Portal de Periódicos da CAPES

Flight-Testing Stability and Controllability of Otto Lilienthal’s Monoplane Design from 1893

2019; American Institute of Aeronautics and Astronautics; Volume: 56; Issue: 4 Linguagem: Inglês

10.2514/1.c035399

1533-3868

Markus Raffel, Felix Wienke, Andreas Dillmann,

Advanced Aircraft Design and Technologies

Open AccessHistory of Key TechnologysFlight-Testing Stability and Controllability of Otto Lilienthal’s Monoplane Design from 1893Markus Raffel, Felix Wienke and Andreas DillmannMarkus RaffelDLR, German Aerospace Center, D-37075 Göttingen, GermanySearch for more papers by this author, Felix WienkeDLR, German Aerospace Center, D-37075 Göttingen, GermanySearch for more papers by this author and Andreas DillmannDLR, German Aerospace Center, D-37075 Göttingen, GermanySearch for more papers by this authorPublished Online:16 Jun 2019https://doi.org/10.2514/1.C035399SectionsRead Now ToolsAdd to favoritesDownload citationTrack citations ShareShare onFacebookTwitterLinked InRedditEmail AboutNomenclatureAfflight altitude; 0≜ to straight legs; feet on grounda0lift slope in lift polar diagramCDdrag coefficient of aircraftCLlift coefficient of aircraftCM,aclongitudinal moment coefficient about aerodynamic centerCM,cglongitudinal moment coefficient about center of gravityDdrag of aircraftLlift of aircraftL′lateral moment about the aircraft’s centerlineMaclongitudinal moment about aerodynamic centerMcglongitudinal moment about center of gravityQpitch rateq∞dynamic pressurettimeV∞freestream velocity relative to aircraftαaircraft’s geometric angle of attack; 0≜ vertical main frameαeaircraft’s trim angle of attackα0aircraft’s absolute angle of attack; 0≜ zero liftθgeometric tail plane angle of attackρair densityI. IntroductionIn 1889, Otto Lilienthal published his book titled Birdflight as the Basis of Aviation [1], describing his experiments proving the concept of curved wings and providing the best available instructions of that time on how to design a winged aircraft. Starting in 1891, Otto Lilienthal made the first, documented, successful flights on a heavier-than-air man-carrying aircraft under the control of a human being. In 1893, he patented the world’s first production aircraft, called the “Normalsegelapparat,” which was a monoplane glider, of which he sold at least nine machines to various places in Europe and America. He performed more than 2000 successful flights, leading to his aerodynamic data and flight reports being circulated around the world. Samuel Langley as well as Octave Chanute and many others corresponded with Lilienthal. Chanute, author of the other most influential book of that time [2]* and later a friend of the Wright brothers, followed Lilienthal’s approach to carefully perform flight tests; with that, he led several younger men to successful flight performances in 1896 and beyond. Wilbur Wright wrote in his article [3]† about Otto Lilienthal: “… he was without question the greatest of the precursors, and the world owes to him a great debt.” According to the patent description, the monoplane glider falls under the category covered by Federal Aviatian Regulations Part 103 [4], which is described in more detail in the Weight-Shift Control Aircraft Flying Handbook (FAA-H-8083-5) of the Federal Aviation Administration [5]. Weight-shift control aircraft are still being flown, safely and successfully, by many thousands of pilots around the world. However, after Otto Lilienthal’s fatal accident in 1896, very little has been known about later attempts to fly this aircraft.II. Glider Dimensions During Wind-Tunnel and Flight TestsWind-tunnel tests gave new insights into the performance, trim state, flight stability, and influence of the permeability of the fabric, which has been woven on an original loom using a formula that was developed based on a careful analysis of fabric taken from an original glider wing. Two parameters that depend on the weight, size, and fitness of the test pilot are the wing area and the pilot’s position in the glider. Their selection was motivated by the analysis of the wind-tunnel data presented in the following. However, in order to focus on the findings gained during the flight tests, only the two diagrams, which helped determining the best possible glider dimensions, will be discussed here.The commonly used classification of Lilienthal’s gliders was introduced by Nitsch ([6] pp. 163–164). According to him the so called “Normalsegelapparat” (“standard soaring apparatus”) has a wingspan of about 6.7 m and a wing area of 13.2 m2. According to Lilienthal’s descriptions, larger dimensions lead to aircraft that are difficult to control in wind gusts. The wind-tunnel model used in the 2017 full-scale wind-tunnel tests at the largest European wind tunnel, the DNW-LLF, had a span width of 6.7 m and a surface area of 13.2 m2. Its main frame location differed slightly from the patent‡ drawing [7] by some centimeters, which has been found to be of great importance to the pilot’s required posture for stable