Document
IT' S T IME T O REIN V EN T THE GENERAL A V I A TI O N AIR P LANE R obert Stengel Princeton Universit y Current designs f or general avia t ion airplanes have become obsolete, and avenues for major redesign must be considered. New designs should incorporate recent advances in electronics, aerodynamics, structures, materials, and propulsion. Future airplanes should be optimized to operate satisfactorily in a positive air traffic control environment, to afford safety and comfort for point-to-point transportation that is at least comparable to auto motive travel, and to take advantage of automated manufacturing techniques and high production rates. These requirements have broad implications for airplane design and flying qualities, leading to a concept for the Modern Equipment Genera l A __.viation (ME(3A) airplane. Synergistic improvements in design, production, and operation can provide a much-needed "fresh start" for the general aviation industry and the traveling public. Achieving these goals requires nothing less than the reinvention of the small airplane, providing new opportunities and requirements for research.
Although the term "general aviation" applies to a wide range of aircraft -- from single-engine, propeller-driven, single-seat planes to business jets -- the focus of this presentation is at the low end of the scale. Except as noted, a small four-place airplane is taken as a reference. Nevertheless, the proposed philosophy for new airplane design applies across the entire spectrum of general aviation.
I T 'S T IME TO REI N VE NT T HE
GE N ERAL A VIA T I ON AIRPLA N E
Declining Practical Importance of
Small General Aviation Airplanes
New T echnologies .for Airplane
Systems
MEGA_PI_¢
Goals and Assumptions Characteristics o f a 4-Place ME_3A-Plane Opportunities for R esearch 8 2 D E CL IN I NG PRAC T I C AL IM PORTA NCE OF SMA LL GENE RAL AVIATION AIRPLANES For all pr a ct i c a l purposes, the production o f sm a ll gener a l avi a tion (GA ) airplanes in the United States has come to an end. The production rate for GA airplanes of all categories is about 1500 planes per year, less than 10 percent of what it was a decade ago. A thousand small GA airplanes were produced in 1986, but few of these were made by the manufacturers that formerly could be considered "The Big Three." Exports accounted for 30% of the US production; in the coming year, the General Aviation Manufacturers Association expects a comparable percentage of foreign imports, with a large concentration in the single-engine category. The average age of US small airplanes is 20 years old.
The production of GA airplanes currently has minimal impact on the domestic economy. Although general aviation is said to be a $15 billion, business, only $200 million of that can be attributed to the sale of small airplanes.
As fixed-base-operator income decreases, the idea of converting airfield real estate to condominiums and shopping centers becomes attractive, driving many small airports near major urban areas out of business. Small airports are losing the business of serious travelers, and an increasing number principally serve weekend pilots and flight schools. This trend is particularly detrimental to the national transportation system in view of the heavy congestion at most major airports, and it accelerates the decline in travel by GA airplane.
An ancillary point is that recent growth in commercial air travel has created a shortage of airline pilots. General aviation has long been a spawn- ing ground for airline pilots, but it is not producing enough well-qualified pilots to meet the demand. One consequence is that relatively low-time pilots are flying in the right seats of many commercial aircraft cockpits.
Although important research continues in applicable technology areas such as stall / spin dynamics and aerodynamic flow control, neither the Federal Aviation Administration (FAA) nor the National Aeronautics and Space Administration (NA S A) currently supports substantial research and develop- ment specifically directed at general aviation. If GA airplanes are no longer produced in the US or if GA flying must be perceived as the domain of the hobbyist and the well-to-do alone, it will be increasingly difficult to justify the expenditure of federal funds for its enhancement.
DE CL I NIN G PR ACT I C AL
IM PO RT ANC E OF SM A LL G E N ER AL
AVIATION AIRPLAN E S
Large-Scale Production of Small
Airplanes has Ended
GA Production Rates are Less Than 10% of What They Were a Decade Ago ( about 1500 planes / yr) Small Airplanes Producti o n is Less Than a Thousand per Year Three Maj o r GA Manufacturers Currently Produce Few Small Airplanes
Small Airports are Closing
Neither FAA nor NASA has Any
Substantial GA R&D Initiatives
RE AS ON S F OR T HE DE CL INE There a re many reasons for t he decl i ne in production of small airplanes. While numerous evolutionary changes have been incorporated in GA airplanes, most examples trace their basic designs to the late 1940s, with technologies established a decade or more earlier. Manufacturers find certification costs higher than ever before, and production costs are in an unstable spiral as the number of produced aircraft decreases. Except in special cases, there is a large disparity between the cost of owning and using a GA airplane and the cost of traveling by competing modes. The GA airplane may well be cost-effective in a small-business setting, particularly where a large area must be covered quickly on a regular basis; lacking a business subsidy, few middle-income travelers consider GA transportation affordable.
