Al started with a little background about himself. He built the usual balsa models when he was young and about the time he should have gotten into radio control models a new movement was beginning in Southern California. This resulted he his first bamboo and plastic hang glider, which his father quickly cut up into six foot chunks. However, his father did go out and buy some aluminum tubing and built a "real" hang glider from an original set of FlexiFlier plans. Al flew hang gliders until about 1976 when he moved into sailing, sailplaning and college. After college he went to work for NASA where he is today.
The paper he was presenting today was not solely his work. It was a collaboration with David Lednicer as a co-author. They had been friends for a while before discovering they were both flying wing nuts and took the opportunity of the flying wing symposium to put together this paper.
As an introduction, Al commented that the number one thing you need to think about when you have a conventional airplane is that you have a tail that serves one function and wing that serves another. When you put these elements together you end up with having to make compromises. These are in performance, stability, control and structure, or how do you put all the pieces into one workable unit and what is the configuration you are left to work with. There are things like planks, "boards", swept (fore or aft) wings, deltas, etc. He went on to present a short chronology of wing designs by Horten, Lippisch, Northrop and the SB-13 as a recent example, just to lay some groundwork for the rest of the paper.
The first compromise to be discussed is performance. It's really important to understand profile drag. You look at the polar for lift to drag; the drag bucket and minimum drag. He showed an example of a laminar drag bucket that has a point where drag drops off sharply indicating there are significant amounts of laminar flow developing. The new computer airfoil design programs can come up with larger drag buckets, in fact you can design them so the entire operating range is within the bucket. However, you have to know what that operating range is and be very careful in how you select it, because if you get outside of what the airfoil can handle you can face severe problems and their consequences.
A high L/D airfoil may not be the best airfoil for your airplane, since there are other compromises that come along which can kind of contaminate what you have to do. These are spanload and trim issues. Next you look for the pitching moment and there are two ways you can go with this. One is with a symmetrical airfoil which has no pitching moment, but somehow you have to trim it to produce positive lift. However, if no pitching occurs then you have a zero lift airfoil which is not very useful. The Horten and Lippisch approaches were to use cambered and reflexed airfoils, the reflex to help control the pitching moment caused by the camber. The good news is you end up with a low pitching moment and it happens at some positive lift coefficient which is very beneficial.
The next thing to look at in airfoil selection is skin friction. Laminar flow is usually a good thing and turbulent flow is sometimes good and bad. You want to avoid laminar separation and minimize the amount of turbulent flow you have to have. This gets into boundary layer control issues and most aircraft we see have natural transition points on the wing.
He then moved on to discuss interference drag which is hard to understand since it is very configuration dependent. This is drag created at corners and other intersections of surfaces like winglets or the fuselage. The air flow doesn't just flow through the corner, but rather it builds up a circulation region. This is putting energy into the air to make it use this swirling motion and that energy is coming from the vehicle that is producing the flow and this is drag.
There is another type of drag created by lift, which is induced drag. One way one to look at this is through the lifting line theory put forth by Ludwig Prandtl. He thought you could describe a wing as a series of vortices. Air goes over the top faster and along the bottom slower and along with this you have the motion of the airfoil through the air. But given this condition, if you remove the motion of the airfoil through the air, what do you have left over? It turns out what you have is this faster motion going forward over the top, but the motion on the bottom is going slower so you subtract it out as a constant, so you a result you can think of as a little bit of motion going backwards. This is called circulation and it is very difficult to calculate, but there are some approximations which help to arrive at a solution. One of these was Schrenk's approximation since it accommodated taper and twist of a normal wing and it would also compare control surface deflections. The one thing it didn't do for you was the effect of sweep. This came up later very strongly for the Hortens because they found with their swept wings they couldn't find everything that was going on.
The affect of sweep was called the middle effect by the Hortens and it influenced their designs over the years. With more modern techniques available through the introduction of digital computers in the '60s and '70s, circulation calculations can be done much more quickly eliminating the need for approximations. Computational fluid dynamics (CFD) has now been introduced, but Al commented that, although not readily advertised, it is still an approximation of what is actually going on, especially in the turbulent flow region. You can't describe the flow motion of the air in a turbulent region due do its chaotic characteristics and the need to average it all out.
Still talking about performance, Al moved on to lift and spanloading. Ludwig Prandtl said if you can describe what this induced loss can be , then what is the minimum induced loss you can get, or the optimum solution? Elliptical spanloading was found to be the minimum induced drag that could be obtained and this was the accepted theory for almost 80 years. Some recent developments have shown that pure elliptical is not really quite the optimum due to the effect of drag on the spanloading. It can be calculated, but it is completely configuration dependent, however, for all practical purposes they look elliptical and elliptical is still optimum.
