[This is an abridged version of the 35th Wilbur Wright Memorial Lecture, which Jack Northrop read to the Royal Aeronautical Society on May 29, 1947. It remains his major statement of his all-wing aircraft. For the full document, with figures and equations, see Dave Bullard's Nurflugel Page or the appendix to Ted Coleman's book, Jack Northrop and the Flying Wing. I've emphasized some statements with bold type, either because they seemed especially far-seeing or because they bear on the crash of the YB-49; the highlighting in all cases is mine and doesn't appear in the original. -- Dan Ford]
In choosing the title, "The Development of All-Wing Aircraft," as the subject of my lecture I run some risk of being accused of writing a company history rather than a paper of the broad scope ordinarily presented before this time-honored institution. This is far from my intent, but being sincerely convinced that the all wing airplane is a valuable step in the development of aeronautics, and desiring to contribute a maximum amount to the available data in the limited time at my disposal, my paper must be confined largely to experience gained by our company in its work on this subject.
Outside the efforts of the Horten Brothers in Germany there has, until a comparatively recent time, been little physical accomplishment in the development of the all-wing airplane except by our company. The contemporary Horten development has been fully described in technical reports emanating from Germany since the close of the European war. In many instances the Horten conclusions were surprisingly similar to our own. Their work was not carried so far, however, and I doubt that they had the sympathetic and responsible governmental backing and the resultant opportunities for development accorded us.
In considering the development of all-wing aircraft I would like first to distinguish between all-wing and tailless airplanes. Most tailless airplanes are not all-wing by our definition. There is a tremendous background of development in tailless types, which has been fully reported by Mr. A. R. Weyl in Aircraft Engineering. These articles outlined a surprising number of reasons for building tailless aircraft which have motivated the various designers and constructors over the years. Only one of the many advantages to be gained through such development has inspired our work, namely improved efficiency of the airplane. . . . Virtually all our efforts have been directed toward the reduction of parasite drag and the improvement of the ratio of the maximum trimmed lift coefficient (Clmax) to the minimum drag coefficient (CDmin). It is natural, then, that we were not interested particularly in tailless airplanes as such; if we could not eliminate vertical tail surfaces, fuselages, and a substantial portion of interference drag, the gains to be made seemed not worth the effort necessary for their accomplishment.
Our work, therefore, through the years has been directed solely to all-wing aircraft, by which I mean a type of airplane in which all of the functions of a satisfactory flying machine are disposed and accommodated within the outline of the airfoil itself. Of course, we have not as yet built any pure all-wing aircraft. All have had some excrescences, such as propellers, propeller drive shaft housings, jet nozzles, gun turrets and the like. We have, however, built a number of airplanes in which the minimum parasite drag coefficient has been reduced to approximately half that ordinarily attained in the best conventional aircraft of like size and purpose, and in some of the designs completed and tested the excrescences and variations from the airfoil contour have been responsible for less than 20 percent of the minimum airplane drag.
A surprisingly large number of people, both within and without the aircraft industry, still appear to question the economic reasons for going to all the trouble to build an all-wing airplane. "Sure," they say, "after a lot of practice people can learn to walk on their hands, but it's most uncomfortable and unnatural, so why do it when nothing is gained thereby?" Actually, there are startling gains to be made in the aerodynamic and structural efficiency of an all-wing type, provided that certain basic requirements can be fulfilled by the type under question. These requirements can be simply stated as follows:
First, the airplane must be large enough so that the all-wing principle can be fully utilized. This is a matter closely related to the density of the elements comprising the weight empty and the useful load to be carried within the wing.
The dimensions of the average human body may also at times be the limiting factor but, ordinarily, in the larger types of transport or bombardment aircraft in which we are most interested, it will be found that excessive sizes are not necessary in order to secure, within a wing of reasonable thickness ratio, adequate volume for a commercial cargo or bomb load plus the necessary fuel.
The extremes explored and satisfactorily flown to date in our experience range from a "buzz" bomb having a span of 29 feet . . . to the 172-foot XB-35 long-range bomber airplane. The buzz bomb was practical because of the comparatively high specific gravity of the warhead, plus the fact that the configuration permitted almost all of the wing to be used as a fuel tank.
The XB-35, on the other hand, is considerably larger than would be necessary to provide ample space for passenger and crew comfort and ample volume for payload, be it cargo or bombs. It was designed larger than necessary because we desired to keep the wing loading comparatively low in this first large experimental venture. It has a normal gross weight of 165,000 lb., an overload gross weight of 221,300 lb., and sufficient volume within the wing envelope so that the maximum gross weight at takeoff might well be increased to over 300,000 lb., somewhat over half of which could be devoted to bombs, fuel and miscellaneous payload. It may be seen, therefore, that there is a practical range of size within which the all-wing airplane can be used. If the requirements of space and volume do not permit the full use of the all-wing principle, a rudimentary nacelle may be added without losing its economic advantages.
The second basic requirement is that the all-wing airplane be designed to have sufficient stability and controlability for practical operation as a military or commercial airplane. We believe this requirement has been fully met by hundreds of flights completed with this type, and we are fully convinced of its practicability after having built a dozen different airplanes embodying scores of different configurations incorporating the all-wing principle.
In comparing all-wing and conventional types, we may fairly assume that spans of comparative aircraft having the same gross weight are equal, and as a further simplification we may for the moment neglect compressibility effects in our comparison to the advantages of all-wing and conventional types of large bombardment or transport aircraft having maximum velocities up to approximately 500 m.p.h.
