NASA SP-2000-4522

Flight Research:Problems Encountered and

What They Should Teach Us

byMilton O. Thompson

withA Background Section by J.D. Hunley

Monographs in Aerospace History # 22

NASA SP-2000-4522

Flight Research:Problems Encountered and

What They Should Teach Us

byMilton O. Thompson

withA Background Section by J.D. Hunley

NASA History DivisionOffice of Policy and Plans

NASA HeadquartersWashington, DC 20546

Monographs inAerospace History

Number 222000

Library of Congress Cataloging-in-Publication Data

Thompson, Milton O. Flight Research : problems encountered and what they should teach us / Milton O.Thompson ; with a background section by J.D. Hunley. p. cm. – (Monographs in aerospace history ;) (NASA history series) (NASA SP-2000 ; 4522) Includes bibliographical references and index.

1. Aeronautics—Research—United States. 2. Airplanes—Flight testing. 3. High-speedAeronautics. I. Title. II. Series. III. Series: NASA history series IV. NASA SP ; 4522.

TL565.T46 2000629.13’07’2073—dc2l 00-048072

For sale by the U.S. Government Printing OfficeSuperintendent of Documents, Mail Stop: SSOP, Washington, DC 20402-9328

ii

Foreword ......................................................................................................................................vi

Background: Flight Research at Dryden, 1946-1979 ....................................................................1

Problems Encountered in Flight Research ..................................................................................19

Aerodynamic Problems ........................................................................................................19

Environmental Problems ......................................................................................................24

Control-System Problems ....................................................................................................27

Structural Problems ..............................................................................................................28

Landing Gear Problems........................................................................................................28

Aerodynamic Heating Problems ..........................................................................................29

Auxiliary-Power-Unit Problems ..........................................................................................30

MH-96 Problems ..................................................................................................................30

Fatigue Problems..................................................................................................................32

X-15 Program Results ..........................................................................................................32

Control-System Problems in General...................................................................................32

First Flight Preparation ........................................................................................................35

First Flight and Envelope Expansion ...................................................................................37

Remotely Piloted Research Vehicles (RPRVs) ....................................................................38

Flight Test Errors..................................................................................................................40

Conclusions ..........................................................................................................................44

The Future ............................................................................................................................46

Index ............................................................................................................................................48

About the Author .........................................................................................................................50

Monographs in Aerospace History ..............................................................................................50

Table of Contents

iii

Foreword

The document by Milt Thompson that is reproduced here was an untitled rough draft found in Thompson’spapers in the Dryden Historical Reference Collection. Internal evidence suggests that it was written about1974. Readers need to keep this date in mind, since Milt writes in the present tense. Apparently, he neveredited the document. Had he prepared it for publication, he would have done lots of editing and refined muchof what he said.

I have not attempted to second guess what Milt might have done in revising the paper, but I have made someminor stylistic changes to make it more readable without changing the sense of what Milt initially wrote.Where I have qualified what Milt said or added information for the reader’s benefit, I have done so either infootnotes or inside square brackets [like these]. The draft itself indicated that it should contain numerousfigures to illustrate what he wrote, but no such figures were associated with the manuscript. I have searchedout figures that appear to illustrate what Milt intended to show, but in some cases I have found none. Whenthat has been the case, I have deleted his references to figures and simply kept his text, which does stand onits own.

For the most part, I have not attempted to bring his comments up to date, although in a few instances I haveinserted footnotes that indicate some obvious changes since he wrote the paper. Despite—or perhaps becauseof—the paper’s age, it offers some perspectives on flight research that engineers and managers not familiarwith the examples Milt provides can still profit from in today’s flight-research environment. For that reason, Ihave gone to the trouble to edit Milt’s remarks and make them available to those who would care to learnfrom the past.

For readers who may not be familiar with the history of what is today the NASA Dryden Flight ResearchCenter and of its predecessor organizations, I have added a background section. Those who do know thehistory of the Center may wish to skip reading it, but for others, it should provide context for the events Miltdescribes. Milt’s biography appears at the end of the monograph for those who would like to know moreabout the author of the document.

Many people have helped me in editing the original manuscript and in selecting the figures. The process hasgone on for so long that I am afraid to provide a list of their names for fear of leaving some important con-tributors out. A couple of them, in any event, requested anonymity. Let me just say a generic ‘thank you’ toeveryone who has assisted in putting this document into its present form, with a special thanks to the DrydenGraphics staff members, especially Jim Seitz, for their work on the figures; to Jay Levine and Steve Lighthillfor laying the monograph out; to Darlene Lister for her assistance with copy editing; and to Camilla McArthurfor seeing the monograph through the printing process.

J. D. Hunley, Historian

iv

speed of sound (Mach 1) in levelflight.1

Even so, during the early 1940s,airplanes like Lockheed’s P-38 Light-ning began to face the problem ofcompressibility in dives—character-ized (among other things) by increaseddensity, a sharp rise in drag, anddisturbed airflow at speeds approach-ing Mach 1. The effects of compress-ibility included loss of elevator effec-tiveness and even the break-up ofstructural members such as the tail,killing pilots in the process. Thisproblem was compounded by theabsence of accurate wind-tunnel data

Milt Thompson’s account of lessons to belearned assumes some familiarity withthe history of flight research at what wasthen called the Flight Research Centerand its predecessor organizations—redesignated in 1976 the Hugh L. DrydenFlight Research Center. The followingaccount provides a brief history of the

subject that perhaps will provide a usefulbackdrop to what Milt had to say.

From the time of the Wright brothers’first flight in 1903 until the end of WorldWar II, airplane technology evolvedconsiderably. The early decades’ mono-and biplanes of wooden framework,typically braced with wire and coveredwith cloth, gradually gave way to an all-metal construction and improved aerody-namic shapes, but most aircraft in WorldWar II still featured propellers and eventhe fastest of them flew at maximumspeeds of about 450 miles per hour. Forexample, the North American P-51Mustang, one of the finest prop fighters

used in the war, had a top speed of 437miles per hour when flying a levelcourse at low altitude. This comparedwith low-level maximum speeds of 514and 585 mph respectively for theMesserschmitt Me 262A and GlosterMeteor F.Mk. jet fighters, both ofwhich thus still flew well below the

1 See, e.g., Laurence K. Loftin, Jr., Quest for Performance: The Evolution of Modern Aircraft (Washington, DC, NASASP-468, 1985), Chs. 1-5 and 9-10, esp. pp. ix, 7-45, 77-88, 128-136, 281-286, 484-490; Roger E. Bilstein, Flight inAmerica: From the Wrights to the Astronauts (rev. ed.; Baltimore, MD.: Johns Hopkins Univ. Press, 1994), pp. 3-40,129-145.

Background:FlightResearch atDryden,1946-1979

J. D. Hunley

A P-51 Mustangon the lakebednext to theNACA High-Speed FlightStation in 1955.(NASA photoE55-2078)

1

for portions of the transonic speedrange in a narrow band on either side ofthe speed of sound.2

This situation led to the myth of a soundbarrier that some people believed couldnot be breached. Since it appeared that jetaircraft would soon have the capability offlying in level flight into the transonicregion—where the dreaded compressibil-ity effects abound—a solution was neededfor the lack of knowledge of transonicaerodynamics. A number of people(including Ezra Kotcher with the ArmyAir Forces [AAF] at Wright Field in Ohio,John Stack at the Langley MemorialAeronautical Laboratory of the NationalAdvisory Committee for Aeronautics[NACA] in Virginia, Robert Woods withBell Aircraft, L. Eugene Root with

Douglas Aircraft, and Abraham Hyatt atthe Navy Bureau of Aeronautics) con-cluded that the solution could best resultfrom a research airplane capable of flyingat least transonically and even supersoni-cally.

The emphases of these different organiza-tions resulted in two initial aircraft—theXS-1 (XS standing for eXperimentalSonic, later shortened to X), for whichBell did the detailed design and construc-tion for the AAF, and the D-558-1Skystreak, designed and constructed byDouglas for the Navy. The XS-1 was thefaster of the two, powered by an XLR-11rocket engine built by Reaction Motorsand launched from a B-29 or later a B-50“mothership” to take full advantage of thelimited duration provided by its rocket

2 James O. Young, Meeting the Challenge of Supersonic Flight (Edwards AFB, CA: Air Force Flight Test CenterHistory Office, 1997), pp. 1-2; John V. Becker, The High-Speed Frontier: Case Histories of Four NACA Programs(Washington, DC: NASA SP-445, 1980), esp. p. 95. It should be noted here that the first studies of compressibilityinvolved tip speeds of propellers and date from 1918 to 1923. On these, see especially John D. Anderson, Jr., “Researchin Supersonic Flight and the Breaking of the Sound Barrier” in From Engineering Science to Big Science: The NACAand NASA Collier Trophy Research Project Winners, ed. Pamela Mack (Washington, DC: NASA SP-4219, 1998), pp.66-68. This article also provides excellent coverage of the early research of John Stack and his associates at the NACA’sLangley Memorial Aeronautical Laboratory on the compressibility issue for aircraft (as opposed to propellers).

A P-38 Light-ning in flight in1943. (NASAphoto E95-43116-2)

2

propulsion. Stack and the other NACAengineers were skeptical about the rocketengine and less concerned about breakingthe sound barrier than gathering flightdata at transonic speeds, so they sup-ported the AAF-Bell project with criticaldesign data and recommendations butwere more enthusiastic about the Navy-

Douglas Skystreak. This was designedwith an early axial-flow turbojetpowerplant and was capable of flyingonly up to Mach 1. However, withcomparably designed wings and amovable horizontal stabilizer recom-mended by the NACA, plus the abilityto fly in the transonic region for a

XS-1 Number 2on the ramp atEdwards AirForce Basewith its B-29mothership.(NASA photoE-9)

A D-558-1 inflight. (NASAphoto E-713)

3

3 Young, Supersonic Flight, pp. 2-18; Becker, High-Speed Frontier, pp. 90-93; Richard Hallion, Supersonic Flight:Breaking the Sound Barrier and Beyond, The Story of the Bell X-1 and the Douglas D-558 (rev. ed.; London andWashington, DC: Brassey’s, 1997), esp. pp. 35-82; Louis Rotundo, Into the Unknown: The X-1 Story (Washington, DCand London: The Smithsonian Institution Press, 1994), esp. pp. 11-33; Toward Mach 2: The Douglas D-558 Program, ed.J. D. Hunley (Washington, DC: NASA SP-4222, 1999), esp. pp. 3-7.

longer period of time, the D-558-1complemented the XS-1 nicely andprovided comparable data.3

To support the research flights, thecontractors, the AAF (after September

1947, the Air Force), and the NACA sentteams of pilots and support personnel tothe Muroc Army Air Field starting inSeptember 1946 to support the XS-1,and then the Navy joined in to help flythe D-558. At Muroc, the 44-square-mile

XS-1 Number 2on Rogers DryLakebed in aphoto that givessome sense of theexpanse ofnatural runwayprovided by thedry lake. (NASAphoto E49-001)

XS-1 Number 1in flight withcopy of “Machjump” papertape data recordof the firstsupersonic flightby Air ForceCapt. ChuckYeager. (NASAphoto E-38438

4

Rogers Dry Lakebed provided an enormousnatural landing field, and the clear skies andsparse population provided an ideal envi-ronment for conducting classified flightresearch and tracking the aircraft.4

The most immediate and dramatic resultof these twin flight research efforts that

proceeded simultaneously at Muroc wasAir Force pilot Chuck Yeager’s breakingthe sound barrier on 14 October 1947 inthe XS-1, for which feat he garnered theCollier Trophy the next year in conjunc-tion with John Stack for the NACA andLarry Bell for his company.5 The flightdispelled the myth about a sound barrier

4 Rotundo, Into the Unknown, pp. 96, 123-132; James R. Hansen, Engineer in Charge: A History of the Langley Aero-nautical Laboratory, 1917-1958 (Washington, DC: NASA SP-4305, 1987), p. 297. Contrary to what is reported in anumber of sources, the initial NACA contingent did not arrive on 30 Sept. 1946, with Walter C. Williams. Harold H.Youngblood and George P. Minalga arrived Sunday, 15 Sept., William P. Aiken, sometime in October after Williams’arrival. Telephonic intvws., Hunley with Youngblood and Aiken, 3 and 4 Feb. 1997.

5 See especially Rotundo, Into the Unknown, pp. 279, 285.

NACA researchaircraft on theramp at the SouthBase area ofEdwards AirForce Base, (leftto right) D-558-2,D-558-1, X-5, X-1, XF-92A, X-4.(NASA photoEC-145)

In-flight photo ofthe X-3. (NASAphoto E-17348)

5

and undoubtedly did much to gain creditfor flight research, resulting in the smallcontingent of NACA engineers, pilots,and support people at Muroc becoming apermanent facility of the NACA and laterNASA.

All of this was extremely important,but even more important than therecord and the glory that went with itwere the data that the NACA garneredfrom the flight research not only withthe several X-1 and D-558-1 aircraft,but also with the Douglas D-558-2,the Bell X-2, the Douglas X-3 “flyingstiletto,” the Northrop X-4semitailless, the Bell X-5 variable-sweep, and the Convair XF-92A delta-winged aircraft. Not all of these air-planes were successful in a conventionalsense, even as research airplanes. But all

of them provided important data foreither validating or correcting informa-tion from wind tunnels and designingfuture airplanes ranging from the Cen-tury series of fighter aircraft to today’scommercial transports, which still fly inthe transonic speed range and feature themovable horizontal stabilizer demon-strated on the X-1 and D-558s. Even theill-fated X-2, of which Dick Hallion haswritten, “its research was nil,” and theX-3, which he has dubbed “NACA’sglamorous hangar queen,” 6 neverthelesscontributed to our understanding of theinsidious problem of coupling dynamics.Furthermore, the Air Force-NACA X-2program featured the first simulator usedfor the various functions of flight-testplanning, pilot training, extraction ofaerodynamic derivatives, and analysisof flight data.7

6 Hallion, On the Frontier: Flight Research at Dryden, 1946-1981 (Washington, DC: NASA SP-4303, 1984), pp. 78,59 respectively. In both cases, Hallion’s characterizations are justifiable in some degree.

7 On the coupling and the computer simulation, Richard E. Day, Coupling Dynamics in Aircraft: A Historical Perspec-tive (Dryden Flight Research Center, CA: NASA SP-532, 1997), esp. pp. 8-15, 34-36. On the value of the X-2 and X-3,see also Ad Inexplorata: The Evolution of Flight Testing at Edwards Air Force Base (Edwards AFB, CA: Air ForceFlight Test Center History Office, 1996), pp. 14, 16. For the other research results, see especially Walter C. Williamsand Hubert M. Drake, “The Research Airplane: Past, Present, and Future,” Aeronautical Engineering Review (Jan.1958): 36-41; Becker, High-Speed Frontier, pp. 42, 95-97; Hallion, On the Frontier, 59-62. In writing this account, Ihave benefited greatly from comments made to me over the years by long-time Dryden research engineer Ed Saltzman.

X-2 in flight.(NASA photoE-2822)

6

Before this account discusses some of theother highlights of flight research at whatbecame NASA’s Dryden Flight ResearchCenter, perhaps it should explain thedifferences and similarities betweenflight research and flight test. Bothinvolve highly trained, highly skilledpilots and sometimes exotic or cutting-edge aircraft, although flight research canuse quite old aircraft modified forparticular kinds of research. There is nohard and fast dividing line separatingflight research from flight test in prac-tice, but flight research, unlike flight testin most applications, is oblivious to theparticular aircraft employed so long asthat airplane can provide the requiredflight conditions. On the other hand,flight test, as the name implies, ofteninvolves testing specific prototype orearly production aircraft (somewhat laterproduction aircraft in the case of opera-tional flight-testing) to see if they fulfillthe requirements of a particular contractand/or the needs of the user. In addition,however, flight testing—at least in theAir Force—involves flying aircraft that

may be quite old to try to improve themand to develop their systems. For ex-ample, the Air Force Flight Test Centerrecently began testing the F-22, a brandnew airplane, while at the same time itcontinued to test the F-15 and its systemseven though various models of F-15s hadbeen in the inventory for more than twodecades.

In partial contrast to flight test, flightresearch sought and seeks fundamentalunderstanding of all aspects of aeronau-tics, and in achieving that understanding,its practitioners may fly experimentalaircraft like the early X-planes and theD-558s or armed service discards likeearly production models of the F-15s,F-16s, and F-18s researchers at Drydenare modifying and flying today. Theymay even fly comparatively new aircraftlike the F-100 in its early days; here,however, the purpose is not to test themagainst contract standards but to under-stand problems they may be exhibiting inoperational flight and learn of ways tocorrect them—a goal very similar to that

This NACAHigh-SpeedFlight Stationphotograph ofthe CenturySeries fightersin formationflight wastaken in 1957(clockwisefrom left —F-104, F-101,F-102, F-100).(NASA photoE-2952)

7

of flight testing in its effort to improve existingaircraft. I should note in this connection thatthe Air Force Flight Test Center (as it is calledtoday) and Dryden (under a variety ofprevious names) have often cooperated inflight research missions, with Air Force andNACA/NASA pilots flying together. Soclearly flight test organizations in these casesparticipate in flight research just as researchpilots sometimes engage in flight tests.8 Itshould also be added that although manyresearchers at what is today Dryden might bequick, if asked, to point out the differencesbetween flight test and flight research, manyof them, including Milt in the account below,often used the two terms as if they wereinterchangeable.

