Chapter 2: The High-Speed Airfoil Program
[3] The first discovery of an aerodynamic anomaly near the speed of sound was made over 200 years ago by the brilliant British scientist Benjamin Robins, inventor of the ballistic pendulum. He observed what we now call the transonic drag rise by firing projectiles into this device and inferring the law of their air resistance as a function of velocity from the deflections of the pendulum (ref. 3). He states:...
...the velocity at which the body shifts its resistance [law from a V2 to a V3 relation] is nearly the same with which sound is propagated through air. Indeed if the [V2 relation] is owing to a vacuum being left behind the body, it is not unreasonable to suppose that the celerity of sound is the very least degree of celerity with which a projectile ... can in some way avoid the pressure of the atmosphere on its hinder parts ... but the exact manner in which the greater and lesser resistances shift into each other must be the subject of further experimental inquiries.
By the end of the 19th century a considerable body of understanding of the differences between subsonic and supersonic flows for projectiles had been built up by the work of Lamb, Ernst Mach, Lord Rayleigh, and others, establishing the speed ratio V/a (later "Mach number") as the controlling nondimensional parameter, and clearly implying drastic changes in the flow in the vicinity of V/a = 1.
The flight speeds of the primitive aircraft of the first two decades of this century were so low that compressibility effects were nil as far as the airframe was concerned. However, by the end of World War I engine powers and propeller diameters had increased to the point where tip speeds as high as the speed of sound were being considered (ref. 4). This appears to have been a matter of particular concern to the British who, [4] perhaps from firsthand acquaintance with Lord Rayleigh's classical studies (ref. 5), or perhaps from his direct personal advices as a member of the British Advisory Committee for Aeronautics, had become aware of a possible critical problem near the speed of sound. That the problem did indeed exist was first demonstrated by Lynam (ref. 4) in free-air zero-advance tests of a low-pitch propeller model at tip speeds up to 1180 ft/sec, the structural limit for the "thoroughly well-seasoned black walnut" test blades. The tests indicated loss of thrust and increase in blade drag, but provided no quantitative data or detailed insight into the phenomena. Wind-tunnel tests of a more representative model propeller at advance ratios in the range of flight operations were recommended.
Contemporary with this early British work, the first American tests pertaining to the propeller problem were undertaken at McCook Field in 1918 by Caldwell and Fales of the U.S. Army's Engineering Division. Almost as though complementary programs had been deliberately planned and coordinated, the Americans chose to make high-speed wind-tunnel tests of stationary propeller blade sections instead of propeller tests. The magnitude of the undertaking was by no means less than that of the British, however, because no high-speed wind tunnel had ever been built, and Caldwell and Fales had to develop the world's first such facility (ref. 6). Exploratory tests using an 8-inch diameter throat were made at the National Bureau of Standards where they were undoubtedly observed with interest by a brilliant young Ph. D. in physics, Hugh L. Dryden, who had recently joined the staff and who would shortly become a pioneer investigator of high-speed aerodynamic phenomena. After exploratory experimental work on all components, a final configuration of the Eiffel-type tunnel was decided upon and constructed at McCook Field.
The tunnel had a 14-inch diameter throat and was powered by a 200-hp motor which produced a maximum speed with test model in place of about 675 ft/sec (Mach .64). (This actual speed was never calculated correctly by Caldwell and Fales. Not knowing how to determine the true air density in the test section they used the ambient air density in the room to calculate an "indicated" airspeed from the measured pressure drop between intake and test section of the tunnel.) [5] Although well below maximum propeller tip speeds, 675 ft/sec was high enough to demonstrate large "compressibility" losses in lift coefficient and increases in drag for the thicker sections and high angles of attack. Caldwell and Fales called the speed at which these changes occurred the "critical speed" and the flow change the high-speed "burble"-terminology which was adopted by succeeding investigators. It is most interesting, however, that they made no mention of the velocity of sound or the speed ratio as a controlling parameter. At the same time, they were not surprised to find changes in the character of the flow as the speed increased. Orville Wright contributed to this outlook by telling them of a hysteresis effect he had seen in his early low-speed wind tunnel tests in which two regimes of flow occurred for certain airfoils at the same test conditions (ref. 6) (now believed to be a separated-flow condition with laminar boundary layer, alternating with an attached flow with turbulent boundary layer).