Based on these assumptions and on the following proved data on the all-wing type, a comparatively simple analysis of advantages may be made.
The ratio of the minimum parasite drag coefficient (CDmin) for all-wing airplanes to that for conventional types is approximately 1:2. Minimum drag coefficients for a number of large bomber and transport aircraft . . . average approximately .023. The minimum drag coefficients for several all-wing types have been measured both in model and full-scale configurations and vary from less than .010 to about .0113, which is the figure for the XB-35 including armament protuberances, drive shaft housings, rudimentary nacelle for gun emplacements, and so on.
The ratio of maximum trimmed lift coefficient (Clmax) for all-wing to conventional types is approximately 1.5:2.3. The latter figure is typical for a number of the large airplanes of conventional arrangement previously mentioned. The former is readily attainable in a configuration such as that of the XB-35 and may be subject to considerable improvement through the use of several types of high lift devices yet to be proved.
For comparative airplanes of the same span and gross weight the selection of the required wing area will depend either on flight conditions, including takeoff without flaps, or landing conditions. If the flight conditions govern, the ratio of required wing areas of all-wing to conventional aircraft will be 1:1 because the two wings are equally effective except under conditions of maximum lift. If landing conditions govern, the ratio will be 3:1, assuming the same landing speed in each case. If takeoff with partial flap deflection governs, the ratio will be somewhere between the above two figures.
In large all-wing bombers and transports, and a growing extent in conventional long-range transports as well, the ratio of gross weight at takeoff to landing weight will approach 2:1. Therefore flight conditions are likely to govern the selection of wing area more than landing conditions. In the following calculations both extremes are used as indicative of the range of advantage to be gained by the use of the all-wing configuration. [Calculations show] that the total minimum parasite drag of the all-wing airplane in terms of the conventional airplane will vary from 50 percent if flight conditions govern, to 77 percent if landing conditions govern. . . .
It is a well-known fact, based on the Breguet range formula, that with conventional reciprocating engines and propellers the speed for maximum range is approximately that at which parasite drag and induced drag are equal. Therefore, at the same cruising speed as the conventional airplane the all-wing type will require from 25 percent to 11 percent less power . . . and with the same amount of fuel will fly from 33 percent to 13 percent farther. . . . If the all-wing airplane is operated at its most economical speed, instead of the most economical speed of the conventional airplane, it will fly 19 percent to 7 percent faster and the range will be from 41 percent to 14 percent greater with the same amount of fuel. . . .
Under high-speed conditions with any type of power plant the parasite drag becomes a much larger percentage of the total drag than for cruising conditions with reciprocating engines. At high speed the parasite drag may account for 80 percent or more of the total, while the induced drag drops to 20 percent or less. Using an assumed figure of 80 percent parasite drag . . . the power required to drive the all-wing airplane at the same speed as the conventional airplane will be from 40 percent to 18-1/2 percent less . . . and the range, at the high speed of the conventional airplane, will be from 66 percent to 22 percent greater. . . . As turbojet and turboprop power plants both operate at relatively high speed for best fuel economy, the advantages of the all-wing configuration, when used in combination with these power plants, will closely approach the above figures for maximum range as well as high speed.
These advantages are all based on the simple aerodynamic values obtained with all-wing airplanes; namely, that; CDmin equals 50 percent of conventional CLmax equals 65 percent of conventional. The probabilities are that the minimum parasite drag can, within a comparatively short time, be reduced, at least for commercial types, to about 40 percent of the conventional figure and that the maximum trimmed lift coefficient (CLmax) may, within a similar short time, be increased to at least 75 percent of conventional values.
One of the most interesting devices for increasing maximum lift is, of course, the judicious use of boundary layer control in conjunction with turbojets or gas turbines. Another involves the development of a better combination of low pitching moment flaps and trimming devices which will permit of "lifting ourselves by our boot straps" in a more successful manner than we have achieved to date. Model configurations tested up to this time, employing such methods, have shown improvements of .1 or .2 CL over the figure now used of 1.5.
A third possibility of rather unconventional nature remains to be proved in the all-wing airplane. This consists of placing the CG behind the aerodynamic center of the wing, eliminating inherent longitudinal stability by so doing and replacing this characteristic, which heretofore we have always considered as an essential to satisfactory aircraft design by highly reliable (and perhaps duplicate) automatic pilots which take over the function of stability from the airframe and may perhaps do a better job of maintaining the proper attitude than the present classical method. While unconventional and possibly a bit horrifying to those unaccustomed to the idea, it may have practical application to very large aircraft where the pilot's skill and strength are largely supplanted by mechanical means of one sort or another, and wherein the pilot controls the mechanism which in turn places the airplane at the desired attitude. If the C.G. is located aft of the aerodynamic center the airplane will trim at a high angle of attack with the flaps or elevator surfaces deflected downward rather than upward from their normal position, thereby increasing the camber and rendering the whole airfoil surface a high-lift device. It is possible that trimmed lift coefficients in the order of 2.0 may be achieved by this method, and experiments completed to date with such a device on conventional aircraft show that the C.G. may be displaced at least 10 percent of the mean aerodynamic chord aft of a normal position without any uncomfortable results in the flying characteristics of the airplane.