To return to specific flight research projectsat Dryden, on 20 November 1953, withNACA pilot Scott Crossfield in the pilot’sseat, the D-558-2 exceeded Mach 2 in aslight dive, and on 27 September 1956,Air Force Capt. Mel Apt exceededMach 3 in the X-2 before losing controlof the aircraft due to inertial couplingand plunging to his death.9 With thethen-contemporary interest in spaceflight, clearly there was a need at thispoint for research into hypersonicspeeds (above Mach 5) and attendantproblems of aerodynamic heating, flightabove the atmosphere, and techniquesfor reentry. In early 1954, therefore, theNACA’s Research Airplane Projects

8 The account that comes closest to what I have said above is Lane Wallace’s Flights of Discovery: 50 Years at the NASADryden Flight Research Center (Washington, DC: NASA SP-4309, 1996), pp. 4-8. On flight test per se, see AdInexplorata, esp. pp. 12-13. For a useful history of both flight testing and flight research, see Richard P. Hallion, “FlightTesting and Flight Research: From the Age of the Tower Jumper to the Age of the Astronaut,” in Flight Test Techniques,AGARD Conference Proceedings No. 452 (copies of papers presented at the Flight Mechanics Panel Symposium,Edwards AFB, CA, 17-20 Oct. 1988), pp. 24-1 to 24-13. Finally, for an early discussion of flight research (despite itstitle) see Hubert M. Drake, “Aerodynamic Testing Using Special Aircraft,” AIAA Aerodynamic Testing Conference,Washington, DC, Mar. 9-10, 1964, pp. 178-188. In writing and refining the above two paragraphs, I have greatlybenefited from AFFTC Historian Jim Young’s insightful comments about flight test, especially as it is practiced today atEdwards AFB, as well as from comments by Ed Saltzman. A point Ed offered that I did not incorporate in the narrative isthat in the obliviousness of flight research to the specific aircraft used, it has more in common with wind-tunnel researchthan with flight test.

9 Hallion, On the Frontier, pp. 308, 316.

The D-558-2Number 2 islaunched fromthe P2B-1 in this1956 NACAHigh-SpeedFlight Stationphotograph.This is the sameairplane thatScott Crossfieldhad flown toMach 2.005 in1953. (NASAphoto E-2478)

8

Panel began discussion of a newresearch airplane that became the X-15.Developed under an Air Force contract withNorth American Aviation, Inc., and flownfrom 1959 to 1968, the X-15 set unofficialworld speed and altitude records of 4,520miles per hour (Mach 6.7) and 354,200 feet(67 miles).10

Much more importantly, however, the joint AirForce-Navy-NASA-North American

program investigated all aspects of pilotedhypersonic flight. Yielding over 765research reports, the 199-flight program“returned benchmark hypersonic data foraircraft performance, stability and control,materials, shock interaction, hypersonicturbulent boundary layer, skin friction,reaction control jets, aerodynamic heating,and heat transfer,”11 as well as energymanagement. These data contributed tothe development of the Mercury,

11 Kenneth W. Iliff and Mary F. Shafer, Space Shuttle Hypersonic Aerodynamic and Aerothermodynamic Flight Re-search and the Comparison to Ground Test Results (Washington, DC: NASA Technical Memorandum 4499,1993), p. 2,for quotation and see also their “A Comparison of Hypersonic Flight and Prediction Results,” AIAA-93-0311, paperdelivered at the 31st Aerospace Sciences Meeting & Exhibit, Jan. 11-14, 1993, in Reno, NV.

10 See esp. ibid., pp. 101-129, 333, 336, and Wendell H. Stillwell, X-15 Research Results (Washington, DC: NASA SP-60, 1965), p. vi and passim.

The X-15 shipNumber 3 (56-6672)is seen here on thelakebed at theEdwards Air ForceBase, California.Ship Number 3 made65 flights during theprogram, attaining atop speed of Mach5.65 and a maximumaltitude of 354,200feet. (NASA photoE-7896)

In this photo theNumber 1 XB-70A (62-0001) isviewed fromabove in cruiseconfigurationwith the wingtips drooped forimproved con-trollability.(NASA photoEC68-2131)

9

Gemini, and Apollo piloted spaceflightprograms as well as the later SpaceShuttle program.12

Overlapping the X-15 program in time,the XB-70 also performed significanthigh-speed flight research. The XB-70was the world’s largest experimentalaircraft. It was capable of flight atspeeds of three times the speed ofsound (roughly 2,000 miles per hour) ataltitudes of 70,000 feet. It was used tocollect in-flight information for use inthe design of future supersonic aircraft,both military and civilian.

The more specific major objectives ofthe XB-70 flight research program wereto study the airplane’s stability andhandling characteristics, to evaluate itsresponse to atmospheric turbulence, andto determine the aerodynamic andpropulsion performance. In addition,

there were secondary objectives tomeasure the noise and friction associ-ated with airflow over the airplane andto determine the levels and extent of theengine noise during takeoff, landing,and ground operations. The first flightof the XB-70 was made on 21 Septem-ber 1964. The Number two XB-70 wasdestroyed in a mid-air collision on 8June 1966. Program management of theNASA-USAF research effort wasassigned to NASA in March 1967. Thefinal flight was flown on 4 February1969. The program did provide a greatdeal of data that could be applied to afuture supersonic transport or a large,supersonic military aircraft. It alsoyielded data on flight dynamics, sonicbooms, and handling qualities.13

Another important high-speed flightresearch program involved theLockheed YF-12 “Blackbird,” precursor

12 See, e.g., John V. Becker, “The X-15 Program in Retrospect,” 3rd Eugen Sänger Memorial Lecture, Bonn, Germany,Dec. 4-5, 1968, copy in the NASA Dryden Historical Reference Collection; Milton O. Thompson, At the Edge of Space:The X-15 Flight Program (Washington, DC, and London: Smithsonian Institution Press, 1992); and the sources citedabove.

13 Hallion, On the Frontier, pp. 185-188; and see, e.g., P. L Lasagna and T. W. Putnam, “Engine Exhaust Noise duringGround Operation of the XB-70 Airplane” (Washington, DC: NASA TN D-7043, 1971) and C. H. Wolowicz and R. B.Yancey, Comparisons of Predictions of the XB-70-1 Longitudinal Stability and Control Derivatives with Flight Resultsfor Six Flight Conditions (Washington, DC: NASA TM X-2881, 1973).

A YF-12A inflight. (NASAphoto EC72-3150)

10

14 On the YF-12, see esp. Berwin M. Kock, “Overview of the NASA YF-12 Program,” YF-12 Experiments Symposium,Vol. 1, (3 vols.; Washington, DC: NASA CP-2054, 1978) plus more specialized papers in the volume; Robert D. Quinnand Frank V. Olinger, “Flight Temperatures and Thermal Simulation Requirements,” NASA YF-12 Flight Loads Program(Washington, DC: NASA TM X-3061, 1974), pp. 145-183; and Hallion, On the Frontier, pp. 196-199, 349-356.

of the SR-71 reconnaissance airplanethat flew at Dryden during the 1990s.Three YF-12s flew at Edwards in a jointNASA-AF research program between1969 and 1979. The aircraft studied thethermal, structural, and aerodynamiceffects of sustained, high-altitude, Mach 3flight. They also studied propulsion, airflow and wind gusts, jet wake dispersion,engine stalls, boundary-layer noise, andmuch else. The 125 research reports theprogram produced contained vastamounts of information used in designingor improving other supersonic aircraft,including the SR-71. Among other things,engineers at Dryden developed a centralairborne performance analyzer to monitorYF-12 flight parameters. It became theforerunner of the on-board diagnosticsystem used on the Space Shuttle.14

Not all of Dryden’s flight research hasconcerned high-speed flight. One crucialflight research project that certainly had

implications for high-speed flight but wasnot restricted to that regime was the F-8Digital Fly-By-Wire project. Drydenengineers replaced all purely mechanicallinkages to flight-control surfaces (rud-ders, ailerons, elevators, and flaps) in anF-8C with electronic ones controlled by adigital flight-control system. Althoughthere had been previous analog flightcontrol systems, this was not only thefirst digital system but also the firstelectronic system without a conven-tional mechanical backup, using ananalog backup instead. Flown in the1970s and into the mid-1980s, the F-8DFBW first used the Apollo computerdeveloped by Draper Lab and then theIBM AP-101 later employed on theShuttle. Flying this system without amechanical backup was important ingiving industry the confidence todevelop its own digital systems sinceflown on the F-18, F-16, F-117, B-2,F-22, and commercial airliners like the

This photo shows the F-8Digital-Fly-By Wire aircraftin flight. The project involv-ing this aircraft contributedsignificantly to the flightcontrol system on the spaceshuttles by testing and gettingthe bugs out of the IBM AP-101 used on the shuttles andby helping the Dryden FlightResearch Center to develop apilot-induced oscillation(PIO) suppression filter thatreduced the likelihood ofpilots overcontrolling theshuttles on landings andthereby creating excursionsfrom the intended landingpath. (NASA photo EC77-6988)

11

Boeing 777—not to mention the X-29and X-31 research aircraft. Some of theseaircraft would be uncontrollable withoutDFBW technology, which is not onlylighter than mechanical systems butprovides more precise and better maneu-ver control, greater combat survivability,and for commercial airliners, a smootherride.15

While the F-8 Digital Fly-By-Wireproject was still ongoing, Drydenhosted the Approach and Landing Testsfor the Space Shuttle in 1977. Theseinvolved testing the 747 Shuttle CarrierAircraft (SCA) modified to carry theShuttle back to its launch location atthe Kennedy Space Center in Florida(following Shuttle landings on the

Rogers Dry Lakebed), plus flying theShuttle prototype Enterprise mated tothe 747, both without and with a crewon the Shuttle, and then five free flightsof the Enterprise after it separated fromthe SCA, including four lakebed land-ings and one on the regular runway atEdwards. Flying without a tail-conefairing around the dummy main enginesas well as landing on the smallerrunway for the first time, astronaut andformer Dryden research pilot FredHaise was keyed up and overcontrolledthe orbiter on the concrete-runwaylanding, resulting in a pilot-inducedoscillation. Once he relaxed his con-trols, the Enterprise landed safely, butonly after some very uneasy mo-ments.16

15 James E. Tomayko, “Digital Fly-by-Wire: A Case of Bidirectional Technology Transfer,” Aerospace Historian (March1986), pp. 10-18, and Computers Take Flight: A History of NASA’s Pioneering Digital Fly-By-Wire Project (Washington,DC: NASA SP-2000-4224, 2000); Proceedings of the F-8 Digital Fly-By-Wire and Supercritical Wing First Flight’s 20thAnniversary Celebration (Edwards, CA: NASA Conference Publication 3256, 1996), Vol. I, esp. pp. 4, 15, 19-20, 34,46-51, 56; oral history interview, Lane Wallace with Kenneth J. Szalai and Calvin R. Jarvis, Aug. 30, 1995, transcript inthe Dryden Historical Reference Collection. See also Lane Wallace’s account in Flights of Discovery, pp. 111-118.

16 Space Shuttle Orbiter Approach and Landing Test, Final Evaluation Report (Houston, TX: NASA JSC-13864. 1978).For shorter, less technical descriptions, see Hallion, On the Frontier, pp. 242-250, and Wallace, Flights of Discovery, pp.134-137.

The Space Shuttleprototype Enter-prise flies free afterbeing released fromNASA’s 747Shuttle CarrierAircraft (SCA)during one of fivefree flights carriedout at the DrydenFlight ResearchCenter, Edwards,California, as partof the Shuttleprogram’s Ap-proach and LandingTests (ALT).(NASA photoECN-8611)

12

This posed a hazard for Shuttle landingsfrom space, because if a keyed-up pilotovercontrolled, the results might be moredangerous. There needed to be a correc-tion to the Shuttle’s flight control system.So the F-8 DFBW and other aircraft werepressed into service to find a solution.Dryden engineers suspected the problemlay in the roughly 270-millisecond timedelay in the Shuttle’s DFBW flightcontrol system, so pilots flew the F-8DFBW research airplane with increasingtime delays to test this belief. When theexperimental time-delay reached 100milliseconds, Dryden research pilot JohnManke was doing a touch-and-go landing-take-off sequence and entered a severepilot-induced oscillation at a high angle ofattack and low speed. As the aircraftporpoised up and down in increasingly

severe oscillations, hearts stopped in thecontrol room until fellow F-8 pilot GaryKrier reminded Manke to turn off the timedelay, allowing him to climb to a safealtitude. The control room remained in astunned silence until Gary keyed up themike again and said, “Uh, John, I don’tthink we got any data on that; we’d likeyou to run it again.” The ensuinglaughter broke the tension. As a resultof this and 13 other flights in Marchand April 1978, Dryden engineers hadthe data they needed to suppress con-trol-surface action resulting fromexcessive pilot inputs. A suppressionfilter reduced the probability of a pilot-induced oscillation without affectingnormal flying qualities, contributing tothe safe landings of the Shuttle eversince.17

17 On this flight research, see especially Wallace’s intvw. with Szalai and Jarvis; Tomayko, “Digital Fly-by-Wire,” p. 17,and Computers Take Flight, pp. 113-114; and Wallace, Flights of Discovery, p. 137.

This 1964 NASAFlight ResearchCenter photo-graph shows theLunar LandingResearch Vehicle(LLRV) Number1 in flight at theSouth Base ofEdwards AirForce Base.(NASA photoECN-506)

13

A much earlier contribution to thenation’s space effort was the LunarLanding Research Vehicle (LLRV). WhenApollo planning was beginning in 1960,NASA began looking for a simulator toemulate the descent to the moon’s sur-face. Three projects developed, but themost important was the LLRV developedby Bell Aerosystems in partnership withthe Flight Research Center. Two LLRVspaved the way for three Lunar LandingTraining Vehicles (LLTVs) supplementedby the LLRVs, which were converted intoLLTVs.

Ungainly vehicles humorously called“flying bedsteads,” they simulated themoon’s reduced gravity on descent byhaving a jet engine provide five-sixths ofthe thrust needed for them to stay in theair. A variety of thrusters then handled therate of descent and provided control. Thevehicles gave the Apollo astronauts aquite realistic feel for what it was like toland on the Moon. Neil Armstrong said

that he never had a comfortable momentflying the LLTVs, and he crashed oneof the LLRVs after it was converted toan LLTV, escaping by means of theejection system. But he said he couldnot have landed on the Moon withoutthe preparation provided by theLLTVs.18

Another very important contribution tothe Shuttles and probably to futurespacecraft came from the lifting bodies.Conceived first by Alfred J. Eggers andothers at the Ames Aeronautical Labora-tory (now the Ames Research Center),Mountain View, California, in the mid-1950s, a series of wingless lifting shapescame to be flown at what later becameDryden from 1963 to 1975 in a jointprogram with the Air Force, other NASAcenters, and both Northrop and Martin onthe industrial side. They included theM2-F1, M2-F2, M2-F3, HL-10, andX-24A and B. Flown at comparativelylow cost, these low lift-over-drag vehicles

18 On the LLRVs and LLTVs, see Donald R. Bellman and Gene J. Matranga, Design and Operational Characteristics ofa Lunar-Landing Research Vehicle (Washington, DC: NASA TN D3023, 1965) and Hallion, On the Frontier, pp. 140-146.

The HL-10landing on thelakebed with anF-104 chaseaircraft. (NASAphoto ECN-2367)

14

demonstrated both the viability and versatil-ity of the wingless configurations and theirability to fly to high altitudes and then toland precisely with their rocket engines nolonger burning. Their unpowered approachesand landings showed that the Space Shuttlesneed not decrease their payloads by carryingfuel and engines that would have been requiredfor conventional, powered landings initiallyplanned for the Shuttle. The lifting bodies

also prepared the way for the later X-33and X-38 technology demonstrator pro-grams that feature lifting-body shapes to beused for, respectively, a potential next-generation reusable launch vehicle and acrew return vehicle from the InternationalSpace Station.19

A very different effort was the F-8Supercritical Wing flight research project,

19 For the details of this remarkable program, see R. Dale Reed with Darlene Lister, Wingless Flight: The Lifting BodyStory (Washington, DC: NASA SP-4220, 1997); Milton O. Thompson with Curtis Peebles, Flight without Wings: NASALifting Bodies and the Birth of the Space Shuttle (Washington, DC: Smithsonian Institution Press, 1999).