The Hortens came up with a bell shaped spanload distribution. The bell curve shows a download at the wingtips but a lot of lift near the centerline and this has a drag penalty associated with it. The further you get from the elliptical distribution the more drag penalty you have to incur, so you have to consider the complete configuration. If you have a conventional design with a tail, it has to have a download in order to trim the airplane, and that download has to be included in the induced drag calculations. It can be compensated for by the designer by looking at the far field wake and how it rolls up and whether it has any discontinuities that can cause additional drag. These discontinuities caused by control surfaces and other design imperfections need to be minimized to achieve the lowest drag component.
Sweep in the flying wing is analogous to a tail in that it allows for trimming the aircraft. It also allows for adjustments to the dampening as described by Irv Culver about 10 years ago. Sweep also allows for putting flaps on the aircraft which can be used for increasing the lift, but you need to be able to trim for the effects of the flap during deployment. There are three ways to go with this, one of them not being very desirable. This is the adverse flap affect where the nose goes over in a tuck as the flaps are put down and is one of the problems that Northrop never quite overcame in his designs. Al related a story about Bob Hoover trying to takeoff in the N9-M on Edwards 12,000' runway with the flaps down. The aircraft just wouldn't come off the runway so he thought maybe he had some type of trim problem and retracted the flaps in an effort to solve it. The instant the flaps came up the aircraft leaped into the air, so obviously there was a trim problem associated with the flaps and the overall wing design.
In modern times, designers have figured out you can actually use flaps on a swept flying wing to your advantage. You can make the flap work as a trim device, and Ilan Kroo integrated them in a way that the nose goes up and the aircraft slows when the flaps are put down. This can be seen in the Bright Star SWIFT. Conversely, you should be able to make it balance out so that you get no pitching moment as in the Flair 30. This is an example of having a clear set of criteria in mind before starting the design of the aircraft.
Al then moved on to stability, not stability AND control. He joked about a guy in his office saying it should be stability OR control since you can't have both. He went on to describe the differences between static stability, neutral stability and dynamic stability. You can have a dynamically stable system where there is enough damping to remove any oscillations that may occur and return the aircraft to a trimmed condition. You can have neutral stability where the oscillation doesn't dampen but it also doesn't grow and get worse over time. It's the dynamic instability that gives you the biggest headache since you can be statically stable and end up with something like flutter and have the whole thing come apart in very short order.
He went on to cover the five classical modes of stability, the first two being longitudinal. There is the short period (Karl Nickel calls it pecking) where when you put in a pulse the aircraft's angle of attack tries to come back. You cannot afford to have dynamic instability here since there is no reaction time to correct the condition before it might become destructive. Then there is the phugoid type which lasts a longer time, maybe up to one or two minutes. It can be dynamically unstable since you have plenty of time to recognize the condition and correct for it before it becomes a real problem.
There are three directional modes. The first is roll control which is the result of a standard roll input that stabilizes into a constant velocity movement. Then there is dutch roll, also known as the dihedral affect, where the aircraft gets hit by a gust and rolls back. This can be damped or undamped and may have more affect on flying wings, such as was seen on the Northrop wings. They weren't really unstable, but rather they had a limit cycle that means the oscillation would build up to a certain size and then just stay there. The last mode is spiral which can be stable or unstable. There are a lot of aircraft that are unstable in this mode, but because it builds up so slowly it is controllable and not a real problem, like in a Cessna 172. The Horten III was spirally stable as demonstrated when it went into the clouds during a contest at Rhön. The pilots (Scheidhauer and Blech) simply held the stick in a constant position and the aircraft continued in a controlled manner until coming out of the clouds.
For the sailplane pilot there are two new modes that need to be considered. These concern the attachment of the tow rope which affects both the longitudinal and lateral modes. He noted these modes had not really been described until about five years ago and there still is no physical flight data for them. There has been some ground testing and measurements taken, but flight data is just now beginning to be gathered for analysis.
Staying with stability, he went on to talk about linearity. This is where the restoring force is proportional to the displacement. Near the outer, ragged edge of the envelope you can get non-linearity, like stalls. There are fixes for some of these non-linearity's so you can get the aircraft to work like you want it too. Devices like fences, flaps, vortex generators and turbulators can be placed on the airfoil to adjust for the problem. Horten put large wing fences on the H Xa to help the elevon control authority.