GA manutacturers a re being found l i able for an increasing number of airplane accidents, and the cost of liability insurance or "self-insurance" has become a significant percentage of the total cost of each new airplane. This is a disincentive not only to the airplane producer but to the potential owner, who must pay the added costs.
While gains have been made, the accident rate for GA aircraft still is substantially higher than that for competing modes. The hazard is comparable to motorcycle riding and somewhat greater than traveling by commercial aircraft, train, or automobile. Air tra ff ic control procedures have become more complex with airline deregulation and increasing commerci a l air travel, and future GA travel is likely to be limited even more in view of productivity and fuel-use considerations. Flying an airplane in poor weather conditions demands a high level of IFR proficiency, something that relatively few GA pilots can achieve and maintain.
Fin a lly, t here i s re a l confusion about the goals of gener a l aviation. Should it be considered as a candidate mode for transporting large numbers of people? Should travel by small airplane be more like sailing to B ermuda or driving to Pittsburgh?
Can general aviation have a major impact on the economy? Should particular classes of general aviation (e.g., air taxis and corporate aircraft vs. personally owned small planes) be singled out for special treatment? Without answers to these and similar questions, general aviation will decline even further.
REASONS FOR TH E D E CLIN E
Slow Incorporation of New
Technologies
High Costs of Certification, Production,
and Operation
High Costs of Liability
High Accident Rates in Comparison to
Other Modes of Transportation
Increasingly Complex Air Traffic
Control Regulations
High Level of Piloting Proficiency
Required for All-Weather Flying
Confusion About GA Goals
NEED F O R A RESURGENT GENERAL AVIATI O N INDUSTRY Gener a l aviat i on provides un i que c a pabilit i es f o r rap i d p oint- to-point travel for small groups of people. It is complementary to the hub-t0-hub and hub-spoke services of the major and feeder airlines. With area navigation and sufficient satellite airports, general aviation airplanes can be routed through under-utilized airspace, avoiding areas of congestion both in the air and on the ground. By diverting significant numbers of travelers from the public carriers, general aviation could actually reduce congestion in the terminal area. Just as automobile travel between two suburbs often is quicker, cheaper, and more efficient than public ground transportation into and out of a neighboring metropolis, general aviation offers a potentially attractive alternative for many trips between points not near large airports.
Impro vi ng rel i ab i l it y and safety are con ti nu in g i ssues in all modes of transportation. No matter what the current level, we always seek to lower the costs of operation, to simplify maintenance, to facilitate on-time performance, and to reduce the risks inherent in travel. In order for new G A airplanes to attract potential buyers, they must provide benefits in comparison to the competition -- used airplanes or new planes produced by other (possibly foreign) manufacturers.
The production of small airplanes could be a multi-billion- dollar business. The two most likely outcomes of not rebuilding the GA industry will be the de facto encouragement of foreign airframe and engine manufacturers to introduce their products to the US market and the loss of additional foreign markets for American products. Yet another business will be converted from a manufactt_ ring to a service industry, with its attendant diminution of technical leadership and long-term economic security.
There is an opportunity -- if not an imperative -- to ask the question, "If we were unfettered by the need to adapt new technologies to old designs in piecemeal fashion, how would we des i gn sma l l ai rplanes?" Or put another way , "Hgw wq old we invent the GA airplane to satisfy the needs of potential users while accounting for the realities of the National Airspace System and pro- viding a reasonable incentive to prospective manufacturers to build such airplanes?"
NEED FO R A RE SU RGENT GENERAL
A V IAT ION INDUS T R Y
Transportation Requirements of the
Public
Continuing Drive for I mproved
Reliabili t y and Safety
Stimulation of Domestic Economy
Increasing Domestic Dependence on Foreign Suppliers L o s t Foreign Marke t Opport u n iti es NEW T ECHNO L OGIES F O R GE N E R A L A V I AT I ON AI R P LA NE SYSTEMS There i s an exciti n g a rr a y o f n ew tech no l o g i es th a t, f o r the m o st part, have n o t been appl i e d t o the pr od u c t io n o f small airplanes. S o me o f them are listed here.