flight. A freestream velocity of 11.5 m/s and an aircraft geometric angle of attack of α=5.82 deg (α=0 deg corresponds to a vertical main frame orientation) leads to sufficient lift to lift the glider (25 kg) plus a person of Otto Lilienthal’s weight (≜80 kg) (see Fig. 1).Fig. 1 Lift coefficient versus geometric angle of attack measured using the wind-tunnel glider at 11.5 m/s.It was discovered during the force and moment measurements that the glider built for the wind-tunnel experiments was difficult to trim. It tended to be tail heavy, even for a small negative angle θ of the horizontal tail plane. Additional tethered flights in the wind tunnel confirmed these findings. During the wind-tunnel run, the test pilot could not continuously hold his body in the most frontal position because of the approximately 1 min time required to accelerate the wind-tunnel flow. In contrast to the glider’s orientation going downhill, in the wind tunnel, the glider and the freestream velocity were inclined upward by approximately 15 deg, which made the pilot’s most forward-leaning posture even more arduous. The tethered flight in the DNW-LLF confirmed that an additional 8.2 kg of trim weight at the front of the cockpit was required to fly the wind-tunnel glider version stably in a horizontal setting. This nosedown force was applied through a ballast cord at the cockpit front pulling downward (see Fig. 2).Fig. 2 Steady tethered flight with an externally applied nosedown moment correction, corresponding to an 8.2 kg trim ballast at the aircraft’s front.The larger weight and body size of the test pilot (see Table 1) led to the following dimensions of the outdoor glider, which later was used for flight testing: the wingspan was chosen to be 7 m and the maximum chord length was 2.5 m, resulting in a wing surface area of 15.6 m2. The fact that the main frame location with respect to the wing area center differed on the wind-tunnel model and the patent drawings (see Fig. 3), and that the wind-tunnel model required a ballast force to reach steady horizontal tethered flight, led to a main frame position of the outdoor glider that was 134 mm further ahead of that of the wind-tunnel glider. The main frame position limits the pilot’s front position, and therefore influences the trim state and static margin and/or the posture required to move the center of gravity in front of the aerodynamic center, which is necessary to obtain a stable and trimmed flight. It is well known that weight-shift control aircraft tend to handle best with small static margins for obvious reasons. However, small static margins also imply that just a few centimeters displacement of the relative center of gravity changes the handling qualities of the aircraft from “easy to control” to “impossible to fly.”Fig. 3 Comparison of wing surface geometry and size at identical main frame positions of three aircraft.Wing surface geometries and sizes at identical main frame positions of the three aircraft versions are depicted in Fig. 3. Instead of the aerodynamic center, the wing’s area center has been chosen as the reference point for the three glider versions because the aerodynamic center was measured only for the wind-tunnel version of glider. However, it is assumed that the wing surface center and aerodynamic center position are displaced at very similar quantities for all three glider versions because of their geometric similarity.Following the wind-tunnel experiments, measurements of the longitudinal center-of-gravity location of the wind-tunnel glider were performed for each of the pilot’s various postures. The evaluation of these data, together with the force and moment data, showed the same deficit of the static margin that was present on the wind-tunnel glider. The necessity of the trim ballast can be explained by the fact that the main frame location, and therefore the pilot’s center of gravity, differed between the wind-tunnel model and the patent drawings depicted in Fig. 3 (top). A detailed analysis of a vertical projection of the glider wings and the location of the glider main frame showed a distance between the wing surface area center and the main frame at the patented glider of 452 mm, as compared to only 348 mm for the wind-tunnel model (see Table 2). A correction to the pilot position within the glider was applied to the measured data, and the results showed that the patented glider could be flown stably at moderate negative tail plane angles without ballast, given a pilot posture as depicted in Fig. 4c (trimmed).Fig. 4 Pilot’s postures during a) start, b) flight middle posture, c) flight trimmed posture, and d) landing.III. Familiarization of the Test Pilot: Tethered Flights and Winch-Supported FlightsThe first thing the test pilot had to learn when attempting to fly the monoplane glider was longitudinal and lateral control by weight shifting. Directional control is only performed indirectly by lateral control. Otto Lilienthal gave the following advice for longitudinal control [2]:The first rule is to