Al then moved on to the area of control. He related this to making crosswind landings where you would need a 3-axis control system. The Hortens used a blended surface where they had multiple trailing edge surfaces that all moved in pitch or roll, but they didn't move proportionally to each other. However, what you notice about the system is that no matter what trim angle of attack you use all the control surfaces have some deflection since you can't choose a trim angle of attack where they are all at zero when tested (the aircraft was out of rig). Northrop was an airfoil aerodynamicist and what he did was use the middle surfaces for pitch and roll, and the outboard surfaces trimmed only in pitch and split to act as drag rudders. If the stick is pulled all the way back what you have done is induce mechanical washout in the middle section and washin at the tips. This is a good recipe for loosing the tip and having a stall/spin problem. One of the solutions was to try and put slots out towards the tips, which helped the stall problem but not that of spin. However, when the airflow over the slots separates it does so with a lot of hysteresis and a small decrease in angle of attack doesn't result in airflow reattachment. It takes a large change in the angle of attack to get the flow to reattach. This was a characteristic of the type of slots that he used. The good news was, of course that it helped the stall.
The next thing on his agenda was adverse yaw. As most sailplane pilots know this is the effect you get when inputting a roll to right and having the nose move towards the left because of the aileron's deflection. Horten had postulated that if you use a bell shaped lift distribution you wouldn't have this problem, and this has been proven to be the case. This is because the tips are loaded down so you end up with a little bit of pro-verse yaw. Another solution is to put verticals on the surface, but now you have additional profile drag from the surface and interference drag from the corners and joints.
Another thing you can encounter is control force reversal. Northrop overcame this problem by using a hydraulic control system. Horten never seemed to have this problem throughout his designs, but Al didn't have an answer as to why this was the case. Horten had obviously found something that worked, but apparently it's not discussed or defined in his works.
He now moved on to trim, which will also have an effect on the drag polar performance. Here he was working from information in Karl Nickel's book. If you have a stable airplane, such as a flying plank, when you deflect the elevator down the drag polar goes up along with the lift coefficient, except you are trimming the nose down. These reactions are one of the reasons flying wings have performance problems. You can have a unstable aircraft that can be made to perform in the better part of the polar by using an artificial control system (active flight controls). Ilan Kroo has done this with an RC model that was 8% unstable, but one of the other problems he encountered was getting actuators that were fast enough. One of the advantages of this type of control system is that you can make the lift distribution what you want it be through control surface movements unavailable with a manual control system.
Now he moved into the area of structures. What can you build versus what to do you want aerodynamically. Ideally, in order to carry the load, you want something that has a thick center section and thin tips. This has adverse affects aerodynamically on airfoil thickness and chord and the stall comes sooner due to the thin tips. Taper ratio is one method for helping solve this problem and Nickel and Wohlfarht favor a low taper design since you can get very close to an elliptical spanload distribution. Obviously Horten favored the high taper ratio, but this has it problems with tip stall that need to be overcome in some other fashion. An elliptical planform is another way of controlling the taper ratio.
Ideally you want to build a spanloader where the local lift distribution carries the load as much as possible, in other words, put the load where the lift is. The Hortens understood this concept when they built some cargo carrying gliders where the ammunition storage bays were evenly distributed across the span. It ended up flying at nearly three times its gross weight due to this distribution method.
Once you have your spanloader, the next issue becomes one of the spar thickness ratio. There is some evidence that you can get energy extraction from a very flexible spar, flexible in bending not torsion, when passing through gusts. Then you have the controversy between monocoque or space frame construction and whether or not the skin or the spar carries the load. Location of the fore and aft spars are also a function of the aerodynamic design and can create a compromise between low or high aspect ratio wings depending on where the pilot is located in relation to the spars.
Now it is all coming together, but it is also where the headaches begin, performance. You have to pick a spanloader, elliptical or bell, but most people don't think about picking one or the other since they probably already have one in mind before starting their project. If you use elliptical then you use taper and twist at your design points to control the spanloading. This elliptical spanloading causes adverse yaw in roll, because you have a fairly large load out at the tip as soon as you deflect the aileron to lift the wing. It increases the spanload locally, increasing the induced drag and you end up going the wrong way. The first thought is to put verticals like winglets which can be good things, some of the time. Of course you now have the profile, induced and interference drag mentioned earlier and you cannot control the angle of the winglet other than the toe-in. This toe-in angle corresponds to a certain spanload and lift coefficient, and this should coincide with the taper and twist design points. For sailplanes this becomes more difficult since they operate over a wide speed range throughout the flight, versus something like a passenger jet that has them set for optimum performance at cruise where the spend most of there time.
Al felt that Horten, with his bell shaped spanloader, was willing to suffer some of the induced drag to prevent adverse yaw without having vertical surfaces, a design compromise. Philosophically, he believed the Horten sat down and figured this would be more advantageous than having verticals. When Horten received his Ph.D., his teacher was Ludwig Prandtl who came up with the elliptical spanload distribution, yet Horten threw out the idea and proceeded with his own design theories.
