. The tailoring process proceeded more easily and quickly than in the cowling work, partly because we were using wooden models. The final optimized profile provided exciting performance. The suction pressure peak that existed at low inlet velocities disappeared completely at velocity ratios greater than about 0.2 (fig. 39), and the critical speed thus became that of the streamline body itself, Mach 0.84 for our particular rather fat body. At the inlet velocity ratio for disappearance of the pressure peak, the transition point jumped rearward to the same location observed for the basic body. And so in less than nine months since his initial visit, we had provided Jacobs with an inlet fulfilling all his ambitious requirements.

Analysis of the drag results revealed an unexpected dividend: the external drag with combinations of the optimized inlets and outlets did not exceed the drag of the basic streamline body, and in some cases was significantly less. All previous work with the NACA cowlings had shown substantial increases in drag; the summary recommendations from the PRT programs suggested a drag coefficient increment of 0.033 for good cowls (30 to 60 percent of typical streamline nacelle drags). Our largest inlet, which was of NACA cowl proportions, added only about one-fourth the PRT value.

[150-151] FIGURE 38-Blower installation in 8-Foot High-Speed Tunnel for investigation of high-speed air inlet and outlet openings, and the principal shapes tested.

[152-153] FIGURE 39.-Typical drag, transition, and pressure data from inlet-outlet investigation.

[154] Another notable discovery was made during analysis of the profiles of the optimized nose shapes. When they were "stretched" analytically to a common length and depth all three had nearly the same profile. The Nose C and B contours were identical within 1 percent of their average ordinates. This implied that an infinite family of optimal nose shapes could be derived from the contours established in these tests. Designers could select the correct shape for their dimensional requirements without the need for any additional testing and development (ref. 179).

My instincts as an aeronautical engineer urged immediate exploitation of these impressive inlets and outlets in aircraft design studies. The Campini system which had triggered the investigation was an obvious application, but at the time it seemed quite remote and doubtful except perhaps to the Jacobs group. To me, the most likely near-term application was a submerged radial engine driving a pusher propeller. In my original -report (ref. 179), I had suggested that the nose inlet supply an of the air requirements for such an installation-carburetor, oil cooling, engine cooling, intercooling, and aircraft ventilation; there would be no drag-producing auxiliary inlets. Both Rogallo's work and the first "drag clean-up" studies of actual aircraft in the full-scale tunnel provided alarming evidence against the use of a multiplicity of small scoops and vents. We proceeded at once with layouts of hypothetical military aircraft employing our new openings.

I had acquired a new colleague in late 1939 in the person of D. D. Baals, freshly out of Purdue. In due course, we worked as a team on several inlet-outlet/internal-flow projects and I found the association to be both profitable and more enjoyable than working alone. One of Baals' first assignments was to design a fighter-type submerged-engine fuselage employing the new openings (fig. 40). This involved considerable stretching of the Nose B profile as recommended in my paper. Baals found that a reference length extending to the maximum diameter station was more convenient than the one first suggested, and this was adopted thereafter. We built and tested a model of the submerged-engine "airflow" fuselage, with gratifying results. All aspects of the new inlet and outlet technology were confirmed (ref. 180).

An important interface between NACA researchers and industry propulsion specialists and layout men was the Power Plant Installation....




[155] FIGURE 40.-NACA concept of submerged radial-engine fighter employing Now B high-speed inlet 1939.

[156] (PPI) group set up at Langley about this time. Organized with the help of the Army Air Corps Liaison Office at Langley in 1940, industry engineers were temporarily assigned to Langley where they pursued advanced installation work, with NACA researchers giving advice on the use of the latest research findings. The group was headed by C. H. Dearborn who frequently called on Baals and me for data and consultation related to the high-speed aspects of cowling, airflow fuselage, and nacelle design. The first company-proposed tentative submerged radial engine installation to appear, after our work had been published, was a pusher-propeller design for the XP-59 incorporating an R-2800 engine in a 22-foot-long nacelle. We provided the design of a 20-inch diameter version of Nose B, and made both internal and external drag calculations, including an estimate of the effect of waste heat recovered as thrust. Among other aircraft installation studies for which we provided similar aid were the B-241d, XB-33, and XB-36.

The era of the submerged radial engine was short-lived, as interest shifted suddenly to jet-engine installations. Following our work on the XP-59 submerged R-2800 nacelle in the spring of 1941, there was a great silence from the Bell Company and the Army as to the progress of this project. Actually the XP-59 had been selected in mid-1941 to become the first U.S. jet-propelled airplane, but such absolute secrecy had been imposed by General H. H. "Hap" Arnold that NACA was not allowed to participate in this project until it was reclassified "Confidential" in 1943 (ref. 41). Our simple high-speed inlets and outlets were ideally adaptable to jet-engine installations, and the submerged engine fuselage arrangement we had developed for the radial engine (ref. 180) became a popular arrangement for jet aircraft. Among the first were the Navy XFJ-l and D-558-1 and the Air Force P-84 and F-86. The jet nacelle or "pod" also afforded an almost ideal application, a recent example being the C-5A. After the first decade of jet aircraft, as speeds moved upward into the supersonic region, both the inlet and jet exit problems developed new complexities involving variable geometry and integration with other features of the airframe which have been the subject of much additional research and development beyond the scope of this review.