keep your legs well extended toward the front, and in landing to throw the upper part of the body backward, so that the front edge raises itself and thus checks the motion, as may be seen whenever a crow alig hts.This leads to the most forward center-of-gravity position during takeoff and the most backward position during landing. They are described by the sketches of postures shown in Fig. 4, which were drawn according to the figures given in the patent drawing by Otto Lilienthal.Lilienthal advised to always use the following procedure to start [2]:The starting and landing must be done exactly against the wind. The fixed vertical rudder will keep the apparatus exactly in the wind when in a state of rest. The horizontal rudder keeps the apparatus from tipping over forward, a thing that arched surfaces are inclined to do.In our case, it was decided to start with tethered flights on a 5×5 m2 platform towed by a passenger car. This allowed for frequent training at reduced risk and costs. The glider as well as the pilot were attached to the platform corners with safety bonds (pilot) and wires (glider), as depicted in Fig. 5. The pilot was wearing a harness connecting him to four safety bonds. The glider had eyebolts anchored to the lower ends of the main frame crossbars through which the steel wires ran from the front corners of the platform to the rear corners on each side. This arrangement limited the glider’s altitude but allowed for directional motion. Two ropes were attached to the car in order to compensate for the aircraft’s drag and to pull the glider forward. Given a glider altitude of less than half of a 1 m above the platform and a lateral position in the center of the platform, the ropes fastened to the platform corners were loose. The pilot could therefore test the influence of weight shifting in longitudinal direction, leading to the postures depicted in Fig. 4.Fig. 5 Practicing at “zero ground speed,” limited height, and limited control authority: platform tests.The two pull ropes were attached to the ends of the glider main frame crossbar by another four ropes, which intersected at a central attachment point 0.7 m in front of the main frame. Therefore, the longitudinal stability was improved and the pilot postures had a reduced influence on the pitching motion.Heading and wind directions did not match during the tethered flight tests because they were conducted on an airfield with just one runway. Therefore, the pilot had to compensate for a crosswind component of up to 5 m/s (see Table 3).The experiences gathered during the tethered flights on the platform were required for the next phase of winch-supported tests. The fact that these flights were performed against the wind (see Table 4) allowed for flights at reduced ground speed (see Fig. 6). The winch tests had to be conducted without the safety wires that, during the platform tests, not only limited the flight altitude but also the roll motion of the glider. The pilot’s lateral control skills became the key issue in cases of moderate gusts on the one hand, as well as in cases of stall that could lead to asymmetric lift on the other hand. (Asymmetric lift during stall occurs especially at low pitch rates Q or initial high attack angles. The influence of the leading-edge vortices inherent to dynamic stall at flare landings will be described later.) Lilienthal described the procedure of lateral control (see Fig. 7) and related difficulties as follows [2]:Fig. 6 Winch speed versus distance traveled.Fig. 7 Practicing lateral control with limited longitudinal control authority: winch tests.The following mistake is to be particularly avoided. The experimenter is soaring in the air and feels himself suddenly raised by the wind, but unequally, as is usually the case--for instance, the left wing more than the right. The inclined position forces him toward the right. The beginner involuntarily stretches his legs to the right, because he foresees that he will strike the earth on the right hand. The result is that the right wing, which is already lower, is loaded still more, and flight tends more and more downward and to the right, until the tips of the right wing strike the earth and are broken. For life and limb there is less danger, as the apparatus forms an efficient guard in every direction, which checks the force of the blow. The correct thing to do is always to extend one’s legs toward the wing that is rising, and thus to press it down again. In the beginning this requires some force of will, but this useful movement soon becomes an unconscious one, after we see how surely the wings can be guided this way and be protected from damage.IV. Trimming of the GliderThe winch tests in Germany were performed with an attachment point using a carabiner clasp placed approximately 0.7 m in front of the main frame center. This additional restoring moment increased the longitudinal stability considerably. Thus, the longitudinal control authority was reduced as long as the pulling rope was under tension. The fact that