One of the things Al felt didn't happen during the early years of flying wing development was a cross-fertilization of ideas between the various designers. There wasn't a great deal of discussion between the designers, and this was partly due to the large distances separating them and the lack of quick transportation. There are compromises to be made and the designer has to understand this right from the start. For instance, Northrop used linear twist at certain design points and straight lined everything in-between which was very simple and easy. He used a symmetrical airfoil which caused most of his trim problems, but it was laminar flow with zero pitching moment at zero lift.
There was another issue on the Northrop designs concerning powerplant stability. The propellers on the XB-35 were a stabilizing influence on the design since they were aft of the center of gravity. On the other hand, the jet engine intakes in the leading edge of the YB -49 caused a bending of the airflow forward of the CG and created a destabilizing influence. Therefore, Northrop had to put the vertical tails on the YB-49 to help recover some of the directional stability. Some people said the verticals were to replace the lost area of the propeller shaft housings, but Al indicated the verticals contained significantly more area then what was lost from the nacelles.
Al felt that John Northrop fully understood the concept of spanloading since he located bomb bays along the wing out to the aileron breaks . He also put some heavy electric gun turrets far out on the wing. He used a monocoque. stressed skinned construction which was a beautiful design.
In contrast, the later model Horten designs with their bell shape lift distribution were giving away performance and drag, but had minimal adverse yaw. The question is whether or not this was a good compromise, and Horten thought it was since he wouldn't have to use verticals. Here is where Al wished Horten and Northrop could have gotten together and discussed airfoils. Horten didn't have access to the same type of airfoils as Northrop, and therefore use Goettingen models which were not analytically designed like the Northrop laminar flow sections. Horten did try to use a P-51 laminar flow airfoil on the H IVb which subsequently spun and crashed killing the pilot. The Mustang airfoil was designed for higher speeds and Reynolds numbers than a sailplane, therefore, were not compatible. This experience turned Horten off to laminar airfoils for the rest of his career, mainly because he didn't understand how they worked. Had he been able to sit down with Northrop, it might have been a different story. Horten, in turn, might have been able to help Northrop with flight control selection and configuration, both designers thereby benefiting from the exchange.
Al went on to talk about the SB-13. Although it performs well, he noted it had some handling problems. They used an optimum modified elliptical span at their design point since they included winglets. It had low taper with twist for spanload control of the lift distribution. The winglets also used rudders which helped in modifying the design point and used a large radius in designing the joining point to minimize interference drag. Although the control mechanisms are very complicated, they allow for very good control surface selection to optimize performance. In order to overcome any flutter problems they built up a unidirectional, monolithic carbon fiber spar (one piece). This controlled the bending moment and then they used cross-plys in the skin to provide the necessary torsional stiffness. The results of all this stiffness to control possible flutter gave them an overstrength spar that tested at 16.5 Gs without failure.
At this point Al talked a little bit about the blended wing body aircraft project started by NASA Langley a few years ago. This is sort of how Al got interested in flying wings and spanloader aircraft. He covered the problem of how big the span has to be in order to have sufficient airfoil depth and chord to handle up to 800 passengers. One way Douglas overcame the problem was to have a large center section that tapered quickly to winglet tips. As he explained earlier this type of design leads to tip stalls, but in this case the use of active flight control systems would be able to overcome the problem.
One of the unusual things about Douglas' design concept was the use of boundary layer control to provide the engine air. In this way almost the whole wing's surface becomes an engine inlet (sucking air through a perforated skin) while at the same time maintaining the laminar layer over a larger portion of the wing. He mentioned the problem of airport infrastructure and its impact on the design considerations since the passenger gate systems can only handle aircraft with a limited amount of span. Then there is passenger acceptance if you try something like folding the wings to keep the span within the desired gate limitations. Ilan Kroo at Stanford has been involved in the project and built two different RC models for testing, one of which is about 23' in span and will be instrumented for future testing. One of the bigger problems that may need to be overcome is the pressurization cycling and, how long a structure that is not a tube will last.
The thing that Al wanted us to get from this presentation was that you've got integrated systems and everything is a compromise. It could be boundary layer control instead of engine inlets, or what you do for spanloading. Optimal performance is not always possible, and perhaps not always desirable, and sometimes you have to compromise on stability or the control system. For control systems, he felt the Horten designs offered some of the best combinations and that Horten really knew what he was doing in this respect.
Upon completing the formal part of the presentation, he showed some color slides of computer aided pictures of pressure distribution, skin
friction coefficient and laminar flow on several different types of flying wings, like the N9-M and Horten IV (#25) which showed the
middle effect mentioned earlier (see note below). He also showed some slides of what is happening in Argentina at this time in trying to restore the Horten H IVc Urubu, and a few historical shots of Horten's work while in Argentina.