This photoshows the M2-F3Lifting Bodybeing launchedfrom NASA’s B-52 mothership atthe NASA FlightResearch Center.(NASA photoEC71-2774)

The X-24Blanding on thelakebed with anF-104 safetychase aircraft.(NASA photoEC75-4914)

15

conducted at the Flight Research Centerfrom 1971 to 1973. This project illus-trates an important aspect of flightresearch at what is today Drydenbecause the design was the work of Dr.Richard Whitcomb at the LangleyResearch Center and resulted from hisinsights and wind-tunnel work. Fre-quently, projects flown at Dryden haveresulted from initiatives elsewhere inNASA, in the armed services, in indus-try, or other places. However, research-ers often discover things in flight thatwere only dimly perceived—or notperceived at all—in theoretical andwind-tunnel work, and flight researchalso can convince industry to adopt a newtechnology when it wouldn’t do so as aresult of wind-tunnel studies alone. In thiscase, Larry Loftin, director of aeronauticsat Langley, said, “We’re going to have aflight demonstration. This thing is sodifferent from anything we’ve ever donebefore that nobody’s going to touch itwith a ten-foot pole without somebodygoing out and flying it.”20

In this case, although there was somediscovery resulting from the flight re-search—e.g., that there was some laminarflow on the wing that was not predicted,in addition to the numerous discrepanciesMilt notes in his account below—gener-ally there was good correlation betweenwind-tunnel and flight data. The SCW hadincreased the transonic efficiency of theF-8 by as much as 15 percent, equating tosavings of $78 million per year in 1974dollars for a 280-passenger transport fleetof 200-passenger airplanes. As a result ofthis study, many new transport aircrafttoday employ supercritical wings. More-over, subsequent flight research withsupercritical wings on the F-111 showedthat the concept substantially improved afighter aircraft’s maneuverability andperformance.21

A final project that should be mentionedhere is the research with the three-eighths-scale F-15/Spin Research Vehicle. Thiswas a sub-scale remotely piloted researchvehicle chosen because of the risks

20 Ted Ayers, “The F-8 Supercritical Wing; Harbinger of Today’s Airfoil Shapes,” Proceedings of the F-8 . . .Supercritical Wing, pp. 69-80, and Richard Whitcomb, “The State of Technology Before the F-8 Supercritical Wing,”ibid., pp. 81-92, quotation from p. 85.

21 Ayers, “Supercritical Wing,” p. 78; Whitcomb, “State of Technology,” pp. 84, 90; Hallion, On the Frontier, pp. 202-208; Wallace, Flights of Discovery, pp. 90-92.

The F-8SupercriticalWing aircraft inflight. (NASAphoto EC73-3468)

16

involved in spin testing a full-scale fighteraircraft. The remotely piloted researchtechnique enabled the pilot to interactwith the vehicle as he did in normal flight.It also allowed the flight envelope to beexpanded more rapidly than conventionalflight research methods permitted forpiloted vehicles. Flight research over anangle-of-attack range of -20 degrees to+53 degrees with the 3/8-scale vehicle—during its first 27 flights through the endof 1975 in the basic F-15 configuration—allowed FRC engineers to test the math-

ematical model of the aircraft in anangle-of-attack range not previouslyexamined in flight research. The basicairplane configuration proved to beresistant to departure from straight andlevel flight, hence to spins. The vehiclecould be flown into a spin using tech-niques developed in the simulator,however. Data obtained during the first27 flights gave researchers a betterunderstanding of the spin characteris-tics of the full-scale fighter. Research-ers later obtained spin data with the

vehicle in other configurations at anglesof attack as large as –70 degrees and +88degrees.

There were 36 flights of the 3/8-scale F-15s bythe end of 1978 and 53 flights by mid-July of1981. These included some in which thevehicle—redesignated the Spin ResearchVehicle after it was modified from the basicF-15 configuration—evaluated the effects ofan elongated nose and a wind-tunnel-designednose strake (among other modifications) on theairplane’s stall/spin characteristics. Results of

flight research with these modificationsindicated that the addition of the nosestrake increased the vehicle’s resistanceto departure from the intended flightpath, especially entrance into a spin. Largedifferential tail deflections, a tail chute, anda nose chute all proved effective as spinrecovery techniques, although it wasessential to release the nose chute once ithad deflated in order to prevent an inadvert-ent reentry into a spin. Overall, remotepiloting with the 3/8th-scale F-15 providedhigh-quality data about spin.22

22 Kenneth W. Iliff, “Stall/Spin Results for the Remotely Piloted Spin Research Vehicle,” AIAA Paper No. 80-1563presented at the AIAA Atmospheric Flight Mechanics Conference, Aug. 11-13, 1980; Kenneth W. Iliff, Richard E.Maine, and Mary F. Shafer, “Subsonic Stability and Control Derivatives for an Unpowered, Remotely Piloted 3/8-ScaleF-15 Airplane Model Obtained from Flight Test,” (Washington, DC: NASA TN D-8136, 1976).

This photographshows NASA’s3/8th-scaleremotely pilotedresearch vehiclelanding onRogers DryLakebed atEdwards AirForce Base,California, in1975. (NASAphoto ECN-4891)

17

In these and many other projects, what istoday the Dryden Flight Research Centerhas shown that while theory, groundresearch facilities, and now ComputationalFluid Dynamics are critical for the design ofaircraft and for advancing aeronautics,flight research is also indispensable. Itserves not only to demonstrate and validatewhat ground research facilities havediscovered but also—in the words of Hugh

Dryden—to “separate the real from theimagined . . .” and to discover in flightwhat actually happens as far as instrumentsand their interpretation will permit.23 Thisessential point is reemphasized in Milt’sstudy from his own particular perspective,but his account also contains a great dealmore that practitioners of flight researchtoday—and perhaps even ground research-ers—would do well to heed.

23 For the quotation, Hugh L. Dryden, “General Background of the X-15 Research-Airplane Project,” in the NACA,Research-Airplane-Committee Report on Conference on the Progress of the X-15 Project (Langley Field, VA: Compila-tion of Papers Presented, Oct. 25-26, 1956): xix. Dryden’s comment related specifically to the X-15 but has more generalapplicability. On the need for interpretation of data from instruments, see Frederick Suppe’s interesting “The ChangingNature of Flight and Ground Test Instrumentation and Data: 1940-1969” on the Internet at http://carnap.umd.edu:90/phil250/FltTest/FltTest1.pdf. Of course, with the use of lasers in a variety of applications today to augment more tradi-tional instrumentation, and with careful calibration of instruments as well as the use of instruments from differentmanufacturers in the same general location on an aircraft, there is less room for assumption and interpretation as well asfor theoretical models to bias the understanding of flight research data than otherwise would be the case. But wheneveraeronautical researchers use instruments in an experimental environment, there is always a need to spend a lot of timeunderstanding what those instruments measure and how they do it to ensure accuracy in using data from them.

Portrait of Dr.Hugh L. Dryden acouple of yearsafter he made theremark quoted inthe narrative.(NASA photo E-4248)

18

Introduction

The NASA Flight Research Center(FRC—formerly the NACA High-SpeedFlight Station [and now known as theNASA Dryden Flight Research Center])has been involved in experimental andresearch flight testing for over 27 years.FRC’s experience began with the X-1series of aircraft and extended through allthe manned X-series aircraft, the D-558series, and most recently the liftingbodies. Other experience was also gainedwith unusual vehicles such as paraglidersand the Lunar Landing Research Vehicles.FRC has flight tested vehicles withoperating speeds ranging from zero to4,500 miles per hour and altitude rangesfrom ground level to 354,000 feet. Over5,000 research flights have been made inover 60 different types of researchaircraft. Only three aircraft and two pilotshave been lost during research testing andnone of these losses were attributable tonegligence or inadequate planning orpreparation.24

This is a remarkable record, especiallyconsidering the extremely hazardousnature of the testing FRC has beeninvolved in. FRC has, however, had anumber of accidents and incidents notinvolving the loss of an aircraft or apilot. Numerous problems have beenencountered in flight that wereunpredicted or unanticipated. This, ofcourse, is the justification for flight-testing. This document will describesome typical examples of the kinds ofproblems we have encountered. Theintent is to make people aware of thekinds of problems we have encounteredso that these same mistakes will not berepeated as they have been so often inthe past.

The kinds of problems that we haveencountered can be categorized intohardware problems, aerodynamic prob-lems, and what might be called environ-mental problems. Hardware problems arethose where a component or subsystemdoes not perform up to expectations. Thecomponent or subsystem doesn’t functionproperly or fails completely. Aerodynamicproblems are those encountered becausethe wind-tunnel predictions were notaccurate or were misinterpreted or eveninadequate. Environmental problems arethose that show up only in flight. Theygenerally result from a lack of foresight orunderstanding of the effects of the envi-ronment on a subsystem or component, orthe vehicle itself.

Of the three types of problems, theemphasis will be on aerodynamic- andenvironmental-type problems. Tworesearch aircraft have been selected as theprime examples, the HL-10 and the X-15.The HL-10 was an unconventionalconfiguration with state-of-the-art off-the-shelf subsystems. Its problems, as youmight suspect, were aerodynamic innature. The X-15 was a relatively conven-tional configuration but most of its sub-systems were newly developed and manypushed the state of the art. Its problemswere mainly with subsystems. Both ve-hicles explored new flight regimes.

Aerodynamic Problems

Figure 1 illustrates an example of aserious problem encountered on the firstflight of the HL-10—flow separation. Theflow separation occurred at the junction ofthe tip fin and the fuselage. It occurred inflight as the pilot began his practice flareat altitude. When this occurred, the pilotessentially lost all pitch and roll control.

ProblemsEncounteredin FlightResearchMilt Thompson

24 Milt did not specify, but presumably he meant Howard Lilly’s crash after takeoff due to compressor disintegration inthe D-558-1 No. 2 on 3 May 1948, which resulted in Lilly’s death; the crash of the M2-F2 without loss of life on 10 May1967; and Michael Adams’ fatal accident in X-15 No. 3 on 15 Nov. 1967. Although badly damaged, the M2-F2 was notlost and was rebuilt with a center fin to make it more stable and a more successful research airplane. This list does notinclude the deaths of Air Force Maj. Carl Cross and NASA pilot Joe Walker as a result of a mid-air collision between anXB-70A and an F-104N in 1966 because that did not occur as part of a research flight.

19

He had almost full right aileron in, and hewas rolling slowly to the left. In this case,we were lucky. The vehicle recovered byitself since the flow separation alsocaused a nose-down pitching moment thatlowered the angle of attack, causing theflow to reattach. A detailed reassessmentof the wind-tunnel data revealed someslight evidence of a potential separationproblem at the flight conditions thatproduced it; however, a substantialamount of additional wind-tunnel testingwas required to confirm this and define afix.

On that same flight we had longitudinal-control-system limit-cycle and sensitivityproblems. The pilot used only one inch oflongitudinal stick deflection from flareinitiation at 300 knots to touchdown at200 knots. The sensitivity and control-

system limit-cycle problems were prima-rily a result of the elevon effectivenessbeing higher than anticipated. I say“anticipated” rather than “predicted”because the measured effectivenesscompared quite well with that measuredin the small-scale wind tunnel; however,we had chosen to believe the full-scalewind-tunnel results. This is an interestingcase since the full-scale wind-tunnel datawere obtained using the actual flightvehicle as the model, and the Reynoldsnumber range was from 20 to 40 million.The small-scale model was a 0.063-scalemodel (16 inches long), and the Reynoldsnumber range was an order of magnitudelower—2 to 4 million. Flight Reynoldsnumbers ranged from 40 to 80 million.25

One might question whether it could be acompressibility effect, but one wouldn’t

25 Reynolds number, named after Osborne Reynolds, is a non-dimensional parameter equal to the product of the velocityof, in this case, an airplane passing through a fluid (air in this instance), the density of the fluid, and a representativelength, divided by the fluid’s viscosity. In shorthand, this is the inertial forces divided by the viscous forces for the massof air acted upon by the vehicle. Among other uses, it served to compare data from wind-tunnel models with that fromfull-sized airplanes or components. The Reynolds number was not determined solely by the viscosity of the air. A largetransport aircraft, for example, would have a much larger Reynolds number when flying through air at a given altitude,location, and time than would a small model simply because of the difference in size and the amount of air displaced.Furthermore, the Reynolds number would be much larger at the rear of a flight vehicle than at the front.

Figure 1: Schematicshowing flowseparation in theoriginal HL-10design and themodification to theHL-10 that solvedthe problem. This isnot the exactillustration Miltintended for thispaper, but it showswhat he was talkingabout. (Originaldrawing by DaleReed; digitalversion by theDryden GraphicsOffice).

20

expect compressibility effects at 0.4 Machnumber. We did, however, see compressibil-ity effects as low as Mach 0.5 on the X-24A. It should be noted that flight-measuredlongitudinal stability was higher thanpredicted by either the small-scale or full-scale tunnels, whose data agreed quite well.

In the case of aileron characteristics, againthe small-scale and full-scale tunnelresults agreed quite well; however, theflight-measured results were higher thaneither, and again we had control-systemlimit-cycle and sensitivity problemsduring flight. The predicted subsoniclongitudinal trim was off by approxi-mately four degrees in angle of attack dueto a combination of discrepancies in zero-lift pitching moment as well as staticstability and control effectiveness. Dis-crepancies in longitudinal trim of roughlythis same magnitude were observed ineach of the lifting bodies.

The HL-10 configuration had over 8,000hours of wind-tunnel testing. One modelthat was tested was actually larger thanthe flight vehicle—28 feet long as com-

pared to a 20-foot flight vehicle, or 1.4scale. The actual flight vehicle was testedin the 40X80-foot tunnel at Ames Re-search Center. You couldn’t get bettermodel fidelity, and yet we still sawdiscrepancies between the predicted andflight-measured data.

Aerodynamic discrepancies were notrestricted to the HL-10 configuration. TheHL-10 was simply used as an example.Each of the other lifting bodies exhibitedsimilar kinds of discrepancies betweenpredicted and flight data. The M2-F2wind-tunnel tests were conducted andanalyzed by another team of expertsincluding people such as [Alfred J.]Eggers, [Clarence] Syvertson, [Jack]Bronson, [Paul F.] Yaggy, and manyothers, and yet again, the predictions werenot perfect. The X-24A configuration wasdeveloped and tested by the MartinCompany for the United States Air Force(USAF). It was a highly optimized andfinely tuned configuration. The X-24Adesigners, for example, detected thepotential for a flow separation problem atthe fin-fuselage juncture and tested overtwenty different fin leading-edge configu-rations before settling on the final leadingedge for the flight vehicle. As meticulousas these designers were, we still saw someslight evidence of unpredicted flowseparation.

On the X-24A, we also observed adiscrepancy in aileron yawing-momentderivative. In terms of the actual numeri-cal value, the discrepancy was small. Interms of percentage, it was an error by aminus 100 percent. In terms of vehiclehandling qualities, the discrepancy was

Figure 2: Control-lability boundariesfor the X-24B atMach 0.95. A

A Adapted and simplified from Christopher J. Nagy and Paul W. Kirsten, “Handling Qualities and Stability Derivativesof the X-24B Research Aircraft” (Edwards AFB, CA: AFFTC-TR-76-8, 1976), p. 56. It is obvious that this was notprecisely the figure Milt had in mind, but it illustrates his point. Note that Nagy and Kirsten comment on p. 54,“Although modeling of the rocket exhaust conditions was not exact (hence the resulting data was not considered to beaccurate), the results were used as guidelines to evaluate the potential loss of stability with the rocket engine on.”They added, “The comparison of the handling qualities boundaries before and after the flight-test program exemplifiesthe need for an incremental envelope expansion approach to flight test of new aircraft. Boundaries determined byactual lateral-directional stability were considerably more restrictive than they were predicted to be. Although power-on wind tunnel test did indicate an effect of the rocket engine, tests of this nature are not conducted for most testprograms.”

21

extremely significant since it caused a pilotinduced oscillation (PIO) in flight.

We have seen some unusual power effects ineach of the lifting bodies. These includelongitudinal trim changes of as much as fourdegrees in angle of attack. Most recently, wehave observed a loss of directional stability inthe X-24B in the Mach number range from 0.9to 1.0 and higher. This is illustrated in Figure 2.These power effects were not due to thrustmisalignments [although some existed]. They

were the result of rocket-plume induced flowseparation over the aft fuselage, fins, andcontrol surfaces. This phenomenon is appar-ently peculiar to lifting-body configurations ornon-symmetrical shapes, since it had not beennoted in earlier rocket aircraft or in missiles toany significant extent.26

Aerodynamic discrepancies are not limited tolifting-body configurations. We saw a reversalof sign in yaw due to aileron on the XB-70 asillustrated in Figure 3. Aileron characteristics

26 Milt seems to be forgetting here that there were rocket-plume effects in the D-558-2 when any other cylinder of theXLR-8 rocket engine fired in a combination including the top cylinder. These effects were most severe at the highestMach number tested—approximately Mach 1.6. The plume effects were small when only the two middle cylinders firedtogether in a horizontal plane. See Chester W. Wolowicz and Herman A. Rediess, “Effects of Jet Exhausts on Flight-Determined Stability Characteristics of the Douglas D-558-II Research Airplane” (Washington, DC: NACA RMH57G09, 1957), esp. pp. 16-17. There apparently were also plume effects on rockets such as the Saturn V.