Ele ct r o n ic s - In little more than a decade, the micr o processor has revolutionized many products and services, but it has done little for the GA airplane. It can provide the focal point for a host of major improvements. Together with concurrent advances in sensors, actuators, displays, and external systems, it can spearhead the drive for a new level of performance, reliability, and safety in GA production and operation. At the same time, flight-critical electronics introduce new concerns that must be addressed during design and operation, including guaranteed uninterupted power, lightning protection, mode switching, and complex control logic, all at a far lower cost than is normally associated with avionics equipment.
Modern Manufacturing Techniques - While it is unlikely that production rates would ever approach those of automobiles, much can be learned from advances being made in the automotive industry. Today ' s GA airplanes are essentially custom - made, deriving little benefit from common production - line concepts; however, modern manufacturing techniques emphasize flexibility, using computers , communication networks, and robotic devices (including numerically controlled machines) to perform a wide variety of functions from preliminary design to painting the end product. This flexibility is precisely what is needed in GA airplane production.
S t ru ct ures an d Ma t erials - New materials not only pr o mise direct benefits: they provide an opportunity for redesigning the basic airplane structure. Furthermore , modern objectives such as enhanced crash survivability and lightning protection can be combined with traditional design considerations like weight, air loads, and fatigue through the use of computational analysis.
Composite materials offer strong and lightweight alternatives to aluminum components, although aluminum remains the cost - effective choice for most primary structures.
NE W T E CH N OLO GIE S FOR G ENERAL
AVIATIO N AIRPLANE SY S TEMS
Elec t ronics
M i croprocess o rs Fiber Op t ics In t egra t ed M oti on Sensors High-Flux-Density Electric Motors Electronic Displays Weather Sensors Precise Long-Range Navigation Air Traffic Control Systems
Modem Manufacturing Techniques
Comp ut er- A ide d Desig n C o mp ut er-Integr at ed Man u facturing Robotics
Structures a nd Materials
Composi t es Aluminum A lloys Honeycomb Integrated Structures NE W T ECHN O L O G I E S FOR GENE RAL A VI A TI ON A I R PLA N E SYSTEMS, continued Ae r odyn am ic s There is a widespre a d misconception t h a t we have learned all we will ever know (or need to know) about subsonic aerodynamics. In fact, some rather dramatic breakthroughs have been made in recent years, and it is likely that there is much more to come. While seemingly redundant, three controllable horizontal surfaces for lift, stability, and control (canard, wing, and tail) provide a number of advantages, including reduced wing size, optimization of cruise condition to reduce trim drag, and stall / spin protection.
Control surfaces need not be coupled mechanically as in the past; hence, there is a high level of redundancy that can be put to good use in improved safety margins. It is now realized that a combination of modern surface finishing techniques and shape selection can provide natural laminar flow over large segments of wings and fairings, reducing drag and improving overall performance. New perspectives on wing design suggest that significant gains in lift / drag ratio can be re a lized by reshaping the planform, particularly in the vicinity of the tips, as well as the airfoil.
Propulsion - Great strides have been made in automobile engines, suggesting avenues for improving the reciprocating engines of small airplanes. Computer-controlled electronic ignition, improved combustion chambers, multiport valving, and turbocharging all could contribute to increased safety, reliability, and efficiency. The new engines could be designed to use automobile gasoline with no loss in performance or economy, solving one of the more pernicious problems of operating small planes today. Computational fluid dynamics, new perspectives on laminar flow control, and recent developments in propeller design for human-powered airplanes, the world-circling Voyager , and wind turbines all can further advance propulsive efficiency.
NEW TECHNOLOGIES, continued
Aerodynamics
Three-Surface Longitudinal C o ntrol Control Surface Redundancy Natural Laminar Flow Planform and Wingtip Design
Propulsion
Automotive Ignition and Combustion Technology Propeller Design TH E MO DER N E Q U I PM E N T G E N E RA L A VIA TIO N AI R PL A N E The princip a l objecti v e for the "re i nvented" G A airpl a ne is to provide a viable alternate form of transportation for a large number of air travelers, in much the same way that automobiles provide a desirable alternative to public ground transportation. The goal is not to redefine our notions of existing GA concepts but to redefine the GA concepts themselves, to make general aviation something that it is not today. The goal is to design a new breed of airplanes that really do make private flying a constructive segment of the National Transportation System through the use of modern technology and manufacturing techniques.