A most interesting feature of the Caldwell/Fales report is inclusion of the first recorded attempt to provide a specific theoretical explanation of the observed critical speed phenomena. Unfortunately the new hypothesis ignored the speed ratio parameter, and attempted to define a "limiting shear stress" in the flow at high speeds beyond which it would separate from the airfoil. The theory was put forward by George de Bothezat, a foreign aerodynamicist of some reputation, author of a textbook on aircraft stability, and a former lecturer at the Polytechnic Institute of Petrograd. De Bothezat had been hired by the newly created NACA and assigned temporarily to McCook Field, since NACA had not yet acquired facilities of its own (ref. 7). Between 1919 and 1921 he published no less than four comprehensive NACA papers (Reports No. 28, 29, and 97, and TN No. 2) which were creditable for their time. He went on to invent the Army helicopter which bore his name and which flew at McCook Field for 2 3/4 minutes at altitudes up to 15 feet in 1923. De Bothezat was almost certainly aware that dynamical similarity suggested the speed ratio as the controlling parameter at high speeds, but he evidently thought the assumptions of similarity were violated by flow separation.
During the same period as the Caldwell/Fales investigations, Sylvanus A. Reed was pursuing a remarkable and unaccountably often overlooked [6] program of high-speed tests of thin-bladed metal propellers (ref. 8). Reed had invented a semi-rigid. metal propeller formed from 5/8-inch thick duralumin billets, tapering to 1/8-inch at the tips. The bending moments due to aerodynamic thrust for the outer portions of the blades were balanced largely by the centrifugal moments due to rotation and blade deflection. This design made it possible to employ extremely thin sections contrasting markedly with the very thick sections of the wooden propellers then in universal use. In the introduction of his paper, Reed made the following revealing observation: "There has been a tradition general among aeronautical engineers that a critical point exists for tip speeds at or near the velocity of sound indicating a physical limit ..., something analogous to what is known in marine propellers as cavitation." Evidently the expectation of the sonic anomaly was so widely known as to be called a "tradition." Reed goes on to state, however, that the only supporting evidence for this "tradition" that he could find were the British propeller tests of Lynam (ref. 4). He notes that Lynam used blunt-edged, thick blades which, by inference from the poor performance of bullets fired blunt end forward, he postulated would have poor sonic and supersonic performance. He therefore conducted a series of high-speed tests of his thin-bladed metal propellers to investigate this postulate.
A series of metal model propellers of 17-inch, 22-inch, and 4-foot diameter were tested in still air at tip speeds up to nearly 1.5 times the speed of sound; and 9-foot diameter full-scale propellers were tested in flight on a Curtiss airplane at near sonic tip speeds with help from the Curtiss Aeroplane and Motor Company. On some of the test propellers the very thin (of the order of 4 percent thickness) outer sections had sharp leading edges. The data showed no significant changes in the thrust/torque coefficient relationships in the region of sonic speed, and only small deterioration at low supersonic tip speeds. The sound generation became very loud and "penetrating" but had none of the "confused and distressing violence" noted in the British tests. Reed concludes that the high-speed problems of the British propeller were due to "the use of [thick, blunt-edged] blades not adapted to high speeds." This remarkable investigation was made before any high-speed section data had been obtained, and it preceded by over 30 years tests of "supersonic" [7] propellers by NACA. Reed appears to have been unaware of the Caldwell/Fales program or perhaps he considered their highest test speed, V/a = .64, too low to be applicable. In any case, Reed had proved that the deterioration of propeller performance at near-sonic tip speeds could be avoided by the use of thin sections. The general failure to accord proper recognition to Reed's work in the subsequent literature may be partly due to the cumbersome and misleading title of his report, and perhaps partly to the rather limited amount of data and analysis it contained.