the outdoor glider was still a bit tail heavy had been overlooked because the pitchup tendency became relevant only at low rope tensions when the glider was in descent and the pitchup was intended and supported by the pilot’s landing posture (Fig. 4d). During the free flight in California, though, this needed to be corrected by either a ballast weight or a reduction of wing area ahead of the pilot. The first measure chosen to increase the static margin for a given stabilizer angle and center of gravity was the reduction of wing area in front of the glider by a v-shaped enlargement of the cockpit “window,” which is the gap in the fabric in front of the pilot. Additionally, after some remaining problems with longitudinal control and stability, the wing area at the glider’s front was further reduced. The front ribs of the wings were moved a few centimeters backward, and the fabric along the leading edge of the wing was cut and reattached accordingly.§ The resulting final shape of the glider is depicted in Fig. 3.Figure 8 depicts CM,cg versus a0 for three different pilot postures for the final version of the outdoor glider. The data have been derived from the wind-tunnel force and moment data measured in 2017, making the following assumptions: The lift increases proportionally with the wing area (15.6/13.24).The weight of the pilot (90 kg) is located further forward than during the wind-tunnel tests because of the altered main frame position of the outdoor glider used for flight tests (134 mm).The dashed line in Fig. 8 corresponds to the middle position (Fig. 4b), the dotted and dashed line corresponds to the trimmed position (Fig. 4c), and the solid line corresponds to the start position (Fig. 4a). It should be noted that the dotted–dashed line intersects with the horizontal axis for CM,cg=0 at a0=17.25 deg. This defines the trim angle of attack; the horizontal tail plane angle; and the pilot posture during trimmed, stable, and controllable downhill flights at 11.5 m/s when the systems weight, drag, and lift forces are in equilibrium. The three necessary criteria for longitudinal balance and static stability (namely, the positive moment coefficient at α0=0, the negative moment curve slope, and the trim angle of attack within the flight range) are fulfilled. The measured data correspond very well to the conditions found during the downhill flights, after the start, but before the landing.Fig. 8 Lift coefficients versus geometric angle of attack derived for the outdoor glider at 11.5 m/s (left). Moment coefficient versus absolute angle of attack derived for the outdoor glider (right).After this modification, the glider could easily be controlled in the longitudinal and lateral directions as well as in free downhill flight, as in cases of flights that were initially accelerated featuring a bungee cord. The latter method was used when the wind and slope of the terrain were too weak for pure foot launching. The bungee cord was a well-suited means to increase the available practicing time and made it possible to get airborne at Dockweiler Beach in the vicinity of Los Angeles and at Tres Pinos in the vicinity of Hollister (in California). Due to the fact that this bungee cord was attached to one central hook directly on the main frame, the trim, stability, and control authority was the same with and without the bungee cord. All final flights were performed freely without any ropes attached at Sand City, north of Monterey (in California).The outdoor glider used in the end for the free flights differed in some other minor aspects from the gliders used by Otto Lilienthal and others at his time:The outdoor glider has been coated with white wood glue. This glue behaves physically just like other types of collodion that was used to coat the Lilienthal gliders. After it is applied to human skin, for example, it acts like a sort of second skin in both cases. The monoplane glider was folded many times, sometimes in very rough conditions, but the coating showed no sign of wear. Ongoing wind-tunnel investigations on the influence of the porosity demonstrate that forces and flowfields are the same for different ways of sealing the fabric. So, this is a chemical difference, but it is aerodynamically of little or no relevance.The skids attached to the lower ends of the main frame were, like the test pilot’s kneepads, only in use after mishaps during landing. So, they were also of little influence during flight. Stainless steel wire ropes of 2 mm in diameter and oval compression sleeves were used instead of steel wires and tension locks (patented by Lilienthal himself).Two parallel short bamboo sticks have been mounted on the outdoor glider on both sides of the cockpit to prevent the pilot from leaning too far back. They substitute the backrests that Lilienthal used and are, like the helmet, a further concession to safety considerations. Luckily, neither the helmet nor the bamboo sticks have ever been needed.The arm rests of our outdoor glider were made more primitively. The pilot