S ta rt i ng o ver i n the design o f s m all a i rpl a nes w il l result in s y nerg i stic benefits that would otherwise be unattainable. In effect, the whole design process is "rubberized", identifying desirable attributes in one system, evalu- ating the impacts of these attributes on other systems, and redefining the lat- ter accordingly. For example, if independent, reconfigurable control surfaces are desirable (which they are) but they require uninterrupted electrical power with extremely low mean failure rates (which they do), existing systems for power supply and distribution are unacceptable. A new standard for power system design is mandatory, and such a system will, no doubt, contain redundancies that are not currently considered necessary. There is good reason to believe that the redesign can be achieved, given modern technology and no predetermined requirement to interface with traditional elements.
Furthermore , by designing for the production process and reasonable production rates, costs will be minimized.
It i s essent ia l to take bold steps i n plan ni ng for wh a t amounts to a maj o r overhaul of the National Transportation System. Clearly, the impact of a large increase in small airplane traffic would be great, and without a com- prehensive approach that considers not only production a nd distribution but operation within the confines of the National Airspace System, that impact would be calamitous. There is a lot of unused airspace that could be used safely and efficiently, with virtually no infringement on airline operations. "Back- of-the-envelope" calculations suggest that a million GA airplanes (about five times the current number) would produce a volumetric density that is on the order of one hundred million times smaller than ground traffic density. Still, this space is unusable without positive assurances that airplanes will not interfere with each other. Consequently, the reinvented GA airplane must have a degree of autonomy and compatibility with other airplanes that is not realized in current designs.
T HE MOD E RN EQ U IPMENT G ENE RAL
AVIATION AIRPLANE
Objective:
Fast, Reliable, Safe, Comfortable, Cheap
Transportation for Large Population of
Travelers
Synergis t ic Use of New Technologies
N o N eed for Compatibili t y wi t h Old Technology Design for Low-Cos t Automated Manufac tu ring All-Electric Control Actuation Reciprocating Engine(s)
Planning for GA t o Become a Major
Componen t of t he National
Transportation System
Production Ra t e of 50,000 Airplanes / yr by 2000 Equilibr i um GA Popul at ion of 1,000,000 Airpl a nes by 2015 _AolP L A_I g , c on t in u ed The reinvented GA airplane should be as simple as possible, containing few clever-but-failure-prone mechanisms. It is, for example, preferable to forego the extra aerodynamic efficiency of an intricate flap deployment mechanism in favor of a simply hinged flap with a single rotational degree of freedom. The latter device is less likely to fail, is easy to fix when it does, and provides a backup roll control device. Systems should contain line- replaceable units that are comprehensive in function, individually reliable, easily understood, and easily replaced. Redundancy should be provided where single-string reliability is inadequate; however, as automatic redundancy management in modern flight control systems often grows to dominate software specifications, the airplane should be designed to allow the pilot to do as much redundancy management as possible. This means that failure modes should be forgiving, allowing time for human decision making. In a similar vein, the airplane's flying qualities must be good enough to allow a relatively inexperienced (or not current) pilot to maintain safe control in a wide range of flight conditions, including those that generate high workload in existing airplanes (weather, traffic, etc.).
The reinvented GA airp l ane must be specificall y designed for ATC sys tem compatibility, or else none of the suggested improvements can be realized.
The area navigation system is as important as the wings and engine of this airplane. It must be integrated with sufficient communication links to allow positive control at all times, and it must provide the pilot with the same sorts of cues that road signs, maps, and traffic lights provide the automobile driver.
There is an important caveat: positive air traffic control may not have the same meaning with this airplane that it has for current airplanes; it may be much less restrictive than the current A TC system. In the future , positive control for small planes operating outside major terminal areas could be more like positive control for today's automobiles, consisting of the equivalent of traffic lights and limited-access highways, with automated up- and down-link of important information, e.g., airplane identifier, location, destination, approved routing, and so on. Furthermore, by adhering to more stringent design and operational requirements, the reinvented GA airplane could be allowed to have preferential departure, routing, and arrival assignments without degrading safety or airline scheduling.