When these improvements in CLmax and CDmin can be realized, further startling gains in performance will accrue, as will be outlined later. It would seem, however, that the present accomplishments offer sufficient incentive to warrant all they have cost in time, effort and money, and that the question, "Why bother with an all-wing airplane?" is already well-answered.
There are other major advantages of the all-wing type which cannot be so definitely evaluated but which can and do contribute appreciably to improvement in efficiency and range. Two of these, namely the elimination of jet-tail surface interference, and the possible elimination of wing-tail surface shock wave interference, have already been mentioned. The third, and the most immediately applicable to designs of the near future, is the improved adaptability of all-wing types to the distribution of major items of weight empty and useful load over the span of the wing. While such distribution can be made to a limited extent in conventional airplanes, it can be much more fully accomplished in the all-wing type. Such weight distribution results in substantial savings in structural weight which have important effects on the ratio of gross weight at takeoff to landing weight. An analysis of the range formula indicates that this ratio is one of the most important range parameters. Competent authority has shown that distribution of fuel in the wings instead of the fuselage of a large conventional modern transport would allow an increase in gross weight of 16 percent without increase to weight empty, with a corresponding increase in range up to 30 percent.
It is fairly obvious that the all-wing airplane provides comparative structural simplicity, plus the possibility of structural material distribution in a most effective way at maximum distances from the neutral axis, plus an opportunity to stow power plant, fuel and payload at desirable intervals along the span of the wing, which cannot be equaled in conventional types. These matters are rather intangible and difficult to illustrate by numerical relationships. They depend to a large extent on the type and size of the airplane, what it is designed to carry, and what the desired high speed may be.
Having demonstrated, perhaps, that the advantages of the all-wing type are fully worth striving for, let us consider the problems involved and their solution. Based on our present experience these difficulties do not appear now of surpassing magnitude, but in 1939 several of them seemed so serious as to discourage the most hardy optimist.
To one testing a swept-back airfoil having a desirable root thickness, taper ratio and symmetrical section, together with reasonable washout at the tips such as might be designed from the then available data, the first results were a bit terrifying. The elevator effect [on the model] was erratic, changed in sign with varying deflections, and was entirely unsuitable for the control of an airplane. It was also seen that the degree of static longitudinal stability indicated by the average slope of the pitching moment curves was less than that considered desirable in a conventional airplane. Experiments involving visual observation of tufts on the model indicated a separation along the training edge of the airfoil which was apparently due to the planform configuration, and which was responsible for the erratic curves. In early experiments a simple addition of 10 percent to the chord length with a straight line contour from approximately the 70 percent chord point to the new 110 percent chord point, almost completely eliminated the difficulty.
It was soon determined that data applicable to conventional wings with little or no sweep were completely unreliable for the degree of sweepback required in practical all-wing designs, and that a whole new technique had to be developed to determine the limits within which taper ratio, sweepback and thickness ratio could be combined for satisfactory results. All these variables were explored in a series of wind tunnel models, and when a reasonably satisfactory group of configurations had been determined it was decided to build our first piloted flying wing, the N-1M (Northrop Model 1 Mockup).
Because of the many erratic answers and unpredictable flow patterns which seemed to be associated with the use of sweepback, it was decided to try to explore most of these variables full scale, and the N-1M provided for changes in planform, sweepback, dihedral, tip configuration, C.G. location, and control surface arrangement. Most of these adjustments were made on the ground between flights; some, such as C.G. location, were undertaken by the shift of ballast during flight. The variations to which this first airplane was subjected involved two extremes of arrangement in which the airplane was found to be quite satisfactory in flight.
It is an interesting commentary on the comparative ease with which the basic problems of controlled flight were solved to note that no serious difficulties were experienced in any flight attempt, or with any of the various configurations used. Some "felt" better to the pilot than others, but at no time was the airplane uncontrollable or unduly difficult to fly. The principal early troubles were related to the cooling of the small "pancake" type air-cooled engines which were buried completely within the wing, and because of the pusher arrangement did not have the benefit of slipstream cooling in taxiing, takeoff and climb. Engine-cooling problems seriously handicapped the early flights but later, somewhat larger engines were installed and the design of the cooling baffles was sufficiently improved so that repetitive sustained flights were accomplished easily.
The first flight was more or less an accident in that, while taxiing at comparatively high speed over the normally smooth surface of the dry desert lake bed used as a testing field, the pilot struck an uneven spot. He was bounced into the air and made a good controlled flight of several hundred yards before returning to earth. Altogether, this first airplane was used in over 200 [sic] flights of substantial duration, during which numerous configurations were tested and a great deal of work was done in the determination of the best types of control surface and surface control mechanism.
From the inception of the work, longitudinal and lateral controls were combined in the "elevon," which word was coined to designate the trailing edge control surface members which operate together for pitch control and differentially for roll control. At no time during early tests did control about the pitch or roll axes give any appreciable difficulty. The control which was least expected to cause difficulty gave the most, namely the rudder.
Early in the test program it was found that the airplane had quite satisfactory two-control characteristics that is, a normal turn resulted from a normal bank without the use of rudder controls and as a result, throughout the program we have often considered the elimination of rudder controls entirely. It was indeed fortunate that the first airplane developed such docile characteristics, for many of the rudder configurations tried proved to be ineffective--or worse, affected the flight characteristics of the airplane adversely.