could lean on them like Lilienthal on his, but he could not lift the glider in the back in order to get the minimum incidence angle for the start at low-wind conditions. During the winch flights and the final downhill flights, this was not necessary because the head wind lifted the rear part of the glider immediately after the first few meters. At low-wind conditions, the pilot used shoulder straps from time to time to lift the tail before start, and thereby facilitate the acceleration.A structural difference between our outdoor glider and the version that was described by Lilienthal in his patents and publications is described in the following. Willow was used only for parts that had to be bent significantly, like the cockpit frame and the vertical and horizontal stabilizers. However, pine wood sticks, which had been bent into shape while being soaked in water for three days, were used instead of willow for the struts of the wings. This increased the weight to 32 kg and might have shifted the center of gravity of the glider a bit toward the rear. Still, it needed to be done due to the test pilot’s weight being approximately 15% greater than Lilienthal’s, and the required lift forces are approximately another 15% higher when flying horizontally (tethered and at the winch rope) than during downhill flights. As for Lilienthal’s gliders, bamboo was used only when parts of a certain length had to be stiff and straight, like the tail.V. Acquirement of Longitudinal Control Skills: Free FlightsWhen practicing on a training hill at Tes Pinos in the vicinity of Hollister (California), the wind tended to be gusty. During long periods over the year, the winds blow from northwest along the valley; but, on the test day, they were dominated by a strong upper western wind leading to flow separation at the western California coast ranges, which created a few occasional rapid changes in local wind strength and direction. A sudden gust in the early flight phase after takeoff created an upward acceleration, which made keeping the pilot’s posture for trimmed condition impossible. The altitude variation could later be reconstructed by correlation-based image processing from the videos (see Fig. 9). A 10th-order polynomial was used to fit the altitude data estimated for each 0.1 s and to derive the rate of climb (or descent, respectively) and the load factor for a time interval of 3.5 s (see Fig. 10).Fig. 9 Stall and poststall behavior.Fig. 10 Altitude, rate of climb/descent, and load factor during the stall event depicted in Fig. 9.The additional inertial forces made it impossible to keep the legs directed straight and forward. The glider was lifted approximately 4 m within 1 s, and the glider’s incidence angle moved from approximately 15 deg downward to 50 deg noseup. As a consequence, the pilot’s weight moved backward within the cockpit. Lilienthal routinely jumped from a roof of more than 4 m high on a routine basis and managed acrobatically to pitch the nose down and avoid roll, even in those vulnerable stalled conditions without lowering his legs. Lilienthal experimented with an actively steered tail plane, flaps, and wing warping, but he intentionally decided to stay with the weight-shifting control because this allowed him to operate the glider reliably during thousands of flights [8]. The many flights that he started by jumping off the building roof at the “flight station” near Steglitz (a suburb of Berlin, Germany) required the start of the glider reliably at zero ground speed. It is hard to imagine that this can be done better using active control means, and thus weight-shift control is still the method of choice for most unpowered foot-launched aircraft. However, foot-launching gliders in gusty wind conditions require a great deal of practice. As a consequence of the pilot’s inability to counteract stall, in Tres Pinos, the glider stopped and fell approximately 4 m like a parachute: the tail plane flipped upward, although no strong nosedown moment occurred, and the wings stayed nearly level. This safety feature of the glider was described in Lilienthal’s U.S. Patent [7] with the words:On the latter is pivoted the tail q in such a manner that it can freely turn upward, but finds downward a point of support on the fixed rudder r. This mode of attaching the tail has the advantage that the tail will have no carrying action when the machine is employed like an ordinary parachute, thereby preventing from turning downward.It should be noted that the tail could not transfer an upward force because it was rigidly coupled with the upper ropes when the horizontal stabilizer acted downward, but it flipped up when a force acted on it upward. The upward flexibility prevents the tail from breaking when it touches the ground during flare landings. It is well understood that these observations cannot substitute an analysis of dynamic stability of the aircraft. However, they indicate that extreme situations can exceed the pilot’s capacity to maintain a posture required for balanced, controlled