__&oP__, c o ntinue d
Simplified Design for Certification and
Operation
Improved Inherent Safety, Reliability, and Maintainability Forgiving Failure Modes "Video Game" Flying Qualities
Design for Air Traffic Control System
Compatibility
Operation Under Positive Control 100% of the Time Fail-Safe Area Navigation Preferential Priorities for M_A-Planes CON C E PT F O R A TYPI C AL 4-PLACE _( 3A- PL ANE To pro v ide a tangible i dea of what a ME_3 A - P l a ne might be, a concept for a 4-place, single-engine design is discussed. This figure is a sketch, not a detailed drawing; it presents a concept for a MEGA- Plane, not the ME_3 A -Plane. The configuration appears similar to existing airplanes, and most of its characteristics have been suggested separately in earlier work. The most obvious difference is the addition of a controllable canard surface in addition to a conven- tional horizontal tail. More subtle visual characteristics include a split rudder, a "T" tail, small wing area, swept wing tips, large window area, fixed gear, and a sizable storage compartment.
i / I t C ONCE P T FOR A T YP ICA L 4-P L ACE _A - PL ANE AERODYNAMIC FEATURES The principal control features of the airplane are its 3-s u rface longi- tudinal control, control redundancy, and simple flaps. Three-surface control provides a number of desirable attributes, including increased allowable center-of-gravity travel, pitch trim for minimum cruise drag, pitch control redundancy, and reduced wing area. (The latter comes about because the canard, unlike the rear tail, can provide positive angular rotation and positive lift for takeoff, eliminating the otherwise necessary download of the tailplane.) Because the canard control surfaces are immediately behind the propeller, they deflect the slipstream, providing strong forces and moments that can be used to reduce takeoff distance, to implement gust alleviation for improved ride qualities, and for stall / spin prevention (or recovery).
Each control surface is independent, i.e., not connected mechanically to any other surface; hence, there are 10 surfaces that can be used in numerous combinations to produce 3 forces and 3 moments for control. (Although not a primary requirement, it would be possible to produce side force for wings- level crosswind landings.) Consequently, ailerons could act like flaps and flaps like ailerons in the event of failure, within possibly reduced limits.
Recent flight rese a rch has shown that dr a m a t i c reduct i ons i n dr a g c a n be achieved by encouraging natural laminar flow over the airplane. Simi- larly, long-held notions of planform effects are being questioned; there is the possibility that swept tips and even more dramatic treatments such as sheared tips, crescent planforms, and serrated trailing edges may further reduce subsonic dr a g.
A lthough a re d und an t fly-by-w i re / l i ght co n trol s y stem is i n d i c a ted , the ME(3A-Plane should have inhe / 'ent aerodynamic stability about all axes, allowing the "forgiving failure modes" mentioned earlier and eliminating the need for a stability augmentation system. (Closed-loop control may be desi- rable for a number of reasons; however, it should not be mandatory for safe flight.) While the airplane should never find itself in a spin , if it does enter the spin, it should have "honest" recovery characteristics, as implied by the unshadowed vertical tail and the full-length rudder.
CONCE P T FOR A TY P ICAL 4 -P LACE
_i__&-PLA N E
NATURAL LAMINAR FLOW AIRFOILS and FAIRINGS '__ UNSHADOWED VERTICAL TAIL REDUNDANT AE RODYNAMI C CONTROL ' STABILITY SURFACES SIMPLE FLAP I CONTROL SURFACES
\
REDUCED WING AREA VORTEX-DIFFUSING WING TIPS VECTORED THREE-SURFACE THRUST LONGITUDINAL CONTROL AERODYNAMIC FEATURES CO NC EPT FOR A T YP IC AL 4 - P L A CE t_,I [ E_A- P L AN E SYSTEM & STRUCTURAL FEATURES The all-electric / fiber-optic control system provides redundant, direct commands and power to all control actuators with a reliability that is at least equivalent to current pushrod-and-cable systems. In overall use, control reliability would be considerably better than that of current mechanical sys- tems because individual control surface failures will not have disasterous effects -- in fact, with one or two random failures, the changes in flying qual- ities would be barely discernable. Electric actuation is possible because the hinge moments for this small airplane need not be large. There would be considerable cost and reliability benefits from using identical actuators on all control surfaces. Fail-safe area navigation would be integrated with the flight control system; both systems would be aided by the use of solid-state motion sensors, as well as external navigation aids like LORAN. The integrated navigation, guidance, and control system would have artificial intelligence attributes as well as anti-wind shear / wake vortex features, which can be provided at minimal cost by existing microprocessors. Cockpit instruments also should be all electric, with a minimal complement of backup air-driven instruments to allow for display failure. Although the "see-and-avoid" approach to collision avoidance has been totally discredited over the years, optical aids that are visible in daylight should be mandatory.