Figure 3: Variationsof XB-70-1 flight-based and pre-dicted aileronyawing-momentcontrol derivativewith Mach numberin hypotheticalclimbout profile. B

B Taken from Chester H. Wolowicz, Larry W. Strutz, Glenn B. Gilyard, and Neil W. Matheny, “Preliminary Flight Evaluation of theStability and Control Derivatives and Dynamic Characteristics of the Unaugmented XB-70-1 Airplane Including Comparisons withPredictions” (Washington, DC: NASA TN D-4578, 1968), p. 64. This may not have been the precise figure Milt had in mind, but itillustrates his point, showing that the predicted aileron yawing-moment control derivative was positive (proverse), whereas the flight-based values were negative (adverse) from a Mach number of about 0.90 through the supersonic range.

Figure 4: Calcu-lated decrement/increment of lift-to-drag ratio resultingfrom the differencebetween predictedand measured basepressure coeffi-cients in the XB-70. Only the lift-to-drag ratio incre-ment in the shadedregions is used forrange-incrementcalculations. C

C Taken from Edwin J. Saltzman, Sheryll A. Goecke, and Chris Pembo, “Base Pressure Measurements on the XB-70 Airplaneat Mach numbers from 0.4 to 3.0” (Washington, DC: NASA TM X-1612, 1968), p. 31. Again, this may not have been theexact figure Milt intended to use, but it makes his point. Notice that there was a favorable increment in lift-to-drag ratio at

22

of delta-wing aircraft have historicallybeen hard to predict since the days of theXF-92, one of the first delta-wing aircraft.A discrepancy in aileron characteristicsmay not seem too significant, and yet anaircraft (a B-58) was lost during theflight-test program because of thisparticular error in prediction.

The B-70 drag discrepancy shown inFigure 4 resulted in a 50-percent reduc-tion in predicted range. This is an excel-lent example of a discrepancy that

couldn’t be completely resolved evenafter the fact. It was suspected that thediscrepancy was primarily due to flexibil-ity or aeroelastic effects. After the flightprogram was completed, a new modelwas constructed and tested in an attemptto get better correlation between wind-tunnel and flight data. The best correla-tion that could be obtained—even know-ing the answer in advance—was 10percent on overall drag, and that stillmeans a big error in overall flight range.

Aerodynamic discrepancies have notdisappeared with time. During tests of theF-8 Supercritical Wing in the 1971-72time frame, we saw numerous discrepan-cies between wind-tunnel and flight dataeven at the optimized design cruisecondition of 0.99 Mach number. Figure 5shows comparisons of wind-tunnel andflight data for some of the aerodynamicderivatives where significant discrepan-cies occurred. Admittedly, the designMach number region is extremely hard towork in. Yet the 50- to 100-percent errorsin such basic stability derivatives as theone for sideslip could hardly be consid-ered acceptable accuracies. At other thandesign cruise condition, a large discrep-ancy was observed in aileron effective-ness and smaller but still significantdiscrepancies in the pitching-momentcoefficients.

Here again, we had a master of the craft,[Richard] Whitcomb, conducting thewind-tunnel tests and analyzing theresults before the fact. Admittedly, theairfoil concept was somewhat revolution-ary; however, Whitcomb had essentiallyunlimited access to any wind-tunnelfacility he needed and should therefore

cruise speeds above Mach 2.5 but that at low supersonic speeds near Mach 1.2 there was the very unfavorable decrementMilt talks about. Thus, even though ground researchers had overestimated base drag at cruise speeds, their underestimate atlow supersonic climbout speeds seriously reduced the aircraft’s range.

Figure 5: Lateral-directional deriva-tives as a functionof angle of attack inthe F-8 Super-critical Wingaircraft. D

D Taken from Neil W. Matheny and Donald H. Gatlin, “Flight Evaluation of the Transonic Stability and Control Characteris-tics of an Airplane Incorporating a Supercritical Wing” (Edwards, CA: NASA Technical Paper 1167, 1978), pp. 42, 43, 46.Once more, this may not be the precise illustration Milt intended to use, but it shows roughly the level of discrepancybetween wind-tunnel and flight data that he had in mind and does so for some of the derivatives he mentions.

23

not have any good excuse other than thefact that the wind tunnels still have someobvious shortcomings.27

More recently, discrepancies in very basicstability characteristics have been observedin the newest and latest aircraft. The[Y]F-16 and [Y]F-17 showed a substantialdifference between predicted and flight-measured longitudinal stability throughout amajor portion of the usable angle-of-attackenvelope.28 The B-1 exhibited much moreadverse yaw due to roll control on its firstflight than had been predicted. This discrep-ancy showed up on a configuration and at aflight condition that should have beenhighly predictable.

In summation, we just haven’t seen evi-dence to prove that wind-tunnel predictionsare improving that much in accuracy orthat we have gotten that much smarterin anticipating all the potential aerody-namic problems.29

Environmental Problems

To turn to the subject of environmentalproblems, I would like to review some that weexperienced with the X-15. There have beennumerous reports published and many papersgiven on the results of the X-15 flight program,but nothing has been published that summa-rized all the problems we had. We went backinto the records to try to identify all the variousproblems. Before discussing them, however,we must recognize that the X-15 was quite anadvanced aircraft for its time, except in termsof its configuration. This was pretty conven-

27 Note that Whitcomb discussed some of the preliminary differences between wind-tunnel and flight data in his “Com-ments on Wind-Tunnel-Flight Correlations for the F-8 Supercritical Wing Configuration,” in Supercritical Wing Technol-ogy: A Progress Report on Flight Evaluations (Washington, DC: NASA SP-301, 1972), pp. 111-120, a report that wasstill classified when Milt was writing this document.

29 If Milt were writing today, he would no doubt add the results of Computational Fluid Dynamics (CFD) to his comments,since the results of CFD have also failed to anticipate many potential aerodynamic problems in vehicles that have used it as adesign tool. On the other hand, many people would argue that wind-tunnel predictions have improved significantly, partly asa result of comparing previous predictions with the actual results of flight research, partly from other sources.

28 The YF-16 and YF-17 were in a very close competition for an Air Force contract, which the YF-16 won in January 1975,and this led to the production F-16As—a fact that Milt could not have known at the time of his writing this document. TheYF-17 later led to the Navy/Marine Corps F/A-18. See the Air Force Flight Test Center History Office’s Ad Inexplorata: TheEvolution of Flight Testing at Edwards Air Force Base (Edwards AFB, CA: AFFTC/HO, 1996), pp. 27-28.

tional as can be seen in Figure 6 with thepossible exception of the upper and lowervertical tails, which were wedge-shaped. Theaircraft had a unique structure for dealing withaerodynamic heating, and it featured many

Figure 6: Three-view and cutawaydrawings of theX-15. E (Seepage 25)

24

new systems that were required to fly to thelimits of the flight envelope. These includedthe reaction control system, the inertial system,the LR-99 rocket engine with throttling, theskid [landing] gear, auxiliary power units, side-arm controller, the ball nose to provide air data,and the MH-96 Flight Control System (a ratecommand system with adaptive gain [thatappeared only in the X-15 Number 3]).

As you might suspect and as will be discussedlater, our major problems were with thesystems rather than with configurationaerodynamics. In most areas, the aerodynamicswere pretty much as predicted. There wasgood correlation between wind-tunnel andflight data throughout the entire Mach range.

The only significant difference was in basedrag, which was 50 percent greater thanpredicted. Again, a characteristic historicallyhard to predict. The lift-to-drag ratio (L/D),however, was higher than predicted—4.5 ascompared to 4.2—which indicates that therewere compensating factors not evident in thewind-tunnel data. Ground-effect and gear-down L/D were also inaccurately predicted.One other important bit of data obtained duringthe X-15 flight program was aerodynamicheating data, which revealed that actual heattransfer rates were substantially lower thanpredicted by theory.30

Figure 7 addresses the X-15 program andsome of the problems encountered. It

E This was taken from Wendell H. Stillwell, X-15 Research Results with a Selected Bibliography (Washington, DC: NASASP-60, 1965), p. 3.

30 Another inaccurate prediction stemmed from the theoretical presumption that the boundary layer (the thin layer of air close to thesurface of an aircraft) would be highly stable at hypersonic speeds because of heat flow away from it. This presumption fostered thebelief that hypersonic aircraft would enjoy laminar (smooth) airflow over their surfaces. Because of this, many designers computedperformance and heating for the hopeful case of laminar flow. At Mach 6, even wind-tunnel extrapolations indicated extensivelaminar flow. However, flight data from the X-15 showed that only the leading edges of the airfoils exhibited laminar flow and thatturbulent flow occurred over the entire fuselage. Small surface irregularities, which produced turbulent flow at transonic and super-sonic speeds, did so equally at speeds of Mach 6. Thus, designers had to abandon their hopeful expectations. On this matter, see JohnV. Becker, “The X-15 Program in Retrospect,” 3rd Eugen Sänger Memorial Lecture, Bonn, Germany, Dec. 4-5, 1968, pp. 8-9; AlbertL. Braslow, “Analysis of Boundary-Layer Transition on X-15-2 Research Airplane” (Washington, DC: NASA TN D-3487, 1966).

Figure 7: X-15Flight Program,found in the MiltThompson Collec-tion of the DrydenHistorical Refer-ence Collection.Note that M=Machnumber, k=1,000,Max q=maximumdynamic pressure,andhmax=maximumaltitude.

25

indicates some milestones in the programand correlates them with the flightnumbers. There was a total of 199 flightsmade with the three aircraft. In the middleof the figure, the various phases of theprogram are shown. On the bottom of thefigure, some of the problem areas arelisted. The bars and dashed lines indicatewhere the problems occurred during theprogram. The solid bar indicates continu-ing significant problems. The dashed lines

indicate continuing minor problems, andthe asterisks represent unique problems.

The Number Two aircraft was severelydamaged on its 31st flight—the 74th X-15flight of the program as a whole—andwas subsequently rebuilt and modified toachieve higher performance.31 It beganflying again shortly after the halfwaypoint in the program as shown in the testphase part of the figure.

31 On the 9 Nov. 1962 flight, Jack McKay could not get the XLR-99 engine to advance its throttle setting beyond 30 percent and hadto make an emergency landing at Mud Lake under X-15 mission rules. He was unable to complete his jettison of propellants after

Figure 8: Timehistory of the flareand touchdown ofX-15-1 on its firstflight.

F

F This was taken from Thomas W. Finch and Gene J. Matranga, “Launch, Low-Speed, and Landing Characteristics Determined from

the First Flight of the North American X-15 Research Airplane” (Washington, DC: NASA TM-195, 1959), Fig. 13 on p. 26. Thisprobably is the figure Milt had in mind to illustrate his point. On pp. 9-10, Finch and Matranga state:

From [the] figure it is obvious that a severe pitching oscillation was induced near the end of the flap cycle. Reducedlongitudinal trim was required as the flaps were being deflected, and the pilot added further airplane nose-down trim toavoid flaring too high. Apparently the oscillation became more severe because of the control input at about 18 secondsbefore touchdown. From this point, the pilot was not able to anticipate the oscillation accurately, which may have beenaggravated by the fact that the control surface was rate-limited to 15° per second. . . . The transient in pitch covered anangle-of-attack range from -1° to 13°, with the amplitude as high as ±5°.

26

The envelope expansion to design speed,altitude, and dynamic pressure concludedwith the 53rd flight—rather early in theprogram. It didn’t take many flights toachieve these design conditions when oneconsiders that 30 of the first flights weremade with the interim engine—the LR-11[two of which flew on each flight], whichlimited the maximum performance toabout Mach 3. Once the LR-99 enginewas available, the flight envelope wasrapidly expanded—roughly half a Machnumber at a time to the design speed ofMach 6, and 30,000 feet at a time to thedesign altitude of 250,000 feet. Afterachieving the design conditions, we beganexploring the total flight envelope andcontinued to expand the altitude envelope,finally achieving an altitude of 354,[200]feet. We had the total impulse available togo even higher; however, the reentry wasbecoming somewhat critical. We alsobegan exposing the aircraft to greater heatloads, going to high Mach numbers atlower and lower altitudes. We also begancarrying piggyback experiments on theaircraft before the 80th flight and from the130th flight on. That’s essentially all thatthe X-15s were used for after that pointsince we had completed the basic aircraftflight-test program.32

Control-System Problems

The first major flight problem we had waswith the control system, and this occurredon the first flight. The pilot got into a PIOduring the landing flare. Very simply, thePIO was due to the limitation of thehorizontal stabilizer to 15 degrees persecond of surface rate and the pilot wasasking for more than 15 degrees persecond as illustrated in Figure 8. The

airplane was almost lost on the first flightas a result of this.

The PIO was a surprise because thesimulation used to define the maximumcontrol surface rate requirement did notadequately stimulate the pilot to get hisown personal gain up. In the real environ-ment on the first flight, his gain was wayup. He was really flying the airplane. Ourexperience has verified that the pilotgenerally demands the maximum controlsurface rates for a given vehicle in theperiod just prior to touchdown, at least forunpowered landings.

In retrospect, this isn’t hard to understand.Just prior to touchdown, the pilot is tryingto control the flight path to within one-half a degree or so to make a goodlanding, five feet per second or less. Anunpowered landing, in our opinion, is oneof the most demanding tasks required of apilot and a flight-control system. Theproblem is that you can’t adequatelysimulate it. Visual simulators don’t havethe necessary resolution near the ground,and even sophisticated flight simulatorssuch as variable stability aircraft can’tseem to get the pilot’s personal gain upsufficiently to thoroughly assess a poten-tial PIO problem in landing. A PIOproblem may not be evident until the firstreal unpowered landing is made. Evenwith a successful first landing one can’tbe sure the problem doesn’t exist, sincewe have found that individual pilot gainvaries considerably and another pilot mayinduce a PIO. The control system of theX-15 was modified after the first flight toincrease the horizontal control surface ratefrom 15 degrees per second to 25 degreesper second.

shutting down the engine, and the excess weight caused him to be high on airspeed. He touched down at 296 miles per hour ratherthan the normal 230. The result subjected the main gear to both a rebound and a high aerodynamic load, causing the left landing gearto collapse, and eventually the aircraft flipped over on its back, injuring McKay and causing the Number Two aircraft to be rebuiltand modified. See Milton O. Thompson, At the Edge of Space: The X-15 Flight Program (Washington and London: SmithsonianInstitution Press, 1992), pp. 227-230 and his further discussion of this flight below in this study.

32 On the other hand, it could be argued that the hypersonic aircraft itself was the primary experiment from flight 1 toflight 199, even when it was carrying piggyback experiments.

27

The next major control-system problemdidn’t show up until the 23rd flight. Theproblem was a structural resonanceproblem wherein the Stability Augmenta-tion System (SAS) was responding to thevibration of the structure on which theSAS box was mounted. This self-sustain-ing control-system problem almost shookthe airplane apart during an entry from169,000 feet. We subsequently added anotch filter to eliminate this problem. Itsurprised us because we did not conduct astructural resonance test. The X-15 SASwas one of the first high-gain, high-authority systems capable of respondingto structural frequencies. Since thatoccurrence, we always conduct resonancetests of an aircraft with SAS on to lookfor such problems.

Structural Problems

We also had basic structural problems. Onthe 4th flight, one of the thrust chambersexploded during engine start, causingengine damage and a fire. The pilot shutdown all the thrust chambers and jetti-soned fuel before making an emergencylanding on Rosamond Dry Lake. He wasunable to jettison all the propellantbecause of the steep nose-down attitude.As a result, the aircraft broke behind thecockpit on nose-gear touchdown.

The aircraft designers had failed toanticipate the nose-down jettisonproblem. The aircraft were subse-quently beefed up to handle this prob-lem.

Landing-Gear Problems

Landing-gear problems plagued usthroughout the X-15 flight program.The landing gear failed on the firstlanding. The landing gear was reworkedand performed satisfactorily until the74th flight. On that flight, after launch,the engine would only develop 30percent thrust. The pilot was told toshut down the engine, jettison propel-lants, and make an emergency landing

Figure 9: TypicalX-15 landing usingwing flaps. Nose-gear touchdown at atime intervalbetween initialmain-gear contactand nose-gearcontact of 1.35seconds (flight 1-30-51 on June 27,1962). Taken fromRichard B. Noll,Calvin R. Jarvis,Chris Pembo, andBetty J. Scott,“Aerodynamic andControl-SystemContributions to theX-15 AirplaneLanding GearLoads” (Washing-ton, DC: NASA TND-2090, 1963), p.26.

at the launch lake. Again, the pilot wasunable to jettison all the propellantsand, to compound the problem, thelanding flaps did not extend whenselected. The main gear failed shortlyafter touchdown and subsequently, thenose gear failed and the aircraft endedup on its back.

This gear failure resulted primarilyfrom the high-speed touchdown due tothe flap failure, and the high grossweight. Touchdown speed was almost300 miles per hour. At main-geartouchdown, with skid-type gear, thenose tends to slam down rather rapidly.