From the start it was determined to eliminate, to the greatest extent possible, vertical fin and rudder surfaces; first, because they violated the all-wing principle and added drag to the basic airfoil; second, because with the moderate sweepback employed in our early designs the moment arm of a conventional rudder about the C.G. was small, and an excessively large vertical surface would have resulted had we tried to achieve conventional yaw control moments. The rudder development was therefore concentrated on finding a type of drag-producing device at the wing tips which would give adequate yawing forces without affecting pitch or roll. To this end we tried 25 or 30 different configurations in flight which were first tested in the wind tunnel. As a result of this experience it was concluded that dynamic reactions were likely to be very different from static reactions; some of the configurations which looked best in the wind tunnel proved to be quite unsatisfactory in flight.
The best and most practical rudder found was one of the simplest in concept and one of the first to be flown, namely a plain split flap at the wing tip which could be opened to produce the desired drag. This flap was later combined with the trimming surface needed to counteract the diving moment of the landing flaps, forming the movable control surfaces at the wing tip of the XB-35.
Among the many flights accomplished with the first experimental airplane were several in tow of other aircraft where the distance to be covered, or the altitude to be gained, made it impractical to depend solely on the airplane's own engines. After a few minutes of acquaintanceship with the slight differences brought about by the presence of the tow cable, the airplane behaved well in tow and several comparatively high altitude flights were made to investigate the spin characteristics. These appeared to be quite normal, based on preliminary tests of this airplane. Later experience, however, indicated that the spin characteristics of tailless types vary from one design to another, in the same fashion as may be expected in conventional types, and that no broad generalization as to spin behavior can be made with safety.
The N-1M was first flown in July 1940 and for about a year was consumed in a combination of aerodynamic tests and attempts to solve engine cooling problems. As soon as good sustained flight demonstrations could be made on schedule the Army Air Forces took active interest in the program and top-flight officers, including General H. H. Arnold and Major General Oliver P. Echols, encouraged us to investigate the application of the all-wing principle to large bomber aircraft. To this end it was decided to construct four scale models of a larger airplane. These were designated N-9M (Northrop Model 9 Mockup) and they duplicated, except for the power plant and propeller arrangement, the aerodynamic configuration of the proposed XB-35 airplane.
The first of these aircraft was completed and test flown on December 27, 1942, and had completed about 30 hours of test flying with pilot (and sometimes an observer) when it crashed, killing the pilot. The machine had been on a routine test flight across the desert away from its base, and was out of sight of technically qualified observers at the time of the accident. However, all evidence pointed to a spin, and the attitude of the airplane on the ground indisputably indicated autorotation at the time of impact.
This loss was a serious setback and work was started immediately to recheck the spin characteristics of the airplane in a spin tunnel. It was later determined, both in the tunnel and in flight, that recovery was good, although a bit unconventional (requiring aileron rather than elevator action), but that the spin parachutes which had been attached to the airplane for the low-speed stalling and stability tests then in progress were ineffective as to size and improperly located.
Subsequent models, over hundreds of flights, gave no trouble. The low-speed stall and spin tests with rear C.G. positions were accomplished without further difficulty and the N-9M proved an invaluable test bed in which various control configurations could be proved in detail. A large number of additional rudder configurations were developed and tested on the N-9Ms; likewise different types of mechanical and aerodynamic boost for the control surfaces were investigated, as well as the general behavior of the airplane in all types of air, and with different C.G. positions.
In connection with the model spin tests of this airplane, an investigation of the tumbling characteristics of the type was made in the spin tunnel. These tests showed that if the model was catapulted into the airstream with an imposed high velocity about the pitch axis in either direction, it would continue to tumble or come out of the maneuver, depending on comparatively minor differences in elevon and C.G. position. In other words, under circumstances of induced rotation about the pitch axis the recovery was marginal. However, it would never tumble from any normal flight condition, such as a stall, spin, or any other to-be-expected maneuver. In some configurations, if dropped vertically trailing edge down into the wind stream, a tumbling action would be induced which might or might not damp out. This was not judged a serious matter in view of the fact that a vertical tail slide is hardly a maneuver to be courted, even by a fighter airplane, let alone a 100-ton bomber.
The three remaining N-9Ms have been flown almost continuously since their completion dates to the present. Only recently have all desirable test programs been completed and the airplanes relegated to a semi-retired status from which they are withdrawn only for the benefit of curious pilots. . . .
During all this development and testing . . . . the design and construction of the big ship had been under way. N-9M airplanes had proved the practicability of the design. They closely approached the XB-35 configuration with the exception that they mounted only two pusher engines, located at positions corresponding to points midway between engines 1 and 2, and engines 3 and 4.
The problem of control-surface actuation on the big bomber involved the development and testing of a complete hydraulic control system, as none of the aerodynamic boosts or balances developed and tested in the N-9M models had proved satisfactory. The system used in the XB-35 employs small valves which are sensitive to comparatively minute movements of the control cable and which, when displaced, permit large quantities of oil to flow into the actuating cylinders. This arrangement eliminates any pilot "feel" of the load on the control surfaces unless a deliberate arrangement for force feedback is made. Rather than undertake this later step, a comparatively simple force mechanism, which is sensitive to accelerations and airspeed, was developed. This device gives the pilot a synthetic feel of the airplane which can be adjusted in intensity to anything he likes, and which has proved satisfactory in flight. For reasons to be outlined shortly, a synthetic feel was much more satisfactory than the feedback of actual control surface loads, particularly at high angles of attack.