While flight loads establish firm constraints, the airplane structure should be designed for inexpensive, automated fabrication and assembly.
Common, "pre-fabricated" components should be used where appropriate, and major elements should be designed to minimize the need for labor-intensive operations. While much enthusiasm has been generated for composite materials lately, it is not clear that they present the minimum-cost solution to small airplane construction when labor and time costs are taken into account.
There is a strong likelihood that large components would be made from aluminum, with small components and panels that can easily be molded into shape being made from composites.
For reasons mentioned earlier, advanced propulsion technology is war- ranted. With the increased importance of electrical power dictated by the control system, an auxiliary power unit and large batteries are appropriate.
The APU should not only provide electrical power; it should be designed to support anti-icing capability for the primary aerodynamic surfaces.
I00
CONCEPT FOR A TY P ICAL 4-PLACE
__3&- P LANE
STROBE LIGHT ADVANCED TECHNOLOGY POWERPLANT cm- FIXED LANDING GEAR ALUMINUM COMPOSITE and HONEYCOMB MAIN F RA_4E COMPONENTS ANTI-ICING AUXILIARY POWER UNIT SYSTEM and STRUCTURA L FEATURES i01 C O NCEP T FO R A T YP ICAL 4 - PLA CE I_, E -PL A N E R EDUNDANT CONTROL(_ANNELS B ecause t he loss of control more often th a n once per some large number of flights (e.g., 109) is an unacceptable risk, it is likely that control system redundancy would be required. While the necessary level of redundancy is system-specific and should be the subject of intensive study, the sort of strategy that can be applied is suggested by this figure. The present objective is not to pre-design the system but merely to indicate that high reliability may be achievable in a fairly simple fashion.
There are two primary systems, each of which commands a different set of control surfaces and both of which operate in parallel in normal conditions. Both can command single units, such as the engine and auxiliary power unit, separately, and both accept the same pilot commands. In the event that either computer fails, a third computer (a "hot spare") would be brought on-line; in normal operation, the third computer would monitor the other two. If either primary string fails altogether, the other string can maintain control with the remaining control surfaces. The implication is that each string in a 2-string system must be individually quite reliable, with a mean-time-between-total-failures on the order of 40,000 hr (about 41 / 2 years of operation). Individual components could fail much more frequently; this large figure applies to all. components in a single string failing simultaneously. Even in this case, the other 40,000-hr string would still be adequate to continue safe flight, and the spare computer would remain at the ready.
CONCEPT FO R A TYPICAL 4-PLACE
__&-P L ANE
<
\
\
REDUNDANT CONTROL CHANNELS OP P OR TUN I T I E S FO R R E S EA RCH Reinventing the small GA airplane is a _ystern,s prqblem, involving airplane design, human factors, and the air traffic control interface. This paper does not provide all the answers. It presents a concept and raises a challenge to the US aeronautical establishment, including government agencies, industry, and universities: conduct the research that is needed to make general aviation the vital contributor to economic, social, and transportation requirements that it can be.
Now in its fifteenth year, the FAA / NASA-sponsored Tri- University Program on Air Transportation Technology can play an important role in this research. Each of the participating universities __ Massachusetts Institute of Technology, Ohio University, and Princeton University -- provides unique perspectives and talents to be applied to the task. Together, we have demonstrated capabilities in literally all of the technologies that must be marshalled to produce this new generation of airplanes and the corresponding upgrade in the air traffic control system.
O P PORTU NITIE S F OR R E S E A R CH
MB(3A-Plane Developmen t is a
Systems Problem, involving
Aircraft Design Hum a n Factors Air Tr a ffic Control Interface Well-Defined Roles for FAA, NASA,