28

As the nose starts to pitch down, theSAS applies nose-up elevator to coun-teract the nose-down pitching moment.The airload at this high speed, resultingfrom the extreme deflection of the hori-zontal stabilizer located immediatelyabove the main landing gear, plus theairload due to the negative three-pointaircraft attitude, added to the normalrebound load from nose-gear impact, wassufficient to break the main landing gear.

A typical time history of loads on themain landing gear is shown in Figure 9.The air load problem due to SAS responsewas not fully appreciated in the initialdesign. A squat switch was later includedto deactivate the SAS on main-geartouchdown. The squat switch workedquite well, but as the airplanes gainedadditional weight during the program dueto added instrumentation, add-on experi-ments, and required modifications,additional fixes were required. The pilotswere first asked to push forward on thestick at touchdown to relieve the air loadson the main landing gear. Later, a stickpusher and a third skid were added toprevent landing-gear failure. We were stillhaving gear problems when the programended after nearly 200 flights.

Aerodynamic Heating Problems

We had a number of problems associ-ated with aerodynamic heating. Theybegan showing up as we intentionally

subjected the airplanes to high heatingrates and temperatures. We had twowindshields shatter, becoming com-pletely opaque as shown in Figure 10,and four that cracked during flight. Theshattering was due to failure of theglass itself at the high temperatures. Aninappropriate choice of material was thecause. The cracking was due to distor-tion of the window frame at hightemperatures. The support structure forthe windshield glass was finally rede-signed.

We had a problem with canopy seals.When the cabin was pressurized, thecanopy leading edge deflected up justenough to allow the air to get to thecanopy seal. At speeds above Mach 3, theair was hot enough to burn the seal,resulting in the loss of cabin pressure. Thefix for this was to add a lip over the front ofthe canopy leading edge that prevented theair from impinging on the canopy seal.

We had a problem with local heating on thewing leading edges. Expansion gaps in thewing leading edge were designed to allowfor the expansion due to aero[dynamic]heating. These gaps, however, triggeredturbulent flow, which caused a hot spotdirectly behind the gaps. This caused thewing skin behind the gap to expand and popthe rivets holding the skin to the leadingedge. Gap covers were added to eliminatethis problem, but it persisted.

Aerodynamic heating also caused problemswith the landing gear. The first problem wasdue to distortion of the nose-gear door. As theairplane got hot, the nose gear door tended tobow, opening a gap between the rear lip of thedoor and the fuselage skin behind the door.This allowed ram air to enter the nose gearcompartment. The hot air cut through electricalwiring and tubing like an acetylene torch. Thenose-gear door and its supporting structurewere finally modified to eliminate this prob-lem.

Another landing-gear problem due to aerody-namic heating resulted in the nose-gear-scoop

Figure 10: Anoteworthy scarfrom the X-15’sfirst flight to Mach6 was this crackedouter panel on theright side of thewindshield.(NASA photoE-7508)

29

door opening at a Mach number of 5.3. Thenose-gear-scoop door was a small doordesigned to assist the extension of the nosegear aerodynamically. When it opened, itscooped ram air into the nose-gear compart-ment. At that speed, the air became hotenough to burn the tires off the nose wheels.The scoop door released due to distortion ofthe uplock linkage system under heatingloads. The uplock system required acomplete redesign.

The scoop door opened another time atMach 4.3 on the modified Number TwoX-15 because of a different problem withthe uplock system. Again, the nose-wheeltires burned and when the pilot extended thelanding gear just prior to touchdown, thenose gear extended very slowly. A nose-gear-up landing was barely averted becausewe had a sharp chase pilot who called theX-15 pilot to hold off until the nose gearwas fully locked. The Number Two aircraft,which had been modified and rebuilt after agear failure that resulted in a roll-over, thushad other gear problems attributed to aero[dynamic] heating.

The nose gear extended in flight at Mach4.3 due to insufficient allowance foradditional structural expansion in thelanding-gear deployment cable system. Thefuselage had been lengthened, but addi-tional compensation for fuselage expansionhad not been included in the landing-gear-cable release system. In another incident,the right-hand main landing gear de-ployed in flight at Mach 4.5 when theuplock hook broke as a result of thebowing of the main-landing-gear strut.The main landing gear on the modifiedNumber Two aircraft had been lengthenedto accommodate the supersonic combus-tion ramjet engine. The additional bend-ing of the longer strut due to differentialheating on the outer and inner portions ofthe strut had not been adequately com-pensated for, and the resulting deflection

in bending of the strut caused the uplockhook to fail in tension.

Auxiliary-Power-Unit (APU) Problems

We had APU problems during the earlyaltitude-buildup phase of the program. Noone had thought of pressurizing the APUgearbox cases. The lubricating oil wasvaporizing at high altitude, and APUs werefailing because of inadequate lubrication.

During a climbout on an altitude flight, the184th flight,33 one APU shut down becauseof an electrical transient that caused anelectrical overload. When the first APUshut down, the electrical load shiftedautomatically to the second APU. Thesecond APU should have accommodatedthe additional load, but because it washeavily loaded as a result of increasedpower demands over the years, it also shutdown. The shutdown of both APUsresulted in a complete loss of hydraulicand electrical power as the aircraft wasclimbing through 100,000 feet. Theaircraft virtually disappeared. The controlroom lost radar tracking, telemetry, andvoice communications. The pilot lost theengine, all electrically driven instruments,and all control except for the manualreaction-control system operated bycables. He managed to get one APUrestarted to regain hydraulic pressure forthe aerodynamic flight-control system andsuccessfully reentered the atmospherefrom an estimated 160,000 feet altitudewith no stability augmentation system andonly a couple of instruments—a “g” meterand his barometric instruments.

MH-96 Problems

The Minneapolis-Honeywell flight-control system [MH-96] was fairlyadvanced for its time. It was a command-augmentation-type control system withadaptive gain scheduling and various

33 This section is moved from a separate heading in the original typescript entitled “Other Problems.” The original saidthis was the 154th flight, but as an anonymous reviewer of this publication correctly pointed out, it was the 184th flight,the 5 being an apparent typographical error.

30

autopilot modes. It was installed in onlythe Number Three X-15 for evaluation.

At launch on the first flight, the stabilityaugmentation portion of the systemcompletely disengaged due to an electri-cal transient. The pilot managed toreengage it, but as you can imagine, it wasquite a shock. Minor problems cropped upduring the next 37 flights of the system.On the 39th flight [125th flight in theoverall program], shown by an asterisk [inFigure 7], the system—you might say—went berserk. The horizontal stabilizersused for both pitch and roll control startedlimit cycling as a result of excessive gainthrough a deflection of ±10 degrees.During this limit cycling, which occurredat Mach 5.5,34 the aircraft was essentiallyout of control in pitch and roll and wasbeing oscillated by the motion of thehorizontal stabilizers in both pitch and rollas shown in Figure 11.

It was quite a ride. Luckily the gainfinally came down and the oscillationstopped. It took a while to find out thereason for this problem. It turned out that

the system had fooled itself into believingit was at a flight condition where maxi-mum system gain was required. Theaircraft had been trimmed in a steady-state4g [acceleration equal to four times theforce of gravity] and the pilot was notmaking any inputs. Normal gain-reducingstimuli for the flight-control system weretotally missing, and the system slowlydrifted to maximum gain. When the pilotfinally made a control input, the systemwent unstable. The electronics weresaturated. This same problem later con-tributed to the structural breakup and lossof the number three airplane. We did notimplement any specific fix for this prob-lem after the first occurrence.

A cure for the problem was, however,discovered after some intensive simulatorinvestigation, and we elected to continueto fly the system without modification.This particular problem could not beduplicated on the Iron Bird simulator afterits initial occurrence. One of the reasons,we later found, that it could not be dupli-cated was that the hydraulic pumpssupplying pressure to the Iron Bird

34 It was actually Mach=5.35. Milt probably was giving just a ballpark figure.

Figure 11: X-15No. 3 on flight 39(3-39-62), 13January 1965—the125th flight of theX-15 program—inwhich the airplanebecame uncontrol-lable in pitch androll for a shorttime.

G

G The dashed line indicates that the roll and pitch rates exceeded the recorded limits and had to be estimated. Taken from

Euclid C. Holleman, “Control Experiences of the X-15 Pertinent to Lifting Entry,” in Progress of the X-15 Research AirplaneProgram (Washington, DC: NASA SP-90, 1965), p. 72. This appears to have been the figure Milt had in mind for this pointin his narrative.

31

simulator could not equal the output ofthe aircraft pumps. On the simulator, wecould not physically rate-limit the controlsurfaces.

Another reason we could not duplicate theproblem on the simulator was that wecould not cause the system gain to hangup at maximum gain as it did in flight. Itwas finally concluded that the simulatorgain would not hang up because of all theextraneous electrical noise in the simula-tion electronics. The electrical noise in theaircraft system was substantially lower.We finally managed to duplicate theproblem on the simulator by physicallypinning the system gain at its maxi-mum. Once the problem was under-stood, the cure was obvious. The pilothad simply to reduce the gain manuallyand the limit cycle would cease. Thepilot flying the Number Three aircrafton its final flight was aware of thispotential problem and the requiredaction should it develop. However, forunknown reasons, he did not take theproper action.

Fatigue Problems

Even though the total flight programincluded only 199 flights and only anaverage of 66 flights per airplane, we hadwhat we considered a couple of fatigueproblems. The 66-flight average in realityprobably involved 200 to 300 systemcycles when you include ground check-outs and aborts. The first fatigue problemwas a rupture of the casing of the engineturbo-pump. The second was a rupture ofa main bulkhead in one of the propellanttanks.

X-15 Program Results

With regard to environmental-typeproblems, the X-15 program has defi-nitely convinced us of the desirability of abuildup-type test program when you havea lot of new systems that you are exposingto flight for the first time. If we had goneto the design speed and altitude on the

first flight and had encountered all of theheating problems and the other subsystemproblems simultaneously, we probablywould have lost the aircraft. Regardless ofall the problems we had, we did make alot of successful flights.

The pilots, because they were designedinto all systems, saved many missions andthe aircraft itself on numerous occasions.Problems notwithstanding, Dr. [Hugh]Dryden [Director of the NACA andDeputy Administrator of NASA] referredto the X-15 flight program as the mostsuccessful research airplane program inthe history of aircraft.

Control-System Problems in General

Flight-control systems are becoming moreand more an integral element of newaircraft. Even now with the currentgeneration of aircraft, the control-systemdesign has in most instances been factoredin to some extent before the configurationis finalized. It is therefore no longerpractical to allow the aerodynamicist, thepropulsion-systems engineer, and thestructures people to design an aircraft asthey have in the past since now flight-control technology has so much to offer.In a control-configured vehicle or active-controls-technology vehicle of the future,the control system will be factored intothe initial design as early and extensivelyas the vehicle’s aerodynamic, structural,performance, propulsion, stability-and-control, and other disciplines to achievethe optimum vehicle. The trend is obvi-ous. Because of this trend, the flight-control system assumes much greaterimportance in the flight testing of a newvehicle. These new systems tend to usehigher gains and authorities. They are thusmore susceptible to such things as struc-tural resonance, limit cycling, and sur-face-rate limiting.

Extensive ground testing of these systemsis required to assure that the controlsystem itself won’t destroy the aircraft inflight. The structural-resonance and limit-

32

cycle problems encountered in flight onthe X-15 are excellent examples of theseriousness of the problem. It is essential,for example, that the flight-control systembe active and monitored in all its variousoperating modes during ground-vibrationtesting. Other ground testing to defineresonance and limit-cycle boundaries isalso a mandatory requirement on anyFlight Research Center flight vehicle.

Ideally, the control surfaces should beunloaded and then loaded during these teststo simulate the total hinge-moment environ-ment that any particular surface can expectto be exposed to. For example, during thecheckout of the M2-F2 for its first flight,some lead weights were placed on the upperflaps. The flight-control system was activeat the time, and immediately, a control-surface oscillation of ±1 degree began. Thatparticular problem resulted from a slightdeflection of the power-actuator supportstructure under load. The support structurehad to be beefed up to eliminate the prob-lem. On another occasion, during limit-cycle testing of the HL-10 flight-controlsystem, a limit cycle was intentionallyinduced at approximately 12 cycles persecond. When the limit-cycle stimulationwas terminated, the limit cycle continued.The stability-augmentation system was thendisengaged in an attempt to stop the limitcycle, but it still persisted. Shutting offhydraulic power to the flight-control systemfinally stopped the limit cycle. That particu-lar problem was a result of the servo-actuator and mechanical-feedback linkagedynamics.

The individual and combined effects of allother subsystem operations on the flight-control system should be examined aswell as start-up transients of each sub-system. Fly-by-wire control systems willrequire additional pre-flight testing toensure that no spurious inputs can get intothe system. Lightning-strike problemshave yet to be defined.

Even after performing all this groundtesting, researchers should anticipate and

make provisions for handling potentialproblems in flight. For high-gain, high-authority stability-augmentation systems,it is essential in our opinion that manual-gain-changing capability be provided thepilot during the flight-research program.The HL-10 limit-cycle problem discussedearlier, resulting as it did from higher-than-predicted control effectiveness, is agood example of the need for suchcapability in the cockpit. If throughsimulation, other potential handling-quality or control problems are identified,provisions should be made to vary thequestionable parameter. The M2-F2required a rudder-aileron interconnect toachieve adequate roll power throughoutthe flight envelope. The lateral-directionalhandling qualities were extremely sensi-tive to interconnect ratio as a function ofangle of attack and dynamic pressure.Because of this and the fact that we couldpossibly have had some variations inpredicted control effectiveness and thus,effective interconnect ratio, we providedthe pilot an adjustable interconnectcontrol.

Any potential PIO problem observed insimulation dictates consideration of ameans of reducing stick gearing inconventional control systems or systemgain in command-augmentation-typecontrol systems. Stick-gearing-changeprovisions, however, are not easilyprovided in conventional controlsystems. We did not provide this capa-bility in the HL-10, although we knewfrom simulation that a control-sensitiv-ity problem might be encountered. Asdescribed earlier, the problem did arise,and only the skill of the pilot preventeda potential disaster. In command-augmentation or fly-by-wire controlsystems, effective gearing-changecapability is relatively easy to imple-ment. Thus, there should be no goodexcuse for a serious or prolonged PIOproblem in an aircraft with such asystem. Yet they have occurred on firstflights, indicating that someone didn’tface up to the facts.

33

To reiterate, any PIO potential revealedthrough simulation should be taken quiteseriously since we and others have beencaught short so many times in the past.The absolute PIO potential is extremelyhard to predict even with the most sophis-ticated moving-base or flight simulator.One of the major unknowns, as mentionedpreviously, is the pilot’s own system gain.No simulator will stimulate the pilot toanything approaching a real-life firstflight. On a first flight, the pilot’s personalgain may be an order of magnitudegreater than anything observed in simula-tion. To further complicate the problem,individual pilot gains vary substantially.One pilot, or even a series of pilots, maysuccessfully fly a vehicle without a hint ofa problem. Then, all of a sudden a pilotappears who can compete with thestability augmentation system in response.We have seen dramatic evidence of this.

Six research pilots successfully flew theM2-F1. The seventh pilot, an experiencedtest pilot, got into a divergent PIO imme-diately after takeoff on two successiveattempts to fly the vehicle. The resultantslow rolls to a landing left even the mostseasoned pilots in the world speechless.The same pilot, flying a different lifting-body vehicle, was actually able to com-pete, in response, with the stabilityaugmentation systems at one cycle persecond.

Command augmentation systems, men-tioned earlier, are becoming quite popular.They are showing up in more and more ofthe newer aircraft. One might questionwhether they are really needed in somecases. Command augmentation systemsgenerally do provide good control charac-teristics and are quite pleasant to fly. Theydo not, however, conform to MIL Specs[military specifications] in all respects,and they do have a number of insidious

characteristics. These systems tend tomask many of the cues the pilot normallyrelies on to give information or warning.

For example, a high-gain command-augmentation-type system tends toeliminate transients or trim changes due togear or flap extension, or center-of-gravity changes. This may not seemsignificant, and yet these trim changes inthe past have informed the pilot that thegear and/or flaps did indeed move whenthe appropriate lever was moved or thatthe center of gravity was not where youwanted it. A subtle thing—yet somewhatdisconcerting when you don’t have thesecues.

These same control systems tend toprovide invariant response throughout theflight envelope. This again would appearto be highly desirable; however, thevariable response of the older controlsystem warned the pilot of an approachinglow-speed stall or overspeed just throughfeel alone. These new systems feelcompletely solid up to and sometimesover the brink of disaster, and thusartificial stall-warning systems aregenerally required. Speed stability is alsogenerally lacking in an aircraft equippedwith this type of system, and unless it isartificially provided, the pilot mustcontinually monitor airspeed.

Normal dihedral effect and ground effectare also masked by systems such as these.A paper discussing many of these insidi-ous characteristics was presented at anAGARD ([NATO] Advisory Group forAerospace Research and Development)Flight Mechanics Panel meeting in 1967,sometime prior to the introduction of theF-111 into operational squadrons.35 Yet atleast one aircraft accident resulted fromeach of the insidious characteristics.Inadvertent high- and low-speed stalls

35 Milt apparently was thinking of his paper with James R. Welsh, “Flight Test Experience with Adaptive ControlSystems,” presented at the AGARD Guidance and Control and Flight Mechanics Panels, Sept. 3-5, 1968, Oslo, Norway,which was a year later than he remembered and also a year later than when the first F-111As entered service with theU.S. Air Force, although only in limited numbers.