There were two other ways of adapting our high-speed nose inlets to [157] radial-engine/propeller installations. The so-called NACA "D" cowling (ref. 163) employed a very large spinner, in part to cover the inefficient hub sections of the propeller, and in part to permit high inlet velocities with their resultant benefit in high critical speed (ref. 163). The contours of both spinner and cowl sections could be derived from our stretched Nose B ordinates. Only a few piston engine installations of the "D" cowl were flown, but it found important later applications in turboprop aircraft.

The other alternative way to use our high-speed inlets in tractor propeller installations was the NACA "E" cowl (ref. 163). (These baptisms had been adopted by the PPI group in 1941.) In this design, the nose inlet lines were extended forward through the plane of the propeller, necessitating a large open-nose spinner. Our purpose was primarily to obtain low drag and high critical speed, but secondarily the hollow spinner offered the possibility of pumping cooling air if it were equipped with appropriate fan blades. This latter possibility had been the prime objective of an earlier test of a "blower spinner" in the PRT (ref. 181). Unfortunately, the PRT model was so crudely designed and constructed that the tests had little meaning. The blower efficiency was on the order of 50 percent and inspection of the external shape suggests a low critical speed. This PRT project is incorrectly said to be the origin of the blower spinner and the "E" cowl in ref.163. Actually, the first blower spinner was developed in 1926 by Magni (ref. 164), and the "E" cowl originated from the 8-foot-tunnel program in 1940. The first investigation of a correctly designed "E" cowl was the work of McHugh in the 19-foot pressure tunnel in 1941 (ref. 182). This was the same design used later in our emergency propeller program in the 8-foot tunnel (Chapter IV, fig. 30). A number of taxing design problems were solved in developing the "E" cowl, one of them the problem of the spinner-body juncture. Several of our engineers favored some sort of sliding seal which was very difficult mechanically. We solved the problem by contouring an open juncture to serve as an efficient outlet for the leakage flow. It was necessary to test this design to convince several skeptics that it involved only negligible drag and pressure losses (ref. 183). The "E" cowl had generally excellent performance but it never found an aircraft application because of its mechanical complexity and [158] vulnerability, and because the advent of the jet eliminated propellers for most high-speed aircraft.

After I left the 8-foot tunnel in July 1943, Baals continued to work the problem of applying our "universal" Nose B profile to a variety of design situations. He sensed the desirability of making it easier for industrial designers to arrive at optimal configurations. With assistance from N. F. Smith and J. B. Wright, he spelled out a system for deriving "NACA 1-series inlets" and produced an appropriate identification code. Some 15 illustrative inlets were laid out and selected inlets were tested to prove the validity of the "stretching" process. Design charts were prepared which made the selection process virtually foolproof (ref. 184). This work gave identity and visibility to the NACA high-speed inlets which would otherwise have been lacking. This system of design has been successfully applied not only to simple nose inlets, but also to scoops, wing inlets, circular inlets, and even to spinners in the "D" cowl. Their performance has also proved acceptable in some cases at supercritical speeds extending above Mach 1 (refs. 185, 186).

COMMENTARY

Like the results of the original NACA cowling tests, the advances achieved in this investigation were there waiting to be discovered and evaluated accurately. There was nothing remarkable in the testing and analyses, but a very important, very simple principle was involved in the initiation and planning of the project which deserves to be underscored. In the words of the first report (ref. 179), "The present investigation was designed....without any restrictions arising from engine dimensions, location, or air-flow requirements." There are many examples of research which could have been greatly enhanced if restrictions relating to current system concepts had not been imposed. A well-known example is the failure of the U.S. propulsion community to involve itself with jet propulsion in the years prior to 1942. Propulsion research was slaved so strongly to the piston engine because of its low fuel consumption that serious attention to jet propulsion was ruled out until the British and German achievements revealed the true potential.

The idea that a single universal inlet profile could be manipulated [159] to fit all sorts of scoops, wing inlets, spinners, etc., and still provide optimum drag and critical-speed performance is, of course, not believable in the exact sense. What is implied in the apparent universality of the Nose B profile as applied in the NACA 1-series system is that "approximately optimum" shapes are adequate in most cases. If one were starting over today, the indicated approach would probably be theory plus the modem computer. It might prove practical by this means to derive the exact optimum profiles for each type of application.