Following Lynam's initial propeller tests in free air the British started immediately to develop a powerful turbine-driven propeller dynamometer suitable for testing 2-foot diameter propellers at high tip speeds in their 7-foot low-speed wind tunnel. Douglas and Wood's report of this investigation (ref. 9) is one of the classical documents of the early years of aeronautical research. The tip section of their wooden test propeller was 10 percent thick and compressibility losses started at about V/a = .78. At their highest tip speed of 1180 ft/sec, V/a = 1.08, the propeller efficiency had dropped from 0.67 to 0.36. The British displayed great ingenuity in their deductions of blade section data from the measured propeller data, aided by pilot surveys and optical measurements of blade twist. The latter measurements made it possible to derive section moment coefficients showing the rearward movement of the center of pressure at the highest speeds. The inclusion of all of the test data and the detailed analysis of results in the Douglas/Wood paper may account for the fact that it is widely referenced, while the Reed paper, which contained only minimal test data and analyses, has seldom been cited in the subsequent literature.
The Caldwell/Fales program had been accomplished under the general direction of Col. Thurman H. Bane, Commander of McCook Field and also the Army Air Service's member of NACA from 1919 to 1922. Bane, is believed to have apprised the Committee of the results and arranged for their publication as a NACA report (ref. 6). Although the need for follow-on wind tunnel tests at higher speeds was quite obvious, none was attempted by the McCook group; presumably they moved on to more pressing problems. The seeds of interest had been sown, however, in both NACA and in the Bureau of Standards. It is likely also [8] that NACA was aware of the continuing British effort on the high-speed problem. The personal relationship between Joseph S. Ames, Chairman of the Physics Department at Johns Hopkins and member of the Executive Committee of NACA and Hugh Dryden of the Bureau, one of Ames' most outstanding recent graduates at Johns Hopkins, was probably a factor in NACA's negotiation of a contract for the Bureau to extend the investigation of propeller sections to high speeds. Authorization for the work was signed in 1922 by George W. Lewis, the recently appointed Executive Director of NACA and also its "Budget Officer" (ref. 10).
Lyman J. Briggs, a senior official at the Bureau (soon to become its Director and a member of NACA), was in charge of the program. He personally designed the compact balance used in the tests and also participated in the testing. The curve plotting, analysis, and evidently the report writing was done mainly by Dryden, aided by G. F. Hull (ref. 11).
Primary emphasis was on extending the Caldwell/Fales data to near sonic speeds. Rather than taking on the costly problem of designing a new wind tunnel or perhaps improving the one at McCook Field, the Bureau of Standards group located a large 5000-hp air compressor capable of continuously supplying air at 2-atmospheres pressure to a 12-inch diameter nozzle. This provided them in effect with a ready-made free-jet wind tunnel having about twice the test Reynolds number of McCook facility and a maximum speed of about Mach .95. A disadvantage was that the airfoil testing had to be done incidentally to developmental testing of the compressor at the General Electric plant at Lynn, Massachusetts. And thus it was that Briggs and Dryden found themselves on Christmas Day, 1923, subjected to the rigors of airfoil testing in an open jet. Shortly afterwards, as Dryden explained later, "We walked down the street in Lynn discussing the jet and noticed passers-by staring at us strangely and shaking their heads. It was some time before we discovered that we'd been shouting at each other at the top of our voices, both temporarily deaf as a result of working with our heads only a few inches from the large jet" (ref. 12).