34

resulted in F-111 aircraft losses. A mal-function in the automatic fuel transfersystem of one F-111 allowed the center ofgravity to move well aft of the normalflight range and finally sufficiently far aftto cause the aircraft to diverge. The pilotof that aircraft was completely unaware ofthe problem until the aircraft diverged.

The inadvertent stall-spin problems in otheraircraft equipped with these systems is stillmuch in evidence. For some reason, theword is not getting around, and everyonehas to learn the hard way. It’s almost as badas the old T-tail problem. The artificial cuesbeing provided to warn the pilots of im-pending stall are in many instances com-pletely inadequate, as are many of the stallinhibitor devices. There is still much morework to be done in these areas. Thesecommand aug[mentation] systems also inmost cases have to be deactivated in a spinsince they tend to apply improper spinrecovery control.

Adaptive-gain flight-control systemspotentially have problems maintainingprogrammed gains. Two different systemswith which we are familiar have hadhistories of excessive gain problems andalso too low a gain at times because ofturbulence or other external stimuli thathave not been adequately compensatedfor in the initial design.

Control Configured Vehicles (CCV) willwithout question be major contenders inthe next generation of military andpossibly commercial aircraft. The firststep has already been taken in the YF-16.The concept is completely feasible;however, these control systems must havethe predicted control moments and powerand cannot be marginal on surface ratesor hinge moments.

And, finally, automatic control systemsare not infallible. Automatic flight-control

systems are only as good as the peoplewho designed them. If the designers havenot anticipated all the possible situationsor flight conditions that the pilot andaircraft can get into, trouble can result.

Admittedly, it is usually easier to makethe desired or necessary changes throughelectronics, but it is still surprising torealize how many changes are madeduring a flight-test program on some ofthe newer systems. Twenty to fiftychanges in the flight-control-systemconfiguration are not uncommon in thesenewer systems. The changes required arein many instances minor changes ortweaking to optimize the system. We areaware, however, of some major changesthat were required such as gearingchanges as high as fifty percent of theoriginal value. The fact that majorchanges such as these are required is quitedisturbing since these aircraft are notexploring new flight regions. Thus, thepredicted aerodynamic data should begood as far as basic stability and controlderivatives are concerned, and these arethe primary requirements for the design ofa flight-control system. The reason forsuch drastic changes being required istherefore not clear. Somewhere, somehow,something is being overlooked or notbeing considered in the design process.

Finally, the primary message on these newcontrol systems is to shake, rattle, and rollthem thoroughly before flight and thenexpect problems in flight and provide thenecessary system-adjustment capability toalleviate the problem if it does occur.36

First Flight Preparation

The aerospace industry has had muchmore first-flight experience than eitherNACA-NASA or the military. It has beenonly recently that the government hasbeen directly involved in preparing a

36 Despite the fact that this section, like the rest of this document, was written about 1973 or 1974, much of what Miltsays is still applicable, although in many cases pilots have adjusted their flying styles to adapt to the circumstancesimposed by control systems, such as those in trim.

35

vehicle for and participating in firstflights, and we may therefore not be themost qualified to define the best approach.The current problem, however, is thatvery few new aircraft have been designedand flown in the past fifteen years, andtherefore even industry has had littlerecent experience. Many aerospacecompanies were essentially without a realflight-test organization during the leanyears, and some did not have companytest pilots. Thus, with the renewal ofaerospace activity, many companies hadto put a flight-test team together fromscratch. That can spell trouble.

We at the NASA FRC have been fortunatein being in a position to actually conduct anumber of first flights over the past tenyears. We made first flights on all of thelifting bodies (the M2-F1, M2-F2, M2-F3,HL-10, the X-24A, and the X-24B) aswell as the F-8 Supercritical Wing, theF-8 Digital Fly-By-Wire, the F-111Transonic Aircraft Technology, and mostrecently the F-15 Remotely PilotedResearch Vehicle. We have gained someappreciation for all of the concerns that goalong with making a first flight and havedeveloped our own procedures andground rules for qualifying a vehicle forflight. These are by no means all-inclu-sive, since in many disciplines we dependon the designer and builder for thenecessary confidence to proceed.

For example, we depend heavily on thecontractor to provide an adequate struc-ture and functional systems. We have inmany cases done the conceptual designbut have never attempted to do the detaildesign since we are not designers—a factthat others in government don’t alwaysadmit. We do monitor and analyze thedesign, do proof loading of criticalportions of the structure, and do func-tional testing of all the subsystems once avehicle is delivered. We also do all theother normal pre-flight things such as taxitests and all-up rehearsals of the firstflight with all systems operating and withall personnel at their appropriate stations.

One of the most important things we do issimulation. We have learned from experi-ence that extensive simulation is the keyto success in flying a new configurationor vehicle. We analyze all the wind-tunneldata and then start with the best fairing ofall the data. Once we complete thatevaluation, we then begin looking at theworst cases. We intentionally vary eachand every derivative over a wide range todetermine the sensitivity of the vehicle’sflying qualities to that particular deriva-tive. The range of variability we investi-gate is much broader than the scatter ofthe wind-tunnel data. From experience,we expect—or I should say we are notsurprised by—discrepancies of 25 to 30percent in predicted derivatives. Inpractice, we vary them as much as 100 to200 percent. In the case of dynamic orrotary derivatives, which are hard tomeasure both in the wind tunnel and inflight, we may vary them even more.Based on experience, we even varycombinations of various derivatives tolook for the worst possible cases. Anypotential handling-quality problemsexposed in this type of investigation arethoroughly explored to determinepossible fixes, recovery techniques,and/or, if necessary, ways to avoid theproblem area. The low-angle-of-attackPIO problem identified during earlysimulations of the M2-F2 was a classicexample. We spent many hours evaluat-ing the problem and determined that wehad two effective recovery techniques.One was to reduce the rudder-aileroninterconnect ratio and the other wassimply to pull up and increase angle ofattack. We could not easily avoid thearea since we had to go to low angles ofattack to pick up the necessary airspeedfor landing flare. We did, however, havelanding rockets as a backup in case wedid have to increase angle of attackfrom the desired pre-flare condition. Inflight, both of these recovery tech-niques were validated by necessity. Thevalue of this type of simulator investi-gation cannot be over-emphasized. Wealso feel that it is essential that the pilot

36

participate extensively in the engineer-ing simulation. We at FRC rely heavilyon the pilot for a successful flightprogram. The importance of the pilot iscritical, particularly in simulation, thedesign or evaluation of the controlsystem, and first-flight preparation.

Pilots generally have a much broader andmore objective view of the overall picture[than do other participants in flightresearch]. We are fortunate to haveexceptionally well qualified and experi-enced pilots, all of whom have made firstflights. They are all well suited to serveas either a project pilot on a new vehicleor a member of the Flight ReadinessReview Board, and in this manner, we getdouble the input. A pilot with first-flightexperience is invaluable. Unfortunately,there aren’t too many active pilots whohave first-flight experience because of thelimited number of new aircraft that havebeen produced in the last fifteen years.Following the extensive engineeringsimulation, we proceed into the proce-dural and flight-planning simulationphases. In these phases, we develop thefirst-flight plan and then look at everyimaginable failure, malfunction, oremergency. In developing the controlsystem and preparing for the first flightof the M2-F2, we spent at least a yearon the simulator, and I as pilot averagedtwo to three hours a day in the simula-tor cockpit.

We have not normally used anything otherthan a fixed-base simulator; however, wewere sufficiently concerned about the lowangle-of-attack PIO problem in the M2-F2 to also investigate it in-flight with avariable-stability aircraft. We haveresorted to variable-stability aircraft andmore sophisticated simulators on a fewother occasions; however, the simplefixed-base simulator has generally beenadequate. We thoroughly exercise theflight-control system to establish limit-cycle boundaries and ensure that we arefree of structural resonance problems asdiscussed earlier.

Another very important part of our pre-flight preparation is a Flight ReadinessReview (FRR). An FRR team is desig-nated at least six months prior to ascheduled first flight. The team isgenerally composed of members of eachof the disciplines involved (aerodynam-ics, stability and control, performance,etc.) as well as subsystem experts,instrumentation experts, and a pilot. Thechairman of the FRR team is at least asenior division-director-level individualwith a broad test background. The FRRteam members are not associated in anyway with the project team and act asdevil’s advocates. The FRR team hasunlimited access to any data, can moni-tor any tests, question any project teammember, and make any recommenda-tions on pre-flight preparation. Inessence, it has carte blanche to examinethe program. The FRR reports directlyto the Center Director, and prior toflight, it submits an oral and writtenreport. The FRR is an extremely effec-tive means of ensuring a safe first flight.

First Flight and Envelope Expansion

Our general philosophy on first flights isthat once the aircraft is airborne, weimmediately begin worrying about how toget it back on the ground again safely.Data maneuvers are of secondary impor-tance. The main area of interest during thefirst flight involves the approach andlanding tasks. Various potential failure orbackup control modes are evaluated in theapproach-and-landing configuration, asare other possible subsystem malfunc-tions. This emphasis on approach andlanding during the first flight is easilyjustified, since on each and every suc-ceeding flight an approach and landingmust be made. Subsystem malfunctionswill ultimately occur during the testprogram, and it is thus wise to assessthese potential malfunctions in a con-trolled manner as early as possible.

Our philosophy on envelope expansionis not unique. We select the most benign

37

Mach number expansion corridor andconcentrate first on validating stability,control, and handling-quality character-istics. We update the simulator follow-ing each flight [to incorporate what welearned on that flight that we did notknow before]. Once the Mach numberenvelope has been explored, the re-mainder of the flight envelope issomewhat systematically expanded interms of angle of attack, dynamicpressure, etc. In the case of the X-15,we began expanding the altitude andangle-of-attack envelope about halfwaythrough the Mach envelope expansion.Once the design Mach and altitude hadbeen achieved, we continued to expandthe altitude envelope and simulta-neously began expanding the dynamicpressure and aerodynamic heatingboundaries. As previously indicated,this type of envelope expansion istypical of the general approach usedthroughout the aircraft industry and is atime-proven way to test aircraft.

In such an incremental buildup of flightresearch, we can, so to speak, poke ournoses into a new area and if we encoun-ter a problem, we can immediately backout of that area and into a known safe-flight region where we have flownbefore. When we do encounter aproblem in flight, we come back down,analyze the data, update the simulator,and then try to determine a fix for theproblem before we again probe into theproblem area. The longitudinal-sensi-tivity problem we observed on the firstflight of the HL-10 is a good example.Flight data confirmed that we had morecontrol effectiveness than we antici-pated. We made a change in the con-trol-system gearing before the nextflight and eliminated the problem.

During the buildup test program in theX-15, we were fortunate to encounterour environmental problems one byone. We burned the canopy seal atMach 3.3, well below the design Machnumber of 6.0. We encountered the

nose-gear-door problem at Mach 5. Wesaw the first indications of the wing-leading-edge and windshield problemsat 5.2 Mach number. The first nose-gearscoop door opened at 5.5 Mach number.If we had gone to Mach 6 on the firstflight, we would probably have had allof these things happen within secondsof one another. Also, each problemwould have been more severe than weactually experienced because we wouldhave had more exposure time.

Remotely Piloted Research Vehicles(RPRVs)

The Flight Research Center has devel-oped a remotely piloted researchtechnique that was first applied in thetesting of an advanced lifting entryconfiguration, the Hyper III. Thetechnique illustrated in Figure 12includes basically a ground cockpit, anuplink for command control signals,and a telemetered downlink that closesthe control loop through the pilot’sinstruments and controls. The cockpithas all the conventional instruments andcontrols normally found in an aircraft,and the pilot thus has complete instru-ment-flight-rules flight capability. Inaddition, a forward-looking televisionmounted in the flight model providesthe ground pilot an out-the-windowview for additional reference. A high-speed computer is included in thecontrol loop to provide or exactlyduplicate any flight control systemaugmentation or automatic controlmode. This allows for a relativelysimple and inexpensive on-boardcontrol system.

This technique has recently beenapplied to spin testing. A 3/8th-scalemodel of the F-15 has been testedthroughout the achievable angle-of-attack range and intentionally spunusing several different control modes.As of this time, the model had not beendeparted or spun using the primarycontrol mode with the operational stall-

38

inhibitor system in the loop. We are stilltrying, however.

The flight program has, in our opinion,been an outstanding success eventhough we have damaged and finallylost the first flight vehicle. We antici-pated losses and originally had threevehicles constructed to ensure complet-ing the planned flight program. In thefourteen flights that have been accom-plished,37 we have thoroughly docu-mented the stability and control charac-teristics of the model from plus 40degrees to minus 20 degrees angle ofattack. We have spun the vehicleupright and inverted and departed the

vehicle at several different g-levels. Wehave validated two different spin modespredicted in the Langley spin studiesand confirmed the proposed recoverytechniques. We have not as yet com-pared model data with full-scale air-plane data, since the full-scale flightdata has so far been unobtainable. The3/8-scale-model data has so far com-pared quite favorably with wind-tunnelpredictions. The simulator developedduring the flight program is probably thefirst good spin simulator ever imple-mented. Flight results have confirmed thevalidity of the simulator, and the simula-tor can and has been used to investigateand develop new spin entry techniques.38

37 Readers who skipped the background section may like to know that the vehicle completed 27 flights by 1975, 53 bymid-July 1981.

Figure 12: Theremotely aug-mented vehicleconcept.

H

H This was taken from Dwain A. Deets and John W. Edwards, “A Remotely Augmented Vehicle Approach to Flight Testing RPV

Control Systems,” paper presented at the AIAA RPV Technology Symposium, Tucson, AZ, 12-14 Nov. 1974 (also published atEdwards, CA, as NASA TM X-56029, Nov. 1974), p. 17.

38 Kenneth W. Iliff, “Stall/Spin Flight Results for the Remotely Piloted Spin Research Vehicle,” paper presented at theAIAA Atmospheric Flight Mechanics Conference, Danvers, MA, 11-13 Aug. 1980 (AIAA Paper 80-1563) gave theresults of flight research with the 3/8-scale F-15, later redesignated the Spin Research Vehicle, most of the way throughits flights. Among other findings it reported were: “the basic airplane configuration was found to be departure and spinresistant. When control authority was increased, the model could be spun using several techniques developed with thesimulator.” Also: “The acquisition of high quality steady spin data for this vehicle was made possible by the remotelypiloted technique.”

39

Much more remains to be done in the F-15 spin program, but whether it will becompleted is questionable because ofother higher-priority flight-programcommitments. Higher-Mach-numberdepartures and spins should be investi-gated. The model should be modified toprovide more representative inertias,since the present inertias are higher thanthat required for dynamic scaling. Theinertias are properly ratioed, however.Additional stability-and-control deriva-tives should be obtained during the actualspin. Wind-tunnel data is lacking at theextreme angle of attack investigated inflight. And finally, wind-tunnel, smallscale-model, 3/8th-scale model, and full-scale data should be compared to com-pletely validate the RPRV technique. Allof this, we feel, is essential to ensure thatin the future accurate simulations cancompletely predict departure and spintechniques as well as recovery tech-niques.

The current FRC position on RPRVs isthat they can be used effectively toprovide meaningful and accurate data.They can be cost effective, and they cansave time and potentially even reduce theamount of full-scale testing required. TheRPRV technique is extremely attractivefor high-risk-type testing such as spintesting, testing of new structural concepts,testing of flutter-suppression systems, etc.A number of such programs have alreadybeen proposed, and we anticipate manymore to materialize. Our problem now isto maintain some reasonable balancebetween unpiloted and piloted flightprograms.39

There is a definite role for RPRV-typetesting based on what we now know. The

RPRV approach is, however, by no meansa panacea for flight testing. RPRV testsmay still have to be supplemented bypiloted testing, and thus it may not alwaysbe most cost effective overall to go theRPRV route. We still have a lot to do indeveloping the technique and reducing thepotential loss rate. It will also be sometime before the reliability of the on-boardpilot can be reproduced in RPRVs.

Flight-Test Errors

The remarkable safety record mentionedin the introduction does not imply that wehave been without fault. The M2-F2landing accident is a good example ofpoor judgment on our part. The M2-F2 atbest had marginal lateral-directionalhandling characteristics. The pilot initi-ated a serious PIO inadvertently on thefirst flight of the vehicle. Another PIOoccurred on a later flight during a data-gathering maneuver involving anotherexperienced research pilot. Followingthe second occurrence, we should havequit flying the aircraft and gone back tothe wind tunnel to look for a fix. Wechose instead to continue flying thevehicle without modification. A thirdpilot, who had previously flown thevehicle, encountered a PIO on finalapproach on the [six]teenth flight. Hesuccessfully recovered, but as a resultof the PIO, he was forced to attempt alanding on an unmarked portion of thelakebed.