The XB-35 was first flown from Northrop Field to the Muroc Army Test Base in June 1946. The first several flights indicated no difficulties whatsoever with the airframe configuration. Indications of trouble with propeller governing mechanisms were discerned at an early date and it was shortly discovered that flights of any substantial duration could not be accomplished because of oil leakage in the hydraulic propeller governing system. On the last flight difficulty with both propellers on one side caused a landing with asymmetrical power, which was accomplished without trouble.
The next six months, from August to March, were spent in a vain attempt to eliminate these difficulties, plus those caused by a series of engine reduction gear failures. To date the XB-35 has not had sufficient time in the air to fully demonstrate its ability to meet its design performance guarantees. However, large-scale model tests in numerous tunnels have indicated the low-drag figures presented earlier in this paper, and preliminary speed versus power tests completed early this month have given gratifying confirmation of our original expectations. Flights accomplished to date have included all maneuvers necessary for large bombardment airplanes. So far, however, violent maneuvers have not been attempted and no exact evaluation of stability and control parameters has been possible.
Two turbojet powered all-wing airplanes, having the same basic shape and size as the XB-35 are virtually complete at this time and will be flying late this summer. They are powered by eight jets having a sea level static thrust of 4,000 lb. apiece. They incorporate small vertical fins to provide the same aerodynamic effect as the propeller shaft housings and propellers of the XB-35.
Let us now turn to considerations of stability and control of the all-wing airplane. They are quite different from those of conventional types and, unless reasonably well understood, may lead to discouragement at an early date concerning projects well worth further evaluation.
In any airplane the primary parameter determining the static longitudinal stability is the position of the center of gravity with respect to the center of lift or the neutral point. Obviously, the neutral point may be shifted aft by adding a tail or by sweeping the wing, or the C.G. may be shifted forward by proper weight distribution, so that from the standpoint of static stability no particular configuration has any special advantage except as it affects the possibilities of proper balance. In an all-wing airplane the elimination of the tail makes the problem of balance somewhat more critical but not excessively so. Unfortunately, for any given airplane the neutral point does not ordinarily remain fixed with variations of power, flap-setting or even lift coefficient, so that the aft C.G. limit for stability is often prescribed by some single flight condition has always occurred for power-off flight at angles of attack approaching the stall.
The pitching instability of a swept wing at high lift coefficients is by now a somewhat familiar phenomenon. The complete mechanisms involved, however, are still somewhat obscure. There are apparently two opposing effects which are of prime importance. They are the tendency for sweepback to increase the relative tip loading and also (by creating a span-wise pressure gradient) to promote boundary layer flow toward the tip. On a plain swept-back wing the latter effect apparently nullifies the former, so that there occurs in the tip portion of the wing a gradual decrease in effective section lift-curve slope with a resulting progressive decrease in stability. The tip, under these circumstances, never completely stalls, as evidenced by the stable pitching moments occurring at the maximum lift coefficient. On the other hand the addition of end plates will prevent to a large extent the effects of span-wise flow, thereby straightening the pitching moment curve but producing the normally expected tip stall, as evidenced by the strongly unstable moments in the vicinity of the maximum lift coefficient. Thus, any modification to the basic wing which affects the span-wise flow will have a noticeable effect on the pitching behavior at high lift coefficients.
In the case of the XB-35 the propeller shaft housings act to inhibit span-wise flow and straighten out the moment curve below the stall as in the case of the end plate; but in order to obtain stability at the stall, a tip-slot is provided to increase the stalling angle of the tip sections. By raising the trim flap in the outer 25 percent span and lowering the main flap in the inner 35 percent span, the stability characteristics are noticeably affected, presumably because of a decrease in spanwise pressure gradient and therefore in boundary layer flow.
Recent investigations have indicated that the problem of static longitudinal instability near the stall for plain swept-back wings depends not only on sweep but also on aspect ratio and it now appears that for a given sweepback the magnitude of the unstable break in the moment curve decreases with decreasing aspect ratio, eventually vanishing.
The possibility of controlling the stalled portions of the wing, as outlined, means that trailing edge flap controls can be laid out to maintain their effectiveness at very high angles of attack. Since a certain portion of this flap must be used to provide high lift and roll control, the amount available for longitudinal trim is limited, so that for the XB-35, for example, the total available nose-up pitching moment coefficient is .15 as compared to .30 for a conventional airplane. This limited control plus the fact that the main wing flaps apparently cannot be made self-trimming and impose a diving moment in the landing condition reduces the available C.G. range . . . as compared with conventional airplanes. The XB-35 has a C.G. range of only 5 percent or 6 percent as compared with conventional values in the order of 10 percent or 12 percent. This comparison is somewhat misleading, however, because the all-wing airplane may have a greater comparative [mean aerodynamic chord] in view of its somewhat lighter wing loading. It is also much easier to arrange weight empty and useful load items spanwise within close m.a.c. limits than in conventional types.
Where manual control of the elevator is employed the stick-free stability and control of all-wing aircraft are impaired by separation of the flow from the upper surface of the wing near the trailing edge, causing up-floating tendencies at higher lift coefficients. If not corrected these up-floating tendencies lead to stick-free instability and, in some cases, to serious control-force reversal at high lift coefficient. Aerodynamic design refinements devised and tested by us to date have not provided a satisfactory solution to the up-floating tendency. For small airplanes these undesirable forces can sometimes be tolerated, but for large aircraft the only solution found so far has been the employment of irreversible full power driven control surfaces.