flight. Due to the pivoted tail plane, stall seems to be less problematic when a pilot behaves cautiously and does not counteract stall in a later stage. Lilienthal’s annual flight reports not only informed others about successful flights but also about occasional failures [9]. One such flight accident described in detail was a rapid dive that occurred after a small change of wing curvature. This can be interpreted in a way that these modifications had an effect on the static stability of the glider. Additionally, it seems likely from the experiences of the authors’ of this article that the required pilot posture to flare, as depicted in Figs. 4b and 4d, will potentially exceed the pilot’s strength in such a situation. The dimensions of the glider used for the free flights corrected geometric imperfections of the wind-tunnel model and increased the static margin to an amount that matched the static margin of the patented glider. The glider has been safely flown after the described incident, but strong wind gusts were more consequently avoided.The final pitch trim was obtained by a moderate negative angle of attack of the horizontal tail plane. As a result, the glider reacted nicely and sensitively to the pilot’s input and could easily be directed against the wind. During several starts on the sand dune, shear winds required counteracting a descending left wing counterintuitively by shifting the pilot’s weight below the right wing. Here, the training during the winch-supported flights paid off; and when the wind lowered one of the wings, the pilot shifted his legs to the other side in an instinctive manner (see Fig. 11). Basically, the control of the roll angle has to be performed like the one of a modern hang glider, but the legs need to travel greater distance sideward in order to create a similar reaction of the glider, due to the smaller weight that is shifted, when repositioning just the legs. In addition to the fact that a modern glider is laterally controlled by the relocation of the pilot’s whole, it also amplifies the roll moment L′ by shifting the keel, and therefore warp the wings accordingly ([5] pp. 3–9). It has to be noted that turns cannot safely be executed while flying at low altitude in the vicinity of a hill. Therefore, the question of whether steep turns can safely be performed remains unanswered by our tests. It is known from Lilienthal’s flight reports that he, who flew much higher, made turns but (still) tried to avoid them because a safe landing has to be conducted against the wind ([10] p. 267).Fig. 11 Preparation, start, and initial flight phases with lateral and longitudinal control.Lilienthal reported that landing requires a similar counterintuitive move, as is the case with turning the glider ([10] p. 266). He reported that the pilot has to move his legs backward to pitch up and decelerate, even if his instincts advise him to have his feet in the front when approaching the ground at higher speeds. However, this depends on the trim of the glider and, in the case of our test flights with the monoplane glider, it is just enough to lean backward with the upper body, and therefore move the weight of the whole body to the rear (see Fig. 12) in contrast to flights with best gliding performance where the pilot is in most forward position (see Figs. 13 and 14) [11]. The only problems that occurred while coordinating the landings at the beginning were the same problems beginners face during their first hang-glider lessons. When initiating the landing too early and too slowly, stall occurs in a way that the flow on the wings separates slowly but massively. As massively separated flow is never steady or two-dimensional, one wing starts sinking earlier than the other and generates more drag at the same time. This leads to an unintended turn at the end of the flight with both Lilienthal’s gliders as well as with any modern weight-shift control aircraft. The trick to landing the Lilienthal glider well is doing this maneuver a little later and at higher pitching rates Q so that the stall occurs dynamically. The dynamic stall vortices along the leading edges of the wings will then force the flow into a two-dimensional condition [12] while creating a short lift overshoot ([13] p. 1]) and an additional pitchup moment. After these lessons were learned, the glider could be flown reliably and steadily for up to 70+ m, and the coordination of gentle straight flare landings were performed routinely.Fig. 12 Straight, controlled downhill flight and landing.Fig. 13 Otto Lilienthal and Paul Beylich: Fliegeberg near Berlin [11] (top). Markus Raffel and Andrew Beem: Sand City near Monterey (bottom).Fig. 14 Preliminary flight tests with Lilienthal’s large biplane.VI. ConclusionsOtto Lilienthal’s patented monoplane glider (Normalsegelapparat) was flown reliably and steadily downhill over distances of up to 72 m. The static margin of the glider as laid out in the patent drawings of Lilienthal can safely be flown by a person with a weight of not more than 80 kg wh