Depth perception on the lakebed isextremely poor without known referencemarks. To further complicate the problem,a rescue helicopter was operating in theimmediate area of the modified landingsite. This distracted the pilot because of a

39 Following the flight research with the 3/8th-scale F-15/Spin Research Vehicle, the Center flew research programswith the Mini-Sniffer, the Oblique-Wing Research Aircraft, Drones for Aerodynamic and Structural Testing, and theHighly Maneuverable Aircraft Technology , among other remotely piloted vehicles. This range of vehicles showed thatalthough Milt’s comments in 1973 or 1974 were accurate as far as they went, sometimes—as he suggests below—sub-scale vehicles could be more expensive and time-consuming than full-scale programs because of the need to developminiature systems to accommodate the smaller spaces in the vehicles. In other cases, however, RPRV operations couldbe cost effective, especially if flights were planned for high data output.

40

concern about a possible collision withthe helicopter. The chase aircraft werealso forced out of position during thePIO and were not in position to call outheight above the lakebed. The result wasa gear-up landing. The pilot suffered thepermanent loss of vision in one eye. Thevehicle was rebuilt with a center verticalfin after additional wind-tunnel testspredicted a significant improvement inflying characteristics. The flight researchwith the modified vehicle proved to becompletely uneventful, since the flyingqualities were quite good, as predicted.

We have had a number of problems inthe past during towing operationsinvolving unconventional flight vehicles.Some of these were due to our overallinexperience with aerial towing. Wereinvented many of the problems wellknown to glider and sailplane peopleeven though we had an experiencedsailplane pilot as Center Director. Wealso used poor judgement when wedecided because of cost to do our owntowing. None of our pilots had any realtowplane experience and only a couplehad ever been exposed in any way totowing operations. As one might expect,we had several problems and one seriousaccident, luckily without any pilot injury.

The events leading up to the accidentstarted with the acquisition of an L-19,which we modified for towing theparaglider. The pilot selected to do thetowing had never flown an L-19, and Iwas elected to check him out. Because ofvarious schedules, only one day wasavailable for checkout. I rode throughtwo flights with the newly selected towpilot and undoubtedly overrode thecontrols, particularly the rudder, duringlanding approaches. I was then calledaway to a meeting and decided on thespot to let him take it alone. On his firstlanding, he ground looped and severelybent one main-gear strut.40

The strut was replaced and a towingflight was scheduled the following day.On the morning of the scheduled flight,the pilot who had been checked outwas rescheduled for a higher priorityflight. Another pilot was selected to dothe towing, and I gave him a quickcheckout in the L-19 prior to the actualtowing flight. After takeoff and uponreaching the edge of the lakebed, thetow-plane pilot, as instructed, began aturn to stay close to the lakeshore incase of a tow-line failure. His rate ofturn was excessive, and within sec-onds, the tow-line was hanging slackbetween us. Since we were only 300feet or so high, the only recourse wasto release and attempt a landing in thesagebrush. The vehicle was extensivelydamaged in the landing attempt.Following this accident, we revertedback to using professional tow pilots.We finally gave up towing altogetherafter two hair-raising incidents whiletowing the M2-F1.

The loss of the Number Three X-15could be attributed to some extent to afaulty experiment that we developedand flew on the aircraft. The totalexperiment did not undergo a completeenvironmental check, although acomponent of the experiment, the drivemotor that caused a problem, hadsuccessfully passed all environmentalchecks and had been used in otherpiggy-back experiments carried on theaircraft. The motor began arcing atapproximately 80,000 feet altitude onthe way up to a planned maximumaltitude of 250,000 feet. The experi-ment was supposedly isolated from allprimary aircraft systems, and yet itcaused faulty guidance-computer andflight-control-system operation. This isanother potential problem to be as-sessed with command augmentationand fly-by-wire control systems. Thefaulty experiment cannot be completely

40 Milt added here, “We found out later that that particular pilot had never flown a tailwheel airplane.” The pilot inquestion wrote beside these words, “Not true. I was the pilot involved. I flew the T-6 210 hours in pilot training.”

41

excused of partial blame for the ulti-mate loss of the aircraft.41

Another example of questionable judgmenton our part involved the maximum speedflight of the Number Two X-15 to a Machnumber of [almost] 7. The flight was madeto demonstrate the capability of the X-15 tocarry a supersonic combustion ramjet

(scramjet) engine to Mach numbersapproaching 8. Figure 13 shows the X-15A-2 with a dummy ramjet on the lowerstub ventral [, eyelid, drop tanks, andablative coating42 for what turned out to bethe Mach 6.7 flight]. In building up to flythis combined configuration, we firstmade a flight with empty tanks todemonstrate tank jettison capability.

41 Here Milt inserted a comment, “Show time history of X-15 #3.” In lieu thereof, perhaps the section of the accidentreport for the aircraft quoted in his At the Edge of Space, p. 263, will better indicate the problems that caused the pilot,Michael Adams, to lose his life in the accident:

The only unusual problem during the ascent portion of the flight was an electrical disturbance that started at analtitude of 90,000 feet and that effected [sic] the telemetered signal, the altitude and velocity computer associ-ated with the inertial platform and the reaction controls that operate automatically in conjunction with the MH-96 adaptive control system. Although the pilot always had adequate displays and backup controls, the conditioncreated a distraction and degraded the normal controls. As the aircraft approached the peak altitude of 266,000feet, it began a slow turn to the right at a rate of about 0.5 degrees per second. The rate was checked by the MH-96 system which operated normally for a brief period so that at peak altitude, the aircraft was 15 degrees offheading. Then the pilot, apparently mistaking a roll indicator for a sideslip (heading) indicator[,] drove theairplane further off in heading by using the manual reaction controls. Thus the aircraft was turned 90 degrees tothe flight path as the aerodynamic forces became significant with decreasing altitude. The aircraft continued toveer and entered what appeared to be a classical spin at an altitude of about 230,000 feet and a Mach number ofabout 5.0.

Some combination of pilot action, the stability augmentation system, and the inherent aircraft stability causedthe aircraft to recover from the spin at an altitude of about 120,000 feet and a Mach number of about 4.7. As theaircraft recovered from the spin, however, a control system oscillation developed and quickly became self-sustaining. At this time the airplane was descending at a rate of about 160,000 feet per minute and dynamicpressure was increasing at nearly 100 pounds per square foot each second. There was a corresponding rapidincrease in the g forces associated with the oscillation, and structural limits were exceeded. The airplane brokeinto many pieces while still at high altitude probably in excess of 60,000 feet, and fell to earth northeast ofJohannesburg, California.

The pilot, probably incapacitated by the high g forces[,] did not escape from the cockpit and was killed onground impact. The accident board concluded that the accident was precipitated when the pilot allowed theaircraft to deviate in heading and subsequently drove it to such an extreme deviation that there was a completeloss of control. The board believes that these pilot actions were the result of some combination of displaymisinterpretation, distraction, and possible vertigo. The board further concludes that the destruction of theaircraft was the result of a sustained control system oscillation driven by the MH-96 adaptive control systemthat caused the divergent aircraft oscillations and aerodynamic loads in excess of the structural limits. Theelectronic disturbance was attributed to the use in one of the scientific experiments of a motor that was unsuitedto very high altitude environments.

Milt said he did not believe that there was any pilot error and that he thought the accident board agreed with him. He didthink that the vertigo contributed to the accident (pp. 263-264).

42 When the X-15A-2 was rebuilt from the Number Two X-15 following its landing accident, it gained an elongatedfuselage and a small internal tank within the plane. Because the ablative coating put on the aircraft to protect it fromsevere heating on the higher-speed flights would char and let off residue at very high velocities, North American hadplaced an eyelid above the left cockpit window. The pilot would keep it closed until just before the approach andlanding, using the right window for visibility during launch and most of the remainder of the flight. Above Mach 6,however, the residue coming from the charred ablator would cover the exposed window and restrict visibility. Hence theneed to open the eyelid for approach and landing.

42

Next we made a flight with full tanks todemonstrate proper fuel and lox [liquidoxygen] transfer. The flight was abortedshortly after launch when there was noindication of flow from one of the tanks.The next full-tank flight was successfulto a Mach number of 6.3. The next flightwas with the eyelid and dummy ramjetto Mach 4.75.

At this point in the program there was astrong desire to put it all together andgo for an all-out flight. The argumentwas that we had adequately demon-strated each of the configurationchanges. We compromised for anotherflight with the ablator, eyelid anddummy ramjet. This flight raised thespeed to Mach 5. We examined theairplane after that flight and saw someindications of localized charring butnothing that we considered significant.We simply made local repairs to theablator and put it all together for aflight to Mach 7.0. Figure 14 shows[some of] the results.

A shock wave off the dummy ramjetinteracting with the boundary layercaused severe localized heating thatburned off all the ablator and burnedthrough the basic Inconel ventral fin

Figure 13: Photoshowing X-15A-2with ablativecoating, drop tanks,and dummy ramjet.(NASA photoECN 1889)

Figure 14: Result of severe heating from the Mach 6.7flight in the X-15A-2. (NASA photo E-17525)

43

structure. We almost lost the airplane.We took too big a step. We essentiallywent from Mach 5 to almost Mach 7 inone step with the dummy ramjet. Themoral is that even though we suppos-edly checked out each individualconfiguration, we should have put themall together and again worked upincrementally in Mach number. Wewould thus have appreciated thesignificance of the shock-impingementheating problem. As mentioned in theintroduction, we have been in the flight-research business over twenty-sevenyears. Many of the people who workedon the X-1s are still here, and yet we stilloccasionally get caught short. We seldomget caught on the same problem, but itseems that we never run out of newproblems.

Conclusions

Irrespective of the fact that our newgeneration of aircraft (F-14, F-15, F-16,F-17 [precursor of the F/A-18], and B-1) is not probing new frontiers, we arestill seeing discrepancies betweenwind-tunnel and flight data as men-tioned earlier. The X-1 achieved aMach number of 2.5 over twenty yearsago and we are still operating withinthat Mach region with most of our newaircraft. The aerodynamic discrepan-cies or problems we are currentlyseeing are not as dramatic as the loss ofdirectional control and consequenttumbling that Chuck Yeager encoun-tered in the X-1.43

We don’t expect surprises such as rollcoupling or aileron reversal, but we areimpressed for example with the unexpected43 This is an apparent reference to Yeager’s flight in the X-1A on 12 December 1953. Bell engineers had warned himbefore the flight that the aircraft might go out of control at speeds above Mach 2.3, but Yeager flew the X-1A to Mach2.44 (1,612 miles per hour) despite the warnings, which proved correct. He shut off the rocket engine, but the aircraftbecame violently unstable, going into something like an oscillatory spin with frequent roll reversals. He was thrownabout the cockpit as the X-1A lost altitude, falling some 50,000 feet (from an altitude of about 76,000 feet). Semicon-scious, Yeager brought the decelerating aircraft into a normal spin, recovering to level flight at about 25,000 feet.Subsequently, he landed on Rogers Dry Lakebed. A pilot with lesser skills and instincts would probably have perished.See Hallion, On the Frontier, pp. 292, 308, and Hallion, Supersonic Flight: The Story of the Bell X-1 and Douglas D-558 (New York: MacMillan, 1972), p.174.

high-angle-of-attack capability of the F-14.This was not anticipated or at least notadvertised. We still are occasionally disap-pointed in actual airplane performance. Westill see substantial discrepancies betweenpredicted and actual basic stability deriva-tives on occasion, which means we haven’timproved our predictive capabilities ortechniques substantially in the last twentyyears regardless of any new or improvedground facilities.

The many discrepancies between wind-tunnel or predicted and flight data dis-cussed in this report undoubtedly give theimpression that we are extremely criticaland/or in opposition to wind-tunneltesting. That is absolutely not the case.Wind tunnels have provided extremelygood data for many years and are continu-ing to do so. As mentioned earlier, thewind-tunnel predictions of the X-15aerodynamics were extremely good.There have been numerous aircraftdeveloped that have had no aerodynamicdiscrepancies whatsoever. We at FRCresort to the wind-tunnel people continu-ously in support of our flight research,and they have bailed us out, so to speak,on many occasions. We always requestwind-tunnel support whenever we makeother than a minor configuration change.We are completely dependent on wind-tunnel predictions in many things that wedo, such as air launch. We depend entirelyon these predictions to assure that wehave no collisions during separation.

We know, too, that the wind-tunnel peopleare not blind to their own limitations. Wehave supported, at their request, a numberof combined wind-tunnel/flight-researchcorrelation tests designed to improve their

44

predictive capability in areas where theyknow they have problems simulating thetotal flight environment.

One might ask, if we are not critical ofthe wind-tunnel results or people, whatare we critical of? The answer is that weare critical of the system. The system isnot closing the loop. When discrepanciesare noted between wind tunnel and flight,they are seldom examined in sufficientdetail to pin down the actual reason forthe discrepancies. The tendency is todismiss the discrepancy with an excusethat the tares were wrong, or that themodel was not representative, or thatpropulsion effects were not duplicated,and so on. There is little enthusiasm to goback and prove it. Both wind-tunnel andflight people are more enthusiastic aboutmoving on to the next program [thanabout investigating the problem on anexisting or completed program]. As aresult, we see reoccurring problemscontinuously.44 Our track record hasn’timproved that much except in suchcatastrophic problem areas as roll cou-pling and the T-tail deep stall problems.Significant aerodynamic discrepanciesstill show up as do performance- andcontrol-related deficiencies.

One of NASA’s primary responsibili-ties is to close the loop from the wind-tunnel to flight, and yet in manyinstances this has never been ad-equately done. An honest attempt to do

this was made with the XB-70 in tryingto explain the large discrepancy be-tween predicted and actual perfor-mance. New models were constructedand a new series of wind-tunnel testswas conducted. These tests showedsomewhat better agreement with flightresults and yet a 10 percent discrepancystill existed; this still would result in asignificant range discrepancy. Nofurther attempts were made to improvethe correlation. A more serious effort tocorrelate wind-tunnel and flight data iscurrently underway with the F-111Transonic Aircraft Technology pro-gram.45 Hopefully this will be carriedthrough to its ultimate conclusion.46

This, of course, does not provide thecorrelation we need at higher Machnumbers. We need additional validationof wind-tunnel predictions in thehypersonic speed region since the onlygood flight data is from the X-15. Wehave successfully flown some small-scale vehicles at hypersonic speeds, butthe flight data obtained was minimaland compromised by lack of accurateair data. As far as environmentalproblems are concerned, we don’tanticipate any significant new problemsin the near future since there are nocurrent plans for higher performanceaircraft.

A potentially serious problem for futureaircraft designers is emerging as the trendtoward contracted wind-tunnel operation

44 At this point in the text, Milt intended to insert a table listing some of the recurring problems, the aircraft involved, andthe period of development of the particular aircraft. Unfortunately, he apparently did not leave behind such a table, and Iam not competent to construct it.

45 Following a great deal of wind-tunnel testing at NASA’s Langley Research Center and by General Dynamics, theFlight Research Center began flight research with an F-111 on 1 November 1973. The program continued until the late1970s and was resumed in a second phase in the mid-1980s. See Richard P. Hallion, On the Frontier: Flight Research atDryden, 1946-1981 (Washington, DC: NASA SP-4303, 1984), pp. 207-209.

46 In a later document in the files he left behind, Milt wrote, “All objectives of the TACT program have been met in FY1978. In general, flight data has validated the improvements to aerodynamic characteristics as predicted by wind-tunneland calculated data.” Draft for “Annual Report of Research and Technology Accomplishments and Applications, FY1978, Hugh L. Dryden Flight Research Center,” p. 3, held in the Dryden Historical Reference Collection.

45

increases. We certainly aren’t going toproduce experts like [Richard] Whitcomb,[Robert T.] Jones, [John] Becker, [Eugene]Love, [Alfred J.] Eggers, [Clarence]Syvertson, [probably Robert W.] Rainey[,]and on and on[,] with contracted wind-tunnel operations. The aerospace contrac-tor will be pretty much on his own inassessing the quality of his data. This isnot to say that an individual contractorcannot do an excellent job on his own.Individual contractors cannot, however,have access to all the data that NASA[does] because of proprietary problems.They thus would be hampered in develop-ing equivalent experts. NASA has in thepast provided continuity and good adviceto many contractors in the development ofnew configurations. NASA has also hadthe luxury of looking at far-future andhigh-risk concepts. This is a luxuryindustry could not afford. Thus, this trendtoward contractor operation of all windtunnels should be halted and hopefullyreversed.

To make the future look even more bleak,there is pressure from some sources toeliminate all government aeronauticalresearch and development. In our opinion,that could mean complete disaster for theU.S. aerospace industry in the interna-tional sale of aircraft and would, as aresult, significantly affect the country’sbalance of payments. We feel the presentcost of government aeronautical researchand development is a very minimal andessential subsidy to our aerospace indus-try. We in the flight-research business dofeel that we will still be in business forsome time to come because of the manyother potential new problem areas alludedto earlier.