It is when considering the lateral stability and control factors that the difference between the all-wing and conventional airplanes becomes most apparent. It is reassuring to state that despite the large differences apparent between the XB-35 and conventional aircraft, the dynamic lateral behavior of the XB-35 type is quite satisfactory, as will be discussed later.
Definite requirements for the weathercock stability . . . depend to a large extent on the airplane's purpose, but positive weathercock stability is always required. The swept-back wing has inherent directional stability which increases with increasing lift coefficient; but this is not considered sufficient for satisfactory flight characteristics under all circumstances and must be supplemented by some additional device. The wingtip fin has been favored by some since it gives the largest yawing lever arm and provides a suitable rudder location. However, as previously pointed out, wingtip fins may be unsatisfactory at the stall. For the XB-35 configuration, effective fin area is provided in large measure by the side force derivative of the pusher propellers. . . .
Considering now the effective dihedral . . . it is apparent that sweepback is the essential difference between the all-wing and conventional airplanes--a difference that will disappear as flight speeds increase and it becomes necessary to employ the desirable high-speed characteristics of swept wings in conventional tailed configurations. . . . Flight ease may indicate that a slightly positive effective dihedral is desirable while dynamic considerations point toward a slightly negative dihedral. Our practice has been to retain positive effective dihedral over the complete flight range.
The rolling control for all-wing airplanes is essentially normal. When elevons are used rather than separated aileron and elevator control, certain variations from conventional craft appear, in that, with the upward elevator deflection required for longitudinal trim, the adverse yaw ordinarily due to aileron deflection disappears. On the other hand, if large up-deflections are required for longitudinal trim, the up-going elevon used as aileron loses effectiveness rapidly, thus reducing the available roll control at high lift coefficients. This is particularly undesirable when considering the increased dihedral effects of swept wings at high lift coefficient.
All-wing airplanes, particularly those without fins, have a very low crosswind derivative; thus a low side force results from sideslipping motion. Some crosswind force is probably important for precision flight, such as tight formation flying, bombing runs, gun training maneuvers, or pursuit. This importance arises because with low side force it becomes difficult to judge when sideslip is taking place, as the angle of bank necessary to sustain a steady sideslipping motion is small. This lack of side forces has been one of the first objections of pilots and others when viewing the XB-35. After flying in the N-9M or XB-35 the objection is removed, except for some of the specific cases mentioned above. For the correction of the lack of sideslip sense, a sideslip meter may be provided for the pilot or automatic pilot, and for very long-range aircraft there is a valuable compensating advantage in being able to fly under conditions of asymmetrical power without appreciable increase in drag.
The free longitudinal motions of any airplane fall into two modes. The first of these is a short-period oscillation. It is highly damped for conventional airplanes and also for all-wing airplanes in spite of the relatively low pitch-damping, Cmq. This somewhat surprising result is due to a coupled motion such that the vertical damping . . . comes into play absorbing the energy from the oscillation. Also, low moment of inertia in pitch makes the small existing Cmq more effective than a similar value would be in conventional types. In tests on the N-9M airplane this short-period oscillation was too rapidly damped to obtain a quantitative check. The combination of low static stability in pitch, as previously described, and low moment of inertia in pitch results in periods of oscillation for all-wing airplanes that are comparable to those of conventional types.
The second mode of longitudinal motion is a long-period oscillation commonly called the phugoid. This is a lightly damped motion even for conventional airplanes, and seems slightly less damped for all-wing airplanes, because of the fact that they have relatively low drag, and drag is the chief means of energy absorption in this mode. N-9M tests indicate that calculation is slightly optimistic in this matter, but still this phugoid motion is sufficiently damped so as to give no serious difficulties. Being a slow motion, it is easily controlled. . . .
The criterion of response is probably the only category in which the flying wing is importantly different from the conventional airplane for longitudinal motion. The action of the two types in an abrupt vertical gust is especially interesting, two factors combining to reduce the accelerations experienced by all-wing airplanes. These factors are the relatively larger wing chord and shorter effective tail length of the all-wing type. The first characteristic increases the time for the transient lift to build up and is the more important in reducing accelerations. The second decreases the time interval between the disturbing impulse at the lift surface and the correcting impulse at the effective tail, so that the airplane tends to pitch into the gust. This latter characteristic is a matter of concern to pilots, since a disturbance in the air is likely to leave them farther from trim attitude, consequently requiring more active pilot control in rough air. It is believed, however, that automatic control will effectively eliminate this difficulty.
The response of the all-wing airplane to elevator deflection seems entirely adequate. It errs, if at all, on the side of over-sensitivity because of low Cmq and low moment of inertia in pitch. An abrupt control movement giving the same final change in trim speed for a conventional and a comparable all-wing airplane results in a larger initial swing in pitch for the all-wing.
As with longitudinal motion, there are two characteristic modes that are of interest laterally. the first of these is the spiral motion which is usually divergent on modern airplanes, thus uncontrolled flight results in a tightening spiral. This slight instability seems favored by pilots. All-wing airplanes have readily acceptable characteristics in this mode requiring from 15 to 20 seconds to double amplitude. In general, any time greater than five seconds to double amplitude is considered acceptable.