The test models were 3-inch chord end-supported wings which extended through the jet boundaries. It was not possible to determine the boundary effects and thus quantitatively meaningful true section data could not be obtained. Qualitatively, however, the results were [9] of great significance, confirming and extending the findings of Caldwell and Fales. The speed ratio, V/a, was used as the primary parameter, and for the first time a hypothesis as to what might be happening was put forward which has stood the test of time (ref.11):
We may suppose that the speed of sound represents an upper limit beyond which an additional loss of energy takes place. If at any point along the wing the velocity of sound is reached the drag will increase. From our knowledge of the flow around air foils at ordinary speeds we know that the velocity near the surface is much higher than the general stream velocity . . . the increase being greater for the larger angles and thicker sections. This corresponds very well with the earlier flow breakdown for the thicker wings and all of the wings at high angles of attack.
This was the first statement of the relation between the critical speed and the known low-speed velocity distribution about the airfoil-one of the fundamental ideas in high-speed airfoil research which was resurrected and exploited in the thirties. Significantly, no mention was made of the apocryphal theory of de Bothezat.
To probe more deeply into the mysteries of the compressibility "burble" and to provide load distribution data, Briggs and Dryden undertook pressure distribution measurements on the same airfoils used in their force tests. The Lynn compressor was no longer available, and a small-capacity plant at Edgewood Arsenal had to be used, capable of supplying only a 2-inch diameter jet. It had the advantage, however, of sufficiently high pressure to achieve low supersonic velocities. Briggs and Dryden designed a converging-diverging (supersonic) nozzle which produced M= 1.08, and their program included the first known aero-dynamic tests in this country at a supersonic speed. There were three important new findings from the pressure data (ref. 13):
There was also one major misinterpretation of the pressure data. The authors stated that the lowest observed upper-surface pressures corresponded approximately to the attainment of the local velocity of sound, and that lower pressures could occur only in "dead air" spaces. "This observation suggests that in an airstream obeying the law of Bernoulli the pressure cannot decrease indefinitely but reaches a limit ... near the critical [sonic] value of 0.53." This is, of course, quite wrong. An examination of their pressure data actually shows quite clearly the existence of supersonic local velocities ahead of the probable locations of the upper surface shocks. Unfortunately, the orifice spacing of 0.25 chord in the aft region of the upper surface precludes any precise examination of the flow and this may explain the misinterpretation.
The pressure data underscored what was already evident from the earlier force data-that the burble phenomena were exceedingly complex, involving shock-boundary layer interactions quite beyond any possibility of theoretical treatment. Future researches would be almost exclusively experimental; not until the later forties, when it was learned that the shocks moved off the airfoil for Mach numbers greater than about 0.95, did valid theoretical solutions appear for Mach 1 and above.
In 1927 a conference of NACA and the military services recommended a final extension of the Briggs/Dryden program to provide force data for additional more recent sections of interest to propeller designers. Included was a typical 10-percent-thick airfoil used by Reed in his metal propellers which was one of the best tested for that thickness ratio (ref. 15). The last extension was a series of tests of circular-arc sections, recommended by the authors for the outer regions of propellers for very high tip speeds (ref. 16). Unaccountably, they made no reference to Reed's work of nearly a decade before suggesting a similar use of sharp-edged sections.
Although NACA continued to sponsor the Briggs/Dryden program until it ended in 1930, it had been decided in 1927 to develop a new high-speed tunnel at Langley and to embark on in-house NACA [11] research at high speeds. The initial direct involvement of the staff with high-speed research was the Jacobs/Shoemaker investigation of thrust augmentors for jet propulsion (ref. 17) in 1926. Although the jet propulsion connection was much ahead of its time, this study stirred in Jacobs the beginnings of a strong interest in high-speed aerodynamics. The thrust augmentor inspired in G. W. Lewis not only keen interest, but also a display of technical imagination and inventiveness seldom seen in administrators at his level. He saw in this device a possible eco-nomical means of powering a large high-speed tunnel, using waste high-pressure air from the frequent blow-downs of the Variable Density Tunnel (VDT) (ref. 18). Dr. Ames, now NACA Chairman, had also followed the high-speed testing of Jacobs, Briggs, and Dryden with interest. All were aware that a major deficiency existed in the Briggs/Dryden investigations, namely the unknown jet boundary effects. The in-house program was therefore launched with the immediate objective of obtaining accurate quantitative high-speed section data for propellers to supplement the comparative results of Briggs and Dryden (ref. 19).