The Future

Flight research in the next five to ten yearsdoesn’t look as though it will be veryexciting. As of now, there are no big

advances planned in terms of flight-envelope expansion for future aircraft.There are no serious efforts to design atriple-sonic fighter. Even more disturbingis that there is no real enthusiasm for ahypersonic research aircraft. True, wehave flown the X-15 to hypersonicspeeds, but the X-15 did not addressmany of the critical disciplines such asstructures, propulsion, etc. KellyJohnson was pretty much on his own indesigning the superb YF-12 and SR-71aircraft. We are currently flying two ofthese aircraft in an attempt to determinewhy they fly as well as they do. So farwe have seen several discrepanciesbetween theoretical and actual data.The boundary layer conditions, forexample, are significantly differentfrom what one would predict.

Considering the number of aerodynamic,propulsion, and performance discrepan-cies we have observed, it is obvious that alot of good, sound engineering judgmentwas applied in the design of the aircraft.It is again only after the fact that we arecapable of explaining why. It is discour-aging to realize that we have to resort tooperational aircraft to obtain data toupdate theoretical and wind-tunnelpredictive capability.

We feel we critically need a new series ofresearch aircraft to stimulate new aircraftdevelopment. The early series of researchaircraft stimulated a wide variety of newsupersonic operational aircraft in the1950s. The swept-wing F-100 was basedon the success of the D-558-2. Thestraight-wing F-104 was based on the X-1and X-3 successes. The F-102 was basedon the X-4 and XF-92 results.47 Andfinally, the F-111 stemmed from themarginally successful X-5 results. Subse-quent to that extremely stimulatingperiod, there have been no real[ly]imaginative developments in aircraftconfigurations.

47 Milt had said XF-91. As an anonymous reviewer commented, this should be the XF-92. “The XF-92 was a delta wing.The XF-91 was a reverse taper prototype.”

46

The current prototype fighter programis a step in the right direction, as arethe Highly Maneuverable AircraftTechnology (HiMAT) and AdvancedFighter Technology Integration (AFTI)programs. However, none of these arereal challenging. They are all stillfocused primarily on the transonicflight regime, as were the F-14 and F-15. To us this is indicative of short-sightedness. We feel that the Vietnam-ese conflict convinced too manyadvanced planners that all futurecombat would take place at transonicspeeds. They have not acknowledgedthe fact that this was inevitable at thetime because of the thrust-to-weightratios of the aircraft involved. The newprototype fighters are capable ofsupersonic combat because of theirhigh thrust-to-weight ratios and canvirtually eat up the highly touted andtransonically optimized F-14 and F-15,with properly developed tactics.

There are those doomsayers who say thatwe will never have supersonic fightercombat. That is ridiculous. Give a fighterpilot an aircraft capable of supersoniccombat and he’ll find a way to use thatadvantage, just as he did with the early jetaircraft against the best propeller-drivenaircraft. You quickly learn not to fight in theopponent’s best arena. If one believes thatphilosophy, then the Spad is still the bestfighter ever conceived.48 In the case of theF-14 and F-15, the philosophy is that if theFoxbat49 comes down to my piece of thesky, I’ll eat him up. If he doesn’t, I’ll shoothim down from below. If that philosophyholds true, the Navy should resurrect theold B-52s, hang a hundred or so Phoenixmissiles on them and have a fleet of flyingbattleships. Better yet, rebuild somedirigibles and have worldwide air superi-ority.

48 There were several models of Spads, but undoubtedly Milt is referring to the Spad XIII built in France at the end ofWorld War I and flown by the French, Italian, and Belgian forces as well as the American Expeditionary Force. Manyfamous pilots flew it, including Captain “Eddie” Rickenbacker. The fighter served well into the 1920s in seven countries.

49 The Foxbat was the NATO reporting name of the Soviet MiG-25, which could climb to over 123,000 feet.

If the U.S. is to retain air superiority, wemust begin developing aircraft that cango up and stick their noses up theFoxbat’s tailpipe. We have the technologyin hand to build a true triple-sonic fighterthat can maneuver aggressively in theFoxbat’s arena. We can give our pilots theperch for the first time since World War I.We lost the perch in World War II andhaven’t regained it since. It’s time we did.

47

IndexAdams, Michael, 19 n., 42 n.Advanced Fighter Technology Integration (AFTI) program, 47Aerodynamic heating, 29-30, 43-44Aiken, William P., 5 n.Air Force, 3, 7, 9-10, 14, 21Air Force Flight Test Center, 7, 8, 14Altitude and equipment problems, 30, 41-42Ames Aeronautical Laboratory, 14Ames Research Center, 14, 21AP-101 flight control computer, 11Apollo, 10

Computer, 11Apt, Mel, 8Army Air Forces, 2-3

B-1, 44B-2, 11B-29, 2, 3 ill.B-50, 2B-52, 47B-58, 23Becker, John, 46Bell Aircraft, 2

Bell Aerosystems, 14Bell, Larry, 5Boeing 747, 12Boeing 777, 12Boundary layer, 25 n.Bronson, Jack, 21

Central airborne performance analyzer, 11Century series of fighters, 6, 7 illCompressibility, 1, 20-21Computational fluid dynamics, 18, 24 n.Control, see flight controlCoupling dynamics, 6, 8, 44Cross, Carl, 19 n.Crossfield, Scott, 8

D-558-1 Skystreak, 2, 3 ill.-4, 5 ill., 6, 19 n.D-558-2 Skyrocket, 5 ill., 6, 8, 22 n., 46Delta-wing aircraft, 23, see also XF-92A, F-102Design, 36, 46Digital fly-by-wire, 11-12Douglas Aircraft, 2Drag, 25, and see lift-to-drag ratioDraper Laboratory, 11Dryden Flight Research Center, Hugh L., 1, 7, passimDryden, Hugh L., 18 ill., 32

Edwards Air Force Base, 5 ill.South Base, 5 ill., 13

Eggers, Alfred J., 14, 21, 46Envelope expansion, 37-38

F-8 Digital Fly-By-Wire (DFBW) project, 11-13, 11 ill.

F-8 Supercritical Wing project, 15-16, 16 ill.F-14, 44, 47F-15, 7, 44, 47F-15, three-eighths-scale, 16-17, 36, 38-40F-16, 6, 11, 24, 35, 44F-18, 7, 11, 44F-22, 7, 11F-100, 7 ill., 46F-101, 7 ill.F-102, 7 ill., 46F-104, 7 ill., 15 ill., 19 n., 46F-111, 16, 34-35, 36, 45, 46F-117, 11First flight preparation, 35-37Flight control, 19-22, 27-28, 30-32, 32-35, 38-39 and see F-8 DFBWFlight Readiness Review, 37Flight research, discussed, 7-8, 16, 19, 36-38, passimFlight Research Center (FRC), 1, 16, 19, passimFlight test, discussed, 7-8, 19, 32, 38Flow separation, 19-20

Gemini, 10Ground effects, 25Ground-vibration testing, 33

Haise, Fred, 12Hallion, Richard P., 6Handling qualities, 21-22, 36Highly Maneuverable Aircraft Technology (HiMAT) vehicle, 47High-Speed Flight Station, 1, 19HL-10, 14 ill., 19-21, 33, 38Hyatt, Abraham, 2Hyper III, 38Hypersonic speeds, 8, 25 n., 27 n., 30, 38, 42-44, 45

Inertial coupling, see coupling dynamicsIron Bird, 31-32

Johnson, Kelly, 46Jones, Robert T., 46

Kotcher, Ezra, 2Krier, Gary, 13

L-19, 41Laminar flow, 25 n.Langley Memorial Aeronautical Laboratory, 2Langley Research Center, 16Lifting bodies, 14-15, 21, 33, 36, and see M2-F1, M2-F2, M2-F3, HL-10, X-24Lift-to-drag ratio, 22-23, 25Lilly, Howard, 19 n.Loftin, Larry, 16Love, Eugene, 46Lunar Landing Research Vehicle (LLRV), 13 ill., 14

48

Lunar Landing Training Vehicles, 14

M2-F1, 14, 34, 41M2-F2, 14, 19 n., 21, 33, 36, 37, 40-41M2-F3, 14, 15 ill., 19 n., 41Manke, John, 13Martin Company, 14, 21Me 262A, 1Mercury, 9Meteor F.Mk., 1MiG-25 Foxbat, 47Minalga, George P., 5 n.Minneapolis-Honeywell adaptive control system, 25, 30- 32Movable horizontal stabilizer, 3, 6Muroc Army Air Field, 4-5

National Advisory Committee for Aeronautics (NACA), 1-2, 5

Research Airplane Projects Panel, 8-9National Aeronautics and Space Administration, 6Navy Bureau of Aeronautics, 2North American Aviation, Inc., 9Northrop, 14

P-38 Lightning, 1, 2 ill.P-51 Mustang, 1 ill.Paresev, 41Pilot-induced oscillation (PIO), 11-13, 22, 27-28, 33-34, 36, 40-41Pilots, reliance upon, 37Pilot saves, 32PIO filter, 13Power effects, 21-22Propellers, 1

Rainey, Robert W. [?], 46Reaction controls, 25Reaction Motors, 2Remotely Piloted Research Vehicles, 38-40, and see F-15, three-eighths-scaleRocket-plume effects, see power effectsRogers Dry Lake, 4 ill., 5Root, Eugene L., 2Rosamond Dry Lake, 28Rudder-aileron interconnect, 33, 36

Saltzman, Ed, 6 n., 8 n.Shuttle Carrier Aircraft, see Boeing 747Simulation, 6, 31-32, 34, 36-37, 39-40Sound barrier, 2, 5Space Shuttle, 11, 13, 14-15

Approach and Landing Tests, 12-13Speed, 1-2, 5, 8, 11

Spin Research Vehicle, 16-17SR-71 Blackbird, 11, 46Stability and control, 21, 24, and see flight controlStability augmentation systems, 28, 33Stack, John, 2-3, 5Structural resonance, 28Supersonic flight, 2, 6, 8, 10-11Syvertson, Clarence, 21, 46

T-6, 41 n.Three-eighths-scale F-15, see F-15Towing, 41Transonic speed range, 2Transport aircraft, 6Trim, 21

Walker, Joseph, 19 n.Whitcomb, Dr. Richard, 16, 23-24, 46Williams, Walter C., 5 n.Wind tunnels, 6, 16, 20-21, 23-25, 36, 39-40, 44-46World War II, 1Wright, Orville and Wilbur, 1

XB-70A, 9 ill., 10, 19 n., 22-23XF-92A, 5 ill., 6, 23, 46XS-1, 2, 3 ill.-4XLR-11 rocket engine, 2, 27XLR-99 rocket engine, 25, 27X-1, 5 ill., 6, 44, 46X-2, 6 ill.X-3, 5 ill., 6, 8, 46X-4, 5 ill., 6, 46X-5, 5 ill., 6, 46X-15, 8-10, 9 ill., 19, 24-32, 24 ill., 29 ill., 38, 41-44

Auxiliary-Power-Unit Problems, 30Fatigue problems, 32Fuel-jettison problems, 28Landing gear problems, 28-29

X-24A, 14, 21-22X-24B, 14, 15 ill.X-29, 12X-31, 12X-33, 15X-38, 15

Yagy, Paul F., 21Yaw reversal, 22Yeager, Charles (Chuck), 4-5, 44YF-12 Blackbird, 10-11, 10 ill., 46YF-16, 24, 35YF-17, 24, 44Young, James, 8 n.Youngblood, Harold H., 5 n.

49

About the Author

Milton O. Thompson was born in Crookston, Minn., on 4 May 1926. He began flying with the U.S. Navy as apilot trainee at the age of 19 and served with the Navy for six years, including duty in China and Japan at theend of World War II. At the conclusion of his period of service with the Navy, he entered the University ofWashington and graduated with a bachelor of science degree in engineering in 1953.

After two years of employment with the Boeing Aircraft Co. as a flight test engineer, Milt became an engineerat the High-Speed Flight Station, predecessor of NASA’s Dryden Flight Research Center. He joined theStation in 1956 and was assigned to the pilots’ office two years later, serving as a research pilot until 1968.Associated with many significant aerospace projects in that period, Milt is perhaps best known as one of theinitiators of and pilots in the lifting-body program and as an X-15 pilot. He was the first pilot to fly a liftingbody—the lightweight M2-F1—and later piloted the heavyweight M2-F2. He flew the X-15 14 times, reach-ing a maximum speed of 3,723 miles per hour and a peak altitude of 214,100 feet.

In 1968 Milt moved on to become the director of Research Projects. He followed that assignment withappointment in 1975 as Chief Engineer. He remained in that position until 1993, the year of his death. Duringthe 1970s, he was a member of NASA’s Space Transportation System Technology Steering Committee, inwhich capacity he was successful in applying the knowledge from the lifting-body and X-15 programs to theShuttle design. That is, he led the effort to design the orbiters for power-off landings rather than increase theirweight with air-breathing engines for airliner-type landings. His work on this committee earned him NASA’shighest award, the Distinguished Service Medal—but one of numerous awards he received.

In the year before he died, Milt published At the Edge of Space: The X-15 Flight Program (Washington, DC,and London: Smithsonian Institution Press, 1992). After his death, Curtis Peebles edited and completed hisFlight without Wings: NASA Lifting Bodies and the Birth of the Space Shuttle (Washington, DC: SmithsonianInstitution Press, 1999), of which Curtis is listed as co-author. In addition, Milt wrote or co-authored about 20technical reports.

Monographs in Aerospace History

Launius, Roger D., and Gillette, Aaron K. Compilers. The Space Shuttle: An Annotated Bibliography. (Mono-graphs in Aerospace History, No. 1, 1992).

Launius, Roger D., and Hunley, J.D. Compilers. An Annotated Bibliography of the Apollo Program. (Mono-graphs in Aerospace History, No. 2, 1994).

Launius, Roger D. Apollo: A Retrospective Analysis. (Monographs in Aerospace History, No. 3, 1994).

Hansen, James R. Enchanted Rendezvous: John C. Houbolt and the Genesis of the Lunar-Orbit RendezvousConcept. (Monographs in Aerospace History, No. 4, 1995).

Gorn, Michael H. Hugh L. Dryden’s Career in Aviation and Space. (Monographs in Aerospace History,No. 5, 1996).

Powers, Sheryll Goecke. Women in Flight Research at the Dryden Flight Research Center, 1946-1995(Monographs in Aerospace History, No. 6, 1997).

Portree, David S.F. and Trevino, Robert C. Compilers. Walking to Olympus: A Chronology of ExtravehicularActivity (EVA). (Monographs in Aerospace History, No. 7, 1997).

50

Logsdon, John M. Moderator. The Legislative Origins of the National Aeronautics and Space Act of 1958:Proceedings of an Oral History Workshop (Monographs in Aerospace History, No. 8, 1998).

Rumerman, Judy A. Compiler. U.S. Human Spaceflight: A Record of Achievement, 1961-1998 (Monographs inAerospace History, No. 9, 1998).

Portree, David S.F. NASA’s Origins and the Dawn of the Space Age (Monographs in Aerospace History, No. 10,1998).

Logsdon, John M. Together in Orbit: The Origins of International Cooperation in the Space Station Program(Monographs in Aerospace History, No. 11, 1998).

Phillips, W. Hewitt. Journey in Aeronautical Research: A Career at NASA Langley Research Center (Mono-graphs in Aerospace History, No. 12, 1998).

Braslow, Albert L. A History of Suction-Type Laminar-Flow Control with Emphasis on Flight Research (Mono-graphs in Aerospace History, No. 13, 1999).

Logsdon, John M. Moderator. Managing the Moon Program: Lessons Learned from Project Apollo (Mono-graphs in Aerospace History, No. 14, 1999).

Perminov, V.G. The Difficult Road to Mars: A Brief History of Mars Exploration in the Soviet Union (Mono-graphs in Aerospace History, No. 15, 1999).

Tucker, Tom. Touchdown: The Development of Propulsion Controlled Aircraft at NASA Dryden (Monographs inAerospace History, No. 16, 1999).

Maisel, Martin D.; Demo J. Giulianetti; and Daniel C. Dugan. The History of the XV-15 Tilt Rotor ResearchAircraft: From Concept to Flight. (Monographs in Aerospace History #17, NASA SP-2000-4517, 2000).

Jenkins, Dennis R. Hypersonics Before the Shuttle: A History of the X-15 Research Airplane. (Monographs in Aerospace History #18, NASA SP-2000-4518, 2000).

Chambers, Joseph R. Partners in Freedom: Contributions of the Langley Research Center to U.S. MilitaryAircraft in the 1990s. (Monographs in Aerospace History #19, NASA SP-2000-4519).

Waltman, Gene L. Black Magic and Gremlins: Analog Flight Simulations at NASA’s Flight Research Center.(Monographs in Aerospace History #20, NASA SP-2000-4520).

Portree, David S.F. Humans to Mars: Fifty Years of Mission Planning, 1950-2000. (Monographs in AerospaceHistory #21, NASA SP-2002-4521).

Those monographs still in print are available free of charge from the NASA History Division, Code ZH, NASAHeadquarters, Washington, DC 20546. Please enclosed a self-addressed 9x12" envelope stamped for 15 ouncesfor these items.

51