The second mode, the "Dutch Roll" oscillation, is more critical for all-wing airplanes, particularly at low speed, high weight and high altitude. All-wing airplanes seem comparatively bad in this respect because of the combination of relatively large effective dihedral and low weathercock stability and, for the conditions noted above as critical, are likely to approach neutral damping in the Dutch Roll mode. However, analytical determinations of this motion, using calculated damping derivatives, indicated less satisfactory characteristics than were obtained in actual flight tests. Because of a relatively low weathercock stability, the Dutch Roll is of a rather long period, in the order of ten seconds for the XB-35. It is usually assumed that for periods of such length, it is not important to have a high rate of damping since control would seem easily "inside" the motion. However, there may be particular instances where this is not true. For instance, in an all-wing airplane in which the rudder is particularly weak, the time of response to rudder control may be of the same order as the period of Dutch Roll motion. This would make directional control extremely difficult in a condition, such as landing, where the roll controls are not usable for changing heading. It is notable that for the very low weathercock stability commonly encountered in all-wing airplanes, the conventional solution of increasing weathercock stability to offset increased dihedral does not hold. . . .
Another factor contributing to the relative lack of damping of all-wing airplanes in Dutch Roll motion is the low value of the damping coefficient in yaw. . . . This appears to be inherent in all-wing designs, particularly if the use of fins is abandoned. For special occasions, when particular airplane steadiness is required (such as a bombing run), it is probable that the equivalence of such damping in yaw may be supplied by an automatic pilot, or by temporarily increasing the drag at the wing tips. This latter effect can be accomplished on the XB-35 by simultaneously opening both rudders and gives deadbeat damping in yaw.
As in the longitudinal motions, the amplitudes of response of an airplane in lateral motion are probably as important as the damping rates in determining free-flight characteristics. All-wing airplanes seem slightly rougher in turbulent air than conventional aircraft of similar weight. This is due chiefly to the reduced wing loading, but high effective dihedral and low weathercock stability may have an added effect. This is a matter of interest in fixing upon analytical criteria for the description of free-flight qualities. As mentioned above, increasing the weathercock stability for all-wing airplanes has a slight effect on the damping rates; however, it affects the amplitudes of response to gusts materially.
Some data from the free-flight tunnel of the National Advisory Committee for Aeronautics indicate that increasing weathercock stability, even for all-wing airplanes, materially helps the "flyability" of the airplane. Another bit of evidence that is of interest in this connection has to do with the magnitude of the side force derivative, Cy.B. Increase of this parameter improves Dutch Roll damping very materially but has virtually no effect on amplitude of response to gusts, according to calculations. Free-flight wind tunnel data again give tentative support to the investigations of response as a criterion by showing little improvement of flight qualities of models with increase of Cy'B.
Flight tests of the all-wing glider in which the vertical fin, located aft on the ship's center line, was varied in size from approximately 2 to 7 percent of the wing area, left the pilot somewhat undecided as to fin requirements except that the larger fin seemed somewhat easier to fly. Presumably, this was, in the light of the foregoing discussion, primarily because of the increased CnD, the coincidental increase in Cy'0 not being effective.
The application of automatic pilot control to an all-wing airplane has certain difficulties which are associated primarily with the low value of C ,B. In conventional applications the fact that the airplane is side slipping is detected by either a lateral acceleration or an angle of bank. In an all-wing airplane neither of these indications exists except in an almost undetectable amount. Accordingly, it is necessary, in order to fly the airplane at zero sideslip, and therefore in the direction of its center line, to provide a yaw-vane signal to which the pilot or automatic pilot will respond. This introduces some difficulty in automatic pilot design because for small disturbances the sideslip angle with respect to the wind, and the yaw angle with respect to a set of fixed axes, are nearly equal and opposite for a flying wing. The customary automatic pilot control on azimuth angle therefore tends to oppose the necessary control on sideslip. To avoid this difficulty it is necessary only to reduce the rate of control on sideslip to approximately one-third that on azimuth. This modification to a conventional automatic pilot was flown on the N-9M with complete success.
Let us now turn to a consideration of the practical limitations in arrangement of the tailless airplane. They may be summarized briefly as sweepforward, sweepback, and a non-swept wing configuration. The sweepforward arrangement requires the use of a large fixed load forward of the leading edge at the center section for proper balancing of the airplane. Therefore, a fuselage with some substantial part of the weight empty of the airplane disposed therein is required. The swept-forward wing itself is unstable directionally and requires some type of fin for weathercock stability. To this must be added more fin area to stabilize the fuselage. In addition, it may be noted that the moment arm of the fin about the C.G. of the airplane is necessarily comparatively small, still further increasing the size of the required fin. If we add to the airfoil a protruding fuselage and an unusually large vertical tail surface, we have departed from our basic all-wing concept. We have incorporated virtually all the elements of drag found in the conventional aircraft and have not accomplished our intent of improving efficiency.. For the above reasons, which could be argued pro and con for hours, our company has done no active design and development work on airplanes with swept-forward wings.
An all-wing configuration embodying a straight, or non-swept wing, has been proposed and flown successfully in model sizes. It offers the serious disadvantage that suitable distribution of weight empty and useful load items is difficult and, if proper balance is to be accomplished, most of the structural weight and useful load must be included in the forward 30 percent or 40 percent of the wing, leaving a large volume of space within the wing unusable. Such a configuration results in an unnecessarily large airplane to accomplish a given job and for this reason has not been considered seriously.