Preliminary trials were made by Jacobs with a 1-inch diameter throat which indicated that the jet-augmentor principle could indeed be successfully applied to drive a high-speed tunnel. Sufficient pressure was available during VDT blow-down to induce supersonic flows, and sonic conditions could be maintained for long periods. Even with a 12-inch throat Jacobs' estimates showed several minutes test duration. The dimensions and configuration selected for the first tunnel coincided with those of the first Briggs/Dryden testing at Lynn: a 12-inch open throat with 3-inch chord wings. The proportions of the open throat and its diffuser inlet were similar to those employed in the NACA VDT and Propeller Research Tunnel (PRT) facilities. However, following Briggs' and Dryden's design, the test wing spanned the jet and was supported at the ends on a photo-recording balance designed by Jacobs and his group (ref. 19). It is unclear now what the rationale was for obtaining more accurate section data with this arrangement since it duplicated the Briggs/Dryden setup in all important respects except for the addition of a diffuser. Several of those interviewed indicated that this was a "real wind tunnel with good flow" while the former was "only an open jet" and this may reflect the early NACA attitudes. Or it may be that the [12] open throat was intended to provide a direct comparison with the earlier test results, prior to the development of an improved closed throat configuration. But this could not be verified in the interviews. In any case, by mid-1928 NACA was ready to begin using its first high-speed wind tunnel (ref. 20).
The combination of the British tests of model propellers at high tip speeds, Reed's tests of thin metal propellers, and the American investigations of blade sections by Caldwell and Fales and by Briggs, Dryden, and Hull constitutes one of the first concerted efforts of the fledgling aeronautical community to solve what was feared to be a serious obstacle to progress. By any standards, the array of talent mustered was truly exceptional. Within the short time of about five years, the problem was accurately delineated and practical solutions had been found. The use of thin sections at low angles of attack in the tip region was the basic prescription, and this was readily practical for the new metal propeller designs that were beginning to appear. Beyond that, however, the use of gearing, and finally variable-pitch and constant-speed propellers eliminated the problem entirely for the airplane speeds foreseeable in 1925. Accordingly, most of the researchers initially involved moved on to more pressing problems in other areas. Briggs and Dryden had developed sufficient scientific and personal interest to carry on for a time under their own momentum, but they both became increasingly involved with other pursuits. The pressure for blade-section research was further diminished when NACA's new "PRT"was placed in operation in 1927.
Certainly there was little comprehension in 1927 that the airframe as well as the propeller would become subject to compressibility problems. Advanced pursuit planes reached speeds of only about 200 mph and it would be six or seven years later before serious speculations regarding the "500-mph airplane" would appear. A scan of the literature of the mid-twenties shows only rare suggestions of very high future speeds. (One overly sanguine prediction found in a NACA republication of a 1924 French document (ref. 21) envisioned aircraft flying at Mach 0.8 or more by 1930, including development of some wholly new but unspecified [12] type of propulsion plus appropriate new high-speed wind tunnels to support these developments.)
The initiation of in-house NACA research in high-speed aerodynamics in 1927, coming in a period where industry pressures for such work were nonexistent (except for extending the Briggs/Dryden program to a logical conclusion), has been called an act of "great foresight" (ref. 20). More probably, the start at this particular time was a natural consequence of Jacobs' 1926 investigation of jet augmentors. This provided both the basis for Dr. Lewis' imaginative suggestion to use VDT blowdowns to actuate a "large" tunnel, and a sufficient level of interest in both men to take on such a project. Jacobs and Lewis also realized intuitively that there was a place in Langley's burgeoning stable of wind tunnels for one that could deal with high-speed problems, eliminating continued dependence on the Bureau of Standards and outside test facilities.