The swept-back arrangement exemplified by the various airplanes previously illustrated and described seems to offer the best configuration for a materialization of our all-wing ideal. It can be balanced satisfactorily within quite wide ranges of sweepback, utilizing almost all available volume within the wing for storage of useful load items. It seems to fly satisfactorily in many different configurations and the arrangement is such that large payloads can be carried virtually over the C.G., with the weight empty items so distributed as to cause little variation in C.G. position between the fully loaded and empty conditions.
As has been pointed out previously, the permissible range of C.G. location is not overly critical in this type of airplane. It is, nevertheless, of great advantage to be able to load the airplane almost at will, without concern as to how the useful load is disposed and the swept-back configuration lends itself most suitably to such loading.
In the case of the XB-35, the useful load, consisting largely of bombs and fuel, can be readily disposed in suitable position about the C.G. While some fuel is located well forward and other fuel well aft of the desired C.G. location, under normal operating conditions the proper balance is readily maintained. In case of failure of one or more engines, it is necessary to pump the fuel from unused tanks to those supplying the remaining engines, but a simple manifolding system provides this facility.
Based on a great many studies of various types and applications of the all-wing principle, some practical limitations may be approximately defined. Where very dense (high specific gravity) payloads are contemplated, such as warheads or similar munitions, quite small units are practical, as demonstrated by the all-wing buzz bombs to which reference has been made. Medium-sized units having a span of perhaps 100 ft. and a gross weight of 50,000 to 60,000 lb., appear entirely practical for medium bombers and freighters. Here again the density of the useful load, both in payload and fuel, is comparatively high.
Airplanes designed to carry people need the largest volume of all. Even individual reclining chair accommodations require a minimum space of perhaps 40 cubic ft. per passenger, which is a density of only about 5 lb. per cubic ft. This is one-half to one-quarter the density of typical air cargo, and only 4 percent or 5 percent of the density of a warhead.
It may be concluded, then, that the all-wing design is immediately applicable and practical for a number of military and cargo-carrying versions, and that the passenger-carrying aircraft are likely to be of rather large size and, in the immediate future at least, will provide only comfortable seating instead of the more luxurious appurtenances associated with long-range ocean travel.
An airplane of the XB-35 configuration and size can carry 50 passengers in comfort in the existing airfoil envelope with adequate headroom for all, and with vision forward through the floor, and upward if desired. Passenger vision in a flying wing may be more satisfactory than in conventional types if we get used to the idea of forward vision rather than that provided by side windows. The really interesting views are likely to be forward and downward rather than to the side. An airplane like the XB-35 will have cargo space for 40,000 to 50,000 lb. of air freight at a density of 10 to 15 lb. per cubic ft., in addition to the necessary crew and space for 50 passengers.
Turning now to future possibilities, it seems that considerable further aerodynamic refinement can be made over that already accomplished in all-wing types. Particularly if turbojets are used as the motive power, the minimum parasite drag may be reduced to .008 or less. This value is obtained by subtracting the drag of propeller shaft housings, gun turrets and other military protuberances from the XB-35 configuration and assuming an improved degree of aerodynamic smoothness of the airfoil section. Boundary layer removal and the use of somewhat thinner wing sections may further appreciably reduce this figure.
A maximum trimmed lift coefficient 1.9 for the all-wing configuration seems attainable by methods already suggested and possibly may be further increased by judicious use of boundary layer control in combination with turbojet power plants. It is our opinion that the ratio of C1max to Cd min may be increased to a value of 235 within the not-too-distant future from our present actual achievement of about 130. In contrast, the years of intensive development of the conventional types already passed promise an improvement of less magnitude within a comparable time. In our judgment a trimmed maximum lift of 2.8 vs. a minimum drag of .020 seems reasonable to expect for large, long-range transport and bombardment aircraft of conventional type.
These estimates are, of course, completely arbitrary and controversial. However, if one cares to assume their validity, the following conclusions may be reached. . . . The total minimum profile drag of the all-wing airplane . . . will be from 40 percent to 59 percent [of the drag of the conventional plane]. The power required by the all-wing to maintain the same cruising speed as the conventional will be from 70 percent to 80 percent and, conversely, the maximum range of the all-wing, at the cruising speed of the conventional airplane, will be 143 percent to 125 percent. The maximum range of the all-wing airplane at its best cruising speed will be 158 percent to 130 percent of the conventional, and the most economic speed will be from 125 percent to 115 percent faster.
Under high speed conditions corresponding to full power of reciprocating, turboprop or turbojet engines, where the induced drag is assumed to be 20 percent and the parasite drag 80 percent of the total, the power required to drive the all-wing airplane at the speed of the conventional airplane will be 52 percent to 67 percent and, conversely, the range will be 192 percent to 149 percent of the conventional airplane. The maximum speed of the all-wing airplane at comparable powers will be 124 percent to 114 percent of its conventional counterpart.
Different assumptions of comparative maximum lift and minimum drag values can be made to suit individual opinion, but it is believed that any reasonable assumptions will always result in an advantage to the all-wing configuration of such magnitude as to fully warrant whatever trials and tribulations may be associated with its development. . . .
It is gratifying to those of us who have been working on all-wing projects for years to recognize the increased interest in the type evidenced in Germany toward the end of the war, and more particularly in England and Canada in recent years. For many years we received scant encouragement and often seriously questioned our own judgment, as well as our ability to achieve a successful solution to the many problems involved in the development of this type. The goals and rewards have always seemed well worth attainment, however, and I believe accomplishments to date have justified the effort required.